15464
J. Phys. Chem. C 2007, 111, 15464-15478
Examination of Tethered Porphyrin, Chlorin, and Bacteriochlorin Molecules in Mesoporous Metal-Oxide Solar Cells Jonathan R. Stromberg,† Andras Marton,† Hooi Ling Kee,‡ Christine Kirmaier,‡ James R. Diers,§ Chinnasamy Muthiah,| Masahiko Taniguchi,| Jonathan S. Lindsey,*,| David F. Bocian,*,§ Gerald J. Meyer,*,† and Dewey Holten*,‡ Departments of Chemistry and Materials Science and Engineering, Johns Hopkins UniVersity, 3400 N. Charles St., Baltimore, Maryland 21218, Department of Chemistry, Washington UniVersity, St. Louis, Missouri 63130-4889, Department of Chemistry, UniVersity of California RiVerside, RiVerside, California 92521-0403, Department of Chemistry, North Carolina State UniVersity, Raleigh, North Carolina 27695-8204 ReceiVed: June 26, 2007; In Final Form: July 29, 2007
The performance of five tetrapyrrole molecules as sensitizers in regenerative solar cells was evaluated. The tetrapyrroles form two sets. One set contains three meso-substituted porphyrins that differ only in the nature of their surface-binding tether: isophthalic acid, ethynylisophthalic acid, or cyanoacrylic acid. The other set includes the ethynylisophthalic acid tether attached to porphyrin, chlorin, and bacteriochlorin macrocycles, which contain zero, one, and two saturated pyrrole rings, respectively. Incident photon-to-current efficiency was measured for each sensitizer loaded onto a mesoporous TiO2 semitransparent electrode in a solar cell. The porphyrin bearing the cyanoacrylic acid tether gives the largest peak and integrated (350-900 nm) photocurrent density of the five tetrapyrrole molecules. For this sensitizer, a quasi-monochromatic power conversion efficiency of 21% was obtained at the Soret maximum (450 nm), along with a fill factor of 0.69. To elucidate the molecular origins of the effects of tether and macrocycle reduction on photocurrent production, the measured redox potentials and optical absorption spectra were analyzed in terms of the characteristics (energies and electron-density distributions) of the frontier molecular orbitals obtained from density functional theory calculations. Additionally, first-principle simulations were performed for the production of photocurrent by hypothetical planar and actual mesoporous films of each sensitizer under AM 1.5 solar irradiation. Collectively, the findings give fundamental insights into the factors that govern the observed differences in photocurrent production characteristics for the five tetrapyrrole sensitizers. In addition, the results provide a framework for further tuning of the properties of these molecules and related sensitizers to enhance solar-cell performance.
I. Introduction The conversion of light energy into electrical energy using dye-sensitized metal-oxide solar cells has been an active area of study.1-5 A common motif is the Gra¨tzel-type regenerative cell. In this device, electron injection by the photoexcited sensitizer dye into the semiconductor is followed by hole transfer from the oxidized dye to a (typically) mobile charge carrier (iodide/triodide/iodine) in an electrolyte solution, which mediates the passage of charge to the counter electrode and, ultimately, through the external circuit. Many factors influence cell efficiency. One parameter is the amount of the solar irradiation that is absorbed. This factor depends on the distribution of wavelengths in the absorption spectrum of the dye (i.e., its overlap with the solar spectrum) and the fraction of the incident light absorbed at each wavelength (which depends on the molecular absorption cross-section, the loading of the dye on the surface, etc.). Other factors affecting cell performance * To whom correspondence should be addressed. Email: G.J.M.:
[email protected]; D.H.:
[email protected]; D.F.B.:
[email protected]; J.S.L.:
[email protected]. † Johns Hopkins University. ‡ Washington University. § University of California. | North Carolina State University.
include the excited-state dye-to-semiconductor electron-injection efficiency, and the efficiency of hole/electron transfer between the oxidized dye and the mobile charge carrier. Some of these factors are directly accessible experimentally, others can be estimated using experimental data, and others are often inferred by comparison. One approach to assess the balance of factors that influence device performance is to compare cell efficiency upon variation of one or two molecular/electronic characteristics among a set of related dye sensitizers. In this study, we have investigated the effect of varying the structural/electronic properties of the surface tether, using a common porphyrin sensitizer. We also probed the effects of increasing red-region and decreasing blue-region absorption of a related set of tetrapyrrole chromophores (porphyrin, chlorin, and bacteriochlorin), employing the same linker to the semiconductor. Gra¨tzel, Officer, and co-workers synthesized several zinc porphyrins and investigated their use in dye-sensitized TiO2 solar cells.6 One of the more effective dyes employed a cyanoacrylic acid linker, which we have also examined in the present study. A characteristic of porphyrins is that they absorb strongly in the blue region but only weakly in the green-red regions. Effective harvesting of solar radiation, as is achieved in plant and bacterial photosynthetic systems, requires the use of
10.1021/jp0749928 CCC: $37.00 © 2007 American Chemical Society Published on Web 10/03/2007
Tethered Porphyrin, Chlorin, and Bacteriochlorin
J. Phys. Chem. C, Vol. 111, No. 42, 2007 15465 these properties are critical for photocurrent generation. The analysis is aided by density functional theory (DFT) calculations that elucidate the characteristics of the frontier molecular orbitals (energies and electron-density distributions). The combined investigations give insights into how the molecular characteristics of tetrapyrrole chromophores may be tuned to enhance their performance as sensitizers in molecular-based solar cells. II. Experimental Methods
Figure 1. Absorption spectra for a porphyrin (blue), a chlorin (magenta), and a bacteriochlorin (green). All three compounds are magnesium chelates: magnesium octaethylporphyrin, chlorophyll a, and bacteriochlorophyll a, respectively. The peak violet-blue (Soret) region molar absorption coefficients7 are 408 mM-1 cm-1 (408 nm), 112 mM-1 cm-1 (428 nm), and 71 mM-1 cm-1 (361 nm), respectively.
pigments that absorb across the visible spectrum, including in the red and near-infrared regions. Chlorins and bacteriochlorins are particularly attractive for absorption in the red and nearinfrared regions, respectively. A comparison of the absorption spectra7 of these three classes of tetrapyrrole molecules (porphyrin, chlorin, and bacteriochlorin) is provided in Figure 1. The predominant use of porphyrins versus chlorins and bacteriochlorins in solar cells (and other applications) stems from the greater synthetic accessibility of the former versus the latter two classes of molecules. The chief limitation to the use of synthetic chlorins and bacteriochlorins has been the lack of methodology for preparing molecules that are both stable to oxidation and can be easily functionalized with the appropriate tethers for surface attachment. Indeed, the chlorins employed to date in dye-sensitized solar cells largely are derived from naturally occurring chlorophylls.8 Recently, one of our groups has developed new synthetic methods that address both of these issues. Stable chlorins9 and bacteriochlorins10 can now be prepared via versatile synthetic methods that enable tailoring of substituents about the perimeter of the hydroporphyrin macrocycle. With these new synthetic methods in hand, we recently prepared a porphyrin (ZnP-EI), a chlorin (ZnC-EI) and a bacteriochlorin (FbB-EI) bearing the same ethynylisophthalic acid tether for incorporation in molecular-based solar cells (Chart 1).11 The use of a single tether type allows a direct comparison of the photocurrent generation in cells that incorporate the three different tetrapyrrole sensitizers. Furthermore, to examine the effect of the tether on photoactivity, we prepared two additional meso-functionalized porphyrins, one bearing a cyanoacrylic acid tether (ZnP-A) and one with an isophthalic acid tether (ZnP-I). We previously characterized the photophysical properties of these five compounds, with a focus on the fluorescence behavior (spectra, quantum yields, and lifetimes) and the excited-state decay pathways.11 In the present study, we examined the photocurrent production in regenerative solar cells, each utilizing a mesoporous TiO2 film loaded with one of the five above-mentioned tetrapyrrole chromophores as the sensitizer. Complementary simulations of the solar-cell efficiency for hypothetical planar and actual mesoporous films of each chromophore were performed to aid in understanding the properties that underpin the observed differences in cell performance for the films. Additionally, we have carried out an in-depth study and analysis of the optical absorption and redox characteristics of the sensitizers, because
A. Synthesis. The syntheses of the five tetrapyrroles bearing carboxylic acid anchors for surface attachment (ZnP-A, ZnP-I, ZnP-EI, ZnC-EI, and FbB-EI) are described elsewhere.11 For each of these compounds, a derivative was also prepared in which the acid moiety is replaced by an ester (ZnP-A′, ZnP-I′, ZnP-EI′, ZnC-EI′, and FbB-EI′). The ester derivatives were used as benchmark compounds for the calculations and solution photophysical and electrochemical studies. B. DFT Calculations. DFT calculations were performed with Spartan ‘06 for Windows (Wavefunction, Inc.)12 on a PC (Dell Optiplex GX270) equipped with a 3.2 GHz CPU and 3 GB of RAM. The hybrid B3LYP functional and a 6-31G* basis set were employed. For the calculations on each complex, the complete structure of the methyl ester of the tetrapyrrole was used in the geometry optimization. The equilibrium geometries were fully optimized using the default parameters of the Spartan ‘06 program. C. Electrochemistry. The redox potentials were measured with square-wave voltammetry; the sample was contained in a standard three-electrode cell using Pt working and counter electrodes and a Ag/Ag+ reference electrode. The solvent/ electrolyte was butyronitrile containing 0.1 M n-BuN4PF6. In the case of ZnP-I′ and ZnP-EI′, the oxidative waves were very poorly defined in butyronitrile/0.1 M n-BuN4PF6; consequently, the oxidation characteristics were also examined in CH2Cl2/ 0.1 M n-BuN4PF6. The oxidative waves of ZnP-I′ and ZnP-EI′ are much better defined in the latter solvent/electrolyte. All potentials reported herein are versus FeCp2/FeCp2+ ) 0.19 V. D. Optical Spectroscopy. Static absorption spectra (Cary 100) were collected for 1-10 µM solutions of the tetrapyrrole in toluene or acetonitrile at room temperature. E. Preparation of TiO2-Sensitizer Films. Mesoporous thin films (10 µm) of ∼20 nm anatase TiO2 nanocrystallites were prepared by a sol-gel technique.13 The nanocrystalline TiO2 was adsorbed onto the side of a glass substrate previously coated with a conducting layer of fluorine-doped tin oxide (FTO). Freshly prepared films were reacted overnight with the tetrapyrroles in concentrated toluene solutions. The films were then rinsed in neat toluene and were utilized for optical and solarcell studies. The surface coverage of each film was on the order of 10-8 mol/cm2 based on comparison of the measured absorptance spectrum of the film with the spectrum calculated for a monolayer of the sensitizer (see below). Thus, each film had on the order of 100 sensitizer layers. Presumably, electron injection into the TiO2 semiconductor occurs primarily from the monolayer of sensitizers that are attached to the surface via the acid group(s). However, because the majority of the sensitizers in each film are in layers removed from the surface, it is impractical to ascertain from spectroscopic measurements the binding motif (number of acid groups involved, distance, orientation, etc) or packing density of the surface bound-sensitizer layer. F. Solar Cell Preparation and Examination. Incident photon-to-current efficiencies (IPCE) of the tetrapyrrole/TiO2/ FTO materials were obtained in a two-electrode configuration at room temperature. A working (photoanode) electrode of
15466 J. Phys. Chem. C, Vol. 111, No. 42, 2007
Stromberg et al.
CHART 1
TABLE 1: Summary of Density Functional Theory Calculations and Measured Redox Potentials compound
HOMO-1 (eV)
HOMO (eV)
LUMO (eV)
LUMO+1 (eV)
ZnTPP ZnP-A′ ZnP-I′ ZnP-EI′ ZnC-EI′ FbB-EI′
-5.12 -5.46 -5.20 -5.25 -5.14 -4.99
-5.05 -5.30 -5.05 -5.00 -5.00 -4.66
-2.09 -2.67 -2.20 -2.45 -2.44 -2.57
-2.09 -2.40 -2.18 -2.20 -1.84 -1.41g
∆Ea (LUMO-HOMO) (eV)
∆EQyb (eV)
∆∆EQyc (eV)
d E0/+1 1/2 (V)
d E0/-1 1/2 (V)
∆Ee (ox-red) (V)
2.96 2.63 2.85 2.55 2.56 2.09
3.00 2.85 2.93 2.80 2.93 2.84
0.07 0.43 0.17 0.50 0.74 1.49
+0.56 +0.65 +0.57f +0.50f +0.41 +0.34
-1.58 -1.21 -1.62 -1.62 -1.59 -1.27
2.14 1.86 2.19 2.12 2.00 1.61
a Difference in the LUMO and HOMO energies. b Average of the (HOMO-1 f LUMO+1) and (HOMO f LUMO) promotion energies. Difference between the (HOMO-1 f LUMO+1) and (HOMO f LUMO) promotion energies. d Measured potentials (versus FeCp2/FeCp2+ ) +0.19 V) in butyronitrile/0.1 M n-BuN4PF6 unless otherwise noted. e Difference in the first oxidation and reduction potentials. f Measured potentials (versus FeCp2/FeCp2+ ) +0.19 V) in CH2Cl2/0.1 M n-BuN4PF6. g This orbital is actually the LUMO+2 of the molecule (see text).
c
known area was immersed in acetonitrile containing 0.5 M LiI/ 0.05 M I2; the counter electrode was a Pt mesh. The photoanode was illuminated through the FTO with a 100 W Xe lamp (PhotoMax) coupled to a 1/4 m monochromator (Oriel Cornerstone). Incident irradiances were measured using an optometer (Graesby Optronics S370 with a United Detector Technology silicon detector), and photocurrents were measured using an electrometer (Keithley 617). IPCE values were calculated using eq 1. IPCE )
1240 (eV nm) × λ (nm) Photocurrent (A) × Photodiode responsivity (V-1) (1) Photodiode photocurrent (A)
Current-voltage characteristics were obtained in a sandwich cell arrangement. The sandwich cell was prepared by pressing a tetrapyrrole-sensitized TiO2/FTO electrode against a Pt-coated conducting glass counter electrode. A Surlyn polymer spacer was placed between the two electrodes. The cell was then heated to 80 °C for 20-25 min until a tight seal was formed between the two electrodes. A solvent/electrolyte solution comprised of valeronitrile:acetonitrile (1:1) containing 0.6 M 3-butyl-1methylimidazolium hexafluorophosphate, 0.5 M p-tert-butylpyridine, 0.6 M LiI, and 0.05 M I2 was introduced between the electrodes through a hole that was drilled through the counter electrode prior to assembly. The hole was then sealed with a Surlyn film and a microscope cover slide. A current-voltage plot was obtained by illuminating the cell with quasimonochromatic light centered at 450 nm. Photocurrents and
photovoltages were measured with an electrometer (Keithley 617) and digital multimeter (Keithley 199 DMM/Scanner), respectively. The incident irradiance at 450 nm for the power conversion efficiencies was measured using the optometer and silicon detector described above. The simulations of the solar-cell performance of the planar and mesoporous films utilized the airmass (AM) 1.5 solar spectrum. This spectrum is available from the National Renewable Energy Laboratory (NREL) website.14 III. Results A. Physicochemical Characteristics of the Tetrapyrroles. 1. Frontier Molecular Orbitals. DFT calculations were performed on the five tetrapyrrole esters to obtain insights into the manner in which changes in the tether (ZnP-A′, ZnP-I′, and ZnP-EI′) and state of macrocycle reduction (ZnP-EI′, ZnC-EI′, and FbB-EI′) influence the characteristics (energy and electrondensity distributions) of the frontier molecular orbitals of the tetrapyrrole macrocycle. In turn, the changes in molecular-orbital characteristics aid in the interpretation of the redox and spectroscopic properties of the compounds. Tetrapyrroles have four frontier molecular orbitals that are regarded as paramount in determining the electronic (e.g., optical and redox) properties of the molecules. The calculated energies of these orbitals for the five tetrapyrrole esters are listed in Table 1. For comparison, the calculated orbital energies for zinc tetraphenylporphyrin (ZnTPP) are also included in the table. The trends in the orbital energies and the associated electrondensity distributions are illustrated in Figure 2. The highest-
Tethered Porphyrin, Chlorin, and Bacteriochlorin
J. Phys. Chem. C, Vol. 111, No. 42, 2007 15467
Figure 2. Electron-density distributions and energies of the four frontier molecular orbitals (HOMO-1, HOMO, LUMO, LUMO+1) obtained from DFT calculations. Note that the symmetry axes for porphyrins ZnP-A′ and ZnP-EI′ are rotated by 45° from those for the other compounds because of the effect of the cyano and ethynyl groups, respectively, on the tether. The effect of the saturated ring(s) in chlorin ZnC-EI′ and bacteriochlorin FbB-EI′ dominate over the effect of the tether so that the symmetry axes remain the same as those for porphyrins ZnTPP and ZnP-I′. Note also that the orbital shown as the LUMO+1 for bacteriochlorin FbB-EI′ is actually the LUMO+2; the LUMO+1 orbital is effectively an orbital on the ethynylaryl group of the tether that has dropped in energy because of electronic interaction with the porphyrin macrocycle.
15468 J. Phys. Chem. C, Vol. 111, No. 42, 2007 filled orbitals of ZnTPP are a1u(π) and a2u(π), and the lowestempty orbitals are egx(π*) and egy(π*) under the D4h symmetry.15 This symmetry group is approximately appropriate for the tetraarylporphyrin (ZnP-I′) and reasonably so for the other two porphyrins (ZnP-A′ and ZnP-EI′). The a2u(π) orbital places considerable electron density at the meso-carbons (to which the tether is attached), whereas the a1u(π) orbital has the greatest electron density at the pyrrole carbons (the site of macrocycle reduction in the chlorin and bacteriochlorin). The chlorin (ZnC-EI′) and bacteriochlorin (FbB-EI′) have approximate C2V symmetry, for which the orbital designations are b1(π), a2(π), b1(π*), and a2(π*). These orbitals mirror the a2u(π), a1u(π), egx(π*) and egy(π*) orbitals of the analogous porphyrin, respectively. Because of the electron-density distributions, the relative energies of the a2u(π) and a1u(π) orbitals (which are the HOMO and the HOMO-1, respectively) and the relative energies of the two eg(π*) orbitals (which are the LUMO and the LUMO+1, respectively) are expected to depend on the substituent pattern and state of macrocycle reduction. For example, the results in Table 1 and Figure 2 indicate that there is (as expected16,17) a switch in the HOMO from the a2u(π) orbital for all three porphyrins (ZnP-I′, ZnP-A′, ZnP-EI′) to the a1u(π)-derived a2(π) orbital for the chlorin and bacteriochlorin (ZnC-EI′, FbB-EI′). In addition to the nature of the macrocycle, the tether plays a role in the molecular-orbital characteristics. Electron density is delocalized to a different extent onto the three tethers [aryl (I′), ethynylaryl (EI′), cyanoacrylate (A′)], and for a given tether this delocalization differs among the orbitals (Figure 2). In this regard, the orbital that is effectively the HOMO on the tether group lies ∼1.3 eV below the porphyrin HOMO in both ZnP-A′ and ZnP-EI′. The LUMO of the tether group lies ∼1.2 and ∼0.8 eV higher in energy than the porphyrin LUMO in ZnP-A′ and ZnP-EI′, respectively. Additionally, the LUMO of the ethynylisophthalate group in bacteriochlorin FbB-EI′ is sufficiently low in energy that it falls between the LUMO and LUMO+1 of the tetrapyrole. Thus, the LUMO of the tether is actually the LUMO+1 of the molecule, and the LUMO+1 of the tetrapyrrole is the LUMO+2 of the molecule (see legend to Figure 2). The effect of the tether can be sufficiently large so as to cause a rotation of the symmetry axes in certain cases. Porphyrins ZnTPP and ZnP-I′ follow the standard convention in which the x-axis bisects pyrrole rings B and D, and the y-axis bisects pyrrole rings A and C (Figure 3A). On the other hand, the cyanoacrylate-bearing tether of ZnP-A′ and the ethynylisophthalate-bearing tether of ZnP-EI′ result in a rotation of the x and y axes to lie along the meso-positions of the macrocycles (Figure 3B). The symmetry-axis rotation is apparent in all of the frontier molecular orbitals of ZnP-A′ and ZnP-EI′, but it is most apparent in the LUMO and LUMO+1. Although the ethynylisophthalate-bearing tether is also present in chlorin ZnC-EI′ and bacteriochlorin FbB-EI′, the pyrrole-ring saturation (ring D of the former and rings B and D of the latter) dominates, and the symmetry axes retain the positions described above for porphyrins ZnTPP and ZnP-I′. The molecular-orbital characteristics illustrated in Figure 2 and Table 1 are discussed below in conjunction with the redox and optical data. 2. Redox Properties. Solution electrochemical studies were performed on ZnP-A′, Zn-PI′, ZnP-EI′, ZnC-EI′, and FbB-EI′. The redox potentials for the first oxidative and first reductive waves are reported in Table 1. For comparison, the redox potentials of ZnTPP are also included in the table. The redox potentials for ZnP-I′ are typical of those observed for tetra-
Stromberg et al.
Figure 3. Structural representations and transition axes for the tetrapyrrole macrocycles. Porphyrins ZnTPP and Zn-PI, chlorin ZnCEI, and bacteriochlorin FbB-EI have directions for the symmetry axes of the molecular orbitals that bisect the N-N axes (panel A), whereas the symmetry axes for porphyrins ZnP-A and ZnP-EI bisect the meso carbons (panel B). The tetrapyrroles have saturated pyrrole rings as follows: bacteriochlorin (B and D), chlorin (D), and porphyrins (none).
arylporphyrins.18 Replacement of the aryl group of ZnP-I′ with the 15-ethynylaryl group of ZnP-EI′ shifts the oxidation potential by 0.07 V to a less positive value, consistent with previous studies of a structurally related porphyrin.19 The finding that ZnP-EI′ is easier to oxidize than ZnP-I′ is reflected in the calculated destabilizing effect of the 15-ethynylaryl group on the a2u(π) HOMO, which places considerable electron density at the meso-carbons (to which the tether is attached; Chart 1). In particular, the HOMO is destabilized (shifted to less negative values) by 0.05 eV (Table 1 and Figure 2). In comparison with the effect on the oxidation potential (and HOMO energy), replacement of the aryl group of ZnP-I′ with the 15-ethynylaryl group of ZnP-EI′ has no measurable effect on the reduction potential. This observation is not well reproduced by the calculations, which predict that the LUMO of ZnP-EI′ should be stabilized relative to that of ZnP-I′ (rendering the former molecule more easily reducible). Saturation of a pyrrole ring of ZnP-EI′ to form the chlorin ZnC-EI′ results in a 0.09 V shift of the oxidation potential to a less positive value. The easier oxidation of ZnC-EI′ versus ZnP-EI′ is consistent with a destabilization of the a1u(π) orbital, which places considerable electron density at the pyrrole carbons (the site of macrocycle reduction), shifting this orbital to above the a2u(π) orbital to become the HOMO in the chlorin (designated a2(π)). This orbital-ordering reversal is consistent with results of the DFT calculations, which places the a2(π) HOMO of ZnC-EI′ 0.14 eV higher (less negative) in energy than the b1(π) (a2u-like) HOMO-1 of the chlorin (Table 1 and Figure 2). The chlorin differs from the porphyrin (ZnC-EI′ versus ZnP-EI′) in exhibiting a small (0.03 V) shift of the reduction potential to a less negative value. This finding is
Tethered Porphyrin, Chlorin, and Bacteriochlorin consistent with the calculated minimal shift in LUMO energy (∼0.01 eV), which arises because pyrrole-ring saturation primarily affects the LUMO+1 and not the LUMO (Figure 2). In the case of bacteriochlorin FbB-EI′, the oxidation potential is even less positive than that of the chlorin ZnC-EI′ (by 0.07 V), indicating a further destabilization of the a2(π) HOMO of the former versus latter molecule. Conversely, the reduction potential of FbB-EI′ is substantially less negative than that of ZnC-EI′ (by 0.32 V), indicating much greater stabilization of the LUMO of the former versus the latter molecule. These trends in the redox potentials are faithfully reproduced by the calculated HOMO and LUMO energies of the chlorin versus bacteriochlorin. Note that quantitative comparison of the redox potentials for FbB-EI′ versus ZnC-EI′ (or ZnP-EI′ and ZnP-I′) reflects not only the macrocyclic reduction state but also the metalation state because FbB-EI′ is a free base, whereas the other three compounds are zinc complexes. The redox properties of the porphyrin bearing the cyanoacrylate tether (ZnP-A′) do not follow the same trends as the other compounds. Compared to porphyrin ZnP-I′, which has an aryl group in the linker, ZnP-A′ has a somewhat more positive oxidation potential (0.08 V) as well as a substantially less negative reduction potential (0.41 V). Thus, ZnP-A′ is both harder to oxidize and easier to reduce than the chlorin or other porphyrins. This trend is reproduced by the calculated energies of the HOMO and LUMO of ZnP-A′ versus the chlorin and other porphyrins. In particular, the calculations predict that the HOMO and LUMO of ZnP-A′ are both stabilized versus the corresponding orbitals of the chlorin and other porphyrins, with the stabilization being the largest for the LUMO (Table 1). The origin of the large effect on the LUMO of ZnP-A′ is apparent by inspection of the electron-density diagrams (Figure 2), which exhibit significant density on the cyanoacrylate. ZnP-A′ also displays optical characteristics that distinguish it from the other sensitizers, as discussed below. 3. Absorption Spectra and Electronic Structure. The absorption spectra of ZnP-A′, ZnP-I′, ZnP-EI′, ZnC-EI′, and FbB-EI′ in toluene are shown in Figure 4. Similar spectra, with modest changes in peak positions and intensities are obtained for the compounds in acetonitrile, in which the central metal of the zinc chelate is solvent coordinated (spectra not shown).11 The optical characteristics of these ester-bearing compounds are essentially the same as those of the analogues containing an acid group in the tether, and thus should be representative of the absorption characteristics of the molecules in the solar cells. The absorption spectra of all the compounds contain a nearultraviolet (UV) Soret (B) band and a series of Q bands to longer wavelengths. The band positions and assignments are given in Table 2. Each absorption band is comprised of a degenerate pair of xand y-polarized transitions for the approximately D4h-symmetry zinc porphyrins ZnP-I′, ZnP-EI′, and ZnP-A′. For these compounds, the long-wavelength and near-UV Soret absorptions are the degenerate Qx,y(0,0) and degenerate Bx,y(0,0) bands, respectively. Because of degeneracy, the same wavelength is listed for the Qx(0,0) and Qy(0,0) bands in Table 2. The Q(0,0) band(s) of the porphyrins shift to longer wavelength along the series ZnP-I′ (591 nm) < ZnP-EI′ (614 nm) < ZnP-A′ (621 nm). The same trend is observed for the B(0,0) band(s): ZnP-I′ (426 nm) < ZnP-EI′ (441 nm) < ZnP-A′ (449 nm). For the nominal C2V-symmetry zinc chlorin ZnC-EI′, the near-UV Soret band (431 nm) contains unresolved x- and y-polarized components, whereas the Qx and Qy manifolds split. The Qx origin and vibronic features of the chlorin remain near the
J. Phys. Chem. C, Vol. 111, No. 42, 2007 15469
Figure 4. Normalized absorption spectra of (A) ZnP-A′, (B) ZnP-I′, (C) ZnP-EI′, (D) ZnC-EI′, and (E) FbB-EI′ in toluene at room temperature.
positions (500-600 nm) found for the analogous porphyrin (ZnP-EI′), whereas the Qy(0,0) is red-shifted to 624 nm. The Qy(0,0) band is further red-shifted (to 757 nm) in the free-base bacteriochlorin FbB-EI′. Furthermore, the x- and y-polarized near-UV Soret components of the bacteriochlorin are split (387 and ∼365 nm, respectively) because of the presence of the two protons in place of the central zinc ion; this splitting results in a much broader Soret envelope for the free-base bacteriochlorin compared to the zinc chlorin and zinc porphyrins. Broader absorption bands (for the same peak value) can have a positive impact on solar-cell performance because more of the solar spectrum can be utilized. Thus, the broad Soret manifold for the free-base bacteriochlorin is useful, albeit this band is located in a region ( ZnP-EI′ (2.00 eV) > ZnC-EI′ (1.98 eV) > ZnP-A′ (1.93 eV) > FbB-EI′ (1.63 eV). This trend deviates somewhat from the calculated average of the (HOMO f LUMO) and (HOMO-1
f LUMO+1) energies (∆EQy, Table 1), which decreases along the series ZnP-I′ (2.93 eV) ∼ ZnC-EI′ (2.93 eV) > ZnP-A′ (2.85 eV) > FbB-EI′ (2.84 eV) > ZnP-EI′ (2.80 eV). The clear outlier in the series is ZnP-EI′; the origin of this deviation between the observed versus calculated trends in the energies is uncertain. In addition to the position of the long-wavelength absorption band, the four-orbital model allows insights into the origin of the change of intensity of this band (both absolute and relative to the Soret) among the five tetrapyrrole complexes. The porphyrin bearing the isophthalic ester tether (ZnP-I′) is effectively a tetraarylporphyrin and serves as a benchmark for comparison purposes. For such complexes, Qx,y(0,0) is extremely weak, exhibiting only ∼20% of the intensity of the Qx,y(1,0) band (Figure 4B). This effect derives from the fact that the a2u(π) HOMO lies only slightly above the a1u(π) HOMO-1, and the LUMO and LUMO+1 are nearly degenerate (Table 1). Thus, the difference in energy of the (HOMO f LUMO) and (HOMO-1 f LUMO+1) one-electron promotions is very small (∆∆EQy ) 0.17 eV, Table 1), and the Qx,y(0,0) band is expected to be very weak, as is observed. The origin of the increase in intensity of the Qx,y(0,0) band upon incorporation of the ethynylaryl tether in ZnP-EI′ (Figure 4C) is more difficult to quantify. The redox data suggest that the a2u(π) HOMO of this molecule is destabilized relative to that of ZnP-I′, whereas the LUMO is little affected. This orbital perturbation would decrease the HOMO-LUMO spacing (∆EQy, Table 1) and nominally increase the energy difference between the (HOMO f LUMO) and (HOMO-1 f LUMO+1) promotions, which is consistent with the observed increase in the intensity of the Qx,y(0,0) band. However, the calculations suggest that the true picture is more complicated and involves shifts of all the frontier molecular orbitals (albeit the calculated LUMO shift is an overestimation given the redox data), with the net result being a significant increase in the energy difference between the (HOMO f LUMO) and (HOMO-1 f LUMO+1) promotions (∆∆EQy increases to 0.50 eV, Table 1). Similarly, the increase in Q(0,0) band intensity along the series ZnP-EI′ to ZnC-EI′ to FbB-EI′ (Figures 4C-E) is predicted to result from a combination of energy shifts of all four frontier molecular orbitals. The net result is that the energy difference between the (HOMO f LUMO) and (HOMO-1 f LUMO+1) promotions increases along the series (∆∆EQy ∼0.50, 0.74, and 1.49 eV, Table 1). The replacement of the aryl ring in the tether of ZnP-I′ with the cyanoacrylate group of ZnP-A′ has somewhat different consequences than the incorporation of the ethynylisophthalate tether in ZnP-EI′; this issue requires closer examination. Although both changes result in a red-shift and intensification of the Qx,y(0,0) band (Figure 4A and Table 2), ZnP-A′ exhibits a small positive shift in the oxidation potential and a much larger positive shift in the reduction potential, whereas ZnP-EI′ exhibits a small negative shift in oxidation potential and no change in reduction potential (Table 1). As noted above, the fact that
Tethered Porphyrin, Chlorin, and Bacteriochlorin ZnP-A′ is slightly more easily oxidized but much harder to reduce than ZnP-I′ is understood in terms of the calculated modest stabilizing effect on the a2u(π) HOMO (-0.25 eV) and the much larger stabilizing effect on the LUMO (-0.47 eV). The calculations also show that the cyanoacrylate group has a small stabilizing effect on the a1u(π) HOMO-1 (-0.26 eV) and on the LUMO+1 (-0.22 eV), comparable to the effect on the HOMO. As a consequence, the average (HOMO f LUMO) and (HOMO-1 f LUMO+1) energy, like the (HOMO f LUMO) gap itself, decreases, giving rise to a red-shift in the Qx,y(0,0) band. In parallel, the difference in the (HOMO f LUMO) and (HOMO-1 f LUMO+1) energies for ZnP-A′ increases as compared to ZnP-I′ (∆∆EQy increases to 0.43 eV from 0.17 eV), resulting in an intensification of the Qx,y(0,0) band. Thus, the combined optical and redox effects of the cyanoacrylate group are driven by its net electron-withdrawing effect on the energies of all the orbitals combined with a differentially large effect on the LUMO (Table 1 and Figure 2). Additionally, the effects of the tether in ZnP-A′ may not derive solely from direct electronic perturbations but may also involve (perhaps interconnected) conformational perturbations. This latter possibility is consistent with the absorption bands of ZnP-A′ being significantly broader than those of the other two porphyrins (noted above). This compound also exhibits a similarly broad and featureless fluorescence profile, a large shift between the Qx,y(0,0) absorption and fluorescence maxima, and a reduced excited-state lifetime (Table 1).11 These effects may derive from the presence of multiple conformers and/or nonplanar conformational distortions with associated electronic perturbations. B. Photocurrents in Mesoporous-Film Regenerative Solar Cells. The light-to-electrical-energy efficiencies were quantified for regenerative solar cells containing ZnP-A, ZnP-EI, ZnP-I, FbB-EI and ZnC-EI anchored to mesoporous (nanocrystalline) TiO2 thin films (hereafter referred to as the “mesoporous films”). Figure 5A shows the IPCE plotted as a function of the excitation wavelength, namely a photocurrent action plot. The wavelengths of the maximum photocurrent in the Soret and Q-band regions are in good agreement with the features in the measured absorptance spectra of the same films (Figure 5B). Absorptance (R) is the fraction of the incident light absorbed, namely R ) 1 T, where T is the transmittance. The absorptance of the tetrapyrrole films is unity at many wavelengths; thus, the shapes and peak positions are not fully revealed. To compare with the full spectral shapes, Figure 5C shows the calculated absorptance spectra of a hypothetical planar film of each sensitizer (hereafter referred to as a “planar film”). The assumptions used to generate this spectrum will be discussed in more detail below. The planarfilm spectra are also used below in simulations of solar-cell performance. The agreement between the IPCE spectra in Figure 5A and the absorptance spectra in Figure 5B (and 5C) indicates that light absorption by the tetrapyrrole molecules gives rise to the observed photocurrents. The IPCE should match the absorptance spectra in both shape and magnitude for an idealized solar cell, namely, one in which every photon absorbed leads to an electron in the external circuit. Thus, the finding that the amplitude of the IPCE at each wavelength across the spectrum is lower than the absorptance, which indicates that factors other than photon absorption by the tetrapyrrole sensitizer are limiting cell performance. One caveat in comparing the plots in Figures 5A and 5B is that the IPCE curves, unlike the absorptance spectra of the same films, were obtained in the presence of the iodide/ triiodide/iodine electrolyte, which competitively absorbs light, particularly in the near-UV Soret (B) region.
J. Phys. Chem. C, Vol. 111, No. 42, 2007 15471
Figure 5. (A) IPCE of mesoporous TiO2 sensitized with ZnP-A (black), ZnP-I (blue), ZnP-EI (orange), ZnC-EI (magenta), FbB-EI (green) in acetonitrile/0.5 M LiI/0.05 M I2. (B) Absorptance (R) of the mesoporous films used for the measurements in panel (A). (C) Calculated absorptance of hypothetical planar films of ZnP-A′, ZnP-I′, ZnP-EI′, ZnC-EI′, FbB-EI′.
IV. Discussion The IPCE measurements indicate that the magnitude of the photocurrent is dependent upon the nature of the tetrapyrrole molecule. For example, the IPCE at each wavelength from approximately 370 to 740 nm is greatest for porphyrin ZnP-A and lowest for porphyrin ZnP-EI. Furthermore, the only sensitizer that gives nonzero IPCE past ∼750 nm is the bacteriochlorin FbB-EI. A number of factors could potentially contribute to these effects. In the sections below, we first give a general description of the factors that control photocurrent generation. We next examine and compare the performance of ideal cells based on planar and mesoporous films of the tetrapyrroles. We then compare the performance of the actual film-based cells to the ideal cells and assess the factors that limit cell performance. Following a description of the power conversion efficiency in the most effective cell, we conclude with some general comparisons. A. Factors Controlling Photocurrent Generation. The photocurrent produced in a regenerative solar cell is the product of the three terms that are indicated in eq 2.
IPCE ) R × φinj × η
(2)
15472 J. Phys. Chem. C, Vol. 111, No. 42, 2007 Here, R is the absorptance, φinj is the excited-state electroninjection efficiency (the fraction of the absorbed photons that result in electron transfer to the semiconductor), and η is the efficiency with which injected electrons are collected in the external cell circuitry (which includes hole transfer from the photooxidized sensitizer to the electrolyte and then to the counter electrode). Below we evaluate each of these three variables to better understand the relationship between the measured photo current and the light-harvesting, photophysical, and redox properties of the tetrapyrrole molecules. As part of this analysis, it is instructive to utilize the known absorption properties of each tetrapyrrole in solution to carry out first-principles calculations of the solar-cell efficiency parameters for two types of hypothetical cells, one utilizing the estimated absorptance of planar films and the other using the measured absorptance of the mesoporous films. These parallel simulations give insight into the molecular factors modulating cell efficiency that complement the cell-efficiency parameters obtained directly from the measured photocurrent action spectra of film-based cells. 1. Light-HarVesting Characteristics. The absorptance (R) contribution to the IPCE in eq 2 is often referred to as the ‘lightharvesting efficiency.’ The films contain a large number of anchored tetrapyrrole chromophores. Thus, as is shown in Figure 5B, the absorptance is effectively unity at many wavelengths. This is particularly true in the strong near-UV Soret region (e.g., 400 nm) of all of the tetrapyrrole sensitizer molecules. The absorptance is also effectively unity in the red Qy(0,0) band of the bacteriochlorin (FbB-EI). As noted above, FbB-EI has greater absorption in this region than the porphyrin and chlorin analogues (Figure 4), both on an absolute basis (i.e., a larger molar absorption coefficient) and relative to the absorption in the violet-blue (Soret) region. Absorptance spectra for hypothetical cells containing a tetrapyrrole planar film can be calculated provided that two quantities are known.22 These parameters are the optical cross section (σ), which is related to the molar absorption coefficient (), and the ‘footprint’ of the molecule on the surface. The relationship between these parameters is given in eq 3, which gives the absorptance at a particular wavelength.
R ) σ(Å2)/footprint (Å2) ) (3.82 × 10-5 × )/footprint (3) The footprint of each tetrapyrrole molecule was estimated based on van der Waals radii using Chemdraw-3D. The estimated footprints of ZnP-A, ZnP-I, ZnP-EI and ZnC-EI are comparable to one another (∼115 Å2), whereas that of FbB-EI is slightly larger (∼132 Å2). The molar absorption coefficients used in the analysis were described above and are given in Table 2.11 The values (with units of mM-1 cm-1) in the long-wavelength Q(0,0) band for the compounds in toluene are as follows: ZnP-I′ (∼4.5), ZnP-A′ (∼31), ZnP-EI′ (∼33), ZnC-EI′ (∼44), and FbC-EI′ (∼120). All polarizations of light may not be absorbed depending on the adsorption geometry on the surface and the orientation of the surface with respect to the incident photon flux. Accordingly, the effective molar absorption coefficient may be lower than the value determined for a collection of non-oriented molecules in solution. If the macrocycle plane of the tetrapyrrole lies parallel to the surface and if the incident light is randomly polarized and impinges along the surface normal, then all polarizations can be absorbed, and the full extinction (greater than that of a randomly oriented solution sample) can be achieved. On the other hand, if a tetrapyrrole macrocycle lies perpendicular to the surface, then the polarization of light perpendicular to the macrocycle plane (the
Stromberg et al. z-direction) will not be absorbed. The actual fraction of the light absorbed in the latter case may be further reduced at particular wavelengths, depending on the orientation of the associated inplane transition dipole with respect to the macrocycle plane and the surface normal. The latter will differ for the zinc porphyrins, in which the x- and y-polarized transitions are degenerate, and the chlorins/bacteriochlorins, in which one or more of the bands with different in-plane polarizations are split. Thus, for the purposes of the calculations at hand, a rough average over all of these possibilities is used. In particular, for each tetrapyrrole sensitizer, the molar absorption coefficient in the longwavelength Q-band is reduced by a factor of 2 from that given in Table 2. That value (/2) is then utilized for that sensitizer along with its footprint to obtain the absorptance at that wavelength using eq 3. The absorptance at all other wavelengths spanning 350-800 nm was then obtained using the relative intensities from the solution-phase spectra of the ester-derivative of that sensitizer in toluene (Figure 4). The resulting calculated absorptance spectra for the planar film of the sensitizers are those shown in Figure 5C. 2. Excited-State Electron-Injection Efficiency (φinj). Gerischer suggests that the excited-state electron injection will be activationless when the excited-state reduction potential E(S+/*) is less than twice the reorganization energy above the conduction band edge.23 The conduction band edge of anatase TiO2 has been reported to be -1.0 V versus SCE in acetonitrile/1.0 M LiClO4 (-0.810 V vs FeCp2+/FeCp2 ) 0.19 V). A reorganization energy of ∼0.25 meV has been estimated for the excitedstate electron injection.24 The excited-state reduction potentials of the porphyrinic molecules can be estimated from the difference in the ground-state redox potentials, E1/2(S+/0), and the energy contained in the excited singlet state (E00), using eq 4.
E(S+/*) ) E1/2(S+/0) - E00
(4)
The E1/2(S+/0) is obtained directly from the voltammetric measurements (Table 1), and the E00 term is estimated by taking the average of the Q(0, 0) absorption and fluorescence energies (Table 2). For all of the tetrapyrrole sensitizers, eq 4 gives E(S+/*) < -1.3 V vs FeCp2+/FeCp2 ) 0.19 V. Therefore, electron injection into the TiO2 conduction band is expected to be thermodynamically favorable and is likely to be activationless for all five tetrapyrrole sensitizers. Kinetically, excited-state electron injection to TiO2 surfaces from photoexcited chromophores has been shown to occur on the picosecond to femtosecond time scales.25 In comparison, the excited-state lifetimes of the tetrapyrrole sensitizers vary between ∼1 and ∼5.5 ns (Table 1).11 Thus, to a first approximation, electron injection is expected to occur with a rate constant at least 1000-fold greater than the inherent excitedstate decay rate (by fluorescence, internal conversion, and intersystem crossing). In this regard, the sensitizer with even the shortest (∼1 ns) excited-state lifetime of all five tetrapyrrole molecules (ZnP-A′) produces the greatest photocurrent. One might also expect the acid derivative ZnP-A, when tethered to a surface, to have one of the largest effective electronic couplings for electron injection (relative to the other linkers), on the basis of (1) a shorter distance from the surface associated with the effective length of the tether (depending on the exact binding motif) and (2) the substantial electron density that resides on the cyanacrylic acid group in the LUMO of this molecule (Figure 2). In the simplest picture, this is the orbital from which the electron will be transferred from the photoexcited sensitizer to the semiconductor. On the basis of these and
Tethered Porphyrin, Chlorin, and Bacteriochlorin related considerations, one may expect differences in the exact electron-injection rates among the five tetrapyrroles studied here, albeit these rates are likely much faster than the inherent excitedstate deactivation rate in each case. Thus, in the absence of direct measurements of these rates, and in view of the kinetic and thermodynamic considerations discussed above, we assume for the present discussion that electron injection is highly efficient for all five tetrapyrrole sensitizers. Accordingly, φinj is assumed to be unity in all of the simulations presented below. 3. Electron-Collection Efficiency (η). The a priori assessment of the electron-collection efficiency is more difficult than the assessment of the excited-state electron-injection efficiency. Factors that affect η include the competition of electron-transfer and charge-recombination processes, such as those involving (1) the oxidized sensitizer and the reduced metal-oxide semiconductor, (2) the reduced metal oxide and the oxidized electrolyte, and (3) the electrolyte and counter electrode. Because there is no obvious reason to expect the latter process to be dependent upon the type of tetrapyrrole, one (or both) of the first two processes would be more likely to limit η. A detailed analysis of the measured photocurrents in the film-based solar cells points to the redox characteristics of the tetrapyrroles being a key factor that limits η. This issue will be discussed in more detail below. B. Calculated Ideal Photocurrent Generation by Hypothetical Planar Films. The absorptance spectrum for a planar film of each tetrapyrrole obtained as described above was used to obtain the photocurrent produced by the absorption process in a hypothetical cell utilizing that sensitizer under AM 1.5 solar irradiation. The AM 1.5 spectrum is typically displayed as power density per wavelength (W m-2 nm-1 ≡ J sec-1 m-2 nm-1). The peak in this spectrum is at ∼480 nm. However, what is most directly relevant to current generation is the number of photons available at each wavelength (λ). In Figure 6A, the AM 1.5 spectrum is presented as photon flux density per wavelength per unit area (photon sec-1 cm-2 nm-1), in which the common practice of expressing the surface area in square centimeters has been used to accommodate a finite-size electrode. The peak in this spectrum is at ∼660 nm. The number of photons available at each wavelength (per second per square centimeter) was then multiplied by the absorptance of the hypothetical planar film to obtain the current density J (coulombs sec-1 cm-2 nm-1 ≡ amperes cm-2 nm-1) produced at that wavelength. This calculation utilized (1) the fact that there are 1.6 × 10-19 coulombs per electron, (2) the assumption that each photon absorbed by the sensitizer gives rise to an electron in the semiconductor, namely an absorbed photon-to-current efficiency (APCE) of unity (φinj ) 1 in eq 2), and (3) the assumption that each electron injected into the semiconductor and associated hole extracted from the photooxidized semiconductor pass with unity efficiency through the cell circuitry (η ) 1 in eq 2). Such values are denoted herein as “ideal” current densities because of assumptions (2) and (3) above. Integration over all wavelengths of light (350-900 nm) gives the total ideal current density for that cell. The ideal current density (µA cm-2 nm-1) generated at each wavelength for a hypothetical cell containing a planar film of each tetrapyrrole is shown in Figure 6A along with the AM 1.5 solar spectrum. The corresponding total (integrated) ideal current density Jtotal (mA/cm2) for each planar film is listed in the second column in Table 3. The value decreases in the order ZnP-A′ (0.14) > ZnP-EI′ (0.098) > FbB-EI′ (0.066) > ZnP-I′ (0.055) ∼ ZnC-EI′ (0.054). This trend tracks the values of the integrated (350-900 nm) absorptance spectra (Figure 5C) for these compounds, which follow the order ZnP-A′ (2.77) > ZnP-EI′
J. Phys. Chem. C, Vol. 111, No. 42, 2007 15473
Figure 6. (A) Calculated theoretical maximum photocurrent density produced by a hypothetical planar film of ZnP-A′ (black), ZnP-I′ (blue), ZnP-EI′ (orange), ZnC-EI′ (magenta) and FbB-EI′ (green) under 1 sun of AM 1.5 solar irradiance, expressed as photon flux density (dark red), for which the scale (right-hand ordinate) has been multiplied by 10-14 (i.e., the value at 800 nm is 5 × 1014). (B and C) Calculated theoretical maximum photocurrent density for mesoporous films of the analogous sensitizers containing the acid form of the tether, ZnP-A, ZnP-I, ZnP-EI, ZnC-EI, and FbB-EI, obtained using the measured absorptances of the films. The curve for 1 sun of AM 1.5 solar irradiance (dark red) is expressed as the maximum derivable current density assuming unity absorptance at each wavelength.
(2.06) > FbB-EI′ (1.63) > ZnP-I′ (1.32) > ZnC-EI′ (1.21). The latter values do not incorporate the precise spectral positions, but they do rely directly on the estimated integrated molar absorption coefficients (via eq 3) obtained for the molecules in solution, and thus modest differences must be viewed with an eye to the uncertainties involved. On the basis of the estimated molar absorption coefficients used in the calculations (Table 2), the predicted larger total ideal current density for ZnP-EI′ versus ZnP-I′ derives from both more intense Soret (B) and Qx,y(0,0) bands. The lower total ideal current density still for ZnC-EI′ is due to an even more reduced Soret absorption for a modestly more intense long-wavelength Q-band compared to ZnP-EI′. Interestingly, the estimated molar absorption coefficient for ZnP-A′ at the Soret maximum is about ∼30% lower than that for ZnP-EI′, and the Q-bands have about the same intensity, as
15474 J. Phys. Chem. C, Vol. 111, No. 42, 2007
Stromberg et al.
TABLE 3: Calculated Solar Cell Performance under AM 1.5 Solar Illumination calculated total ideal photocurrent densitya
sensitizer
planar film
ZnP-A ZnP-I ZnP-EI ZnC-EI FbB-EI
measured total photocurrent density
mesoporous film
mesoporous films
Jtotal (mA/cm2)b
% of max for AM 1.5c
Jtotal (mA/cm2)d
% of max for AM 1.5c
Jtotal (mA/cm2)e
% of max for AM 1.5c
% of idealf
0.14 0.055 0.098 0.054 0.066
0.42 0.16 0.29 0.16 0.20
18 12 15 15 22
54 36 45 45 66
7.6 2.8 0.035 0.99 1.6
23 8.4 0.10 3.0 4.8
42 23 0.23 6.6 7.3
a Simulated total current density for hypothetical planar films (using the ester analogues of the sensitizers listed) and actual mesoporous films were obtained assuming that each absorbed photon (350-900 nm) results in an electron in the external circuit (φinj ) η ) 1 in eq 1). b Calculated total current density for a planar film of the tetrapyrrole sensitizer, using the estimated footprint, molar absorption coefficients (Table 2), and the absorption spectrum for the ester analogue in toluene (Figure 2), to obtain the absorptance spectrum. c The percentage of the total current density in the previous column with the maximum value (33.4 mA/cm2) that can be derived for AM 1.5 irradiation assuming a unity absorptance at each wavelength (350-900 nm) and that each photon absorbed produces an electron current in the cell circuitry. d Calculated current density for each sensitizer-loaded mesoporous film using the experimentally measured absorptance spectrum of the film (Figure 5B). e The calculated current density for each mesoporous film using the experimentally measured IPCE spectrum (Figure 5A). f The value (in percent) is 100 times the entry in column 6 divided by that in column 4. [This value is also 100 times the entry in column 7 divided by that in column 5, because, in both cases the second value is derived from the first by the common factor described in footnote c.] This value is the percentage of the ideal photocurrent obtained for the mesoporous film (obtained assuming unity electron-injection and -collection efficiencies, namely φinj ) η ) 1 in eq 1) that is actually realized for that same film (with no assumptions). Thus, the values listed are a measure of non-unity φinj and/or η; the tentative analysis is that the values largely reflect the electron-collection efficiency η (expressed as percent).
reflected in the relative peak absorptance values (Figure 5C) and peak ideal current densities (Figure 6A). Nonetheless, the calculated total ideal current densities for cells based on hypothetical planar films of the two porphyrins is reversed, namely the current for ZnP-A′ is ∼30% larger than that for ZnP-EI′. A small fraction of this difference resides in the fact that the Soret band of ZnP-A′ is red-shifted slightly from that for ZnP-EI′, and thus lies in a slightly more photon-rich portion of the solar spectrum (Figure 6A). However, the major effect derives from the fact that all of the absorption bands of the former compound are significantly broader than those for the latter. For example, the full-width-at-half-maximum (fwhm) of the Soret band of ZnP-A′ (1670 cm-1) is roughly twice that for porphyrins ZnP-EI′ and ZnP-I′ (680 and 780 cm-1, respectively). Similar differences exist in the Q bands. Placing these bandwidth differences in another perspective, if the Soret bands of the three porphyrins had the same peak absorption coefficient, the integrated (350-900 nm) absorbance (or absorption) of ZnP-A′ would be about twice the values for ZnP-EI′ and ZnP-I′, and this is predominantly because of differences in bandwidths. The importance of the spectral widths is also seen, but to a lesser extent, in the calculated total ideal current density for a planar film of FbB-EI′. The Soret band of this compound is reduced in intensity and blue-shifted further into the near-UV range, where the solar intensity has fallen off dramatically (Figure 6A). The Soret contribution to the total ideal photocurrent density is compensated to some degree by the more significant width of this band as compared to ZnP-EI′, ZnP-I′, and ZnC-EI′ (and even ZnP-A′). More important to the prediction that the total ideal current density for FbB-EI′ remains appreciable (Table 3) is the fact that an enhanced red contribution (where the solar spectrum has more photons at each wavelength) offsets the effect of a reduced Soret contribution. The contribution of red photons to the total ideal photocurrent density for a planar film of FbB-EI′ would have been even greater except for the fortuitous overlap with a sharp notch in the solar spectrum that was previously attributed to absorption by oxygen or water vapor (Figure 6A).26 C. Calculated Ideal Photocurrent Generation with Mesoporous Films. In principle, the absorptance spectra of the mesoporous films and the currents generated under AM 1.5 illumination could be calculated by a procedure analogous to
that used above for planar films. However, eq 3 cannot be utilized directly because the exponential decay of the incident irradiance through the film must be taken into account. We have recently performed such an analysis in simulations of the photoresponse of cells containing surface-bound tetrapyrrole rods.22 However, given the uncertainties in the exact packing of the chromophores in the films studied here, such a calculation would necessarily contain additional assumptions. Thus, it is prudent to utilize the measured absorptance spectra (Figure 5B) to obtain the cell performance parameters for the films and to compare the results with those obtained above from firstprinciples simulations for planar films of the same molecules. Figure 6B shows the resulting ideal photocurrent density that would be produced at each wavelength (350-900 nm) for individual cells that contain a mesoporous film of each of the three porphyrin sensitizers [ZnP-A (black line), ZnP-I (blue), and ZnP-EI (orange)]. Figure 6C shows similar calculated ideal photocurrent-density spectra for cells that incorporate the two reduced tetrapyrrole sensitizers [ZnC-EI (magenta) and FbB-EI (green)]. Again, the measured absorptance spectrum of the film of a given sensitizer was used to obtain the spectra shown in Figures 6B and 6C, so the only assumptions are that φinj ) η ) 1. Integration of each calculated spectrum (350-900 nm) gives the total ideal integrated photocurrent density for that film. These ideal photocurrent densities are listed in the fourth column in Table 3. The values range from 12 mA/cm2 for ZnP-I to 22 mA/cm2 for FbB-EI. D. Comparison of Ideal Photocurrent Densities for Planar and Mesoporous Films. The calculations described in the two sections above give the total ideal photocurrent densities produced by AM 1.5 illumination (350-900 nm) of each of the five tetrapyrroles in hypothetical planar and actual mesoporous films. In both cases, the calculated ideal photocurrent densities are the maximum possible for a given absorptance profile (and AM 1.5 irradiation) because the calculations assumed that each photon absorbed produces an electron in the semiconductor and subsequent current generation in the external circuit with unit efficiency (φinj ) η ) 1 in eq 2). In the case of the simulations for a planar film, the factors that control the ideal photocurrent are (1) the magnitude of the absorptance at each wavelength, which depends on the molar absorption coefficient and the footprint (eq 2), (2) the number of photons
Tethered Porphyrin, Chlorin, and Bacteriochlorin available in the AM 1.5 spectrum at that wavelength (Figure 6A), and (3) the absorptance profile, namely, the overlap of the sensitizer and solar-irradiation spectra. For the mesoporous films, the same factors are involved, plus the loading (concentration) of the molecule on the surface. Because of the loading effects, the mesoporous films have a much larger absorptance than the planar films at a given wavelength, which results in calculated total ideal current densities that are ∼200 times larger for the former cells. In particular, the calculated total ideal current density averaged among the five tetrapyrroles is ∼16 mA/cm2 for the mesoporous films (Table 3, column 4) versus 0.08 mA/cm2 for the planar films (Table 3, column 2). It should be noted, however, that the increased current density for the mesoporous film does not necessarily scale linearly with the concentration of the sensitizer. When the absorptance of a film at a given wavelength (or across a given spectral region) effectively reaches unity, further increasing the loading does not result in a significant absorption of more photons (i.e., there is a saturation effect). Once saturation occurs in the vicinity of a spectral peak, what becomes most important for further increases in the production of photocurrent is the inherent (e.g., dilute-solution) spectral width of the absorption band and the resulting more extensive overlap with the solar spectrum. The above points can be seen in the calculated ideal photocurrent spectra for the five tetrapyrroles. This comparison is facilitated via the reformulation of the AM 1.5 solar spectrum as presented in Figure 6, panels B and C (brown). This spectrum gives the ideal photocurrent that would be produced by AM 1.5 illumination if a film (mesoporous or planar) had an absorptance of unity at each wavelength. Figure 6, panels B and C, shows that this AM 1.5 plateau value is reached at many wavelengths for the films studied because they have effective unity absorptance in certain regions. At a given wavelength, a calculated ideal photocurrent density lower than the AM 1.5 solar-irradiation plateau value simply results from a non-unity absorptance. Integration of the AM 1.5 plateau photocurrent density spectrum (350-900 nm) gives a maximum photocurrent density of 33.4 mA/cm2. This value would be attained if every available photon in the AM 1.5 spectrum over this region were absorbed by the mesoporous film and produced an electron current with unity efficiency (φinj ) η ) 1). Based on this value (33.4 mA/cm2), the simulated total ideal photocurrent densities for cells containing the same films can be recast in terms of the percentage of the maximum possible given the available AM 1.5 photon density (Table 3, column 5). Similarly, the calculated total current densities for planar films can be recast in terms of percentages of the maximum possible for AM 1.5 illumination (Table 3, column 3). Inspection of Table 3 shows that the percentage of the maximum ideal photocurrent achieved by the mesoporous films of the tetrapyrroles decrease in the order FbB-EI (66%) > ZnP-A (54%) > ZnP-EI (45%) ∼ ZnC-EI (45%) > ZnP-I (36%). In comparison, the percentage of the maximum achieved for the planar films decreases in the order ZnP-A′ (0.42%) > ZnP-EI′ (0.29%) > FbB-EI′ (0.20%) > ZnC-EI′ (0.16%) ∼ ZnP-I′ (0.16%). Thus, not only is there less variation in ideal photocurrents among the five tetrapyrroles in mesoporous films as compared to planar films, but the order of the values is different. The smaller variation for the mesoporous films derives from the saturation effect described above because the films have effective unity absorbance at many wavelengths (Figures 6B and 6C), whereas no planar film approaches that level. The altered ordering in relative ideal photocurrents derives from the
J. Phys. Chem. C, Vol. 111, No. 42, 2007 15475 effective widths of the (saturated) absorptance bands and the spectral positions under the solar umbrella. For example, in the films, FbB-EI absorbs all the available violet-blue photons between 350 and ∼430 nm, and also all of the AM 1.5 photonrich light in the green (510-560 nm) and red (720-790 nm) regions (Figure 6C). This effective use of broad regions of the solar spectrum, particularly the red region, in a loaded film gives FbB-EI the highest predicted ideal photocurrent density among all five of the tetrapyrrole sensitizers. The most notable comparisons are with porphyrin ZnP-EI and chlorin ZnC-EI, which share the same ethynylisophthalic acid tether and have approximately the same total ideal photocurrent as each other (Table 3). In contrast, for the planar films (which do not afford the absorptance-saturation effect), FbB-EI′ produces only ∼20% more current than chlorin ZnC-EI′ and ∼30% less current than porphyrin ZnP-EI′. The above findings address one of the goals of this study, which has been to probe cell photoresponse for a set of tetrapyrrole molecules that employ the same tether but incorporate a progressive increase in the relative red/blue absorption intensities achieved via macrocycle saturation (porphyrin f chlorin f bacteriochlorin). Collectively, the simulations for idealized solar cells (i.e., when only absorptance matters) suggest that the chlorin has the least favorable absorption characteristics of the set for both planar films (Figure 6A) and mesoporous films (Figure 6C). The extensive red-region absorption of the free-base bacteriochlorin results in a better predicted performance than the zinc porphyrin for the idealized cell when the surface loading results in saturated absorbance in the blueviolet region. These factors appear to mask (1) the inherently more intense (larger peak molar absorption coefficient) of the porphyrin in the Soret band, and (2) a possibly lower surface loading of the bacteriochlorin due to its larger footprint, although the relative loading levels were not ascertained. The results of the simulations also allow insights concerning a second goal of the study, which has been to probe the effects of three different surface-attachment tethers [cyanoacrylic acid (A), isophthalic acid (I), ethynylisophthalic acid (EI)] for the same porphyrin sensitizer. The simulated total photocurrent densities for idealized cells follows the order ZnP-A (or ZnP-A′) > ZnP-EI (or ZnP-EI′) > ZnP-I (or ZnP-I′) for both the mesoporous and planar films (Table 3). The dominance of the porphyrin containing the cyanoacrylic acid/ester tether rests primarily in the roughly 2-fold greater spectral bandwidths derived from this tether as compared to the other two. The larger photocurrents for the ethynylisophthalic acid/ester versus the isophthalic acid/ester tether can be traced primarily to the larger estimated molar absorption coefficient in both the Soret and Qx,y(0,0) bands, the latter of which lies in a photon-rich region of the AM 1.5 spectrum. E. Determinants of the Observed Photocurrents in Mesoporous Films. The insights obtained above from simulations of the ideal photocurrents in mesoporous (and planar) films of the five tetrapyrroles provide a framework for evaluating the measured IPCE curves for the same films (Figure 5A). In the analysis of these data, each measured IPCE curve replaces the measured absorptance curve in otherwise the same procedure performed above (convolution with the AM 1.5 solar spectrum) to simulate the ideal photocurrent generated. The resulting IPCEderived current-density spectra are again integrated to obtain the total current densities, which then can be compared with the maximum allowable for AM 1.5 irradiation (33.4 mA/cm2 from 350 to 900 nm) to obtain the percentage of the maximum
15476 J. Phys. Chem. C, Vol. 111, No. 42, 2007 actually achieved. These results are listed in Table 3, columns 6 and 7. It is important to emphasize that the IPCE-derived total current densities represent measured values and do not incorporate the assumptions φinj ) η ) 1 (eq 2) that were used in simulations of the performance of the idealized solar cells. Thus, the electron-injection and electron-collection efficiencies now supplement the efficiency of absorption of AM 1.5 photons (i.e., the absorbance) in impacting the measured total current density for each mesoporous film (eq 2). Comparison with the predicted photocurrent for the idealized cell (φinj ) η ) 1) gives insights into the combined electron-injection/collection efficiency (φinj × η) for that sensitizer. This latter comparison is given in the last column of Table 3, which lists the percentage of the total ideal photocurrent density that is actually achieved from the total IPCE-derived photocurrent density of the mesoporous films. These percentages decrease in the order ZnP-A (42%) > ZnP-I (23%) > FbB-EI (7.3%) > ZnC-EI (6.6%) > ZnP-EI (0.23%); thus, the combined electron-injection/collection efficiencies (φinj × η) also decrease in this order. Furthermore, inspection of Figure 5A indicates that the integrated (350-900 nm) percentages of ideal photocurrent closely track the magnitudes of the peak IPCE values in the blue-violet (Soret) region. In that region, each mesoporous film has effective unity absorbance, which means that all of the available photons of that color are essentially utilized, and thus, losses due to the transmission of light are minimal. Thus, the effect of the electron-injection and electron-collection efficiencies on the relative photocurrent production by the five tetrapyrroles is reflected similarly in both the spectrally integrated and the peak values of the photocurrent. The key observation is that porphyrin ZnP-A has the largest combined electron-injection/collection efficiency (φinj × η). The value for this sensitizer is roughly twice that of porphyrin ZnP-I, which has an efficiency that is roughly 3-fold greater than those of bacteriochlorin FbB-EI and chlorin ZnC-EI. The combined electron injection/collection efficiency of porphyrin ZnP-EI is surprisingly low by comparison, and this result is directly attributable to the very low-amplitude IPCE spectrum for the cell containing a mesoporous film of this sensitizer (Figure 5A). As previously discussed, the electron-injection yield (φinj) is assumed to be effectively unity. Under this assumption, the electron-collection efficiency (η) would then control the percentage of the ideal total photocurrent density that is realized experimentally for the mesoporous films (Table 3, column 8). The same argument applies to the non-unity values of the photocurrent at the Soret maximum, where the absorptance is effectively unity. Accordingly, the efficiency with which the injected electrons are collected in the external cell follows the trend ZnP-A (42%) > ZnP-I (23%) > FbB-EI (7.3%) > ZnC-EI (6.6%) > ZnP-EI (0.23%). A plausible source of a reduced electron-collection efficiency is inefficient oxidation of the iodide and subsequent reduction of the tetrapyrrole cation. If the iodide oxidation is sluggish, this process would not compete as favorably with recombination involving the hole on the oxidized tetrapyrrole and the electron injected into the semiconductor, thereby lowering the photocurrent production. In this regard, all of the tetrapyrrole reduction potentials are more positive than the I-/I3- equilibrium potential (roughly -0.4 V vs FeCp2/FeCp2+ ) +0.19 V) implying thermodynamic feasibility of iodide oxidation. However, in general, the relative electron-collection efficiencies ZnP-A (42%) > ZnP-I (23%) > FbB-EI (7.3%) > ZnC-EI (6.6%) > ZnP-EI (0.23%) might be expected to track the first oxidation potentials,
Stromberg et al.
Figure 7. Photocurrent-voltage characteristics of a ZnP-A-sensitized TiO2 slide in a sandwich cell configuration with 450 ( 5 nm illumination in valeronitrile:acetonitrile (1:1) containing 0.6 M 3-butyl1-methylimidazolium hexafluorophosphate, 0.5 M p-tert-butylpyridine, 0.6 M LiI, and 0.05 M I2. The calculated single-wavelength efficiency is 21% with a measured fill factor of 0.69.
which follows the trend ZnP-A (+0.65 V) > ZnP-I (+0.57 V) > ZnP-EI (+0.50 V) > ZnC-EI (+0.41 V) > FbB-EI (+0.34 V). These two trends parallel each other, except for porphyrin ZnP-EI, which exhibits an anomalously low photocurrent (Figure 5A and Table 3) and, by implication, a proportionately low η value. Accordingly, the relatively low values of the redox potentials of the reduced tetrapyrroles (ZnC-EI and FbB-EI) might possibly be the source of the low η values for these molecules. In contrast, porphyrin ZnP-A is the strongest oxidant (and most difficult to oxidize) among the tetrapyrroles, suggesting a more favorable electron-collection efficiency. This factor, along with the favorable optical characteristics noted above, are consistent with the observation that this molecule produces the greatest total photocurrent density (and Soret peak photocurrent). F. Power Conversion Efficiency. Of the five tetrapyrrole sensitizers studied, ZnP-A produces the largest total photocurrent density in a regenerative solar cell (Table 3). It is therefore of interest to quantify the absolute power conversion efficiency. Figure 7 shows a current-voltage curve measured using 450 nm light excitation (irradiance I450 ) 1.58 × 10-4 W cm-2). This wavelength is essentially at the Soret maximum for this sensitizer. The cell employed electrolytes that have been better optimized for porphyrins (Figure 7 legend).6 A fill factor (ff) of 0.69 and an open circuit voltage (VOC) of 0.48 V were measured. The photocurrent density at this wavelength is Jsc ) 101 µA cm-2 (Figure 5). Using these values, the quasimonochromatic conversion efficiency (η450) was calculated using eq 5.
η450 )
JSCVOCff × 100 I450
(5)
The calculated result using the values given above is 21%. This single-wavelength (Soret-maximum) power-conversion efficiency along with the measured fill factor (0.69) are noteworthy and are among the best measured for dye-sensitized solar cells. G. Comparison with Other Work. A number of recent studies of solar-cell performance have employed tetrapyrrole sensitizers.6,27-33 One relevant previous study involved TiO2 solar cells containing a derivative of ZnTPP that bears a
Tethered Porphyrin, Chlorin, and Bacteriochlorin cyanoacrylic acid tether at a β-pyrrole position (denoted ZnP-βA)6 rather than at a meso position as used here for zinc porphyrin ZnP-A (Chart 1). Solar cells containing ZnP-βA as a sensitizer gave a ff ) 0.70, which is virtually the same as that obtained here for ZnP-A (Figure 7). A global power conversion efficiency (1 sun AM 1.5 irradiation) of 5.6% was obtained for the ZnP-βA-based cell. Although this value cannot be compared directly with the monochromatic power-conversion efficiency obtained here for ZnP-A, a comparison can be made of the IPCE curves. In particular, ZnP-βA gave an IPCE at the Soret maximum of ∼90%, whereas ZnP-A gives ∼60% (Figure 5A). Additionally, for AM 1.5 illumination, the cell containing ZnP-βA gave a total photocurrent density of 13 mA cm-2, whereas our value calculated from the measured IPCE curve is 7.6 mA cm-2 (Table 3). The origin of the differences in photocurrent efficiencies for ZnP-βA versus ZnP-A is unclear. In general, optimized metalloporphyrin-based sensitizers previously studied have given fill factors in the range of 0.60 - 0.75 and global power conversion efficiencies of ∼5% (1 sun of AM 1.5 irradiation). However, some porphyrin sensitizers give lower fill factors, such as the values of ∼0.4 derived in a recent study of the effect of surface-attachment tether.27 The most optimized of the porphyrins in the latter study gave IPCE values at the Soret maximum that are comparable to those found here for ZnP-A and ZnP-I (50-60%). Turning to chlorin sensitizers, the IPCE values in the blue-violet Soret maximum and in the red Qy(0,0) maximum of ZnC-EI (Figure 5A) are comparable to those obtained for cells based on a number of naturally occurring chlorins.8 Perhaps the most efficient dyesensitized solar cell is based on a coumarin sensitizer, achieving a 5.6% global power conversion efficiency (1 sun AM 1.5 irradiation) with an even smaller 0.63 fill factor than that derived here for the cell based on porphyrin ZnP-A.34 V. Conclusions and Outlook Our initial studies on cells utilizing porphyrin ZnP-A show that this sensitizer has a combination of factors that positively impact solar-cell performance. These factors include enhanced Qx,y(0,0) absorption as compared to tetraarylporphyrins and broad optical bands that facilitate absorption of solar irradiation over a wider range. Furthermore, ZnP-A is the strongest oxidant (of the mobile charge carrier) of the tetrapyrrole compounds studied here while maintaining the ability to inject electrons into the semiconductor. In this regard, the electron density shifted to the cyano group of the tether should enhance the electronic coupling for the electron-injection process. Collectively, the findings on ZnP-A and the other four complexes investigated suggest that the state-of-the-art in solar-cell performance should be extendable using properly tuned tetrapyrrole chromophores. This tuning process may involve the incorporation of additional auxochromes and/or redox-tuning substituents such as those found to be effective in recent studies of a large family of synthetic chlorins.21,35 For example, incorporation of a cyanoacrylic acid tether in synthetic bacteriochlorins should afford favorable properties, including strong absorption in the near-infrared spectral region. Such chromophores should complement the properties of similarly tuned chlorins and porphyrins to efficiently produce photocurrent throughout the solar spectrum. Acknowledgment. This research was supported by grants from the Division of Chemical Sciences, Office of Basic Energy Sciences, Office of Energy Research, U.S. Department of Energy to JSL (DE-FG02-96ER14632), DFB (DE-FG02-05ER15660), GM (DE-FG02-96ER14662), and DH (DE-FG02-05ER15661).
J. Phys. Chem. C, Vol. 111, No. 42, 2007 15477 References and Notes (1) Gra¨tzel, M. Nature 2001, 414, 338-344. (2) Gra¨tzel, M. J. Photochem. Photobiol. C: Photochem. ReV. 2003, 4, 145-153. (3) Campbell, W. M.; Burrell, A. K.; Officer, D. L.; Jolley, K. W. Coord. Chem. ReV. 2004, 248, 1363-1379. (4) Meyer, G. J. Inorg. Chem. 2005, 44, 6852-6864. (5) Gra¨tzel, M. Mater. Res. Soc. Bull. 2005, 30, 23-27. (6) Wang, Q.; Campbell, W. M.; Bonfantani, E. E.; Jolley, K. W.; Officer, D. L.; Walsh, P. J.; Gordon, K.; Humphry-Baker, R.; Nazeeruddin, M. K.; Gra¨tzel, M. J. Phys. Chem. B 2005, 109, 15397-15409. (7) Dixon, J. M.; Taniguchi, M.; Lindsey, J. S. Photochem. Photobiol. 2005, 81, 212-213. (8) (a) Kay, A.; Gra¨tzel, M. J. Phys. Chem. 1993, 97, 6272-6277. (b) Kay, A.; Humphry-Baker, R.; Gra¨tzel, M. J. Phys. Chem. 1994, 98, 952959. (c) Amao, Y.; Yamada, Y. Langmuir 2005, 21, 3008-3012. (d) Amao, Y.; Yamada, Y. Biosensors Bioelectr. 2007, 22, 1561-1565. (9) (a) Strachan, J.-P.; O’Shea, D. F.; Balasubramanian, T.; Lindsey, J. S. J. Org. Chem. 2000, 65, 3160-3172. (b) Taniguchi, M.; Ra, D.; Mo, G.; Balasubramanian, T.; Lindsey, J. S. J. Org. Chem. 2001, 66, 73427354. (c) Taniguchi, M.; Kim, M. N.; Ra, D.; Lindsey, J. S. J. Org. Chem. 2005, 70, 275-285. (d) Laha, J. K.; Muthiah, C.; Taniguchi, M.; McDowell, B. E.; Ptaszek, M.; Lindsey, J. S. J. Org. Chem. 2006, 71, 4092-4102. (e) Laha, J. K.; Muthiah, C.; Taniguchi, M.; Lindsey, J. S. J. Org. Chem. 2006, 71, 7049-7052. (f) Ptaszek, M.; McDowell, B. E.; Taniguchi, M.; Kim, H.-J.; Lindsey, J. S. Tetrahedron 2007, 63, 3826-3839. (g) Taniguchi, M.; Ptaszek, M.; McDowell, B. E.; Lindsey, J. S. Tetrahedron 2007, 63, 38403849. (h) Taniguchi, M.; Ptaszek, M.; McDowell, B. E.; Boyle, P. D.; Lindsey, J. S. Tetrahedron 2007, 63, 3850-3863. (i) Muthiah, C.; Bhaumik, J.; Lindsey, J. S. J. Org. Chem. 2007, 5839-5842. (10) (a) Kim, H.-J.; Lindsey, J. S. J. Org. Chem. 2005, 70, 5475-5486. (b) Fan, D.; Taniguchi, M.; Lindsey, J. S. J. Org. Chem. 2007, 72, 53505357. (11) Muthiah, C.; Taniguchi, M.; Kim, H.-J.; Schmidt, I.; Kee, H. L.; Holten, D.; Bocian, D. F.; Lindsey, J. S. Photochem. Photobiol. 2007, 83, doi: 10.1111/j.1751-1097.2007.00195.x. (12) Kong, J.; White, C. A.; Krylov, A. I.; Sherrill, D.; Adamson, R. D.; Furlani, T. R.; Lee, M. S.; Lee, A. M.; Gwaltney, S. R.; Adams, T. R.; Ochsenfeld, C.; Gilbert, A. T. B.; Kedziora, G. S.; Rassolov, V. A.; Maurice, D. R.; Nair, N.; Shao, Y.; Besley, N. A.; Maslen, P. E.; Dombroski, J. P.; Daschel, H.; Zhang, W.; Korambath, P. P.; Baker, J.; Byrd, E. F. C.; Van Voorhis, T.; Oumi, M.; Hirata, S.; Hsu, C.-P.; Ishikawa, N.; Florian, J.; Warshel, A.; Johnson, B. G.; Gill, P. M. W.; Head-Gordon, M.; Pople, J. A. J. Computational Chem. 2000, 21, 1532-1548. (13) Heimer, T. A.; D’Arcangelis, S. T.; Farzad, F.; Stipkala, J. M.; Meyer, G. J. Inorg. Chem. 1996, 35, 5319-5324. (14) National Renewable Energy Laboratory, www.nrel.gov (accessed May 18, 2007). (15) Gouterman, M. In The Porphyrins; Dolphin, D., Ed.; Academic Press, Inc.: New York, 1978; Vol 3, pp 1-165. (16) Gouterman, M. J. Chem. Phys. 1959, 30, 1139-1161. (17) Gouterman, M. J. Mol. Spectrosc. 1961, 6, 138-163. (18) Loewe, R. S.; Lammi, R. K.; Diers, J. R.; Kirmaier, C.; Bocian, D. F.; Holten, D.; Lindsey, J. S. J. Mater. Chem. 2002, 12, 1530-1552. (19) Roth, K. M.; Yasseri, A. A.; Liu, Z.; Dabke, R. B.; Malinovskii, V.; Schweikart, K.-H., Yu, L.; Tiznado, H.; Zaera, F.; Lindsey, J. S.; Kuhr, W. G.; Bocian, D. F. J. Am. Chem. Soc. 2003, 125, 505-517 and references therein. (20) Yang, S. I.; Seth, J.; Strachan, J.-P.; Gentemann, S.; Kim, D.; Holten, D.; Lindsey, J. S.; Bocian, D. F. J. Porphyrins Phthalocyanines 1999, 3, 117-147. (21) Kee, H. L.; Kirmaier, C.; Tang, Q.; Diers, J. R.; Muthiah, C.; Taniguchi, M.; Laha, J. K.; Ptaszek, M.; Lindsey, J. S.; Bocian, D. F.; Holten, D. Photochem. Photobiol. 2007, 83, 1125-1143. (22) Hasselman, G. M.; Watson, D. F.; Stromberg, J. R.; Bocian, D. F.; Holten, D.; Lindsey, J. S.; Meyer, G. J. J. Phys. Chem. B 2006, 110, 2543025440. (23) (a) Gerischer, H. Photochem. Photobiol. 1972, 16, 243-260. (b) Gerischer, H. Pure Appl. Chem. 1980, 52, 2649-2667. (24) (a) Clark, W. D. K.; Sutin, N. J. Am. Chem. Soc. 1977, 99, 46764682. (b) Sonntag, L. P.; Spitler, M. T. J. Phys. Chem. 1985, 89, 14531457. (25) (a) Tachibana, Y.; Moser, J. E.; Gra¨tzel, M.; Klug, D. R.; Durrant, J. R. J. Phys. Chem. 1996, 100, 20056-20062. (b) Benko¨, G.; Kallioinen, J.; Korppi-Tommola, J. E. I.; Yartsev, A. P.; Sundstro¨m, V. J. Am. Chem. Soc. 2002, 124, 489-493. (c) Kuciauskas, D.; Monat, J. E.; Villahermosa,
15478 J. Phys. Chem. C, Vol. 111, No. 42, 2007 R.; Gray, H. B.; Lewis, N. S.; McCusker, J. K. J. Phys. Chem. B 2002, 106, 9347-9358. (d) Furube, A.; Katoh, R.; Hara, K.; Sato, T.; Murata, S.; Arakawa, H.; Tachiya, M. J. Phys. Chem. B 2005, 109, 1640616414. (26) (a) Burlov-Vasiljev, K. A.; Matvejev, Y. B.; Vasiljeva, I. E. Sol. Phys. 1998, 177, 25-40. (b) Gueymard, C. A. Sol. Energy 2001, 71, 325346. (27) Rochford, J.; Chu, D.; Hagfeldt, A.; Galoppini, E. J. Am. Chem. Soc. 2007, 129, 4655-4665. (28) Wang, X.-F.; Matsuda, A.; Koyama, Y.; Nagae, H.; Sasaki, S.-I.; Tamiaki, H.; Wada, Y. Chem. Phys. Lett. 2006, 423, 470-475. (29) Tanaka, M.; Hayashi, S.; Eu, S.; Umeyama, T.; Matano, Y.; Imahori, H. Chem. Commun. 2007, 2069-2071. (30) Luo, L.; Lo, C.-F.; Lin, C.-Y.; Chang, I.-J.; Diau, E. W.-G. J. Phys. Chem. B, 2006, 110, 410-419.
Stromberg et al. (31) He, J.; Benko¨, G.; Korodi, F.; Korodi, F.; Polı´vka, T.; Lomoth, R.; Åkermark, B.; Sun, L.; Hagfeldt. A.; Sundstro¨m, V. J. Am. Chem. Soc. 2002, 124, 4922-4932. (32) Reddy, P. Y.; Giribabu, L.; Lyness, C.; Snaith, H. J.; Vijaykumar, C.; F. Chandrasekharam, M.; Lakshimikantam, M.; Yum, J.-H.; Kalyanasundaram, K.; Gra¨tzel, M.; Nazeeruddin, M. K. Angew. Chem., Int. Ed. 2007, 46, 373-376. (33) Imahori, H.; Hayashi, S.; Umeyama, T.; Eu, S.; Oguro, A.; Kang, S.; Matano, Y.; Shishido, T.; Ngamsinlapasathian, S.; Yoshikawa, S. Langmuir 2006, 22, 11405-11411. (34) Hara, K.; Sayama, K.; Ohga, Y.; Shinpo, A.; Suga, S.; Arakawa, H. Chem. Commun. 2001, 569-570. (35) Kee, H. L.; Kirmaier, C.; Tang, Q.; Diers, J. R.; Muthiah, C.; Taniguchi, M.; Laha, J. K.; Ptaszek, M.; Lindsey, J. S.; Bocian, D. F.; Holten, D. Photochem. Photobiol. 2007, 83, 1110-1124.