Increased light harvesting in dye-sensitized solar cells with energy relay dyes

Increased light harvesting in dye-sensitized solar cells with energy relay dyes

(Parte 1 de 3)

Increased light harvesting in dye-sensitized solar cells with energy relay dyes

Brian E. Hardin1,2,E ric T.H oke1,P aulB .A rmstrong3,J un-Ho Yum2,P ascalC omte2,TomasTorres4, Jean M. J. Frechet3, Md Khaja Nazeeruddin2,M ichael Gratzel2 and Michael D. McGehee1*

Conventional dye-sensitized solar cels have excelent charge colection efficiencies, high open-circuit voltages and good fil factors. However, dye-sensitized solar cells do not completely absorb all of the photons from the visible and near-infrared domain and consequently have lower short-circuit photocurrent densities than inorganic photovoltaic devices. Here, we present a new design where high-energy photons are absorbed by highly photoluminescent chromophores unattached to the titania and undergo Forster resonant energy transfer to the sensitizing dye. This novel architecture allows for broader spectral absorption, an increase in dye loading, and relaxes the design requirements for the sensitizing dye. We demonstrate a 26% increase in power conversion efficiency when using an energy relay dye (PTCDI) with an organic sensitizing dye (TT1). We estimate the average excitation transfer efficiency in this system to be at least 47%. This system offers a viable pathway to develop more efficient dye-sensitized solar cells.

ye-sensitized solar cells (DSCs) work on the basis of light harvesting using a sensitizing dye (SD) attached to a widebandgap semiconductor1–5. DSCs are composed mainly of abundant, non-toxic materials and offer an inexpensive route to the development of highly efficient photovoltaic cells. State-of-theart DSCs, which absorb light from 350 to 700 nm, have validated power conversion efficiencies of over 1% (ref. 6). A key to improving the efficiency of DSCs is to increase their spectral absorption range. To reach power conversion efficiencies of 15% using an

I2/I32 redox couple, DSCs must absorb 80% of the solar spectrum from 350 to 900 nm (ref. 7). Light absorption in DSCs is determined by the molar extinction coefficient of the SD, the surface coverage of the dye (dye molecules nm22), and the total surface area of the oxide film8. Films comprising TiO2 nanoparticles enhance the surface area; 10-m-thick films have surface areas

1,0 greater than that of a flat junction. SDs generally pack tightly on the TiO2 surface with a density of 0.5–1 dye molecules nm22 (ref. 8). The SD has traditionally been made from ruthenium-based complexes (for example, N719 and Z907)6,9 that have fairly broad absorption spectra (Dl 350 nm) but low molar extinction coefficients (5,0–20,0 M21 cm21). Organic dyes have recently been developed with substantially higher molar extinction coefficients (50,0–200,0 M21 cm21) but narrow spectral bandwidths (Dl 250 nm)10–13. As a general rule, dyes that absorb strongly do not typically exhibit broad absorption.

Co-sensitization of titania by dyes with complimentary absorption spectra has been demonstrated to enhance light absorption and broaden the spectral response of organic DSCs (ref. 14). However, the limited number of sites on the titania surface to which dye molecules attach places a constraint on the light absorption achievable by co-sensitization. Furthermore, co-sensitization requires that each dye adsorb strongly on the surface, transfer charge efficiently into the TiO2 (refs 15–18), have slow recombination (that is, in the millisecond time domain)17,19–21, and regenerate with the redox couple22. Few dyes exist that are both excellent absorbers and possess the requisite energy levels and chemical anchoring groups to be good SDs. A recent study has demonstrated the use of Forster resonant energy transfer (FRET) between covalently linked energy donor molecules to the SD attached on the titania surface23. Siegers and colleagues23 were able to demonstrate a high excitation transfer efficiency (.89%) between attached dye molecules and an improvement in the device external quantum efficiency of 5–10% between 400 and 500 nm. However, the overall power conversion efficiency enhancement of the DSC was low (,9%) and linked more to an increase in the opencircuit voltage rather than an increase in the short-circuit photocurrent density.

In this Article we demonstrate that unattached, highly luminescent chromophores (PTCDI) inside the liquid electrolyte can absorb high-energy photons and efficiently transfer the energy to the anchored near-infrared sensitizing zinc phthalocyanine dye (TT1), increasing the absorption bandwidth of the DSC. Figure 1 shows two routes for charge generation incorporated in this system. In typical DSCs, light is absorbed by the SD (1), which transfers an electron into the titania and a hole into the electrolyte. In the new design, the unattached energy relay dye (ERD) is excited by higher energy (blue) photons and then undergoes Forster energy transfer (2) to the SD. This design is analogous to photosynthesis in purple bacteria, where an aggregate of light-harvesting pigments transfer their energy to the reaction centre, initiating charge separation24. In particular, the pigment LH-I is not in direct contact with the reaction centre, and transfers its excitation by means of an intermediate pigment (LH-I) in under 100 ps with 95% efficiency25,26. We recently proposed using unattached ERDs and long-range energy transfer to increase light absorption27. Placing the ERDs inside the electrolyte has several important advantages. First, because the attached dye only has to absorb light over a smaller spectral region, it can be chosen to have a stronger and narrowerDepartment of Material Science and Engineering, Stanford University, Stanford, California, 94305-4045, USA, Laboratoire de Photonique et Interfaces, Ecole Polytechnique Federale de Lausanne, CH-1015, Lausanne, Switzerland, Department of Chemistry, University of California, Berkeley, California 94720- 1460, USA, Departamento de Quımica Organica (C-I) and Departamento de Fısica de Materiales (C-IV), Facultad de Ciencias, Universidad Autonoma de

Madrid, Cantoblanco, 28049 Madrid, Spain. *e-mail:


NATURE PHOTONICS | ADVANCE ONLINE PUBLICATION | 1 © 2009 Macmillan Publishers Limited. All rights reserved.

absorption spectrum. Second, the SD can be redshifted compared to the commonly used dyes because the ERD can absorb higher energy photons. Furthermore, it is possible to place multiple ERDs with complementary absorption spectra to tailor light absorption inside the device. Finally, the ERD does not need to be attached to the titania surface and with no additional processing steps can be mixed in very large concentrations inside the electrolyte. In summary, the addition of ERDs into the electrolyte makes the overall absorption spectrum wider and stronger for the same film thickness. It is important to note that the ERDs do not participate in the charge transfer or collection process and thus do not require precise energy levels or specialized attachment groups28. ERDs should be designed to be soluble in and not greatly quenched by the electrolyte. The ERD concept is particularly applicable to solid-state DSCs (refs 3,29,30), which are currently restricted to a thickness of 2 m and are not able to absorb all of the light, even at the peak of the dyes’ absorption spectrum. Incorporating long-range energy transfer into the solid-state DSC will require ERDs that avoid charge transfer into the hole transporter. The ERD system is also extremely useful for nanostructured systems (for example, TiO2 nanotubes31, ZnO nanorods32) that have less available surface area and thus poorer light absorption.

FRET involves dipole–dipole coupling of two chromophores, known as the donor and acceptor, through an electric field33.A n excitation of the donor, or in our case the ERD, can be transferred non-radiatively through the field to the acceptor, or SD, if there is overlap between the emission spectrum of the donor and the absorption spectrum of the acceptor. Efficient energy transfer over 3–8 nm can be achieved with strong spectral overlap and high donor emission efficiencies, for an isotropic alignment between individual chromophores in solution. If, however, the single acceptor chromophore is replaced by a dense two-dimensional array (that is, SDs tightly packed on the titania surface) FRET can become efficient well over 25 nm from the interface34,35.

High FRET transfer rates (kFRET) are essential to quickly transfer the energy before the excited ERD non-radiatively decays. The

FRET rate is dependent upon the Forster radius (Ro) between the ERD and the SD, the separation distance between the ERD and the SD/TiO2 interface, which is a function of pore size and geome- try, and the natural fluorescence decay rate of the ERD, k0 ¼ 1/t0. The Forster radius, or the distance in which Forster energy transfer

is 50% probable between individual chromophores, can be calculated33 using equation (1):

ð FDðlÞ1AðlÞl4 dl ð1Þ where n is the index of refraction of the host medium (1.4–1.5 for the DSC electrolyte), k2 the orientational factor (2/3 for random orientation), NA Avogadro’s number, QD the photoluminescence

(PL) efficiency, FD the emission profile of the donor and e(l)i s the molar extinction coefficient.

A previously reported36 derivative of perylene-3,4,9,10-tetracarboxylic diimide (PTCDI; Fig. 2b) was synthesized (see Methods) for use as an ERD. PTCDI is an ideal ERD candidate because of its extremely high PL efficiency (.90%), fast fluorescence lifetime (4.8 ns), excellent photo and air stability and relatively strong absorption coefficient (50,0 M21 cm21 at 580 nm)37 .I tsb ulky alkyl phenyl substituents were designed to reduce chromophore interactions between adjacent dye molecules in order to prevent aggregate formation and reduction of fluorescence. A zinc phthalocyanine dye (TT1; Fig. 2c) was chosen as the SD for its high molar extinction coefficient of 191,500 M21 cm21 centred at 680 nm (ref. 14). One would prefer a dye with a smaller energy gap, but such dyes are not readily available yet with the necessary anchoring groups. When attached to titania, the TT1 dye absorption broadens (as shown in Fig. 2a) and significantly overlaps the PL emission of the PTCDI. Given the absorption and emission profile of the TT1 and PTCDI, respectively, the Forster radius is estimated to be 8.0 nm. Time-resolved PL measurements on solutions with varying concentration of TT1 deter- mined Ro to be 7.5–7.6 nm (see Supplementary Fig. S1). Once excited, the ERD can transfer its energy to the SD by means of FRET, emit a photon or non-radiatively decay. Non-radiative decay in the DSC system is greatly increased due to the presence of triiodide in the electrolyte. Triiodide is a highly mobile ion that is known as a ‘perfect quencher’, meaning that collisions with the ERD have a near unity probability of quenching the excited state38. Given the high concentrations of triiodide in the DSC electrolyte, the quenching rate of chromophores can be 20–2,0 times greater than the natural decay rate. Collisional quenching of the PTCDI by triiodide is described by the Stern–Volmer equation (2) (refs 39,40),

Titania nanoparticle =

Energy relay dye =Sensitizing dye = FTO (front contact)

Electrolyte-containing energy relay dye

Platinum-coated FTO (back contact) (2) FRET

(1) SD absorption

Sensitizing dye covered TiO2

Figure 1 | Schematic representation of a dye-sensitized solar cell (DSC) with energy relay dyes (ERDs). The right side of the figure shows the typical absorption process for lower energy (red) photons in the DSC: light is absorbed by the sensitizing dye (1), transferring an electron into the titania, and a hole is transported to the back contact through the electrolyte. The ERD process is similar, except that higher energy (blue) photons are first absorbed by the ERD, which undergoes Forster energy transfer (2) to the sensitizing dye (SD).


NATURE PHOTONICS | ADVANCE ONLINE PUBLICATION | © 2009 Macmillan Publishers Limited. All rights reserved.

where PL0 is the photoluminescence in the absence of a quencher, PL the photoluminescence for a given quencher concentration

[Q], t0 the natural fluorescence lifetime, t the fluorescence lifetime

stant and the electrolyte concentrations are relatively fixed, a short t0 is important for minimizing the fluorescence quenching. We determined the fluorescence lifetime of the PTCDI to be 4.8 ns

(see Supplementary Fig. S2). Figure 3 shows that the fluorescence intensity and lifetime are both reduced with increasing concentrations of the 1-methyl-3-propyl imidazolium iodide (PMII) and

used in the DSC (0.6 M PMII, 0.05 M I2) the non-radiative decay rate due to quenching (kQUENCH) is calculated to be 30 times greater than the natural fluorescence decay rate (kQUENCH ¼ 30k0). The excitation transfer efficiency (ETE) is the probability that an excited ERD will transfer its energy to a SD. The ETE for a single relay dye molecule at position ~x is dependent on the rate of Forster energy transfer, kFRET, relative to the combined rate of all decay mechanisms including the natural decay rate and quenching rate:

ETEð~xÞ¼ kFRETð~xÞ k0 þ kQUENCH þ kFRETð~xÞ ð3Þ

The FRET rate is a function of the separation distance between the ERD molecule and nearby acceptor molecules. The rate of Forster energy transfer between isolated chromophores, known as point- to-point transfer, is given by kFRET ¼ k0(Ro)6/r6, where r is the separation distance. When multiple acceptor molecules are present, the

FRET rate is equal to the sum of the transfer rates to each of the acceptors. ERDs within the Forster radius of the SD array will transfer their excitation with high efficiency, whereas ERDs in the middle of a large pore may be quenched before energy transfer occurs. We have developed a model that approximates the nanopores as either cylinders or spheres to calculate FRET rate profiles, kFRETð~xÞ, and excitation transfer efficiency profiles, ETEð~xÞ, using equation (3), and assuming uniform SD coverage over the pore walls. The morphology of the pores has important implications on ERD/SD array separation distance. Assuming a homogeneous ERD concentration, the average separation distance between the ERD and the closest

SD/TiO2 interface in a spherical pore is a quarter of the pore radius, whereas in a cylinder the average separation distance is one- third of the pore radius. Figure 4 shows how the average excitation transfer efficiency, ETE, depends on the pore diameter for cylindrical and spherical pores using the parameters calculated for the PTCDI– TT1 DSC system. Although the excited ERD has a non-radiative decay half-life of only 0.15 ns (4.8 ns/31) when placed in the electrolyte, it has an expected ETE between 5 and 70% in a 30 nm pore. The titania film comprised 20-nm particles to ensure close proximity of the ERD to the SD. The 20-nm TiO2 particles produce pore diameters between 2 and 38 nm, a film porosity of

68% (without the addition of the dye), and a roughness factor of 97 m21 (see Supplementary Figs S3,S4). A 10-m-thick layer of 20-nm particles and a 5-m-thick layer of 400-nm scattering

0 /PL

Quencher concentration (M)

PMII(PL0 /PL) PMII(τ0 /τ)

Figure 3 | Quenching of PTCDI by electrolyte species. The PTCDI photoluminescence is reduced with increasing concentration of PMII

(half-filled blue circles) and I2 (green squares). The reduction in photoluminescence (PL0/PL) by PMII is equivalent to the reduction in excitation lifetime (t0/t) shown as the red triangles. The PTCDI concentration was 1 1024 M in gamma-butyrolactone.

O O t-Bu i-Pr i-Pr i-Pr t-Bu t-Bu t-Bu i-Pr

PTCDI (Energy relay dye)

NZn t-Bu t-Bu t-Bu

TT1 (Sensitizing dye) cAbsorption, emission (rel) Wavelength (nm)

PTCDI absorption PTCDI emission TT1 absorption

Figure 2 | PTCDI and TT1 properties. a, PTCDI absorption (blue), PTCDI emission (red) in chloroform and TT1 absorption (black) on titania nanoparticles. b,c, Chemical structures of the energy relay dye PTCDI (b), and sensitizing dye TT1 (c).


(Parte 1 de 3)