2.1.

Synthesis Details

As the simplest dye in this series, P1 has a –COOH group connected directly to the pentacene backbone. We were unable to prepare this compound by straightforward halide/CO₂ carboxylation of a 2-halopentacene in the presence of BuLi or Grignard reagent. Instead, a catalytic formylation³¹ of iodide 1 (Ref. 32) afforded aldehyde 2 in reasonable yield, which was followed by Ag₂O oxidization to acid P1. Dye P2 was obtained in two steps; a Heck reaction between 1 and t-butyl acrylate yielded ester 3, which was hydrolyzed to acrylic acid P2. Cyanoacrylic acid P3 was obtained by Knoevenagel condensation between aldehyde 2 and cyanoacetic acid. P7 and P8 were prepared in the same manner. Alkyne-linked dye P4 was prepared by hydrolysis of methyl ester 4, which was accessible by Sonogashira coupling between 1 and methyl propiolate. Dye P5 was also made by Sonogashira coupling using 4-ethynyl benzoic acid. Thiophene aldehyde 5 was obtained by Suzuki coupling of a thiophene carboxaldehyde in moderate yield and then oxidized to provide dye P6.

2.2.

Device Fabrication

Vertically oriented TiO₂ nanowire arrays were synthesized by a hydrothermal method on fluorine doped tin oxide (FTO)-coated glass (TEC-8, 8 ohm/square) substrates. The substrates were initially cleaned by sonication in acetone, 2-propanol, and methanol, rinsed with deionized water, and dried in a nitrogen stream. These substrates were dipped in 0.1 M TiCl₄ solution for 8 h and then heated at 500°C for 1 h to obtain a layer of TiO₂ ∼ 20 nm thick. To grow nanowire arrays, these treated FTO substrates were loaded into a sealed Teflon^® reactor (23 ml volume) containing 10 ml of toluene, 1 ml of titanium tetrachloride (1 M in toluene), 1 ml of tetrabutyl titanate, and 1 ml of hydrochloric acid (37%). We conducted the growth of single-crystalline nanowires on transparent conducting oxide (TCO) solid substrate by a nonpolar solvent/hydrophilic solid interfacial reaction mechanism.³³ (A very different alkali hydrothermal growth process has also been used to synthesize single-crystalline rutile nanowire arrays.)³⁴ Because it is desirable to have a high surface area and close packing³⁵ of the nanowires for liquid solar cell applications, a high concentration of Ti⁴⁺ precursor was used. The reaction took place at 160°C for 10 h. After the reaction was completed, the nanowire-covered FTO substrates were carefully washed with ethanol and annealed at 450°C for 30 min in air to remove the surface-adsorbed organic species. The nanowire arrays were ∼3.5 μm in length, 20 nm in width, and 5–10 nm in interwire spacing. All samples were further treated with oxygen plasma for 5 min at an RF power of 80 W to clean and hydroxylate the surface for better dye adhesion. The DSSCs were prepared using the following two procedures:

Method A: Nanowire array samples were coated with dye by immersion for 30–60 min in a 1-mM tetrahydrofuran (THF) solution of the respective pentacene dye (P1–P6). Liquid-junction solar cells were prepared by infiltrating the dye-coated TiO₂ electrode with redox electrolyte containing lithium iodide (0.1 M), iodine (0.01 M), tri-n-butyl phosphate (0.4 M), butyl methyl imidazodium iodie (0.6 M), and guanidinium thiocyanate (0.1 M) in a mixture of acetonitrile and valeronitrile (v/v = 15/1). Conductive glass slides, sputter coated with 100 nm of Pt were used as the counter electrode. Electrode spacing between the nanowire and counter electrodes was ensured by using a 25-μm-thick SX-1170 spacer (Solaronix Inc., Aubonne, Switzerland). Photocurrent density and photovoltage were measured with active sample areas of 0.2 cm² using AM-1.5 simulated sunlight produced by a 500-W Oriel Solar Simulator, calibrated with a national renewable energy laboratory (NREL)-certified silicon solar cell.

Method B: Similar to method A except that nanowire array samples were coated with dye by immersion for 5 h in a 0.4-mg/mL toluene solution of the respective pentacene dye (P3, P7, and P8) at 70°C.

2.3.

Computational Details

Density functional theory (DFT) calculations were performed in the Gaussian03 program package.³⁶ The geometries of P1–P6 were optimized at the B3LYP level using a 3–21G* basis set in vacum using without any symmetry constraints. Electronic structure calculations were performed using a 6–31G* basis set. Molecular orbitals were visualized by GaussView 3.0 software.

3.1.

Solar Cell I–V Characterization

P1–P6 did act as effective sensitizers in DSSCs. Although their sensitizing capabilities seem to be limited, structure-property relationships are obvious. The linker that couples the pentacene chromophore and the anchor group plays a critical role in determining the sensitizing effect. The solubility of these dyes seems to be another important factor.

3.1.1.

Linker effect

Initially, the solar cells were fabricated using method A. The subtle change in linker structure of sensitizers P1–P6 dramatically affects their performance in the devices (Fig. 3). Table 1 summarizes the key photovoltaic parameters for these materials. P1, which has no linker, shows the highest short-circuit current density (J_SC = 1.49 mA/cm²) as well as the highest open-circuit voltage (V_OC = 0.6 V); however, its fill factor (FF) is only 0.38, the lowest in the group. In the case of P2 where a simple –C=C– linker is introduced, comparable PCE (η = 0.30%) is seen, although J_SC drops and FF rises. It is surprising to see a poor result from P3, which has an extra cyano group on the vinyl linker. Incorporating cyanoacrylic acid is a well-known strategy in organic sensitizer development because the strong electron-withdrawing nature of –C≡N has been demonstrated to facilitate the electron-injection process.³⁷ The unexpected low J_SC leads to overall poor performance by P3, indicating that a characteristic electronic property of the pentacene chromophore may mitigate the impact of the nitrile substituent. Switching to an alkyne linker (P4) results in the lowest PCE, arising from both poor J_SC and V_OC. This poor performance is also seen in alkyne-containing P5. Direct aryl attachment (P6), however, exhibits sensitizing ability similar to that observed in vinyl-linked P2, indicating that heteroacenes could serve as efficient electron donors to titania.

Table 1

Photovoltaic parameters of DSSCs based on P1–P6.

Methods	Dye	VOC (V)	JSC (mA/cm2)	ff	η (%)
A	P1	0.60	1.49	0.38	0.34
	P2	0.60	1.05	0.47	0.30
	P3	0.58	0.45	0.48	0.13
	P4	0.51	0.36	0.52	0.10
	P5	0.55	0.42	0.59	0.14
	P6	0.57	0.95	0.53	0.29
B	P3	0.59	1.02	0.52	0.31
	P7	0.61	1.69	0.49	0.50
	P8	0.58	1.74	0.47	0.47
	JK-2	0.77	7.79	0.56	3.34
	N719	0.76	6.57	0.66	3.30

Fig. 3

J-V characteristics of DSSCs based on P1–P6.

3.2.

Absorption and Electrochemical Properties

As mentioned above, poor light-harvesting abilities and/or inefficient electron injection also can cause the relatively low performance of these pentacene dyes. The absorption spectra of P1–P6 solutions in THF [Fig. 4] possess an intense S₀→S₁ band (ɛ is ∼10⁴ M⁻¹cm⁻¹) with fine vibronic progression in the near-IR region (between 550 and 700 nm). Figure 4 illustrates the absorption properties of these dyes chemisorbed onto the 3.5-μm nanocrystalline TiO₂ thin film.

Fig. 4

(a) Normalized absorption spectra of P1–P6 in THF. (b) Absorption spectra of P1–P6 on TiO₂ films (systematically shift along the y-axis for clarity). (c) IPCE spectra of DSSCs sensitized with P1 and P3.

There is no clear shift in the thin-film spectrum in comparison to the corresponding solution spectrum, indicating no obvious aggregation of dye molecules. For dyes whose photovoltaic performance in DSSCs is negatively impacted by aggregation, coadsorption of deoxycholic acid is often used to disrupt the aggregation and thus improve performance.^{38, 39} However, TiO₂ nanowire array solar cells coated with a mixed monolayer of deoxycholic acid and P1–P6 were not found to exhibit any improved performance over the plain (P1–P6)–sensitized solar cells. This, together with the absence of clear shifts in the absorption data, rules out aggregation as the reason for low photocurrents. The broad absorptions extending well beyond 650 nm should yield good light harvesting for these materials.

The orbital energy levels were determined by a combination of differential pulse voltammetry (DPV) measurements and optical spectrum data (Table 2). The downhill energy offset by the LUMOs of P1–P6 (>–3.50 eV versus vacuum), relative to the conduction band edge (–4.00 eV versus vacuum) of TiO₂, ensures sufficient thermodynamic driving force for electron injection. The uphill energy offset by the HOMOs (<–5.00 eV versus vacuum) with respect to that of iodide (–4.60 eV versus vacuum) supplies a desired negative Gibbs energy change for dye regeneration.⁴⁰ Computational results lead to the same conclusion on the favorable energy alignment of these pentacene sensitizers.

Table 2

Experimental and computational energy levels of P1–P6.

However, the incident photon to current efficiency (IPCE) of these sensitizers reach only 5% in the near-IR absorption window of pentacene chromophore. Examples using P1 and P3 are depicted in Fig. 4. Although the absorption properties and orbital energies of the pentacene sensitizers are acceptable, our results point to certain electronic properties of these pentacene dyes accounting for inefficient electron injection, leading to poor IPCEs and overall photocurrents.

In large conjugated systems, depending on their conformational and electronic structures, the electron-density distribution in the frontier orbitals can have a huge impact on molecular orbital mixing with the t2g/eg levels constituting the TiO₂ conduction band, as well as the geminate charge recombination between oxidized dye molecules and photoinjected electrons in TiO₂.⁴¹ Therefore, the relevant photophysical process will determine the output photocurrent.

As shown in Fig. 5, an ideal example of the electron-density distribution in the HOMO and LUMO of a sensitizer is shown by the high-performance organic dye JK-2 (calculation based on reported geometry).¹⁸ The excited-state orbital (LUMO) populates the anchor, facilitating significant mixing with the titania conduction band and, therefore, efficient electron injection. On the contrary, P3 clearly shows a less favorable pattern. It appears that the HOMO→LUMO excitation does not significantly redistribute the electron population from the pentacene core to the carboxylic acid anchor. Although the anchoring group gains a little electron population in the LUMO, most of the orbital still extends over the entire molecule. As a result, there may be less orbital mixing potentially resulting in reduced photoexcited electron injection from sensitizer into TiO₂. The pentacene core seems to hold delocalized electrons too strongly, even when a powerful electron-withdrawing group (−C≡N) is introduced. Similar situations were observed for all the other pentacene dyes reported here. Therefore, future work involving further molecular modification to provide a system with a better distinguished ground/excited states is necessary for a highly efficient pentacene-based sensitizer. The surprisingly high efficiency of the thiophene-substituted pentacene (P6) hints that heteroacenes, with their larger, more polarizable constituents, may be more amenable for use as dyes for DSSCs.

Fig. 5

HOMO/LUMO isodensity surface plots of JK-2 and P3.