Improved conversion efficiency of 10 for solid-state dye-sensitized solar cells utilizing P-type semiconducting CuI and multi-dye consisting of novel porphyrin dimer and organic dyes

To realize highly efficient solid-state dye-sensitized solar cells (SDSCs), the absorption range of the dye should be extended to the near-IR range to increase short-circuit current density (Jsc); a high Jsc in turn requires a highly conductive p-type semiconductor. A newly developed dye (DIPDAB2) with a porphyrin dimer structure provided higher absorption coefficients than the conventional dye with a similar framework (DTBC) in the long wavelength range of 700–800 nm, leading to higher incident photon-to-current conversion efficiencies. The dip in the absorption spectrum of DIPDAB2 located at 500– 700 nm between the Soret band and Q band was filled by combining with two kinds of organic dyes (D131 and D358). The multi-dye consisting of the three dyes realized a high Jsc over 20 mA cm . The use of copper iodide that has a higher conductivity than p-type organic semiconductors and copper complexes secured a high filling factor. Introduction of Li ions into the TiO2 photoelectrodes improved the open-circuit voltage (Voc) along with a slight increase in Jsc. Light soaking also contributed to a higher Voc. The conversion efficiency of the present SDSC was as high as 10%.


Introduction
Dye-sensitized solar cells (DSCs) have various advantages of color choice and transparency, as well as low costs and low energy consumption in production, 1,2 compared with inorganic semiconductor-based solar cells. Using these unique characteristics, we have developed new application products such as light-emitting signboards, self-emitting guidance lights, etc., in which DSCs are installed as stand-alone power sources. 3 Smallsize DSCs provide a high conversion efficiency (h) of 14.7%. 4 However, h and durability are in a trade-off relationship, as summarized in Fig. S1 †. [5][6][7] DSCs using a liquid electrolyte of a highly volatile solvent such as 3-methoxypropionitrile achieve a high h but suffer from low durability caused by solvent evaporation. 5 The use of g-butyrolactone extends the lifetime up to 3-5 years in an outdoor environment, while it appreciably decreases h. 6 DSCs using ionic liquids are further durable; the lifetime estimated by an accelerated aging test is as long as 15 years. 7 However, h is further lower owing to low diffusivity of the redox ions in viscous ionic liquids. The liquid electrolytes raise another issue of DSC modules that wide sealants separating neighbouring cells are required for high durability, resulting in narrowing of active areas.
Replacement of the liquid electrolytes with solid-state p-type semiconductors can realize both high efficiency and high durability, 8,9 by solving the evaporation issue with securing a high hole conductivity. Therefore, studies on solid-state DSCs (SDSCs) have been very active. Table 1 summarizes electric properties of candidate p-type materials and photovoltaic properties of SDSCs. Among them, copper iodide (CuI) is highly conductive and stable. 10 The highest h of SDSCs using CuI is 7.4% coupled with the N3 dye. 11 Although CsSnI 3 perovskite is also highly conductive and achieves a higher h of 8.5% with the N719 dye, 12 it is an unstable material; easily oxidized under ambient air condition. On the other hand, conductivities of ptype organic semiconductors such as 2,2 0 ,7,7 0 -tetrakis-(N,N-dip-methoxyphenyl-amine)-9,9 0 -spirobiuorene (spiro-OMeTAD) and the copper complexes are much lower than those of the inorganic materials. 9,[13][14][15] In the present study, we have focused on four items to improve h of SDSCs. The priority for this purpose is to widen the absorption range of the dye to increase the short-circuit current density (J sc ). Therefore, the rst item is a new dye (DIPDAB2) with a porphyrin dimer structure to achieve higher absorption coefficients in the near-IR range. A conventional indoline dye (D149) has a narrow absorption band ranging up to 780 nm, 16 resulting in a low J sc of 13 mA cm À2 . We have developed DIP-DAB2 based on a donor-acceptor type porphyrin dimer dye: Nfused carbazole-zinc porphyrin-free-base porphyrin triad (DTBC) that sensitizes up to 900 nm. 17 The molecular structure of DIPDAB2 is compared with that of DTBC in Fig. 1. There are renements in DIPDAB2. A stronger electron donor unit of bis(4-hexylphenyl)amine moiety and a p conjugated system extended by a triple-bond-connected acceptor of benzoic acid red-shi the Q band. DTBC and other dyes of the porphyrin dimer structures previously developed: YDD6 (ref. 18) and LDD1 (ref. 19) are equipped with common substituents on the two porphyrin rings as shown in Fig. 1. The unique point of DIP-DAB2 is that two different substituents are attached to the two porphyrin rings, which may inhibit aggregation of the dyes on the TiO 2 electrodes. The second is the use of a multi-dye consisting of DIPDAB2 and two kinds of organic dyes: a yellow indoline dye (D131) 20 and a red indoline rhodanine dye (D358), 21 to cover a wide wavelength range from the visible to the near-IR. It has been demonstrated that co-sensitization can improve the performance of DSCs. 18,19 We have selected cosensitizers of D131 and D358 dyes to match DIPDAB2.
A LiI treatment aer the dye-absorption on the TiO 2 electrodes is employed to increase the open-circuit voltage (V oc ), which is the third item. The last is light soaking that also increases V oc . The effects of the LiI treatment and light soaking on the present SDSCs are contrasting to those on liquid-type DSCs of improvements in J sc with lowering in V oc .
Another important point is the selection of the p-type semiconductor. CuI and copper complexes are stable materials and hence the candidates, as stated above. Our strategy of the improvement in h is to increase J sc , which requires high conductivity of the p-type material, because detrimental impacts of the inner resistance are more notable at a higher J sc . Copper complexes show high efficiency, [13][14][15] however, the conductivity is much lower than that of CuI. Therefore, we selected CuI for the p-type semiconductor. We have previously overcome the difficulties in lling the porous TiO 2 electrodes with CuI by using TiO 2 particles with a larger size than those for DSCs using liquid electrolytes, as shown in Fig. S2. † 22

Synthesis of DIPDAB2
We synthesized DIPDAB2 dye by the scheme illustrated in

Fabrication of DSCs with a liquid electrolyte
To compare the sensitization performance of DIPDAB2 with that of DTBC, DSCs using a liquid electrolyte were fabricated. Porous TiO 2 electrodes consisting of nanoparticles of ca. 20 nm in diameter were screen-printed on glass substrates covered with transparent conductive oxide (TCO) layers, followed by a surface treatment using a TiCl 4 aqueous solution. Then, the TiO 2 electrodes of 10 mm Â 7 mm in size were immersed in a 0.2 mM dye (DTBC or DIPDAB2) and 2 mM chenodeoxycholic acid (CDCA) acetonitrile/tert-butyl alcohol (1 : 1) solution for 12 hours. The counter electrodes were prepared by chemical deposition of platinum onto the TCO glasses using a 0.05 M hexachloroplatinic acid solution at 400 C. The dye-adsorbed TiO 2 electrodes and platinum counter electrodes were faced to each other with polyolen-based sealants. An liquid electrolyte containing I À /I 3 À redox couples using 0.6 M DMPII and 25 mM I 2 , 0.1 M LiI and 0.5 M 4-tert-butylpyridine in acetonitrile was injected into the cells.

Fabrication of SDSCs with p-type semiconducting CuI
Double-layered TiO 2 electrodes were prepared by screen printing on the TCO glasses. The rst layers of around 5 mm in thickness were deposited using TiO 2 particles with a size distribution ranging from 35 nm to 400 nm. The average pore size was 160 nm. Larger TiO 2 particles of 400 nm in size were used for the 3 mm-thick second layers. The TiCl 4 post treatments were repeated three times. The TiO 2 electrodes of 10 mm Â 7 mm in size were immersed in a 0.2 mM DIPDAB2 dye and 2 mM CDCA acetonitrile/tert-butyl alcohol (1 : 1) solution for 1 hour, and then in 0.1 mM D131 and 0.3 mM D358 mixed ethanol solution for 16 hours. Aer the multi-dye adsorption, the TiO 2 electrodes were immersed into 0.1-0.5 M LiI acetonitrile solutions. Solid-state CuI was injected into the dyeadsorbed porous TiO 2 electrodes by dropping a CuI solution (CuI and 1-ethyl-3-methylimidazoliumthiocyanate in acetonitrile) followed by evaporation of the solvent as shown in S2. † [22][23][24] Finally, the TiO 2 -dye-CuI electrodes were coupled with platinum foils as the counter electrodes to complete the SDSCs. SDSCs using D149 were also fabricated for comparison. LiI concentration distribution in the SDSCs was analysed by time-of-ight secondary ion mass spectrometry (TOF-SIMS).

Evaluation of the photovoltaic performance
The current-voltage (I-V) characteristics of the DSCs were determined by measuring the photocurrent using an I-V tester (IV-9701, WACOM) under the simulated solar insolation of AM1.5G produced using a 1 kW xenon lamp and a 400 W halogen lamp (WXS-155S-L2, WACOM). The incident photon-tocurrent conversion efficiency (IPCE) was measured in a wavelength range of 300-900 nm using an equipment (CEP-2000, BUNKOU-KEIKI) specially designed for DSCs consisting of a xenon lamp, a monochromator, a digital lock-in amplier, and a Keithley 2400 source meter.

Results and discussion
As mentioned above, at rst we have designed and synthesized DIPDAB2 with a porphyrin dimer structure to achieve higher absorption coefficients in the near-IR range than those of conventional DTBC. 17 UV-vis absorption spectra of DIPDAB2 and DTBC adsorbed on porous TiO 2 lms shown in Fig. 2 conrm the red-shied Q band with higher absorbance of  Table S1. † The absorption coefficient of DIPDAB2 at the Q band was much higher than that of YDD6 and comparable with that of LDD1. Photovoltaic performance of the liquid DSCs are shown in Fig. 3 and Table S1. † The IPCE and conversion efficiency for DIPDAB2 were much higher than those for DTBC, reecting the notable enhancement in the absorbance. Similar trends were observed for YDD6 (ref. 18) and LDD1 (ref. 19) according to the absorption coefficients.
The second item is the use of a multi-dye. Although DIPDAB2 exhibits high absorbance in the near-IR range as described above, there is a dip between the Soret band centered at 500 nm and the Q band located at around 700 nm. To ll the gap and secure highly efficient sensitization in the whole of the visiblenear-IR range, a multi-dye consisting of DIPDAB2 and two kinds of organic dyes: D131 and D358 was used. The molecular structures of D131 and D358 are displayed in Fig. S9. † Fig. 4 compares the IPCE spectrum of the SDSC using the multi-dye (DIPDAB2 + D131 + D358) with those for D149 and DIPDAB2. D131 and D358 well complemented the dip between the Soret band and Q band originating from the porphyrin dimer in DIPDAB2. As a result, the IPCE spectrum for the multi-dye covered a notably wider range than that for D149, leading to a dramatic improvement in J sc to over 20 mA cm À2 estimated from the IPCE spectrum. We consider that the multi-dye inhibits the aggregation of the single DIPDAB2 dye. 18,19 As the result, the IPCE for the multi-dye in the absorption range of the Q band was much higher than the single DIPDAB2 dye.
The third item is the post treatment using LiI. The dyeadsorbed TiO 2 electrodes were immersed in LiI acetonitrile solutions. Fig. 5 shows the I-V curves of the SDSCs with and  It is well known that introduction of LiI into an electrolyte consisting of I À /I 3 À redox couples used for DSCs increases J sc while it decreases V oc . 25 Li ions adsorbed on the TiO 2 electrodes lowers the conduction-band minimum (CBM) of the TiO 2 , leading to a higher efficiency of electron injection from the photo-excited dyes to the TiO 2 and consequently a higher J sc . On the other hand, V oc lowers as it is determined from the energy difference between the CBM of the TiO 2 and the redox level. The contrasting result of increasing V oc for the present SDSCs suggests a different mechanism of the LiI treatment effect, like reduction of carrier recombination at the interfaces between the TiO 2 and CuI. The fourth item is light soaking. During the light soaking at 1 sun, V oc gradually increased with only a slight decrease in J sc , as shown in Fig. 6a. Aer a half hour, V oc reached 0.65 V (an improvement by 0.07 V) with securing a high J sc of 22 mA cm À2 for the best-performance cell. The I-V curves of the SDSCs using the multi-dye and D149 are compared in Fig. 6b. Reecting the wider sensitivity range, J sc for the multi-dye was much higher than that for D149. In addition, there was no degradation of the lling factor (FF) in spite of the higher J sc , owing to the high conductivity of CuI. As a result, h for the multi-dye was signicantly improved to be 10.1% compared with that for D149, and substantially exceeded the previously reported highest value of 7.4% using CuI and the N3 dye. 11 The reproducibility was fairly good as summarized in Table S2. † It is worthy to note that the effect of the light soaking on the SDSCs were also contrasting to that of higher J sc and lower V oc on the DSCs using liquid electrolytes, although the present mechanism is not clear.
The photovoltaic performance of the SDSCs employing the four items is compared with previously reported ones in Table 1. The present J sc of 22 mA cm À2 is extremely higher than those of the other SDSCs, with no remarkable degradation of FF compared with the other SDSCs using CuI. The disadvantage for CuI is lower V oc than those for the other p-type semiconductors, caused by carrier recombination at the CuI surfaces. 26 It has been reported that CuI surfaces act as hole trapping sites. 26 The maximal V oc determined from the difference between the CBM of TiO 2 (À4.2 eV from the vacuum level) and the VBM of CuI (À5.2 eV) is 1.0 V. 8 If V oc could be improved to close to the ideal value with securing a high J sc of 22 mA cm À2 , h of 15% would be attained. Thus, the SDSCs using the multi-dye consisting of the new porphyrin dimer dye DIPDAB2 and organic dyes with CuI are promising for highly efficient and cost-effective solar cells.

Conclusions
We have established a strategy to improve the conversion efficiency of SDSCs; to widen the absorption range leading to a high current density and to use highly conductive p-type CuI for solid-state electrolytes. The multi-dye consisting of the newly developed DIPDAB2 with a porphyrin dimer structure and two kinds of organic dyes (D131 and D358) covered notably wider range with high IPCE compared with conventional porphyrin dimer dye DTBC and organic dyes, leading to an extremely high current density over 22 mA cm À2 . The CuI electrolyte fully injected into the porous TiO 2 electrodes secured a high FF. The LiI treatments to the dye-absorbed TiO 2 electrodes and light   soaking of the completed SDSCs improved V oc with a slight increase in J sc . Eventually, the conversion efficiency reached as high as 10%.

Conflicts of interest
There are no conicts to declare.