Catherine Bonniera, Devin D. Machina, Omar K. Abdia, Kiyoshi C. D. Robsonb and Bryan D. Koivisto*a
aDepartment of Chemistry and Biology and the Centre for Urban Energy (CUE) and Ryerson University, 350 Victoria St. Toronto, Ontario, Canada M5B 2K3. E-mail: bryan.koivisto@ryerson.ca; Fax: +1-416-979-5044; Tel: +1-416-979-5000 ext 4625
bDepartment of Chemistry, University of Calgary, Calgary, AB, Canada
First published on 23rd August 2013
A family of seven organic triphenylamine-based dyes suitable for dye-sensitized solar cell (DSSC) applications is reported. The donor portion of these dyes has been systematically modified using polymerisable thienyl subunits. The physicochemical properties and device performance are discussed with device efficiencies ranging from 5.51 to 6.65%.
Concomitantly, organic-based photosensitizers have become increasingly popular for light-harvesting applications owing to an improved understanding of the requisite structure–property relationships.9 For example, all organic dyes for DSSC applications are comprised of a redox-active donor/chromophore (D) that is coupled through a π-conjugated spacer to an acceptor (A) capable of anchoring to TiO2.9,10 Furthermore, there have been an increasing number of examples that suggest that organic dyes can be systematically tuned11,12 to have improved interactions with I−/I3−, and in many cases can outperform their ruthenium-based counterparts when using novel electrolytes.4,7,8
While a great deal of synthetic effort has been put into developing unique π-spacers capable of increasing charge separation and bathochromically shifting the absorption profile,13,14 the modification of the redox-active donor with electron-rich π-systems has been less studied.15–17 Triphenylamines (TPAs) are the most commonly employed donors in organic motifs because of their intense absorption and rich redox behaviour.18
Our motivation for this work began with a report by Liu and Ramakrishna, where they found that an indoline-based dye could effectively sensitize TiO2, and outperform Z907, when using polyEDOT as an HTM (in situ polymerization of EDOT using photochemical or electrochemical methods).7 In that report they observed a titania thickness dependence on device performance, and suggested that the nature of the sensitizer could affect the effective penetration depth of the polymer HTM.
We wanted to systematically study this hypothesis and propose that if a donor, like the TPA in the previously reported 1,19 is equipped with polymerizable subunits, the resulting family of dyes may be able to integrate with the polymer film leading to increased pore filling, decreased dye–HTM interfacial capacitance, minimized recombination and improved long term stability. To this end, we have prepared a family of dyes based on 1 where the TPA donor is functionalized with pendant thiophene arms that are conjugated to the donor. Before exploring these dyes in HTM-based devices, we have examined the effect of donor modification on the physicochemical properties and device performance with the traditional I−/I3− electrolyte. The results of this benchmark study are reported herein.
![]() | ||
Fig. 1 A family of TPA dyes where the donor is substituted with various thienyl-subunits. |
![]() | ||
Scheme 1 Synthesis of dye families 4a, 4b,226b, 8a and 8b. Reaction conditions: (i) A (1.15 or 2.3 eq.), K2CO3 (5 or 10 eq.), Pd(PPh3)4 (10 or 20 mol%), THF–H2O 9![]() ![]() ![]() ![]() ![]() ![]() |
![]() | ||
Scheme 2 Synthesis of dye family 12. Reaction conditions: (i) C (1.x eq.), K2CO3 (5 eq.), Pd(PPh3)4 (10 mol%), THF–H2O 9![]() ![]() ![]() ![]() |
Compound | E1/2 (V vs. NHE)a | UV-Visb |
---|---|---|
TPA•+/TPA | λmax/nm (ε × 103/M−1 cm−1) | |
a Data collected using 0.1 M NBu4PF6 DCM solutions at 100 mV s−1 and referenced to a [Fc]/[Fc]+ internal standard followed by conversion to NHE; [Fc]/[Fc+] = +765 mV vs. NHE in DCM.b Data collected in DCM solutions. | ||
1 | 1.27 | 479 (22.1), 346 (12.3) |
4a | 1.24 | 487 (37.4), 350 (30.6) |
4b | 1.21 | 492 (37.3), 360 (45.6) |
6b | 1.25 | 489 (39.3), 369 (54.2) |
8a | 1.26 | 487 (32.3), 329 (25.6) |
8b | 1.23 | 492 (33.9), 340 (37.6) |
12a | 1.29 | 486 (36.8), 371 (38.8) |
12b | 1.32 | 491 (33.9), 389 (53.8) |
For all dyes, with the exception of 1, sweeps to higher potential (∼1.3 V), beyond the reversible TPA oxidation, resulted in a pseudo reversible wave that grows in intensity with multiple scans. This observation suggests thiophene polymer formation at the electrode surface and is consistent with the formation of thiophene–TPA linkages.23 An examination of the working electrode surface (ESI; Fig. S1†) after multiple scans shows the formation of an insoluble red polymeric film, which further supports the notion that these dyes could be successfully integrated with an HTM during an in situ polymerization.
A representative set of UV-Vis spectra recorded in DCM for 4b, 6b, 8b, and 12b are presented in Fig. 2 and tabulated in Table 1 (UV-Vis profiles of dyes 1, 4a, 8a, and 12a can be found in Fig. S2†). The UV-Vis spectra of all dyes show two significant absorptions: one at high energy and one at low energy centred at ca. 500 nm. When compared to the benchmark dye 1, it was satisfyingly observed that modifying the TPA donor increased the extinction coefficients and moderately redshifted the low energy absorption envelope. The low energy absorption has been assigned as the HOMO–LUMO transition and DFT calculated (B3LYP/6-31G) frontier orbitals are presented in Fig. 3. However, the most significant differences due to modification of the TPA are observed in the high energy transition which has been assigned as the HOMO–LUMO + 1 transition.
![]() | ||
Fig. 2 UV-Vis spectra in DCM for the di-substituted TPA dyes. |
![]() | ||
Fig. 3 Frontier molecular orbitals as predicted by DFT (B3LYP/6-31G) for di-substituted dyes 4b, 6b, 8b, 12b. |
The HOMO of these dyes is centred on the TPA, but delocalized over the entire molecule (Fig. 3). As a result of an increased optical cross-section, larger extinction coefficients are observed. The LUMO is appropriately located on the cyanoacetic anchor in all cases. The LUMO + 1 is primarily localized on the polymerizable subunits. When the conjugation lengths between the TPA and thienyl-subunit are increased, a significant bathochromic shift is observed in the higher energy absorption. A similar trend is observed for the mono-substituted TPA derivatives (4a, 8a, and 12a); however, this high energy band is less intense in all mono-substituted derivatives.
With a detailed understanding of the physicochemical properties in solution, DSSC device performance was examined for our new family of dyes. The relevant parameters were extracted from the measured current–voltage traces under standard AM 1.5 sunlight and are detailed in Table 2. We were satisfied to observe that each of our modified dyes had a higher device efficiency than the benchmark dye 1. The highest performing dyes were 4b and 8b, but surprisingly, 12b had the lowest performance of this new family.
Compound | Jsc (mA cm−2) | Voc (V) | FF | η (%) |
---|---|---|---|---|
a Device fabrication and data collection procedures can be found in the ESI. | ||||
1 | 10.93 | 0.68 | 0.70 | 5.15 |
4a | 11.65 | 0.70 | 0.72 | 5.84 |
4b | 12.59 | 0.71 | 0.73 | 6.48 |
6b | 12.12 | 0.71 | 0.72 | 6.23 |
8a | 11.90 | 0.69 | 0.71 | 5.89 |
8b | 12.86 | 0.71 | 0.72 | 6.56 |
12a | 10.80 | 0.70 | 0.74 | 5.59 |
12b | 11.33 | 0.68 | 0.72 | 5.51 |
Several structure–property relationships can be observed in this series. The first is that disubstituted TPAs outperform mono-substituted derivatives owing to increased light harvesting (see Table 1). Furthermore, within the error of the measurement, there is no significant difference between 2- and 3-substituted thiophenes. 12a and 12b, with the olefin spacer, were designed to offer conformational flexibility for future in situ polymerizations; therefore, it was a little disappointing to see this dye with the lowest performance. While differing dye–electrolyte interactions cannot be ruled out, all dyes were studied using the same Z1137 I−/I3− electrolyte and it is reasonable to assume that because these dyes have similar hydrophobicity, owing to their similar polyaromatic structure, regeneration mechanisms would not be rate limiting. Furthermore, assuming that we are operating within a normal Markus region, regeneration for 12b should be fastest because it has the most stabilized HOMO. Heavy atom quenching (by the bromine atoms) of the excited state can also be ruled out because the same poor performance was not observed in 6b.
The incident photon-to-electricity conversion efficiency (IPCE) spectrum for disubstituted dyes is presented in Fig. 4. Despite having a similar profile to 4b and 8b, the current response for 12b is ∼10% lower for most of the visible spectrum.
![]() | ||
Fig. 4 Photocurrent action spectrum obtained for di-substituted dyes attached to TiO2 films using the Z1137 electrolyte. |
When considering the hydrodynamic volume of 12b, the most rational explanation for the low Jsc and Voc (and ultimately performance) is a reduced dye loading on the TiO2 surface. Unfortunately, the preprogrammed conformational flexibility meant to facilitate integration with the HTM would also prohibit dye loading, once again reaffirming the classical organic reasoning that both electronics and sterics are important considerations in molecular design.
Footnote |
† Electronic supplementary information (ESI) available: Experimental data and characterization is available. See DOI: 10.1039/c3ob41288a |
This journal is © The Royal Society of Chemistry 2013 |