2.1.

TCO-less Glass Fiber Tandem DSC With Light-Scattering Waveguide

Figure 1 shows a glass-fiber-type dye-sensitized solar cell structure and the working principle. The cell consists of a glass fiber/porous TiO₂ layer stained with a dye/a porous Ti electrode (anode)/a gel electrolyte sheet/a Ti sheet (counter electrode, cathode). Light is introduced from the glass rod edge and absorbed by a TiO₂ layer stained with dyes fabricated on the glass rod. Photogenerated electrons are collected by a porous Ti electrode on a porous TiO₂/dye layer. The preparation process of the porous Ti electrode is described in our previous paper.²¹ Collected electrons are directed to a counter Ti electrode with a Pt layer and are injected into I₂ in an electrolyte to form I^–. I^– species diffuse in an electrolyte layer and give electrons to oxidized dye molecules. The structure is similar to back-contact-type DSCs fabricated on a flat glass substrate reported previously.^{22, 23, 24}

Fig. 1

TCO-less glass-fiber DSC and working principle.

Figure 2 shows a TCO-less glass fiber tandem DSC structure (TCO-less TAN 1). Two cells are placed on a glass fiber and connected in series. A top cell was prepared in the following procedure. TiO₂ paste (Ti-nanooxide D - Solaronix, Aubonne, Switzerland) was coated on a glass tube (8 mm o.d., 6.3 mm i.d, 30 mm length), and the glass tube was baked at 450°C for 30 min. The process was repeated to prepare a porous TiO₂ electrode with 6 μm thickness. A mixture solution of tetrapod-shaped ZnO (Panatetra WZ-0501 - Panasonic, Matsushita Electric Industrial Co., Ltd., Toyonaka, Osaka, Japan) and P25 - Degussa (Evonik Degussa Corporation, Parsippany, NJ, USA) in ethyl alcohol was sprayed on the porous TiO₂ layer, where the tetrapod-shaped ZnO acts as a template of nanoholes.²¹ The glass tube was baked again at 450°C for 30 min. Ti was sputtered on the layer by rotating the glass tube [SH-250-T04 (Sputter–ULVAC, Kanagawa, Japan) 200 W, 6.6 × 10^–4 Pa]. After that, ZnO crystals were removed by dipping the tube into a solution consisting of 40% acetylacetone and 60% methanol to make a porous Ti electrode. The size of a ZnO crystal is from 1 to 10 μm. At least one of the four hands of a ZnO crystal sticks out from a sputtered Ti film (500 nm thick). The ZnO crystal was washed away from portions sticking out from a sputtered Ti film. Then, the substrate was dipped in a dye 2 (Fig. 3) solution [a mixture solution of dye 2 (0.25 mM) and chenodeoxycholic acid (37.5 mM) in ethanol] to stain the porous TiO₂ layer. Then, the substrate was covered with a porous poly(tetrafluoroethylene) film [(PTFE), Advantec, Tokyo, Japan, 0.1-μm pore size, 35 μm thick], followed by a Ti sheet (Nilaco, Tokyo, Japan, 50 μm thick). A bottom cell was fabricated in the same way by using dye 1 (Fig. 3). A mixture of dye 1 (0,25 mM) and chenodeoxycholic acid (37.5 mM) in ethanol was used for the dye staining. After that, an electrolyte solution (Electrolyte C) consisting of 500 mM LiI, 50 mM I₂, 580 mM t-butylpyridine, 600 mM ethylmethylimidazolium dicyanoimide in acetonitrile was injected into the porous PTFE film.

Fig. 2

TCO-less glass-fiber tandem DSC structure (TCO-less TAN 1): (A) Polystyrene dispersion (132 nm diam) in water, (B) polystyrene dispersion (592 nm diam) in water, (C) porous titania stained with dye 2, (D) porous titania stained with dye 1, (E) porous Ti electrode, (F) Ti sheet (counter electrode), (C + E) or (D + E) porous titania sheet.

Fig. 3

Model dye structures of dyes 1 and 2.

The absolute photovoltaic performances for these two model dyes were not high. These two dyes were selected because visible-light absorption of dye 2 does not overlap with that of dye 1. This simplified the analyses of tandem performances. Finally, these two cells were connected in series, as shown in Fig. 2. Light-splitting units consist of two compartments (a glass tube 2.8 mm o.d., 6.3 mm i.d., 11 mm length and a glass tube 1: 6 mm o.d., 4.8 mm i.d., 11 mm length) in which polystyrene particles with 132 and 592 nm diam (Seradyn Inc., Indianapolis, IN, USA 10% dispersion in water) were dispersed. A commercially available polystyrene dispersion in water (132 nm diam) were diluted 1000 times with water and the latter dispersion (592 nm diam) was diluted with water 200 times. The two light-splitting units were placed into the right positions of a glass tube where two TCO-less DSCs were placed, as shown in Fig. 2.

2.2.

TCO-Less Pillar Tandem DSC with Light-Splitting Waveguide (TCO-less TAN 2)

A TCO-less TAN 2 is composed of TCO-less DSCs and light-splitting waveguide consisting of dichroic mirrors (Fig. 4). Flat TCO-less DSCs (TCO-less DSCs 2-1, 2-2, and 2-3) were prepared by coating titania paste (Titania D, Solaronix) on a stainless mesh sheet (25-μm-diam stainless wire, 40-μm sheet thickness, 25-μm space between the wire, product of ASADA, Osaka, Japan). The stainless wire was protected with a TiO_x thin layer, where x changed from 0 to 2 gradually from the position contacting stainless wire surface to that contacting the surface of a porous titania layer.²⁵ This TiO_x thin layer effectively blocked charge recombination between electrons in the stainless steel surface and iodine in an electrolyte.²⁵ The TiO_x layer was fabricated by Ti sputtering (ULVAC model SH-250) using a Ti target by a method described in our previous report.²⁵ The electrode was baked at 450°C for 1 h. Three electrodes were stained by dipping these substrates into solutions of dyes shown in Fig. 5, respectively. The thickness of the porous titania layers was 6 μm. A glass (a cover glass for a microscope, 1 mm thick), a porous titania sheet stained with a dye, a gel electrolyte sheet, and a counter electrode (Pt-sputtered Ti sheet, NILAKO, t:100 μm) were merely piled up to prepare TCO-less DSCs 2-1, 2-2, and 2-3. Model dyes 1, 2, and 3 (Fig. 5) were not high-efficiency dye molecules. They were selected because their visible absorptions did not overlap each other. A gel electrolyte sheet was prepared by infiltrating Electrolyte C in a porous polymer sheet, which was described in Sec. 2.1. Three cells were connected in series on a wall of a specially designed square glass tube, as shown in Fig. 6. A 5 × 5 mm mask was placed on each DSC. Dichroic mirrors were placed in a glass tube, as shown in Fig. 4, in a square glass tube.

Fig. 4

Tandem DSCs with three dichroic mirrors (light-splitting structure).

Fig. 5

Light-absorption spectra of three dyes used for TCO-less TAN 2.

Fig. 6

Connection and structures of TCO-less DSC 21, 2-2,and 2-3 in TCO-less TAN 2.

Photovoltaic performances were monitored with a simulator KHP-1 (Bunko Keiki, Tokyo, Japan) equipped with a xenon lamp (XLS-150A). The exposure light was adjusted to be AM1.5 (100 mW/cm²). The solar simulator spectra and power were adjusted using an Eiko Seiki solar simulator spectroradiometer LS-100. Exposure power was also corrected with an amorphous Si photodetector (Bunko-Keiki BS-520 S/N 007), which has visible-light sensitivity similar to that of DSC. After the cell fabrication, the cell area was precisely recalculated using the obtaining photograph image. A mask was placed on the inlet of light of fiber or pillar glasses. Short-circuit current (J_sc) was calculated by using this area. Spectra of split light were monitored by a solar simulator spectroradiometer LS-100 (Eiko Seiki, Tokyo, Japan). Dyes 1, 2, and, 3 were synthesized in our lab, according to a method reported in our previous papers.^{26, 27}

3.1.

TCO-less Fiber Tandem DSC with Light-Scattering Waveguide (TCO-less TAN 1)

Figure 7 shows scattered light spectra from polystyrene particles of a top and bottom cells in Fig. 2. A top waveguide consisted of 132-nm-diam polystyrene particles in water mainly scattered light in the range of 400–600 nm, and the scattered light was visibly blue, as shown in an inset of Fig. 7. A bottom waveguide consisted of 592-nm-diam polystyrene particles of scattered light of wavelength of >400 nm, and the scattered light was visibly white. Figure 8 shows an incident photon-to-current efficiency (IPCE) curve for a TCO-less TAN 1 shown in Fig. 2. Two peaks were observed at around 430 and 650 nm, which correspond to dye 2 (λ_max: 429 nm) and dye 1 (λ_max: 646 nm), respectively, clearly demonstrated that two electrodes generated electrons.

Fig. 7

Scattered light spectra from polystyrene particles of (A) top cell and (B) bottom cell.

Fig. 8

IPCE curves for TCO-less TAN 1.

Figure 9 shows photovoltaic performances for TCO-less TAN 1 and corresponding single cells consisting of top and bottom cell structures. Open-circuit voltage (V_oc) of TCO-less TAN 1 was 1.15 V, which was the same as the sum of 0.59 V for a top single cell and 0.56 V for a bottom single cell. These results on IPCE curves and I–V curves show that a TCO-less TAN 1 structure worked as a tandem cell. Short-circuit current (J_sc) of a flat TCO-less DSC stained by dyes 1 or 2 was about 5–7 mA/cm² when solar light was introduced vertically to the flat TCO-less DSC (normal configuration). Low J_sc of 0.8–1.0 mA/cm² in this TCO-less DSC is explained by low light-scattering properties of polystyrene particles. A waveguide structure in which light is effectively scattered is needed to obtain high-efficiency cells. Precise photovoltaic performances were as follows: Top single cell (efficiency 0.37%, FF 0.66, V_oc 0.59 V, J_sc 0.96 mA/cm²), bottom single cell (efficiency 0.3%, FF 0.65, V_oc 0.56 V, J_sc 0.84 mA/cm²), and TCO-less TAN 1 (efficiency 0.69%, FF 0.66, V_oc 1.15 V, J_sc 0.92 mA/cm²).

Fig. 9

Photovoltaic performances (I–V curves) for TCO-less TAN 1 and corresponding single cells.

3.2.

Photovoltaic Performances of TCO-less TAN 2 with Dichroic Mirrors

Figure 10 shows split light spectra from dichroic mirrors 1, 2, 3, and a whole light spectrum from a solar simulator. Dichroic mirrors 2, 3, and 1 split lights from 350 to 520 nm, from 480 to 600 nm, and >550 nm, respectively. These dichroic mirrors were coupled with TCO-less DSCs stained by dyes 2, 3, and 1, whose absorption spectra are shown in Fig. 5. Dye 2 (350–450 nm) absorbs light split from a dichroic mirror 2 (light from 350 to 520 nm). Dye 3 (420–550 nm) absorbs light split from a dichroic mirror 3 (light from 480 to 600 nm). Dye 1 (550–670 nm) absorbs light split from a dichroic mirror 1 (light longer than 550 nm). IPCE curves for TCO-less TAN 2 had three peaks of 450, 550, and 650 nm, which correspond to those of single cells of TCO-less DSC 2-2 stained with dye 2, TCO-less DSC 2-3 stained with dye 3, and TCO-less DSC 2-1 stained with dye 1, as shown in Fig. 11. These three IPCE peaks demonstrate that three TCO-less DSCs (TCO-less DSCs 2-2, 2-3, and 2-1, shown in Fig. 6) generated electrons. Photovoltaic performances for these cells were as follows: TCO-less DSC 2-1: efficiency 0.42%, FF 0.54, V_oc 0.49 V, J_sc 1.63 mA/cm², TCO-less DSC 2-2: efficiency 0.43%, FF 0.61, V_oc 0.55 V, J_sc 1.28 mA/cm²; TCO-less DSC 2-3: efficiency 0.33%, FF 0.55, Voc 0.53 V, J_sc 1.13 mA/cm²; and TCO-less TAN 2: efficiency 1.25%, FF 0.58, V_oc 1.59 V, J_sc 1.36 mA/cm². V_oc of a TCO-less TAN 2 was 1.59 V, which was almost the sum of corresponding single cells, as shown in Fig. 12. These results show that TCO-less DSC TAN 2 structure works as a three-connected tandem cell. J_sc of TCO-less DSC 2-1 was a little higher than that of TCO-less TAN 2. Because of reproducibility of cell alignment position on the photovoltaic measurement in a solar simulator, J_sc has some experimental errors. However, the experimental errors do not deny the fact that TCO-less TAN 2 showed tandem performance because V_oc is closely related to the demonstration of tandem properties.

Fig. 10

Split light spectra from dichroic mirrors 1, 2, 3, and the whole light spectrum for solar simulator.

Fig. 11

IPCE curves for TCO-less TAN 2, TCO-less DSC 2-2 (single cell), TCO-less DSC 2-3 (single cell), and TCO-less DSC 2-1 (single cell).

Fig. 12

Photovoltaic performances (I–V curves) for TCO-less TAN 2, TCO-less DSC2-2, TCO-less DSC 2-3, and TCO-less DSC 2-1.