Volume 1 photovoltaic solar energy 1 21 – mesoporous dye sensitized solar cells

Volume 1 photovoltaic solar energy 1 21 – mesoporous dye sensitized solar cells Volume 1 photovoltaic solar energy 1 21 – mesoporous dye sensitized solar cells Volume 1 photovoltaic solar energy 1 21 – mesoporous dye sensitized solar cells Volume 1 photovoltaic solar energy 1 21 – mesoporous dye sensitized solar cells Volume 1 photovoltaic solar energy 1 21 – mesoporous dye sensitized solar cells Volume 1 photovoltaic solar energy 1 21 – mesoporous dye sensitized solar cells

A Hagfeldt, UB Cappel, and G Boschloo, Uppsala University, Uppsala, Sweden

L Sun, KTH—Royal Institute of Technology, Stockholm, Sweden; Dalian University of Technology (DUT), Dalian, China

L Kloo, KTH—Royal Institute of Technology, Stockholm, Sweden

H Pettersson, Swerea IVF AB, Mölndal, Sweden

EA Gibson, University of Nottingham, Nottingham UK

© 2012 Elsevier Ltd

1.21.2 Mesoporous Dye-Sensitized Solar Cells

1.21.2.1 Overview of Current Status and Operational Principles

1.21.2.2 The Kinetic Model – Electron-Transfer Processes

1.21.2.2.1 Electron injection and excited state decay

1.21.2.2.2 Regeneration of the oxidized dyes

1.21.2.2.3 Electron transport through the mesoporous oxide film

1.21.2.2.4 Loss reactions

1.21.2.2.5 Transport of the redox mediator and reactions at the counter electrode

1.21.2.3 Basic Characterization of DSC Devices

1.21.2.3.1 Efficiency measurements

1.21.2.3.2 External and internal quantum efficiencies

1.21.2.3.3 Overview of the concept of a DSC toolbox

1.21.2.4 Development of Material Components and Devices

1.21.2.4.1 Mesoporous oxide working electrodes

1.21.2.4.3 Electrolytes

1.21.2.4.4 Counter electrodes

1.21.2.4.5 Development of modules

References

1.21.1 Introduction

The umbrella of solar energy conversion encompasses solar thermal, solar fuels, solar-to-electricity (photovoltaic (PV)) technology, and the great many subcategories below those PVs, or solar cells, are advancing quickly, both with regard to industrialization and research Globally, the total PV installation is around 40 giga Watts (GW), and an annual growth rate of 45% has been experienced over the recent years Solar cell technologies can be divided into three generations The first is an established technology such as crystalline silicon, the second includes the emerging thin-film technologies that have just entered the market, while the third generation covers future technologies that are not yet commercialized A link for PV updates is www.solarbuzz.com, and our own contribution for a review of PV technologies with special emphasis on the materials science aspects is Reference 1

When comparing different PV technologies, a figure of merit is the production cost per peak Watt of solar electricity produced

−1 For the so-called second-generation thin-film solar cells, production costs down to and even below $1 Wpeak are reported To be competitive with conventional energy sources for large-scale electricity production, new PV technologies need to aim at production

−1 costs below $0.5 Wpeak To give an example, this means a cost of $70 m−2 at a module efficiency of 14% The dye-sensitized solar cell

−1 (DSC) is a molecular solar cell technology that has the potential to achieve production costs below $0.5 Wpeak

DSC is based on molecular- and nanometer-scale components Record cell efficiencies of 12%, promising stability data and energy-efficient production methods, have been accomplished In the present table of record solar cell efficiencies [2], in which the solar cell area must be at least 1 cm2, the record is held by the Sharp company in Japan at 10.9% [3] The record cell efficiencies for a DSC module is 9.9% achieved by Sony, Japan [4] Key advantages of the DSC technology over others include the prospect of low investments and fabrication costs and a short energy-payback time (<1 year) DSCs can be designed with great flexibility in shape, color, and transparency Because they are relatively easy to integrate into a range of different products, new commercial opportunities for niche applications have been made possible Ultimately, the most important comparison of different energy sources is the cost in relation to energy production, e.g., the production cost per kWh For DSC technology, it is advantageous to compare energy cost rather than cost per peak Watt since DSCs perform relatively better compared with other solar cell technologies under diffuse light conditions and at higher temperatures

DSC research groups have been established around the world and the field is growing fast, which can be illustrated by the fact that about two or three research articles are being published every day From a fundamental research perspective, we can conclude that the physical chemistry of several of the basic operations in the DSC device remains far from fully understood The energetics

and kinetics of many specific model and reference systems under controlled conditions have been carefully investigated However, it

is still not possible to predict accurately how a small change to the system – e.g., replacing one component or changing the electrolyte composition – will affect the DSC performance Such investigations have demonstrated the sheer chemical complexity of DSCs, and the main challenge for future research is to understand and master this complexity, in particular, at the oxide-dye­ electrolyte interface A challenging but realizable goal for the present DSC technology is to achieve efficiencies above 15% We have for many years known where the main losses in the state-of-the-art DSC device are: the potential drop in the regeneration process and the recombination loss between electrons in the TiO2 and acceptor species in the electrolyte With our breakthrough of using one-electron-transfer redox systems such as Co complexes, in combination with a dye that efficiently prevents the recombination loss, we may now have found the path to increase the efficiency significantly [5] With the recent world record of 12.3 by Grätzel and co-workers [6] using Co complexes, the main direction of the research field is now to explore this path

The industrial interest in DSCs is strong and several companies present encouraging results, in particular with regard to upscaling with world record minimodule efficiencies above 9% (Sony and Fujikura) In this context, we emphasize that world record efficiencies are not the same as stable efficiencies obtained after durability tests The stable module efficiencies that are reported vary significantly in the literature and are difficult to judge The best values listed in the literature are about 5%, although presentations at conferences report better results Several companies are dedicated to setting up manufacturing pilot lines G24i

is a company based in Cardiff, Wales, that focuses on consumer electronics On its website (www.g24i.com), such niche products are now for sale Companies that sell material components, equipment, and consultancy services have increased and are growing There are several recent reviews on DSCs, and the reader is directed to these papers for further information [7–19] In this chapter, we present the general concepts and principles of DSC, introduce the basic characterization methods, and overview materials and device development This also means that the reference list gives more of a general summary of the field, including mainly books, review articles, and some original papers

1.21.2 Mesoporous Dye-Sensitized Solar Cells

Attempts to develop mesoporous DSCs have been made before [9, 20–22] the breakthrough of O’Regan and Grätzel in 1991 [23] The basic problem was the belief that only smooth semiconductor surfaces could be used The light-harvesting efficiency (LHE) for a monomolecular layer of dye sensitizer is far less than 1% of the AM1.5 spectrum Attempts to harvest more light by using multilayers of dyes were in general unsuccessful Indications of the possibilities to increase the roughness of the semiconductor surface so that a larger number of dyes could be adsorbed directly to the surface and simultaneously be in direct contact with a redox electrolyte had also been reported before 1991 For example, Matsumura et al [24] and Alonso et al [25] used sintered ZnO electrodes to increase the efficiency of sensitization by rose bengal and related dyes But the conversion yields from solar light to electricity remained well below 1% for these systems Grätzel, Augustynski, and co-workers presented results on dye-sensitized fractal-type TiO2 electrodes with high surface area in 1985 [26] For DSC, there was thus an order-of-magnitude increase when

O’Regan and Grätzel in 1991 reported efficiencies of 7–8% [23] With regard to stability, a turnover number of 5
106 was measured for the ruthenium complex sensitizer This was followed up by the introduction of the famous N3 dye, giving efficiencies around 10% [27] For more than a decade, the ruthenium complex N3, Ru(Lbip)2 (NCS)2 with Lbip being a dicarboxylated bipyridyl ligand, its salt analog N719, and the so-called black dye RuLter(NCS)3 with L being a terpyridyl ligand, were state-of-the-art sensitizers Recently developed Ru complexes such as C101 show now higher performances both in terms of efficiency and stability [17, 28] The molecular structures of N3, the black dye, and C101 are shown in Figure 1

1.21.2.1 Overview of Current Status and Operational Principles

Since the initial work in the beginning of the 1990s, a wealth of DSC components and configurations have been developed

At present, several thousands of dyes have been investigated Fewer, but certainly hundreds of electrolyte systems and mesoporous films with different morphologies and compositions have been studied and optimized For DSCs at present, in the official table of world record efficiencies for solar cells, the record is held by the Sharp company in Japan at 10.9  0.3% [2, 3] A criterion to qualify for these tables is that the solar cell area is at least 1 cm2 For smaller cells, conversion efficiencies above 12% have been reached using the so-called C101 sensitizer as the sensitizer and with Co complex-based electrolytes [6] (see Section 1.21.2.4.3)

A schematic of the interior of a DSC showing the principle of how the device operates is shown in Figure 2

The typical configuration is as follows At the heart of the device is the mesoporous oxide layer composed of a network of TiO2 nanoparticles These are deposited on a transparent conducting oxide (TCO) substrate and an Ohmic contact is formed between them during the sintering step, which also fuses together the nanoparticles to enable electronic conduction The mesoporous film thickness is, typically, ca 10 µm and the nanoparticle diameter is 10–30 nm giving a porosity of 50–60% The mesoporous layer is deposited on a transparent conducting oxide on a glass or plastic substrate A typical scanning electron microscopy (SEM) image of a mesoporous TiO2 film is shown in Figure 3

Attached to the surface of the nanocrystalline film is a monolayer of the charge-transfer dye

Photoexcitation of the latter results in the injection of an electron into the conduction band of the oxide leaving the dye in its oxidized state The dye is restored to its ground state by electron transfer from the electrolyte, usually an organic solvent containing

COOH

S

O O–

Ru

C

COOH

COOH

Dye-sensitized

−

E

Light

e

-e

- I- /

-I3

2 Dye Electrolyte

300 nm

Figure 1 Molecular structures of N3, the black dye, and C101

Figure 2 A schematic of the interior of a DSC showing the principle of how the device operates

Figure 3 SEM picture of a typical mesoporous TiO2 film

Energy Semiconductor dye electrolyte counter electrode

ΔV

4 2 (S+/S*)

5

0 1

6

3 (I−/I

3 −) e

(S+/S)

7

e−

−

e−

−0.7

0.3 1.1

e−

2.7

V vs NHE

hν

10−8 s

10−12 s

10−13 s

10−11 10−12 s

10−3 s

10−2 s

I3 −/I−

10−4 s

I2 −/I− 10−5 s

10−6 s TiO2

the iodide–triiodide redox system The regeneration of the sensitizer by iodide intercepts the recapture of the conduction band electron by the oxidized dye The I3 − ions formed by oxidation of I− diffuse a short distance (<50 μm) through the electrolyte to the cathode, which is coated with a layer of catalytic platinum, where the regenerative cycle is completed by electron transfer to reduce I3 −

to I− For a DSC to be durable for more than 15 years outdoors, the turnover number required is 108 This is satisfied by the ruthenium complexes mentioned above

The voltage generated under illumination corresponds to the difference between the electrochemical potential of the electron at the two contacts For DSCs, this is generally the difference between the Fermi level of the mesoporous TiO2 layer and the redox potential of the electrolyte Overall, electric power is generated without permanent chemical transformation

As mentioned above, a huge number of material components, dyes, mesoporous and nanostructured electrodes, electrolytes, and counter electrodes have been synthesized and developed for DSC applications The material component variations of a DSC device are therefore endless It is important to keep in mind that the description above, and the one below in the next section, of the conventional DSC device with a mesoporous TiO2, a Ru complex sensitizer, I−/I3 − redox couple, and a platinized TCO counter electrode is only valid for this particular combination of material components As soon as one of these components are modified or completely replaced by another component, the picture has changed; energetics and kinetics are different and need to be determined for the particular system at hand To generalize a result in DSC research is therefore difficult and can many times be misleading 1.21.2.2 The Kinetic Model – Electron-Transfer Processes

The basic electron-transfer processes in a DSC, as well as the potentials for a state-of-the-art device based on the N3 dye adsorbed on TiO2 and I−/I3 − as redox couple in the electrolyte, are shown in Figure 4 The corresponding kinetic data for the different electron-transfer processes taking place at the oxide–dye–electrolyte interface are summarized in Figure 5

Figure 4 Simple energy-level diagram for a DSC The basic electron-transfer processes are indicated by numbers (1–7) The potentials for a DSC based

on the N3 dye, TiO2, and the I−/I3 − redox couple are shown Reprinted with permission from Hagfeldt A, Boschloo G, Sun L, et al (2010) Chemical Reviews 110: 6595 American Chemical Society

Figure 5 Summary of the kinetic data for the different electron-transfer processes depicted in Figure 4 Reprinted with permission from Hagfeldt A, Boschloo G, Sun L, et al (2010) Chemical Reviews 110: 6595 American Chemical Society

Besides the desired pathway of the electron-transfer processes (processes 2, 3, 4, and 7) described in Figure 4, the loss reactions 1,

5, and 6 are indicated Reaction 2 is electron injection from the excited state of the dye to the semiconductor, 3 is regeneration of the oxidized dye by the electrolyte, 4 is electron transport through the mesoporous oxide layer, and 7 is the reduction of the electrolyte

at the counter electrode Reaction 1 is direct recombination of the excited dye reflected by the excited state lifetime Recombination reactions of injected electrons in the TiO2 with either oxidized dyes or with acceptors in the electrolyte are numbered 5 and 6, respectively In principle, electron transfer to I3 − can occur either at the interface between the nanocrystalline oxide and the electrolyte or at areas of the TCO contact that are exposed to the electrolyte In practice, the second route can be suppressed by using a compact blocking layer of oxide deposited on the anode by spray pyrolysis [29, 30] Blocking layers are mandatory for DSCs that utilize one-electron redox systems or for cells using solid organic hole-conducting media [18]

1.21.2.2.1 Electron injection and excited state decay

One of the most astounding findings in DSC research is the ultrafast injection from the excited ruthenium complex in the TiO2 conduction band Although the detailed mechanism of this injection process is still under debate, it is generally accepted that a fast, femtosecond (fs) component is observed for these types of sensitizers, directly attached to an oxide surface [31–34] This would then

be one of the fastest chemical processes known to date There is a debate for a second, slower, injection component on the picosecond (ps) timescale The reason for this could, on the one hand, be due to an intersystem crossing of the excited dye from a singlet to a triplet state The singlet state injects on the fs timescale, whereas the slower component arises from the relaxation time of the singlet to triplet transition and from a lower driving force in energy between the triplet state and the conduction band of the TiO2 [34] Another view of the slower component is that it is very sensitive to sample condition and originates from dye aggregates

on the TiO2 surface [33] For DSC device performance, the timescales of the injection process should be compared with direct recombination from the excited state of the dye to the ground state This is given by the excited state lifetime of the dye, which for typical ruthenium complexes used in DSC is 30–60 ns [8] Thus, the injection process itself has not generally been considered to be a key factor limiting device performance Koops et al observed a much slower electron injection in a complete DSC device with a time constant of around 150 ps This would then be slow enough for kinetic competition between electron injection and excited state decay of the dye, with potential implications for the overall DSC performance [35] These results are debated since they are obtained with a single-photon counting technique that is a nondirect measurement of the injection process A slower injection in the sub-ps time regime could be even more severe for organic dyes, for which the excited state decay time could be less than nanoseconds Therefore, research on the electron injection process will continue to be an important topic in DSC research and needs to be extended to other classes of sensitizers beside the ruthenium complexes

1.21.2.2.2 Regeneration of the oxidized dyes

The interception of the oxidized dye by the electron donor, normally I−, is crucial for obtaining good charge collection yields and long cycle life of the sensitizer For a turnover number (cycle life of the sensitizer in the DSSC device) above 108 (which is required for a DSC lifetime of >15 years in outdoor conditions), the lifetime of the oxidized dye must be longer than 100 s if the regeneration time is 1 μs [36] This is achieved by the best-performing ruthenium complexes such as C101

A large number of sensitizers are efficiently regenerated by iodide, as is evident from their good solar cell performance Most of these sensitizers have oxidation potentials that are similar to or more positive than that of the standard N3 sensitizer Ru(Lbip)2

0

(NCS)2 (Vredox = +1.10 V vs normal hydrogen electrode (NHE)) Because the redox potential of the iodide–triiodide electrolyte with organic solvent is about +0.35 V versus NHE, the driving force ΔG0 for regeneration of N3 is 0.75 eV It is of interest to estimate how much driving force is needed Kuciauskas et al [37] investigated regeneration kinetics of a series of Ru and Os complexes and found that Os(Lbip)2(NCS)2 with a ΔG0 = 0.52 eV is not (or is very slowly) regenerated by iodide, while Os(Lbip)2(CN)2 (ΔG0

= 0.82 eV) is regenerated The black dye, RuLter(NCS)3, with ΔG0

= 0.60 eV, shows rapid regeneration [38] Clifford et al [39] studied regeneration in a series of Ru sensitizers and found that Ru(Lbip)2Cl2 with ΔG0 = 0.46 eV gave slow regeneration (>100 μs), leading to a low regeneration efficiency The results suggest that 0.5–0.6 eV driving force is needed for regeneration of Ru complex sensitizers in iodide–triiodide electrolyte The need for such a large driving force is probably and outcome of the initial regeneration reaction, involving the I−/I2 − redox couple, having a more positive potential than I−/I3 − [40]

Fast regeneration kinetics are also found for the one-electron redox mediators Cobalt(II)-bis[2,6-bis(1′-butylbenzimidazol-2′-yl)

pyridine] ðCoðdbbipÞ2 Þ gave regeneration times of some microseconds and regeneration efficiencies of more than 0.9 [41, 42] Ferrocene and phenothiazine gave rapid regeneration, while cobalt(II)-bis(4,4′-di-tert-butyl-2,2′-bipyridine) was slow [43] Interestingly, mixtures of this Co complex with ferrocene and phenothiazine were efficient in DSCs, suggesting that a mix of redox mediators can be a viable approach in DSCs [43] Very rapid dye regeneration was observed in the case of the solid-state DSCs (sDSCs) where the redox electrolyte is replaced by the solid hole conductor spiro-MeOTAD [44] Bach et al found that hole injection from the oxidized Ru(Lbip)2 (SCN)2 dye to the spiro-MeOTAD proceeds over a broad timescale, ranging from less than 3 ps to a few nanoseconds [45]

Very recently, several papers have been published dealing with the regeneration of oxidized dyes with the iodide–triiodide electrolyte [40, 46, 47]

1.21.2.2.3 Electron transport through the mesoporous oxide film

The mesoporous semiconductor electrode consists of numerous interconnected nanocrystals Because these particles are typically not electronically doped and surrounded by ions in the electrolyte, they will not have an internal electrical field and will not display any significant band bending Electrons photoinjected into the nanoparticles from the dye molecules are charge-compensated by

ions in the electrolyte Photocurrent will be detected in the external circuit once the electrons are transferred into the conducting substrate The gradient in electron concentration appears to be the main driving force for transport in the mesoporous oxide films – that is, electron transport occurs by diffusion [11,18, 48] Because the electrons in the mesoporous electrode are charge-compensated by ions in the electrolyte, the diffusion processes of electrons and ions will be coupled through a weak electric field This will affect transport of charge carriers The measured electron diffusion can thus be described by an ambipolar diffusion model [49, 50]

In contrast to the notion that electron transport occurs by diffusion, it is observed that the electron transport depends on the incident light intensity, becoming more rapid at higher light intensities [51, 52] This can be explained by a diffusion coefficient that is light intensity-dependent or, more correctly, dependent on the electron concentration and Fermi level in the TiO2 The measured value of the diffusion coefficient is orders of magnitude lower than that determined for single-crystalline TiO2 anatase (0.4 cm2 s−1) [53] These observations are usually explained by using a multiple-trapping model [52, 54–57] In this model, electrons are considered to be mostly trapped in localized states below the conduction band, from which they can escape by thermal activation Experiments suggest that the density and energetic location of such traps is described by an exponentially decreasing tail of states below the conduction band [54, 56] The origin of the electron traps remains obscure at present: they could correspond to trapping of electrons at defects in the bulk, grain boundaries, or surface regions of the mesoporous oxide or

to Coulombic trapping due to local field effects through interaction of electrons with the polar TiO2 crystal or with cations of the electrolyte [58–60]

1.21.2.2.4 Loss reactions

During their relatively slow transport through the mesoporous TiO2 film, electrons are always within only a few nanometers distance of the oxide–electrolyte interface Recombination of electrons with either oxidized dye molecules or acceptors in the electrolyte is therefore a possibility The recombination of electrons with the oxidized dye molecules competes with the regeneration process, which usually occurs on a timescale of submicroseconds to microseconds The kinetics of the back electron-transfer reaction from the conduction band to the oxidized sensitizer follow a multiexponential time law, occurring on a microsecond to millisecond timescale, depending on electron concentration in the semiconductor and, thus, the light intensity The reasons suggested for the relatively slow rate of this recombination reaction are as follows: (1) weak electronic coupling between the electron in the solid and the Ru(III) center of the oxidized dye, (2) trapping of the injected electron in the TiO2, and (3) the kinetic impediment due to the inverted Marcus region [61] Application of a potential to the mesoporous TiO2 electrode has a strong effect [62–65] When the electron concentration in the TiO2 particles is increased, a strong increase in the recombination kinetics is found Under actual working conditions, the electron concentration in the TiO2 particles is rather high and recombination kinetics may compete with dye regeneration

Recombination of electrons in TiO2 with acceptors in the electrolyte is, for the I−/I3 − redox system, generally considered to be an important loss reaction, in particular under working conditions of the DSC device when the electron concentration in the TiO2 is high The kinetics of this reaction are determined from voltage decay measurements and normally referred to as the electron lifetime Lifetimes observed with the I−/3 − system are very long (1–20 ms under 1 sun light intensity) compared with other redox systems used in DSCs, explaining the success of this redox couple The mechanism for this recombination reaction remains unsettled but appears to be dominated by the electron trapping–detrapping mechanism in the TiO2 [66] Recently, a lot of attention has been drawn to the effects of the adsorbed dye on the recombination of TiO2 electrons with electrolyte species There are several reasons: first, adsorption of the dye can lead to changes in the conduction band edge of TiO2 because of changes in surface charge This will lead to a larger driving force for recombination Second, dyes can either block or promote reduction of acceptor species in the electrolyte [67] The size of the oxide particle, and thus the surface-to-volume ratio, is also expected to have a significant effect on electron lifetime [68, 69]

1.21.2.2.5 Transport of the redox mediator and reactions at the counter electrode

Transport of the redox mediator between the electrodes is mainly driven by diffusion Typical redox electrolytes have a high conductivity and ionic strength so that the influence of the electric field and transport by migration is negligible In viscous electrolytes such as ionic liquids, diffusion coefficients can be too low to maintain a sufficiently large flux of redox components, which can limit the photocurrent of the DSC [70]

In the case of the iodide–triiodide electrolyte, an alternative and more efficient type of charge transport can occur when high mediator concentrations are used: the Grotthuss mechanism In this case, charge transport corresponds to the formation and cleavage

of chemical bonds In viscous electrolytes, such as ionic liquid-based electrolytes, this mechanism can contribute significantly to charge transport in the electrolyte [70–73] In amorphous hole conductors that replace the electrolyte in sDSCs, charge transport takes place through hole hopping In the most investigated molecular hole conductor for DSCs, spiro-MeOTAD, mobility is increased 10-fold by the addition of a Li salt [74] Resistance, however, in the hole-transporting layer can be a problem in sDSCs

At the counter electrode in standard DSCs, triiodide is reduced to iodide The counter electrode must be catalytically active to ensure rapid reaction and low overpotential Pt is a suitable catalyst as iodine (triiodide) dissociates to iodine atoms and iodide upon adsorption, enabling a rapid one-electron reduction The charge-transfer reaction at the counter electrode leads to a series resistance in the DSC, the charge-transfer resistance RCT Ideally, RCT should be ≤1 Ω cm2 to avoid significant losses A poor counter electrode will affect the current–voltage characteristics of the DSC by lowering the fill factor (FF)

1.21.2.3 Basic Characterization of DSC Devices

In this section, we describe the basic solar cell measurements – that is, the determination of solar-to-electrical energy conversion efficiency, η, and the quantum efficiency As mentioned above, there are a huge number of material components and combinations which can be used to prepare a DSC device To illustrate some results from basic efficiency measurements we have included a comparison between liquid and solid-state DSC [75] The results are obtained for DSC devices which contained an organic sensitizer, D35, a liquid I−/I3 − electrolyte or a solid-state, spiro-MeOTAD, hole conductor The molecular structures of D35 and spiro-MeOTAD are shown in Figure 6

For the liquid cell, a platinized fluorine-doped tin oxide TCO substrate is used as counter electrode, and for the sDSC, an evaporated silver layer on top of the hole conductor is used For the working electrode, 1.8 μm mesoporous TiO2 films screen printed on dense TiO2 blocking layers were used The mesoporous TiO2 films were treated with a TiCl4 solution [76] The electrolyte concentrations were 0.05 M I2, 0.5 M LiI, and 0.5 M 4-tert-butyl pyridine (4TBP) in 3-methoxypropionitrile (MPN), while the spiro-MeOTAD solution used for spin coating consisted of 150 mg spiro-MeOTAD per milliliter of chlorobenzene with 15 mM LiTFSI and 60 mM 4TBP added

1.21.2.3.1 Efficiency measurements

Current–voltage measurements (I–V measurements) under illumination are used to determine the efficiencies of solar cells A lamp which simulates the AM1.5 solar spectrum is used for illumination and is calibrated to an intensity of 1000 W m−2 for measure­ ments at 1 sun intensity A Newport solar simulator of class B was used for the results presented below A voltage is then applied between the working and counter electrode of the solar cell and the current output is measured The voltage range should include the voltage at which the current is 0 (the open-circuit voltage, VOC) and 0 V at which the short-circuit current density (JSC) is measured The resulting current–voltage curve is usually referred to as an I–V curve The conditions for measuring the current– voltage characteristics of a DSC device should be carefully checked The I–V characteristics of DSC can be quite sensitive to scan rate, preconditioning of the scan (which potential is applied and for how long), as well as changes occurring after repeated scans – see, for example, discussions in Reference 75

Measurements can also be carried out in the dark, and the measured data are accordingly called a ‘dark current curve’ Figure 7

shows an example of I–V curves under illumination and in the dark for the solid and the liquid-electrolyte cell with the D35 dye The efficiency of a solar cell, η, is given by

Pmax ðJ
V Þmax

where Pin is the illumination intensity and Pmax is the maximum power output of the solar cell at this light intensity To describe the efficiency of a solar cell in terms of VOC and JSC, a quantity called the fill factor (FF) is introduced, relating Pmax

to VOC and JSC:

ðJ
V Þmax

Figure 6 Molecular structures of the organic dye D35 and hole conductor spiro-MeOTAD

J / mA

8

6

4

2

−2

0 Liquid

Solid

4

2

0

−2

Figure 7 I–V curves of an sDSC (gray) and a liquid-electrolyte DSC (black) with D35 as sensitizer under 1 sun illumination (solid line) and in the dark (dashed line) Courtesy of Dr Ute Cappel

The efficiency can then be written as

JSC
VOC
FF

in

For the example in Figure 7, the efficiencies of the liquid DSC and sDSC were 2.9% and 3.6%, with VOC of 0.77 V and 0.93 V, JSC

of 7.0 and 7.0 mA cm−2, and FF of 0.54 and 0.55, respectively Thus, the sDSC has a higher VOC than the liquid-electrolyte DSC, which is the reason for the higher efficiency of the sDSC It should be noted that the film thickness is only 1.8 μm, so the optical density of the film is relatively low, reducing the overall efficiency What was also observed in Reference 75 is that in consecutive scans, the short-circuit current of the solid-state cell decreases, while the FF and the overall efficiency increase, demonstrating how care must be taken in measuring I–V curves for DSCs, in particular for sDSCs

With regard to illumination of the DSC cell, the cells should be masked A mask size that is 1 mm on each side bigger than the active area is recommended in Reference 77 Using thin TCO glass (∼1 mm) and a device size of at least 5
5 mm is also recommended This will reduce optical artefacts that can enhance or diminish the power conversion efficiency To be qualified in the official table of world record efficiencies for PVs, the solar cell area must be at least 1 cm2 For efficiency measurements of solar cells in general, we refer to Reference 78 For DSC specifically, we summarize the discussions above according to Reference 79

1 The rate of change of the bias applied must be slow, and the IV curve should be scanned in both directions to determine if it is slow enough

2 The efficiency depends on the premeasurement state The temperature should be 25 ˚C, and there should be a bias light of one-sun at Pmax Procedures that approximate this can be used It should be noted that VOC or JSC might not give the same results as preconditioning at Pmax

3 Light may penetrate from outside the defined area The magnitude of this should be determined by looking at JSC with or without

an aperture

4 The monochromatic quantum efficiency dependence on light intensity (see 1.21.2.3.2) may be nonlinear, and control measure­ ments should be made at different light intensities to check this It is recommended that quantum efficiency measurements should be done with bias light and preferably chopped monochromatic beam

1.21.2.3.2 External and internal quantum efficiencies

The incident photon-to-current conversion efficiency (IPCE), sometimes also called the external quantum efficiency of the solar cell, describes how many of the incoming photons at one wavelength are converted to electrons:

Electrons out ðλÞ JSCðλÞ hc JSCðλÞ JSCðλÞ½Åcm− 2

Incident photons ðλÞ qΦðλÞ q λPinðλÞ λ½nm
PinðλÞ½W cm− 2 where JSC is the short-circuit current density, Ф is the photon flux, Pin is the light intensity at a certain wavelength λ, q the elementary charge, and h and c are the Planck’s constant and speed of light, respectively

IPCEs are made by typically using a xenon or halogen lamp coupled to a monochromator The photon flux of light incident on the samples is measured with a calibrated photodiode, and measurements are typically made at 10 or 20 nm wavelength intervals between 400 nm and the absorption threshold of the dye Since DSCs are devices with relatively slow relaxation times, it is important to make sure that the measurement duration for a given wavelength is sufficient for the current to be stabilized (normally

5–10 s) If it is observed that IPCE depends on light intensity, then the measurements should be made with additional bias light to

100

80

60

40

Solid-state DSC

20

Liquid-electrolyte DSC

0

λ / nm

350 400 450 500 550 600 650

Figure 8 IPCE spectra of the sDSC and the liquid-electrolyte DSC with D35 as sensitizer Courtesy of Dr Ute Cappel

ascertain that IPCE is determined at relevant light intensity conditions The reasons for light intensity-dependent IPCE may be that the charge collection efficiency (process 4 in Figure 4) increases with light intensity due to faster electron transport, or that there are mass transport limitations in the electrolyte, decreasing IPCE with light intensity

The magnitude of the IPCE spectrum depends on how much light is absorbed by the solar cell and how much of the absorbed light is converted to electrons, which are collected:

Where LHE is equal to 1 – 10−A with A being the absorbance of the film, φinj and φreg the quantum yields for electron injection and dye regeneration, respectively, and ηCC the charge collection efficiency

IPCE spectra of the liquid DSC and sDSCs sensitized with D35 are shown in Figure 8 [75]

The spectra are slightly different in shape, although the same TiO2 thickness and the same dye were used The spectrum of the sDSC

is lower at around 380 nm and higher at the red edge of the spectrum than the spectrum of the liquid-electrolyte DSC These differences can be explained with Equation [5]: at around 380 nm, spiro-MeOTAD absorbs strongly, decreasing the transmittance at this wavelength in the solid-state device compared with the liquid-electrolyte DSC and therefore decreasing the IPCE LHE at the absorption maximum of D35 was close to 1 for the devices, resulting in IPCE maxima of 80% However, at longer wavelengths, light harvesting was incomplete In the sDSC, the reflecting back contact increased the light harvesting at these wavelengths and therefore also the IPCE

The short-circuit current of a solar cell can be calculated by integrating over the product of the IPCE and the AM1.5 solar spectrum:

where φph,AM1.5 is the photon flux in AM1.5 at wavelength λ For the DSC presented in Figure 8, the integrated JSC were determined to be 7.75 mA cm−2 for the sDSC and 7.4 mA cm−2 for the liquid-electrolyte cell These currents are slightly higher than the currents determined in the I–V measurements For the sDSC, this might be due to the fact that the IPCE measurement was carried out prior to the I–V measurements, so the analysis of the data must be checked according to the discussions above

From a fundamental viewpoint, the so-called absorbed photon-to-current conversion efficiency (APCE) values provide further insight into the properties of the device APCE (or internal quantum efficiency) shows how efficiently the numbers of absorbed photons are converted into current APCE is obtained by dividing the IPCE number by the LHE (0–100%) The IUPAC name for LHE is ‘absorptance’ Thus,

IPCE

Quantitative in situ measurement of the LHE of complete devices is complicated because of light scattering by the mesoporous oxide film and light absorption by the other cell components For fundamental studies, it is therefore advisable to use transparent mesoporous TiO2 films There are several descriptions of the procedures to obtain LHE in the literature, including on how to take into account scattering effects – see, for example, References 80–83

1.21.2.3.3 Overview of the concept of a DSC toolbox

The dye-sensitized solar cell is a complex, highly cooperative system To understand the precise working mechanism of the DSC and to optimize its performance, it is important to map the energetics of the different components and interfaces and the kinetics

of the different electron-transfer reactions for complete DSC devices working under actual operating conditions The so-called toolbox of characterization techniques is used to investigate the kinetics of different reactions in situ in DSC devices These studies are particularly fruitful, as the interactions between different components can be studied Toolbox methods are

continuously being developed by several research groups, and for two recent reviews we refer to References 18 and 84 Examples

of tool-box techniques are:

• Photovoltage and photocurrent as a function of light intensity

• Small-modulation photocurrent and photovoltage transients to investigate electron transport and recombination

• Steady-state, quantum efficiency measurements to determine injection efficiency, collection efficiency, and electron diffusion length

• Electron concentration measurements

• Determination of the internal potential (quasi-Fermi level) in the mesoporous electrode

• Photo-induced absorption spectroscopy to obtain information on recombination reactions and regeneration of the oxidized dye

by the electrolyte

A set of very powerful toolbox techniques is based on electrochemical impedance spectroscopy (EIS) The reader is referred to the works of Bisquert and co-workers on this topic, and as examples of references we propose References 84–86 In EIS, the potential applied to a system is perturbed by a small sine-wave modulation, and the resulting sinusoidal current response (amplitude and phase shift) is measured as a function of modulation frequency The impedance is defined as the frequency domain ratio of the voltage to the current and is a complex number For a resistor (R), the impedance is a real value, independent

of modulation frequency, while capacitors (C) and inductors (L) yield an imaginary impedance with values that vary with frequency The impedance spectrum of an actual system – that is, the impedance measured in a wide range of frequencies – can be described in terms of an equivalent circuit consisting of series- and parallel-connected elements R, C, L, and W, which is the Warburg element that describes diffusion processes Using EIS, the following parameters can be obtained: series resistance, charge-transfer resistance of the counter electrode, diffusion resistance of the electrolyte, the resistance of electron transport and recombination in the semiconductor, and the chemical capacitance of the mesoporous electrode One of the advantages of impedance spectroscopy is that it allows simultaneous characterization of electron transport in the mesoporous oxide and of recombination of the electrons from the oxide to the hole-conducting medium The transport and interfacial transfer of electrons

in the mesoporous oxide layer can be modeled using a distributed network of resistive and capacitive elements in the form of a finite transmission line

1.21.2.4 Development of Material Components and Devices

Since the initial work in the beginning of the 1990s, a wealth of DSC components and configurations have been developed Perhaps

a key concept for the future success of DSC is ‘diversity’ At present, several thousands of dyes have been investigated, as well as numerous types of mesoporous films with different morphologies and compositions The last year has seen some very interesting breakthroughs in the use of alternative redox systems, and this field is now opened up after almost 20 years of I−/I3 − dominance With such a diversity to explore, the DSC technology can be expected to progress rapidly, be it through design of new materials and combinations based on fundamental insights or by ‘evolution’ – that is, trial and error – or better, with the use of combinatorial approaches

In the following sections, we briefly overview the development of the material components and devices in DSC and in general refer to the many recent reviews on these topics

1.21.2.4.1 Mesoporous oxide working electrodes

The key to the breakthrough for DSCs in 1991 was the use of a mesoporous TiO2 electrode with a high internal surface area to support the monolayer of a sensitizer Typically, the increase of surface area by using mesoporous electrodes is about a factor of

1000 in DSCs TiO2 still gives the highest efficiencies, but many other metal oxide systems have been tested, such as ZnO, SnO2, and

Nb2O5 Besides these simple oxides, ternary oxides, such as SrTiO3 and Zn2SnO4, have been investigated, as well as core–shell structures, such as ZnO-coated SnO2 For recent reviews on the development of nanostructured metal oxide electrodes for DSC, the reader is referred to References 1, 18 and 87–90 During the last few years, large efforts have been made to optimize the morphology

of the nanostructured electrode, and a large range of nanostructures has been tested from random assemblies of nanoparticles to organized arrays of nanotubes and single-crystalline nanorods These studies are motivated by the expectation of an improved and directed charge transport along the rods and tubes and by an improved pore filling of hole conductor materials for sDSC The general reviews for preparation techniques and structures are given, for example, by Chen et al for TiO2 [91] and Ozgur et al for ZnO [92]

TiO2 is a stable, nontoxic oxide that has a high refractive index (n = 2.4–2.5) and is widely used as a white pigment in paint, toothpaste, sunscreen, self-cleaning materials, and food (E171) Several crystal forms of TiO2 occur naturally: rutile, anatase, and brookite Rutile is thermodynamically the most stable form Anatase is, however, the preferred structure in DSCs, because it has a larger band gap (3.2 vs 3.0 eV for rutile) and a higher conduction band edge energy, Ec This leads to a higher Fermi level and VOC in DSCs for the same conduction band electron concentration

For state-of-the-art DSCs, the employed architecture of the mesoporous TiO2 electrode is as follows [Reference 18 and references therein]: