CORE/SHELL PHOTOANODES 1 6.1. Introduction
Dye-sensitized photoelectrosynthesis cells (DSPECs) offer a solution to solar energy storage by integrating molecular chromophores and catalysts with mesoporous films of wide bandgap oxide semiconductors for light-driven water splitting and carbon dioxide reduction.1,2 In the simplest case of water splitting, the overall reaction can be divided into two half-reactions: water oxidation at a photoanode (Equation 6.1) and proton reduction at a photocathode (Equation 6.2).
Equation 6.1
2H2O → O2 + 4H+ + 4e–
Equation 6.2 4H+ + 4e– → 2H2
At the photoanode in a water-splitting DSPEC, oxidation of the catalyst must occur prior to water oxidation. Activation of the catalyst is typically achieved through a series of charge transfer events.3 First, a molecular chromophore bound to a wide bandgap semiconductor (e.g. TiO2, SnO2) absorbs a photon of light, forming a photo-excited state. Electron transfer (a.k.a.
1 This chapter presents work for a manuscript in preparation with co-authors Melissa K. Gish, M. Kyle Brennaman, Byron H. Farnum, Joseph L. Templeton, Thomas J. Meyer, and John M. Papanikolas.
electron injection) can occur from the photo-excited chromophore to the conduction band of the semiconductor, leaving the chromophore oxidized. Hole transfer, or oxidation of the molecular catalyst by the chromophore, can now occur. For O2 production, this sequence of events must occur four times. The injected electrons can either diffuse to the back contact of the mesoporous semiconductor film or recombine with oxidized molecular species, a deleterious process known as back electron transfer (BET). BET limits the efficiency of DSPEC water oxidation
photoanodes by preventing the necessary accumulation of four holes on the catalyst; it is also thought to limit efficiency in dye-sensitized solar cells (DSSCs), which require only single- electron transfer events rather than charge accumulation.4,5 Despite rapid excited-state electron injection and charge transfer among molecular species bound to TiO2 nanoparticles (Ecb ≈ –0.15 V vs. NHE at pH 1), BET to the oxidized catalyst still occurs on the microsecond timescale at pH 1.6,7
Core/shell oxides are one method for inhibiting BET to oxidized molecules on
photoanodes. In a core/shell structure, an underlying material (core) is physically coated with another material. If each material (core and shell) is a distinct metal oxide semiconductor, misalignment of their conduction band edges creates an energy gradient at their interface. This design principle is derived from Nozik’s “photochemical diode” in which two semiconductors with mismatched conduction band edges are sandwiched together, forming a rectifying junction.8 In DSPEC photoanodes, it is advantageous for Ecb(core) to be more positive than that of
Ecb(shell), an arrangement that promotes electron transfer from shell to core (Figure 6.1a). The electron effectively becomes “trapped” in the core, leading to long-lived charge separation.
Figure 6.1. a) Schematic depiction of the photophysical processes of excitation and charge transfer occurring at the chromophore-SnO2/TiO2 core/shell interface with excitation and
forward electron transfer indicated with green numbers and excited-state decay and back electron transfer indicated with red numbers (adapted from reference 9); b) Molecular structure of RuP2+; c) Schematic depiction of physical core/shell structure.
One commonly employed core/shell structure utilizes a SnO2 core and TiO2 shell. While formation of the TiO2 shell has historically been performed by chemical deposition,10,11 recently atomic layer deposition (ALD) has allowed for low temperature deposition of conformal,
thickness-controlled shells in high aspect ratio structures.5,12,13 Transient absorption spectroscopy on the nanosecond timescale has been used to investigate the mechanism of charge
SnO2 TiO2 Chr Chr* 1 2 3 4 5 6 1) Light-absorption 2) Excited-state decay
3) Electron injection (shell)
4) Electron injection (core)
5) BET (shell) 6) BET (core) –E N N N N N N Ru 2+ P P O HO HO O HO HO Core Shell
a)
b)
c)
recombination in nanoSnO2/TiO2 core shell structures derivatized with the molecular
chromophore [Ru(2,2’-bipyridine)2(4,4’-dpbpy)][Cl]2 (RuP2+, Figure 6.1b; 4,4’-dpbpy = [2,2’- bipyridine]-4,4’-diylbis(phosphonic acid)).9 In this report, recombination of the injected electron with the oxidized dye occurs by electron tunneling through the TiO2 for shells less than 3.2 nm thick. For thicker shells, direct recombination from the TiO2 shell is observed. Despite these promising conclusions, ultrafast transient absorption spectroscopy data for electron injection and recombination with core/shell structures are needed to corroborate the proposed mechanism.
Herein we investigate the electron injection and recombination kinetics from the 250 fs to ms timescale following 480-nm laser excitation of RuP2+-derivatized nanoSnO2/TiO2 core/shell electrodes in 0.1 M HClO4. Our results suggest that an ultrafast BET pathway is accessible for
nanoSnO2/TiO2|-RuP2+ core/shell electrodes with TiO2 shells between 1 and 2 nm thick;
however this pathway is not evident for thin (<0.5 nm) shells. Annealing these photoanodes prior to dye loading resulted in a diminished contribution of the ultrafast BET pathway, suggesting the structure and composition of the TiO2 shell plays a major role in charge separation lifetime on
nanoSnO2/TiO2 core/shell photoanodes. 6.2. Results and Discussion
6.2.1. Construction of Core/Shell Electrodes
Core/shell electrodes were constructed by performing ALD of TiO2 on mesoporous metal oxide films. The thickness of the TiO2 shell was controlled by varying the number of
TiTDMA/H2O cycles. Shell thickness was estimated by ellipsometry measurements performed on witness Si. Growth characterized for individual electrodes was less than expected; however across all samples the slope (thickness/cycles) was approximately the expected 0.6 Å cycle-1 (Figure 6.2).14,15 Slow growth could be attributed to the low temperature (150 °C) used for
deposition.16 Electrodes were not annealed following TiO2 deposition (unless otherwise indicated), resulting in amorphous, slightly colored TiO2 shells.17
Figure 6.2. Plot of expected (blue diamonds) and observed (red squares) TiO2 thickness as a function of TiTMDA/H2O cycles as determined by ellipsometry on planar Si. The red line is a linear fit of the observed data. (conditions: Treactor = 150 °C, TTiTDMA = 75 °C; each reactant was held in the chamber for 20 s, then purged for 60 s)
The molecular chromophore RuP2+ was synthesized as its chloride salt by literature procedures.18 Dye loading was achieved by soaking the desired electrode in a methanol solution of RuP2+ (~1 mM) overnight, followed by rinsing with methanol and drying under a stream of air. The amount of RuP2+ loaded onto the core/shell electrodes was determined using UV-visible absorption spectroscopy and Equation 6.3 by subtracting the nanoMOx/TiO2 background
spectrum from the nanoMOx/TiO2|-RuP2+ absorption spectrum.19 For the unannealed
nanoSnO2/TiO2|-RuP2+ electrodes, Γ(RuP2+) = (5.6 ± 0.8) × 10-8 mol cm-2. For the unannealed
nanoZrO2/TiO2|-RuP2+ electrodes, Γ(RuP2+) = (5.1 ± 0.7) × 10-8 mol cm-2. These data suggest that similar amounts of RuP2+ are adsorbed to the core/shell electrode films.
Equation 6.3
Γ = A(λ) × (ε(λ) × 1000)-1
6.2.2. Electron Injection Results
The kinetics and apparent relative yields of electron injection of RuP2+* on core/shell electrodes of both nanoSnO2/TiO2 and nanoZrO2/TiO2 as a function of TiO2 shell thickness (0 to 1.8 nm) in argon-deaerated 0.1 M HClO4 were investigated by transient absorption spectroscopy. Following 480-nm excitation (100 nJ/pulse, 150-fs pulse width), the differences between the white light absorption spectrum with and without excitation (i.e. pump on and pump off) were monitored from 200 fs to 1.3 ns. There are two main features of the transient absorption
spectrum of RuP2+* on TiO2: a positive feature for a bpy•– radical (λmax = 380 nm) and a negative ground state bleach (GSB; λmax = 450 nm) corresponding to the MLCT transition of RuP2+
(Equation 6.4).20 Injection of photo-excited electrons into the semiconductors’ conduction bands was evaluated by monitoring the π-π* absorption feature of the reduced bpy ligand characteristic of the excited state of RuP2+, namely RuP2+*, at 376 nm. Excited-state electron injection occurs
from the surface-bound 4,4’-dpbpy•– ligand (Equation 6.5),21 causing the bpy•– absorbance to decay as injection occurs. The GSB intensity is maintained throughout the injection process because the metal center remains oxidized. The pump wavelength used in all transient absorption