ARTICLE pubs.acs.org/Langmuir
Chiral Imprinting of Diblock Copolymer Single-Chain Particles Gabriel Njikang, Guojun Liu,* and Liangzhi Hong Department of Chemistry, Queen’s University, 90 Bader Lane, Kingston, Ontario K7L 3N6, Canada
bS Supporting Information ABSTRACT: This Article reports the molecular imprinting of polymer single-chain particles that have a radius ∼3.7 nm. For this, the template L-phenylalanine anilide or L-ΦAA and a diblock copolymer PtBA-bP(CEMA-r-CA) were used. Here, PtBA denotes poly(tert-butyl acrylate), and P(CEMA-r-CA) denotes a random block consisting of cinnamoyloxyethyl methacrylate (CEMA) and carboxyl-bearing (CA) units. In CHCl3/ cyclohexane (CHX) with 64 vol % of CHX or at fCHX = 64%, a block-selective solvent for PtBA, PtBA-b-P(CEMA-r-CA) formed spherical micelles. The core consisted of the insoluble P(CEMA-r-CA) block and L-ΦAA, which complexed with the CA groups. Pumping slowly this micellar solution into stirred CHCl3/(CHX) at fCHX = 64% triggered micelle dissociation into single-chain micelles, which comprised presumably a solubilized PtBA tail and a collapsed P(CEMA-r-CA)/L-ΦAA head. Because the solvent reservoir was under constant UV irradiation, the photo-cross-linkable units in the P(CEMA-r-CA) head cross-linked, and the singlechain micelles were converted into cross-linked single-chain micelles or tadpoles. Synchronizing the micelle addition and photoreaction rates allowed the preparation, from this protocol, of essentially pure tadpoles at high final polymer concentrations. Imprinted tadpoles were procured after L-ΦAA was extracted from the tadpole heads. Under optimized conditions, the produced imprinted tadpoles had exceptionally high binding capacity and high selectivity for L-ΦAA. In addition, the rates of L-ΦAA release from and rebinding by the particles were high.
I. INTRODUCTION Molecular imprinting is traditionally done by polymerizing, in the presence of a porogen, a monomer and a cross-linker around a template (analyte) to yield a polymer monolith.1,2 The monolith is then ground and sieved to yield particles that are tens of micrometers in diameter. After the removal of the template by extraction, molecularly imprinted polymer (MIP) particles, which contain cavities with shapes resembling that of the template and surface functional group arrangements that are best suited for template rebinding, are obtained. MIPs rebind with the template or analyte under optimized conditions with high specificity. They have been used as alternatives for antibodies in immunoassays35 and are called plastic antibodies.1 Traditional MIPs are orders of magnitude larger than antibodies. Because of the large size of the MIP particles, the accessibility of their internal binding sites is limited, resulting in low analyte binding capacities. This large size also slows reagent transport into and out of the particles and introduces other complications. For example, it is hard to use these nondispersible MIP particles in a sandwich assay, one of the widely practiced immunoassay protocols.3 These limitations can, in principle, be overcome by solvent-dispersible smaller MIP particles. Thus, we report in this article a novel method for preparing imprinted single-chain polymer particles. Because of the advantages of small MIP particles, there have been quite a few reports on their preparation and characterization.6 Using a multistep synthetic procedure, Ishi-i et al.7 were able to prepare triad molecules consisting of a saccharide unit sandwiched by two anchors and then anchored one triad on the r 2011 American Chemical Society
surface of one fullerene C60 molecule. This was followed by the cleavage of the boronic ester groups that connected the saccharaide unit to the anchors and the creation of an imprinted site. Particles containing an imprinted single site have also been prepared by Zimmerman et al.8 Their procedure involved the cross-linking of the periphery of a dendrimer and then the removal of the dendritic core to yield an imprint. The singlesite particles from the above two approaches were unfortunately not very selective. Recently, Wulff et al.9 performed precipitation polymerization under high dilution conditions in the presence of template and produced nanogels, which had molecular weights down to 4 104 g/mol. Under controlled conditions, they were able to produce nanogels that contained, on average, one imprinted site per particle. The imprinted sites demonstrated selective catalysis. This nanogel imprinting method has been extended by Carboni et al.10 to another catalytic system and by Hoshino et al.11 and Cutivet et al.12 to aqueous systems for biomolecule imprinting. Preparation of MIPs with somewhat larger diameters has also been reported. Zhou et al.13,14 and Li et al.15 have, for example, reported the imprinting of the cores of diblock copolymer micelles, which have diameters typically around tens of nanometers. Traditional precipitation polymerization,11,16,17 dispersion polymerization,18 mini-emulsion polymerization,19 and suspension polymerization20 have also been used to prepare imprinted Received: February 22, 2011 Revised: April 9, 2011 Published: April 29, 2011 7176
dx.doi.org/10.1021/la2006887 | Langmuir 2011, 27, 7176–7184
Langmuir nanoparticles. Even the cylindrical pores of anodized alumina membrane have been used as the mold for fabricating columnar and tubular particles that were imprinted.21,22 Our group develops general methodologies for fabricating, from block copolymers, architectural polymers, nanostructures, and microstructures.2328 To prepare solvent-dispersible singlechain particles,29 we used a diblock copolymer with one block photo-cross-linkable. In a block-selective solvent for the other block, the diblock copolymer formed micelles with the photocross-linkable block as the core.3033 Pumping slowly this micellar solution into a solvent reservoir triggered micelle dissociation into single-chain micelles, which presumably comprised a soluble tail block and an insoluble head block. Because the solvent reservoir was under constant UV irradiation, the head was quickly cross-linked to yield cross-linked single-chain micelles or tadpoles, which were dispersed in the solvent phase by the solubilized tail. Synchronizing the micelle addition and photoreaction rates allowed the preparation, from this protocol, of essentially pure tadpoles free of crosslinked micelles at final polymer concentrations that were tens or probably hundreds of times higher than the critical micellar concentration. This project was initiated mainly to determine if our tadpole preparation methodology could be extended for creating the more complex and potentially more useful imprinted tadpoles. For tadpole imprinting, the chiral template L-phenylalanine anilide (L-ΦAA) was used.
L-ΦAA was used because MIPs imprinted by L-ΦAA have been extensively studied.3436 This would facilitate the comparison between the performance of our particles and the traditional MIPs. The diblock copolymer used was poly(tert-butyl acrylate)block-poly[(2-cinnamoyloxyethyl methacrylate)-ran-[2-(20 -carboxybenzoyloxy)ethyl methacrylate]], PtBA-b-P(CEMA-r-CA):
The PtBA was targeted because it is soluble in a wide range of solvents including cyclohexane and hexane and would facilitate block-selective solvent selection. The CEMA units were used to lock in the single-chain micelle heads by taking advantage of their photo-cross-linking properties.23 The CA units were incorporated into the second block so that they would undergo electrostatic interaction and hydrogen bonding with the amino group and the carbonyl oxygen of L-ΦAA, respectively.1,36,37
II. EXPERIMENTAL SECTION Materials. Cyclohexane (CHX), chloroform, methanol, and methylene chloride were of Fisher Scientific reagent grade and were used without further purification. Pyridine (g99%, Sigma-Aldrich) was freshly distilled over CaH2 before use. Cinnamoyl chloride (98% predominantly trans),
ARTICLE
N,N0 -dicyclohexylcarbodiimide (99%, DCC), and phthalic anhydride (99þ%) were purchased from Sigma-Aldrich and used without further purification. Chloroform-d (D, 99.8%) and deuterated N,N-dimethyl formamide or DMF-d7 (D, 99.5%) were purchased from Sigma-Aldrich and Cambridge Isotope Laboratories Inc., respectively. L-Phenylalanine anilide (L-ΦAA) and D-phenylalanine anilide (D-ΦAA) were synthesized following the protocol of Katz and Davis35 using N-(tert-butoxycarbonyl)L-phenylalanine and N-(tert-butoxycarbonyl)-D-phenylalanine, purchased from Sigma-Aldrich, as the precursors. PtBA-b-P(CEMA-r-CA). PtBA-b-P(CEMA-r-CA) was derived from PtBA-b-PHEMA, where PHEMA denotes poly(2-hydroxyethyl methacrylate). The precursor to PtBA-b-PHEMA was PtBA-b-P(HEMA-TMS), where P(HEMA-TMS) denotes poly(2-trimethylsiloxyethyl methacrylate). PtBA-b-P(HEMA-TMS) was prepared by anionic polymerization.38 The hydrolysis of PtBA-b-P(HEMA-TMS) in THF/ methanol/water yielded PtBA-b-PHEMA.39 Because the preparation of PtBA-b-PHEMA has been reported on many prior occasions by us, the procedures are not repeated here.38,40 To prepare PtBA-b-P(CEMA-r-CA), PtBA-b-PHEMA was reacted with a limiting amount of cinnamoyl chloride to convert 80% of the HEMA units into CEMA following procedures reported before.40 After purification by precipitation from methanol/water at v/v = 1/2 and vacuum drying, the resultant PtBA-b-P(CEMA-r-HEMA), 300 mg containing 0.19 mmol of hydroxyl groups, was dissolved in 5.0 mL of dry pyridine. This was followed by the addition of 300 mg (2.0 mmol) of phthalic anhydride. The reaction mixture was stirred at 50 C overnight under nitrogen. The polymer was recovered by precipitation from a 9:1 v/v methanol/water mixture. It was washed thrice with the same solvent mixture before it was dissolved in THF and precipitated in 0.1 M HCl. The precipitate was dried overnight in a vacuum oven to yield 292 mg of a white flaky material. Imprinted Tadpoles. The imprinted tadpoles were prepared on the basis of a method that was recently reported by us. The method had its roots in the methodologies of Tao et al.41 and Harth et al.42 Figure 1S of the Supporting Information shows the apparatus used for imprinted tadpole preparation. The light beam was from 500 W Hg lamp in an Oriel 6140 lamp housing and powered by an Oriel 6128 power supply. The beam was prefiltered with a 270 nm cutoff filter. The photoreactor had a total internal capacity of 120 mL and was regulated to 22 C by circulating water. The micellar solution from a feed bottle was delivered into the photoreactor by a peristaltic pump via a syringe needle. The tip of the needle was covered by an aluminum foil trumpet. The feed bottle and the reactor formed a closed system to minimize solvent escape during the photolysis. Two types of imprinted tadpoles, Type I and Type II tapdoles, were produced. To prepare Type II imprinted tadpoles, PtBA-b-P(CEMA-rCA) (42.9 mg containing 25 μmol of carboxylic groups) and L-ΦAA (1.5 mg or 6.3 μmol) were dissolved and equilibrated with each other for 6 h in 3.6 mL of CHCl3 before 6.4 mL of CHX was added dropwise under vigorous stirring. The micellar solution was equilibrated overnight before it was pumped at 0.22 mL/min into 40 mL of a CHCl3/CHX solvent mixture that also had a CHX volume fraction (fCHX) of 64%. After the micellar solution was completely pumped into the reactor in 45 min, the mixture was further irradiated for 75 min to achieve a CEMA double bond conversion of 70%, which was determined from the absorbance decrease of PCEMA at 274 nm.23 The irradiated solution was concentrated to 2.0 mL by rotary evaporation and was then dialyzed against 50 mL of methanol/CHCl3 at v/v = 2/8 to remove the imprint molecules. The solvent mixture was changed five times within 3 days. The amount of template extracted was monitored by UV spectroscopic measurements, and dialysis was stopped when changes in the absorbance reading at 242 nm for the external methanol solution became negligible. The imprinted tadpole solution was further concentrated and precipitated from a v/v = 1/9 methanol/water mixture. The precipitate 7177
dx.doi.org/10.1021/la2006887 |Langmuir 2011, 27, 7176–7184
Langmuir was dried in a vacuum oven and stored at 4 C. The amount of extract obtained after rotary evaporation of the solvent was dried in a vacuum and dissolved in methanol. The total amount of template extracted was quantified directly from its absorbance at 242 nm and the molar extinction coefficient of L-phenylalanine anilide at 242 nm in methanol. To prepare Type I imprinted tadpoles, the micelles were prepared initially at a L-ΦAA to CA carboxyl group molar ratio rΦc of 1/7.3. The other half of L-ΦAA was added into the solvent reservoir. After all of the copolymer and the L-ΦAA template were mixed, rΦc was 1/3.6. The Type II imprinted tadpoles were prepared by mixing all L-ΦAA with the diblock copolymer at rΦc = 1/4.0 during the micellization stage, and no L-ΦAA was present initially in the solvent reservoir. Imprinted Nanospheres. The method described above for the preparation of L-ΦAA-containing micelles was used to prepare 10.0 mL of a micellar solution at a polymer concentration of 4.3 mg/mL and rΦc = 1/4.0. It was then irradiated by light under stirring at 22 C for 30 min to cross-link the micelles to yield imprinted nanospheres. Our comparison of the absorbance at 274 nm for PCEMA before and after the photolysis indicated that the PCEMA double bond conversion was 74% for the nanospheres. Extraction of the template from imprinted nanospheres was also done by dialysis. Before drying, the concentrated imprinted nanosphere solution was precipitated from water. Sorption Isotherms. To determine the sorption isotherms in CDCl3, 0.50 mL of the imprinted tadpole solution in CDCl3 at 10.0 mg/mL was used. A given amount of ΦAA, L-ΦAA or D-ΦAA, was dissolved in 100 μL of CDCl3, and then mixed with the imprinted tadpole solution. The amount of free ΦAA present in a sample after overnight equilibration was quantified using 1H NMR spectroscopy by comparing the intensity of the peaks of ΦAA and the internal standard CH2Cl2, which, at 10.0 μL as a 0.52 M solution in CDCl3, was added prior to the measurements. In such an experiment, we first determined the ratio, R0, of the intensities of the ΦAA peaks relative to that of the added internal standard CH2Cl2 in the absence of imprinted tadpoles. The relative intensities, R, were determined again after the addition of imprinted tadpoles. The amount of ΦAA, q, sorbed per gram of imprinted tadpole heads was calculated using: c0 V0 R 1 ð1Þ q¼ R0 m where c0 and V0 are the initial ΦAA solution concentration and volume, respectively, and m is the mass of the imprinted tadpole heads. Only the mass of the heads was used in this calculation because the PtBA tails should not contribute significantly to ΦAA binding.
Release Kinetics of L-ΦAA from Tadpoles and Nanospheres. Imprinted tadpoles or nanospheres (10.0 mg containing
5.8 μmol of CA units) were equilibrated with 0.50 μmol of L-ΦAA in 0.50 mL of CDCl3 overnight. At this stage, the presence of free L-ΦAA could not be detected by 1H NMR spectroscopy. The CDCl3 solvent was then evaporated to a volume 0.25 mL. To this concentrated solution was added 0.25 mL of deuterated DMF (DMF-d7). The tube was swirled for 10 s before it was inserted into the NMR instrument, and data acquisition was started 1 min after DMF-d7 addition. Instruments and Techniques. UV absorbance measurements were performed on a Varian-300 UVvisible spectrophotometer using 1.00 cm Hellma quartz cells. 1H NMR measurements were conducted using a Bruker Avance-500 instrument. Peak integrals were obtained after the spectra had been line fitted with the NMR utilities transform software NUTs. The specific refractive index increment of the copolymer was measured in THF using a Wyatt Optilab rEX refractive index detector. Molecular weights and polydispersity indices were obtained by size exclusion chromatographic measurements (SEC), performed on a Wyatt instrument equipped with an Optilab rEX refractive index detector and a Dawn Heleos-II light scattering detector (LSD).
ARTICLE
III. RESULTS AND DISCUSSION Polymer Characterization. The PtBA-b-PHEMA precursor for PtBA-b-P(CEMA-r-CA) was fully cinnamated to yield PtBAb-PCEMA. PtBA-b-PCEMA was characterized by SEC and 1H NMR to yield the number of repeat units m and n for the first and second blocks. The characterization of PtBA-b-PCEMA was preferred, because both the CEMA and the tBA units, unlike CA units, had good solubility in THF and CDCl3, the solvents used for the characterization experiments. The repeat unit number ratio m/n between PtBA and PCEMA was determined to be 0.62 by comparing the 1H NMR peak integrals of the two blocks. The polydispersity index Mw/Mn, the specific refractive index increment dnr/dc, and the LS molecular weight Mw were determined in THF and were 1.07, 0.137 mL/g, and 2.1 105 g/mol, respectively. The weight-average repeat unit numbers m and n for the PtBA and the PCEMA blocks were calculated to 380 and 620 from the m/n and Mw values determined from 1H NMR and LS, respectively. To prepare PtBA-b-P(CEMA-r-CA), a fraction of the PtBA-bPHEMA hydroxyl groups was reacted with a controlled amount of cinnamoyl chloride to produce PtBA-b-P(CEMA-r-HEMA). The partially cinnamated polymer was then reacted with excess phthalic anhydride to introduce carboxylic functionalities to the remaining HEMA units. Our 1H NMR analysis indicated that the molar fraction of CA was 20% in the second block. The PHEMA block should be randomly cinnamated for the following reasons. First, our past experience indicated that the complete cinnamation of PHEMA was slow and required overnight. Second, the cinnamation was performed by adding dropwise under rigorous stirring a cinnamoyl chloride solution in pyridine into a PtBA-b-PHEMA solution in pyridine. This procedure avoided a local buildup of reactant concentrations and ensured a faster reagent dispersing than reaction rate and therefore the random cinnamation of the PHEMA hydroxyl units. Imprinted Tadpole Preparation. In a block-selective solvent, a diblock copolymer would normally form micelles.30,31,4347 Coexisting with the micelles at the critical micellar concentration (cmc) are some single-chain micelles or unimolecular micelles.48,49 Because the cmc is normally very low for block copolymers, unimolecular micelles have been largely ignored previously. It was only recently that the intramolecular self-assembly of block copolymers in a block-selective solvent has been used to prepare architectural polymers including tadpole molecules29 and macrocycles.50 Because P1 had a low cmc in chloroform/cyclohexane (CHX) at fCHX = 64%, we started tadpole preparation with a micellar solution. This micellar solution with L-ΦAA in the core (structure A, Figure 1) was then pumped slowly into the photoreactor that contained CHCl3/CHX with fCHX also at 64% and optionally L-ΦAA. Immediately after micelle addition, the micelles were in the region protected by the aluminum foil (Figure 1S) and were not irradiated. Thus, the photo-cross-linking of the micelles (process AfF in Figure 1) was prevented. Rather, the micelles dissociated into unimers (AfB). As the unimers diffused away from the protected region, they became exposed to the irradiation and were converted into cross-linked tadpoles (BfC). Imprinted tadpoles were produced after the removal of L-ΦAA from the tadpole heads by dialysis against CHCl3/CH3OH (CfD). More than 85% of the L-ΦAA introduced to the copolymer was removed this way. The micelles dissociated at the early stage of the experiment inside the solvent reservoir, because there was no polymer present 7178
dx.doi.org/10.1021/la2006887 |Langmuir 2011, 27, 7176–7184
Langmuir
ARTICLE
Figure 2. Comparison of SEC traces of P1 and its resultant tadpoles. Figure 1. Schematic diagram of processes for imprinted tadpole preparation and those that may occur simultaneously.
and the micelle/unimer equilibrium favored unimer formation. The added micelles dissociated throughout the preparation because the unimers produced earlier would have been converted into cross-linked tadpoles. Thus, the system remained starved of unimers, provided that the polymer addition rate was lower than that of tadpole formation. As the concentration of tadpoles increased, the tadpoles could also form micelles (CfE). As was demonstrated before, tadpole micelle cross-linking (EfF) could be prevented by optimizing fCHX.29 At an optimized fCHX, the heads of the tadpole molecules were not tightly associated and would not undergo intertadpole cross-linking to yield cross-linked tadpole micelles. This was reasonable because CEMA dimerization is not a facile reaction. It occurs only if the double bonds of two CEMA units are within 0.4 nm and are aligned during the triplet lifetime of an excited CEMA unit.51 The intertadpole cross-linking process could also be eliminated by introducing some inert and soluble groups into the head block. These groups should be concentrated on the surface of the tadpole heads to prevent the coupling of the different heads.52 Two types of imprinted tadpoles were produced. Type I was produced by adding initially one-half of the desired L-ΦAA into the photoreactor. Type II imprinted tadpoles were produced by mixing initially all L-ΦAA with the diblock copolymer micelles. All samples were prepared at a final P1 concentration of 0.85 mg/mL. This should be orders of magnitude higher than the cmc of the polymer as was demonstrated previously for a polymer with a similar composition under similar solvation conditions.29 Figure 2 compares the size exclusion chromatography (SEC) traces of P1 and an imprinted Type I tadpole sample. These traces were obtained using the refractive index detector, whose signal intensity is proportional to the polymer amount eluted. The data suggested the following: First, the tadpole peak was delayed relative to the P1 peak by 0.5 min. Second, no SEC peaks were observed on the higher molecular weight or shorter elution time side for the tadpole sample even though it was prepared at a final polymer concentration of 0.85 mg/mL. The tadpole peak shifted to the lower molecular weight side because the P(CEMA-r-CA) block in the tadpoles was crosslinked, and the tadpoles were smaller than their precursor in the SEC eluant THF, which solubilized both PtBA and un-crosslinked P(CEMA-r-A). Our analysis of the SEC traces yielded the
PS-equivalent molecular weights of 9.8 104 and 1.22 105 g/ mol at the maxima of the tadpole and P1 peaks. The hydrodynamic volumes Vp of the samples at their elution peak maxima could be calculated from the following equation: Vp ¼ ½ηp Mp =ð2:5NA Þ
ð2Þ
where NA was the Avogadro number; and the intrinsic viscosity [η]p of PS, in THF, with molecular weight Mp could be obtained from:53 ½ηp ¼ 1:10 102 Mp 0:725 mL=g
ð3Þ
Using eqs 2 and 3 and the PS-equivalent molecular weights for the P1 and the tadpoles, we computed a 32% reduction in the hydrodynamic volume of the tadpoles relative to P1.29 If a spherical shape was assumed for the solvated tadpole, which consisted of a head and a tail, and Vp = (4/3)πRh3, the sphereequivalent hydrodynamic radius Rh was calculated to be 8.9 nm for the tadpoles. That the tadpoles consisted indeed of single chains was verified before.41 In that study, we determined that the molecular weights of the precursory polymer and the resultant tadpoles were essentially identical despite their hydrodynamic volume difference. The lack of a high molecular weight peak for the tadpole sample of Figure 2 and other tadpole samples suggested that the tadpoles were produced essentially free of cross-linked micelles and multimers and the effectiveness of our strategy for singlechain particle preparation. We also analyzed gravimetrically an imprinted tadpole sample eluted out from the SEC columns between the retention times of 20.5 and 24.0 min. The injected sample was essentially quantitatively recovered. This confirmed that the absence of higher molecular weight peaks for the tadpole sample was indeed due to the low concentration of these species and was not an artifact derived from the fact that these species were retained by the SEC filter or columns.52 It should be mentioned that obtaining results of Figure 2 required quite some optimization. As established before,29 the most effective method to eliminate cross-linked micelles was to reduce fCHX. Once an fCHX yielding a significant fraction of tadpoles was identified, the tadpole yield was further increased by reducing the rate of micelle addition into the photoreactor. This could be achieved more effectively by adding a larger amount of a more dilute micellar solution than by pumping in more slowly a more concentrated micellar solution. 7179
dx.doi.org/10.1021/la2006887 |Langmuir 2011, 27, 7176–7184
Langmuir
ARTICLE
Figure 3. L-ΦAA and D-ΦAA sorption isotherms for the Type I (left) and Type II (right) imprinted tadpoles. The solid curves represent the biLangmuir and Langmuir fit to the L-ΦAA and D-ΦAA sorption data, respectively. The dashed curves were obtained by switching the data treatment model from Langmuir to bi-Langmuir or vice versa.
Detection of ΦAA Uptake by Imprinted Tadpoles. Figure 2S in the Supporting Information compares 1H NMR spectra of L-ΦAA in CDCl3 with and without imprinted Type I tadpoles. The intensities of the L-ΦAA peaks decreased drastically after tadpole addition, and no peaks were seen for the P(CEMA-r-CA) heads. No tadpole heads were seen, because the tumbling motion of the P(CEMA-r-CA) protons inside the heavily cross-linked heads was too slow and the peaks were excessively broadened out. The L-ΦAA peak intensities decreased sharply after the addition of the imprinted tadpoles and disappeared eventually with the addition of more imprinted tadpoles because the tumbling motion of the sorbed L-ΦAA was slow as well. Thus, only the free ΦAA in the solvent phase exhibited NMR signals. This allowed the quantification, by 1H NMR, of the amount of unbound ΦAA, which can be either L-ΦAA or D-ΦAA, present in any mixtures of ΦAA and imprinted tadpoles. Adsorption Isotherms. Using this detection method, we constructed isotherms of L-ΦAA and D-ΦAA sorption by the two types of imprinted tadpoles. The isothermal data were measured at room temperature in CDCl3, a good solvent for both PtBA and P(CEMA-r-CA). Figure 3 shows the amount of ΦAA sorbed per gram of tadpole heads, q, against the ΦAA concentration (cf) in the solvent phase at the sorption equilibrium. Figure 3 revealed that the L-ΦAA-imprinted tadpoles had a higher sorption capacity (q) for L-ΦAA than for D-ΦAA at any cf. This was the case for both the Type I and the Type II imprinted tadpoles. Thus, the polymer single-chain particles were successfully imprinted. Models for Sorption Data Analysis. The isothermal data can be treated to yield the capacities and sorption equilibrium constants for the different binding sites. Traditional MIPs are known to have a wide range of sorption sites with different binding energies or constants (K).54,55 Their sorption isotherms should be treated, in principle, by an integral equation containing the site distribution function and local sorption models.54 Assuming the validity of the Langmuir model for all of the binding sites, the integral equation for q at the sorbent equilibrium concentration cf is:54 Z ¥ Kcf q¼ qðKÞ dK ð4Þ 1 þ Kcf 0
where q(K) is the binding capacity distribution function, which describes how the binding capacity varies as a function of binding constant K.
In reality, the isothermal data are plagued by errors, contain insufficient data points, and cover a narrow concentration cf range. Because of these, the data cannot be treated by the integral equation to yield the distribution function q(K). Rather, the integral equation is approximated by several terms corresponding to several types of binding sites: X q i Ki c f ð5Þ q¼ 1 þ Ki c f i Thus, the treatment of sorption data by assuming the presence of only one type of binding site, or the Langmuir model, yields the binding capacity q1 and binding equilibrium constant K1 for the sites. The assumption of two types of binding sites, or a biLangmuir model, would yield four fitting parameters, which are q1, K1, q2, and K2. Evidently, the treatment of sorption data derived from n types of binding sites requires the use of 2n fitting parameters. If n is large and the number of data points in a data series are limited, obtaining accurate qi and Ki values becomes difficult. In addition, curve fitting does not allow the accurate determination of the qi and Ki values if a particular type of binding sites does not contribute significantly or contribute to a constant degree to the sorbed amount, q, over the experimentally probed cf range. The latter scenario occurs if Ki is very large, so that Kicf . 1 over the cf range of the data. The former scenario can, for example, correspond to the case when qi is small or when qi is reasonably large but Ki is small so that Kicf , 1 over the cf range studied. Thus, the Ki values that can be probed with confidence are only those that have values between approximately 1/cmax and 1/cmin, where cmax and cmin are the maximum and minimum cf concentrations used in an isothermal data set. Intuitively, only nonspecific binding sites should exist for D-ΦAA sorption by the L-ΦAA-imprinted tadpoles. Properly and nonproperly imprinted sites should exist for L-ΦAA sorption by the L-ΦAA-imprinted tadpoles. A properly imprinted site has not only the correct cavity shape but also the proper cavity surface functional group arrangement to bind L-ΦAA tightly. These sites may lie most likely in the interior of the tadpole heads. The improperly imprinted sites might have the right surface carboxyl group arrangement but lacked the right cavity shape for accommodating L-ΦAA. Alternatively, the cavity shape might have been correct but the surface carboxyl groups might be mis-aligned. 7180
dx.doi.org/10.1021/la2006887 |Langmuir 2011, 27, 7176–7184
Langmuir
ARTICLE
Table 1. Results from Treating the Data of ΦAA Sorption by L-ΦAA-Imprinted Tadpoles sorbent
q1 (μmol/g)
103 K1 (M1)
a L-ΦAA
Type I
297
0.142
L-ΦAA
Type I
237
0.60
0.9907
D-ΦAA
Type I
58
0.73
0.9865
sorbate
D-ΦAA
Type I
56
0.49
L-ΦAA
Type II
238
0.98
D-ΦAA
Type II
132
0.59
q2 (μmol/g) 66
8.2 28
103 K2 (M1)
Rv2
3.6
0.9993
2.8
0.9910
3.6
0.9933 0.9959
a
Data in bold were derived from treating the L-ΦAA and D-ΦAA sorption data by the bi-Langmuir and Langmuir models, respectively. Data in regular font were obtained by switching the data treatment model from the Langmuir to the bi-Langmuir model or vice versa.
Thus, the D-ΦAA and L-ΦAA sorption data were treated by the Langmuir and bi-Langmuir model, respectively. Nonproper imprinting can occur for various reasons, including the deformation of the polymer network in an imprinted tadpole head after template extraction. A mechanism unique to our system might also have existed. This could be due to the mobility of the template L-ΦAA molecules during tadpole preparation. In the preparation of traditional MIPs, the L-ΦAA and COOH concentrations are essentially constant throughout the monomer cross-linking process. Meanwhile, these two concentrations would change continuously in our preparations. For example, the feed molar ratio for L-ΦAA and COOH for our Type I tadpoles at the end of the preparation should be 1/3.6. The tadpoles prepared at the early stage should have a [L-ΦAA]/[COOH] value much larger than 1/3.6, because L-ΦAA was added into the photoreactor before the micelles. At the end of the tadpole preparation, this ratio should not deviate much from 1/3.6 among different tadpoles. This was possible only if some L-ΦAA molecules dissociated from the tadpoles prepared at the early stage. Thus, the association state and the association partner(s) of one L-ΦAA molecule might change many times during tadpole preparation. These changes in the association state and partners would be facilitated by the fact that reagent diffusion into and out of the ultrafine particles was fast. Results of Isothermal Data Analysis. Shown in bold in Table 1 are the parameters generated from treating the L-ΦAA and D-ΦAA sorption data by the bi-Langmuir and Langmuir model, respectively. The results from treating the L-ΦAA and D-ΦAA sorption data using the Langmuir and bi-Langmuir model, respectively, are also shown but in regular font for comparison. In every case, the coefficient of determination, Rv2, for curving fitting was given.56 A good fit was supposed to produce a Rv2 value very close to 1. The treatment, by the Langmuir model, of the data of L-ΦAA sorption by the Type I imprinted tadpoles yielded a Rv2 value of 0.9907. This value increased to 0.9993 when the bi-Langmuir model was used. The fitting curves derived from the two models are shown in Figure 3. The bi-Langmuir curve agreed with the experimental data substantially better. In contrast, only a small increase in the Rv2 value from 0.9865 to 0.9910 was observed when the fitting function was changed from the Langmuir to the bi-Langmuir model while treating the data of D-ΦAA sorption by the Type I imprinted tadpoles. A further examination of the curves (Figure 3) generated from these two models revealed that the two curves differed little at cf > 0.38 mM, the lowest cf value used. They differed only in the low cf range. For example, the bi-Langmuir curve had an upward swing at cf values
