Tuning the Hydrophilic, Hydrophobic, and Ion Exchange Properties of

Mar 31, 2009 - Different charged guest molecules were incorporated on top or into the supported membranes. The host−guest interactions were found to...
0 downloads 13 Views 2MB Size
Download PDF
pubs.acs.org/Langmuir © 2009 American Chemical Society

Tuning the Hydrophilic, Hydrophobic, and Ion Exchange Properties of Mesoporous TiO2 Dereje HailuTaffa, Murugavel Kathiresan, and Lorenz Walder* :: :: Institute of Chemistry, University of Osnabruck, Barbarastrasse. 7, D-49069 Osnabruck, Germany Received November 17, 2008. Revised Manuscript Received February 24, 2009 Alkyl phosphonic acids (Pho-Cn-R) of different chain length (6, 10, and 14 carbons) bearing neutral, positive, and negatively charged head groups (R = -H, R- = sulfonate, R+ = pyridinium) were prepared and anchored to the inner walls of randomly sintered mesoporous TiO2 thin films. Quartz crystal microbalance (QCM) and Fourier transform infrared (FT-IR) measurements show that a monolayer coverage was achieved. The monolayer crystallinity is lower as compared to alkyl thiols on gold, but it increases with the length of the carbon chain. The neutral phosphonic acid modifier makes the TiO2 highly hydrophobic and suppresses electrochemistry in aqueous media, and the alkyl phosphonic acids with charged head groups render the TiO2 film as an ion exchanger with a phase separated hydrophilic and hydrophobic portion. Different charged guest molecules were incorporated on top or into the supported membranes. The host-guest interactions were found to be electrostatic, hydrophobic, or both. Highly charged electroactive metal complexes ([Fe(CN)6]4-, [IrCl6]2-) and purpose-synthesized organic electrochromophores (dialkylated viologens with variable chain length, C1-V+2-Cn, Cn-V+2-Cn, n = 6, 10, and 14) were used as molecular guests, and the assemblies were characterized by cyclic voltammetry and FT-IR. Using the preconcentration phenomenon, [Fe(CN)6]4- concentration as low as 200 nM can be detected on a Pho-C14-R+ modified TiO2 electrode by conventional cyclic voltammetry. The new surface modification technique simplifies the molecular requirements for functional surface modifiers considerably. Using a limited set of organic anchors with orthogonal coordination properties and adjustable hydrophobicity, a broad range of electrochromophores, redox active wiring compounds, or sensitizers can be adsorbed onto TiO2.

Introduction Mesoporous TiO2 thin films composed of randomly sintered nanosized particles have been surface functionalized with different compounds for various applications such as photovoltaics, electrochromics, and sensors.1-5 The main features of mesoporous TiO2 thin films are high surface area (roughness factor of 100 per μm thickness), chemical stability, nontoxicity, low price, ease of fabrication, and semiconducting properties with a band gap of 3.0-3.2 eV which makes them transparent for visible light.6 Many of the above-mentioned applications require a molecular layer of a functional compound on the inner walls of the mesoporous system. A key step in the surface modification of ∼20 nm sized TiO2 particles is creating a strong and stable binding of the functional moiety to the surface via an anchor group. Two of the most commonly used anchor groups are carboxylic and phosphonic acid functionalities, with the phosphonate showing a stronger and selective coordination.7-9 This is attributed to the low pKa value which allows them to participate in acid-base *Corresponding author. E-mail: [email protected]. (1) Meier, K. R.; Gratzel, M. ChemPhysChem 2002, 3, 371–374. (2) Pechy, P.; Rotzinger, F. P.; Nazeeruddin, M. K.; Kohle, O.; Zakeeruddin, S. M.; Humphrybaker, R.; Gratzel, M. J. Chem. Soc., Chem.Commun. 1995, 1093– 1093. (3) Bonhote, P.; Moser, J. E.; Humphry-Baker, R.; Vlachopoulos, N.; Zakeeruddin, S. M.; Walder, L.; Gratzel, M. J. Am. Chem. Soc. 1999, 121, 1324–1336. (4) Moller, M. T.; Asaftei, S.; Corr, D.; Ryan, M.; Walder, L. Adv. Mater. 2004, 16, 1558–1561. (5) Lancelle-Beltran, E.; Prene, P.; Boscher, C.; Belleville, P.; Buvat, P.; Lambert, S.; Guillet, F.; Boissiere, C.; Grosso, D.; Sanchez, C. Chem. Mater. 2006, 18, 6152–6156. (6) Diebold, U. Surf. Sci. Rep. 2003, 48, 53–229. (7) Galoppini, E. Coord. Chem. Rev. 2004, 248, 1283–1297. (8) Park, H.; Bae, E.; Lee, J. J.; Park, J.; Choi, W. J. Phys. Chem. B 2006, 110, 8740–8749. (9) Pawsey, S.; Yach, K.; Reven, L. Langmuir 2002, 18, 5205–5212. (10) Gawalt, E. S.; Avaltroni, M. J.; Koch, N.; Schwartz, J. Langmuir 2001, 17, 5736–5738.

Langmuir 2009, 25(9), 5371–5379

interaction with the surface hydroxyl groups.10,11 Besides TiO2, phosphonic acids have been used to modify mesoporous metal oxides such as ZrO2 and Al2O3.12,13 They form disordered monolayers if the chain length is n < 16.14 Mono-, bi-, or tridentate binding modes have been proposed for phosphonates on TiO2, and these are supported by a variety of characterization techniques such as IR,15 X-ray photoelectron spectroscopy (XPS),16 solid state magic-angle spinning (MAS) NMR,17,18 and atomic force microscopy (AFM).19 In order to check a new functional compound (sensitizer, electrochromophore, redoxshuttle) as a candidate for its efficiency, it has to be equipped with a phosphonate anchoring group, and this may be a complicated synthetic task. Bifunctional molecular linkers have been widely used to modify metal and metal oxide surfaces to impart peculiar properties to surfaces and to extend the coordination properties.20,21 Recently, increased flexibility for coordinating a functional moiety on mesoporous TiO2 has been introduced using bridging anchors (11) Pellerite, M. J.; Dunbar, T. D.; Boardman, L. D.; Wood, E. J. J. Phys. Chem. B 2003, 107, 11726–11736. (12) Marcinko, S.; Fadeev, A. Y. Langmuir 2004, 20, 2270–2273. (13) Shyue, J. J.; Tang, Y.; De Guire, M. R. J. Mater. Chem. 2005, 15, 323–330. (14) Gao, W.; Dickinson, L.; Grozinger, C.; Morin, F. G.; Reven, L. Langmuir 1997, 13, 115–118. (15) Mann, J. R.; Watson, D. F. Langmuir 2007, 23, 10924–10928. (16) Koh, S. E.; McDonald, K. D.; Holt, D. H.; Dulcey, C. S.; Chaney, J. A.; Pehrsson, P. E. Langmuir 2006, 22, 6249–6255. (17) Pawsey, S.; McCormick, M.; De Paul, S.; Graf, R.; Lee, Y. S.; Reven, L.; Spiess Hans, W. J. Am. Chem. Soc. 2003, 125, 4174–84. (18) Holland, G. P.; Sharma, R.; Agola, J. O.; Amin, S.; Solomon, V. C.; Singh, P.; Buttry, D. A.; Yarger, J. L. Chem. Mater. 2007, 19, 2519–2526. (19) Demidenok, K.; Bocharova, V.; Stamm, M.; Jahne, E.; Adler, H. J. P.; Kiriy, A. Langmuir 2007, 23, 9287–9292. (20) Wanunu, M.; Livne, S.; Vaskevich, A.; Rubinstein, I. Langmuir 2006, 22, 2130–2135. (21) Colvin, V. L.; Goldstein, A. N.; Alivisatos, A. P. J. Am. Chem. Soc. 1992, 114, 5221–30.

Published on Web 3/31/2009

DOI: 10.1021/la8038126

5371

Article

Taffa et al.

Scheme 1. (left) Structure of the Phosphonic Acids and (right) Alkyl Viologen with Variable Carbon Chain n = 6, 10, and 14 Used in This Study

Scheme 2. Synthesis of Alkyl Phosphonic Acids

with orthogonal coordination properties, such as thiolated alkyl phosphonates for CdSe on TiO2, or carboxylated sensitizers with hydrophobic alkyl chains for hole conductors on TiO2.15,22,23 The main aim of this work is to extend this idea, that is, to introduce a flexible approach that allows one to immobilize many functional compounds characterized by their charge and hydrophobicity on the TiO2 surface using a limited set of orthogonal organic anchors. For this purpose, terminally charged phosphonic acids with variable chain length (n = 6, 10, and 14) were synthesized and grafted to mesoporous TiO2 thin films. In Scheme 1, we present the compounds that are used for the modification. Depending on their carbon chain length, they occupy a considerable space of the average pore volume (ca. 5-10%). Their head groups (R) determine the “inner-wall” surface properties and potential interactions with species in solution. In this study, we investigate the influence of the pore wall modifiers on the accumulation of electroactive guests on or in the membranes using electrochemical and IR techniques. Our results indicate that (i) for methyl terminated alkyl phosphonates the aqueous electrolyte does not enter the porous system, (22) Dibbell, R. S.; Soja, G. R.; Hoth, R. M.; Watson, D. F. Langmuir 2007, 23, 3432–3439. (23) Schmidt-Mende, L.; Zakeeruddin, S. M.; Gratzel, M. Appl. Phys. Lett. 2005, 86, 013504/1–013504/3.

5372

DOI: 10.1021/la8038126

(ii) for alkyl phosphonates with a charged headgroup a stable noncovalent anchoring can be achieved depending on the electrostatic and hydrophobic interaction, (iii) such systems can be used for preconcentrating an analyte on the sensing layer, and (iv) efficient electron transfer is possible by lateral electron hoping (wiring) as well as by electron injection into TiO2.

Experimental Section Materials and Reagents. Potassium ferrocyanide (Merck), potassium ferricyanide, and 1-ferrocenyl ethanol (Fluka) were analytical grade and used as received. n-Hexylphosphonic acid (Alfa), 1,6-dibromohexane, 1,10-dibromodecane, 1-bromohexane, 1-bromodecane, 1-bromotetradecane, 4,40 -bipyridine, pyridine, sodium sulfite, triethylphosphite, HBr (Sigma-Aldrich), methyl iodide (Merck), 1,14-tetradecanedioic acid (ABCR Chemicals), and silica gel (0.063-0.200 mm, Baker) were analytical grade and used as received. Spectroscopic grade CH3CN used for the reactions and tetrahydrofuran, pyridine, and dichloromethane were distilled using standard procedures before use. Aqueous solutions were prepared with deionized water. Synthesis. (a) Phosphonic Acid (Surface Modifiers). All alkane phosphonic acids with terminal charged groups (Scheme 2) were prepared in our laboratory: 6-phosphonohexane-1-pyridinium bromide (Pho-C6-R+Br-), 10-phosphonodecane-1-pyridinium bromide (Pho-C10-R+Br-), 14phosphonotetradecane-1-pyridinium bromide (Pho-C14-R+Br-), Langmuir 2009, 25(9), 5371–5379

Taffa et al.

Article Scheme 3. Synthesis of Mono- And Dialkyl Viologens

Table 1. Surface Coverage of the Different Phosphonic Acids on Mesoporous TiO2 Thin Filmsa charge

phosphonic acids

Γmesoporous/IR (10-7 mol cm-2)

Γcorr/IR (10-10 mol cm-2)

Γcorr/QCM (10-10 mol cm-2)

neutral

Pho-C14-H

2.7

6.6

6

positively charged

Pho-C6-R+ Pho-C10-R+ Pho-C14-R+

0.7 1.1 1.1

1.8 2.8 2.8

2.5 3 2.7

0.3 0.8 0.7 Pho-C6-R0.6 1.5 0.8 Pho-C10-R0.7 1.8 2.5 Pho-C14-R a Γmesoporous: surface concentration on 4 μm thick TiO2 film (projected area). Γcorr = Γmesoporous/400 (400 = roughness factor).

negatively charged

1-(6-phosphonohexyl) sulfonic acid (Pho-C6-R-), 1-(10-phosphonodecyl) sulfonic acid (Pho-C10-R-), and 1-(14-phosphonotetradecyl) sulfonic acid (Pho-C14-R-). Additionally, methyl terminated phosphonic acids were prepared: decylphosphonic acid (Pho-C10-H) and tetradecylphosphonic acid (Pho-C14-H). (b) Mono- and Dialkyl Viologens (Guest Molecules). 1-Hexyl-10 -methyl-4,40 -bipyridinium dibromide (C1-V2+C6 2Br-), 1-decyl-10 -methyl-4,40 -bipyridinium dibromide (C1-V++-C10 2Br-), 1-tetradecyl-10 -methyl-4,40 -bipyridinium dibromide (C1-V++-C14 2Br-), 1,10 -dihexyl-4,40 -bipyridinium dibromide (C6-V++-C6 2Br-), 1,10 -didecyl-4,40 -bipyridinium dibromide (C10-V++-C10 2Br-), and 1,10 -ditetradecyl4,40 -bipyridinium dibromide (C14-V++-C14 2Br-) were prepared according to Scheme 3. For detailed synthetic procedures and 1H NMR data of all new compounds, see the Supporting Information. Preparation of Mesoporous TiO2 Thin Films. Mesoporous TiO2 thin films were prepared on TEC glass (F-doped SnO2 (FTO), 2.2 mm, 15 Ω cm-2) (LOF) or glass (Omni Laboratory) substrate as follows. The substrates were sequentially cleaned in 1% Mucasol solution, distilled water, 0.1 M HCl, distilled water, and acetone, each for 10 min in an ultrasonic bath. They were then heated at 450 °C for 10 min using a hot air gun. The titanium dioxide paste (particle size 14.5-21.6 nm, surface area 103 m2 g-1, NTERA, Dublin, Ireland) was applied to the substrates by the doctor blade method and masked with scotch tape and air-dried for 15 min. After removing the mask, the sample was heated with a hot air gun or in an oven for 30 min at 450 °C. The resulting TiO2 films have a thickness of ∼4-5 μm in accordance to the supplier information. Previous reports Langmuir 2009, 25(9), 5371–5379

showed that films prepared analogously have an average pore diameter of 10-15 nm.24,25 Adsorption of Alkyl Phosphonic Acid on TiO2 Thin Films. The as prepared TiO2 films were immersed in a freshly prepared 1 mM ethanol or methanol solution of the phosphonic acids overnight, washed with the same solvent, and dried at 80 °C for 15 min before further use. IR Measurements. IR spectra were measured with a VERTEX-70 (BRUKER) infrared spectrometer with a resolution of 4 cm-1; 16 scans were averaged. TiO2 covered glass substrates (Omni Laboratory) which were transparent in the region of 2500-3100 cm-1 were used for the preparation of the modified films. Sub-millimolar solutions of the acids were prepared in carbon tetrachloride (spectroscopic grade) for the solution phase IR measurement. The extinction coefficient of C-H absorption bands for the charged phosphonic acid was approximated by the ε value of the noncharged counterpart with the same C number. We found the extinction coefficients of hexylphosphonic acid (Pho-C6-H), decylphosphonic acid (Pho-C10-H), and tetradecylphosphonic acid (Pho-C14-H) to be 217, 449, and 630 M-1 cm-1, respectively, at the νa(CH2) band maxima. These values are in good agreement with previously reported values and are used to calculate the surface concentration (see the Supporting Information).26 (24) Gratzel, M. Prog. Photovoltaics 2000, 8, 171–185. (25) Kavan, L.; Rathousky, J.; Gratzel, M.; Shklover, V.; Zukal, A. Microporous Mesoporous Mater. 2001, 44-45, 653–659. (26) Hastings, S. H.; Watson, A. T.; Williams, R. B.; Anderson, J. A. Anal. Chem. 1952, 24, 612–618.

DOI: 10.1021/la8038126

5373

Article

Taffa et al.

Figure 1. IR spectra of modified mesporous TiO2 films with different phosphonic acid bearing charged head groups of pyridinium (left) and sulfonate (right) showing -CH2 stretch between 2920 and 2940 cm-1 and aromatic -CH between 3050 and 3600 cm-1.

Figure 2. Single scan voltammograms of 0.1 mM K4[Fe(CN)6] at a bare TiO2 electrode (black dotted, a) and at Pho-C14-H (black solid, c), Pho-C14-R+(red, d), and Pho-C14-R- (blue line, b) modified electrodes in 0.1 M KCl at v = 50 mV s-1 and corresponding models (a-d) showing homogeneous (a, b, d) and heterogeneous (c) wetting regimes, as well as repulsion and accumulation of [Fe(CN)6]4-. Quartz Crystal Microbalance (QCM) Measurements. The loading of the phosphonic acid on TiO2 was determined by QCM measurements according to a modified procedure.27 The 6 MHz, AT-cut quartz crystal was cleaned with acetone and coated on one side with TiO2 and then subjected to a temperature of 450 °C for 30 min. Prior to modification with the acids, the fundamental frequency (f0) of each crystal before and after coating with TiO2 was measured when a stable frequency reading was obtained under Ar. The quartz electrodes were mounted in a microbalance holder type 225 with an integrated circuit (Institute of Physical Chemistry, Polish Academy of Sciences, Warsaw, Poland). The resonance frequency was monitored with a PeakTech 2060 universal frequency counter. The quartz was then treated in a 1 mM solution of phosphonic acids in ethanol or methanol overnight at room temperature, washed with the respective solvents, and dried overnight at 80 °C. The new frequency of the modified quartz was measured after the frequency was stable for 30 min. The difference in frequency of the modified and unmodified quartz crystals was then used to calculate the amount of phosphonic acid adsorbed on the TiO2 surface (see the Supporting Information).28,29 Electrochemical Measurements. Electrochemical experiments were carried out with an Autolab Potentiostat PGSTAT 20 interfaced with a personal computer running under GPES for Windows, version 4.9 (ECO Chemie, Netherlands). A standard three electrode electrochemical cell with a Pt wire (27) Schon, P.; Michalek, R.; Walder, L. Anal. Chem. 1999, 71, 3305–3310. (28) Sauerbrey, G. Z. Phys. 1959, 155, 206–222. (29) Miller, J. B.; Schwartz, J. Inorg. Chem. 1990, 29, 4579–4581.

5374

DOI: 10.1021/la8038126

counter electrode and a Ag/AgCl reference electrode separated by a salt bridge were used. The TiO2 modified electrode with a geometric area of 0.78 cm2 served as working electrode and electrical contact was made by a Ti holder. The electrolyte was 0.1 M KCl. It was bubbled with Ar at least for 5 min and Ar was blown over the electrolyte during the measurement.

Results and Discussion Characterization of the Modified TiO2 Surface. Monolayers of phosphonic acid were adsorbed on the inner walls of mesoporous TiO2 and characterized by Fourier transform infrared (FT-IR) and QCM measurements (see the Experimental Section and Supporting Information for details). The surface coverage was calculated from IR by using a modified form of the Beer-Lambert equation: R A ð1Þ Γ ¼ εint  1000 where the Γ Ris the surface coverage on the projected area (mol cm-2), A is the integrated absorbance in the C-H stretching region (2800-3000 cm-1), and εint is the integrated absorption coefficient (M-1 cm-2) from 2800 to 3000 cm-1. The integrated absorption coefficients for Pho-C6-H, PhoC10-H, and Pho-C14-H were 1.37  104, 2.3  104, and 3.45  104 M-1 cm-2, respectively. TiO2 on glass substrate used as a Langmuir 2009, 25(9), 5371–5379

Taffa et al.

Article

Figure 3. CVs of 0.5 mM Fc-EtOH at a bare TiO2 electrode (black dashed line) and at Pho-C14-R- (blue, a) and Pho-C14-R+ (red, b) TiO2 modified electrodes in 0.1 M KCl at v = 50 mV s-1. Interaction model: TiO2/Pho-C14-R- and Fc-EtOH (a) and TiO2/Pho-C14-R+ and Fc-EtOH (b).

Figure 4. CV of monoalkyl viologen modified TiO2 electrodes TiO2/Pho-C14-R-/C1-V+2-C6 (cyan, a), TiO2/Pho-C14-R-/C1-V+2-

C10 (orange, b), and TiO2/Pho-C14-R-/C1-V+2-C14 (green, c) in 0.1 M KCl at v = 50 mV s-1. Interaction model: (a) TiO2/Pho-C14-Rvs C1-V+2-C6, (b) TiO2/Pho-C14-R- vs C1-V+2-C10, and (c) TiO2/Pho-C14-R- vs C1-V+2-C14.

support for the modifier, and the analysis is limited to the spectral range 2500-3100 cm-1, that is, only C-H stretching modes are considered. We observed the CH2 symmetric stretching band (2850-2865 cm-1), CH2 asymmetric stretching band (2920-2940 cm-1), and aromatic C-H stretching of a pyridinium group (3050-3060 cm-1) (see the Supporting Information). Table 1 summarizes the surface coverage of the different phosphonic acids on the mesoporous TiO2. The quality of the QCM results is definitely lower than that of the IR measurements, but they show a reasonable correlation with the IR results. The surface concentration of the neutral phosphonic acid is 2.7  10-7 mol cm-2 on the mesoporous film, which gives 7  10-10 mol cm-2 after correction for the roughness factor (400) in agreement with literature values (Table 1).15,30 Upon introduction of the charged head groups, the surface coverage drops by a factor of 2-3 as compared to the methyl terminated alkyl (30) Sotomayor, J.; Hoyle, R. W.; Will, G.; Fitzmaurice, D. J. Mater. Chem. 1998, 8, 105–110.

Langmuir 2009, 25(9), 5371–5379

phosphonic acid. This is probably due to the repulsive electrostatic interaction and the weakening of hydrophobic interactions as carbon chain length decreases. The organization of the monolayers (crystallinity and disorder) can be judged from the frequency of the symmetric and asymmetric C-H stretching bands (Figure 1). For all modified TiO2 surfaces, the C-H asymmetric stretching band appears at higher frequency than the reference 2918 cm-1, which indicates that the monolayers are disordered.31,32 The disorder is more pronounced for shorter chain phosphonic acids. In addition, the phosphonate groups have a large surface area (0.25 nm2) which hampers close packing of the alkyl chain.14 The IR measurements ensure the incorporation of the phosphonic acids into the inner walls of the mesoporous TiO2 thin films. In the following section, we demonstrate the

(31) Quinones, R.; Gawalt, E. S. Langmuir 2007, 23, 10123–10130. (32) Raman, A.; Dubey, M.; Gouzman, I.; Gawalt, E. S. Langmuir 2006, 22, 6469–6472.

DOI: 10.1021/la8038126

5375

Article

Taffa et al.

Figure 5. CV of dialkyl viologen modified TiO2 electrodes TiO2/Pho-C14-R-/C6-V+2-C6 (cyan, a), TiO2/Pho-C14-R-/C10-V+2-C10

(orange, b), and TiO2/Pho-C14-R-/C14-V+2-C14 (green, c) in 0.1 M KCl at v = 50 mV s-1. Interaction model: (a) TiO2/Pho-C14-R- vs C6-V+2-C6, (b) TiO2/Pho-C14-R- vs C10-V+2-C10, and (c) TiO2/Pho-C14-R- vs C14-V+2-C14.

incorporation of neutral or charged molecular guests on or into the membranes. Pure Electrostatic Interaction. Similar surface concentrations were achieved for phosphonic acids with carbon chain lengths 10 and 14; however, the latter was expected to be more stable due to high hydrophobic interaction between carbon chains. Thus, we used these modifiers for further electrochemical studies. All TiO2 electrodes modified with phosphonic acid of chain n = 14 exhibit stability in water toward hydrolysis over days as long as the pH is kept below 8.12 Figure 2 shows the current response of [Fe(CN)6]4- in solution at bare and modified TiO2 surfaces. The redox couple [Fe(CN)6 ]4- shows a quasi reversible wave with E° = 0.305 V, with a peak to peak separation of 104 mV, and with a peak current density of ∼40 μA cm-2 at the unmodified TiO2 electrode (Figure 2a). It is worthwhile to mention the fact that the redox potential is observed in a range in which the TiO2 electrode behaves as an insulator rather than as a metal.33,34 The current density varies with the pH (increasing with lower pH, decreasing with higher pH (not shown in Figure 2)), but it is generally in the range expected for a flat gold electrode. This indicates that the pores are open for diffusing [Fe(CN)6]4- ions, that heterogeneous charge transfer occurs at the FTO solution interface or at defect areas on TiO2 exhibiting unexpected metallic behavior, and that there are only minor interactions with the unmodified TiO2 walls. Electrodes modified with the methyl terminated phosphonic acid (Pho-C14-H) show no current for [Fe(CN)6]4-. Obviously, water cannot enter into the mesoporous system. In fact, we observed contact angles in the range of 130° for water droplets on the Pho-C14-H modified TiO2 plate (see the Supporting Information) (Figure 2c). Such phenomena have been reported for mesoporous systems in general and TiO2 in particular.35-38 If the pores are modified with alkyl chains carrying either a negatively or positively charged head group (33) Yan, S. G.; Hupp, J. T. J. Phys. Chem. 1996, 100, 6867–70. (34) Boschloo, G.; Fitzmaurice, D. J. Electrochem. Soc. 2000, 147, 1117–1123. (35) Otal, E. H.; Angelome, P. C.; Bilmes, S. A.; Soler-Illia, G. Adv. Mater. 2006, 18, 934–938. (36) Zhang, X.; Kono, H.; Liu, Z.; Nishimoto, S.; Tryk, D. A.; Murakami, T.; Sakai, H.; Abe, M.; Fujishima, A. Chem. Commun. 2007, 4949–4951. (37) Feng, X.; Zhai, J.; Jiang, L. Angew. Chem., Int. Ed. 2005, 44, 5115–5118. (38) Sun, W.; Zhou, S.; Chen, P.; Peng, L. Chem. Commun. 2008, 603–605.

5376

DOI: 10.1021/la8038126

Table 2. Surface Concentration of Different Alkyl Viologens on Mesoporous Pho-C14-R Modified Thin Film TiO2 Electrodesa viologen

Γ (10-8 mol cm -2)

Γcorr (10-10 mol cm-2)

C1-V++-C6 C1-V++-C10 C1-V++-C14 C6-V++-C6 C10-V++-C10 C14-V++-C14 Pho-C-V++-C1 a Γcorr = Γ/400.

0.37 2.9 2.7 1.1 2.2 1.5 7.5

0.1 0.8 0.7 0.3 0.6 0.4 1.9

(Figure 2 b/d)), the pore walls are wetted and the [Fe(CN)6]4ions are allowed to enter the pores as in the case of unmodified TiO2 (Figure 2a). In case of the negatively charged head groups (Figure 2d), the current density is reduced (ca. 8 μA cm-2) as compared to the plain TiO2. Obviously, the electrostatic repulsion reduces the concentration of [Fe(CN)6]4- ions in the channels and/or on the wall surface. TiO2 electrodes modified with Pho-C14-R+, with its pH-independent positively charged head group, result in a large enhancement of the faradic current (ca. 240 μA cm-2). It is explained by the electrostatic attractive interaction between the positively charged inner walls and [Fe (CN)6]4- ions in solution, leading to an increased concentration of [Fe(CN)6]4- ions in the channels and/or on the inner walls. The preconcentration process can be monitored; that is, the peak currents in the cyclic voltammogram grow steadily under forced convection over ∼10 min to reach a maximum of 240 μA cm-2 (inset Figure 2). The standard potential of [Fe(CN)6]4-/3- on unmodified, Pho-C14-R+ and Pho-C14-R- modified TiO2 electrodes were 0.304, 0.174, and 0.255 V, respectively. The potentials are negatively shifted compared to the unmodified electrode. These shifts are attributed to a combination of Donnan potential and preferential stabilization of one of the oxidation states over the other. 39,40

(39) Redepenning, J.; Tunison, H. M.; Finklea, H. O. Langmuir 1993, 9, 1404– 1407. (40) Vengatajalabathy Gobi, K.; Ohsaka, T. J. Electroanal. Chem. 1999, 465, 177–186.

Langmuir 2009, 25(9), 5371–5379

Taffa et al.

Article

Figure 6. CV of 0.5 mM C10-V+2-C10 at TiO2 electrode modified with Pho- C6-R- (cyan, a), Pho- C10-R- (orange, b), and Pho- C14R- (green, c) in 0.1 M KCl at v = 50 mV s-1. Interaction model: (a) TiO2/Pho-C6-R- vs C10-V+2-C10, (b) TiO2/Pho-C10-R- vs C10V+2-C10, and (c) TiO2/Pho-C14-R- vs C10-V+2-C10.

Even at bare mesoporous TiO2 electrodes, the electrochemistry of [Fe (CN)6]4-/3- is controversially discussed in the literature with respect to its standard potential, the rate of electron transfer, and the site of heterogeneous electron transfer considering the “isolating state” of the TiO2 in the positive potential range, and our contribution cannot solve this debate.41,42 However, the shape of the cyclic voltammogram (CV) and the maximum current observed using our membrane modified mesoporous TiO2 points to a combination of diffusion, thin layer behavior, and surface confinement of electroactive species that determine the shape of the CV and its dependence on scan rate.43 In order to distinguish the current (or charge) contribution from the mesoporous walls and from the channels of the mesoporous system, a Pho-C14-R+ modified electrode was exposed to K4[Fe(CN)6] (c = 100 μM, i.e., where saturation is achieved) and then studied by cyclic voltammetry in pure electrolyte. A current density of ∼180 μA cm-2 was observed, indicating that a large portion of the current is related to channel wall confined [Fe(CN)6]4-. The preconcentration phenomenon described above has been applied to trace analysis of K4[Fe(CN)6] on the Pho-C14-R+ modified TiO2 electrode. Complete uptake of the [Fe(CN)6]4ion was achieved within 10 min under convective conditions. The electrodes were then checked in pure aqueous electrolyte assuring that all currents stem from pore wall confined [Fe(CN)6]4-. A saturation of the [Fe(CN)6]4- signal corresponding to Γ = 1.6  10-8 mol cm-2 was reached at 100 μM solution concentration. This value is higher than those reported for functionalized mesoporous silica films.44 Compared to ΓPhoC14-R+(Table 1) and assuming full charge compensation by the ferrocyanide ions, the expected surface concentration is ∼2.7  10-8 mol cm-2, and hence, the experimental value indicates 60% of the theoretical coverage. Using traditional cyclic voltammetry, the detection limit of [Fe(CN)6]4- is ∼200 nM (see the Supporting Information). Similar results were obtained for other electroactive complex ions such as [IrCl6]2-. (41) Kubota, L. T.; Gushikem, Y. Electrochim. Acta 1992, 37, 2477–2480. (42) Etienne, M.; Grosso, D.; Boissiere, C.; Sanchez, C.; Walcarius, A. Chem. Commun. 2005, 4566–4568. (43) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd ed.; J. Wiley: New York, 2001; Chapter 14. (44) Fattakhova-Rohlfing, D.; Rathousky, J.; Rohlfing, Y.; Bartels, O.; Wark, M. Langmuir 2005, 21, 11320–11329.

Langmuir 2009, 25(9), 5371–5379

Figure 7. IR spectra of TiO2/Pho-C14-R- before (black) and

after the incorporation of C1-V++-C10 (red) and C10-V++C10 (blue), and the respective interaction models.

Hydrophobic Interaction Combined with Minor Electrostatic Interactions. In order to check the influence of the hydrophobic phase of our membranes, we used the neutral redox marker 1-ferrocene ethanol, Fc-EtOH.45 As mentioned above, the Pho-C14-H modifier blocks water from entering the mesopores and therefore no electrochemistry is observed, even if the Fc-EtOH would be incorporated in the membrane. Notably, in other systems, entrance of water into alkyl-chain modified (tailored) mesoporous TiO2 was found, but this may be related to differences in the TiO2 architecture and/or of the modifiers structure.35 With Fc-EtOH in solution (c = 0.5 mM), PhoC14-R+ and Pho-C14-R- modified TiO2 electrodes both show higher currents than the untreated TiO2 (the anodic peak current increases from 29 μA on unmodified TiO2 to 93 and 110 μA on TiO2-Pho-C14-R- and TiO2-Pho-C14-R+, respectively, Figure 3). This is definitely related to the partitioning of the uncharged redox probe in the hydrophobic part of the monolayer. In agreement with this hypothesis, electrodes exposed to a 0.5 mM solution of Fc-EtOH for 10 min and then transferred to pure aqueous electrolyte solution show (45) Takehara, K.; Takemura, H.; Ide, Y. Electrochim. Acta 1994, 39, 817–822.

DOI: 10.1021/la8038126

5377

Article

Taffa et al. Table 3. Absorbance Increase after Intercalation of Viologens into Membrane Modified TiO2a modified TiO2 films

R

A(1)

R

A(2)

R R Rexpt ( A(2)/ A(1))

Rcorr

x

-

1.784 TiO2/Pho-C14-R 2.817 0.213 0.075 0.28 1.4 TiO2/Pho-C14-R-/C1-V++-C10 ++ 3.320 0.154 0.046 0.17 2 TiO2/Pho-C14-R /C10-V -C10 0.591 0.211 0.357 3.73 1 TiO2/Pho-C6-V++-C1 R R a A(1): integrated absorbance from 2800 to 3000 cm-1. A(2): integrated absorbance from 3000 to 3100 cm-1. x: number of membrane chains per incorporated viologen (eq 2).

electrochemical activity (not shown). Upon oxidation, Fc-EtOH becomes positively charged and is therefore expelled from the membrane with positive head groups resulting in an asymmetric CV wave (Figure 3, curve b); while the negatively charged membrane does not show this behavior (Figure 3, curve a) and equilibration of the electrode at 0.6 V for 5 min does not affect the shape of the CV, indicating that for the ferrocene couple the hydrophobic interaction is more important than the 1:1 charge interaction. The electrostatic stabilization/destabilization of the oxidized species at TiO2/Pho-C14-R- and TiO2/Pho-C14-R+ can also be read from the E° potential difference which the couple exhibits on the two membranes, E-° = 0.227 V and E+° = 0.257 V, respectively, indicating that the oxidized species is slightly stabilized by the sulfonated membrane as compared to the pyridinium terminated membrane. Modulated Hydrophobic Interaction Combined with Substantial Electrostatic Interactions. In this section, we study the modulated hydrophobic interactions combined with constant electrostatic contributions. Previously partitioning of long chain alkyl viologens into disorganized monolayers of alkyl silane on ITO due to hydrophobic interaction was reported by Markovich and Mandler.46 The cyclic voltammogram of the dimethyl viologen on a Pho-C14-R- modified TiO2 electrode shows no significant difference from that on an unmodified electrode (CV not shown). Hence, we prepared viologens with different alkyl substituents (Scheme 1). They exhibit the same charge, but their hydrophobicity is controlled by the number of chains and the chain length. Notably, the chains (legs) have length comparable to those of the TiO2 modifiers, and according to our model they allow intercalation and efficient van der Waals contact. This interaction combined with the electrostatic attraction between the bipyridinium moiety and the modifier sulfonate head groups could lead to a strong partitioning of the guest in the membrane, mimicking biological systems. Figure 4 demonstrates the influence of the chain length of monolegged viologens on the partitioning of these guests into the membrane electrode TiO2-Pho-C14-R-. The alkyl viologens were adsorbed on the modified electrode from 0.5 mM solution in methanol for 10 min under convective conditions. They were then transferred to the aqueous electrolyte solution for CV measurements (Figure 4). It is obvious that the faradic current (and the surface concentration) associated with the first reduction of the viologen to its radical cation increases as the number of carbons on the alkyl chain increases from 6 to 10. However, it decreases slightly when the chain length is increased from 10 to 14. Thus, additional hydrophobic interaction does not bring higher surface concentration, possibly because of a bad fit of the leg in the membrane (Figure 4c). The scan was not extended over the second wave for stability reasons concerning the monolayer (reductive desorption) and loss of electrostatic interaction (neutral oxidation state).47 In Figure 5, the cyclic (46) Markovich, I.; Mandler, D. Analyst 2001, 126, 1850–1856. (47) Bird, C. L.; Kuhn, A. T. Chem. Soc. Rev. 1981, 10, 49–82.

5378

DOI: 10.1021/la8038126

voltammograms obtained with the two-legged viologens in the same membrane are represented. In comparison with the monolegged viologen, the two-legged viologens have lower saturation packing densities, because (i) the double-legged viologen occupies more space within the membrane to intercalate its legs in the free space available between the chains of the modifier, and (ii) the bipyridinium group is forced to sit flat on the modifier head groups and therefore it occupies a larger projected area. This is further illustrated in their surface concentration calculated from the integrated currents in the voltammograms (Table 2). A similar behavior was earlier reported from our group for viologens linked to TiO2 with two phosphonate anchoring groups resulting in lower surface concentration as compared to viologens with a single phosphonic acid anchoring group.48,49 In the current study, the surface coverage of the viologens was compared to a directly attached viologen on the TiO2 film through the phosphonate linker, that is, 1-methyl-10 -(6-phosphonohexyl)-4,40 -bipyridinium dibromide (Pho-C6-V+2-C1 2Br-). The surface coverages of the intercalated violgens Γ(C1V+2-C10) and Γ(C10-V+2-C10) are 2.5-3.5 times lower than the surface concentration of the directly attached viologens, or viologens on a gold surface.50 This is expected because of the relatively small space available in the presence of the membrane (Table 2). Alternatively, we studied the influence of the modifier chain length on the surface coverage of the double-legged C10-V+2C10. The short chain modifier (n = 6 and 10) does not offer enough space for efficient hydrophobic interactions and shows consequently a reduced peak current in the CV (Figure 6a and b). The maximum interaction and therefore the highest peak current is reached for the C10-V+2-C10/TiO2-PhoC14-R- interaction. The above results demonstrate clearly the importance of the match or mismatch situation with respect to the electrostatic and hydrophobic interactions. It is obvious that (a) the closest approach between the sulfonate head group and the positive charge(s) on the viologen and (b) the hydrophobic interaction between the membrane chains and the viologens legs are of crucial importance. The schematic interaction models in Figures 4-6 are simplified graphical representations which reflect the hydrophobic interaction and the resulting surface concentration. The number of viologens drawn per membrane chain and the conformation of the alkyl chains represent the trends only. The intercalation of the viologen into the monolayers can also be monitored by a comparison of the IR spectra of the modified TiO2 before and after exposure to the viologen solutions. We observe the appearance of the -CH stretching band (3000-3100 cm-1) related to the aromatic -CH vibration (48) Felderhoff, M.; Heinen, S.; Mulisho, N.; Webersinn, S.; Walder, L. Helv. Chim. Acta 2000, 83, 181–192. (49) Vlachopoulos, N.; Nissfolk, J.; Moeller, M.; Briancon, A.; Corr, D.; Grave, C.; Leyland, N.; Mesmer, R.; Pichot, F.; Ryan, M.; Boschloo, G.; Hagfeldt, A. Electrochim. Acta 2008, 53, 4065–4071. (50) Delong, H. C.; Buttry, D. A. Langmuir 1992, 8, 2491–2496.

Langmuir 2009, 25(9), 5371–5379

Taffa et al.

Article

of the bipyridinium group.51 Selected spectra are shown in Figure 7. In addition, an increase of the absorption band related to -CH2 stretching (legs + membrane) is observed (2850-2940 cm-1) upon loading the membrane with viologens (Table 3). Using the calibrated extinction coefficients, we were able to quantify the stoichiometric ratio of the membrane chains as compared to the viologen chains. For this purpose, we used Pho-C6-V++-C1 as a standard with a fixed ratio of aromatic -CH to aliphatic -CH2 groups (8:6), and we calculated the number of chains of the membrane (x) per incorporated viologen using eq 2. 8 ð2Þ Rcorr ¼ 13x þ nCH2 where Rcorr is the corrected integrated absorbance ratio, x is the number of membrane chains per viologen, and nCH2 is the number of CH2 groups attached to the viologen. Alternatively, the same information can be obtained from the increase in integrated absorbance in the region 2800-3000 cm-1 after the R R intercalation of the viologen ( A(m) f A(m + vio)), eq 3, R

R

AðmÞ 13y ¼ Aðm þ vioÞ 13y þ znCH2

ð3Þ

where y is the number of chains in the membrane and z is the number of chains from the viologen. The results suggest that for each viologen with one alkyl leg incorporated in the membrane there are 1.4 chains from the membrane while for the viologen with two alkyl chains there are 2 membrane chains, that is, in accordance with our model of interaction (Figures 4-7). The absorbance difference before and after the incorporation of the viologen was also used to estimate the absolute surface concentration of the viologen applying by eq 1; taking the average integrated absorbance coefficients of Pho-C10-H and Pho-C14-H, 2.88  104 M cm-2, we found 2.7  10 -8 mol cm-2 and 3.6  10-8 mol cm-2 for C10-V++-C10 and C10-V++-C1, respectively. These values are 20-25% higher than the values obtained from CV measurements, which indicates that not all adsorbed viologens are electrochemically accessible. The above results demonstrate that high surface concentrations of a membrane-intercalating guest molecule can be achieved using combined electrostatic and hydrophobic binding modes. This approach could be further exploited to attach other electrochromic materials, electron-transfer catalysts, sensitizers, or biomolecules to TiO2, via combined electrostatic and hydrophobic interactions, opening new roads in the field of mesoporous TiO2 based electrochromic displays,4,52 solar cells, battery application, or biomimetic sensors. (51) Silverstein, R. M.; Webster, F. X.; Kiemie, D. Spectrometric Identification of Organic Compounds, 7th ed.; J. Wiley: Danvers, MA, 2002; Chapter 2. (52) Lieder, M.; Grzybkowski, W.; Schlaepfer, C. W. Electroanalysis 1998, 10, 486–491.

Langmuir 2009, 25(9), 5371–5379

Conclusion New classes of phosphonic acids with positive and negatively charged head groups and variable alkyl chain lengths ranging from 6-14 were synthesized, and are strongly anchored to the inner surfaces of mesoporous TiO2 films via phosphonate-TiO2 complexation. The surface coverage is stable over days, and Γ is in the range of (0.3 -1.1)  10 -7 mol cm-2 ((0.8 -2.8)  10 -10 mol cm-2 corrected for the TiO2 roughness factor), depending on the charge and the alkyl chain length. Thus, the molecules build up a membrane consisting of closely packed alkyl phosphonic acids. The head groups of the modifier molecules can tune the properties of the TiO2 surface from hydrophilic to superhydrophobic. The membrane modified pores can be used as a matrix to incorporate different guest molecules with variable charge and hydrophobic properties. It is shown that a balanced mix of electrostatic and hydrophobic interactions can lead to high loading of guest molecules at the outer membrane border or within the membrane. Stabilities over days can be achieved for matched membrane-guest interactions. The possible applications are obvious, as many interesting applications of mesoporous TiO2 require a functional compound on the TiO2 surface, for example, a sensitizer for photovoltaics or an electrochromophore for electrochromics. For the direct attachment, it is necessary to synthesize phosphonated functional compounds, whereas the new method presented here allows one to incorporate these molecules via electrostatic/hydrophobic interactions (i.e., the molecular requirement is less specific). Remarkably, the surface concentration of the guest compounds is only slightly reduced when compared to their direct attachments as shown for alkyl viologens (electrochromic materials) in this work. Other applications such as photoinduced electron injection into TiO2 by a sensitizer partitioning in the membrane or shuttling of electrons along the membrane surface via bound redox couples are definitely challenging task, in the view of our TiO2 modification technique. Finally, the mimicking of a biological membrane spanned over the inner walls of the TiO2 pores offers an opportunity to enhance membrane signals by the TiO2 roughness factor. Acknowledgment. D.H.T. thanks the federal state of Lower Saxony (Lichtenberg-Stipendium), and M.K. thanks the graduate college 612 funded by the DFG for financial support. The :: assistance of P. Pavel and R. Kerstin (University of Osnabruck) with the IR measurements is greatly acknowledged. Supporting Information Available: Details concerning IR measurements, QCM measurements, electrochemistry, as well as representative micrographs, synthesis of pyridinium alkyl phosphonic acids and phosphonoalkyl sulfonic acids, and synthesis of viologen derivatives. This material is available free of charge via the Internet at http://pubs.acs.org.

DOI: 10.1021/la8038126

5379