Memorization of coordination environments in

RIKEN Review No. 45 (March, 2002): Focused on Nanomaterials Research in RIKEN

Memorization of coordination environments in ultrathin titania films Junhui He,∗1 Izumi Ichinose,∗1 Aiko Nakao,∗2 and Toyoki Kunitake ∗1 ∗1

Topochemical Design Laboratory, RIKEN Frontier Research System ∗2

Surface Characterization Division, RIKEN

We have previously demonstrated that metal oxide ultrathin films, by themselves, can produce imprinted cavities for organic molecules. In the present study, memorization of the coordination environments of metal ions was achieved by incorporation of diamine chelating ligands in thin films of metal oxides. The films imprinted with Cu2+ or Zn2 + showed selective binding to the template over the control metal ion. The selectivity is attributed to the immobilized geometry of ethylenediamine ligands. The ligand that is doubly linked to the matrix showed enhanced stability of binding capacity relative to that of a singly linked ethylenediamine ligand.

Introduction The aim of molecular imprinting is to immobilize shapes and functionalities of templates in solid matrices via covalent 1) and noncovalent 2) interactions. This strategy has been applied to a wide range of organic templates, such as pharmaceuticals, pesticides, peptides, nucleotide bases, steroids, and sugars. Metal ions have been considered to be important targets, although they have received much less attention.3, 4) The conventional imprinting technique is associated with some drawbacks, such as low binding efficiency, a slow binding process and difficult fabrication, as the imprinted sites are embedded in bulk (usually hydrophobic) polymer matrices, and ready access of guest molecules to imprinted sites is often suppressed. These drawbacks are relieved at least partially by the use of surfaces, e.g., metal-ion imprinting on the surface of polymer particles 5) and on the inner wall of microporous silica.6) Some years ago, we developed an imprinting technique for organic molecules in ultrathin TiO2 -gel matrices.7) Recently, we succeeded in the imprinting of metal ions in ultrathin TiO2 -gel films by incorporation of metal-ion ligands.8) The present article gives a full account of the metal-ion imprinting.

that was continued for 6 h. Then, 0.8525 mL of Ti(OBu)4 was added with stirring, and the mixture was stirred at room temperature overnight (12 h). The final composition was: Cu2+ /10 mM, SiEN/20 mM, Ti(O-n Bu)4 /100 mM, H2 O/100 mM. n

Imprinted films were assembled layer-by-layer on substrates by the surface sol-gel process. A quartz crystal microbalance (QCM, 9 MHz) device manufactured by USI System, Japan, was used for monitoring film assembly. The surface sol-gel process was conducted by dipping mercaptoethanol-modified, gold-coated QCM electrodes in precursor solutions at 30◦ C for 10 min, followed by rinsing in toluene, drying with nitrogen gas, and hydrolysis in air (Fig. 1). The frequency was recorded when its change due to progress of the hydrolysis of the surface alkoxide groups became insignificant. Chemisorption, rinsing, drying, and hydrolysis constitute one cycle of the adsorption process. Ten cycles were repeated unless otherwise indicated. To remove template metal ions, the imprinted film was immersed in aqueous HCl (pH 4) for 3–10 min, rinsed with pure

Experimental Zn(NO3 )2 ·6H2 O, 2-mercaptoethanol, Cu(NO3 )2 ·3H2 O, aqueous HCl, and NaOH of 1 N were purchased from Kanto Chemical. N-(2-Aminoethyl)-3-aminopropyltrimethoxysilane (SiEN), bis[3-(trimethoxysilyl)propyl]ethylenediamine (2SiEN) (62% in methanol) and Ti(O-n Bu)4 were obtained from Gelest. All these chemicals were used as purchased. Ultrapure water with a specific resistance of 18.3 MΩ·cm was produced using Yamato-WQ500 (Millipore, Japan). Aqueous solutions of given pHs were prepared from ultrapure water and 1 N HCl or 1 N NaOH. A typical precursor solution for film assembly was prepared as follows: 0.0604 g of Cu(NO3 )2 ·3H2 O was dissolved in 12 mL of ethanol, 0.1121 mL of SiEN was added, and the mixture was stirred for 1 h. After addition of 12 mL toluene, 0.0451 mL of pure water was added with stirring

Fig. 1. Schematic procedure of film assembly.

37

water and dried with nitrogen gas. It was then treated with aqueous NaOH (pH 10) for 3–10 min. After rinsing with pure water and drying by nitrogen gas, the frequency was measured. Rebinding ability was evaluated by immersing the imprinted film in a metal-ion ethanolic solution (10 mM) for 10 min, followed by rinsing with ethanol, drying by flushing nitrogen gas, and the frequency measurement. The bound mass was calculated from the frequency shift using the Sauerbrey equation.9) A frequency decrease of 1 Hz corresponds to a mass increase of 0.87 ng in our system. Films on QCM electrodes were cleaved, and coated with 2 nm of Pt using a Hitachi E-1030 ion-coater. Then, cross sections of the films were observed on a Hitachi S-900 scanning electron microscope (SEM). Competitive binding was conducted in a mixture of Cu(NO3 )2 and Zn(NO3 )2 (5 mM each) in ethanol, followed by rinsing and drying. Imprinted films were assembled on microscopic glass slides. Then, the X-ray photoelectron spectroscopy (XPS) measurements were carried out on ESCALAB 250 (VG) using Al Ka (1486.6 eV) radiation. The applied power was operated at 15 kV and 20 mA. The base pressure in the analysis chamber was less than 10−8 Pa. Smoothing, background removal and peak fitting were carried out with a VG analysis software package, ECLIPS. Peak positions were normalized to the carbon peak at 285 eV.

Results and discussion Preparation of metal-ion imprinted films We selected Cu2+ and Zn2+ ions as templates because they have identical charges, almost the same ionic radii, and constitute one of the most stringent tests for the imprinting concept. SiEN and 2SiEN were chosen as the ligands. They are

Scheme 1.

ethylenediamine (en) derivatives, one or two nitrogen atoms of which are linked to silicon methoxide groups through trimethylene chains (Scheme 1). When Cu(NO3 )2 ·3H2 O and SiEN or 2SiEN were mixed in the 1 : 2 molar ratio in solvent, [Cu(SiEN)2 ]2+ and [Cu(2SiEN)2 ]2+ complexes were formed, respectively.8) Ti(On Bu)4 was used as the matrix-forming monomer. Silicon methoxide groups link the complexes to the matrix. Film assembly was carried out in the absence and the presence of the templates under the same conditions. As shown in Fig. 2, regular film growth was achieved in all these experiments. However the average frequency shift per cycle increased in this order: 592 Hz (Fig. 2 a, none) < 1295 Hz (Fig. 2 b, Cu2+ ) < 2808 Hz (Fig. 2 c, Zn2+ ), indicating that the metal ions accelerated hydrolysis of alkoxides, with Zn2+ having a greater effect than Cu2+ . In fact, selfcondensation of SiEN was observed 10 min after SiEN and Zn2+ were mixed in ethanol, resulting in the formation of fine SiEN/Zn2+ aggregates, which were attached onto the

38

Fig. 2. QCM frequency shifts during assembly of unimprinted (a), Cu2 + -imprinted (b), and Zn2 + -imprinted (c) films: metal ion/0 or 10 mM, SiEN/20 mM, Ti(O-n Bu)4 /100 mM, H2 O/100 mM. (d) Assembly of Zn2+ -imprinted film by a modified procedure: Zn2 + /10 mM, SiEN/20 mM, Ti(O-n Bu)4 /100 mM, H2 O/0 mM.

TiO2 -gel film but easily removed by treatment with aqueous HCl. To fabricate robust Zn2+ -imprinted films, the precursor solution was prepared by a modified procedure: 0.0744 g of Zn(NO3 )2 ·6H2 O was dissolved in 12 mL of ethanol. To this solution were added first 0.8525 mL of Ti(O-n Bu)4 and second 0.1121 mL of SiEN, with stirring. Then 12 mL of toluene was added and the mixture was stirred. The final concentrations were: Zn2+ /10 mM, SiED/20 mM, Ti(On Bu)4 /100 mM, H2 O/0 mM. No self-condensation of SiEN was noted in this procedure. SiEN was probably stabilized by attaching to Ti(O-n Bu)4 molecules. The obtained mixture was used for dipping at room temperature. As shown in Fig. 2 d, a better linearity was achieved for film assembly with an average frequency shift of 416 Hz per cycle than in the case of Fig. 2 c. For these films, the thickness was estimated to be 95 nm (Fig. 2 a), 210 nm (Fig. 2 b), 450 nm (Fig. 2 c), and 67 nm (Fig. 2 d).10) Similar results were also obtained when 2SiEN was used as the ligand. The film thicknesses obtained under the identical conditions were 98 nm (Cu2+ imprinting, 5 cycles) and 30 nm (Zn2+ imprinting, 6 cycles), respectively. The film thickness can be controlled by changing the number of adsorption cycles. It can also be regulated by water content in precursor solution and by the rinsing solvent. Table 1 shows the effect of water content in the precursor solution on film thickness. When toluene was used as rinsing solvent, the thickness of each layer increased with the water content. In contrast, when good solvents such as 2propanol, ethanol and toluene/ethanol (1/1, v/v) were used, the layer thickness (3.2–4.4 ˚ A) was surprisingly independent of the water content. The Ti–O bond length is 1.960 ˚ A in the tetrahedral structure of TiO2 . The thickness of the TiO2 monolayer formed in the surface sol-gel process should be less than 3.920 ˚ A. Therefore monolayer assembly appears to be achieved in good solvents due to removal of loosely bound aggregates. Nine to twelve cycles were repeated on QCM electrodes. The obtained films (3.0–5.3 nm) were transparent and fairly uniform. The metal ions were removed by acidifying the amine groups with aqueous HCl (pH 4), followed by neutralization with aqueous NaOH (pH 10). The QCM frequency increased after removing the metal ions. This frequency shift is related to the removed mass of template and denoted as ∆fremoved . The total frequency shift during film assembly is related to

Table 1. Effect of water content and rinsing solvent on layer and film thicknesses.a Water contentb Rinsing solvent d/cycle (˚ A) d (nm) 1 toluene 210 210 2 toluene 1200 470 1 2-propanol 3.3 3.0 2 ethanol 4.4 5.3 2 toluene/ethanol (1/1, v/v) 3.2 3.2 a Cu2+ /10 mM, SiEN/20 mM, Ti(O-n Bu) /100 mM. b The water content is repre4 sented by the molar ratio of H2 O to Ti(O-n Bu)4 . Table 2. QCM frequency changes due to film assembly and removal of metal ion. Water content Rinsing solvent −∆ffilm (Hz)a ∆fremoved (Hz) ∆fremoved /−∆ffilm 1 toluene 12951 1550 1/8 5013 1/6 2 toluene 29332b 2 toluene/ethanol 202 31 1/7 2 ethanol 330 40 1/8 a 10 cycles unless otherwise indicated, b 4 cycles. Cu2+ /10 mM, SiEN/20 mM, Ti(O-n Bu) / 4 100 mM.

the film mass and denoted by −∆ffilm . As shown in Table 2, thicker films incorporate more template metal ions. However, the ratio, ∆fremoved /−∆ffilm is constant (1/6–1/8) for all samples. The ratio, mtemplate /mfilm was estimated to be 1/7 from the composition of precursor solution, where mtemplate and mfilm are mass of the incorporated template and film mass, respectively.11) ∆fremoved /−∆ffilm agrees with mtemplate /mfilm , implying that the template has been completely removed. When 2SiEN was used as the ligand, films of various thicknesses were prepared by changing the number of cycles repeated. Identically, the ∆fremoved /−∆ffilm values agree well with mtemplate /mfilm at all the film thicknesses. The above results are supported by XPS experiments (Fig. 3). In the XPS spectrum of an as-prepared Zn2+ imprinted film (Fig. 3 (a)), peaks were observed at 101.6 eV (Si 2p) and 399.9 eV (N 1s), 1022.4 eV (Zn 2p3) and 406.7 eV (N–O),

Fig. 3. XPS spectra of (a) an as-prepared Zn2 + imprinted film and (b) the same film after removal of Zn2 + ions by aqueous HCl (pH 4).

showing that the ligand and Zn2+ /NO− 3 had been incorporated into the TiO2 -gel film (458.6 eV/Ti 2p3) during the film assembly. After the film was treated with aqueous HCl and NaOH, the Zn2+ peak disappeared (Fig. 3 (b)).

Metal-ion recognition by imprinted films When an imprinted film was immersed in an ethanolic solution of metal ion (10 mM), a decrease of the QCM frequency was recorded. From this frequency shift, the mass of bound metal salt was estimated. By dividing the mass with the molecular weight of the metal salt, we could obtain the mole number of the bound mass. The effect of

Fig. 4. Repeated binding of Cu2 + and Zn2 + onto Cu2 + ((a) 120 nm) and Zn2 + ((b) 67 nm) imprinted SiEN films and onto Cu2 + ((c) 98 nm) and Zn2 + ((d) 30 nm) imprinted 2SiEN films. The extent of binding was estimated from QCM frequency changes.

39

imprinting was studied by means of rebinding experiments. When SiEN was used as the ligand, the Cu2+ imprinted film (Fig. 4 (a)) demonstrates greater binding to Cu2+ than to Zn2+ . The selectivity calculated from the first two bindings is 4.3 (m1−Cu2+ /m2−Zn2+ ). The Zn2+ -imprinted film (Fig. 4 (b)), on the other hand, demonstrates greater binding to Zn2+ than to Cu2+ . The selectivity (m1−Zn2+ /m2−Cu2+ ) is 3.2. Deterioration of the imprinting effect was also noted in Fig. 4 (a) and (b). After 4 cycles, the binding capacity drops to 59% (m5−Cu2+ /m1−Cu2+ , Fig. 4 (a)) and 21% (m5−Zn2+ /m1−Zn2+ , Fig. 4 (b)) of the original value. This might be due to smaller stability of imprinted sites (ethylenediamine groups). When 2SiEN was used as the ligand, however, both selectivity and stability of binding capacity were enhanced (Fig. 4 (c) and (d)). The selectivity is 10 (m1−Cu2+ /m2−Zn2+ ) and 1.3 (m1−Zn2+ /m2−Cu2+ ) for the Cu2+ - and Zn2+ - imprinted films, respectively. After 4 cycles, 93% of the binding capacity remains for the Cu2+ -imprinted film (m5−Cu2+ /m1−Cu2+ = 93%), and 71% of the capacity remains for the Zn2+ -imprinted film (m5−Zn2+ /m1−Zn2+ = 71%). Competitive binding was also tested by immersing the Cu2+ -imprinted film in a mixture of Cu(NO3 )2 and Zn(NO3 )2 (5 mM each) in ethanol, followed by XPS measurements. The atomic ratio of bound Cu2+ to Zn2+ , as determined by XPS, was 7.5 (SiEN film, 105 nm) and 8.2 (2SiEN film, 40 nm), respectively, in agreement with the selectivity values, 4.3 (SiEN) and 10 (2SiEN), obtained from QCM studies.

Effect of film thickness It was found that the frequency shift due to rebinding (−∆frebinding ) was smaller than that due to template removal (∆fremoved ). This indicates that only part of the imprinted sites functioned. −∆frebinding /∆fremoved may be used as a qualitative estimate of effective binding sites. Figure 5 shows the results for Cu2+ -imprinted SiEN films. The film thicknesses were estimated from QCM frequency changes during film assembly. At a film thickness of 3.2 nm, all the imprinted sites were accessible and effective. However, with increasing film thicknesses, binding efficiency decreased sharply. It is apparent that metal ions cannot access the lower layers of the film. From Fig. 5, the thickness of accessible upper lay-

Fig. 5.

40

Dependence of binding efficiency on film thickness.

Fig. 6. Scanning electron micrograph of a Zn2 + -imprinted SiEN film.

ers was roughly estimated to be less than 30 nm under the present experimental conditions.

SEM observation of imprinted films SEM observation showed that imprinted ultrathin films are generally flat, smooth and uniform. The SEM image of a Zn2+ -imprinted SiEN film is shown in Fig. 6 as an example. The quartz crystal, gold electrode, and the imprinted film are all clearly seen. Because the film is very thin, cleaving the QCM electrode caused some fractures along the film edge. However, uniform film thickness of around 60 nm is still observed in a large area, which agrees with that (67 nm) estimated from the QCM frequency shift.

Nature of imprinted sites Cu2+ and Zn2+ have the identical charges and almost the same ionic radii. However, they assume different configurations as ethylenediamine chelates. The selectivity of imprinted films may arise from the configurational difference. [Cu(en)2 ]2+ has a square planar configuration (Fig. 7 a), while [Zn(en)2 ]2+ has a tetrahedral configuration (Fig. 7 b). Al-

Fig. 7. Configurations of [Cu(en)2 ]2 + (a) and [Zn(en)2 ]2 + (b) and schematic representations of Cu2 + -imprinted SiEN (c) and 2SiEN (e) films and Zn2 + -imprinted SiEN (d) and 2SiEN (f) films.

though the stability constant of [Cu(en)2 ]2+ (β2 = 19.6) is much larger than that of [Zn(en)2 ]2+ (β2 = 10.62), the Zn2+ imprinted film shows greater binding towards Zn2+ . During the surface sol-gel process, different configurations of the diamine ligand are fixed in matrices, depending on the template metal ion. Removal of the template leaves behind nanocavities in which amine ligands are immobilized in the same configuration as the template/ligand complex. Therefore, the imprinted film can provide selective rebinding to its template rather than to the control (Fig. 7 c, d, e, and f). In the SiEN molecule, only one end of the diamine is linked to the TiO2 gel network (Fig. 7 c and d), resulting in less fixed imprinted configurations. The imprinted site might gradually lose its original configuration upon treatment with aqueous HCl and NaOH or it may be spatially rearranged by repeated binding of metal ions (post-imprinting). When 2SiEN is used, however, the double linkages of the diamine group to the TiO2 -gel network help fix these configurations more stably within ultrathin films, providing a stable binding capacity (Fig. 7 e and f). Other interactions, such as hydrogen bonding and van der Waals interactions might also help fix the configurations, contributing to the selectivity and stability of binding capacity.

Conclusion The surface sol-gel process was previously shown to provide a useful tool for imprinting organic molecules in metal oxide matrices themselves. It is clear that incorporation of chelating function made it possible to create imprinted cavities for metal ions. The ultrathin characteristic of imprinted films provides ready access of metal ions to imprinted sites compared with conventional imprinted bulk materials. By chang-

ing the conditions for film assembly, film thickness can be controlled. 2SiEN molecules can effectively enhance the selectivity and stability of the imprinting effect by their double linkages to the matrix. The imprinted films can be assembled on planar surfaces as well as curved surfaces. This feature enables us to modify materials of various morphologies with imprinted functionalities. Furthermore, the present strategy could be extended to the imprinting of various targets by carefully choosing functional alkoxides. J. He is grateful to the Japan Science and Technology Corporation (JST) for a STA fellowship. References 1) G. Wulff: Angew. Chem., Int. Ed. Engl. 34, 1812 (1995). 2) G. Vlatakis, L. I. Anderson, R. M¨ uller, and K. Mosbach: Nature 361, 645 (1993). 3) W. Kuchen and J. Schram: Angew. Chem., Int. Ed. Engl. 27, 1695 (1988). 4) H. Chen, M. M. Olmstead, R. L. Albright, J. Devenyi, and R. H. Fish: Angew. Chem., Int. Ed. Engl. 36, 642 (1997). 5) K. Y. Yu, K. Tsukagoshi, M. Maeda, and M. Takagi: Anal. Sci. 8, 701 (1992); K. Uezu, M. Yoshida, M. Goto, and S. Furusaki: Chemtech 29, 12 (1999). 6) S. Dai, M. C. Burleigh, Y. Shin, C. C. Morrow, C. E. Barnes, and Z. Xue: Angew. Chem., Int. Ed. Engl. 38, 1235 (1999). 7) S.-W. Lee, I. Ichinose, and T. Kunitake: Langmuir 14, 2857 (1998). 8) J. He, I. Ichinose, and T. Kunitake: Chem. Lett. 2001, 850. 9) G. Sauerbrey: Z. Phys. 155, 206 (1959). 10) I. Ichinose, H. Senzu, and T. Kunitake: Chem. Lett. 1996, 831. 11) mtemplate /mfilm = CCu(NO3 )2 ·3H2 O ×MCu(NO3 )2 /[CTi(O-n Bu)4 × MTiO2 + C2SiEN × (M2SiEN-6CH3 ) + CCu(NO3 )2 ·3H2 O × MCu(NO3 )2 ], C: concentration, M : formula weight.

41