Effect of Surface Modification on Back Electron Transfer Dynamics of

Electron injection and back electron transfer (BET) dynamics have been carried out for dibromo fluorescein. (DBF) sensitized TiO2 nanoparticles capped...
0 downloads 8 Views 88KB Size
1430

Langmuir 2004, 20, 1430-1435

Effect of Surface Modification on Back Electron Transfer Dynamics of Dibromo Fluorescein Sensitized TiO2 Nanoparticles G. Ramakrishna,† Amit Das,‡ and Hirendra N. Ghosh*,† Radiation Chemistry and Chemical Dynamics Division, Solid State Physics Division, Bhabha Atomic Research Centre, Mumbai 400 085, India Received July 3, 2003. In Final Form: November 27, 2003 Electron injection and back electron transfer (BET) dynamics have been carried out for dibromo fluorescein (DBF) sensitized TiO2 nanoparticles capped (modified) with sodium dodecyl benzene sulfonate using transient absorption techniques in picosecond and microsecond time domain. BET dynamics have been compared with bare (unmodified) nanoparticles for the same DBF/TiO2 system. It has been observed that BET reaction is slow on the modified surface compared to a bare surface in earlier time domain (picosecond). This observation has been explained by the fact that on surface modification the energy levels of the semiconductor nanoparticles are pushed up in energy. As a result, the free energy of reaction (-∆G°) for BET reaction of a dye/SM-TiO2 system increases as compared to the dye/bare TiO2 system. High exoergic BET reaction in dye-sensitized TiO2 nanoparticles surfaces fall in the Marcus inverted regime, so with increasing free energy of reaction, BET rate decreases on the modified surface. However, a reversible trend in BET dynamics has been observed for the above systems in the longer time domain (microsecond). In microsecond time domain BET reaction is faster on the modified surface as compared to on the bare surface. Modification of this surface reduces the density of deep trap states. Recombination dynamics between deep-trapped electron and parent cation is slow due to low coupling strength of BET reaction. As the density of deep-trapped electrons is high in bare particles, BET reaction is slow in longer time domain.

Introduction Semiconductor nanoparticles have attracted significant interest in the past 15 years due to their potential applications in solar energy conversion, nonlinear optics, and heterogeneous photocatalysis.1 Surface modification of semiconductor nanoparticles can change their optical, chemical, and photocatalytic properties significantly.1c Surface modification of nanoparticles can lead to the following effects: (i) It may enhance their excitonic and defect emission by blocking nonradiative electron/hole (e-/ h+) recombination at the defect sites (traps) on the surface of the semiconductor nanoparticles.2 (ii) It may increase the photostability of semiconductor nanoparticles.2 (iii) It may create new traps on the surface of the nanoparticles leading to the appearance of new emission bands.3 (iv) It may boost the selectivity and efficiency of light-induced reactions occurring on the surface of semiconductor nanoparticles.1c,4 The performance of most semiconductor-containing devices is critically dependent on the electronic properties of the semiconductor surface band bending (Vs) and the surface recombination velocity (SRV).5 These properties in turn depend on the density and energy distribution of * Corresponding author. E-mail: [email protected]. † Radiation Chemistry and Chemical Dynamics Division. ‡ Solid State Physics Division. (1) . (a) Oregan, B.; Gratzel, M. Nature 1991, 353, 737. (b) Henglein, A. Chem. Rev. 1989, 89, 1861. (c) Bawendi, M. G.; Steigerwald, M. L.; Brus, L. E. Annu. Rev. Phys. Chem. 1990, 41, 477. (c) Kamat, P. V. Chem. Rev. 1993, 93, 267. (d) Alivisatos, A. P. J. Phys. Chem. 1996, 100, 13226. (e) Fox, M. A. Top. Curr. Chem. 1987, 72, 142. (2) Spanhel, L.; Haase, M.; Weller, H.; Henglein, A. J. Am. Chem. Soc. 1987, 109, 5649. (3) Spanhel, L.; Weller, H.; Fojtik, A.; Henglein. Ber. Bunsen-Ges. Phys. Chem. 1987, 91, 88. (4) Fox, M. A.; Dulay, M. T. Chem. Rev. 1993, 93, 341. (5) (a) Sze, S. M. Physics of Semiconductor Devices, 2nd ed.; John Wiley & Sons: New York, 1981. (b) Brillson, L. J. Contacts to Semiconductors; Noyes Publications: Park Ridge, NJ, 1993.

surface states. As the surface state properties are controlled by the chemistry of the surface, much effort has been devoted to modify the surface states by chemical treatments.6 The use of suitable organic or organometallic molecules as surface treatments holds great promise for fine tuning the desired surface electronic properties. Thus, a group that optimizes molecular binding to the surface can be augmented with auxiliary groups, can provide control over molecular dipole moments, frontier orbital energy levels, light sensitization properties, hydrophilic/ hydrophobic character, etc. Rajh et al.7 have reported a new route to improve the optical response of nanocrystalline TiO2 in the visible region. The approach involves direct electron transfer from ascorbate,7b mercaptocarboxylic,7c and enediol ligand7e modifier of TiO2 into the conduction band of nanocrystalline TiO2 particles. Chelation of surface Ti atoms with electron-donating bidentate ligands in these systems changes the electronic properties of the nanocrystalline particles. Interfacial electron-transfer dynamics in dye-sensitized semiconductors have been studied by us8 and many other groups.9 In the present investigation we have studied electron injection and back electron transfer (BET) reaction of dibromo fluorescein (DBF) sensitized bare and (6) (a) Murphy, C. J.; Lisensky, G. C.; Leung, L. K.; Kowach, G. R.; Ellis, A. B. J. Am. Chem. Soc. 1990, 112, 8344. (b) Lorenz, J. K.; Kuech, T. F.; Ellis, A. B. Langmuir 1998, 14, 1680. (c) Cohen, R.; Kronic, L.; Shanzer, A.; Cahen, D.; Liu, A.; Rosenwaks, Y.; Lorenz, J. K.; Ellis, A. B. J. Am. Chem. Soc. 1999, 121, 10545. (d) Ellis, A. B.; Kaiser, S. W.; Boltz, J. M.; Wrighton, M. S. J. Am. Chem. Soc. 1977, 99, 2839. (e) Natan, M. J.; Thackeray, J. W.; Wrighton, M. S. J. Phys. Chem. A 1986, 90, 4089. (7) (a) Rajh, T.; Ostafin, A. E.; Micic, O. I.; Teide, D. M.; Thurnauer, M. C. J. Phys. Chem. 1996, 100, 4538. (b) Rajh, T.; Nedeljkovic, J. M.; Chen, L. X.; Poluektov, O.; Thurnauer, M. C. J. Phys. Chem. B 1999, 103, 3515. (c) Thurnauer, M. C.; Rajh, T.; Teide, D. M. Acta Chem. Scand. 1997, 51, 610. (d) Rajh, T.; Poluektov, O.; Dubinski, A. A.; Wiederrecht, G.; Thurnauer, M. C.; Trifunac, A. D. Chem. Phys. Lett. 2001, 344, 31. (e) Rajh, T.; Chen, L. X.; Lukas, K.; Liu, T.; Thurnauer, M. C.; Teide, D. M. J. Phys. Chem. B 2003, 106, 10543.

10.1021/la035190g CCC: $27.50 © 2004 American Chemical Society Published on Web 01/17/2004

Effect of Surface on Electron Dynamics

sodium dodecyl benzene sulfonate (DBS) modified TiO2 nanoparticles using transient absorption techniques detecting in the picosecond and microsecond time domain. BET dynamics of both the systems have been compared in both the time domains. The electron injection to both bare and surface-modified TiO2 nanoparticles by DBF have been confirmed by transient absorption studies, where conduction band electrons in the nanoparticles have been detected in the visible region and near-IR region. BET rate constants have been measured from the bleach recovery kinetics and found to be multiexponential. It has been observed that BET rate is slower for DBF-sensitized surface-modified TiO2 nanoparticles compared to the bare one. Slow BET for the case of surface-modified TiO2 can be explained by the fact that free energy (-∆G°) is increased on surface modification compared to that of bare one for DBF/TiO2 systems. On surface modification pinning of the Fermi level of semiconductor takes place, where the energy levels (conduction band edge, Fermi level, shallow and deep trap state) of semiconductor nanoparticles pushed up in energy. As a result, the free energy (-∆G°) of BET reaction increases. BET reaction for the DBF/TiO2 system comes under the Marcus inverted regime. Hence, BET reaction becomes slow on the modified TiO2 surface compared to the bare one. To follow BET dynamics in longer time domain, we have carried out transient absorption experiments in microsecond time domain. BET dynamics in microsecond time scale can also be fitted with multiexponential time constants and was found to be faster on a modified surface compared to the bare surface. The reversible trend in BET has been explained in the following way. In the case of bare TiO2, density of the deep trap states are much higher compared to that of modified TiO2 due to the pinning effect of surface modification.10 The HOMO of the surface modifier molecules interacts with the unfilled deep surface states. The density of deep trap states will be decreased in the case of modified TiO2 nanoparticles due to surface passivation. Recombination reaction (BET) is much slower for the electrons in the deep trap state and the parent cation due to lower coupling matrix for BET reaction. As a result, the BET reaction is slower in the longer time domain for DBFsensitized bare TiO2 nanoparticles compare to the modified one. 2. Experimental Section (a) Materials. Dibromo fluorescein was obtained from Aldrich and was used without further purification. Titanium(IV) tetraisopropoxide {Ti[OCH(CH3)2]4} (Aldrich, 97%) and isopropyl alcohol (Aldrich) were purified by distillation. Dodecyl benzene sulfonic acid (DBS) was obtained from Aldrich. Chloroform (CHCl3), dimethylformamide (DMF), and pyridine were obtained from Spectrochem India Ltd. and were used without further purification. (b) Nanoparticle Preparation. We have synthesized DBScapped TiO2 nanoparticles and reported on that earlier.11 Briefly, (8) (a) Ghosh, H. N. J. Phys. Chem. B 1999, 103, 10382. (b) Ramakrishna, G.; Ghosh, H. N. J. Phys. Chem. B 2001, 105, 7000. (c) Ramakrishna, G.; Ghosh, H. N.; Singh, A. K.; Palit, D. K.; Mittal, J. P. J. Phys. Chem. B 2001, 105, 12786. (d) Ramakrishna, G.; Ghosh, H. N. J. Phys. Chem. A 2002, 106, 2545. (e) Ramakrishna, G.; Singh, A. K.; Palit, D. K.; Ghosh, H. N. J. Phys. Chem. B 2004, 108, 1701. (9) (a) Ghosh, H. N.; Asbury, J. B.; Lian, T. J. Phys. Chem. B 1998, 102, 6482. (b) Asbury, J. B.; Hao, E.; Wang, Y. Q.; Ghosh H. N.; Lian, T. J. Phys. Chem. B 2001, 105, 4545 (c) Benko, G.; Hilgendroff, M.; Yartsev, A. P.; Sundstrom, V. J. Phys. Chem. B 2001, 105, 967. (d) Dang, X.; Hupp, J. T. J. Am. Chem. Soc. 1999, 121, 8399. (e) Hannapel, T.; Burfeindt, B.; Storck, W.; Willig, F. J. Phys. Chem. B 1997, 101, 6799. (10) Bard, A. J.; Bocarsly, A. B.; Fan, F. F.; Walton, E. J.; Wrighton, M. S. J. Am. Chem. Soc. 1980, 102, 3671. (11) Ramakrishna, G.; Ghosh, H. N. Langmuir 2003, 19, 505.

Langmuir, Vol. 20, No. 4, 2004 1431 in a 500 mL mixture of freshly prepared TiO2 colloids in water synthesized as reported earlier,8,12 250 mL of toluene was added in a round-bottom flask. The resulting mixture was stirred slowly for 15-20 min. In the stirred solution, 100 mL of 0.2 M DBS (C12H25C6H4SO3Na, sodium dodecyl benzene sulfonate) was added, and the final mixture had been stirred slowly for 3 h. DBS can be easily dissolved in water, because of its ionic nature. As the surface of the TiO2 nanoparticles is positively charged, the sulfonic group (SO3-) of the DBS molecules can easily bind with the nanoparticles. This has been confirmed by FTIR data.14 The DBS captured TiO2 nanoparticles (in water) were extracted to toluene, with the help of a separating funnel. At this stage the organic phase looked a little cloudy. The organic phase was dried in CaCl2 and transformed to an optically clear solution. The organic phase was then refluxed for 2 h, and the solvent was taken out with the help of a rotary evaporator in N2 atmosphere. Dry TiO2 particles capped by DBS, which were left in the flask, could be dispersed in many nonaqueous solvents to get colloidal solution in that particular solvent. The newly capped TiO2 nanoparticles looks like a reverse micelle. To prepare sensitized nanoparticles, DBF was added to TiO2 colloid and sonicated for 1 min. Typical concentration of DBF was kept ∼12 µM in steady-state fluorescence measurements. Concentrations of about 30 µM of dye and 10 g/L of TiO2 are used in flash photolysis experiments. For all the measurements the sample solutions were deoxygenated by continuously bubbling high-purity nitrogen (99.95% Iolar grade from Indian Oxygen Co. Ltd., India) through the solutions. The solutions were flowed through a 1 cm × 1 cm quartz cell during all the measurements. (c) Picosecond Visible Spectrometer. Picosecond laser flash photolysis experiments were carried out using a pumpprobe spectrometer, described elsewhere.13 Briefly, the second harmonic output (532 nm, 8 mJ, 35 ps) of an active-passive modelocked Nd:YAG laser (Continuum, USA, model 501-C-10) was used for the excitation of the samples. The transients produced in the irradiated samples were detected by their optical absorption. A white light continuum (400-950 nm) produced by focusing the residual fundamental (1064 nm) of the Nd:YAG laser onto a 10 cm length quartz cell containing 50:50 (v/v) H2O-D2O mixture was used as the monitoring light source. The probe light was passed through a variable optical delay line (1 m long) and then split into two parts using a 50:50 beam splitter. One part of the monitoring light was used as the reference beam and the other was used as the analyzing beam (passing through the irradiated sample). Both the reference and the analyzing beams were dispersed through a spectrograph and monitored using a dual diode array based optical multichannel analyzer, which is interfaced to a personal computer to process the data. Time-resolved experiments in the microsecond time domain were carried out using the same picosecond laser for excitation, a tungsten lamp as analyzing light source, a Bausch & Lomb monochromator (0.20 m), a PMT (Hamamatsu R-928), and a digital oscilloscope (Tektronics TDS-500, 500 MHz bandwidth).

3. Results and Discussion (a) Dye-Nanoparticle Interaction. To study the dyesensitized electron-transfer reaction in the excited state, it is very important to know the type of interaction between the dye and the nanoparticles. Reports are available on interfacial ET dynamics for dye-sensitized TiO2 nanoparticles mainly in aqueous solution8 and on thin film.9b,c,e In the present investigation, we have carried out dye sensitization experiments in some nonaqueous solvents for the surface-modified (SM) TiO2 nanoparticles. The steady-state optical absorption spectra of dibromo fluorescein (DBF) in chloroform, DBF in DBS/chloroform solution, and SM TiO2 nanoparticles colloidal solution in chloroform are shown in Figure 1. The DBF molecule shows optical absorption up to 560 nm with a peak at 480 (12) Ghosh, H. N.; Adhikari, S. Langmuir 2001, 17, 4129. (13) Ghosh, H. N.; Pal, H.; Sapre, A. V.; Mittal, J. P. J. Am. Chem. Soc. 1993, 115, 11722. (14) Ramakrishna, G.; Ghosh, H. N. Supporting Information.

1432

Langmuir, Vol. 20, No. 4, 2004

Figure 1. Optical absorption spectra of (a) dibromo fluorescein (DBF) dye in chloroform. (b) DBF-sensitized surface-modified TiO2 nanoparticles (10 g/L) in chloroform. (c) DBF in sodium dodecyl benzene sulfonate (DBS)/chloroform solution. Emission spectra of (d) DBF in DBS/chloroform solution and (e) DBF sensitized surface-modified TiO2 nanoparticles. Concentration of DBF dye molecules in curves b-d was ∼12 µM.

nm in chloroform (Figure 1a). The solubility of DBF in chloroform is poor. Now on addition of SM TiO2 nanoparticles, the optical density of DBF (Figure 1b) increases and is red shifted. In presence of SM TiO2 nanoparticles the solubility of DBF also increases. At high concentration of the SM TiO2 nanoparticles (10 g/L) the optical absorbance goes beyond 640 nm with a peak at 500 nm (Figure 1b). We have also observed that in the presence of DBS, the solubility of DBF also increases in chloroform (Figure 1c). We would like to make a point that the concentration of DBF in Figure 1b and Figure 1c is same. To show the change of spectral properties of DBF in the presence of TiO2 nanoparticles at same dye concentration (∼12 µM), we have introduced curve c in Figure 1. It is interesting to see that the optical absorption spectra of DBF do not change in the presence of DBS. This indicates that there is negligible interaction between DBF and DBS in ground state. We have also compared the dye-nanoparticle interaction for the bare TiO2 nanoparticles in aqueous solution, which is shown in Supporting Information. It is also observed that optical absorption spectra of DBF do not change in the presence of DBS in water. We have already explained that xanthene dyes8b (including DBF) have good interaction with TiO2 nanoparticles. Due to the electronic interaction between TiO2 nanoparticle and adsorbed DBF molecule, the optical density increases and the absorption spectrum is shifted to the long wavelength region. To find out about the dye-nanoparticle interaction in the excited state, we have carried out steady-state fluorescence experiments for the above systems. We have observed in our earlier studies8b that DBF gives a broad (450-650 nm) emission with a peak around 560 nm. In the presence of TiO2 nanoparticles, the emission quantum yield of DBF was drastically reduced. This was attributed to the electron injection from the excited state of the dyes to the conduction band of the nanoparticle. We have also carried out similar measurements for SM TiO2 nanoparticles in chloroform. On excitation, DBF molecules in DBS/ chloroform give an emission band (450-650 nm) with a peak at 550 nm Figure 1d). But in the presence of SM TiO2 nanoparticles in chloroform, the emission quantum yield was drastically reduced (Figure 1e). The emission quantum yield of DBF is decreased by a factor of 6 (Figure 1). This reduction in emission quantum yield has been attributed to the electron injection from DBF to SM TiO2 nanoparticles.

Ramakrishna et al.

Figure 2. Transient absorption spectra of dibromo fluorescein (DBF) sensitized surface-modified (SM) TiO2 nanoparticles in chloroform at (a) 0, (b) 66, (c) 132, (d) 330ps (e) 1.32, (f) 2.31, (g) 3.63, and (h) 5.28 ns after excitation at 532 nm. The spectrum at each time delay consists of a small positive peak at 500 nm, a bleach at 510-560 nm region centered around 530 nm, and a broad positive absorption feature in the whole spectral region (570-700 nm). These features are assigned to the cation radical and the ground-state bleach of dibromo fluorescein dye and injected electron in the nanoparticles. Inset: Kinetic trace of the bleach recovery at 530 nm.

(b) Electron Injection and Charge Recombination. It has been demonstrated by us8b and many workers9c that on optical excitation of xanthene dye molecules adsorbed on colloidal and thin film TiO2 surface, electrons are injected into the conduction band of the nanoparticles. In the present investigation, we have carried out picosecond laser flash photolysis experiments exciting at 532 nm for DBF-sensitized surface-modified TiO2 nanoparticles in different nonaqueous solvents and bare TiO2 nanoparticles in water to follow the interfacial electrontransfer dynamics on the semiconductor surface. Figure 2 shows the transient absorption spectra of dibromo fluorescein (DBF) sensitized SM TiO2 nanoparticles in chloroform from 470 to 700 nm wavelength region at different time delays. A bleach (negative absorption) in the spectral region 510-560 nm centered around 530 nm and a broad positive feature in the spectral region 570700 nm have been observed. The negative absorption change has been attributed to the bleaching of the ground state of the dye upon excitation by the laser pulse. The broad spectral absorption in the 570-700 nm region is attributed to the conduction band electrons in the nanoparticles.8,9e It has already been shown by many workers that the conduction band electrons can be detected both by visible8,9e and infrared absorption.9a,b However, a small positive absorption appearing below 510 nm with a peak at 500 nm can be unambiguously assigned to the oxidized state of DBF. The optical density of the cation peak is small due to the ground-state absorption of the dye molecules in that region. To confirm the transient absorption peak for the cation radical, we have carried out the pulse radiolysis experiment of DBF in water saturated with N2O in the presence of NaN3.14 Under such experimental conditions, only the cation radical of DBF is produced following pulse radiolysis. Transient absorption spectra obtained from the above solution show a strong absorption peak at around 500 nm and a small peak at 440 nm. In the present investigation, the electron injection process can be given by the following scheme hν

TiO2 + DBF S [TiO2-DBF]adsorb 9 8 electron injection e-cb(TiO2) + DBF+ (1)

Effect of Surface on Electron Dynamics

Langmuir, Vol. 20, No. 4, 2004 1433

Figure 3. Comparison of bleach recovery kinetics of DBFsensitized TiO2 nanoparticles at 530 nm (a) bare and (b) modified particles as measured by transient picosecond flash photolysis.

where DBF+ is the cation radical of dibromo fluorescein dye and e-cb is the conduction band electron in TiO2 nanoparticles. In the present investigation we could not monitor the cation radical properly due to low intensity of the probe light in the region of cation absorption and also due to the huge ground-state absorption of the dye molecule. However, the bleach at 530 nm recovers (Figure 2 inset) and is due to the back reaction involving recapture of the conduction band electrons by the DBF cation radical.

Figure 4. Comparison of bleach recovery kinetics of DBFsensitized TiO2 nanoparticles at 510 nm (a) and bare (b) modified particles as measured by transient microsecond flash photolysis. Table 1. Bleach Recovery Kinetics of DBF-Sensitized Bare and Surface-Modified (SM) TiO2 Nanoparticles in Different Solvents after Exciting with 532 nm Laser Light

system/solvent bare TiO2/water SM-TiO2/CHCl3

e-cb(TiO2)

+ recombination

+ DBF 98 [TiO2-DBF]Adsorb

(2)

The recovery of the observed bleach can be fitted by a multiexponential function with time constants of 172 ps (28.4%) and >5 ns (71.6%), although up to 6 ns (our maximum time limit) only 70% of the bleach recovers. In our earlier reports8b we have observed the recombination dynamics of DBF-sensitized bare TiO2 nanoparticles in water is multiexponential with typical time constants of 72 ps (13.9%), 1.54 ns (46.5%), and >5 ns (39.6%) as measured by time-resolved picosecond absorption spectrometer. It has been observed above that more time constants are required to fit the decay in the case of bare particles compared to the modified one in the same time domain (up to 6 ns). This may be due to more heterogeneity of bare nanoparticles due to higher density of surface states. We have compared the recombination dynamics in Figure 3 for bare and modified TiO2 nanoparticles sensitized by DBF molecules. It is interesting to see that recombination dynamics is slow for the DBF/TiO2 nanoparticle system on the modified surface. We have also carried out sensitization experiments for the DBF and SM-TiO2 by dispersing the nanoparticles in different nonaqueous solvents by using picosecond flash photolysis. Picosecond time-resolved spectra have been obtained for all the above systems in the 500-800 nm wavelength region. In all the above cases bleach around 500-570 nm and a broad absorption band in the 570-800 nm wavelength region for the electron have been observed. Bleach recovery dynamics has been determined following their respective bleach peaks and is shown in Table 1. To compare the recombination dynamics in longer time scale for DBF/TiO2 systems both for bare and SM particles, we have carried out flash photolysis experiments in the microsecond time domain. The bleach recovery kinetics of the above systems are shown in Figure 4 with monitoring at 510 nm. We have observed that the recombination reaction can be fitted multiexponentially with typical time constants of 1.55 µs (58.6%) and 10 µs (41.4%) for surfacemodified nanoparticles in chloroform (Table 1). We have also carried out recovery kinetics for the DBF/TiO2 system on a bare surface. The recombination kinetics has been fitted with a multiexponential function, and corresponding time constants are given in Table 1. It is interesting to see

SM-TiO2/pyridine SM-TiO2/DMF

lifetime of the transient in shorter time (ns)

lifetime of the transient in longer time (µs)

τ1 ) 0.072 (13.9%) τ2 ) 1.54 (46.5%) τ3 > 5.00 (39.6%) τ1 ) 0.172 (28.4%) τ2 > 5.00 (71.6%) τ1 ) 0.156 (68.3%) τ2 >5.00 (21.7%) τ1 ) 0.221 (62.3%) τ2 > 5.00 (37.7%)

τ1 ) 3.29 (66.5%) τ2 > 20 (33.5%) τ1 ) 1.54 (68.8%) τ2 > 20 (31.2%) τ1 ) 1.94 (75.8%) τ2 > 20 (24.2%) τ1 ) 1.76 (74.1%) τ2 > 20 (25.9%)

that the recombination reaction is slow for the DBF/TiO2 system on bare surface compared to the modified one (Figure 4). We can observe the reversible trend in recombination dynamics for the above systems in picosecond and microsecond time domains. We have carried out bleach recovery kinetics on modified surface by changing the solvent and observed that the recombination dynamics were very similar (Figure 4, Table 1). The detailed mechanisms of back electron transfer reaction for the above systems are discussed in the following section. (c) Mechanism of Charge Recombination Reaction. According to the semiclassical theory15 formulation for the back electron transfer (BET) rate constant kBET is given in eq 3 as

kBET )

( )

{

}

(∆G0 + Λ)2 2π 1 [HAB]2 exp p 4ΛkT (4πΛkT)1/2

(3)

where Λ is the total reorganization energy, HAB is the coupling element, and ∆G° is the overall free energy of reaction ()EC - ES/S+), where EC is the potential of electrons in the conduction band of the semiconductor (-0.49 V),16 and ES/S+ is the redox potential of the adsorbed dye (Scheme 1). In the present investigation we have compared BET dynamics for the DBF/TiO2 system for bare nanoparticles with the modified one. As they are same dye-TiO2 nanoparticle system, we can imagine that Λ, the total reorganization energy, and HAB, the coupling element, will be same for the systems studied. In such a case, kBET will depend on the overall free energy of reaction (∆G°). We have observed that recombination reaction (BET) between the injected electron and the parent cation is faster on the bare nanoparticle surface compared to the modified one. The results can be explained by invoking a model in which the energy levels in the semiconductor nanoparticles are shifted due to surface modification (15) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985, 811, 265. (16) Hagfeldt, A.; Gratzel, M. Chem. Rev. 1995, 95, 49.

1434

Langmuir, Vol. 20, No. 4, 2004

Scheme 1. Mechanistic Scheme of Electron Transfer (ET) and Effect of Surface Modification on ET in Dibromo Fluorescein (DBF) Sensitized Bare (left) and Modified (right) TiO2 Nanoparticlesa

a Fermi level pinning is seen due to surface modification on TiO2 nanoparticles (right). The energy levels of the modified nanoparticles shifts toward more negative and effective free energy (-∆G°) increases. On surface modification, deep trap states are passivated.

(Scheme 1). Dimitrijevic et al.17 have observed that the position of the Fermi level in modified TiO2 colloids was shifted by at least negative 0.1 V compared to the unmodified one. It has been reported previously by Ellis et al.6a-d and Natan et al.6e that strong adsorption of negative counterions on the electrode surface shifted the flat band potentials (Vfb) to more negative values. Yan and Hupp18 have reported that with increasing pH on the semiconductor electrode surface, the flat band potentials (Vfb) move toward more negative values. In the present investigation, DBS molecule (modifier) adsorbed strongly to the nanoparticles through a sulfonic acid group (SO3H-, negatively charged). So on surface modification the different energy levels of the semiconductor nanoparticles such as deep trap states, shallow trap states, Fermi level, and conduction band edge will be changed and will move up in energy due to interaction with the modifier molecule. As the conduction band edge of the modified colloids is pushed up in energy, the overall free energy of BET reaction of DBF/SM-TiO2 will be increased compared to the DBF/TiO2 system (-∆G2° > -∆G1° Scheme 1). According to the Marcus electron transfer (ET) theory, ET rates ultimately decrease with the increasing thermodynamic driving force (-∆G°).15 This high exoergic region is often termed “inverted regime”. Back electron transfer processes in dye-sensitized TiO2 nanoparticle surfaces fall in the Marcus inverted regime for its high free energy of reaction.8b,9d,19 In this region, with increasing driving force (-∆G°) of reaction, the rate of BET decreases. As a result, BET rate on the modified surface is slower compared to that on the bare one. We have carried out dye sensitization experiments on surface-modified nanoparticles in different solvents to see the effect of dielectric constant of the medium on BET kinetics. However we have not observed much difference in BET rate constants (Table 1) on the modified surface with changing the polarity of the solvents. Here the total reorganization energy Λ may change marginally. So kBET on the modified surface will depend mostly on free energy (-∆G°) of the (17) Dimitrijevic, N. M.; Savic, D.; Micic, O. I.; Nozik, A. J. J. Phys. Chem. 1984, 88, 4278. (18) Yan, S. G.; Hupp, J. T. J. Phys. Chem. 1996, 100, 6867. (19) Lu, H.; Prieskorn, J. N.; Hupp, J. T. J. Am. Chem. Soc. 1993, 115, 4927.

Ramakrishna et al.

reaction. In the present investigation we have observed little difference in recombination dynamics (BET) in different polar solvents on modified nanoparticles surface both in pico- and in microsecond time domains (Table 1). It is interesting to follow the recombination dynamics of the DBF/TiO2 system in longer time domain (microsecond) for the bare and surface-modified nanoparticles. BET dynamics was found to be multiexponential with time constants of 3.29 µs (66.5%) and >20 µs (32.5%) for bare TiO2 in water and 1.55 µs (68.6%) and >20 µs (31.4%) for surface-modified nanoparticles in chloroform. On a modified surface, the injected electron and the parent cation recombine faster than that on a bare surface in longer time domain (microsecond). We have already observed (previous paragraph) that the recombination dynamics of the DBF/TiO2 system is faster on the bare surface compared to that on the modified surface in shorter time scale (picosecond). This reversible trend in BET dynamic has been explained in the following way. On surface modification the orbital interaction takes place between the HOMO of the surface modifier molecules and the unfilled deep surface states (which acts as LUMO).6c On interaction, the energy level of LUMO goes up and that of the HOMO goes down. So energy levels of the surface states move toward more negative. As a result the density of deep trap states decreases (Scheme 1). Energetically shallow trap states are higher in energy compared to the deeper one. So, on surface modification, interaction between the modifier molecular orbital and the deep trap states is much more compared to that with the shallower one. As a result, the density of the deep trap states decreases more than that of the shallow trap states. In the microsecond time domain, greater number of the injected electrons will reside in shallow trap states for modified particles and in deep trap states for bare particles. It has been observed by us8c,e and Moser et al.20 that the recombination reaction (BET) is much slower for the deep trap state electron and the parent cation due to low coupling matrix for BET. As a result BET reaction is slower for the case of DBF-sensitized bare TiO2 nanoparticle compared to the modified one in the microsecond time domain. Although free energy (-∆G°) of the BET reaction on modified particles favors slow recombination as we have observed in early time domain, weak coupling of deep trapped state electrons slows the recombination dynamics on the bare nanoparticle surface. The results can also be explained by adopting the model as suggested by Durrant et al.21 and Tachiya et al.22 for slow recombination reaction in the dye-sensitized electrontransfer process, where the injected electrons can diffuse out some distance on the surface of the nanoparticles before getting trapped in a particular trap state. Under that condition, the electrons are apart from the parent cations so that the electronic coupling is negligibly small and recombination through tunneling cannot take place. But the trapped electron can move assisted by thermal energy; namely, they can move through trap and detrap process. So the diffusion of the injected electrons from trap-to-trap may take place. As we know that the density of trap states in the surface-modified nanoparticles are less compared to the bare one, so the probability of diffusion from one (20) Moser, J. E.; Gratzel, M. Chem. Phys. 1993, 176, 493. (21) (a) Haque, S. A.; Tachibana, Y.; Willis, R. L.; Moser, J. E.; Gratzel, M.; Klug, D. R.; Durrant, J. R. J. Phys. Chem. B 2000, 104, 538. (b) Nelson, J.; Haque, S. A.; Klug, D. R.; Durrant, J. R. Phys. Rev. B 2001, 63, 205321. (22) Barzykin, A. V. Tachiya, M. J. Phys. Chem. B 2002, 106, 4356.

Effect of Surface on Electron Dynamics

trap to another trap will be less. As a result the recombination will be faster on the modified particle surface. From the Figure 4b, one can see that the recombination is not complete even on the modified surface indicating the presence of deep trap states even in the surfacemodified nanoparticles. This is because we were unable to remove trap states quantitatively though surface modification. Recently we are carrying out dye-sensitized electron transfer on the modified nanoparticle surface by changing the modifier molecule. We are on the process of finding a suitable modifier molecule that can remove the surface states of a nanoparticles more efficiently. 5. Conclusion Pico- and microsecond transient absorption spectroscopy have been carried out to study the effect of surface modification on electron injection and back electron transfer (BET) dynamics of dibromo fluorescein (DBF) sensitized TiO2 nanoparticles capped (modified) with sodium dodecyl benzene sulfonate (DBS). Electron injection has been confirmed by direct detection of electrons in the conduction band, cation radical of the adsorbed dye, and a bleach of the dye in real time as monitored by transient absorption spectroscopy in the visible and nearIR regions. Charge recombination (BET) dynamics have been measured by monitoring the bleach recovery kinetics of the adsorbed dye at 530 nm and were found to be

Langmuir, Vol. 20, No. 4, 2004 1435

multiexponential. BET dynamics have been found to be slow for modified particles compared to the bare one in the earlier time domain (pico- and nanosecond). On surface modification, the flat band potential of the nanoparticles pushed up in energy. As a result, the free energy (-∆G°) of reaction increases. BET reaction in DBF/TiO2 system falls in the inverted regime of ET reaction, where with increasing free energy, the BET rate decreases. However, a reversible trend in BET dynamics has been observed for the above systems in the longer time domain (microsecond), where the BET reaction is faster for modified particles compared to the bare one. Surface modification removes many of the deeper trap states, which are responsible for the long time recombination dynamics of injected electrons (deep-trapped) and the parent cation due to low coupling strength of BET reaction. So BET reaction is found to be slow on a bare particle surface compared to the modified one. Acknowledgment. We are thankful to Dr. T. Mukherjee and Dr. J. P. Mittal for encouragement. Supporting Information Available: Transient absorption spectrum of DBF cation radical as obtained from pulse radiolysis in N2O saturated aqueous solution. This material is available free of charge via the Internet at http://pubs.acs.org. LA035190G