Multimolecular assemblies on high surface area metal oxides and their role in interfacial energy and electron transfer

Abstract

High surface area metal oxides offer a unique substrate for the assembly of multiple molecular components at an interface. The choice of molecules, metal oxide, and the nature of the assembly method can have a profound influence on the mechanism, rate, and efficiency of photoinduced energy and electron transfer events at the interface. Owing to their diversity and high level of control, these interfacial assemblies are of interest for numerous applications including solar energy conversion, photoelectrosynthesis, photo-writable memory, and more. Although these assemblies are generated with very different goals in mind, they rely on similar surface binding motifs and molecular structure–property relationships. Therefore, the goal of this review is to summarize the various strategies (i.e. co-deposition, axial coordination, metal ion linkages, electrostatics, host–guest interactions, etc.) for assembling chromophores, hosts, electron donors/acceptors, and insulating co-adsorbent molecules on mesoporous metal oxide substrates. The assembly, synthesis, and characterization, as well as subsequent photoinduced events (i.e. cross-surface energy/electron transfer, interchromophore energy transfer, electron injection, and others) are discussed for the various assembly strategies.

---

Jamie C. Wang

Jamie C. Wang received her BS degree in Applied Chemistry from National Chi Nan University (Nantou, Taiwan) in 2013. She is currently a PhD candidate under the supervision of Prof. Kenneth Hanson at Florida State University. Her current research focuses on understanding and controlling electron transfer at dye–semiconductor interfaces.

Sean P. Hill

Sean P. Hill received his BS from Louisiana State University, Baton Rouge in 2013. He is currently a PhD candidate under the supervision of Prof. Kenneth Hanson at Florida State University in the Department of Chemistry and Biochemistry. His research focuses on harnessing molecular photon upconversion using self-assembled bilayer on metal oxide surfaces.

Tristan Dilbeck

Tristan Dilbeck received her BS in Chemistry from California Polytechnic State University, San Luis Obispo in 2014. Upon coming to Florida State University, she joined the lab of Prof. Kenneth Hanson to study energy and electron transfer dynamics where she has focused on measuring the rate of energy and electron transfer in photon upconversion self-assembled bilayers on high surface area metal oxides.

Omotola O. Ogunsolu

Omotola O. Ogunsolu obtained her BS degree at University of Ibadan, Nigeria in 2010 and her MS degree from University of Sheffield, UK in 2013. She is currently a PhD candidate studying Materials Science and Engineering at Florida State University. Her current research focuses on influencing and characterizing interfacial electron transfer events in dye-sensitized solar cells.

Tanmay Banerjee

Tanmay Banerjee received his BS and MS degree from University of Calcutta, India and his PhD from National Chemical Laboratory, Pune, India in 2014. He was a postdoctoral research associate in Prof. Kenneth Hanson's Lab at Florida State University (2014–2016). He currently works as a scientist in the group of Prof. Bettina V. Lotsch at the Max Planck Institute for Solid State Research, Stuttgart, Germany. His research interests include electron and energy transfer processes in DSSC and photocatalysis.

Kenneth Hanson

Kenneth Hanson received a BS in Chemistry from Saint Cloud State University (2005), his PhD from the University of Southern California (2010), followed by an appointment as a postdoctoral scholar at the University of North Carolina at Chapel Hill (2010–2013). His independent research career began in 2013 at Florida State University as a member of the Department of Chemistry & Biochemistry and is affiliated with the Materials Science & Engineering program. His current research interests include the design, synthesis, and characterization of photoactive molecules/materials with particular emphasis on manipulating energy and electron-transfer dynamics at organic–inorganic interfaces using multilayer self-assembly.

---

1. Introduction

Photoinduced energy and electron transfer are critical processes in photosynthesis, bioimaging, solar energy conversion, chlorometric/fluorometric sensing, and other applications. Molecular species, as opposed to solid-state materials, are ideal to study/control these phenomena because exquisite synthetic variation of their structures can be used to tune the photo- and electrochemical properties of the chromophores/donors/acceptors.

Over the past century, the seminal work of Marcus,^1,2 Dexter,³ Förster,⁴ and others has been key to understanding the thermodynamic and kinetic parameters that dictate intermolecular energy and electron transfer events. The majority of this research has been conducted on molecules dissolved in solution. However, incorporating molecules in media/lattices offers many potential advantages over solutions, including fixed molecular orientations and distances, and the ability to harness and observe kinetic events that are not diffusion limited. The earliest and most notable example of fixed molecular structures of this type is that of natural photosynthetic systems. Through millions of years of evolution these light harvesting assemblies have developed a protein scaffolding to control intermolecular distances and orientations in order to maximize the efficiency of photoinduced energy and electron transfer events.⁵

Understanding and controlling intermolecular interactions in rigid scaffoldings is crucial for maximizing the efficiency of applications such as artificial solar energy conversion. As a result there is growing interest in generating rigid chromophore–donor/acceptor systems like metal–organic frameworks,⁶ molecular crystals,^7,8 phospholipid bilayers,^9,10 derivatized proteins/DNA,^11,12 and others. Adhering molecules to a nanocrystalline metal-oxide surface is a particularly appealing strategy because complex, multimolecular structures can be prepared through a simple soaking and self-assembly procedure. Using this strategy, molecules are bound to a photo-/electroactive interface, where intermolecular interactions can be used to perturb electrochemical and photophysical properties of the molecules. Interactions between molecules at the interface can have a profound influence, both positive and negative, on molecule–metal oxide based applications including dye-sensitized solar cells (DSSCs), dye-sensitized photoelectrosynthesis cells (DSPECs), sensing, photo-writable memory, and more. These seemingly disparate applications are the result of similar surface binding motifs and molecular structure–property relationships at the interface.

Despite the increasing popularity and importance of these interfaces there has yet to be a comprehensive review on how multimolecular assemblies on high surface area metal oxides can influence photoinduced energy and electron transfer events at the interface. Therefore, the aims of this review are to (1) compile the current literature for various types of multimolecular interfacial assemblies, (2) summarize their role in energy and electron transfer events, (3) share commonalities between these different types of assemblies/events, (4) describe the plethora of measurement techniques used to characterize these interfaces, and (5) encourage readers to develop new strategies to characterize and harness molecular interactions at interfaces.

1.1 Scope

In terms of scope, this review will focus on multimolecular species on high surface area metal oxides and their role in photoinduced events. The discussion has been restricted to high surface area metal oxides because, although planar surfaces are important for a number of applications including electrochemistry,¹³ sensing,¹⁴ and others, high surface area is necessary to achieve sufficient absorption/emission for standard spectroscopic measurements and high absorbance is an essential precursor for efficient solar energy conversion.

Multimolecular, in this case, is defined as the adsorption/binding of one or more electrochemically and photophysically distinct small molecules (i.e. not polymers,¹⁵ quantum dots,^16,17 atomic layer deposited films,^18,19etc.) onto the metal oxide surface. The discussion of pre-synthesized assemblies (i.e. dimers, dyads, etc.) will be limited to only a few examples of covalently linked molecules containing electronically distinct molecular moieties. The molecules of interest are categorized as chromophores (C; used interchangeable with dye), hosts, electron donors (D), electron acceptors (A), and “insulating” molecules that are photo- and electrochemically inert under standard measurement conditions for metal oxide interfaces. Catalysts are also an important class of molecules for these types of assemblies, particularly for application in dye-sensitized photoelectrosynthesis cells, but they have been recently reviewed elsewhere and will be omitted from this discussion.²⁰

The review will be partitioned by photophysical process rather than by the method of assembly formation/interaction (e.g. co-deposition, electrostatics, metal ion coordination, etc.). First, in Section 2 we will discuss homochromophoric events like aggregation (Section 2.1), multi-photon/electron generation (Section 2.2), and cross surface energy and electron migration (Section 2.3). Then in Section 3 we cover multi-molecular strategies to control interfacial electron transfer including co-adsorbents (Section 3.1), chromophore–donor/acceptor systems (Section 3.2), and supersensitization (Section 3.3). In Section 4, the co-deposition (Section 4.1), antenna assembly (Section 4.2), and device engineering (Section 4.3) strategies to harness cooperative effects of complementary chromophores will be covered. Finally, supramolecular interactions like host–guest type assemblies will be discussed in Section 5.

The application cited for a majority of these multimolecular films is in dye-sensitized solar cells (DSSCs). Since the characterization and discussion of most of the literature are reported in the context of these devices, a general understanding of DSSCs is necessary to comprehend the entirety of the material reviewed below. Therefore, the next section will contain a brief introduction into the dynamic events, performance metrics, and standard characterization techniques for DSSCs. More comprehensive reviews on DSSCs can be found elsewhere.²¹

1.2 Dye-sensitized solar cells

The standard components of a DSSC are depicted in Fig. 1a. In a typical device, transparent conducting substrates are coated with a nanocrystalline semiconducting metal oxide film (typically TiO₂) and a regenerative catalyst (typically platinum) as the anode and cathode, respectively. The lack of visible absorption by the large band gap semiconductor is supplemented by functionalizing the metal oxide with a monolayer of one or more types of dye molecules. Between the two electrodes is an electrolyte solution, or polymer, containing an electro-active molecule also known as a redox mediator.²²

Fig. 1 The (a) components and (b) relative energetics and dynamic events for a standard DSSC. Advantageous and loss events are depicted in black and red, respectively (E_f = Fermi level, k_ex = excitation rate, k_inj = electron injection rate, k_r = radiative relaxation rate, k_nr = non-radiative relaxation rate, k_BET = back electron transfer rate, k_reg = dye regeneration rate, k_rec = recombination rate, k_red = mediator reduction rate).

The relative energetics of these components, as well as the dynamic events of a functioning cell under operational conditions, are depicted in Fig. 1b. Photon absorption by the dye molecule (k_ex) generates an excited state with sufficient potential²³ to inject an electron (k_inj) into the conduction band of the metal oxide. The now oxidized dye is reduced via electron transfer from the mediator, effectively regenerating the ground state dye (k_reg). Concurrently the electron in the metal oxide enters the external circuit to do useful work and then reduces the oxidized mediator at the cathode (k_red).^22,24,25

In addition to the favorable events described above, there are a number of unfavorable processes that hinder the performance of DSSCs. For example, prior to electron injection, the excited state can decay back to the ground state via radiative (k_r) or non-radiative (k_nr) relaxation.²⁵ Prior to entering the external circuit, the injected electron can transfer to the oxidized dye or mediator via back electron transfer (k_BET) and recombination (k_rec), respectively.^22,24

The first and perhaps most important data typically acquired for DSSCs, or any solar cell for that matter, is the current density–voltage (J–V) relationship. For this measurement, the anode and cathode are connected to an electrochemical workstation that sweeps the potential between the electrodes while monitoring the current response both in the dark and under illumination.²⁶ An example J–V curve under irradiation as well as important performance metrics are shown in Fig. 2.

Fig. 2 Current density–voltage curve for a solar cell under illumination (J_sc = short-circuit current density, V_oc = open-circuit voltage, P_max = maximum power point, V_max = voltage at maximum power, J_max = current density at maximum power, P_opt = optimal/theoretical power point, FF = fill factor).

The open circuit voltage (V_oc) is the maximum potential difference between the anode and cathode that can be achieved under illumination when no current is flowing through the cell. In a DSSC, the maximum V_oc is dependent on the energy difference between the quasi-Fermi level in the semiconducting metal oxide and the potential of the redox mediator in the electrolyte. Since the Fermi level is determined by the electron occupancy in the metal oxide, the V_oc is dictated by the ratio between the rate of electrons entering (Rate_e⁻ in) and exiting (Rate_e⁻ out) the metal oxide as described by the modified diode equation (eqn (1))²⁷ where k_B is the Boltzmann constant, T is the temperature, and e is the elementary charge.

To increase the V_oc under a given solar flux, it is necessary to maximize k_inj and k_reg (Rate_e⁻ in) while minimizing k_BET and k_rec (Rate_e⁻ out). As noted by Hupp and others,²⁸ every order of magnitude decrease in the rate of loss pathways (k_bet and k_rec) increases the device V_oc by ∼59 mV.

J _sc, the maximum current flowing while under illumination when there is zero potential, is proportional to the product of several variables including the light harvesting efficiency (LHE(λ)), quantum yields for electron injection (φ_inj), regeneration (φ_reg), and the charge collection efficiency (η_col) as shown in eqn (2).²¹

J_sc ∝ LHE(λ)·φ_injφ_regη_col (2)

In a DSSC, the light harvesting efficiency, more generally described as absorptance, is the percent of incident photons absorbed by the dye molecules at a given wavelength which is dependent on the metal oxide film thickness as well as the surface coverage and extinction coefficient of the dye molecule.²⁷ The φ_inj is effectively the ratio between k_inj from the excited state of the dye to the sum of all excited state decay pathways (i.e. k_inj, k_nr, k_r, etc.) and η_col is the fraction of electrons injected into the metal oxide that enter the external circuit and are not lost through back electron transfer or recombination. Increasing dye coverage, extinction coefficient, solar spectral overlap, and k_inj and decreasing k_nr, k_r, k_BET, and k_rec are all effective strategies for increasing J_sc. A quick note of clarification, for standardization/comparison purposes and to account for the active area of the cell, herein all reported values of J_sc will refer to the short circuit current density (current per area illuminated).

Power conversion efficiency (PCE or η), the primary metric of interest for most solar cell researchers, is the ratio of maximum power output from the device (P_max) to input power from the light hitting the active material (P_in) and can be calculated using eqn (3).

For lab research purposes, an input power of 1 Sun has been standardized to the AM1.5 solar spectrum having an integrated intensity of 100 mW cm⁻².²¹

Since the goal of a DSSC is to generate as much power as possible from solar irradiance, a functioning solar cell is not operating at V_oc or J_sc since the power (P = J × V) at these points is zero. Instead, the solar cell is operated under conditions where the product of current (J_max) and voltage (V_max) is maximized (P_max = J_max × V_max). An important metric relating J_sc, V_oc, and P_max is the fill factor (FF). The FF (eqn (4)) is a fractional term between 0 and 1 that describes how close an actual cell is behaving relative to an ideal diode with the same J_sc and V_oc.

The less than ideal behavior of real DSSCs (FF < 1) is due to shunt resistance (i.e. shorts across the cell) and series resistance (i.e. non-Ohmic contacts between interfaces) that are not infinite and zero, respectively.²⁹

Incident photon-to-current efficiency measurement (IPCE, also known as external quantum efficiency or EQE) is another common characterization technique that effectively measures the photocurrent generation efficiency at a given monochromatic wavelength. If each photon absorbed by the dye(s), regardless of wavelength, has equal probability of generating current, the IPCE spectrum should match the absorptance spectrum.²⁷ A related metric, the absorbed photon-to-current efficiency (APCE, also known as internal quantum efficiency or IQE) is the IPCE divided by the absorptance spectrum. These measurements are particularly useful for quantifying the relative contributions from multi-chromophoric DSSCs.

An additional, but less commonly reported, measurement is electrochemical impedance spectroscopy (EIS). EIS is a relatively complex measurement which will only be briefly mentioned here, but we encourage the reader to learn more about this powerful technique using the references cited here.^30,31 During an EIS measurement, photocurrent response is monitored with respect to an applied potential oscillated at a range of frequencies. The phase difference between applied potential and current response is due to the impedance of the materials/interfaces. Fitting the data using an equivalent circuit composed of resistors and capacitors helps determine the electron lifetime and recombination dynamics at interfaces in the DSSC.^30,31

There are a number of additional spectroscopic techniques mentioned below that are not necessarily performed on complete devices, but are critical in studying the energy and electron transfer events at the interface.²⁷ For example, time-resolved emission measurements, irradiating a sample with a short pulse of light and monitoring the emission intensity with respect to time after the pulse, are used to quantify the intrinsic radiative and non-radiative decay rates of molecules as well as any additional quenching mechanisms, like energy and electron transfer. Transient absorption spectroscopy (TA) is a pump–probe technique that uses two beams of light, one to excite the sample and the second to monitor changes in absorption spectra over time. This technique is critical for monitoring transient species that are non-emissive, like the cation species formed after electron injection, and the lifetimes of those states. These, and other spectroscopic techniques are often used in conjunction with device measurements to generate a complete understanding of the static and dynamic events dictating the device performance.

2. Chromophore–chromophore interactions

Intermolecular interactions between two chromophores that have the same structures can occur in mediums ranging from solutions to neat solids. Metal oxides are particularly effective at facilitating these chromophore–chromophore interactions due to the close proximity and relatively fixed orientation of dyes bound to the surface. These interactions can facilitate ground state aggregate or excimer formation (Section 2.1), multi-photon/electron processes (Section 2.2), and cross-surface energy and electron transfer (Section 2.3).

2.1 Ground and excited state aggregation

Aggregation is the self-association of molecules that occurs when strong van der Waals forces are present between them. In solutions, aggregation is observed typically only at high solute concentrations or when the molecules have low solubility. On metal oxide surfaces the abundance of metal ion binding sites results in close packing and nearly complete monolayer coverage of chromophores. Most chromophores are composed of large aromatic systems that, when fixed in proximity on a surface, can aggregate due to π–π stacking. Aggregation can strongly influence photophysical and electrochemical properties and ultimately improve or hinder device performance.

Several possible intermolecular interactions that can occur between chromophores (C) on a metal oxide surface (MO) and are summarized in eqn (5)–(8) with a vertical bar separating independent chromophores.

MO–C|C → MO–C*|C → MO(e⁻)–C⁺|C → MO–C|C (5)

MO–CC → MO–CC* → MO(e⁻)–CC⁺ → MO–CC (6)

MO–C|C → MO–CC* → MO(e⁻)–CC⁺ → MO–C|C (7)

MO–C|C → MO–C|C* → MO(e⁻)–CC⁺ → MO–C|C (8)

If molecules are in proximity but do not strongly interact (C|C in eqn (5)), a single chromophore will undergo excitation, electron injection, and BET as described in Section 1.2. During this process the excited state and cation remain localized on a single chromophore. Other molecules that exhibit strong interactions in the ground state will delocalize the absorption event and the resulting excited state will exist across two or more molecules (CC and CC* in eqn (6)). Alternatively, molecules that do not strongly interact in the ground state can, following excitation, form excimer states where the excited state is delocalized across two molecules (CC* in eqn (7)). Then following electron injection and BET, the chromophores dissociate into their independent ground states (C|C).

Although we have depicted a cation dimer (CC⁺) in eqn (6) and (7), the nature of this post-electron injection state is not well characterized likely due to the difficulty differentiating between transient cation dimers (CC⁺) and localized cations (C⁺|C). Along these lines, a fourth possibility (eqn (8)) is that following excitation and electron injection from a single chromophore, a cation dimer is formed with the cation delocalized across two chromophores. We found no reports identifying this species on a metal oxide surface. However, the generation of a stabilized cation dimer may be one strategy to increase the driving force for regeneration by the redox mediator as well as slow BET by pushing the energetics further into the Marcus inverted region.^32,33

Ground state aggregation (eqn (6)) is the most commonly reported intermolecular interaction at a metal oxide interface. This is most likely because it can readily be identified, even by visual inspection, due to a shift in absorption features. The two most common types of aggregates are H- and J-type which can readily be described by using Kasha's exciton theory (Fig. 3).³⁴

Fig. 3 Exciton band energy diagram for molecular aggregates with parallel and in-line transition dipoles.

Chromophore aggregation shifts and broadens the absorption features relative to the monomer due to interaction between transition dipoles of the molecules.³⁵ H-aggregates exhibit face-to-face stacking (parallel) with the out of phase and aligned dipoles resulting in lower and higher energy excited states, respectively. In contrast, J-aggregates are due to head to tail stacking (in-line) with the out of phase dipole alignment increasing the energy and the in phase dipole alignment decreasing the energy of the excited state. Photophysically, because only the aligned dipole transition is allowed, H- and J-aggregation manifests as a hypsochromic and bathochromic shift, respectively, relative to the monomer species.

The first report of aggregation on a high surface area metal oxide surface was by Tributsch in 1969 with pseudocyanine dyes on high roughness ZnO surfaces.³⁶ The ability of these dye aggregates to produce photocurrent was investigated through IPCE and time-resolved photocurrent measurements, with results suggesting photocurrent was produced via direct electron transfer between excited levels in the dye aggregate and the conduction band of the semiconductor.

Since these early studies, there have been hundreds of reports of dye aggregation on high surface area metal oxides. The general structure of several classes of dyes that tend to aggregate are shown in Fig. 4 including cyanines,³⁷ indolines,³⁸ porphyrins,³⁹ phthalocyanine,³⁹ chalcogens,⁴⁰ thiazines,⁴¹ and squaraines.⁴² These complexes all share the characteristics of being relatively planar, large aromatic systems that are susceptible to π–π stacking.

Fig. 4 General structure for dyes with propensity to aggregate on metal oxide surfaces.

In terms of solar energy conversion, a majority of dye aggregation hinders dye-sensitized device performance.^42–48 Low efficiency is often attributed to fast non-radiative decay from the excimer state (MO–CC* → MO–CC) prior to electron injection into the metal oxide (MO–CC* → MO(e⁻)–CC⁺). The decreased injection yield reduces the photocurrent and lowers the overall DSSC efficiency.

There are however examples where, by two different mechanisms, dye aggregation can improve the device performance. The first is by acting as an antenna-like species where aggregates absorb higher energy light, transfer that energy to a monomer which can then inject an electron into the metal oxide.⁴⁹ This is only beneficial when the quantum yield of energy transfer between aggregates and monomers is near unity. The second case involves aggregation induced broadening of absorption transitions and both the monomer and the aggregated state can inject electrons and contribute to photocurrent generation.^{37,38,41,49–56}

For example, Cai and coworkers reported that herring-bone aggregates of cyanine dyes on TiO₂ induced both red- and blue-shifted absorption features.⁵⁰ IPCE measurements indicate that both monomeric and aggregated forms contribute to charge separation and photocurrent generation. A maximum energy conversion efficiency of 2.9% was obtained with a cell whose photocurrent was generated by both the aggregate and monomer.

Recently, using the plant pigment betanin, McHale⁵⁷ demonstrated a remarkable 2.5 fold increase in power conversion efficiency for a DSSC composed of predominantly aggregated dyes relative to the monomer dominated device. The improvement is attributed to red- and blue-shifted absorption features for the aggregates, Fig. 5. Not only was the IPCE higher, so was the APCE, suggesting that improved photoconversion efficiency is not merely the result of increased light harvesting but is also due to more efficient electron injection from the aggregated dye compared to the monomer. The reported efficiency of 3% is the best yet achieved for a natural dye containing DSSC, which the authors believe can be further increased through changes in electrolyte composition, surface treatment, and other modifications.

Fig. 5 (a) Structure of betanin and (b) spectral deconvolution of the absorption spectrum of betanin on TiO₂. Adapted with alterations with permission from ref. 57. Copyright 2016, American Chemical Society.

Although dye aggregation on metal oxide surfaces has become somewhat ubiquitous and sometimes advantageous, systematic studies that couple spectroscopy and photoelectrochemistry to understand the role of aggregates on device performance are relatively limited. Perhaps one of the most extensively studied examples of aggregation at an interface is in cresyl violet dyes.^41,53,56

Using TA spectroscopy Liu et al. demonstrate that cresyl violet⁵⁶ and rhodamine 6G aggregates⁵⁵ on SnO₂ quickly undergo intersystem crossing and are capable of injecting electrons from the triplet excimer state with a rate constant of 2.0 × 10⁸ s⁻¹ and 5.5 × 10⁹ s⁻¹, respectively. Overlaying the IPCE with the absorptance spectra shows that electron injection from the aggregated dye is responsible for photocurrent generation, but the efficiency was still <1%. TA measurements indicate that, due to unfavorable kinetics for regeneration of the oxidized dye by iodide, back electron transfer was highly competitive with photocurrent generation.

Given the sometimes advantageous nature of dye aggregates on metal oxide surfaces, there are some efforts to control aggregate formation. Variations in solvent, dye concentration, and soaking times can be used to manipulate the relative amounts of monomer and aggregate species.⁵⁷ Treatments of metal oxide film prior to dye loading, such as hydroxylation or salt impregnation, can also influence the formation of aggregates.⁵⁸

Variations in dye structure can be a powerful tool in manipulating chromophore–chromophore interactions. For example, Mann et al.⁵⁴ controlled the aggregation of several chalcogenoxanthylium dyes on TiO₂ by varying the position of the surface-binding group and the identity of the chalcogen atoms (Fig. 6). The 1-E series of dyes readily formed H-aggregates upon binding to the surface whereas the 2-E series gave amorphous monolayers on the left and right in Fig. 6, respectively. The H-aggregated dyes exhibited broader absorption bands, increased light-harvesting efficiencies, and improved photoelectrochemical performance compared to the dyes which adsorbed in amorphous monolayers. The increased performance is partially attributed to a 2-to-3-fold increase of φ_inj for the aggregates, relative to the non-aggregated dye.⁴⁰

Fig. 6 Chemical structure of chalcogenoxanthylium dyes from Mann et al., left 1-E and right 2-E. Adapted with alteration with permission from ref. 54. Copyright 2009, American Chemical Society.

While some aggregates are advantageous, a majority of them are non-productive. Because of this, a number of strategies have been developed to inhibit aggregation including co-adsorption of electro- and photochemically inert molecules like CDCA or CHENO and others listed in Section 3.1 that act as a spacer between chromophores to prevent aggregation.^59–68

Unfortunately, due to the limited surface area, co-adsorbents can also decrease the concentration of dyes and reduce total absorption (see Section 3.1.3).⁶⁹ Alternatively, aggregation can be hindered by synthetically modifying the dye itself such that π–π stacking between molecules is inhibited with bulky ligands^63,70–76 or alkyl side chains.^75,77–79 Close packing between dyes can also be hindered by engineering dyes with unique shapes that prevent aggregation including H^80–82 and starburst shaped molecules.⁸³

While much is known about dye aggregation on high surface area metal oxides, a majority of reports simply note that aggregation is occurring and co-adsorbents are used to inhibit it. To effectively utilize aggregates, more systematic studies are needed in order to understand their influence on electron injection, BET, and ultimately device performance. Complementing spectroscopic, electrochemical, and device measurements with theoretical modeling will no doubt be crucial to this effort. Ab initio modeling of a more than 200 atom system is computationally expensive and therefore theoretical calculations of aggregates on surfaces has been relatively limited.^{42,57,69,72,73} Only within the last 5–10 years have more in depth studies been used to identify reasons for aggregation and how it affects optical and electronic properties of dyes on metal oxide surfaces.^{49,70,84–89} Future increases in computational power will allow for prediction of not only the aggregate formation, but also their influence on absorption and subsequent electron transfer events.

2.2 Multi-photon/electron processes

In contrast to the largely non-productive nature of aggregates/excimers discussed in the previous section, over the last 5–10 years there has been an increased interest in using intermolecular interactions on metal oxide surfaces to facilitate productive processes like molecular singlet fission (SF)⁹⁰ and photon upconversion via triplet–triplet annihilation (TTA-UC).⁹¹ Binding SF and TTA-UC molecules to a metal oxide surface is particularly appealing because it can act as a charge separation interface. Additionally, the interface offers a means of controlling the proximity and orientation of molecules which are critical factors dictating the rate and efficiency of these processes. Many of these reports study photophysical events on metal oxides such as nanocrystalline ZrO₂, whose relatively high conduction band energy inhibits excited state electron transfer to the substrate.⁹² More recently, with the goal of solar energy conversion in mind, TiO₂ has been employed as the charge separation interface to extract charge from these multiphoton and multielectron processes.

2.2.1 Singlet fission. In SF, a high energy photon is absorbed by a chromophore resulting in a singlet excited state (S₁). If a second chromophore of the appropriate energetics is in proximity then, through a spin-allowed process, the singlet excited state can be split into two triplet states (T₁) on two adjacent molecules. Systems analyses indicate that if these triplet states undergo charge separation in a device, then two charge carriers can be harvested per absorbed photon and theoretical solar conversion efficiencies can be increased from 33%⁹³ to upwards of 43%.⁹⁴ The phenomena of SF was first observed more than 60 years ago,^95,96 and has since been extensively investigated in molecular crystals, crystalline/amorphous films, and even in molecular dimers in solution.^94,97–102

In 2015, Johnson and coworkers published the first evidence of SF on a metal oxide surface. In their report the authors physisorbed 1,3-diphenylisobenzofuran, a dye known to undergo SF with 100% efficiency in crystals,¹⁰³ onto a metal oxide surface (Fig. 7b).¹⁰⁴ The introduction of a ZrO₂ “insulating layer” of increasing thickness between the dye and TiO₂, first decreases (<10 Å) and then increases the photocurrent response (10–15 Å) from the dye. This atypical behavior was rationalized as the ZrO₂ layer decreasing the rate of electron injection from the singlet state of the dye to TiO₂via tunnelling through the ZrO₂ layer. But, with a 10–15 Å insulating layer, they reached a balance where injection from the singlet state (k^inj_S in Fig. 7a) is sufficiently decreased such that singlet fission (k_SF) and electron injection from two long lived triplet states (k^inj_T) can occur, effectively increasing the photocurrent generated.

Fig. 7 (a) Energy level diagram and interfacial charge transfer events for an isobenzofuran-sensitized metal oxide (k_SF = rate of singlet fission, k^inj_S = rate of singlet injection, k^inj_T = rate of triplet injection, DPIBF = 1,3-diphenylisobenzofuran). Adapted with alterations with permission from ref. 104. Copyright 2015, American Chemical Society, and (b) a TiO₂–ZrO₂ core–shell metal oxide functionalized with isobenzofuran undergoing singlet fission (SF) and subsequent electron injection.

Simulations based on TA data and thickness dependent electron injection kinetics are used to support this mechanism but unfortunately the electron injection yields (electrons generated per photons absorbed) were still less than 100%. Similar measurements using a 1,3-diphenylisobenzofuran derivative with a hexoic acid surface-binding functional group were promising but less conclusive. The authors suggest that with other dyes, more rigid surface binding groups, or by manipulating the deposition of a mixture of species on the surface it may be possible to increase the rate and efficiency of SF at an interface and generate a viable SF DSSC. As a side note, it is important to mention that to realize the theoretical efficiency increase to 43%, the SF dye must be supplemented by photocurrent generation from an additional low energy absorbing dye, making multi-chromophoric assembly strategies (Section 4) even more crucial to this endeavour.

2.2.2 Triplet–triplet annihilation upconversion. In what can loosely be thought of as the inverse of SF, photon upconversion via triplet–triplet annihilation (TTA-UC) is the process of combining two lower energy triplet excited states into one higher energy, singlet excited state. TTA-UC typically requires two molecules: a low energy absorbing sensitizer and a triplet acceptor molecule, also known as an emitter or annihilator.⁹¹ During TTA-UC, a sensitizer molecule absorbs low energy light, undergoes intersystem crossing which is followed by sensitizer-to-acceptor, intermolecular triplet energy transfer (TET). Two triplet acceptor molecules in proximity can then undergo TTA-UC to generate one ground state and one singlet excited state. Emission from the resulting singlet excited state is hypsochromically shifted relative to the excitation light and thus the photon energy is upconverted during the process. While there are many potential applications for TTA-UC,⁹¹ its use in solar energy conversion is particularly intriguing because it offers a means of harnessing sub-band gap photons, increasing the maximum theoretical solar cell efficiencies from 33% to upwards of 48%.^105–107

TTA-UC was first demonstrated in the mid 1960s,¹⁰⁸ but the field largely laid dormant for several decades. A complete rebirth of interest in TTA-UC over the past decade has been fuelled by the development of new, efficient sensitizer molecules, and new strategies for combining TTA molecular pairs.^91,109,110 Of particular interest for application in solar energy conversion is the deposition of TTA-UC molecular pairs on metal oxide substrates.

To date, three different strategies have been used to achieve TTA-UC on a metal oxide surface and each are depicted in Fig. 8. The first example was reported by Morandeira and coworkers in 2011, where a green-to-blue TTA-UC efficiency of 6 × 10⁻⁴% was achieved by co-adsorbing the sensitizer and acceptor molecules onto a ZrO₂ substrate (Fig. 8a).¹¹¹ Chemically linking the dyes to ZrO₂via carboxylate groups resulted in a nearly four-fold increase in the UC efficiency (2 × 10⁻³%).¹¹²

Fig. 8 Schematic depiction of (a) co-deposited, (b) solution sensitized, and (c) self-assembled bilayer strategies for facilitating TTA-UC on a metal oxide substrate.

Although the TTA-UC efficiency was relatively low, these reports demonstrated that the close proximity of molecules on a metal oxide could be used to facilitate TET and TTA-UC. The low efficiency can be attributed, at least in part, to co-deposition effectively diluting the acceptor molecule concentration and limiting cross-surface triplet density and energy transfer rates. To remedy this shortcoming, the authors introduced a heterogeneous TTA-UC system with the acceptor molecule bound to the surface and the sensitizer dissolved in solution (Fig. 8b).¹¹² The significant increase in TTA-UC efficiency (0.04%) was attributed to increased acceptor concentration density on the surface. Interestingly, the efficiency limiting process of the heterogeneous scheme was diffusion limited sensitizer-to-acceptor TET and/or non-productive sensitizer–sensitizer TTA.¹¹³

Expanding upon this work, Nattestad and coworkers incorporated the heterogeneous TTA-UC scheme into a DSSC. Using TiO₂ as the substrate and with the addition of a cobalt based redox mediator into the solution, they observed photocurrent of up to 0.1 μA cm⁻² under 560–640 nm illumination at ambient solar intensities.¹¹⁴ Transient absorption and excitation intensity dependent measurements support a TTA-UC photocurrent generation mechanism, making this the first example of an integrated TTA-UC solar cell that directly extracts charge from an upconverted state.

The third strategy, introduced by Hill et al. in 2015, is through self-assembly of molecular multilayers via metal ion linkages (Fig. 8c).¹¹⁵ The multilayer films are formed using a simple, step-wise soaking procedure.^116–118 Briefly, a monolayer of acceptor molecules is bound to the metal oxide (ZrO₂ or TiO₂) via phosphonate groups. Then Zn²⁺ metal ions are coordinated to a terminal functional phosphonate groups of the acceptor molecules. Finally, a sensitizer layer is bound to the Zn²⁺ ions via carboxylate groups.

These self-assembled bilayer films are an effective strategy to facilitate TTA-UC emission (QY = 0.23%) and transient photocurrent generation.¹¹⁹ Self-assembled bilayers on TiO₂ were incorporated into a DSSC and a sustained photocurrent of 0.009 mA cm⁻² was observed under low energy light (485–550 nm) at solar intensities.¹²⁰ The increased photocurrent relative to the heterogeneous scheme was attributed to having a higher sensitizer concentration.

The incorporation of a second, complementary sensitizer in a trilayer architecture (Fig. 9), increased photocurrent to upwards of 0.074 mA cm⁻², while the system reached maximum TTA-UC efficiency below solar intensities (AM1.5, 100 mW cm⁻²).¹²¹ Efforts demonstrated that the choice of redox mediator in these TTA-UC solar cells has a profound influence on their performance with photocurrents as high as 0.158 mA cm⁻² under one Sun irradiation with maximum TTA-UC efficiency achieved at excitation intensities well below solar irradiance (0.8 mW cm⁻²).¹²² The photocurrent response is particularly notable in that it surpasses the device relevant threshold as outlined by Schmidt and coworkers.¹²³

Fig. 9 Self-assembled bilayer and trilayer TTA-UC DSSC. Reproduced from ref. 121 with permission from the Royal Society of Chemistry.

Recently, Dilbeck et al. quantified rates of energy and electron transfer in TTA-UC self-assembled bilayer films using a combination of time-resolved emission and TA spectroscopy.¹²⁴ Although sensitizer–sensitizer TTA can occur and is recognized as a non-productive process with high concentrations of sensitizer in solution, it was found that this process was negligible in the bilayer films presumably due to the limited mobility of the sensitizer in the bilayer architecture. A relatively slow sensitizer to acceptor triplet energy transfer rate limits the energy transfer efficiency and, as a result, the overall TTA-UC efficiency. However, electron injection from the singlet acceptor is ultra-fast and occurs with near unity quantum efficiency. These results indicate that designing new TTA-UC bilayers containing upconversion pairs with increased sensitizer to acceptor energy transfer rates, decreased sensitizer and acceptor non-radiative rates, and increased TTA rates is a key step towards realizing efficient TTA-UC photocurrent generation and improving solar energy conversion efficiencies.

2.3 Cross surface energy and electron migration

In addition to aggregation/excimer formation (Section 2.1) and multi-photon/electron processes (Section 2.2), the densely-packed nature of dyes on metal oxides can effectively facilitate efficient energy and electron transfer events between the same or different dye molecules on the surface. Interest in quantifying the rates of these migration events and increasing our ability to harness them continues to grow because of applications like photon upconversion and photocatalysis that require multiple excited states or the build-up of multiple redox equivalents at a single site at the interface. Below we will discuss the current methodology used to quantify these energy (Section 2.3.1) and electron (Section 2.3.2) transfer events as well as the variables that influence these processes.

2.3.1 Cross-surface energy transfer. One of the first experimental observations of cross-surface energy transfer in dyes bound to nanocrystalline metal oxide surfaces was published by Gerald J. Meyer and coworkers in 1999.¹²⁵ The excited state decay kinetics for a monolayer of Ru(deeb)(bpy)₂²⁺ (deeb = 2,2′-bipyridine-4,4′-diethylester) bound to nanocrystalline TiO₂ in MeCN (i.e. minimal electron injection in the absence of Li⁺) was strongly dependent on the intensity of the excitation light. The complex kinetics were attributed to a mixture of unimolecular and bimolecular excited-state decay of Ru(deeb)(bpy)₂²⁺via radiative/non-radiative decay and presumably TTA, respectively (Fig. 10). At the highest irradiances studied, only 10–20% of the surface bound molecules were excited, making the likelihood of having two adjacent excited state molecules low. Thus, the authors concluded that cross-surface energy migration must be responsible for the bimolecular event.

Fig. 10 Cross-surface energy transfer and TTA of Ru(deeb)(bpy)₂²⁺ on TiO₂. Adapted with permission from ref. 125. Copyright 1999 American Chemical Society.

Later in the same year, the Meyer group used a co-sensitization of ruthenium and osmium polypyridyl dyes to directly demonstrate cross-surface energy transfer (Fig. 11).¹²⁶ In their films, it is energetically favorable for Ru* to transfer energy to the Os dye which effectively acts as an emissive trap. Selective excitation of the Os dye resulted in emission characteristic of the Os complex only. When exciting both Ru and Os dyes equally, in the absence of energy transfer, the Ru emission should be 80 times higher than Os. However, in the co-deposited film, Ru emission was significantly quenched (Γ_Ru : Γ_Os, 0.3 : 1) with near unity efficiency suggesting a Ru to Os energy transfer rate of >10⁸ s⁻¹. Although both inter- and intramolecular energy transfer were well known in solutions^127–129 and polymers^130–132 of Ru and Os dyes, this was the first time this phenomenon was observed on a metal oxide surface.

Fig. 11 Energy transfer events for a co-deposited film composed of Ru^II and Os^II polypyridyl dyes on a metal oxide surface. Adapted with permission from ref. 133. Copyright 2001 American Chemical Society.

Expanding on this effort, Thomas J. Meyer and coworkers used time-resolved emission in conjunction with Monte Carlo simulations to model cross-surface energy transfer between [Ru(bpy)₂(4,4′-(PO₃H₂)₂-bpy)]²⁺ and [Os(bpy)₂(4,4′-(COOH)₂-bpy)]²⁺ on nanocrystalline ZrO₂.¹³³ The excited state lifetime of Ru* on ZrO₂ was dependent on several factors including the Ru surface coverage, concentration of Os quencher, and monitoring wavelength. Above 70% Ru loading (i.e. above the percolation threshold) Ru*-to-Ru energy migration (k_mig ∼ 10⁷ s⁻¹) was followed by Ru*-to-Os energy transfer (k_en ∼ 10⁶ s⁻¹). The rate constant for energy transfer was exponentially dependent on the distance between dyes, consistent with a dominant role for a Dexter energy transfer mechanism.

To confirm the proposed mechanism from their previous work, the Gerald J. Meyer group revisited their Ru only and co-deposited Ru and Os TiO₂ films using the time-resolved emission and Monte Carlo modelling (MCS) methodology.¹³⁴ For a fully loaded film, they estimated a rate constant of (30 ns)⁻¹ for isoenergetic Ru* → Ru energy transfer across the nanocrystalline surface. Monte Carlo simulations indicated that due to its relatively long lifetime (τ = 1 μs) a Ru* excited state could transfer energy to any of the ∼800 sensitizers on an ∼20 nm anatase nanoparticle surface.

In 2011, the Gerald J. Meyer group introduced an alternative technique, time-resolved spectroscopic anisotropy, as a means to directly probe cross-surface energy transfer on nanocrystalline metal oxides.¹³⁵ Briefly, a fraction of the surface bound molecules with the appropriate dipole orientation are excited using plane polarized light and polarized absorption/emission are then monitored both parallel and perpendicular to the incident light. Since the surface bound molecules cannot undergo rotation, any change in the intensity of polarized absorption/emission is indicative of excited state migration across the spherical particle's surface. The anisotropy data for a fully loaded film in conjunction with the Monte Carlo simulation showed a nearest-neighbor hopping rate constant of (120 ns)⁻¹. In contrast to previous co-deposited films, this technique allows researchers to monitor energy migration with a film containing only one dye molecule, mitigating any concerns with the emissive trap inhibiting percolation or disturbing surface packing.

While cross-surface energy transfer is likely a common phenomenon in dye-sensitized devices, it has only been investigated on metal oxide surfaces a handful of times. This is partially due to the difficulty in measuring the isoenergetic energy transfer. However, as shown above, the emergence of the transient anisotropy and time-resolved emission combined with Monte Carlo simulations has made these analyses more straightforward. Understanding the extent to which cross-surface energy transfer occurs will likely become important in maximizing the efficiency of DSSCs and certainly is important when implementing phenomena like triplet–triplet annihilation upconversion on metal oxide surfaces (Section 2.2).

2.3.2 Cross-surface electron/hole transfer. Understanding cross-surface electron and hole transfer is becoming increasingly important due to emerging technologies like dye-sensitized photoelectrosynthesis cells (DSPECs) and solid-state DSSCs (ssDSSCs). In DSPECs, multiple holes/electrons must diffuse through a dye monolayer to a single catalytic site at a rate that is competitive with recombination events.^136–138 In ssDSSCs, cross surface hole diffusion is necessary for the oxidized species to make direct contact with the hole transporting material.^139,140 Dyes with good hole transport properties may even be used as the hole transport layer themselves.¹⁴¹

Unlike energy transfer, where emissive species can readily be observed using time-resolved spectroscopy, cross-surface electron transfer is a “dark” event. Transient absorption can be used to monitor electron/hole hopping,^{135,142–144} but it is much more commonly measured using electrochemical techniques like cyclic voltammetry (CV)¹⁴⁵ and chronoabsorptometry (CA).¹⁴⁶ For these measurements, the dye functionalized nanocrystalline metal oxide is situated on a planar conductive substrate, usually fluorine doped tin oxide (FTO) coated glass. At sufficiently positive applied potentials, electrons are transferred from the immediately adjacent dyes to the conductive substrate (Fig. 12, top). If the dye surface loading is above the percolation threshold, then electrons diffuse through the molecular monolayer on the nanocrystalline metal oxide surface towards the underlying conductive substrate (Fig. 12, middle), and the process is repeated (Fig. 12, bottom).¹⁴⁷

Fig. 12 Mechanism of cross-surface electron transfer in a monolayer of dye bound to a MO on a conductive substrate under positive applied potential (+V).

The rate of this charge migration is most commonly quantified using the apparent diffusion coefficient (D_app) as described in eqn (9) and (10).

where D_phys is the physical displacement of the redox centers, δ is the distance from the electrode, C_E is the concentration of the redox active species, and k_et is the bimolecular rate constant for electron self-exchange.¹⁴⁸ For surface bound molecules the physical displacement can be assumed to be zero which simplifies eqn (9) to eqn (10).

Experimentally, D_app is typically determined by monitoring the changes in current by CV or, for electrochromic molecules, monitoring changes in absorption with respect to time under applied potential using eqn (11) and (12) respectively:

where j_p is the peak current density of the cyclic voltammogram, q is the electron elementary charge, k_B is the Boltzmann constant, T is temperature, c₀ is the dye loading expressed as volume concentration and ν is the scan rate.where ΔA is the change in absorbance (A_max − A_(t)), t is the time in seconds, and d is the thickness of the film.

When reporting/interpreting these values, care must be taken in noting the measurement technique since D_app can vary by up to an order of magnitude for the same film measured by different methods.^149–151 A summary of the apparent diffusion coefficients reported for a number of metal oxide bound dyes in various solvents as measured by CV and/or CA are shown in Table 1.

Table 1 A summary of D_app values for a variety of dyes, surface coverages, and solvents measured by cyclic voltammetry (CV) or chronoabsorptometry (CA). Molecules are listed in the table with the labels used in the corresponding reference but complete names for each molecule can be found in the footnote

Entry	Molecule	Γ	Solvent	D app (cm^2 s^−1)	CA/CV	Ref.
a Sodium(3-(4-(bis(4-methoxyphenyl)amino)phenoxy)propyl)phosphonate. b 1-Ethyl-2-methylimidazolium bis(trifluoromethylsulfonyl)imide. c 5-[[4-[4-(2,2-Diphenylethenyl)phenyl]-1,2,3,3a,4,8b-hexahydrocyclopent[b]indol-7-yl]methylene]-4-oxo-2-thioxo-3-thiazolidineacetic acid. d 2-Cyano-3-[4-[4-(2,2-diphenylethenyl)phenyl]-1,2,3,3a,4,8b-hexahydrocyclopent[b]indol-7-yl]-2-propenoic acid. e 5-[[4-[4-(2,2-Diphenylethenyl)phenyl]-1,2,3,3a,4,8b-hexahydrocyclopent[b]indol-7-yl]methylene]-2-(3-ethyl-4-oxo-2-thioxo-5-thiazolidinylidene)-4-oxo-3-thiazolidineacetic acid. f 5-[[4-[4-(2,2-Diphenylethenyl)phenyl]-1,2,3,3a,4,8b-hexahydrocyclopent[b]indol-7-yl]methylene]-2-(3-octyl-4-oxo-2-thioxo-5-thiazolidinylidene)-4-oxo-3-thiazolidineacetic acid. g (2,2′-Bipyridyl-4,4′-dicarboxylic acid)-bis(2,2′-bipyridyl)ruthenium(II). h Bis(2,2′-bipyridyl-4,4′-dicarboxylic acid)-(2,2′-bipyridyl)ruthenium(II). i Tris(2,2′-bipyridyl-4,4′-dicarboxylic acid)ruthenium(II). j (2,2′-Bipyridyl-4,4′-dimetyhl-dicarboxylic acid)-bis(2,2′-bipyridyl)ruthenium(II). k Bis(2,2′-bipyridyl-4,4′-dimethyl-dicarboxylic acid)-(2,2′-bipyridyl)ruthenium(II). l Tris(2,2′-bipyridyl-4,4′-dimethyl-dicarboxylic acid)ruthenium(II). m cis-Di(thiocyanato)(2,2′-bipyridyl-4,4′-dicarboxylic acid)-(2,2′-bipyridyl-4,4′-tridecyl)ruthenium(II). n cis-Di(thiocyanato)(2,2′-bipyridyl-4,4′-dicarboxylic acid)-(2,2′-bipyridyl-4,4′-bis(vinyltriphenylamine))ruthenium(II). o Bis(isothiocyanato)(4,4′-dicarboxylato-2,2′-bipyridine)(4,4′-dinonyl-2,2′-bipyridine)ruthenium(II). p cis-Di(thiocyanato)-(2,2′-bipyridyl-4,4′-dicarboxylic acid)(4,4′-dimethyl-2,2′-bipyridyl)-ruthenium(II). q Zinc phthalocyanine-tri-tert-butyl-2-carboxylic acid. r Ruthenium phthalocyanine-tri-tert-butyl-2-carboxylic acid. s RuPcA2; see reference for structure. t RuPcA5; see reference for structure. u 4,4′-[(4-[5′-([N-(2,6-Diisopropylphenyl)]-9-perylenyl-3,4-dicarboxyimide)-3,4′-dihexyl-2,2′-bithien-5-yl]phenyl)imino]dibenzoate 4,4′-[(4-[5′-([N-(2,6-diisopropylphenyl)]-9-perylenyl-3,4-dicarboxyimide)-3,4′-dihexyl-2,2′-bithien-5-yl]phenyl)imino]dibenzoate. v cis-Dithiocyanato(4,4′-dicarboxy-2,2′-bipyridine)-(4,4′-dinonyl-2,2′-bipyridyl)Ru(II). w cis-Dithiocyanato(4,4′-dicarboxy-2,2′-bipyridyl)-(4,4′-dimethyl-2,2′-bipyridyl)Ru(II). x cis-Dicyanato(4,4′-dicarboxy-2,2′-bipyridyl)-(4,4′-dimethyl-2,2′-bipyridine)Ru(II). y (4,4′-Dicarboxy-2,2′-bipyridyl)-bis(4,4′-dimethyl-2,2′-bipyridyl)Ru(II). z Bis(4,4′-dicarboxy-2,2′-bipyridyl)-(4,4′-ditridecyl-2,2′-bipyridyl)Ru(II). aa cis-Dithiocyanato-4,4′-bis[p-carboxylic acid]-2,2′:6′,2′′:6′′,2′′′-quaterpyridine Ru(II). ab (2,2′-Bipyridyl-4,4′-dicarboxylic acid)-bis(4,4′-di-tert-butyl-2,2′-bipyridy)ruthenium(II).
1	1^a	3.3 × 10^−10 mol cm^−2	Diethylene glycol dimethyl ether	2.6 × 10^−8	CA	147
2	1^a	3.3 × 10^−10 mol cm^−2	EtMeIm^+Tf2N^−^b	2.8 × 10^−8	CA	147
3	1^a	3.3 × 10^−10 mol cm^−2	EtMeIm^+Tf2N^− : 4-methyl-2-pentanone (1 : 1)	5.0 × 10^−8	CA	147
4	1^a	3.3 × 10^−10 mol cm^−2	EtMeIm^+Tf2N^− : 3-methoxypropionitrile (1 : 1)	9.1 × 10^−8	CA	147
5	1^a	3.3 × 10^−10 mol cm^−2	EtMeIm^+Tf2N^− : MeCN (1 : 1)	11 × 10^−8	CA	147
6	D102^c	3.0 × 10^−8 mol cm^−2	0.1 M NBu4PF6/MeCN	≥5 × 10^−9	CV	155
7	D131^d	4.3 × 10^−8 mol cm^−2	0.1 M NBu4PF6/MeCN	≥5 × 10^−9	CV	155
8	D149^e	2.0 × 10^−8 mol cm^−2	0.1 M NBu4PF6/MeCN	≥5 × 10^−9	CV	155
9	D205^f	3.5 × 10^−8 mol cm^−2	0.1 M NBu4PF6/MeCN	≥5 × 10^−9	CV	155
10	Ru(bpy)2(Ad-tripod-phen)^2+	3.1 × 10^−8 mol cm^−2	0.1 M TBAPF6/MeCN	∼10^−11	CA	157
11	Ru(bpy)2(Ad-tripod-bpy)^2+	3.9 × 10^−8 mol cm^−2	0.1 M TBAPF6/MeCN	∼10^−11	CA	157
12	Ru(bpy)2(C-tripod-phen)^2+	4.2 × 10^−8 mol cm^−2	0.1 M TBAPF6/MeCN	∼10^−11	CA	157
13	Ru(bpy)2(C-tripod-bpy)^2+	2.0 × 10^−8 mol cm^−2	0.1 M TBAPF6/MeCN	∼10^−11	CA	157
14	1^g	2.2 × 10^−8 mol cm^−2	Aqueous 0.1 M HClO4	13.3 × 10^−10	CA	146
15	2^h	2.4 × 10^−8 mol cm^−2	Aqueous 0.1 M HClO4	3.6 × 10^−10	CA	146
16	3^i	2.4 × 10^−8 mol cm^−2	Aqueous 0.1 M HClO4	2.7 × 10^−10	CA	146
17	1c^j	3.7 × 10^−8 mol cm^−2	Aqueous 0.1 M HClO4	12.1 × 10^−10	CA	146
18	2c^k	3.7 × 10^−8 mol cm^−2	Aqueous 0.1 M HClO4	2.9 × 10^−10	CA	146
19	3c^l	3.7 × 10^−8 mol cm^−2	Aqueous 0.1 M HClO4	0.6 × 10^−10	CA	146
20	N621^m	∼70%	0.1 M TBAP/MeCN	1.2 × 10^−9	CA	156
21	HW456^n	∼60%	0.1 M TBAP/MeCN	2.6 × 10^−8	CA	156
22	Z-907^o	1.5 × 10^−4 mol cm^−3	0.1 M TBAP/MeCN	2.0 × 10^−8	CV	145
23	N820^p	1.6 × 10^−4 mol cm^−3	0.1 M TBAP/MeCN	3.5 × 10^−8	CV	145
24	D131^d	1.5 × 10^−4 mol cm^−3	0.1 M TBAP/MeCN	2.8 × 10^−8	CV	145
25	D149^e	1.1 × 10^−4 mol cm^−3	0.1 M TBAP/MeCN	27 × 10^−8	CV	145
26	TT-1^q	2.2 × 10^−4 mol cm^−3	0.1 M TBAP/MeCN	0.39 × 10^−8	CV	145
27	TT-35^r	7.3 × 10^−5 mol cm^−3	0.1 M TBAP/MeCN	3.9 × 10^−8	CV	145
28	A2^s	1.2 × 10^−4 mol cm^−3	0.1 M TBAP/MeCN	1.7 × 10^−8	CV	145
29	A5^t	4.5 × 10^−5 mol cm^−3	0.1 M TBAP/MeCN	1.3 × 10^−8	CV	145
30	C60-Diels–Alder monocarboxylic acid	500 mM	50 : 50 EC : DMC and 1 M lithium imide/MeCN	6.5 × 10^−9	CV	153
31	Bis(2,2′-bipyridine)(4,4′-diphosphonato-2,2′-bipyridine)ruthenium(II)	9 × 10^−8 mol cm^−2	0.1 M LiClO4/MeCN	11 × 10^−10	CA	138
32	[Os^II(bpy)2(4,4′-(CO2H)2bpy)](PF6)2 loaded in EtOH	7 × 10^−11 mol cm^−2	0.1 M TBAH/MeCN	1.4 × 10^−11	CA	158
33	[Os^II(bpy)2(4,4′-(CO2H)2bpy)](PF6)2 loaded in MeCN	1.1 × 10^−10 mol cm^−2	0.1 M TBAH/MeCN	1.4 × 10^−9	CA	158
34	PMI-T2-tPA^u	1.5 × 10^−10 mol cm^−2	0.1 M EMITFSI/MeCN	2.2 × 10^−8	CV	159
35	1^v	—	0.1 M EMITFSI/MeCN	4.1 × 10^−9	CV	149
36	2^w	—	0.1 M EMITFSI/MeCN	11.4 × 10^−9	CV	149
37	3^x	—	0.1 M EMITFSI/MeCN	1.9 × 10^−9	CV	149
38	5^y	—	0.1 M EMITFSI/MeCN	0.3 × 10^−9	CV	149
39	6^z	—	0.1 M EMITFSI/MeCN	0.02 × 10^−9	CV	149
40	7^aa	—	0.1 M EMITFSI/MeCN	0.5 × 10^−9	CV	149
41	bpy^g	1.9 × 10^−7 mol cm^−2	0.1 M LiClO4/MeCN	2.2 × 10^−9	CA	151
42	bpy^g	1.9 × 10^−7 mol cm^−2	0.1 M LiClO4/MeCN	1.1 × 10^−10	CV	151
43	dmb^y	9.5 × 10^−8 mol cm^−2	0.1 M LiClO4/CH3CN	5.3 × 10^−9	CA	151
44	dmb^y	9.5 × 10^−8 mol cm^−2	0.1 M LiClO4/CH3CN	3.1 × 10^−10	CV	151
45	dtb^ab	7.7 × 10^−8 mol cm^−2	0.1 M LiClO4/CH3CN	2.4 × 10^−10	CA	151
46	dtb^ab	7.7 × 10^−8 mol cm^−2	0.1 M LiClO4/CH3CN	1.4 × 10^−11	CV	151

---

As can be seen in Table 1, D_app can vary by several orders of magnitude from 2.7 × 10⁻⁷ cm² s⁻¹ for D149 on TiO₂ in 0.1 M TBAP in acetonitrile at room temperature measured using CV¹⁴⁵ to 1.4 × 10⁻¹¹ for [Ru(4,4′-di-tert-butyl-2,2′-bipyridine)₂(dcbH₂)](PF₆)₂ on TiO₂ in 0.1 M LiClO₄ in acetonitrile at room temperature measured with CV.¹⁵¹ In addition to the dye structure, film porosity, and nanoparticle size,¹⁵¹D_app is dependent on temperature,¹⁵² polarity and pH of the solvent,^{144,147,153,154} molecular packing and distance between dyes,^151,155,156 and, for some dyes, metal ions present in solution.¹⁵⁶

For example, in a study by Li et al.¹⁶⁰ (Table 1, entries 20 and 21), the percolation dynamics (measured with CA) of a ruthenium complex (N621) are compared with those observed for an analogous dye with an additional triphenyl amine (TPA) donor moiety (HW456), whose structures can be seen in Fig. 13. The increased electronic coupling between the TPA donor moieties made the rate of hole transfer 20 times faster in a monolayer of HW456 compared to a monolayer of N621. With the introduction of mercuric ions into the solvent, hole percolation was quenched for N621 films and enhanced for HW456 due to Hg²⁺ coordination to the thiocyanate moiety on both dyes. This suggests that the hole percolation pathway is along the overlapped thiocyanate groups for the N621 molecules, whereas in HW456 molecules cation percolation proceeds between intermolecular TPA ligands.

Fig. 13 Molecules from Li et al. showing the electron transfer pathways for (a) N621 and (b) HW456.¹⁶⁰

In another systematic study (entries 41–46 in Table 1), DiMarco et al. investigated the self-exchange intermolecular Ru^III/II electron transfer for three homologous Ru diimine compounds of the general form [Ru(LL)₂(dcbH₂)](PF₆)₂, where LL is 2,2′-bipyridine (bpy), 4,4′-dimethyl-2,2′-bipyridine (dmb), or 4,4′-di-tert-butyl-2,2′-bipyridine (dtb) and dcbH₂ is 2,2′-bipyridyl-4,4′-dicarboxylic acid, in order to probe how the electron transfer rate varies as the distance and aliphatic bulk between the Ru centers is increased (bpy < dmb < dtb).¹⁵¹D_app was determined using both CV and CA and the same trend of dtb < bpy < dmb was found with both techniques, but the value for the latter was an order of magnitude lower than the former. The authors suggest that the large difference in D_app for these two techniques could lie in the assumption that the semi-infinite diffusion boundary approximation holds true with both techniques.¹⁴⁵ That is, the analysis is only valid if the oxidized molecules do not reach the outer edge of the film, an assumption that holds true in CV measurements where only about 5% of the film is oxidized but not necessarily in CA measurements where almost all of the film is oxidized by the end of the measurement.

For molecules where distinct absorption spectral shifts accompany changes in redox state, the rate of electron/hole diffusion through a monolayer film can also be measured using a combination of TA spectroscopy and Monte Carlo simulations. Similar to the experiments described above to determine energy transfer rates (Section 2.3.1) electron migration and site-to-site hopping rates can be determined for single component films using TA anisotropy¹³⁵ or for a co-deposited film containing the dye of interest and a second molecule that acts as a redox sink.^135,142,143 In comparison to electrochemical techniques, spectroscopic determinations of electron migration do not suffer from the boundary limitations described above for D_app and also, in conjunction with Monte Carlo simulations, allow for the determination of site-to-site electron transfer rates rather than just a bulk diffusion constant.

As an aside, it is worth noting that there have been considerable advances in modeling cross-surface electron transfer processes using ab initio computational methods. More details can be found in the review article recently published by Pastore et al.¹⁶¹

In summary, it is known but perhaps underappreciated that cross-surface electron and hole transfer through dye monolayers on high surface area electrodes can play an important role in emerging dye–semiconductor technologies. Though a powerful experimental approach, D_app is dependent on numerous variables and can vary by an order of magnitude depending on the measurement/analysis methodology. A combination of TA and MCS may give a more detailed description of the electron hopping events but require additional computational efforts as well as expensive, time consuming, and often difficult spectroscopic measurements. Overall, it is generally safe to suggest that in order to directly compare redox migration events at interfaces with respect to solvent, molecular structure, surface loading, etc., the samples should be measured under the same set of conditions and with the same technique.

3. Interfacial electron transfer

Molecules bound to a metal oxide surface is a geometrically and spatially controllable scaffolding to study and manipulate electron transfer dynamics at interfaces. These electron transfer events can take place between the surface/dye and an electroactive molecule in solution, or between dyes and the surface, and between two molecules on the surface. Regardless of if it occurs in solution or at an interface, electron transfer dynamics can generally be described using Marcus electron transfer theory (eqn (13)).¹ The physical parameters that dictate the electron transfer rate (k_ET) between a donor (D) and acceptor (A) are the Gibbs free energy change (ΔG°), reorganization energy (λ), temperature (T), and electronic coupling between the donor and acceptor molecules (V_DA).

|V_DA|² = |V₀|² exp(−βR_DA) (14)

The electronic coupling term can loosely be thought of as the height and width of the electron tunneling barrier between the donor and acceptor. As shown in eqn (14), the coupling at a given distance (V_DA) is dictated by the coupling at close contact (V₀), the attenuation coefficient (η), and perhaps most importantly for the discussion below, the distance (R_DA) between D and A. Given that there is an exponential relationship between V_DA² and R_DA, and in turn k_ET, the distance between D and A can have a profound influence on electron transfer rates and ultimately device performance. Therefore, modifications in the molecular structure at the interface can effectively be used to facilitate or hinder electron transfer by decreasing or increasing R_DA, respectively.

Below we summarize efforts at manipulating electron transfer events at molecule–metal oxide interfaces using two different types of multimolecular interactions: (1) co-deposition of a dye and an electrochemically inactive “insulator” molecule (Section 3.1), and (2) binding chromophores and donors/acceptors (Section 3.2) to a metal oxide surface. We also discuss how these interactions can be utilized to access unique electron transfer processes like supersensitization (Section 3.3).

3.1 Co-adsorbents

3.1.1 Introduction to co-adsorbents. In Section 2.1 we describe the use of electro- and photochemically inert co-adsorbent molecules as a means of inhibiting aggregation and increasing electron injection yields.^59–68 These inert co-adsorbents have also become a common method for slowing recombination in DSSCs^162–171 with the proposed mechanism shown in Fig. 14.

Fig. 14 The proposed recombination process at the dye–metal oxide interface in the (a) absence and (b) presence of co-adsorbent molecules.

Under optimized loading conditions, dye molecules form a closely packed, nearly complete monolayer on the metal oxide surface. Despite having high surface loadings there are still exposed portions of TiO₂, either in the spaces between molecules or in vacancy defects in the monolayer packing, that are in direct contact with the solvent. These open sites allow for diffusional contact (small R_DA) between the oxidized redox mediator and electrons in TiO₂ (Fig. 14a) which can result in recombination. The deposition of co-adsorbents in the space between dye molecules not only hinders aggregation, but also effectively inhibits the redox mediator from approaching the surface (large R_DA) which slows recombination (Fig. 14b).

In 1978, Miyasaka et al. published one of the earliest reports of using inert co-adsorbents on metal oxide surfaces.¹⁷² They demonstrated that, despite having lower dye loading, tin oxide films functionalized with chlorophyll dye and stearic acid as an inert co-adsorbent exhibited both increased photocurrent and photovoltage compared to films with the dye alone. Despite these promising results, the use of inert co-adsorbents in photoelectrochemical cells laid largely dormant until the introduction of high surface area DSSCs by O’Regan and Gratzel in 1991.¹⁷³ Since then, there has been a sharp increase in use of inert co-adsorbents in DSSCs. A majority of these studies utilize co-adsorbents to prevent dimer/excimer formation in large planar aromatic dyes like porphyrins, coumarins, carbazole, indoline, squaraine and phthalocyanines^{60,61,63,67,68,174–178} which are susceptible to aggregation through π–π stacking. Most ruthenium polypyridyl dyes, with the exception of black dye,⁵⁹ do not readily aggregate on the surface and thus co-adsorbents are largely used for slowing recombination at the interface.^162–171

3.1.2 Classes of inert co-adsorbents. We have somewhat arbitrarily classified the most common inert co-adsorbents into five different classes with select molecules shown in Fig. 15.

Fig. 15 Molecular structure of the most common electro- and photochemically inert co-adsorbents. The surface binding moiety is highlighted in red.

A majority of co-adsorbents are composed of sp³ carbon rings/chains or benzene rings, presumably because they lack absorption in the visible wavelength region, and are electrochemically inactive under operational potentials of dye-sensitized devices. These molecules are typically anchored to the surface via carboxylic acids,^61–65 phospho acids,^{163,167,179–181} or ester groups.^{171,180,182–184} Both carboxylic and phospho acids bind through covalent linkages.^{167,182,184,185} In contrast, ester binding is largely assumed to be through hydrogen bonding.^180,182 Phosphonic acids are known to have the strongest binding affinity for the surface and thus offer increased DSSC stability,^{25,167,179–181} however there is currently no definitive evidence that any of these binding groups are superior in terms of device efficiencies.

Of these co-adsorbents, cholic acids, in particular deoxycholic acid (DCA) and chenodeoxycholic acid (CDCA), are by far the most commonly used presumably because (1) their steroidal skeleton offers greater steric bulk compared to alkyl chains, (2) they are commercially available by-products found in the bile of mammals,¹⁸⁶ and (3) they have a strong history of significantly improving device performance relative to dye only systems.¹⁸⁷

An interesting and distinct type of co-adsorbents are zwitterion molecules.^164,169,188 This unique class of molecules allows researchers to tune both the steric and electrostatic properties at the dye interface which directly influences recombination dynamics.

A majority of the co-adsorbents studied to date have relied on commercially available carboxylate and phosphonate functionalized molecules. This is presumably at least partially due to a disconnect between synthetic chemists and researchers studying devices. However, there have been efforts to synthesize novel co-adsorbents with specific design principles in mind.^{167,168,181,188–190}

3.1.3 Co-adsorbent loading strategies. There are a number of strategies to generate metal oxide films functionalized with both dye and co-adsorbent molecules (Fig. 16). By far the most common is a co-loading procedure where the metal oxide film is soaked in a solution containing both dye and co-adsorbent (Fig. 16C).^{59–63,65–68,162,167,168} An alternative strategy is to first fully load the metal oxide surface with co-adsorbent molecules and then soak the passivated film in the dye solution (Fig. 16B).¹⁷⁴ Conversely, a film fully loaded with dye can be soaked in co-adsorbent (Fig. 16A).^168,191 Finally, some researchers incorporate the co-adsorbent into the electrolyte solution during DSSC assembly effectively forming the co-adsorbed film in situ (Fig. 16D).^192–195

Fig. 16 Common loading procedures for dyes (dark blue) and co-adsorbents (gray) on metal oxide substrates (M = mediator).

The final dye/co-adsorbent loading ratio for all four of these strategies is strongly dependent on competitive binding between the two species which is dictated by binding affinities of molecules and concentrations of loading solutions. One common trend, regardless of film preparation strategy, is a decrease in total dye loading in co-adsorbent films, relative to the dye only films, presumably due to limited surface area and competitive desorption. Since the primary goal of these solar cells is to maximize light absorption, special care must be taken to optimize loading conditions to maintain high dye concentrations while still reaping the benefits of co-adsorbent molecules. As far as we can tell, current research efforts rely largely on trial and error to determine the appropriate set of conditions for a given dye/co-adsorbent.^{61,162,190,196}

One of the fundamental limitations hindering the progress in controlling co-adsorbent loading is the difficulty in quantifying surface loading ratios. The lack of absorption and electrochemical signal for the co-adsorbent limits the techniques available to quantify their surface coverage. At least a few reports indicate that co-adsorbent concentrations can be determined using photoelectron spectroscopy (PES),^167,185 and scanning tunneling microscopy (STM)¹⁹⁷ but have been somewhat limited in their use.

In one study, Marinado et al. used PES measurements to quantify the relative concentration of atoms specific to the dye molecules (e.g. S2p from thiophene) and the co-adsorbents (e.g. P2p from phosphorus binding groups).¹⁸⁵ They found that with the same concentration (0.05 mM of co-adsorbent to 0.2 mM D29 dye) of two different co-adsorbents, decylphosphonic acid (DPA) and bis(3,3-dimethyl-butyl)phosphonic acid (DINHOP) (Fig. 15), dye loading was decreased by 53% and 25% respectively. The dye/co-adsorbent ratio also varied between 1 : 3.5 for DPA and 3 : 1 for DINHOP. The authors also investigated CDCA as a co-adsorbent but quantification of surface loading by PES was difficult due to a lack of unique atoms in CDCA. Regardless the authors clearly demonstrate the utility of PES in quantifying co-adsorbent concentrations and that the structure, size, and affinity of the co-adsorbent play important roles in dictating the surface coverages and loading ratios. Additionally, while most researchers only report the concentrations in the loading solutions, this work demonstrates that the dye/co-adsorbent ratio can vary significantly between solution and the surface.

3.1.4 Co-adsorbents influence on J_sc, V_oc and FF. In DSSCs the primary goal of using these co-adsorbents is to improve device performance. As such the most common strategy to evaluate their success is to measure device performance with and without co-adsorbent loading. Depending on the dye, the presence of co-adsorbents can have a profound influence on J_sc, V_oc, FF, and/or efficiency of the devices. The highest improvement in device performance with co-adsorbents was reported by Wu et al. where they observe a more than two fold increase in J_sc (8.46 mA cm⁻² to 17.5 mA cm⁻²) and a 8% increase in V_oc (601 mV to 651 mV) which combine to give a >200% improvement in efficiency (3.86% to 8.56% PCE).¹⁷¹ Some additional, notable device improvements are summarized in Table 2.

Table 2 Performance characteristics of DSSCs with and without co-adsorbent molecules

Dye	Co-adsorbent		J sc (mA cm^−2)	V oc (mV)	Fill factor	PCE (%)	Ref.
The percent change in the presence of co-adsorbents are noted in parentheses.a 2-Cyano-3-[5-[4-(9-pentyl-9H-carbazol-3-yl)-1,2,3,3a,4,8b-hexahydrocyclopent[b]indol-7-yl]ethenyl-2-thienyl]-2-propenoic acid.b Ruthenium(II) cis-dithiocyanato(4,4′-dicarboxy-2,2′-bipyridine)(1,10-phenanthroline).c 1-[(5-Carboxy-1,3-dihydro-3,3-dimethyl-1-octyl-2H-indol-2-ylidene)methyl]-3-[(1-ethyl-1,3dihydro-3,3-dimethyl-2H-indol-2-ylidene)methyl]-2,4-dihydroxy-cyclobutenediylium.d Zinc(II) 2-carboxy-9(10),16(17),23(24)-tris-[2-(N,N-bis-(9,9-dimethylfluoren-2-yl)-4-aminophenyl)-5,5′-bithiophene]phthalocyanine.e Zinc(II) [10-[2-[10,20-bis[3,5-bis(1,1-dimethylethyl)phenyl]-15-[bis(4-octylphenyl)amino]-porphin-5-yl]ethynyl]-9-anthracenecarboxylato].f Zinc(II) 4,4′-[10,20-bis[4-(dimethylamino)phenyl]-porphine-5,15-diyl]bis-benzoic acid.
C-CA^a	DCA	w/o	14.16	628	0.66	5.87	206
		w	18.53 (31%)	649 (3%)	0.71 (8%)	8.49 (45%)
AR20^b	Citric acid	w/o	9.34	610	0.485	2.8	166
		w	14.69 (57%)	650 (7%)	0.634 (31%)	6.1 (118%)
SQ1^c	CDCA	w/o	8.02	509	0.69	2.82	203
		w	10.23 (28%)	595 (17%)	0.69 (0%)	4.23 (50%)
ZnPc 1^d	CDCA	w/o	3.26	503	0.68	1.11	65
		w	5.25 (61%)	541 (8%)	0.73 (7%)	2.07 (86%)
YD 13^e	CDCA	w/o	4.22	631	0.71	1.89	207
		w	6.27 (49%)	652 (3%)	0.71 (0%)	2.9 (53%)
P2^f	CDCA	w/o	11.1	660	0.66	4.83	208
		w	12.4 (12%)	720 (9%)	0.68 (3%)	6.07 (26%)

---

From device performance alone it is difficult to determine the role of co-adsorbents in preventing aggregation, inhibiting recombination, or other effects because of the intertwined relationship between J_sc and V_oc.^{179,198–201} As discussed in Section 2.1 aggregation can decrease electron injection efficiency due to competitive, fast non-radiative decay from the excimer state. In a device, decreased injection yield will manifest as lower J_sc and lower overall DSSC efficiency.^{65,175,184,199,202} Co-adsorbents can be used to prevent dye aggregation but interestingly, a majority of concentration dependent studies indicate that with increasing co-adsorbent there is an initial increase in J_sc followed by a decrease.^{59,177,179,201,203} The initial increase in J_sc is often attributed to an increase in electron injection yield, but it is also important to note that red/blue shifts in the absorption spectrum and changes in extinction coefficient upon co-adsorbent addition may also play a role in influencing the absorption overlap with the solar spectrum.^204,205

An eventual decrease in J_sc at high co-adsorbent concentration is due to a decrease in dye loading and a lower light harvesting efficiency of the film. Because of this increase-decrease response curve, optimal device performance and co-adsorbent loading conditions are often determined empirically through concentration dependent studies. In a number of studies, lower J_sc have been obtained at all concentrations of the co-adsorbents. The decreased J_sc and minimal change in the absorption spectra are indicators that the dye molecules do not aggregate on the surface.^209,210 For these systems, typically containing ruthenium dyes such as N719, N749 and Z-907,^{164,169,190,201} co-adsorbents are introduced as a means of slowing recombination and increasing V_oc.^167,180 Since V_oc can be improved by a number of factors including negative conduction band edge shifts,¹⁸⁰ higher J_sc, and slower recombination,¹⁷⁸ it is important to quantify the role of recombination using techniques like electrochemical impedance spectroscopy (EIS) and transient photovoltage measurements.^{164,169,190,201} For example, Shen et al. found, via EIS, that a decrease in the number of protons from PPA (2H⁺) to DPPA (1H⁺) to PATEE (0H⁺) was accompanied by an increase in V_oc due to a negative shift in the conduction band energy level of TiO₂.¹⁸⁰ A similar trend was reported by Ren et al. where a 26 mV positive shift and a 33 mV negative shift in conduction band energy was observed for deoxycholic acid (DCA) and sodium deoxycholate, (DCNa) respectively (Fig. 17)¹⁸⁴ Neale et al. used infrared transmittance measurements with respect to tetrabutylammonium chenodeoxycholate (TBACDC) loading, to differentiate the role of recombination and conduction band energy shifts on the device V_oc.¹⁸³ Interestingly, the increased V_oc in the presence of TBACDC was due to a negative shift in conduction band energy and not to slower recombination. In fact, counter to the expectation shown in Fig. 17, there was a fivefold increase in recombination rate in the presence of co-adsorbent.

Fig. 17 Influence of co-adsorbents on the conduction band energies, recombination rates, and over all V_oc. Reprinted with permission from ref. 183. Copyright 2005, American Chemical Society.

3.2 Chromophore–donor/acceptor assemblies

Electron transfer between a chromophore and donor or acceptor molecules has been extensively investigated in solution either as solvated molecules^211,212 or with chemically linked chromophore–donor/acceptor assemblies.^213–218 These assemblies are particularly appealing because the distance is fixed by the molecular structure and not dictated by the concentration of the solution, random molecular orientations, and diffusion rates. Similar structural control can be achieved at interfaces using several different strategies including covalent linkages (Section 3.2.1), co-deposition (Section 3.2.2), and supramolecular assemblies (Section 3.2.3) of chromophore and donor/acceptor moieties. The metal oxide surface adds an additional variable in that it can also act as an electron acceptor, donor, or insulator.

3.2.1 Covalent chromophore–donor/acceptor systems. The most common method for adhering a chromophore and donor/acceptor to a surface is by first synthesizing a covalently linked dyad (C–D or C–A) and then binding it to a metal oxide. Several examples of common chromophores and donors used in C–D complexes can be seen in Fig. 18. This class of compounds can technically still be classified as single molecules but we have included them in the discussion here because the chromophore and donor/acceptor moieties are, for the most part, photo- and electrochemically insulated from each other in the ground state. That is, the absorption spectrum and electrochemical traces are simply the sum of its parts with no new transitions or major energetic shifts observed. This is in contrast to D–π–A, or push–pull chromophores where the donor and acceptor portions of the molecule are strongly coupled and exhibit new absorption transitions. D–π–A molecules have been extensively reviewed elsewhere and will not be discussed here.^219–223

Fig. 18 Chromophores (left), donors (middle), and acceptors (right) used in covalent D–C/A dyads.

3.2.1.1 Chromophore–donor/acceptor dyads on n-type MO. A majority of these covalent dyad systems contain a donor, rather than acceptor moiety presumably due to their targeted use as TiO₂ photoanodes in DSSCs. The typical stepwise events for electron transfer events between donor (D), chromophore (C), and a n-type metal oxide (MO) are summarized in eqn (15)–(19).

MO–C–D → MO–C*–D (15)

MO–C*–D → MO(e⁻)–C⁺–D (16)

MO(e⁻)–C⁺–D → MO(e⁻)–C–D⁺ (17)

MO(e⁻)–C–D⁺ → MO–C–D (18)

MO(e⁻)–C⁺–D → MO–C–D (19)

Upon excitation, the chromophore (C* in eqn (15)) has sufficient excited state potential to transfer an electron to the conduction band of MO generating the chromophore cation (C⁺ in eqn (16)). Electron transfer from D to the oxidized dye regenerates the ground state chromophore (C in eqn (17)). MO(e⁻) can then either enter the external circuit or recombine with D⁺ by back electron transfer returning the system back to its initial state (eqn (18)). Additional competitive processes like MO(e⁻) to C⁺ back electron transfer can also occur (eqn (19)).

In terms of dye-sensitized devices, the primary goal of the MO–C–D architecture is to slow recombination dynamics (eqn (18) and (19)) by reducing the electronic coupling between MO(e⁻) and the cation species. This is typically achieved through a directional electron transfer cascade where the spatial separation, R_DA, in MO(e⁻)–C–D⁺ is greater than for MO(e⁻)–C⁺. The general design strategy for MO–C–D is to select the components such that the oxidation potential of the donor moiety is more negative, i.e. more easily oxidized, than the chromophore to ensure that electron transfer from C* to MO, and from D to C⁺, is favorable.

One of the first examples of covalent MO–C–D assemblies was introduced by Gerald J. Meyer and coworkers.²²⁴ The complex, Ru(4-(CO₂⁻)-4′-(CO₂H)-2,2′-bipyridine)₂(4-CH₃-4′-CH₂-PTZ,-2,2′-bipyridine), is composed of a ruthenium chromophore linked, via methyl spacer, to a phenothiazine (PTZ) donor group. Electron transfer from PTZ to Ru(III) is thermodynamically favorable by ∼360 mV. The charge separated state, TiO₂(e⁻)–Ru^II–PTZ⁺ as observed by TA spectroscopy, is generated in less than 30 ns and recombines with a rate constant of 3.6 × 10³ s⁻¹. Electron–hole recombination is slowed by a factor of ∼2000 compared to the complex without PTZ (TiO₂(e⁻)–Ru^III). When incorporated into a DSSC, there was a 100 mV increase in V_oc which was attributed to its slower recombination dynamics relative to the complex without donor.

Since then, a number of MO–C–D systems have been examined (Fig. 18). These complexes are typically composed of a ruthenium polypyridyl chromophore and tri-aryl and/or alkyl amine donor moiety.^{221,225–229} Presumably, they are selected as the archetypical C and D because of their well understood and well behaved single-electron reversible electrochemistry.^230,231 There are several other compounds that utilize zinc porphyrin,^221,226,232 copper phenanthroline,²³³ organic D–π–A,²³⁴ and ruthenium pyrazole/pyrrole²³⁵ complexes as the chromophore. A majority of reports utilizing C–D molecules on surfaces focus on studying the influence of the donor group on DSSC performance. As reported in the Meyer system described above, they often observe increase in V_oc which is attributed to the slowed electron–hole recombination MO(e⁻)–C–D⁺. Rarely, but occasionally, this conclusion is supported by TA measurements.^{227,236–239} As expected from Marcus theory, the MO(e⁻) to D⁺ recombination rate decreases exponentially with the distance between the surface and donor group.²³⁹ It is interesting to note that for these devices, J_sc was usually either unchanged or decreased upon the addition of the donor group. The decrease in J_sc is often attributed to bulkier donor moieties reducing surface loading, resulting in a lower absorption cross-section in the film.²⁴⁰

A particularly interesting subset of these compounds are those containing multiple donor moieties.^{233,239–245} Compared to parent molecules without donor groups, these complexes exhibit similar changes in photophysical and device characteristics as the single donor systems (i.e. longer lived charge separation, increased V_oc, etc.).^246,247 In a relatively unique study, Song et al. made a comparison between Ru(4-(CO₂⁻)-4′-(CO₂H)-2,2′-bipyridine)(4-R₁-4′-R₂-2,2′-pipyridine) (NCS)₂ chromophores substituted with one (R₁ = CH₃, R₂ = CH₂NCH₃Ph) or two (R₁, R₂ = CH₂NCH₃Ph) dimethyl aniline donor groups. In DSSCs, the compound with two donor groups had an approximately 5% greater J_sc, V_oc and FF with a ∼20% greater overall efficiency compared to the singly substituted complex.²⁴⁶ Later TA measurements suggest improved performance was due to an increased D to C⁺ electron transfer rate for the double (k = 3.4 × 10⁷ s⁻¹) versus single (k = 2.6 × 10⁷ s⁻¹) donor system.²⁴⁷ These results indicate that although the MO–C–D compounds only undergo single photon-single electron transfer events, the rate/probability of these events increases with more donor moieties.

The above discussion focuses on TiO₂–C–D assemblies. Analogous but less studied are the metal oxide–chromophore–acceptor assemblies that are of interest for application in photochromic devices, optical read–write–reset devices, and others. For example, Will et al. synthesized a D–A complex composed of a ruthenium tris-bipyridyl chromophore linked to a methylviologen acceptor group by a methyl spacer.^248,249 After loading the complex on TiO₂ they were able to demonstrate directional, potential controlled electron transfer (Fig. 19).

Fig. 19 Intramolecular electron transfer of RV on TiO₂ with applied bias. Adapted with alterations with permission from ref. 248. Copyright 1998 American Chemical Society.

Under open circuit conditions or a positive applied potential, electron transfer from C* to MO was significantly faster than from C* to A resulting in a 95% electron injection yield. In contrast, under a negative bias, the TiO₂ conduction band energy is shifted and the electron injection rate is sufficiently slowed such that 52% of C excited states transfer electrons to the methylviologen moiety. The reduced methylviologen exhibits a color change that effectively demonstrates a MO–C–A assembly can be used as a photo-electrochromic switching material.

3.2.1.2 Chromophore–donor/acceptor dyad on p-type MO. An increasingly popular use of C–A or D–C dyads is for lengthening the charge separation lifetime in the photocathodes of p-type and tandem DSSCs (Section 4.3).^250,251 In contrast to the electron accepting nature of n-type materials (i.e. TiO₂ and SnO₂), p-type semiconductors, like NiO, act as an electron donor, effectively inverting the typical DSSC architecture.

Electron transfer events between A, C, and a p-type metal oxide (pMO) in pMO–C–A assemblies are summarized in eqn (20)–(24).

pMO–C–A → pMO–C*–A (20)

pMO–C*–A → pMO(h⁺)–C⁻–A (21)

pMO(h⁺)–C⁻–A → pMO(h⁺)–C–A⁻ (22)

pMO(h⁺)–C⁻–A → pMO–C–A (23)

pMO(h⁺)–C–A⁻ → pMO–C–A (24)

In this scheme, the excited chromophore (eqn (20)) is reduced via electron transfer from the metal oxide, also known as hole injection (eqn (21)). An electron is then transferred from C⁻ to A, affording the pMO(h⁺)–C–A⁻ charge separated state (eqn (22)). The cation/hole can either diffuse through the pMO and enter the external circuit, or recombine as in eqn (23) or (24) (Fig. 20).

Fig. 20 Electron transfer dynamics of PMI and PMI-NDI dyad on NiO. Reprinted with permission ref. 429. Copyright 2009 American Chemical Society.

Odobel and coworkers later generated a squaraine dye–peryleneimide–naphthalenediimide triad.²⁵² In contrast to the dyad, the triad exhibited a charge separation mechanism where excitation of the squaraine was followed by electron transfer to the perylene unit, and then subsequent perylene to naphthalene and NiO to squaraine electron transfer events. In an I⁻/I₃⁻ based p-type DSSC the SQ-PMI-NI triad exhibited a more than three fold higher J_sc but a >30% reduction in V_oc relative to the PMI-NI dyad.

Ji et al. also relied on naphthalenediimide as the acceptor unit but with a ruthenium terpyridine complex as the chromophore.²⁵³ On NiO the charge separated NiO(+)–C–A⁻ was generated in 1.1 ps but interestingly, the back electron transfer rate was only nominally slower in the dyad (426 ps) than the chromophore only film (371 ps).

An alternative strategy to increase the charge separation lifetime on NiO is by using a donor–chromophore dyad effectively generating a pMO–D–C assembly. Proposed electron transfer events between D, C, and pMO are summarized in eqn (25)–(29).

pMO–D–C → pMO–D–C* (25)

pMO–D–C* → pMO–D⁺–C⁻ (26)

pMO–D⁺–C⁻ → pMO(h⁺)–D–C⁻ (27)

pMO(h⁺)–D–C⁻ → pMO–D–C (28)

pMO–D⁺–C⁻ → pMO–D–C (29)

The number of examples of D–C dyads where each component is electronically unique is relatively limited. In 2013, Ji et al. published a dyad that was composed of ruthenium(II) with two 2,2′-bipyridine or 1,10-phenanthroline ligands and one 2-phenylpyridine cyclometallating ligand functionalized with a di(carboxylic acid)-triphenylamine (TPA) surface binding moiety (Fig. 21).²⁵⁴ UV-vis, electrochemistry, DFT calculations, and IPCE measurements indicate that the ruthenium chromophore was largely independent of the TPA group.

Fig. 21 pMO–D–C assembly and proposed electron transfer events by Wu and coworkers. Adapted with alterations with permission from ref. 254. Copyright 2013 American Chemical Society.

In a p-type DSSC, the dyad had a >5-fold increase in J_sc and an order of magnitude higher solar energy conversion efficiency (0.099%) compared to the same dye directly on the NiO surface (0.009%).²⁵⁵ The improved performance is attributed to a longer lived charge separated state, later supported by TA measurements.²⁵⁶ However, it is important to note that while the mechanism proposed in eqn (25)–(29) is feasible, the cation feature for TPA was not observed during TA measurements indicating that it either did not form, or more likely, that reduction of TPA⁺ by NiO is a faster event.

Odobel^257–259 and others^260–262 have generated additional dyad-like dyes for use on NiO and in p-type DSSCs, however since the components are linked via acetylene or aromatic bridges, it is not easy to distinguish C–A and D–C from D–π–A systems so they will not be discussed here.

3.2.2 Co-adsorption of chromophores with donor/acceptors. An alternative strategy to the pre-formed C–A or C–D assemblies is to co-adsorb both the dye and the donor or acceptor molecules directly on the metal oxide surface, herein denoted as MO–C|D or MO–C|A, respectively. In the covalent assemblies the D/A group is spatially separated/decoupled from the surface. In contrast, for the co-deposited system there is direct contact between the donor/acceptor molecule and the metal oxide which allows for competitive electron transfer events to/from the surface. The electron transfer mechanism is dependent on the nature of the electroactive molecule (electron donor or acceptor), chromophore, and metal oxide. Here the discussion will be partitioned into first MO–C|D and then MO–C|A systems.
3.2.2.1 Chromophores co-adsorbed with donors. Like the covalent C–D systems described above, the primary goal cited for co-depositing C and D on TiO₂ (TiO₂–C|D) is to increase the rate of regeneration and inhibit recombination in DSSCs. The observed electron transfer events, in the absence of mediator, can similarly be described using eqn (15)–(19) from Section 3.2.1. However, unlike the numerous reports of covalent C–D systems, reports of co-depositing C and D are largely limited to research efforts by the groups of Fitzmaurice^263–265 and Kim.^266,267

In 1999, Grätzel and coworkers published a thorough and systematic study comparing the electron transfer rates for a ruthenium(II)bis-terpyridine complex as the chromophore and substituted triphenylamine as the donor in both a covalent and co-deposited films (Fig. 22).²²⁷ For the co-deposited film, excitation and electron injection from C (eqn (15) and (16)) was followed by relatively fast (<5 ns) electron transfer from D to C⁺ (eqn (17)). Interestingly, despite having similar or faster electron transfer rates, the quantum yield for generating the charge separated state was near unity for the covalent dyad (TiO₂(e⁻)–C–D⁺) but only 60% for the codeposited films (TiO₂(e⁻)–C/D⁺). The films also exhibit dramatic differences in the lifetime of the charge separated state (eqn (17)) being 3 μs and 30 μs for the co-deposited and covalent systems respectively. The shorter lifetime for the co-deposited film was attributed to the close proximity between TiO₂(e⁻) and D and the well-known distance dependence on electron transfer rate as described in the introduction to this section. Despite this difference, charge recombination is four orders of magnitude slower than D to S⁺ electron transfer, thus the charge separation yield of TiO₂(e⁻)–C/D⁺ would still be expected to be near unity. This relatively low yield (60%) was hypothesized to be due to reductive quenching of S* by D (see Section 3.3, eqn (38)) prior to electron injection.

Fig. 22 (a) Coadsorbed ruthenium(II) bis-terpyridine complex with 3-(4-(N,N-di-p-anisylamino)phenoxy)-propyl-1-phosphonate donor and (b) covalent dyad structures and their respective electron transfer rates. Reprinted with permission from ref. 227. Copyright 1999 American Chemical Society.

In 2011 Kim and coworkers studied the influence of co-deposition of C and D on DSSC performance (Fig. 23a and d).²³⁹ Despite having a lower chromophore loading and decreased absorbance, the DSSC with an anode composed of co-deposited films with 4-[3,6-bis[4-(2-ethylhexyloxy)phenyl]-9H-carbazol-9-yl]benzoic acid (HC-A) as the donor and oligothiophene-containing coumarin (NKX2677) as the chromophore, exhibited an increased J_sc, V_oc, and a factor of 1.33 efficiency improvement over the NKX2677 only film. The improvement was attributed to several factors including (1) the co-adsorbent decreasing aggregation/π–π stacking of planar chromophores, (2) slowed charge recombination between injected electrons and the redox mediator,²⁶⁶ (3) at least some photocurrent generation from direct excitation of HC-A at 350 nm, and finally (4) HC-A acting as a hole conductor, effectively reducing the oxidized dyes and relaying holes to the electrolyte. Regarding the latter point, although the driving force for HC-A to NKX2677⁺ electron transfer is energetically unfavorable (360 mV), oxidized HC-A was observed using TA spectroscopy.

Fig. 23 Molecular structures of co-adsorbent (blue) (a) HC-A1, (b) HC-A′, (c) HC-A′′, and chromophore (black) (d) NKX2677 and (e) FP.²³⁹

In an extension of this work, Kim and coworkers synthesized two additional derivatives of HC-A, HC-A′ and HC-A′′, with varying degrees of π-conjugation (biphenyl and tetramethylbiphenyl) between the carbazole donor group and the surface binding carboxylate group (Fig. 23b and c).²⁶⁷ When co-deposited on TiO₂ with a porphyrin based dye, both derivatives were less effective at inhibiting charge recombination compared to the parent dye. The increase in current and voltage, relative to the chromophore only device, was attributed to the increased spectral absorption range, and the decrease of π–π stacking induced aggregation in NKX2677.

In an effort to increase the driving force of D to C⁺ electron transfer, Kim and coworkers then synthesized two new phenothiazine (PTZ) based donor molecules.²⁶⁸ Using TA spectroscopy they demonstrated that both PTZ donors reduce oxidized dye in ∼100 ns, and exhibit slower BET rate compared to using DCA and HC-A. Despite the slowed recombination and slightly higher photocurrent contribution, devices with PTZ donors had similar performance metrics as the HC-A device.

3.2.2.2 Chromophores co-adsorbed with acceptors. For C|A deposited on TiO₂, both A and TiO₂ can act as an electron acceptor and thus at least two electron transfer pathways can be envisioned and have been demonstrated. The events depicting excited state electron transfer from C to the metal oxide (eqn (30)–(32)) or the acceptor molecule (eqn (33)–(36)) are given below.

MO–C|A → MO–C*|A (30)

MO–C*|A → MO(e⁻)–C⁺|A (31)

MO(e⁻)–C⁺|A → MO–C⁺|A⁻ (32)

MO–C⁺|A⁻ → MO–C|A (33)

MO–C*|A → MO–C⁺|A⁻ (34)

MO–C⁺|A⁻ → MO(e⁻)–C⁺|A (35)

MO(e⁻)–C⁺|A → MO–C|A (36)

Upon excitation of C (eqn (30)), the electron can transfer directly to the metal oxide surface (eqn (31)), as with a standard DSSC, or to the adjacent acceptor molecule (eqn (34)). In the former case, the electron in the metal oxide can then transfer to a surface bound acceptor (eqn (32)) to generate a surface bound charge separated state (MO–C⁺|A⁻). In the latter case the reduced acceptor can transfer an electron to the metal oxide resulting in a MO(e⁻)–C⁺|A charge separated state (eqn (35)). Regardless of the nature of the charge separated state, in the absence of an external circuit or sacrificial donor, it will decay back to the original state via charge recombination (eqn (33) and (36)).

One of the earliest examples of C and A codeposited on nanocrystalline TiO₂ was reported by Fitzmaurice and coworkers. In their system they chose bis[(4,4′-dicarboxy-2,2′-bipyridine)(4,4′-dimethyl-2,2′-bipyridine)-ruthenium(II)] as the chromophore with a methyl viologen monomer²⁶³ or oligomer²⁶⁴ as the acceptor species (Fig. 24a). They observed a long lived (>15 minutes) TiO₂–C⁺|A⁻ charge separated state and proposed a step-wise mechanism proceeding via excitation of C (eqn (30)), electron transfer to the metal oxide (eqn (31)), and then metal oxide mediated electron transfer to the acceptor (eqn (32)). Later experiments, applying a bias to the TiO₂, lent support to this mechanism by demonstrating that electrochemically populating the CB of TiO₂ resulted in electron transfer from the metal oxide to the acceptor moiety (TiO₂(e⁻)–A → TiO₂–A⁻).²⁶⁵ In a similar TiO₂–C|A system, using Ru(bpy)₃²⁺ donor and MV²⁺ acceptor precipitated on TiO₂, Zhang et al. used EPR spectroscopy to monitor electron transport through the TiO₂ conduction band and further supported a metal oxide mediated charge separation process.²⁶⁹

Fig. 24 Electron transfer at the interface for co-deposited chromophores and acceptors via three different pathways (a) metal oxide mediated electron transfer from Ru(bpy)₃ to MV²⁺ (ref. 264) (b) alzarin electron transfer to MV²⁺ (ref. 271) and (c) zinc porphyrin to fullerene electron transfer and direct electron injection to TiO₂ (ref. 272).

In later work, Bignozzi and coworkers, using a ruthenium based chromophore and a mixed valence acceptor, established no charge separated state was observed on a ZrO₂ substrate due to the lack of both direct C to A electron transfer (eqn (34)) and electron injection into the metal oxide (eqn (31)).²⁷⁰ In contrast, on TiO₂, a TiO₂–C⁺|A⁻ state was observed whose lifetime (microseconds) was significantly increased (>40 hours) by depositing PMMA as an oxygen exclusion over layer. The authors concluded that the extremely long lifetime demonstrated that TiO₂–C|A films could be used as a self-erasable optical memory device with a lifetime controlled by oxygen diffusion through a polymeric protecting layer.

In contrast to the previous systems, Matylitsky et al. used an electrochemically inert Al₂O₃ substrate to show that intermolecular electron transfer between an excited alizarin chromophore and methyl viologen acceptor occurred in ∼100 fs (eqn (34)).²⁷¹ On TiO₂ (Fig. 24b), electron injection from C* (eqn (31)) was competitive (occurred in 100 fs) with direct charge separation. Interestingly, TA measurements showed no evidence for TiO₂(e⁻) to acceptor electron transfer (eqn (32)) because recombination dynamics of injected electrons were similar for both TiO₂–C and TiO₂–C|A films.

Most recently, Stranius et al. utilized this direct C* to A electron transfer mechanism to generate a directional electron transfer cascade from a porphyrin chromophore, to a C₆₀ acceptor, and finally to the TiO₂ substrate (Fig. 24c).²⁷² The relatively long surface binding functional group on the porphyrin significantly slowed the recombination event (eqn (36)) and increased the charge separated lifetime to upwards of 20 ms. Presumably this type of TiO₂–C|A architecture could be used in a DSSC to decrease recombination (increase V_oc) while still maintaining a high electron injection yield (high J_sc).

Through the above work, it is clear that anchoring donor/acceptor moieties onto a semiconductor surfaces is a simple but effective means of manipulating/controlling electron transfer events at an interface. This can be achieved by either introducing an additional charge separation mechanism or facilitating an electron transfer cascade that is not present in the chromophore only systems. The mechanism and rate of these processes is dependent on variables including the energy level of the conduction band, ground and excited state potentials of the chromophore, and redox properties of the donor/acceptor. One under mentioned variable in these studies is D/A loadings which will directly influence electron transfer rates between co-deposited species simply by influencing the distance between D/A and C. As mentioned above with co-adsorbents in Section 3.1, the lack of visible absorption features for the D/A molecules inhibits direct quantification on TiO₂. However, desorption/quantification of molecules and then correlation of performance with loading ratios as well as orientation information about the C and D/A species will be a necessary next step in establishing design principles to fully utilize co-deposited donor/acceptor species for applications in membrane design, energy storage, optical memories and photovoltaics.

3.2.3 Supramolecular assemblies of C and D/A. A third strategy for generating C–D/A on metal oxide surfaces is through supramolecular assemblies. Instead of using covalent bonds (Section 3.2.1) or co-deposition (Section 3.2.2), these assemblies rely on hydrogen bonds, π–π interaction, electrostatics, and/or coordination bonds between C and D and/or A, one of which is typically bound directly to the surface.²⁷³ In terms of solar cells, these supramolecular assemblies on nanocrystalline metal oxides are most commonly described as a hybrid between a bulk-heterojunction solar cell (BHJSC) and a DSSC, and are sometimes called a dye-sensitized bulk-heterojunction solar cell (DSBHJSC).²⁷⁴ The hybrid description for these cells is partially due to the films exhibiting two possible charge separation mechanisms: (1) electron transfer from/to C* to/from A/D in the molecular assembly (eqn (34) from Section 3.2.2) like that observed in a BHJSC, and (2) electron transfer from/to C* to/from the metal oxide substrate (eqn (16) from Section 3.2.1) like in DSSCs. Additionally, the electron/hole transport medium is not necessarily supported by the nano-assembly as in BHJSCs, but instead is through the high surface area nanocrystalline semiconducting substrate. The most studied supramolecular assemblies on metal oxide films are composed of porphyrin–fullerene pairs as C and A, respectively. These molecules are chosen for their favorable interactions that facilitate the formation of supramolecular assemblies,²⁷³ and their high electron transfer efficiencies in part due to small reorganization energies.²⁷⁵ Reviews published by Imahori,^276,277 Hasobe^273,278 and Fukuzumi et al.²⁷⁹ extensively discuss the research progress for porphyrin–fullerene assemblies in solution, on planar surfaces, and on nanocrystalline metal oxides prior to 2010. Given the depth of these reviews we will only briefly touch upon previous work and highlight more recent progress since.

There are two common approaches for generating porphyrin–fullerene assemblies on metal oxides: (1) pre-assembled structures that are then deposited on a surface, or (2) in situ supramolecular assembly on the surface (Fig. 25).

Fig. 25 Examples of (a) pre-formed²⁸⁰ and (b) in situ²⁸⁹ assemblies on metal oxide surfaces.

The first method offers direct control over the C and A/D distances and has been achieved using dendrimers,^280–283 peptides oligomers,^284,285 dyads,^286,287 and polymers.²⁸⁸ The latter method does not require pre-assembly, but instead dissolves both molecular species in solution, and through non-covalent intramolecular interactions, supramolecular assemblies are formed/deposited directly on the surface.^289–295

For the most part DSBHJSCs composed of C–A assemblies generate more photocurrent than the sum of devices made with C and A only.^{284,289,292,296,297} Photocurrent enhancements are attributed to facile charge separation, that is photoinduced electron transfer occurs from the singlet excited state of porphyrin (−0.7 to −1.0 V) to C₆₀ (−0.4 to −0.2 V) which typically occurs within less than 2 ps.²⁹⁷ However, it is difficult to deconvolute other contributing factors such as regeneration/recombination events and additional absorption and photocurrent generation by the fullerene acceptor molecule.^280,296

In contrast, Ooyama et al. recently demonstrated that porphyrin dimer–C₆₀ supramolecular assemblies (Fig. 26) can also inhibit device performance.²⁹⁸ In this example, the incorporation of C₆₀ to surface-bound porphyrin dimers (H₄–C₄–CDP_py(TEO) and H₄–Ptz–CPDPy(TEO)) significantly reduced device performance because generation of the charge separated H₄–C₄–CDP_py(TEO)⁺–C₆₀⁻ and Ptz–CPD_py(TEO)⁺–C₆₀⁻ states, and subsequent recombination, were competitive with direct injection from the porphyrin which significantly decreased the J_sc.

Fig. 26 Structure of the Ooyama system (left: H4-C4-CDPpy(TEO), right: H4-Ptz-CPDPy(TEO)). Reprinted from ref. 298 with permission from the Royal Society of Chemistry.

Regardless of specific C and D/A, these supramolecular assemblies share many of the same drawbacks including poor intercalation of large assemblies into pores of the nanocrystalline electrodes²⁹⁹ and limited control over directionality of electron transfer,³⁰⁰ both of which can hinder device performance.

A relatively new, alternative supramolecular strategy, briefly discussed in Section 2.2 and seen again in Sections 4.2 and 5, is to generate self-assembled multilayers via metal ion linkages.¹¹⁵ In contrast to the strategies above, this step-wise assembly procedure generates supramolecular structures from the metal oxide-up in a layer-by-layer fashion. Therefore, it is not limited by diffusion into the pores and the order of molecular deposition can be used to directionally control electron transfer.

Thomas J. Meyer and coworkers recently used this strategy to generate self-assembled trilayers of A–C–D or D–C–A on nanocrystalline indium tin oxide (ITO) electrodes using Zr⁴⁺ linking ions (Fig. 27).³⁰¹ The films effectively behave as molecular analogues of semiconductor p/n junctions where, following excitation of C, a A⁻–C–D⁺ or D⁺–C–A⁻ charge separated state is observed presumably via a stepwise C* to A and D to C⁺ electron transfer pathway. Under an applied potential of 0 V vs. Ag/AgCl, the films exhibited remarkably slow recombination times of 5.6 s and 26 ms for ITO–D⁺–C–A⁻ and ITO–A⁻–C–D⁺ respectively.

Fig. 27 Layer-by-layer molecular p/n junction assemblies on mesoporous nanoITO electrodes. Reprinted with permission from ref. 301. Copyright 2016 Nature Publishing Group.

In a follow-up to this work Shan et al. generated a D–C bilayer on nanocrystalline NiO for use as photocathodes in DSSCs.³⁰² Upon excitation of the chromophore (NiO–D–C*), the films exhibited vectorial electron transfer away from the surface in a stepwise manner by first generating NiO–D⁺–C⁻ and then NiO⁺–D–C⁻ in 15% yield under open circuit conditions. The lifetime of the charge separated state was on the tens of millisecond time scale, several orders of magnitude longer than for C directly on the NiO surface.

More recently, Lim et al. demonstrated a chromophore–donor assembly at TiO₂, interface via axial coordination of aluminum–porphyrin with bis(p-anisole)-aminopyridine donor (Fig. 28).³⁰³ Electron injection into TiO₂ from the surface bound C* followed by electron transfer from D to C⁺ took place in 0.7 ps and 56 ps, respectively. Interestingly, electron injection is 1 ps faster with the TiO₂–C–D assembly compared to reference TiO₂–C without the donor.

Fig. 28 Electron transfer events of TiO₂ surface bound aluminum porphyrin axial coordinated to anisolamonipyridine electron donor. Reprinted with permission from ref. 303. Copyright 2017 American Chemical Society.

3.3 Reduced sensitizer injection

In addition to facilitating the formation of long lived charge separated states, co-deposition of chromophores and D/A can result in alternative electron transfer and photocurrent generation mechanisms. One example is reduced sensitizer injection, also known as supersensitization. In a standard dye-sensitized device, photoexcitation is followed by electron injection into the metal oxide and then reduction of subsequently oxidized dye by a donor species (eqn (15)–(17)). In contrast, during supersensitization, the photoexcited dye is first reduced by a donor species, then the reduced dye transfers an electron to the metal oxide (eqn (37) and (38)). Supersensitization increases the theoretical efficiency from 33% to 66% if the electron–hole pair is converted to electric power prior to band edge thermalization.^304,305 Spectroscopically these mechanisms can be differentiated by observing the dye cation and anion for excited state and reduced sensitizer injection, respectively.

MO–C|D → MO–C*|D (37)

MO–C*|D → MO–C⁻|D⁺ (38)

MO–C⁻|D⁺ → MO(e⁻)–C|D⁺ (39)

MO(e⁻)–C|D⁺ → MO–C|D (40)

MO–C⁻|D⁺ → MO–C|D (41)

Because the photocurrent generation step, i.e. electron transfer to the metal oxide, is a dark event in reduced sensitizer injection, devices operating by this mechanism are sometimes referred to as a photogalvanic cell, dye-sensitized photogalvanic regenerative solar cell, and/or a photo-galvanic dye-sensitized solar cell (P-DSSC).²⁷ A notable advantage of reduced sensitizer injection is that since the oxidized form of the dye is never generated, oxidatively unstable dyes can still be used. Additionally, the reduced sensitizer typically has more than 200 mV higher potential/driving force for electron transfer to the metal oxide than the excited state dye, thus the name supersensitization. Despite these and other potential advantages, research into supersensitization is somewhat limited compared to standard DSSCs.

While there may be a few earlier examples alluding to a supersenstization mechanism,³⁰⁶ arguably the first definitive demonstration was reported by Nasielski-Hinkens and coworkers in 1985.³⁰⁷ The excited state of the sensitizer molecule, Ru(TAP)₃ (TAP = 1,4,5,8-tetraazaphenanthrene) has insufficient driving force for electron injection into the CB of SnO₂. However, in the presence of hydroquinone (HQ) donor molecules the dye emission is completely quenched and a photocurrent of 0.3 μA was measured. Stern–Volmer quenching studies suggest that HQ can reduce the excited state of the Ru(TAP)₃ with the resulting anion having >600 mV more driving force for electron injection, leading them to attribute the photocurrent to a supersensitization mechanism.

In a system composed of Ru(deeb)(bpy)₂²⁺ (deeb = 4,4′-(CO₂C₂H₅)₂-2,2′-bipyridine) and phenothiazine (PTZ) as the donor species, Gerald J. Meyer were the first to use TA measurements to monitor reduced sensitizer injection (Fig. 29).³⁰⁸ For solution mixtures, TA measurements clearly demonstrated the reduced dye, Ru(deeb)(bpy)₂¹⁺, and PTZ cation were quickly generated via reductive quenching. On TiO₂, despite the dye lacking the excited state potential to inject into the conduction band (without the addition of Li⁺ ions),³⁰⁹ the formation of the charge separated PTZ⁺ and TiO₂(e⁻) state was attributed to reduced sensitizer injection. It is important to note that for the surface bound measurements, no feature of Ru(deppb)(bpy)₂¹⁺ was directly observed, which was credited to fast (<8 ns) electron transfer from the reduced dye to TiO₂.

Fig. 29 Ru(deep)(bpy)₂ direct photoinduced electron injection (left) and with PTZ reduced sensitizer injection mechanism (right). Adapted with permission from ref. 308. Copyright 1999 American Chemical Society.

In a follow up to this work, Ru(deeb)(bpy)₂²⁺ and other sensitizers were incorporated into a P-DSSC with TiO₂ as the substrate.³¹⁰ Interestingly, despite >99% excited state quenching of the dyes by PTZ, there is only <5% electron injection efficiency from the reduced dye and minimal photocurrent was observed. One of the rationalizations given for the low injection yield was that the CB of TiO₂ (−1.0 V vs. SCE) is not energetically accessible for reduced states of the dye. Presumably, to remedy this shortcoming, devices were then prepared using SnO₂ whose conduction band energies are ∼0.5 V lower in energy (i.e. more positive) than TiO₂.³¹¹ However, the excited state potential of these dyes were sufficient to directly transfer electrons into SnO₂ and thus no supersensitization was reported.

Iodide, one of the components of the common I⁻/I₃⁻ based DSSC redox mediator, is also an effective reducing agent for excited chromophores. For example, in 2000 Kamat et al. demonstrated that under an applied bias of 0.15 V vs. Ag/AgCl, C₆₀ clusters deposited on SnO₂ with an I⁻/I₃⁻ mediator electrolyte generated photocurrents of up to 0.2 mA cm⁻².³¹² Since the excited state of C₆₀ has insufficient potential to directly inject electrons into SnO₂, they attribute the photocurrent generation to electron injection from reduced C₆₀.

Gerald J. Meyer and coworkers observed excited state quenching of Ru(deeb)(bpy)₂²⁺* by I⁻ which was tentatively attributed to reductive quenching³¹³ after the formation of π adducts between the dye and iodide ([Ru(deeb)(bpy)₂, I]⁺).³¹⁴ The reduced sensitizer, Ru(deeb)(bpy)₂⁺, did not inject electrons into TiO₂ but rather decayed through a slow recombination with I₃⁻.

Grätzel and coworkers have also observed the reduction of excited state dye by I⁻.³¹⁵ Using TA spectroscopy they concluded that approximately 25% of the Ru(2,2′-bipyridyl-4,4′-dicarboxylato)(4,4′-di-nonyl-2′-bipyridyl)(NCS)₂ (Z-907) excited states were reductively quenched at high I⁻ concentration. The decay of the Z-907 anion occurred on the millisecond time scale via recombination with I₃⁻ but it was unclear why the reduced dye did not inject electrons into TiO₂.

A more recent example of supersensitization was reported by Youngblood and coworkers with fullerene and naphthacenequinone sensitizers on TiO₂ with solid-state spiro-MeOTAD or I⁻/I₃⁻ in solution as the donors/redox mediators, respectively.³¹⁶ For a mixture of dye and spiro-MeOTAD, the reduced dye is generated <1 ps after excitation and then recombines on a nanosecond time scale. Upon addition of TiO₂, the dye anion decayed significantly faster than the spiro-MeOTAD cation, due to electron transfer from the fullerene anion to TiO₂. The close proximity and fast sensitizer reduction in the spiro-MeOTAD device were given as the reason for higher photocurrents than the I⁻/I₃⁻ based device, though the overall performance was still low.

Despite the potential advantages of supersensitization over the traditional excited state electron injection mechanism, an efficient P-DSSC has yet to be demonstrated. From the above reports, it is clear that low injection yields from reduced sensitizers to semiconductors is one of the fundamentally limiting factors responsible for the low photocurrent generation efficiency. Why electron injection yields from the reduced state are so much lower than those from the excited state remains unanswered. Answering this question and subsequently designing new systems with high injection yields will be a crucial step in realizing an efficient P-DSSC.

A majority of studies mentioned in this section involve surface bound dye molecules with donor in solution. However, designing new covalent (Section 3.2.1), co-deposited (Section 3.2.2), or supramolecular (Section 3.2.3) C–D assemblies may be key to realizing efficient supersensitization that is not diffusion limited. At the very least, these assemblies will need to have a C–D pair that can readily undergo reductive electron transfer and a reduced dye with sufficiently negative potentials for electron injection. As an aside, similar oxidative mechanisms could be envisioned for photocathodes composed of dye and acceptor molecules on a p-type semiconductor like NiO but as far as we know that area has been unexplored.

As a final note on this topic, it is important to acknowledge that some currently reported DSSCs could be operating by the supersensitization mechanism, but in the absence of spectroscopic measurements to complement device measurements, the standard electron injection mechanism is assumed.

4. Complementary chromophores

The most commonly published examples of multimolecular assemblies on metal oxide surfaces are those containing two or more chromophores. While these chromophores can be used to quantify cross-surface energy/electron migration (Section 2.3) and other applications (Section 2.2), the primary goal of this effort is to increase broadband light absorption for use in solar energy conversion. In contrast to inorganic crystalline materials, which exhibit broad absorption features (FWHM > 1 eV),³¹⁷ molecules typically exhibit relatively narrow absorption peaks (FWHM < 0.5 eV). For solar cells containing a single dye molecule, these narrow features, and corresponding absorption gaps, typically result in light transmission losses which leads to decreased photocurrent (lower J_sc) and lower overall device efficiencies.³¹⁸ In DSSCs, incorporating two or more different chromophores at the metal oxide interface is an increasingly popular and effective strategy to mitigate these losses. The chromophores are typically selected such that the peak of one chromophore fills in the absorption gap of another and vice versa.

In terms of interfacial architecture, we have partitioned the multiple chromophore strategies into two primary categories. The first, and by far the most common, is co-deposition of two or more chromophores directly on the metal oxide surface (Section 4.1). The second, is by generating antenna like structures (Section 4.2) where one chromophore is bound to the surface and then, distal to that surface, a second chromophore is attached to the first by either an axial coordination (Section 4.2.1), metal ion linkages (Section 4.2.2), click chemistry (Section 4.2.4), electrostatic interactions (Section 4.2.5), or hydrogen bonding (Section 4.2.6).

4.1 Co-deposition

Co-deposition of complementary chromophores typically entails the chemical linkage of the dye molecule directly to the metal oxide surface, usually via carboxyl (CO₂H) or phosphonic acid (PO₃H₂) functional groups, or to a lesser extent, physisorption. Functionalizing a surface with two or more dyes is readily achieved by either (A) partial loading of one dye followed by soaking in the second dye, (B) a co-loading procedure where the metal oxide film is soaked in a solution containing both dyes, or (C) fully loading of one dye, partial desorption of that dye, followed by soaking in the second dye. Regardless of the loading procedure, the outcome is a distribution of both dye molecules directly bound to the surface as depicted in Fig. 30a.

Fig. 30 (a) Co-deposition of two different dyes on a MO via stepwise loading (slide A), co-loading from a mixed solution (slide B), or full loading of dye 1, followed by selective desorption, then loading of dye 2 (slide C). (b) Absorption spectra of N3 (squares) and p1M (circles) adsorbed on mesoporous TiO₂ films and their spectra (triangles) after co-deposition.

A distinct advantage of co-deposition, as opposed to those described below, is that the chromophores can undergo excitation and electron injection independently of each other. Consequently, the energetic design restrictions remain the same as for a single dye on the surface (i.e. the dye must have sufficient excited state potential for electron injection into the metal oxide and an oxidation potential more positive than the redox mediator for dye regeneration). An additional advantage is that each dye can be synthesized, or purchased, independently with the only rigid structural restriction being that they must contain a surface binding group.

While the co-deposited films do exhibit broadband absorption, one disadvantage is that the absorbance contribution from each chromophore is reduced relative to the fully loaded single sensitizer films, as shown in Fig. 30b. That is, the total dye loading is limited by the available surface area of the metal oxide and, by generating a 50 : 50 co-deposited film of similarly sized dyes, it effectively reduces the absorption contribution of each by half. Absorbance can again be increased by increasing film thickness, but increased thickness can also be a detriment by slowing redox mediator diffusion into the mesoporous films.³¹⁹ Given that the highest efficiency DSSCs to date utilize co-deposition (vide infra), the advantages of broadband absorption clearly outweigh the absorbance decrease.

4.1.1 History and record efficiencies. Although the first realized DSSC was published in 1991, co-deposition of two or more chromophores on high surface area nanocrystalline electrodes did not begin to take off until the mid 1990's. Some of the earliest examples, reported by Deng et al., utilized co-loading of tetrasulfonated gallium phthalocyanine (GaTsPc) and tetrasulfonated zinc porphyrin (ZnPor) onto nanocrystalline TiO₂.^320–322 DSSCs composed of the mixed films exhibited a >10% increase in light harvesting efficiency and short-circuit current compared to the single dye DSSCs. The enhancement was attributed to both increased broad-band absorption and to co-sensitization decreasing non-productive GaTsPc aggregation by effectively diluting the surface with ZnPor. Deng et al. later expanded on this work co-sensitizing with metallated and unmetallated tetrasulfonated porphyrins.³²⁰

Over the past two decades, co-deposition of complementary chromophores has significantly increased in popularity.^{44,50,323–345} The highest efficiency co-deposited DSSCs typically contain a mixture of organic D–π–A dyes,^338,346 zinc porphyrins,^{223,347–350} or ruthenium thiocyanate dyes^351–354 with several notable examples depicted in Fig. 31.

Fig. 31 Absorption range and peak maxima of chromophores used in the highest efficiency co-deposited DSSCs. Red, blue, and yellow dots represent individual dye absorption peak maxima, purple dots represent absorption peak maxima for the co-deposited films. Lines represent the approximate absorption ranges.

The highest efficiency DSSC reported to date, 14.3% PCE published by Kakiage et al. in 2015,³⁴⁶ was achieved by co-depositing ADEKA-1 and LEG4 dyes (Fig. 32). This is a notable accomplishment, not only as the most efficient DSSC to date, but also for being composed of purely organic molecules, with the previous record being 8.14%, using BP-2 and SQ2 dyes.³⁵⁵ Prior to the 14.3% efficient device, the DSSC efficiency record was held by Grätzel and coworkers at 12.3% PCE using a co-deposited film containing a porphyrin dye and an organic dye (YD2-o-C8 and Y123 in Fig. 32b).²²³ It is notable that efficiencies over 13% were measured at lower light intensities indicating that the device efficiency was limited, in part, by mass transport of the cobalt mediator into the porous metal oxide film. The dye absorption spectral features and device performance metrics for other notable co-deposited DSSCs are summarized in Fig. 31 and Table 3. It is important to note that these are the highest efficiencies reported in literature and are not certified values.

Fig. 32 Molecular structure of co-deposited dyes used in the highest efficiency DSSCs (a) ref. 346 adapted with permission from the Royal Society of Chemistry Copyright 2015 and (b) ref. 223. Reprinted with permission from The American Association for the Advancement of Science Copyright 2011.

Table 3 Chromophores used in highest efficiency co-deposited DSSCs

Dye	J sc (mA cm^−2)	V oc (V)	FF	η (%)	Ref.
ADEKA-1/LEG4	18.27	1.014	0.77	14.3	346
Y123	—	—	—	—	223
YD-o-C8	17.3	0.965	0.71	11.9
Y123/YD-o-C8	17.66	0.935	0.74	12.3
BD	20.61	0.682	0.71	10	352
D131	9.94	0.658	0.72	4.7
BD/D131	23.23	0.676	0.71	11.1
C1	11.21	0.780	0.65	5.7	349
XW4	16.22	0.702	0.70	7.9
C1/XW4	20.15	0.736	0.71	10.5
CD4	16.52	0.734	0.74	5.8	356
YDD6	9.66	0.671	0.72	4.7
YD-o-C8	16.6	0.758	0.71	8.8
CD4/YDD6/YD-o-C8	19.28	0.753	0.72	10.4
LD31	20.02	0.699	0.71	9.95	348
AN-4	8.85	0.692	0.73	4.44
LD31/AN-4	20.27	0.704	0.72	10.3

---

4.1.2 Systematic studies. One unfortunate observation about the collective of record efficiency chasing devices is that there is a large degree of trial and error where two known DSSC dyes are combined in different ratios until the maximum efficiency is achieved. For example, the highest efficiency device to date was the result of one such trial and error effort as noted by the authors:

“we fabricated cells sensitized by [ADEKA-1] and with [carboxy-anchor organic sensitizing dyes which have been reported to have high sensitizing properties] using an electrolyte solution containing an I₃/I redox mediator…Among the cells, a significant and the largest improvement in the [IPCE] was observed in the cell photosensitized by [ADEKA-1] with [LEG4].”³⁴⁶

Additionally, often only the best performing cells are reported. As in, only a particular dye and dye loading ratio for the best performing DSSC is published. However, there is likely much insight to be gained through systematic studies on how the dye structure and loading ratios/conditions influence the photophysical and electrochemical properties of the films as well as their performance in devices. Despite the potential importance of these types of studies, the number of reports remains relatively limited.^{37,347,356–360}

In one example, Spitler and coworkers, reported the synthesis and characterization of six different cyanine dye derivatives that were then loaded on TiO₂ (Fig. 33).³⁷ Interestingly, time dependent dye loading studies indicate that B1, G6, and R4 initially load as monomers and then form H-aggregates as the surface loading increases. In contrast, G2, G7, and R5, with ethyl substituents on the cyanine core, exhibit the inverse aggregate-to-monomer transition with increased surface loading. Mixtures of these dyes were effective in not only increasing broadband absorption in DSSCs but also in inhibiting dye aggregation resulting in a J_sc increase of 2 mA cm⁻² in the mixed B1-G7 DSSC over a comparative N3 control device.

Fig. 33 (a) Molecular structure and (b) absorption spectrum of the cyanine dyes used by Splitler and coworkers. Adapted with permission from ref. 37. Copyright 2001 American Chemical Society.

In a similar vein, Delcamp et al. synthesized eight different dyes with structural variation in the π bridges between the squaraine chromophore and the surface anchoring group.³⁵⁹ While a majority of the dyes aggregated at high surface loadings, incorporation of hexyl-substituted cyclopentadithiophene into the bridge disrupted aggregation, and improved device performance was achieved through co-sensitization.

There are several reports where the concentration of porphyrins, squaraines, cyanines, and organic D–π–A dye co-deposited mixtures were systematically varied^357,358 and even in one exceptional case, where they studied the influence of dye loading order on device performance.³⁴⁷ The loading ratios can often have a profound impact on the device J_sc and η with a large range of optimum loading ratios from 1 : 1³⁴⁷ to 1 : 20³⁵⁶ which is presumably dictated by variables such as the spectral overlap, inhibition of dye aggregation, dye regeneration rates, and others. Unfortunately, while these experiments are useful in maximizing the device performance, the discussion/understanding as to why a particular loading condition is best remains relatively limited.

4.1.3 Limitations and outlook. From the general trend in maximizing device efficiencies, it has become clear that co-deposited films are more promising than single sensitizer devices. There is a strong likelihood the DSSC that finally breaks the 15% efficiency threshold will contain two or more complementary chromophores co-deposited on the surface. As we progress towards that goal, there are more insights to be gained by reflecting on how the current devices are characterized, and what new or underutilized experiments can be performed to understand how co-deposition dictates device performance.

For example, most manuscripts only report the loading solution concentrations, with few papers^352,356 quantifying the amount of each dye absorbed on the surface. Unfortunately, the loading solution concentrations do not provide any information for the actual ratio of chromophores on the surface or their distribution across the surface (i.e. islands vs. homogenous).

A second under reported/studied phenomena is energy and electron transfer events between co-deposited dyes. There are several examples where a dye has insufficient excited state potential to quickly or efficiently inject an electron into TiO₂ but can transfer that excited state energy to a second dye that has much higher injection yields.^361–363 There are other examples where, after electron injection, the oxidative equivalent is transferred between two different dyes prior to regeneration by the mediator.³⁶⁴ In at least one system, by Hardin, et al., with the appropriate energetics, both intermolecular energy and electron transfer are observed, Fig. 34.³⁶⁵ Computational studies have been applied to co-deposited systems that help to predict energy transfer rates for specific molecular pairs and dyads, which could drive research into these phenomena forward with the benefit of increased charge injection.^161,362

Fig. 34 NIR dye attached to a TiO₂ surface absorbs NIR photons and uses short-range energy transfer to a sensitizing dye, which is responsible for electron transfer into the TiO₂ and hole regeneration with the electrolyte (k_reg). Reprinted with permission from ref. 365. Copyright 2011 American Chemical Society.

Given that these events can either improve or hinder device performance, it is clear that steady-state emission, time-resolved emission, and TA measurements could be used to provide fundamental insights into these events and help researchers understand how they influence and can be used to control device performance metrics.

Finally, it is well known that the orientation of dyes relative to the surface can have a profound influence on electron injection yields.³⁶⁶ Similarly, the aggregation and energy/electron transfer events described above are strongly dependent on the orientation of the dyes relative to the surface and to each other. Currently, presumably due to the complexity of the system, there is minimal direct structural information about multiple chromophores at the interface. However, future measurements (NMR, IR, XRD, etc.) or theoretical calculations could be used to correlate the interfacial structure with the rate of these events, and that knowledge could be used to design dyes that exhibit favorable interactions.

4.2 Antenna systems

Molecular antennas are composed of two or more chromophores that are linked via covalent bonds, metal ion coordination, electrostatics, hydrogen bonding, or other interactions.³⁶⁷ These antenna structures have been studied in solution,³⁶⁸ membranes,³⁶⁹ and other motifs but given the topic of this review, we will exclusively focus on chromophore antenna systems on nanocrystalline metal oxide films. This typically entails one chromophore bound to the metal oxide via surface binding groups and then, through various types of interaction, two or more chromophores are linked to the surface bound chromophore.

Similar to the co-deposited films from Section 4.1, the primary goal of these assemblies is to increase broadband absorption using complementary absorption features of each chromophore. However, the antenna systems differ from co-deposition in two distinct ways. The first is that, unlike co-deposited films where the chromophores must share surface area which effectively reduces the absorbance of each chromophore by half for a 1 to 1 loaded film, the antenna systems are generated by extending chromophore networks distal from the surface so the absorbance of each type of dye is cumulative. Secondly, in the context of DSSCs, because the antenna molecules are not directly bound to the surface, they must rely on either an energy or electron transfer cascade towards the surface or long distance electron transfer to generate photocurrent.

Some of the earliest chromophores used in DSSCs could be considered antenna systems composed of trinuclear ruthenium complexes with a surface-bound core and two antenna structures coordinated via cyanide ligands.^370–372 These complexes, as well as covalent assemblies (i.e. dimers, dyads, etc.) are not electronically distinct molecular moieties and thus will not be discussed here. Instead, below we will focus on antenna systems generated using axial coordination (Section 4.2.1), metal ion linkages (Section 4.2.2), click chemistry (Section 4.2.3), electrostatic interactions (Section 4.2.4), and hydrogen bonding (Section 4.2.5). A schematic depicting these different antenna strategies are shown in Fig. 35.

Fig. 35 Chromophore antenna systems linked via (a) axial coordination, (b) metal ion linkages, (c) click chemistry, (d) electrostatic interactions, and (e) hydrogen bonding.

4.2.1 Axial coordination. Axial coordination of pyridine to the metal center of porphyrins and phthalocyanines has been known for nearly twenty years.³⁷³ However, it was not until 2009 that Rawling et al. were the first to use axial coordination to generate two chromophore antenna assemblies on nanocrystalline metal oxides.³⁷⁴ Their assemblies were composed of a structural analogue of N3, where one bipyridyl ligand of the bis(2,2′-bipyridine)ruthenium(II) (NCS)₂ core is functionalized with surface binding carboxylate groups, and the other with pyridine which can readily coordinate to the second ruthenium phthalocyanine dye. In their studies, the C–C dyad was first generated in solution and then bound to a nanocrystalline TiO₂ surface. Despite the complementary absorption features and appropriate energetics for directional electron transfer to the surface, the J_sc, V_oc and η for the dyad were lower by a factor of ∼2 compared to the ruthenium dye alone. Because their “per-molecule photocurrent” (roughly the equivalent of APCE) was higher in the dyad, they attribute the lower performance of their antenna system to the lower surface loading. Presumably the large footprint of the dyad, relative to the single dye, inhibits high surface loadings.³⁷⁵

Later, Panda et al. took an alternative, stepwise approach of first binding a peryleneimide with a terminal pyridyl group to the TiO₂ surface and then linking either a zinc porphyrin³⁷⁶ or zinc phthalocyanine³⁷⁷ to the first layer via axial coordination. The antenna systems outperformed their individual components, with increases in J_sc, V_oc, and PCE. Supersensitization (Section 3.3) was proposed as the possible photocurrent generation mechanism but no spectroscopic support was provided.

4.2.2 Metal ion linkages. The coordination of CO₂H or PO₃H functionalized dyes to metal ions offers a means to self-assemble antenna structures on metal oxide substrates. The common strategies to utilize these metal ion linkages can be partitioned into (1) self-assembled multilayers and (2) metal–organic frameworks.

Self-assembled multilayers are formed using a simple, step-wise soaking procedure. For a bilayer film, a monolayer of the first molecule is bound to a high surface area metal oxide. Metal ions are then coordinated to the terminal CO₂H or PO₃H₂ groups of the first molecular layer. Finally, a different molecular layer is bound to the first through the linking metal ions. This approach has been exploited by Lee et al.^116,378 and Haga et al.^117,379 to functionalize planar gold and silicon surfaces, but was more recently translated to high surface area metal oxide electrodes by Hanson et al.¹¹⁵ The bilayer films were composed of phosphonate derivatized ruthenium polypyridyl complexes linked by Zr^IV ions. By tuning the energetics of each dye, they were able to control the direction of energy/electron transfer to or away from the metal oxide surface. The simple and modular nature of the bilayers were illustrated using phosphonated Ru(bpy)₃, carboxylated Ru(bpy)₃, N719, and ruthenium water oxidation catalysts.

The bilayer strategy was later used to generate an energy and electron transfer cascade DSSC composed of phosphonate derivatives of 1M (p1M) and N3 (Fig. 36), in which the bilayer increased J_sc and V_oc compared to each of the dyes alone.³⁸⁰ The multilayer structure was particularly effective at slowing recombination between the electrons in TiO₂ and the redox mediator, presumably due to increased steric hindrance of the bilayer. The self-assembled bilayer or trilayer films have subsequently been used to control electron transfer (Section 3.2.3) and facilitate photon upconversion (Section 2.2.2) on a metal oxide substrate.

Fig. 36 A bilayer assembly composed of pN3 and p1M dyes.³⁸⁰

The same metal ion coordination chemistry is common in metal organic frameworks (MOFs), and in 2015, within one month of each other, two different strategies emerged for functionalizing nanocrystalline metal oxide substrates with MOFs.^381,382

Morris and coworkers used solvothermal reactions to grow a ruthenium polypyridine MOF onto TiO₂.³⁸² The MOFs were generated by either (1) heating a mixture of 2,2′-bipyridine-5,5′-dicarboxylic acid, ZrCl₄, Ru(bpy)₂(2,2′-bipyridine-5,5′-dicarboxylic acid), and a TiO₂ film in “one pot”, or (2) growing the 2,2′-bipyridine-5,5′-dicarboxylate-Zr⁴⁺ MOF on TiO₂ and then incorporating ruthenium via thermal reaction with Ru(bpy)₂Cl₂. Although the device efficiencies were low (≤0.125%), this work did demonstrate that MOFs can be successfully implemented as an antenna structure in DSSCs (Fig. 37). Time-resolved emission measurements indicate that the triplet excited state can migrate up to 25 nm within the MOF framework, resulting in efficient emission quenching (>90%), presumably due to electron injection into TiO₂.

Fig. 37 A Ru-MOF sensitized solar cells. Reprinted from ref. 382 with permission from the Royal Society of Chemistry.

Alternatively, Han and coworkers employed the layer-by-layer technique to grow Cu^II, Ru^II, and Co^II based MOFs.^381,383,384 These films were generated by soaking TiO₂ films in alternating solutions of metal ions and trimesic, terephthalic, or naphthalenedicarboxylic acids. Powder X-ray diffraction measurements indicate that the MOFs grown via layer-by-layer were structurally the same as those grown by solvothermal methods. One of the key observations of this work was that iodine intercalation or “doping” of the MOFs created cooperative electrical conductivity between the π-electrons from aromatic rings and iodine, shifting the electronic behavior of the MOFs from insulating to conducting. When paired with I⁻/I₃⁻ electrolyte, the Ru-benzene-1,3,5-tricarboxylate MOF gave the best result with a PCE of 1.22% with a 10 layer MOF.³⁸⁴

4.2.3 Click chemistry. “Click” chemistry is a general class of metal catalyzed coupling reactions that are biocompatible, occur in high yield, and create minimal byproducts.³⁸⁵ Given the simplicity of this reaction, it is not surprising that it has been used to generate new D–π–A^386,387 and ruthenium polypyridyl dyes³⁸⁸ for use in DSSCs. Nickolaou, et al.³⁸⁹ have also used copper(I) catalyzed coupling of azide and acetylene-containing porphyrins to generate 1,2,3-triazoles linked to zinc porphyrin dyads. The dyads were first synthesized in solution, and then bound to TiO₂ as depicted in Fig. 38a. DSSCs composed of the porphyrin dyads had power conversion efficiencies of up to 5.16%. The performance of the dyads was dependent on the bridge length, with the shorter being more efficient. It is important to note that the dyad performance was not compared to the monomer only, so the contribution of the antenna structure was not clearly delineated.

Fig. 38 (a) Click then surface binding,³⁸⁹ and (b) click reaction on the surface.³⁹⁴

Click reactions have also been effective at building multi-chromophore antenna structures on planar metal oxide surfaces.^390–393 To the best of our knowledge, there is only one example of using click chemistry to directly generate a chromophore antenna structure on nanocrystalline metal oxides. In their report, Unger, Stack, and coworkers first functionalized the TiO₂ with a squaraine dye and 4-azidobenzoic acid co-adsorbent.³⁹⁴ The co-adsorbent served two purposes, (1) it prevented H-aggregation of the squaraine dyes (Section 2.1), and (2) it has a terminal azide group that can undergo click chemistry. The co-functionalized TiO₂ was then immersed in a solution of copper(I) catalyst and acetylene functionalized organic donor–acceptor dye (Fig. 38b). The two dyes worked cooperatively via efficient (>90%) energy transfer, slowed recombination, and improved overall device performance by 124% relative to a device that did not undergo the click reaction.

4.2.4 Electrostatic interactions. Electrostatic interactions between cationic and anionic dyes is an alternative means of generating an antenna structure at the surface. In 2004, the first example of an electrostatic assembly was demonstrated by Tennakone and coworkers who bound the sodium salt of an anionic xanthene dye to a metal oxide surface.³⁹⁵ Then through ion exchange of Na⁺ for the cationic dye, a second layer composed of methyl violet (MV) was deposited. In the absence of the xanthene dye, there was minimal deposition of MV on the surface, indicating that electrostatics was the primary driving force for the multi-chromophore assembly. IPCE spectra for the supramolecular assembly was the sum of both of the dyes absorption features and efficiencies increased up to 1.37% compared to its constituent devices (<0.6%).

Gnichwitz, et al. later used cationic pyridinium dendrimers on ZnO to electrostatically assemble anionic porphyrin molecules on the surface.³⁹⁶ Due to their high energy absorption (<400 nm), the dendrimers had minimal contribution to the photocurrent, and primarily only acted as an assembly scaffold.

D’Souza and coworkers used the complementary absorption features of a surface bound, anionic phthalocyanine, and a cationic porphyrin to generate a panchromatic supramolecular solar cell, Fig. 39.³⁹⁷ After comparing the monolayer and supramolecular assemblies of both the zinc and freebase porphyrin/phthalocyanine, they observed the highest J_sc, V_oc, and η with the zinc phthalocyanine/zinc porphyrin assembly, which was attributed to both higher IPCE across a broader wavelength range and slowed recombination at the interface.

Fig. 39 Electrostatically assembled bilayer from D'Souza.³⁹⁷

In an alternative approach, Kawashima, et al. generated nanoclusters (100–200 nm) composed of anionic C₆₀ and cationic zinc phthalocyanines, which were then electrochemically deposited on the SnO₂ surface.³⁹⁸ The IPCE for the clusters was more than 3 times larger than the sum of its parts with measurable photocurrent to nearly 800 nm. The proposed mechanism was charge separation in the cluster followed by electron transfer to the metal oxide, which was indirectly supported by TA measurements of the clusters in suspended solution.

4.2.5 Hydrogen bonding. Hydrogen bonding is a non-covalent interaction that is very common in biological systems (i.e. DNA, proteins, etc.), but despite its ubiquity, there is only one example of a hydrogen bonded supramolecular assembly on a metal oxide surface. Charalambidis et al. showed that peptide-based porphyrins on TiO₂ could seed the growth of a zinc porphyrin nanospherical antenna (Fig. 40).³⁹⁹ Interestingly, while attempting to generate DSSCs, the addition of electrolyte solution completely disrupted the hydrogen bonding network. To remedy this shortcoming, they measured devices with a solid-state hole transporting material, tetrakis(N,N-di-p-methoxyphenylamine)-9,9′-spirobifluorene (spiro-OMeTAD). The nanospherical antenna improved the J_sc and PCE compared to the monolayer alone due, at least in part, to slowed recombination dynamics at the interface. However, the performance was still very low (0.19%).

Fig. 40 Hydrogen bonding antenna. Reprinted from ref. 399 with permission from the Royal Society of Chemistry.

This section shows that there are many different non-covalent strategies that can be used to generate antenna structures at metal oxide interfaces. Generally, these assemblies increase the absorption, IPCE, and J_sc compared to the constituent monolayer devices. Additionally, a majority of these films slow recombination dynamics at the MO–dye–mediator interface due to the increased steric hindrance afforded by the supramolecular assembly. While promising, the highest performing DSSCs are still with co-deposited films (Section 4.1) and not assemblies. Given their additive absorption and slowed recombination, there is no reason to believe that an antenna structure with the appropriate dyes and structure could not be as efficient as or better than co-deposited films. However, the energetic and functional group requirements for antenna systems effectively increase the synthetic complexity and the number of steps needed to assemble the film/device. Additional energy and electron transfer events in the assembly introduce new loss pathways, and make it more difficult to characterize the films/mechanism. For example, unlike in the co-deposited films where both dyes are directly adjacent to the surface, the antenna systems require inter-chromophore energy and electron transfer events that are both orientation and distance dependent. Finally, when working with nanocrystalline substrates, there are concerns about the assemblies becoming too large and clogging the pores of the film, which will hinder diffusion of the redox mediators. Regardless of the challenges, there is much left to be learned about the function and assembly of chromophore antennas, not only for application in DSSCs but also in sensors, bio-recognition, catalysis, electroluminescence, and others.

4.3 Device engineering

Up until this point, our discussion has focused on the deposition of two or more molecules in proximity on a metal oxide surface. There are a number of deviations from the standard DSSC architecture that can also be used to effectively harness the complementary absorption features of two dye molecules. For the most part, these strategies can be described as device engineering with a common theme of spatially separating the dye molecules, as schematically summarized in Fig. 41. Spatial separation of dyes offer several advantages over proximity loading. The first is that the dye molecules will not suffer from any possible non-productive intermolecular interactions that can shift absorption spectra, reduce electron injection yields or increase recombination rates. The second is that competitive absorption losses can be mitigated by orienting the high energy absorber first and the low energy absorber second relative to the incident light. Many of these device architectures were reviewed in 2013 by Balasingam et al. and will only be briefly mentioned and updated here.⁴⁰⁰

Fig. 41 Schematic representation of multiple chromophore containing DSSCs (DCF = down-conversion filter, UCF = up-conversion filter).

The simplest device modification is to incorporate upconversion^260,401 or down conversion^402,403 filters in front and/or behind the DSSC, respectively (Fig. 41a). These filters are used to hypso- or bathochromically shift the incident light for improved spectral overlap with the absorption of the dye molecule on the metal oxide surface. As opposed to the antenna strategies mentioned in Section 4.2, which rely on non-emissive Dexter or Förster energy transfer, these filters require absorption and then emission from the filter towards the photoactive component of the solar cell. Given that the emission is isotropic, additional design components like reflectors or wave guides are required to efficiently harness the up/down converted light.⁴⁰⁴

Multilayer anodes are composed of metal oxides, typically TiO₂, functionalized with two or more distinct layers of dye molecules as depicted in Fig. 41b. Despite being deposited on the same metal oxide film, the contact between dye one and two only occurs at the interface between the two layers, effectively limiting any possible non-productive interactions. These films have been prepared using several different strategies. One is to utilize slow diffusion rates of a loading/desorption solution to preferentially adsorb/desorb dye on the top of the metal oxide film. The films are then soaked in a second dye to load the remaining surface area.^{329,354,405–408} Alternative strategies involve pressing,^{335,409–412} cold spray depositing,⁴¹³ doctorblading,^414,415 or electrophoretic depositing⁴¹⁶ a second layer of TiO₂ (or Al₂O₃)^417,418 on a TiO₂ layer that that is functionalized with dye 1. The second layer is then loaded with dye 2. IPCE measurements of these multilayer devices indicate that photocurrent is generated from both of the dye layers and, when corrected for filter effects, the J_sc is approximately the sum of each layers of the device. The photocurrent is highly sensitive to the thickness of each photoactive layer, as well as the total thickness of the anode.

An alternative multi-chromophore strategy is to incorporate two anodes, sensitized with two different dyes, into a device (Fig. 41c–f). The stacked tandem architecture (Fig. 41c), first introduced by Dürr et al. in 2004,⁴¹⁹ is effectively two adjacent DSSCs with two separate electrolyte compartments having a shared transparent substrate. The two DSSCs can be connected in series^420–423 or parallel.⁴²⁴ There are also several designs that utilize a single electrolyte compartment, but with the anodes arranged in a face-to-face architecture wired in parallel (Fig. 41d), as well as with a permeable TiO₂ (Fig. 41e)⁴²⁵ or TiO₂/Pt floating anode (Fig. 41f) wired in parallel and series respectively.^426,427 All of these devices are designed such that the incident irradiation is first absorbed by the higher energy dye/anode and the lower energy light is transmitted to the second low energy absorber. As noted above, these multiple anode devices can be connected in either a parallel or series circuit. With the parallel wiring, the J_sc is the sum of the current densities from each component but the V_oc is dictated by the lower voltage component. Alternatively for the cells connected in series, the V_oc is the sum of the voltages from each anode but current matching of each individual cell is necessary to minimize power losses.⁴²¹

An alternative tandem strategy, first introduced by Lindquist and coworkers,⁴²⁸ is to replace the photo-passive platinum cathode of a standard DSSC with a photoactive cathode typically composed of a molecular dye on a p-type semiconductor electrode like NiO (Fig. 41g).⁴²⁹ The electrodes are connected in series so the voltage is approximately the sum of the two single electrode devices, but the photocurrent is dominated by the weaker performing electrode. The overall device efficiency was low (0.39%), but this was a crucial proof of concept for a new class of p–n tandem DSSCs. Since the initial report, the device performance has been improved by modifying the dye, the metal oxide electrodes, and electrolyte.^430–434 The current record efficiency for a p–n tandem DSSC is held by Bach and coworkers at 1.91%.⁴³⁵ This performance is well below anode-only DSSCs, but there are many opportunities for improvements particularly by slowing down the non-productive recombination processes at the dye–metal oxide interface in the photocathode.⁴²⁹

Regardless of the device architecture shown in Fig. 41, the absorption, charge separation and regeneration events are dictated by the dye–semiconductor interface. As outlined above, there is considerable effort to improve the efficiency of these events using multimolecular interactions on the metal oxide surface. These molecular interactions are often tested in a standard DSSC architecture, but can readily be incorporated into a majority of the device described in Fig. 41. Therefore, our understanding of multimolecular interactions at the interface can be utilized in conjunction with new cell architectures to further increase device performance.

5. Supramolecular interactions

Until this point we have discussed chromophores interacting with themselves (Section 2), donors, acceptors, or insulating co-adsorbents (Section 3), or other chromophores (Section 4) on a nanocrystalline metal oxide substrate. The final category of multimolecular assemblies is comprised of photo- and electro-inactive molecules that are involved in adhering chromophores to the interface. Despite the “inert” nature of these molecules, they can effectively (1) prevent aggregation, (2) slow BET and recombination dynamics, (3) increase photochemical stability, and/or (4) provide a means of adsorbing non-surface binding molecules to the metal oxide surface. These assemblies have been further partitioned into host–guest interactions (Section 5.1), metal–ligand axial coordination (Section 5.2), and self-assembled multilayers (Section 5.3) as depicted in Fig. 42.

Fig. 42 Supramolecular assemblies from (a) host–guest interactions, (b) metal–ligand axial coordination, and (c) self-assembled multilayers with the dye molecule in blue and the “inert” molecule in grey.

5.1 Host–guest interactions

Supramolecular host–guest type interactions between molecules in solution have been studied for more than 50 years^436,437 with considerable interest in their use as artificial receptor sites and/or drug delivery scaffolds.⁴³⁸ More recently, host–guest interaction has emerged as a strategy to manipulate the surface binding and electron transfer dynamics of dyes on semiconducting interfaces. Work in this area prior to 2011 has been summarized in a perspective by Freitag and Galoppini,⁴³⁹ and thus will only briefly be described here with additional updated references.

In these MO–host–guest systems, the host molecules are typically composed of macrocyclic organic molecules with a hydrophobic porous interior and a hydrophilic exterior. Example host structures cyclodextrin (CD), cucur[7]bituril (CB[7]) and octa acid are shown in Fig. 43.

Fig. 43 Structures of host molecules used for host–guest assemblies on nanocrystalline metal oxide films. Reprinted from ref. 439 with permission from the Royal Chemical Society.

On metal oxide surfaces, the guest is typically a dye molecule, electron donor/acceptor with an aromatic core, or ligand that is small enough to insert into the host. These host–guest assemblies on the interface can either be formed in solution and then deposited onto a metal oxide substrate, or the host bound to the surface and then the guest molecule is added.

In 2003, Piotrowiak and coworkers published one of the first examples of a host–guest assembly on TiO₂ nanoparticles which was composed of a hemicarceplex host and an azulene dye as the guest.⁴⁴⁰ This host–guest complex was first generated in solution and then bound to the metal oxide substrate. For both azulene directly bound to the surface and in the assembly (Fig. 44), the quenched emission from azulene was attributed to fast and efficient excited state electron transfer from the dye to the metal oxide substrate. Due to the increased spatial separation between the dye and TiO₂, recombination of the charge separated state was three orders of magnitude slower in the host–guest assembly.^439,441

Fig. 44 (a) Azulene host–guest assembly and (b) carboxylated azulene on nanocrystalline TiO₂. Reprinted from ref. 440 with permission from The Royal Society of Chemistry.

At approximately the same time, Haque et al. were the first to use cyclodextrin, an oligosaccharide derived from starch, as a host for an azobenzene based dye on TiO₂ (Fig. 45).⁴⁴² The host–guest assembly was first synthesized in solution by mixing α-cyclodextrin with diaminoazobenzene in aqueous solution, followed by capping the ends of the guest molecule with relatively large dimethylnaphthylamine groups, sterically inhibiting any reversible release of the dye. For the dye molecule only, there was nominal loading on TiO₂. In contrast, monolayer surface coverage was achieved for the host–guest complex, which was attributed to hydrophilic interactions between the TiO₂ surface and the OH groups of cyclodextrin, or presumably hydrogen bonding, that is absent in the dye alone. The slowed electron recombination dynamics in the host–guest assembly was due to spatial separation between the dye cation and TiO₂ surface.

Fig. 45 Cyclodextrin–azobenzene host–guest assembly. Reprinted from ref. 442 with permission from Wiley-VCH.

Choi et al. used the alternative approach of first depositing cyclodextrin (CD) on a TiO₂ surface and then soaking the CD functionalized TiO₂ in an organic D–π–A dye (JK-2).⁴⁴³ The surface loading of JK-2 was higher with β-CD (9.78 × 10⁻⁸ mol mg⁻¹) than with αCD (7.98 × 10⁻⁸ mol mg⁻¹) or γCD (6.97 × 10⁻⁸ mol mg⁻¹), whose inner ring diameters are 6.0 Å, 4.7 Å, and 7.5 Å, respectively, but all of which exhibited lower loadings than JK-2 (2.04 × 10⁻⁷ mol mg⁻¹) in the absence of CD. Despite the lower loading, DSSCs composed of βCD-(JK-2) assemblies exhibited higher J_sc, V_oc, and PCE than the JK-2 only or the JK-2 + DCA co-deposited films. The improved performance was attributed to slowed recombination and increased electron lifetimes in TiO₂ for the host–guest film.

Similar efforts using carboxymethyl-β-CD (CMBCD) functionalized TiO₂ have been effective in adhering ruthenium(II)⁴⁴⁴ and iron(II)⁴⁴⁵ polypyridyl complexes to TiO₂. In contrast to CD, CMBCD has surface binding carboxylate groups that presumably increase surface loading and stability of the host molecule. DSSCs using the ruthenium–CMBCD films had increased V_oc and, in spite of 9.5 times lower surface coverage, it had comparable J_sc to the dye only devices. Unfortunately, overall device efficiencies were low (<0.3%) due to the low redox potential of their dye (0.48 V vs. SHE) which likely slows regeneration and may increase BET rates.

Using an alternative host with carboxylate groups, Porel et al. demonstrated that coumarin dyes (e.g. C-153, C-480, and C-1) in octa acid hosts can also be used to adhere dyes to the interface.⁴⁴⁶ The host–guest complexes were first prepared in solution as demonstrated by DOSY NMR experiments. Emission spectra from the assembly on ZrO₂ were similar to that observed in solution which suggested that the C-153 molecules were encapsulated and shielded from the local environment in both cases. On TiO₂, the assembly was still effective in facilitating excited state electron transfer from the dye to the substrate.

Galoppini and Freitag have demonstrated that host–guest interactions are also an effective means of generating electrochemical cells that are promising for application as electrochromic windows.⁴⁴⁷ In their films, cucur[7]bituril (CB[7]) and methyl viologen (MV²⁺) or methyl tolyl viologen (MTV²⁺) were selected as the host and guest molecules, respectively (Fig. 46a).

Fig. 46 (a) MV²⁺–CB[7] and MTV²⁺–CB[7] on TiO₂. (Reprinted with permission from ref. 447 Copyright 2010 American Chemical Society) and (b) the electrochromic response of MTV²⁺–CB[7] on TiO₂/FTO before (left) and after (right) an applied bias of −0.8 V vs. SCE (reprinted from ref. 450 with permission from The Royal Society of Chemistry).

There are a few reports of binding carboxylated methylviologen molecules directly to the TiO₂ surface^448,449 but, as noted by the authors, using the CB[7] host molecule has two distinct advantages: (1) MV²⁺ and MTV²⁺ can be adhered to the metal oxide in spite of the lack of surface binding groups, and (2) the host prevents degradation and dimerization of the reduced species.⁴⁵⁰ When used as an anode in an electrochemical cell under an applied bias of −0.8 V vs. NHE, a strong broad absorption feature with a peak at 600 nm appeared, and was accompanied by a distinctive, yellow to purple, reversible color change as shown in Fig. 46b. Since the applied potential is sufficiently negative, the mechanism is presumably reduction of the TiO₂ followed by reduction of MTV²⁺ at the dye–metal oxide interface, and not via cross surface electron transfer as described in Section 2.2. The color change was reversible for at least 20 switching cycles with minimal degradation.

The final example of a guest–host interaction to be described here was recently reported by Sun and coworkers.⁴⁵¹ In their system, a phosphonated ruthenium chromophore covalently lined to a βCD unit was first bound to TiO₂. Then upon soaking this film in a ruthenium based catalyst solution, the host–guest assembly was generated by insertion of the phenylpyridine ligand of the catalyst into the cyclodextrin cavity, as depicted in Fig. 47. Based on electrochemical measurements, the ratio of the host chromophore and the guest catalyst was estimated to be around 3 : 1. Under a small applied bias, the assemblies were shown to be effective for light driven water oxidation catalysis, albeit at low efficiencies due to competitive recombination. While this example is technically a chromophore–catalyst assembly, which in the introduction we mentioned we would omit from this review, we have included it here because this type of chromophore–host–molecule motif may offer new opportunities in generating chromophore–chromophore and chromophore–D/A type structures.

Fig. 47 A DSPEC composed of a chromophore–catalyst, host–guest assembly. Reprinted from ref. 451 with permission. Copyright 2016 Wiley-VCH.

5.2 Metal–ligand axial coordination

Metal–ligand axial coordination was briefly mentioned in Section 4.2.1 as a strategy for combining multiple chromophores. This versatile assembly motif is much more commonly used for attaching chromophores to metal oxide surfaces. The most obvious advantage of having a separate axial binding motif as the surface attachment unit is in the simplification of synthesis of the chromophore, but it can also help prevent aggregation. Multi-chromophoric arrays can be constructed in a stepwise, controlled manner, and the orientation of the chromophore relative to the surface can be modulated. Examples of two of the most common strategies for using axial coordination to adhere chromophores to a surface are depicted in Fig. 48.

Fig. 48 General depiction of (a) chromophore–metal–ligand and (b) metal–ligand axial coordination on metal oxide substrates.

The first involves an anchor–metal–chromophore type of architecture (Fig. 48a) that is generated by binding the nitrogen containing heterocyclic ligand to the metal oxide substrate, followed by treatments with a metal ion and then the chromophore.^452–455 The alternative, and more common strategy, is the anchor–chromophore type architecture that is typically generated by dipping a bare metal oxide film in a solution of the axial ligand and then in a solution of the metalloporphyrin or phthalocyanine dye (Fig. 48b).^456–460 While a thorough study using the same chromophore but comparing the two aforesaid binding modes is not available, spectroscopic and DSSC data suggest that the anchor–chromophore exhibits better performance than the anchor–metal–chromophore motif presumably due to reduced electronic coupling and slowed electron injection dynamics in the latter.⁴⁵⁵

In 2009, Brumbach et al. were the first to demonstrate axial ligand coordination of a dye to nanocrystalline TiO₂ surfaces.⁴⁵⁶ For bare or benzoic acid functionalized TiO₂, they observed minimal zinc or ruthenium porphyrin dye loading. In contrast, with carboxylated N-heterocycle functionalized TiO₂, metallated porphyrin readily adheres to the surface with the amount of surface coverage primarily dictated by the formation constant of axial ligation. The peak splitting of the metal porphyrin Soret band, as well as the lack of nonmetallated porphyrin loading, support an axial–metal binding architecture. Interestingly, with the surface binding motif, phosphonic acid or carboxylic acid, had minimal influence on the dye loading. Axial binding motif was also used to generate mixed chromophore films that exhibited 1.4 times higher efficiencies in DSSCs than individual sensitizers alone.

At about the same time, D’Souza and coworkers employed axial coordination to construct dye–TiO₂ electrodes, but with a series of imidazole and pyridine based amines with different basicities as the axial linker.⁴⁵⁷ As compared to linkers 1–3, higher photocurrents and photovoltages were obtained from DSSCs composed of linker 4 based anodes. The photovoltaic performance using the different linkers follows their metal–ligand binding constants with the dye. The highest device performance (V_oc = 0.67 V, J_sc = 1.6 mA cm⁻², η = 0.56%) was achieved with the zinc porphyrin–ferrocene dyad 8 and linker 4. The higher performance of 8 was attributed to electron transfer from the ferrocene donor to the oxidized porphyrin, effectively slowing recombination dynamics as described in Section 3.2.1. Electron injection times between 560 and 700 fs indicate there is relatively good electronic coupling between the dye and TiO₂ surface.⁴⁵⁸

D'Souza and coworkers later expanded this axial coordination methodology to zinc porphyrin–oxoporphyrinogen⁴⁵⁹ (Fig. 49a) and zinc porphyrin–zinc phthalocyanine dyads (ZnP–ZnPc; Fig. 49b).⁴⁶⁰ For the later system, the higher binding affinity of the imidazole linker to ZnP, was supported using competitive binding/displacement studies of ZnP and ZnPc monomers. The energy and electron transfer dynamics in these chromophore dyad systems were then studied using time-resolved emission/absorption techniques.

Fig. 49 Axial-metal coordinated (a) zinc porphyrin–oxoporphyrinogen (reprinted from ref. 459 by permission of The Royal Society of Chemistry), and (b) zinc porphyrin–zinc phthalocyanine dyads on TiO₂. (Reprinted from ref. 460 with permission. Copyright 2012 American Chemical Society.)

Crabtree, Schmuttenmaer, and coworkers did a systematic study varying the anchoring group (carboxylic acid, phosphonic acid, acetylacetone and hydroxamic acid) of the axially ligating pyridine molecule.⁴⁶¹ Interestingly, a higher binding constant was obtained with more basic guest molecules such as phenylimidazole compared to pyridine. Minimal dye desorption was observed for any of the linkers, even after soaking in water for several days, which was attributed to the hydrophobicity of the porphyrin dye.

While electron injection rates could not be resolved with transient THz spectroscopy, electron injection efficiencies with carboxylate and hydroxamate linkers were found to be higher than with phosphonate or acetylacetonate linkers. The efficiency of electron injection was later found to correlate with the calculated single molecule conductance of the linkers.⁴⁶²

5.3 Self-assembled multilayers

Self-assembled multilayers via metal ion linkages have already been introduced in Sections 2.2 and 4.2.2 as a means of facilitating triplet–triplet annihilation and generating chromophore–chromophore assemblies, respectively. This strategy is also effective for manipulating electron transfer dynamics at the dye–semiconductor interface.^463,464 These films are prepared by stepwise soaking of a metal oxide film in a solution of di-phosphonated bridging molecule, that is photochemically and electrochemically inert under standard conditions, a metal ion, and a carboxylated or phosphonated chromophore as depicted in Fig. 50. Both the nature of the “inert” molecule and the metal ion play a role in the rate, efficiency and mechanism of electron transfer.

Fig. 50 (a) The stepwise loading procedure and (b) a self-assembled bilayer composed of a bridging molecule, metal ion, and chromophore.

In 2015, Wang et al. reported the first example of manipulating electron transfer dynamics in a self-assembled bilayer using a bridging molecule.⁴⁶⁴ The formation of the bilayer architecture on TiO₂ and ZrO₂ was supported by ATR-IR and UV-vis measurements. By incrementally increasing the number of benzene rings in the bridging molecule from 1,4-phenylenediphosphonic acid to biphenyl-4,4′-diphosphonic acid to terphenyl-4,4′′-dicarboxylic acid, the authors slowed excited state electron transfer from RuC dye to the TiO₂ substrate, as demonstrated by time-resolved emission measurements. The relative efficiency of electron transfer events could even be observed visually (Fig. 51). The exponential decrease in both forward and backward electron transfer rates with respect to the length of the bridging molecule was attributed to the bridge and the Zr^IV metal linking ion acting as an electron tunneling barrier between the dye and metal oxide. In DSSCs, using N3 as the dye, increasing the bridge length manifests as an increased lifetime and diffusion length for the electrons in TiO₂ as well as an increase in open circuit voltage due to the slowed recombination dynamics.⁴⁶³ However, the overall device performance also decreased with bridge length due to the lower electron injection rate/yield and significant decrease in J_sc.

Fig. 51 Relative emission quenching in monolayers and self-assembled bilayers on ZrO₂ and TiO₂. Reprinted with permission from ref. 464. Copyright 2015 American Chemical Society.

While a majority of the self-assembled bilayers are generated using Zr^IV and Zn^II linking ions, follow up work by the same group demonstrated that that is not always the case. For example, it was shown Cu^II linking ions significantly quench the excited state of the dye even on ZrO₂.⁴⁶⁵ Despite the near unity quenching of the excited state, electron injection into TiO₂ and photocurrent generation did occur in a DSSC suggesting either electron tunneling is competitive with the excited state quenching process or potentially there was a metal ion-mediated hopping mechanism. Copper ions were also shown to slow both back electron transfer and recombination but a mechanism was not reported.

In a follow-up study, investigating a series of metal ions (Cd^II, Cu^II, Fe^II, La^III, Mn^II, Sn^IV, Zr^IV and Zn^II), excited state quenching and slowed recombination was similarly observed with Mn^II and Fe^II.^466,467 The nature of the metal ion also had an influence on the photostability of the films with bilayers composed of Zr^IV and La^III ions exhibiting the highest and lowest stability on both TiO₂ and ZrO₂, respectively. While there is still much to be learned about the role of the bridge and metal ions in these films, it is clear that self-assembled bilayers offer a simple, straightforward, and modular method for manipulating electron transfer dynamics at dye–semiconductor interfaces.

6. Conclusions

Over the past several decades, there has been a sharp rise in interest towards generating multi-molecular assemblies on high surface area metal oxide substrates. The assembly methods are largely based on intermolecular interactions that are well established in solvated systems, but the introduction of a substrate offers new opportunities as well as new challenges for generating, studying, and harnessing these assemblies. Synthetic control of surface binding groups as well as intermolecular interacting moieties of the chromophores, donors/acceptors, and other molecules offers exquisite control over the organic–inorganic interface. A combination of electrochemical as well as steady-state and time-resolved absorption/emission spectroscopic techniques have been crucial in measuring the rate and efficiency of energy and electron transfer events at these assembly–metal oxide interfaces. Key design strategies for improving advantageous while hindering detrimental events in device applications are already emerging.

Despite this progress, there is still much to be learned. The compilation of published data indicates a clear need to characterize and report the ratios and surface loadings of the various chromophores, donors/acceptors, and other species in the assemblies. This is particularly challenging for molecules that are photochemically and electrochemically inert under standard measurement conditions, but techniques like NMR and XPS show promise in addressing this shortcoming. Additionally, presumably due to the complexity of the interface and the difficulty of characterizing mesoporous materials, there is a notable lack of structural information about the assemblies at the interface. Theoretical modeling of these many atom systems is increasingly shedding light on the interface, but new developments in experimental characterization techniques for the structure at the interface will have a profound role in establishing structure–property relationships for the assemblies. Regardless of these shortcomings, it is clear that multi-molecular assemblies show great promise for influencing/controlling energy and electron transfer events for a plethora of applications.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This effort was supported in part by the Army Research Office under Grant No. W911NF-14-1-0660 and the National Science Foundation Graduate Research Fellowship under Grant No. DGE-1449440.

References

1. R. A. Marcus, J. Chem. Phys., 1956, 24, 966.
2. R. A. Marcus, J. Chem. Phys., 1965, 43, 679.
3. D. L. Dexter, J. Chem. Phys., 1953, 21, 836–850.
4. T. Forster, Discuss. Faraday Soc., 1959, 27, 7–17.
5. M. F. Hohmann-Marriott and R. E. Blankenship, Annu. Rev. Plant Biol., 2011, 62, 515–548.
6. H. Miyasaka, Acc. Chem. Res., 2013, 46, 248–257.
7. R. Foster, J. Phys. Chem., 1980, 84, 2135–2141.
8. L. Y. Zhu, Y. P. Yi, A. Fonari, N. S. Corbin, V. Coropceanu and J. L. Bredas, J. Phys. Chem. C, 2014, 118, 14150–14156.
9. B. Armitage and D. F. O'Brien, J. Am. Chem. Soc., 1991, 113, 9678–9679.
10. K. C. Hwang and D. Mauzerall, Nature, 1993, 361, 138–140.
11. F. D. Lewis, R. L. Letsinger and M. R. Wasielewski, Acc. Chem. Res., 2001, 34, 159–170.
12. E. Galoppini and M. A. Fox, J. Am. Chem. Soc., 1996, 118, 2299–2300.
13. M. Tadatsugu, Semicond. Sci. Technol., 2005, 20, S35.
14. S. K. Arya, S. Saha, J. E. Ramirez-Vick, V. Gupta, S. Bhansali and S. P. Singh, Anal. Chim. Acta, 2012, 737, 1–21.
15. C.-Y. Hsu, Y.-C. Chen, R. Y.-Y. Lin, K.-C. Ho and J. T. Lin, Phys. Chem. Chem. Phys., 2012, 14, 14099–14109.
16. J. D. Moras, B. Strandberg, D. Suc and K. Wilson, Science, 1996, 271, 933.
17. J. Jasieniak, M. Califano and S. E. Watkins, ACS Nano, 2011, 5, 5888–5902.
18. J. R. Bakke, K. L. Pickrahn, T. P. Brennan and S. F. Bent, Nanoscale, 2011, 3, 3482–3508.
19. M. D. Losego and K. Hanson, Nano Energy, 2013, 2, 1067–1069.
20. D. L. Ashford, M. K. Gish, A. K. Vannucci, M. K. Brennaman, J. L. Templeton, J. M. Papanikolas and T. J. Meyer, Chem. Rev., 2015, 115, 13006–13049.
21. A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo and H. Pettersson, Chem. Rev., 2010, 110, 6595–6663.
22. J. Albero, P. Atienzar, A. Corma and H. Garcia, Chem. Rec., 2015, 15, 803–828.
23. M. K. Nazeeruddin, R. Humphry-Baker, P. Liska and M. Grätzel, J. Phys. Chem. B, 2003, 107, 8981–8987.
24. Z. Zhang, S. M. Zakeeruddin, B. C. O'Regan, R. Humphry-Baker and M. Gratzel, J. Phys. Chem. B, 2005, 109, 21818–21824.
25. V. S. Manthou, E. K. Pefkianakis, P. Falaras and G. C. Vougioukalakis, ChemSusChem, 2015, 8, 588–599.
26. J. A. Christians, J. S. Manser and P. V. Kamat, J. Phys. Chem. Lett., 2015, 6, 852–857.
27. S. Ardo and G. J. Meyer, Chem. Soc. Rev., 2009, 38, 115–164.
28. T. W. Hamann, O. K. Farha and J. T. Hupp, J. Phys. Chem. C, 2008, 112, 19756–19764.
29. F. Fabregat-Santiago, J. Bisquert, E. Palomares, L. Otero, D. Kuang, S. M. Zakeeruddin and M. Grätzel, J. Phys. Chem. C, 2007, 111, 6550–6560.
30. F. Fabregat-Santiago, G. Garcia-Belmonte, I. Mora-Sero and J. Bisquert, Phys. Chem. Chem. Phys., 2011, 13, 9083–9118.
31. F. Fabregat-Santiago, J. Bisquert, G. Garcia-Belmonte, G. Boschloo and A. Hagfeldt, Sol. Energy Mater. Sol. Cells, 2005, 87, 117–131.
32. S. G. Yan, J. S. Prieskorn, Y. Kim and J. T. Hupp, J. Phys. Chem. B, 2000, 104, 10871–10877.
33. G. B. S. Edward, H. Yonemoto, R. H. Schmeh, S. M. Hubig, B. L. I. R. L. Riley and T. E. Mallouk, J. Am. Chem. Soc., 1994, 116, 4786–4795.
34. M. Kasha, H. R. Rawls and M. Ashraf El-Bayoumi, Pure Appl. Chem., 1965, 11, 371.
35. N. J. Hestand and F. C. Spano, Acc. Chem. Res., 2017, 50, 341–350.
36. H. Tributsch, Bunsen-Ges. Phys. Chem. Ber., 1969, 73, 582–590.
37. A. Ehret, L. Stuhl and M. T. Spitler, J. Phys. Chem. B, 2001, 105, 9960–9965.
38. A. Konno, G. R. A. Kumara, S. Kaneko, B. Onwona-Agyeman and K. Tennakone, Chem. Lett., 2007, 36, 716–717.
39. M. V. Martinez-Diaz, G. de la Torre and T. Torres, Chem. Commun., 2010, 46, 7090–7108.
40. K. R. Mulhern, M. R. Detty and D. F. Watson, J. Phys. Chem. C, 2011, 115, 6010–6018.
41. D. Liu and P. V. Kamat, J. Electrochem. Soc., 1995, 142, 835–839.
42. T. Maeda, Y. Hamamura, K. Miyanaga, N. Shima, S. Yagi and H. Nakazumi, Org. Lett., 2011, 13, 5994–5997.
43. S. Tatay, S. A. Haque, B. O'Regan, J. R. Durrant, W. J. H. Verhees, J. M. Kroon, A. Vidal-Ferran, P. Gavina and E. Palomares, J. Mater. Chem., 2007, 17, 3037–3044.
44. K. Sayama, S. Tsukagoshi, T. Mori, K. Hara, Y. Ohga, A. Shinpou, Y. Abe, S. Suga and H. Arakawa, Sol. Energy Mater. Sol. Cells, 2003, 80, 47–71.
45. A. C. Khazraji, S. Hotchandani, S. Das and P. V. Kamat, J. Phys. Chem. B, 1999, 103, 4693–4700.
46. H. Choi, J. K. Lee, K. H. Song, K. Song, S. O. Kang and J. Ko, Tetrahedron, 2007, 63, 1553–1559.
47. S. Alex, U. Santhosh and S. Das, J. Photochem. Photobiol., A, 2005, 172, 63–71.
48. Z.-S. Wang, Y. Cui, Y. Dan-oh, C. Kasada, A. Shinpo and K. Hara, J. Phys. Chem. C, 2007, 111, 7224–7230.
49. G. Marotta, M. A. Reddy, S. P. Singh, A. Islam, L. Han, F. De Angelis, M. Pastore and M. Chandrasekharam, ACS Appl. Mater. Interfaces, 2013, 5, 9635–9647.
50. M. Guo, P. Diao, Y.-J. Ren, F. Meng, H. Tian and S.-M. Cai, Sol. Energy Mater. Sol. Cells, 2005, 88, 23–35.
51. M. Kawasaki and S. Aoyama, Chem. Commun., 2004, 988–989,  10.1039/B400071D.
52. Y.-S. Kim, K. Liang, K.-Y. Law and D. G. Whitten, J. Phys. Chem., 1994, 98, 984–988.
53. D. Liu, R. W. Fessenden, G. L. Hug and P. V. Kamat, J. Phys. Chem. B, 1997, 101, 2583–2590.
54. J. R. Mann, M. K. Gannon, T. C. Fitzgibbons, M. R. Detty and D. F. Watson, J. Phys. Chem. C, 2008, 112, 13057–13061.
55. C. Nasr, D. Liu, S. Hotchandani and P. V. Kamat, J. Phys. Chem., 1996, 100, 11054–11061.
56. D. Liu and P. V. Kamat, J. Chem. Phys., 1996, 105, 965–970.
57. N. A. Treat, F. J. Knorr and J. L. McHale, J. Phys. Chem. C, 2016, 120, 9122–9131.
58. F. Nüesch, J. E. Moser, V. Shklover and M. Grätzel, J. Am. Chem. Soc., 1996, 118, 5420–5431.
59. S. Y. Bang, M. J. Ko, K. Kim, J. H. Kim, I. H. Jang and N. G. Park, Synth. Met., 2012, 162, 1503–1507.
60. D. Daphnomili, G. D. Sharma, S. Biswas, K. R. J. Thomas and A. G. Coutsolelos, J. Photochem. Photobiol., A, 2013, 253, 88–96.
61. T. Geiger, S. Kuster, J. H. Yum, S. J. Moon, M. K. Nazeeruddin, M. Gratzel and F. Nuesch, Adv. Funct. Mater., 2009, 19, 2720–2727.
62. L. Giribabu, V. K. Singh, C. V. Kumar, Y. Soujanya, P. Y. Reddy and M. L. Kantam, Sol. Energy, 2011, 85, 1204–1212.
63. J. J. He, G. Benko, F. Korodi, T. Polivka, R. Lomoth, B. Akermark, L. C. Sun, A. Hagfeldt and V. Sundstrom, J. Am. Chem. Soc., 2002, 124, 4922–4932.
64. Z. S. Huang, H. L. Feng, X. F. Zang, Z. Iqbal, H. P. Zeng, D. B. Kuang, L. Y. Wang, H. Meier and D. R. Cao, J. Mater. Chem. A, 2014, 2, 15365–15376.
65. M. Ince, F. Cardinali, J. H. Yum, M. V. Martinez-Diaz, M. K. Nazeeruddin, M. Gratzel and T. Torres, Chem. – Eur. J., 2012, 18, 6343–6348.
66. M. Ishida, S. W. Park, D. Hwang, Y. B. Koo, J. L. Sessler, D. Y. Kim and D. Kim, J. Phys. Chem. C, 2011, 115, 19343–19354.
67. T. Maeda, Y. Hamamura, K. Miyanaga, N. Shima, S. Yagi and H. Nakazumi, Org. Lett., 2011, 13, 5994–5997.
68. M. K. Nazeeruddin, R. Humphry-Baker, M. Gratzel and B. A. Murrer, Chem. Commun., 1998, 719–720,  10.1039/a708834e.
69. A. Morandeira, I. Lopez-Duarte, B. O'Regan, M. V. Martinez-Diaz, A. Forneli, E. Palomares, T. Torres and J. R. Durrant, J. Mater. Chem., 2009, 19, 5016–5026.
70. S. Feng, Q.-S. Li, P.-P. Sun, T. A. Niehaus and Z.-S. Li, ACS Appl. Mater. Interfaces, 2015, 7, 22504–22514.
71. E. Palomares, M. V. Martinez-Diaz, S. A. Haque, T. Torres and J. R. Durrant, Chem. Commun., 2004, 2112–2113,  10.1039/B407860H.
72. Y. Ooyama, A. Ishii, Y. Kagawa, I. Imae and Y. Harima, New J. Chem., 2007, 31, 2076–2082.
73. J. H. Yum, P. Walter, S. Huber, D. Rentsch, T. Geiger, F. Nuesch, F. De Angelis, M. Gratzel and M. K. Nazeeruddin, J. Am. Chem. Soc., 2007, 129, 10320–10321.
74. J. Shi, J. Chen, Z. Chai, H. Wang, R. Tang, K. Fan, M. Wu, H. Han, J. Qin, T. Peng, Q. Li and Z. Li, J. Mater. Chem., 2012, 22, 18830–18838.
75. H. Choi, C. Baik, S. O. Kang, J. Ko, M.-S. Kang, M. K. Nazeeruddin and M. Grätzel, Angew. Chem., Int. Ed., 2008, 47, 327–330.
76. L.-N. Yang, S.-C. Li, Z.-S. Li and Q.-S. Li, RSC Adv., 2015, 5, 25079–25088.
77. K. Sayama, S. Tsukagoshi, K. Hara, Y. Ohga, A. Shinpou, Y. Abe, S. Suga and H. Arakawa, J. Phys. Chem. B, 2002, 106, 1363–1371.
78. N. Koumura, Z.-S. Wang, S. Mori, M. Miyashita, E. Suzuki and K. Hara, J. Am. Chem. Soc., 2006, 128, 14256–14257.
79. S. Qu, C. Qin, A. Islam, Y. Wu, W. Zhu, J. Hua, H. Tian and L. Han, Chem. Commun., 2012, 48, 6972–6974.
80. M. Fang, H. Li, Q. Li and Z. Li, RSC Adv., 2016, 6, 40750–40759.
81. Q. Li, J. Shi, H. Li, S. Li, C. Zhong, F. Guo, M. Peng, J. Hua, J. Qin and Z. Li, J. Mater. Chem., 2012, 22, 6689–6696.
82. B. L. Watson, B. D. Sherman, A. L. Moore, T. A. Moore and D. Gust, Phys. Chem. Chem. Phys., 2015, 17, 15788–15796.
83. Z. Ning, Q. Zhang, W. Wu, H. Pei, B. Liu and H. Tian, J. Org. Chem., 2008, 73, 3791–3797.
84. M. Pastore and F. De Angelis, ACS Nano, 2010, 4, 556–562.
85. S. Agrawal, M. Pastore, G. Marotta, M. A. Reddy, M. Chandrasekharam and F. De Angelis, J. Phys. Chem. C, 2013, 117, 9613–9622.
86. T. Etienne, L. Chbibi, C. Michaux, E. A. Perpète, X. Assfeld and A. Monari, Dyes Pigm., 2014, 101, 203–211.
87. S. Feng, Q.-S. Li, L.-N. Yang, Z.-Z. Sun, T. A. Niehaus and Z.-S. Li, J. Power Sources, 2015, 273, 282–289.
88. L. Zhang and J. M. Cole, ACS Appl. Mater. Interfaces, 2014, 6, 15760–15766.
89. H. Kusama and K. Sayama, J. Phys. Chem. C, 2012, 116, 23906–23914.
90. M. B. Smith and J. Michl, Annu. Rev. Phys. Chem., 2013, 64, 361–386.
91. T. N. Singh-Rachford and F. N. Castellano, Coord. Chem. Rev., 2010, 254, 2560–2573.
92. R. Katoh, A. Furube, T. Yoshihara, K. Hara, G. Fujihashi, S. Takano, S. Murata, H. Arakawa and M. Tachiya, J. Phys. Chem. B, 2004, 108, 4818–4822.
93. W. Shockley and H. J. Queisser, J. Appl. Phys., 1961, 32, 510–519.
94. M. B. Smith and J. Michl, Chem. Rev., 2010, 110, 6891–6936.
95. C. E. Swenberg and W. T. Stacy, Chem. Phys. Lett., 1968, 2, 327–328.
96. S. Singh, W. J. Jones, W. Siebrand, B. P. Stoicheff and W. G. Schneider, J. Chem. Phys., 1965, 42, 330–342.
97. I. Paci, J. C. Johnson, X. Chen, G. Rana, D. Popović, D. E. David, A. J. Nozik, M. A. Ratner and J. Michl, J. Am. Chem. Soc., 2006, 128, 16546–16553.
98. S. T. Roberts, R. E. McAnally, J. N. Mastron, D. H. Webber, M. T. Whited, R. L. Brutchey, M. E. Thompson and S. E. Bradforth, J. Am. Chem. Soc., 2012, 134, 6388–6400.
99. E. A. Margulies, C. E. Miller, Y. Wu, L. Ma, G. C. Schatz, R. M. Young and M. R. Wasielewski, Nat. Chem., 2016, 8, 1120–1125.
100. N. V. Korovina, S. Das, Z. Nett, X. Feng, J. Joy, R. Haiges, A. I. Krylov, S. E. Bradforth and M. E. Thompson, J. Am. Chem. Soc., 2016, 138, 617–627.
101. J. Zirzlmeier, D. Lehnherr, P. B. Coto, E. T. Chernick, R. Casillas, B. S. Basel, M. Thoss, R. R. Tykwinski and D. M. Guldi, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 5325–5330.
102. S. N. Sanders, E. Kumarasamy, A. B. Pun, M. T. Trinh, B. Choi, J. Xia, E. J. Taffet, J. Z. Low, J. R. Miller, X. Roy, X. Y. Zhu, M. L. Steigerwald, M. Y. Sfeir and L. M. Campos, J. Am. Chem. Soc., 2015, 137, 8965–8972.
103. J. C. Johnson, A. J. Nozik and J. Michl, J. Am. Chem. Soc., 2010, 132, 16302–16303.
104. J. N. Schrauben, Y. Zhao, C. Mercado, P. I. Dron, J. L. Ryerson, J. Michl, K. Zhu and J. C. Johnson, ACS Appl. Mater. Interfaces, 2015, 7, 2286–2293.
105. N. J. Ekins-Daukes and T. W. Schmidt, Appl. Phys. Lett., 2008, 93, 063507.
106. V. Gray, D. Dzebo, M. Abrahamsson, B. Albinsson and K. Moth-Poulsen, Phys. Chem. Chem. Phys., 2014, 16, 10345–10352.
107. T. F. Schulze and T. W. Schmidt, Energy Environ. Sci., 2015, 8, 103–125.
108. C. A. Parker, C. G. Hatchard and T. A. Joyce, Nature, 1965, 205, 1282–1284.
109. C. E. McCusker and F. N. Castellano, Inorg. Chem., 2015, 54, 6035–6042.
110. J. Zhao, W. Wu, J. Sun and S. Guo, Chem. Soc. Rev., 2013, 42, 5323–5351.
111. J. S. Lissau, J. M. Gardner and A. Morandeira, J. Phys. Chem. C, 2011, 115, 23226–23232.
112. J. S. Lissau, D. Nauroozi, M.-P. Santoni, S. Ott, J. M. Gardner and A. Morandeira, J. Phys. Chem. C, 2015, 119, 25792–25806.
113. J. S. Lissau, D. Nauroozi, M.-P. Santoni, T. Edvinsson, S. Ott, J. M. Gardner and A. Morandeira, J. Phys. Chem. C, 2015, 119, 4550–4564.
114. C. Simpson, T. M. Clarke, R. W. MacQueen, Y. Y. Cheng, A. J. Trevitt, A. J. Mozer, P. Wagner, T. W. Schmidt and A. Nattestad, Phys. Chem. Chem. Phys., 2015, 17, 24826–24830.
115. K. Hanson, D. A. Torelli, A. K. Vannucci, M. K. Brennaman, H. Luo, L. Alibabaei, W. Song, D. L. Ashford, M. R. Norris, C. R. K. Glasson, J. J. Concepcion and T. J. Meyer, Angew. Chem., Int. Ed., 2012, 51, 12782–12785.
116. H. Lee, L. J. Kepley, H. G. Hong, S. Akhter and T. E. Mallouk, J. Phys. Chem., 1988, 92, 2597–2601.
117. K. Terada, K. Kobayashi, J. Hikita and M. Haga, Chem. Lett., 2009, 38, 416–417.
118. L. A. Vermeulen, J. L. Snover, L. S. Sapochak and M. E. Thompson, J. Am. Chem. Soc., 1993, 115, 11767–11774.
119. S. P. Hill, T. Banerjee, T. Dilbeck and K. Hanson, J. Phys. Chem. Lett., 2015, 6, 4510–4517.
120. S. P. Hill, T. Dilbeck, E. Baduell and K. Hanson, ACS Energy Lett., 2016, 1, 3–8.
121. T. Dilbeck, S. P. Hill and K. Hanson, J. Mater. Chem. A, 2017, 5, 11652–11660.
122. S. P. Hill and K. Hanson, J. Am. Chem. Soc., 2017, 139, 10988–10991.
123. L. Frazer, J. K. Gallaher and T. W. Schmidt, ACS Energy Lett., 2017, 2, 1346–1354.
124. T. Dilbeck, J. C. Wang, Y. Zhou, A. Olsson, M. Sykora and K. Hanson, J. Phys. Chem. C, 2017, 121, 19690–19698.
125. C. A. Kelly, F. Farzad, D. W. Thompson and G. J. Meyer, Langmuir, 1999, 15, 731–737.
126. F. Farzad, D. W. Thompson, C. A. Kelly and G. J. Meyer, J. Am. Chem. Soc., 1999, 121, 5577–5578.
127. C. Creutz, M. Chou, T. L. Netzel, M. Okumura and N. Sutin, J. Am. Chem. Soc., 1980, 102, 1309–1319.
128. V. Balzani, A. Juris, M. Venturi, S. Campagna and S. Serroni, Chem. Rev., 1996, 96, 759–834.
129. J. P. Sauvage, J. P. Collin, J. C. Chambron, S. Guillerez, C. Coudret, V. Balzani, F. Barigelletti, L. De Cola and L. Flamigni, Chem. Rev., 1994, 94, 993–1019.
130. L. M. Dupray, M. Devenney, D. R. Striplin and T. J. Meyer, J. Am. Chem. Soc., 1997, 119, 10243–10244.
131. W. E. Jones, S. M. Baxter, G. F. Strouse and T. J. Meyer, J. Am. Chem. Soc., 1993, 115, 7363–7373.
132. L. A. Worl, G. F. Strouse, J. N. Younathan, S. M. Baxter and T. J. Meyer, J. Am. Chem. Soc., 1990, 112, 7571–7578.
133. S. A. Trammell, J. Yang, M. Sykora, C. N. Fleming, F. Odobel and T. J. Meyer, J. Phys. Chem. B, 2001, 105, 8895–8904.
134. G. T. Higgins, B. V. Bergeron, G. M. Hasselmann, F. Farzad and G. J. Meyer, J. Phys. Chem. B, 2006, 110, 2598–2605.
135. S. Ardo and G. J. Meyer, J. Am. Chem. Soc., 2011, 133, 15384–15396.
136. W. Song, A. Ito, R. A. Binstead, K. Hanson, H. Luo, M. K. Brennaman, J. J. Concepcion and T. J. Meyer, J. Am. Chem. Soc., 2013, 135, 11587–11594.
137. S. E. Bettis, K. Hanson, L. Wang, M. K. Gish, J. J. Concepcion, Z. Fang, T. J. Meyer and J. M. Papanikolas, J. Phys. Chem. A, 2014, 118, 10301–10308.
138. J. R. Swierk, N. S. McCool, T. P. Saunders, G. D. Barber and T. E. Mallouk, J. Am. Chem. Soc., 2014, 136, 10974–10982.
139. D. Moia, U. B. Cappel, T. Leijtens, X. Li, A. M. Telford, H. J. Snaith, B. C. O’Regan, J. Nelson and P. R. F. Barnes, J. Phys. Chem. C, 2015, 119, 18975–18985.
140. L. Yang, U. B. Cappel, E. L. Unger, M. Karlsson, K. M. Karlsson, E. Gabrielsson, L. Sun, G. Boschloo, A. Hagfeldt and E. M. J. Johansson, Phys. Chem. Chem. Phys., 2012, 14, 779–789.
141. D. Moia, T. Leijtens, N. Noel, H. J. Snaith, J. Nelson and P. R. F. Barnes, Adv. Mater., 2015, 27, 5889–5894.
142. S. Ardo and G. J. Meyer, J. Am. Chem. Soc., 2010, 132, 9283–9285.
143. B. J. Brennan, A. C. Durrell, M. Koepf, R. H. Crabtree and G. W. Brudvig, Phys. Chem. Chem. Phys., 2015, 17, 12728–12734.
144. B. J. Brennan, K. P. Regan, A. C. Durrell, C. A. Schmuttenmaer and G. W. Brudvig, ACS Energy Lett., 2017, 2, 168–173.
145. D. Moia, V. Vaissier, I. Lopez-Duarte, T. Torres, M. K. Nazeeruddin, B. C. O'Regan, J. Nelson and P. R. F. Barnes, Chem. Sci., 2014, 5, 281–290.
146. K. Hanson, M. K. Brennaman, A. Ito, H. Luo, W. Song, K. A. Parker, R. Ghosh, M. R. Norris, C. R. K. Glasson, J. J. Concepcion, R. Lopez and T. J. Meyer, J. Phys. Chem. C, 2012, 116, 14837–14847.
147. P. Bonhôte, E. Gogniat, S. Tingry, C. Barbé, N. Vlachopoulos, F. Lenzmann, P. Comte and M. Grätzel, J. Phys. Chem. B, 1998, 102, 1498–1507.
148. D. N. Blauch and J. M. Saveant, J. Am. Chem. Soc., 1992, 114, 3323–3332.
149. Q. Wang, S. M. Zakeeruddin, M. K. Nazeeruddin, R. Humphry-Baker and M. Grätzel, J. Am. Chem. Soc., 2006, 128, 4446–4452.
150. V. Vaissier, E. Mosconi, D. Moia, M. Pastore, J. M. Frost, F. De Angelis, P. R. F. Barnes and J. Nelson, Chem. Mater., 2014, 26, 4731–4740.
151. B. N. DiMarco, T. C. Motley, R. S. Balok, G. Li, M. A. Siegler, R. M. O’Donnell, K. Hu and G. J. Meyer, J. Phys. Chem. C, 2016, 120, 14226–14235.
152. X. Wang, L. Guo, P. F. Xia, F. Zheng, M. S. Wong and Z. Zhu, J. Mater. Chem. A, 2013, 1, 13328–13336.
153. N. Papageorgiou, M. Grätzel, O. Enger, D. Bonifazi and F. Diederich, J. Phys. Chem. B, 2002, 106, 3813–3822.
154. P. Qu and G. J. Meyer, Langmuir, 2001, 17, 6720–6728.
155. A. Fattori, L. M. Peter, H. Wang, H. Miura and F. Marken, J. Phys. Chem. C, 2010, 114, 11822–11828.
156. X. Li, M. K. Nazeeruddin, M. Thelakkat, P. R. F. Barnes, R. Vilar and J. R. Durrant, Phys. Chem. Chem. Phys., 2011, 13, 1575–1584.
157. E. Galoppini, W. Guo, W. Zhang, P. G. Hoertz, P. Qu and G. J. Meyer, J. Am. Chem. Soc., 2002, 124, 7801–7811.
158. S. A. Trammell and T. J. Meyer, J. Phys. Chem. B, 1999, 103, 104–107.
159. Q. Wang, S. M. Zakeeruddin, J. Cremer, P. Bäuerle, R. Humphry-Baker and M. Grätzel, J. Am. Chem. Soc., 2005, 127, 5706–5713.
160. X. Li, M. K. Nazeeruddin, M. Thelakkat, P. R. F. Barnes, R. Vilar and J. R. Durrant, Phys. Chem. Chem. Phys., 2011, 13, 1575–1584.
161. M. Pastore, T. Etienne and F. De Angelis, J. Mater. Chem. C, 2016, 4, 4346–4373.
162. J. H. Delcamp, Y. R. Shi, J. H. Yum, T. Sajoto, E. Dell'Orto, S. Barlow, M. K. Nazeeruddin, S. R. Marder and M. Gratzel, Chem. – Eur. J., 2013, 19, 1819–1827.
163. D. B. Kuang, C. Klein, S. Ito, J. E. Moser, R. Humphry-Baker, S. M. Zakeeruddin and M. Gratzel, Adv. Funct. Mater., 2007, 17, 154–160.
164. J. Lim, Y. S. Kwon and T. Park, Chem. Commun., 2011, 47, 4147–4149.
165. S. L. Lu, T. Wu, B. F. Ren and R. Geng, J. Mater. Sci.: Mater. Electron., 2013, 24, 2346–2350.
166. A. Reynal and E. Palomares, Energy Environ. Sci., 2009, 2, 1078–1081.
167. H. P. Shen, H. Lin, Y. Z. Liu, X. Li, J. Zhang, N. Wang and J. B. Li, Electrochim. Acta, 2011, 56, 2092–2097.
168. Y. B. Shi, M. Liang, L. N. Wang, H. Y. Han, L. S. You, Z. Sun and S. Xue, ACS Appl. Mater. Interfaces, 2013, 5, 144–153.
169. M. K. Wang, C. Gratzel, S. J. Moon, R. Humphry-Baker, N. Rossier-Iten, S. M. Zakeeruddin and M. Gratzel, Adv. Funct. Mater., 2009, 19, 2163–2172.
170. P. Wang, S. M. Zakeeruddin, P. Comte, R. Charvet, R. Humphry-Baker and M. Gratzel, J. Phys. Chem. B, 2003, 107, 14336–14341.
171. K. L. Wu, W. P. Ku, J. N. Clifford, E. Palomares, S. T. Ho, Y. Chi, S. H. Liu, P. T. Chou, M. K. Nazeeruddin and M. Gratzel, Energy Environ. Sci., 2013, 6, 859–870.
172. T. Miyasaka, T. Watanabe, A. Fujishima and K. Honda, J. Am. Chem. Soc., 1978, 100, 6657–6665.
173. B. Oregan and M. Gratzel, Nature, 1991, 353, 737–740.
174. H. Y. Li, Y. Z. Yang, Y. Q. Hou, R. L. Tang, T. N. Duan, J. N. Chen, H. Wang, H. W. Han, T. Y. Peng, X. G. Chen, Q. Q. Li and Z. Li, ACS Sustainable Chem. Eng., 2014, 2, 1776–1784.
175. W. Q. Li, B. Liu, Y. Z. Wu, S. Q. Zhu, Q. Zhang and W. H. Zhu, Dyes Pigm., 2013, 99, 176–184.
176. Y. Ooyama, Y. Hagiwara, T. Mizumo, Y. Harima and J. Ohshita, New J. Chem., 2013, 37, 2479–2485.
177. Z. S. Wang, Y. Cui, Y. Dan-oh, C. Kasada, A. Shinpo and K. Hara, J. Phys. Chem. C, 2007, 111, 7224–7230.
178. Z. J. Ning, Y. Fu and H. Tian, Energy Environ. Sci., 2010, 3, 1170–1181.
179. Y. R. Liu, J. R. Jennings, X. Z. Wang and Q. Wang, Phys. Chem. Chem. Phys., 2013, 15, 6170–6174.
180. H. P. Shen, X. Li, J. B. Li, W. L. Wang and H. Lin, Electrochim. Acta, 2013, 97, 160–166.
181. M. K. Wang, X. Li, H. Lin, P. Pechy, S. M. Zakeeruddin and M. Gratzel, Dalton Trans., 2009, 10015–10020,  10.1039/b908673k.
182. H. Lin, Y. Z. Liu, C. J. Liu, X. Li, H. P. Shen, J. Zhang, T. L. Ma and J. B. Li, J. Electroanal. Chem., 2011, 653, 81–85.
183. N. R. Neale, N. Kopidakis, J. van de Lagemaat, M. Gratzel and A. J. Frank, J. Phys. Chem. B, 2005, 109, 23183–23189.
184. X. M. Ren, Q. Y. Feng, G. Zhou, C. H. Huang and Z. S. Wang, J. Phys. Chem. C, 2010, 114, 7190–7195.
185. T. Marinado, M. Hahlin, X. A. Jiang, M. Quintana, E. M. J. Johansson, E. Gabrielsson, S. Plogmaker, D. P. Hagberg, G. Boschloo, S. M. Zakeeruddin, M. Gratzel, H. Siegbahn, L. C. Sun, A. Hagfeldt and H. Rensmo, J. Phys. Chem. C, 2010, 114, 11903–11910.
186. P. Lefebvre, B. Cariou, F. Lien, F. Kuipers and B. Staels, Physiol. Rev., 2009, 89, 147–191.
187. A. Kay and M. Gratzel, J. Phys. Chem., 1993, 97, 6272–6277.
188. Z. P. Zhang, S. M. Zakeeruddin, B. C. O'Regan, R. Humphry-Baker and M. Gratzel, J. Phys. Chem. B, 2005, 109, 21818–21824.
189. B. J. Song, H. M. Song, I. T. Choi, S. K. Kim, K. D. Seo, M. S. Kang, M. J. Lee, D. W. Cho, M. J. Ju and H. K. Kim, Chem. – Eur. J., 2011, 17, 11115–11121.
190. L. Y. Han, A. Islam, H. Chen, C. Malapaka, B. Chiranjeevi, S. F. Zhang, X. D. Yang and M. Yanagida, Energy Environ. Sci., 2012, 5, 6057–6060.
191. S. Ruhle, M. Greenshtein, S. G. Chen, A. Merson, H. Pizem, C. S. Sukenik, D. Cahen and A. Zaban, J. Phys. Chem. B, 2005, 109, 18907–18913.
192. T. Daeneke, T. H. Kwon, A. B. Holmes, N. W. Duffy, U. Bach and L. Spiccia, Nat. Chem., 2011, 3, 211–215.
193. I. R. Perera, A. Gupta, W. C. Xiang, T. Daeneke, U. Bach, R. A. Evans, C. A. Ohlin and L. Spiccia, Phys. Chem. Chem. Phys., 2014, 16, 12021–12028.
194. P. Salvatori, G. Marotta, A. Cinti, C. Anselmi, E. Mosconi and F. De Angelis, J. Phys. Chem. C, 2013, 117, 3874–3887.
195. P. Salvatori, G. Marotta, A. Cinti, E. Mosconi, M. Panigrahi, L. Giribabu, M. K. Nazeeruddin and F. De Angelis, Inorg. Chim. Acta, 2013, 406, 106–112.
196. S. Ito, H. Miura, S. Uchida, M. Takata, K. Sumioka, P. Liska, P. Comte, P. Pechy and M. Graetzel, Chem. Commun., 2008, 5194–5196,  10.1039/b809093a.
197. M. Ikeda, N. Koide, L. Han, A. Sasahara and H. Onishi, Langmuir, 2008, 24, 8056–8060.
198. A. Dessi, G. B. Consiglio, M. Calamante, G. Reginato, A. Mordini, M. Peruzzini, M. Taddei, A. Sinicropi, M. L. Parisi, F. F. de Biani, R. Basosi, R. Mori, M. Spatola, M. Bruzzi and L. Zani, Eur. J. Org. Chem., 2013, 1916–1928,  DOI:10.1002/ejoc.201201629.
199. K. Hara, Y. Dan-Oh, C. Kasada, Y. Ohga, A. Shinpo, S. Suga, K. Sayama and H. Arakawa, Langmuir, 2004, 20, 4205–4210.
200. K. Hara, H. Sugihara, Y. Tachibana, A. Islam, M. Yanagida, K. Sayama, H. Arakawa, G. Fujihashi, T. Horiguchi and T. Kinoshita, Langmuir, 2001, 17, 5992–5999.
201. K. M. Lee, V. Suryanarayanan, K. C. Ho, K. R. J. Thomas and J. T. Lin, Sol. Energy Mater. Sol. Cells, 2007, 91, 1426–1431.
202. R. Y. Y. Lin, H. W. Lin, Y. S. Yen, C. H. Chang, H. H. Chou, P. W. Chen, C. Y. Hsu, Y. C. Chen, J. T. Lin and K. C. Ho, Energy Environ. Sci., 2013, 6, 2477–2486.
203. J. H. Yum, S. J. Moon, R. Humphry-Baker, P. Walter, T. Geiger, F. Nuesch, M. Graetzel and M. D. K. Nazeeruddin, Nanotechnology, 2008, 19.
204. Y. Numata, A. Islam, H. Chen and L. Y. Han, Energy Environ. Sci., 2012, 5, 8548–8552.
205. V. S. Saji and M. Pyo, Curr. Appl. Phys., 2010, 10, S410–S413.
206. B. Liu, Q. B. Liu, D. You, X. Y. Li, Y. Naruta and W. H. Zhu, J. Mater. Chem., 2012, 22, 13348–13356.
207. H. P. Lu, C. L. Mai, C. Y. Tsia, S. J. Hsu, C. P. Hsieh, C. L. Chiu, C. Y. Yeh and E. W. Diau, Phys. Chem. Chem. Phys., 2009, 11, 10270–10274.
208. M. K. Panda, G. D. Sharma, K. R. J. Thomas and A. G. Coutsolelos, J. Mater. Chem., 2012, 22, 8092–8102.
209. S. Eu, T. Katoh, T. Umeyama, Y. Matano and H. Imahori, Dalton Trans., 2008, 5476–5483.
210. S. Mori, M. Nagata, Y. Nakahata, K. Yasuta, R. Goto, M. Kimura and M. Taya, J. Am. Chem. Soc., 2010, 132, 4054–4055.
211. Y. Nagasawa and H. Miyasaka, Phys. Chem. Chem. Phys., 2014, 16, 13008–13026.
212. B. S. Brunschwig and N. Sutin, Coord. Chem. Rev., 1999, 187, 233–254.
213. F. D. Lewis, T. Wu, Y. Zhang, R. L. Letsinger, S. R. Greenfield and M. R. Wasielewski, Science, 1997, 277, 673–676.
214. M. R. Wasielewski, Chem. Rev., 1992, 92, 435–461.
215. M. Gilbert and B. Albinsson, Chem. Soc. Rev., 2015, 44, 845–862.
216. M. Natali, S. Campagna and F. Scandola, Chem. Soc. Rev., 2014, 43, 4005–4018.
217. A. Arrigo, A. Santoro, M. T. Indelli, M. Natali, F. Scandola and S. Campagna, Phys. Chem. Chem. Phys., 2014, 16, 818–826.
218. B. Albinsson and J. Mårtensson, J. Photochem. Photobiol., C, 2008, 9, 138–155.
219. M. Liang and J. Chen, Chem. Soc. Rev., 2013, 42, 3453–3488.
220. A. Mahmood, Sol. Energy, 2016, 123, 127–144.
221. T. Ripolles-Sanchis, B.-C. Guo, H.-P. Wu, T.-Y. Pan, H.-W. Lee, S. R. Raga, F. Fabregat-Santiago, J. Bisquert, C.-Y. Yeh and E. W.-G. Diau, Chem. Commun., 2012, 48, 4368–4370.
222. Y. Tang, Y. Wang, X. Li, H. Ågren, W.-H. Zhu and Y. Xie, ACS Appl. Mater. Interfaces, 2015, 7, 27976–27985.
223. A. Yella, H.-W. Lee, H. N. Tsao, C. Yi, A. K. Chandiran, M. K. Nazeeruddin, E. W.-G. Diau, C.-Y. Yeh, S. M. Zakeeruddin and M. Grätzel, Science, 2011, 334, 629–634.
224. R. Argazzi, C. A. Bignozzi, T. A. Heimer, F. N. Castellano and G. J. Meyer, J. Am. Chem. Soc., 1995, 117, 11815–11816.
225. S. A. Haque, S. Handa, K. Peter, E. Palomares, M. Thelakkat and J. R. Durrant, Angew. Chem., Int. Ed. Engl., 2005, 44, 5740–5744.
226. J.-H. Yum, I. Jung, C. Baik, J. Ko, M. K. Nazeeruddin and M. Gratzel, Energy Environ. Sci., 2009, 2, 100–102.
227. P. Bonhôte, J.-E. Moser, R. Humphry-Baker, N. Vlachopoulos, S. M. Zakeeruddin, L. Walder and M. Grätzel, J. Am. Chem. Soc., 1999, 121, 1324–1336.
228. G. Li, K. Hu, C. Yi, K. L. Knappenberger, G. J. Meyer, S. I. Gorelsky and M. Shatruk, J. Phys. Chem. C, 2013, 117, 17399–17411.
229. Y. Hao, E. Gabrielsson, P. W. Lohse, W. Yang, E. M. J. Johansson, A. Hagfeldt, L. Sun and G. Boschloo, Adv. Sci., 2015, 2, 1500174.
230. A. Juris, V. Balzani, F. Barigelletti, S. Campagna, P. Belser and A. von Zelewsky, Coord. Chem. Rev., 1988, 84, 85–277.
231. K. Yuan Chiu, T. Xiang Su, J. Hong Li, T.-H. Lin, G.-S. Liou and S.-H. Cheng, J. Electroanal. Chem., 2005, 575, 95–101.
232. J. Luo, M. Xu, R. Li, K.-W. Huang, C. Jiang, Q. Qi, W. Zeng, J. Zhang, C. Chi, P. Wang and J. Wu, J. Am. Chem. Soc., 2014, 136, 265–272.
233. L. N. Ashbrook and C. M. Elliott, J. Phys. Chem. C, 2013, 117, 3853–3864.
234. Y. Hao, E. Gabrielsson, P. W. Lohse, W. X. Yang, E. M. J. Johansson, A. Hagfeldt, L. C. Sun and G. Boschloo, Adv. Sci., 2015, 2, 1500174.
235. T. A. Shoker and T. H. Ghaddar, RSC Adv., 2014, 4, 18336–18340.
236. R. Argazzi, C. A. Bignozzi, T. A. Heimer, F. N. Castellano and G. J. Meyer, J. Phys. Chem. B, 1997, 101, 2591–2597.
237. H. J. Snaith, C. S. Karthikeyan, A. Petrozza, J. Teuscher, J. E. Moser, M. K. Nazeeruddin, M. Thelakkat and M. Gratzel, J. Phys. Chem. C, 2008, 112, 7562–7566.
238. M. Song, D. Kim, K. Ihn, S. Jo and D. Kim, Nanotechnology, 2004, 15, 1861–1865.
239. S. A. Haque, S. Handa, K. Peter, E. Palomares, M. Thelakkat and J. R. Durrant, Angew. Chem., Int. Ed., 2005, 44, 5740–5744.
240. R. Sivakumar, A. Manivel, M. Melendrez, J. Martinez-Oyanedel, M. Bunster, C. Vergara and P. Manidurai, Polyhedron, 2015, 87, 135–140.
241. S. Handa, H. Wietasch, M. Thelakkat, J. R. Durrant and S. A. Haque, Chem. Commun., 2007, 1725–1727.
242. J. H. Yum, I. Jung, C. Baik, J. Ko, M. K. Nazeeruddin and M. Gratzel, Energy Environ. Sci., 2009, 2, 100–102.
243. K. Willinger, K. Fischer, R. Kisselev and M. Thelakkat, J. Mater. Chem., 2009, 19, 5364–5376.
244. A. El-Shafei, M. Hussain, A. Atiq, A. Islam and L. Han, J. Mater. Chem., 2012, 22, 24048–24056.
245. L. Giribabu, T. Bessho, M. Srinivasu, C. Vijaykumar, Y. Soujanya, V. G. Reddy, P. Y. Reddy, J. H. Yum, M. Gratzel and M. K. Nazeeruddin, Dalton Trans., 2011, 40, 4497–4504.
246. S. Lin-Qing, X. Pu. Hui, W. Xue-Song, H. Yuan-Jun, Z. Bao-Wen, C. Yi, L. Wei-Ying, Z. Jing-Bo, X. Xu-Rui and L. Yuan, Chin. J. Chem., 2003, 21, 644–649.
247. L. Q. Song, H. Y. Ding, X. S. Wang, B. W. Zhang, Y. Cao, J. Feng, X. C. Ai and J. P. Zhang, J. Photochem. Photobiol., A, 2004, 165, 137–142.
248. G. Will, G. Boschloo, R. Hoyle, S. N. Rao and D. Fitzmaurice, J. Phys. Chem. B, 1998, 102, 10272–10278.
249. G. Will, G. Boschloo, S. N. Rao and D. Fitzmaurice, J. Phys. Chem. B, 1999, 103, 8067–8079.
250. M. Buchalska, J. Kuncewicz, E. Świętek, P. Łabuz, T. Baran, G. Stochel and W. Macyk, Coord. Chem. Rev., 2013, 257, 767–775.
251. F. Odobel, Y. Pellegrin, E. A. Gibson, A. Hagfeldt, A. L. Smeigh and L. Hammarstrom, Coord. Chem. Rev., 2012, 256, 2414–2423.
252. J. Warnan, J. Gardner, L. Le Pleux, J. Petersson, Y. Pellegrin, E. Blart, L. Hammarstrom and F. Odobel, J. Phys. Chem. C, 2014, 118, 103–113.
253. Z. Ji and Y. Wu, J. Phys. Chem. C, 2013, 117, 18315–18324.
254. Z. Ji, G. Natu and Y. Wu, ACS Appl. Mater. Interfaces, 2013, 5, 8641–8648.
255. Z. Ji, G. Natu, Z. Huang, O. Kokhan, X. Zhang and Y. Wu, J. Phys. Chem. C, 2012, 116, 16854–16863.
256. M. He, Z. Ji, Z. Huang and Y. Wu, J. Phys. Chem. C, 2014, 118, 16518–16525.
257. L. Favereau, J. Warnan, Y. Pellegrin, E. Blart, M. Boujtita, D. Jacquemin and F. Odobel, Chem. Commun., 2013, 49, 8018–8020.
258. Y. Farré, L. Zhang, Y. Pellegrin, A. Planchat, E. Blart, M. Boujtita, L. Hammarström, D. Jacquemin and F. Odobel, J. Phys. Chem. C, 2016, 120, 7923–7940.
259. T. T. T. Pham, S. K. Saha, D. Provost, Y. Farré, M. Raissi, Y. Pellegrin, E. Blart, S. Vedraine, B. Ratier, D. Aldakov, F. Odobel and J. Bouclé, J. Phys. Chem. C, 2017, 121, 129–139.
260. A. Nattestad, Y. Y. Cheng, R. W. MacQueen, T. F. Schulze, F. W. Thompson, A. J. Mozer, B. Fückel, T. Khoury, M. J. Crossley, K. Lips, G. G. Wallace and T. W. Schmidt, J. Phys. Chem. Lett., 2013, 4, 2073–2078.
261. L. Le Pleux, A. L. Smeigh, E. Gibson, Y. Pellegrin, E. Blart, G. Boschloo, A. Hagfeldt, L. Hammarstrom and F. Odobel, Energy Environ. Sci., 2011, 4, 2075–2084.
262. C. J. Wood, M. Cheng, C. A. Clark, R. Horvath, I. P. Clark, M. L. Hamilton, M. Towrie, M. W. George, L. Sun, X. Yang and E. A. Gibson, J. Phys. Chem. C, 2014, 118, 16536–16546.
263. R. Hoyle, J. Sotomayor, G. Will and D. Fitzmaurice, J. Phys. Chem. B, 1997, 101, 10791–10800.
264. R. Hoyle, G. Will and D. Fitzmaurice, J. Mater. Chem., 1998, 8, 2033–2036.
265. J. Sotomayor, G. Will and D. Fitzmaurice, J. Mater. Chem., 2000, 10, 685–692.
266. H. M. Song, K. D. Seo, M. S. Kang, I. T. Choi, S. K. Kim, Y. K. Eom, J. H. Ryu, M. J. Ju and H. K. Kim, J. Mater. Chem., 2012, 22, 3786–3794.
267. I. T. Choi, M. J. Ju, S. H. Kang, M. S. Kang, B. S. You, J. Y. Hong, Y. K. Eom, S. H. Song and H. K. Kim, J. Mater. Chem. A, 2013, 1, 9114–9121.
268. I. T. Choi, M. J. Ju, S. H. Song, S. G. Kim, D. W. Cho, C. Im and H. K. Kim, Chem. – Eur. J., 2013, 19, 15545–15555.
269. Z. Jinlong, H. Bin, Z. Qin and H. Deyin, Dyes Pigm., 1998, 38, 41–47.
270. M. Biancardo, R. Argazzi and C. A. Bignozzi, Inorg. Chem., 2005, 44, 9619–9621.
271. V. V. Matylitsky, L. Dworak and J. Wachtveitl, ChemPhysChem, 2010, 11, 2027–2035.
272. K. Stranius, L. George, A. Efimov, T.-P. Ruoko, J. Pohjola and N. V. Tkachenko, Langmuir, 2015, 31, 944–952.
273. T. Hasobe, Phys. Chem. Chem. Phys., 2010, 12, 44–57.
274. H. Imahori and T. Umeyama, J. Phys. Chem. C, 2009, 113, 9029–9039.
275. H. Imahori, Angew. Chem., Int. Ed., 2002, 41, 2344–2347.
276. T. Hasobe, H. Imahori, S. Fukuzumi and P. V. Kamat, J. Phys. Chem. B, 2003, 107, 12105–12112.
277. H. Imahori, J. Mater. Chem., 2007, 17, 31–41.
278. T. Hasobe, H. Murata, S. Fukuzumi and P. V. Kamat, Mol. Cryst. Liq. Cryst., 2007, 471, 39–51.
279. S. Fukuzumi and T. Kojima, J. Mater. Chem., 2008, 18, 1427–1439.
280. T. Hasobe, H. Imahori, P. V. Kamat and S. Fukuzumi, J. Am. Chem. Soc., 2003, 125, 14962–14963.
281. T. Hasobe, Y. Kashiwagi, M. A. Absalom, J. Sly, K. Hosomizu, M. J. Crossley, H. Imahori, P. V. Kamat and S. Fukuzumi, Adv. Mater., 2004, 16, 975–979.
282. H. Imahori, A. Fujimoto, S. Kang, H. Hotta, K. Yoshida, T. Umeyama, Y. Matano and S. Isoda, Adv. Mater., 2005, 17, 1727–1730.
283. T. Hasobe, H. Imahori, P. V. Kamat, T. K. Ahn, S. K. Kim, D. Kim, A. Fujimoto, T. Hirakawa and S. Fukuzumi, J. Am. Chem. Soc., 2005, 127, 1216–1228.
284. T. Hasobe, P. V. Kamat, V. Troiani, N. Solladié, T. K. Ahn, S. K. Kim, D. Kim, A. Kongkanand, S. Kuwabata and S. Fukuzumi, J. Phys. Chem. B, 2005, 109, 19–23.
285. T. Hasobe, K. Saito, P. V. Kamat, V. Troiani, H. Qiu, N. Solladie, K. S. Kim, J. K. Park, D. Kim, F. D'Souza and S. Fukuzumi, J. Mater. Chem., 2007, 17, 4160–4170.
286. H. Imahori, T. Hasobe, H. Yamada, P. V. Kamat, S. Barazzouk, M. Fujitsuka, O. Ito and S. Fukuzumi, Chem. Lett., 2001, 784–785.
287. H. Imahori and S. Fukuzumi, Adv. Funct. Mater., 2004, 14, 525–536.
288. A. Kira, T. Umeyama, Y. Matano, K. Yoshida, S. Isoda, J. K. Park, D. Kim and H. Imahori, J. Am. Chem. Soc., 2009, 131, 3198–3200.
289. H. Imahori, J.-C. Liu, H. Hotta, A. Kira, T. Umeyama, Y. Matano, G. Li, S. Ye, M. Isosomppi, N. V. Tkachenko and H. Lemmetyinen, J. Phys. Chem. B, 2005, 109, 18465–18474.
290. H. Hotta, S. Kang, T. Umeyama, Y. Matano, K. Yoshida, S. Isoda and H. Imahori, J. Phys. Chem. B, 2005, 109, 5700–5706.
291. H. Imahori, M. Ueda, S. Kang, H. Hayashi, S. Hayashi, H. Kaji, S. Seki, A. Saeki, S. Tagawa, T. Umeyama, Y. Matano, K. Yoshida, S. Isoda, M. Shiro, N. V. Tkachenko and H. Lemmetyinen, Chem. – Eur. J., 2007, 13, 10182–10193.
292. A. Kira, M. Tanaka, T. Umeyama, Y. Matano, N. Yoshimoto, Y. Zhang, S. Ye, H. Lehtivuori, N. V. Tkachenko, H. Lemmetyinen and H. Imahori, J. Phys. Chem. C, 2007, 111, 13618–13626.
293. H. Hayashi, A. Kira, T. Umeyama, Y. Matano, P. Charoensirithavorn, T. Sagawa, S. Yoshikawa, N. V. Tkachenko, H. Lemmetyinen and H. Imahori, J. Phys. Chem. C, 2009, 113, 10819–10828.
294. A. Varotto, L. Todaro, M. Vinodu, J. Koehne, G.-Y. Liu and C. M. Drain, Chem. Commun., 2008, 4921–4923.
295. T. Hasobe, S. Hattori, H. Kotani, K. Ohkubo, K. Hosomizu, H. Imahori, P. V. Kamat and S. Fukuzumi, Org. Lett., 2004, 6, 3103–3106.
296. T. Hasobe, Y. Kashiwagi, M. A. Absalom, J. Sly, K. Hosomizu, M. J. Crossley, H. Imahori, P. V. Kamat and S. Fukuzumi, Adv. Mater., 2004, 16, 975–979.
297. T. Hasobe, H. Imahori, P. V. Kamat, T. K. Ahn, S. K. Kim, D. Kim, A. Fujimoto, T. Hirakawa and S. Fukuzumi, J. Am. Chem. Soc., 2005, 127, 1216–1228.
298. Y. Ooyama, K. Uenaka, T. Kamimura, S. Ozako, M. Kanda, T. Koide and F. Tani, RSC Adv., 2016, 6, 16150–16158.
299. H. Imahori and T. Umeyama, J. Phys. Chem. C, 2009, 113, 9029–9039.
300. H. Imahori and S. Fukuzumi, Adv. Funct. Mater., 2004, 14, 525–536.
301. B. H. Farnum, K.-R. Wee and T. J. Meyer, Nat. Chem., 2016, 8, 845–852.
302. B. Shan, B. H. Farnum, K.-R. Wee and T. J. Meyer, J. Phys. Chem. C, 2017, 121, 5882–5890.
303. G. N. Lim, S. Hedström, K. A. Jung, P. A. D. Smith, V. S. Batista, F. D’Souza, A. van der Est and P. K. Poddutoori, J. Phys. Chem. C, 2017, 121, 14484–14497.
304. W. Shockley and H. J. Queisser, J. Appl. Phys., 1961, 32, 510–519.
305. R. T. Ross and A. J. Nozik, J. Appl. Phys., 1982, 53, 3813–3818.
306. H. Gerischer and F. Willig, in Physical and Chemical Applications of Dyestuffs, ed. F. P. Schäfer, H. Gerischer, F. Willig, H. Meier, H. Jahnke, M. Schönborn and G. Zimmermann, Springer Berlin Heidelberg, Berlin, Heidelberg, 1976, pp. 31–84 DOI:10.1007/BFb0046057.
307. A. Kirsh-de Mesmaeker, D. Maetens and R. Nasielski-Hinkens, J. Electroanal. Chem. Interfacial Electrochem., 1985, 182, 123–132.
308. D. W. Thompson, C. A. Kelly, F. Farzad and G. J. Meyer, Langmuir, 1999, 15, 650–653.
309. C. A. Kelly, F. Farzad, D. W. Thompson, J. M. Stipkala and G. J. Meyer, Langmuir, 1999, 15, 7047–7054.
310. B. V. Bergeron, C. A. Kelly and G. J. Meyer, Langmuir, 2003, 19, 8389–8394.
311. B. V. Bergeron, A. Marton, G. Oskam and G. J. Meyer, J. Phys. Chem. B, 2005, 109, 937–943.
312. P. V. Kamat, S. Barazzouk, K. G. Thomas and S. Hotchandani, J. Phys. Chem. B, 2000, 104, 4014–4017.
313. C. C. Clark, A. Marton and G. J. Meyer, Inorg. Chem., 2005, 44, 3383–3385.
314. A. Marton, C. C. Clark, R. Srinivasan, R. E. Freundlich, A. A. Narducci Sarjeant and G. J. Meyer, Inorg. Chem., 2006, 45, 362–369.
315. P. Wang, B. Wenger, R. Humphry-Baker, J.-E. Moser, J. Teuscher, W. Kantlehner, J. Mezger, E. V. Stoyanov, S. M. Zakeeruddin and M. Grätzel, J. Am. Chem. Soc., 2005, 127, 6850–6856.
316. S. A. Berhe, H. B. Gobeze, S. D. Pokharel, E. Park and W. J. Youngblood, ACS Appl. Mater. Interfaces, 2014, 6, 10696–10705.
317. B. Henderson and G. F. Imbusch, Optical spectroscopy of inorganic solids, Clarendon Press, Oxford University Press, Oxford, Oxfordshire, New York, 1989.
318. N. Sharifi, F. Tajabadi and N. Taghavinia, ChemPhysChem, 2014, 15, 3902–3927.
319. B. M. Klahr and T. W. Hamann, J. Phys. Chem. C, 2009, 113, 14040–14045.
320. H. Deng, H. Mao, Z. Lu, J. Li and H. Xu, J. Photochem. Photobiol., A, 1997, 110, 47–52.
321. D. Huihua, Z. Hong, L. Bingjie and L. Zuhong, Jpn. J. Appl. Phys., 1998, 37, L132.
322. H. Deng, Y. Zhou, H. Mao and Z. Lu, Synth. Met., 1998, 92, 269–274.
323. T. Hasobe, H. Imahori, S. Fukuzumi and P. V. Kamat, J. Phys. Chem. B, 2003, 107, 12105–12112.
324. T. Horiuchi, H. Miura and S. Uchida, Chem. Commun., 2003, 3036–3037.
325. H. Otaka, M. Kira, K. Yano, S. Ito, H. Mitekura, T. Kawata and F. Matsui, J. Photochem. Photobiol., A, 2004, 164, 67–73.
326. V. P. S. Perera, P. K. D. D. P. Pitigala, M. K. I. Senevirathne and K. Tennakone, Sol. Energy Mater. Sol. Cells, 2005, 85, 91–98.
327. S. Kim, J. K. Lee, S. O. Kang, J. Ko, J. H. Yum, S. Fantacci, F. De Angelis, D. Di Censo, M. K. Nazeeruddin and M. Grätzel, J. Am. Chem. Soc., 2006, 128, 16701–16707.
328. J.-J. Cid, J.-H. Yum, S.-R. Jang, M. K. Nazeeruddin, E. Martínez-Ferrero, E. Palomares, J. Ko, M. Grätzel and T. Torres, Angew. Chem., Int. Ed., 2007, 46, 8358–8362.
329. K. Lee, S. W. Park, M. J. Ko, K. Kim and N.-G. Park, Nat. Mater., 2009, 8, 665–671.
330. T. Bessho, S. M. Zakeeruddin, C.-Y. Yeh, E. W.-G. Diau and M. Grätzel, Angew. Chem., Int. Ed., 2010, 49, 6646–6649.
331. P. J. Holliman, M. L. Davies, A. Connell, B. V. Velasco and T. M. Watson, Chem. Commun., 2010, 46, 7256–7258.
332. X.-F. Wang, Y. Koyama, O. Kitao, Y. Wada, S.-I. Sasaki, H. Tamiaki and H. Zhou, Biosens. Bioelectron., 2010, 25, 1970–1976.
333. M. Planells, L. Pelleja, P. Ballester and E. Palomares, Energy Environ. Sci., 2011, 4, 528–534.
334. K.-M. Lee, Y.-C. Hsu, M. Ikegami, T. Miyasaka, K. R. Justin Thomas, J. T. Lin and K.-C. Ho, J. Power Sources, 2011, 196, 2416–2421.
335. F. Huang, D. Chen, L. Cao, R. A. Caruso and Y.-B. Cheng, Energy Environ. Sci., 2011, 4, 2803–2806.
336. M. J. Griffith, A. J. Mozer, G. Tsekouras, Y. Dong, P. Wagner, K. Wagner, G. G. Wallace, S. Mori and D. L. Officer, Appl. Phys. Lett., 2011, 98, 163502.
337. P. J. Holliman, M. Mohsen, A. Connell, M. L. Davies, K. Al-Salihi, M. B. Pitak, G. J. Tizzard, S. J. Coles, R. W. Harrington, W. Clegg, C. Serpa, O. H. Fontes, C. Charbonneau and M. J. Carnie, J. Mater. Chem., 2012, 22, 13318–13327.
338. M. Kimura, H. Nomoto, N. Masaki and S. Mori, Angew. Chem., Int. Ed., 2012, 51, 4371–4374.
339. V. Saxena, P. Veerender, A. K. Chauhan, P. Jha, D. K. Aswal and S. K. Gupta, Appl. Phys. Lett., 2012, 100, 133303.
340. R. Yeh-Yung Lin, H.-W. Lin, Y.-S. Yen, C.-H. Chang, H.-H. Chou, P.-W. Chen, C.-Y. Hsu, Y.-C. Chen, J. T. Lin and K.-C. Ho, Energy Environ. Sci., 2013, 6, 2477–2486.
341. L. Wei, Y. Yang, R. Fan, P. Wang, L. Li, J. Yu, B. Yang and W. Cao, RSC Adv., 2013, 3, 25908–25916.
342. X. Lu, T. Lan, Z. Qin, Z.-S. Wang and G. Zhou, ACS Appl. Mater. Interfaces, 2014, 6, 19308–19317.
343. L. Yang, W. W.-F. Leung and J. Wang, Nanoscale, 2013, 5, 7493–7498.
344. A. Islam, M. Akhtaruzzaman, T. H. Chowdhury, C. Qin, L. Han, I. M. Bedja, R. Stalder, K. S. Schanze and J. R. Reynolds, ACS Appl. Mater. Interfaces, 2016, 8, 4616–4623.
345. J.-H. Yum, S.-R. Jang, P. Walter, T. Geiger, F. Nuesch, S. Kim, J. Ko, M. Gratzel and M. K. Nazeeruddin, Chem. Commun., 2007, 4680–4682.
346. K. Kakiage, Y. Aoyama, T. Yano, K. Oya, J.-i. Fujisawa and M. Hanaya, Chem. Commun., 2015, 51, 15894–15897.
347. S. Chang, H. Wang, Y. Hua, Q. Li, X. Xiao, W.-K. Wong, W. Y. Wong, X. Zhu and T. Chen, J. Mater. Chem. A, 2013, 1, 11553–11558.
348. C.-L. Wang, J.-Y. Hu, C.-H. Wu, H.-H. Kuo, Y.-C. Chang, Z.-J. Lan, H.-P. Wu, E. Wei-Guang Diau and C.-Y. Lin, Energy Environ. Sci., 2014, 7, 1392–1396.
349. Y. Wang, B. Chen, W. Wu, X. Li, W. Zhu, H. Tian and Y. Xie, Angew. Chem., Int. Ed., 2014, 53, 10779–10783.
350. C.-M. Lan, H.-P. Wu, T.-Y. Pan, C.-W. Chang, W.-S. Chao, C.-T. Chen, C.-L. Wang, C.-Y. Lin and E. W.-G. Diau, Energy Environ. Sci., 2012, 5, 6460–6464.
351. R. Y. Ogura, S. Nakane, M. Morooka, M. Orihashi, Y. Suzuki and K. Noda, Appl. Phys. Lett., 2009, 94, 073308.
352. H. Ozawa, R. Shimizu and H. Arakawa, RSC Adv., 2012, 2, 3198–3200.
353. D. D. Babu, D. Elsherbiny, H. Cheema, A. El-Shafei and A. V. Adhikari, Dyes Pigm., 2016, 132, 316–328.
354. S.-Q. Fan, C. Kim, B. Fang, K.-X. Liao, G.-J. Yang, C.-J. Li, J.-J. Kim and J. Ko, J. Phys. Chem. C, 2011, 115, 7747–7754.
355. R. Y.-Y. Lin, Y.-S. Yen, Y.-T. Cheng, C.-P. Lee, Y.-C. Hsu, H.-H. Chou, C.-Y. Hsu, Y.-C. Chen, J. T. Lin, K.-C. Ho and C. Tsai, Org. Lett., 2012, 14, 3612–3615.
356. H.-P. Wu, Z.-W. Ou, T.-Y. Pan, C.-M. Lan, W.-K. Huang, H.-W. Lee, N. M. Reddy, C.-T. Chen, W.-S. Chao, C.-Y. Yeh and E. W.-G. Diau, Energy Environ. Sci., 2012, 5, 9843–9848.
357. Y. Chen, Z. Zeng, C. Li, W. Wang, X. Wang and B. Zhang, New J. Chem., 2005, 29, 773–776.
358. J. Chang, C.-P. Lee, D. Kumar, P.-W. Chen, L.-Y. Lin, K. R. J. Thomas and K.-C. Ho, J. Power Sources, 2013, 240, 779–785.
359. J. H. Delcamp, Y. Shi, J.-H. Yum, T. Sajoto, E. Dell'Orto, S. Barlow, M. K. Nazeeruddin, S. R. Marder and M. Grätzel, Chem. – Eur. J., 2013, 19, 1819–1827.
360. A. Connell, P. J. Holliman, M. L. Davies, C. D. Gwenin, S. Weiss, M. B. Pitak, P. N. Horton, S. J. Coles and G. Cooke, J. Mater. Chem. A, 2014, 2, 4055–4066.
361. M. D. Brown, P. Parkinson, T. Torres, H. Miura, L. M. Herz and H. J. Snaith, J. Phys. Chem. C, 2011, 115, 23204–23208.
362. C. Siegers, U. Würfel, M. Zistler, H. Gores, J. Hohl-Ebinger, A. Hinsch and R. Haag, ChemPhysChem, 2008, 9, 793–798.
363. M. Shrestha, L. Si, C.-W. Chang, H. He, A. Sykes, C.-Y. Lin and E. W.-G. Diau, J. Phys. Chem. C, 2012, 116, 10451–10460.
364. J. N. Clifford, A. Forneli, H. Chen, T. Torres, S. Tan and E. Palomares, J. Mater. Chem., 2011, 21, 1693–1696.
365. B. E. Hardin, A. Sellinger, T. Moehl, R. Humphry-Baker, J.-E. Moser, P. Wang, S. M. Zakeeruddin, M. Grätzel and M. D. McGehee, J. Am. Chem. Soc., 2011, 133, 10662–10667.
366. L. Martin-Gomis, F. Fernandez-Lazaro and A. Sastre-Santos, J. Mater. Chem. A, 2014, 2, 15672–15682.
367. V. Balzani, L. Moggi and F. Scandola, in Supramolecular Photochemistry, ed. V. Balzani, Springer Netherlands, Dordrecht, 1987, pp. 1–28 DOI:10.1007/978-94-009-3979-0_1.
368. D. Astruc, E. Boisselier and C. Ornelas, Chem. Rev., 2010, 110, 1857–1959.
369. T. Mirkovic, E. E. Ostroumov, J. M. Anna, R. van Grondelle, Govindjee and G. D. Scholes, Chem. Rev., 2017, 117, 249–293.
370. R. Amadelli, R. Argazzi, C. A. Bignozzi and F. Scandola, J. Am. Chem. Soc., 1990, 112, 7099–7103.
371. M. K. Nazeeruddin, P. Liska, J. Moser, N. Vlachopoulos and M. Grätzel, Helv. Chim. Acta, 1990, 73, 1788–1803.
372. C. A. Bignozzi, R. Argazzi, J. R. Schoonover, G. J. Meyer and F. Scandola, Sol. Energy Mater. Sol. Cells, 1995, 38, 187–198.
373. M. K. Nazeeruddin, R. Humphry-Baker, M. Gratzel, M. K. Nazeeruddin, R. Humphry-Baker, M. Gratzel and B. A. Murrer, Chem. Commun., 1998, 719–720.
374. T. Rawling, C. Austin, F. Buchholz, S. B. Colbran and A. M. McDonagh, Inorg. Chem., 2009, 48, 3215–3227.
375. C. J. Kleverlaan, M. T. Indelli, C. A. Bignozzi, L. Pavanin, F. Scandola, G. M. Hasselman and G. J. Meyer, J. Am. Chem. Soc., 2000, 122, 2840–2849.
376. D. K. Panda, F. S. Goodson, S. Ray, R. Lowell and S. Saha, Chem. Commun., 2012, 48, 8775–8777.
377. D. K. Panda, F. S. Goodson, S. Ray and S. Saha, Chem. Commun., 2014, 50, 5358–5360.
378. H. Lee, L. J. Kepley, H. G. Hong and T. E. Mallouk, J. Am. Chem. Soc., 1988, 110, 618–620.
379. T. Ishida, K.-i. Terada, K. Hasegawa, H. Kuwahata, K. Kusama, R. Sato, M. Nakano, Y. Naitoh and M.-a. Haga, Appl. Surf. Sci., 2009, 255, 8824–8830.
380. O. O. Ogunsolu, I. A. Murphy, J. C. Wang, A. Das and K. Hanson, ACS Appl. Mater. Interfaces, 2016, 8, 28633–28640.
381. D. Y. Lee, I. Lim, C. Y. Shin, S. A. Patil, W. Lee, N. K. Shrestha, J. K. Lee and S.-H. Han, J. Mater. Chem. A, 2015, 3, 22669–22676.
382. W. A. Maza, A. J. Haring, S. R. Ahrenholtz, C. C. Epley, S. Y. Lin and A. J. Morris, Chem. Sci., 2016, 7, 719–727.
383. D. Y. Lee, D. V. Shinde, S. J. Yoon, K. N. Cho, W. Lee, N. K. Shrestha and S.-H. Han, J. Phys. Chem. C, 2014, 118, 16328–16334.
384. D. Y. Lee, E.-K. Kim, C. Y. Shin, D. V. Shinde, W. Lee, N. K. Shrestha, J. K. Lee and S.-H. Han, RSC Adv., 2014, 4, 12037–12042.
385. H. C. Kolb, M. G. Finn and K. B. Sharpless, Angew. Chem., Int. Ed., 2001, 40, 2004–2021.
386. T. Duan, K. Fan, Y. Fu, C. Zhong, X. Chen, T. Peng and J. Qin, Dyes Pigm., 2012, 94, 28–33.
387. T. Duan, K. Fan, K. Li, W. Cao, C. Zhong, X. Chen, T. Peng and J. Qin, Dyes Pigm., 2015, 117, 108–115.
388. I. Stengel, N. Pootrakulchote, R. R. Dykeman, A. Mishra, S. M. Zakeeruddin, P. J. Dyson, M. Grätzel and P. Bäuerle, Adv. Energy Mater., 2012, 2, 1004–1012.
389. V. Nikolaou, P. A. Angaridis, G. Charalambidis, G. D. Sharma and A. G. Coutsolelos, Dalton Trans., 2015, 44, 1734–1747.
390. E. E. Beauvilliers, M. R. Topka and P. H. Dinolfo, RSC Adv., 2014, 4, 32866–32875.
391. P. K. B. Palomaki and P. H. Dinolfo, Langmuir, 2010, 26, 9677–9685.
392. P. K. B. Palomaki and P. H. Dinolfo, ACS Appl. Mater. Interfaces, 2011, 3, 4703–4713.
393. P. K. B. Palomaki, M. R. Civic and P. H. Dinolfo, ACS Appl. Mater. Interfaces, 2013, 5, 7604–7612.
394. E. L. Unger, S. J. Fretz, B. Lim, G. Y. Margulis, M. D. McGehee and T. D. P. Stack, Phys. Chem. Chem. Phys., 2015, 17, 6565–6571.
395. P. K. D. D. P. Pitigala, M. K. I. Seneviratne, V. P. S. Perera and K. Tennakone, Langmuir, 2004, 20, 5100–5103.
396. J.-F. Gnichwitz, R. Marczak, F. Werner, N. Lang, N. Jux, D. M. Guldi, W. Peukert and A. Hirsch, J. Am. Chem. Soc., 2010, 132, 17910–17920.
397. N. K. Subbaiyan and F. D'Souza, Chem. Commun., 2012, 48, 3641–3643.
398. Y. Kawashima, K. Ohkubo, V. M. Blas-Ferrando, H. Sakai, E. Font-Sanchis, J. Ortíz, F. Fernández-Lázaro, T. Hasobe, Á. Sastre-Santos and S. Fukuzumi, J. Phys. Chem. B, 2015, 119, 7690–7697.
399. G. Charalambidis, K. Karikis, E. Georgilis, B. L. M'Sabah, Y. Pellegrin, A. Planchat, B. Lucas, A. Mitraki, J. Boucle, F. Odobel and A. G. Coutsolelos, Sustainable Energy Fuels, 2017, 1, 387–395.
400. S. K. Balasingam, M. Lee, M. G. Kang and Y. Jun, Chem. Commun., 2013, 49, 1471–1487.
401. Y. Y. Cheng, A. Nattestad, T. F. Schulze, R. W. MacQueen, B. Fuckel, K. Lips, G. G. Wallace, T. Khoury, M. J. Crossley and T. W. Schmidt, Chem. Sci., 2016, 7, 559–568.
402. T.-H. Wang, T.-W. Huang, Y.-C. Tsai, Y.-W. Chang and C.-S. Liao, Chem. Commun., 2015, 51, 7253–7256.
403. G. Griffini, F. Bella, F. Nisic, C. Dragonetti, D. Roberto, M. Levi, R. Bongiovanni and S. Turri, Adv. Energy Mater., 2015, 5, 1401312.
404. M. J. Y. Tayebjee, D. R. McCamey and T. W. Schmidt, J. Phys. Chem. Lett., 2015, 6, 2367–2378.
405. F. Inakazu, Y. Noma, Y. Ogomi and S. Hayase, Appl. Phys. Lett., 2008, 93, 093304.
406. N. Yusuke, I. Keita, O. Yuhei, S. P. Shyam and H. Shuzi, Jpn. J. Appl. Phys., 2009, 48, 020213.
407. P. Se Woong, L. Kyungtae, L. Doh-Kwon, K. Min Jae, P. Nam-Gyu and K. Kyungkon, Nanotechnology, 2011, 22, 045201.
408. F. Lodermeyer, R. D. Costa, J. Malig, N. Jux and D. M. Guldi, Chem. – Eur. J., 2016, 22, 7851–7855.
409. Q. Miao, L. Wu, J. Cui, M. Huang and T. Ma, Adv. Mater., 2011, 23, 2764–2768.
410. E. Y. Kim, S. Yu, J. H. Moon, S. M. Yoo, C. Kim, H. K. Kim and W. I. Lee, Electrochim. Acta, 2013, 111, 261–267.
411. J. Lim, M. Lee, S. K. Balasingam, J. Kim, D. Kim and Y. Jun, RSC Adv., 2013, 3, 4801–4805.
412. K. Cao, J. Lu, H. Li, Y. Shen and M. Wang, RSC Adv., 2015, 5, 62630–62637.
413. G.-J. Yang, S.-Q. Fan, C.-J. Li and C.-X. Li, Nanosci. Nanotechnol. Lett., 2011, 3, 483–486.
414. J. Bandara, K. Willinger and M. Thelakkat, Phys. Chem. Chem. Phys., 2011, 13, 12906–12911.
415. N. C. Jeong, H.-J. Son, C. Prasittichai, C. Y. Lee, R. A. Jensen, O. K. Farha and J. T. Hupp, J. Am. Chem. Soc., 2012, 134, 19820–19827.
416. Z. Xue, L. Wang and B. Liu, Nanoscale, 2013, 5, 2269–2273.
417. J. N. Clifford, E. Palomares, M. K. Nazeeruddin, R. Thampi, M. Grätzel and J. R. Durrant, J. Am. Chem. Soc., 2004, 126, 5670–5671.
418. H. Choi, S. Kim, S. O. Kang, J. Ko, M. S. Kang, J. N. Clifford, A. Forneli, E. Palomares, M. K. Nazeeruddin and M. Grätzel, Angew. Chem., Int. Ed., 2008, 47, 8259–8263.
419. M. Dürr, A. Bamedi, A. Yasuda and G. Nelles, Appl. Phys. Lett., 2004, 84, 3397–3399.
420. M. Yanagida, N. Onozawa-Komatsuzaki, M. Kurashige, K. Sayama and H. Sugihara, Sol. Energy Mater. Sol. Cells, 2010, 94, 297–302.
421. T. Yamaguchi, Y. Uchida, S. Agatsuma and H. Arakawa, Sol. Energy Mater. Sol. Cells, 2009, 93, 733–736.
422. W. S. Choi, I. T. Choi, B. S. You, J.-W. Yang, M. J. Ju and H. K. Kim, Isr. J. Chem., 2015, 55, 1002–1010.
423. B. Ajay Kumar, F. Naotaka, H. Azwar, O. Yuhei, S. P. Shyam, M. Tingli and H. Shuzi, Appl. Phys. Express, 2015, 8, 102301.
424. W. Kubo, A. Sakamoto, T. Kitamura, Y. Wada and S. Yanagida, J. Photochem. Photobiol., A, 2004, 164, 33–39.
425. X. D. Zhang, W. M. Liao, W. Mu, D. J. Zheng, Y. S. Zhou, B. L. Xue, W. Liu, Z. Q. Lin and Y. L. Deng, J. Mater. Chem. A, 2014, 2, 11035–11039.
426. K. Uzaki, S. S. Pandey and S. Hayase, J. Photochem. Photobiol., A, 2010, 216, 104–109.
427. U. Kenshiro, S. P. Shyam, O. Yuhei and H. Shuzi, Jpn. J. Appl. Phys., 2010, 49, 082301.
428. J. He, H. Lindström, A. Hagfeldt and S.-E. Lindquist, Sol. Energy Mater. Sol. Cells, 2000, 62, 265–273.
429. F. Odobel, L. Le Pleux, Y. Pellegrin and E. Blart, Acc. Chem. Res., 2010, 43, 1063–1071.
430. A. Nakasa, H. Usami, S. Sumikura, S. Hasegawa, T. Koyama and E. Suzuki, Chem. Lett., 2005, 34, 500–501.
431. E. A. Gibson, A. L. Smeigh, L. Le Pleux, J. Fortage, G. Boschloo, E. Blart, Y. Pellegrin, F. Odobel, A. Hagfeldt and L. Hammarström, Angew. Chem., Int. Ed., 2009, 48, 4402–4405.
432. S. Powar, R. Bhargava, T. Daeneke, G. Götz, P. Bäuerle, T. Geiger, S. Kuster, F. A. Nüesch, L. Spiccia and U. Bach, Electrochim. Acta, 2015, 182, 458–463.
433. C. J. Wood, G. H. Summers and E. A. Gibson, Chem. Commun., 2015, 51, 3915–3918.
434. H. Choi, T. Hwang, S. Lee, S. Nam, J. Kang, B. Lee and B. Park, J. Power Sources, 2015, 274, 937–942.
435. A. Nattestad, A. J. Mozer, M. K. R. Fischer, Y. B. Cheng, A. Mishra, P. Bauerle and U. Bach, Nat. Mater., 2010, 9, 31–35.
436. C. J. Cram and J. M. Cram, Science, 1974, 183, 803–809.
437. D. J. Cram, Angew. Chem., Int. Ed. Engl., 1988, 27, 1009–1020.
438. G. Wenz, Clin. Drug Invest., 2000, 19, 21–25.
439. M. Freitag and E. Galoppini, Energy Environ. Sci., 2011, 4, 2482–2494.
440. P. Piotrowiak, K. Deshayes, Z. S. Romanova, C. Pagba, S. Hore, G. Zordan, I. Place and A. Farrán, Pure Appl. Chem., 2003, 75, 1061–1068.
441. C. Pagba, G. Zordan, E. Galoppini, E. L. Piatnitski, S. Hore, K. Deshayes and P. Piotrowiak, J. Am. Chem. Soc., 2004, 126, 9888–9889.
442. S. A. Haque, J. S. Park, M. Srinivasarao and J. R. Durrant, Adv. Mater., 2004, 16, 1177–1181.
443. H. Choi, S. O. Kang, J. Ko, G. Gao, H. S. Kang, M.-S. Kang, M. K. Nazeeruddin and M. Grätzel, Angew. Chem., Int. Ed., 2009, 48, 5938–5941.
444. J. A. Bonacin, S. H. Toma, J. N. Freitas, A. F. Nogueira and H. E. Toma, Inorg. Chem. Commun., 2013, 36, 35–38.
445. S. H. Toma, J. A. Bonacin, K. Araki and H. E. Toma, Surf. Sci., 2006, 600, 4591–4597.
446. M. Porel, A. Klimczak, M. Freitag, E. Galoppini and V. Ramamurthy, Langmuir, 2012, 28, 3355–3359.
447. M. Freitag and E. Galoppini, Langmuir, 2010, 26, 8262–8269.
448. D. Cummins, G. Boschloo, M. Ryan, D. Corr, S. N. Rao and D. Fitzmaurice, J. Phys. Chem. B, 2000, 104, 11449–11459.
449. F. Campus, P. Bonhôte, M. Grätzel, S. Heinen and L. Walder, Sol. Energy Mater. Sol. Cells, 1999, 56, 281–297.
450. H.-J. Kim, W. S. Jeon, Y. H. Ko and K. Kim, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 5007–5011.
451. H. Li, F. Li, Y. Wang, L. Bai, F. Yu and L. Sun, ChemPlusChem, 2016, 81, 1056–1059.
452. W.-T. Deng, J.-C. Liu, J. Cao, D.-C. Hu, R.-Z. Li and N.-Z. Jin, Dalton Trans., 2014, 43, 626–631.
453. J. Cao, D.-C. Hu, J.-C. Liu, R.-Z. Li and N.-Z. Jin, Org. Electron., 2014, 15, 509–516.
454. J. Cao, J.-C. Liu, W.-T. Deng, R.-Z. Li and N.-Z. Jin, Org. Electron., 2013, 14, 2713–2720.
455. J. Cao, D.-C. Hu, J.-C. Liu, R.-Z. Li and N.-Z. Jin, J. Coord. Chem., 2013, 66, 4211–4219.
456. M. T. Brumbach, A. K. Boal and D. R. Wheeler, Langmuir, 2009, 25, 10685–10690.
457. N. K. Subbaiyan, C. A. Wijesinghe and F. D'Souza, J. Am. Chem. Soc., 2009, 131, 14646–14647.
458. H. B. Gobeze, S. K. Das and F. D’Souza, J. Phys. Chem. C, 2014, 118, 16660–16671.
459. N. K. Subbaiyan, J. P. Hill, K. Ariga, S. Fukuzumi and F. D'Souza, Chem. Commun., 2011, 47, 6003–6005.
460. C. B. Kc, K. Stranius, P. D’Souza, N. K. Subbaiyan, H. Lemmetyinen, N. V. Tkachenko and F. D’Souza, J. Phys. Chem. C, 2013, 117, 763–773.
461. L. A. Martini, G. F. Moore, R. L. Milot, L. Z. Cai, S. W. Sheehan, C. A. Schmuttenmaer, G. W. Brudvig and R. H. Crabtree, J. Phys. Chem. C, 2013, 117, 14526–14533.
462. C. F. A. Negre, R. L. Milot, L. A. Martini, W. Ding, R. H. Crabtree, C. A. Schmuttenmaer and V. S. Batista, J. Phys. Chem. C, 2013, 117, 24462–24470.
463. O. O. Ogunsolu, J. C. Wang and K. Hanson, ACS Appl. Mater. Interfaces, 2015, 7, 27730–27734.
464. J. C. Wang, I. A. Murphy and K. Hanson, J. Phys. Chem. C, 2015, 119, 3502–3508.
465. J. C. Wang, K. Violette, O. O. Ogunsolu and K. Hanson, Phys. Chem. Chem. Phys., 2017, 19, 2679–2682.
466. J. C. Wang, K. Violette, O. O. Ogunsolu, S. Cekli, E. Lambers, H. M. Fares and K. Hanson, Langmuir, 2017 DOI:10.1021/acs.langmuir.7b01964.
467. O. O. Ogunsolu, J. C. Wang and K. Hanson, Inorg. Chem., 2017 DOI:10.1021/acs.inorgchem.7b01531.

---