Evaluation of Fractal Structured Dye-Sensitized Solar Cells Made by Fused Transparent Filament Fabrication

Abstract

Flexible dye-sensitized solar cells (DSSCs) offer several benefits in terms of cheap fabrication, mass production, lightweight, and, finally, conforming to uneven surfaces. However, these cells are mainly fabricated using commercially polymer substrates with limited functional properties. DSSCs performance is highly influenced by the dye adsorbent capability of nano-crystalline oxide semiconductors (TiO₂), which require a large surface area, and this can be achieved by developing texture or microstructures on substrates. Further, texture-based substrates reduced the optical reflection, increased the optical path of light, and trapped large amounts of light. Different transparent filaments are used to print flexible substrates that possess high transparency using a versatile fused filament fabrication. Facile laser-engraved fractal textures are developed on printed polymer substrates, which act as photoanodes and counter electrodes. The maximum attained power conversion efficiency and short-circuit current density (Jsc) are 3.90% and 9.34 mA/cm², respectively, for fractal anode-fractal cathode based DSSCs, which are 82.2% and 47.7%, respectively, higher than as-printed anode- as-printed cathode-based DSSCs.

Keywords

solar cell dye-sensitized transparent filament 3D printing texture fractal pattern ITO coating

Introduction

Dye-sensitized solar cells (DSSCs) are a subset of organic solar cells that have become prominent for reasons such as the use of electron conducting materials such as the cheaply and abundantly available titanium oxide, mass-production possibilities by the roll-to-roll process at ambient conditions, abilities to function at lower light intensities including artificial light, and the low costs of dye sensitizer.¹ Conventional DSSCs are mainly based on glass-based substrates but suffer from several drawbacks. Contrarily, polymer substrate-based DSSCs are lighter, cheaper, more flexible to conform to uneven surfaces, and resistant to impact.^2,3 Also, the roll-to-roll processing of polymers is easier and faster. Polymer plates made of materials such as polyethyleneterephthalate (PET) and polyethylene-naphthalate (PEN) can be used as transparent substrates for DSSCs, but due to gas permeation, they may lose some of their performance characteristics.⁴ The schematic arrangement depicted in Figure 1 explains the basic principle of working of DSSCs. The dye sensitizer gets excited after exposure to sunlight and transfers electrons to the conducting band of mesoporous TiO₂, which then diffuses into the anode and passes on to the cathode through the external load.

FIG. 1.

Schematic of the working principle of dye-sensitized solar cells.

The commercially available polymer substrates have standard shapes, sizes, and applications with limited customization of forms. Further, these substrates or electrodes are bare, with no special architecture, depriving the achievement of additional benefits such as large reaction sites, reduced optical reflection, increased optical path of light, and trapping large amounts of light.⁵ Substrate modifications are needed achieving designed texture or microstructure to develop a high surface area. DSSCs performance is highly influenced by the dye adsorbent capability of nanocrystalline oxide semiconductors (TiO₂), which require a large surface area. Pu and Chen⁶ developed a texture of TiO₂ using a KrF excimer laser. The results showed an increased efficiency of 24% for a laser-textured photoanode. In addition, due to the rougher texture, the photoanode becomes less transparent increasing the trapped light. Lee et al.⁷ introduced a textured underlayer of TiO₂ between the conducting substrates and the nanocrystalline TiO₂. The underlayer thickness impacts reflectance, open-circuit voltage (V_oc), and fill factor (FF), and an optimum underlayer thickness gives a power conversion efficiency (PCE) of 7.10%. Next, macro-porous silicon (PSi) substrate-based photoanode was introduced to enhance dye-loading capacity and light trapping, increasing PCE up to 50%.⁵ Wooh et al.⁸ fabricated a three-dimensional (3D) pyramid-shaped patterned TiO₂ photoanode using soft lithographic technique for efficient light trapping, eventually improving the PCE by up to 36%. In a further study, 3D printing technology was used to fabricate TiO₂ thin films of optimized periodic submicron 3D pyramid design, resulting in better light scattering and tapping. This design confirms an improvement of up to 25% in the photocurrent compared with planar TiO_2.⁹ Kong et al. developed DSSCs on patterned FTO electrodes to enhance the PCE of the cells by increasing effective surface area.¹⁰ Wooh et al. proposed a novel textured 3D photoanode made of TiO₂ nanoparticles which offered higher light absorbance and photocurrent than the ordinary two-dimensional (2D) flat photoanode counterparts.⁸

In addition to textured photoanodes, textured counter electrodes (CE) were also deployed in the earlier literature, aiming to enhance the active surface areas and, eventually, the catalytic activity of electrolytes and charge transfer at the counter electrode and electrolyte interface. Tsai et al. developed nanotextured Pt CE by simply depositing Pt thin films on textured substrates and using the same in DSSCs demonstrated a 9–10% increase in both PCE and short-circuit current (SCC) results.¹¹ In another study, Nitrogen-doped graphenes (NGs) were used as a counter electrode instead of Platinum for DSSCs, rendering mesoporous textures with large specific surface areas that promote the electrocatalytic reduction of $I_{3}^{-}$ electrolytes.¹² James and Contractor fabricated nature-inspired three-dimensional (3D) fractal-based CE by 3D printing of conductive PLA. Conductive copper and catalyst graphite were selectively electroplated on the fractal structures printed and demonstrated better catalytic reactions from the active surface area.¹³

Evidently, substrate modification via textured structures has been deployed in both photoanode and counter electrode, aiming mainly to enhance the reaction sites, and consequently the PCE and SCC responses. However, the development of textures involves complex physical and chemical processes such as mechanical grooving,¹⁴ lithography,¹⁵ and electrochemical etching.¹⁶ A facile method is needed for developing controlled textures on substrates, while a substrate that inherently renders textured surfaces could be a step change. Fused filament fabrication (FFF) is one process where the fabricated substrates can be controlled to acquire possess textured surfaces. In addition, customized texture patterns can be printed by maneuvering the infill direction of struts. Furthermore, 3D texture pattern-based substrates can be printed simply by digital data from the computer-aided design (CAD) model design.

Nature-inspired textures, i.e., fractal structures, have been incorporated into numerous fields. Fractal architectures exhibit the same pattern at any observation scale. Fractal designs have been explored for heat exchangers, reactor and catalysis engineering, photobioreactor, electrodes in solar cells, fern fractal leaf-inspired antenna design, and thin-film stretchable electronics.^17–19 The fractal-based design enables a high surface area-to-volume ratio, leading to efficient thermal and electrical energy collection in energy devices.²⁰ Thekkekara et al. developed leaves fern-based laser-scribed graphene electrodes for solar energy storage. This self-powered device has an energy density of 30 times better than a planar structure.²¹ Esthetic qualities of fractal design have been adopted to design a hybrid pattern combining fractal and bus-bar arrangements of the solar cell electrode. This hybrid design renders the aesthetic property of fractal and matches the electrical properties of bus bars in light transmission.²² Nanocrystalline tin oxide thin film electrodes of highly porous fractal structures were fabricated for solid-state DSSCs exhibiting a power energy conversion efficiency comparable with TiO₂-based electrodes.²³

Despite these benefits and ready processing methods such as etching, lithography, and plasma grafting technologies, the fabrication of fractal structures is cumbersome. Berenschot et al. developed 3D fractal structures using recursive micromachining of single crystalline silicon. The advantage of this process is that this fractal shape can be transferred as a mold for other materials.²⁴ Islam and Gandhi developed multiscale fractal-like structures by controlling the interface of mixing fluid with different viscosities. A spatially distributed pit on the cell plates was developed that controls mixing, and thus, an ordered fractal-like pattern of multiple generations could be created.²⁵ Han et al. generated a quasi-fractal design by extracting leaf fern design by etching, metal film deposition, and finally transferring it to substrates using drum printing.²⁰

Using the above methods for fractal structures poses several practical constraints and is expensive. Moreover, large-scale fabrication could be challenging, as some processes require a vacuum. A more versatile manufacturing process is needed. Additive manufacturing used to fabricate shapes irrespective of the complexity can be a viable option for the physical reproduction of fractal designs. The fractal geometry can be remodeled through the iterated function system (IFS) of a CAD software to generate a CAD model and the digital data from which can directly be used on a 3D printer for the physical prototyping.²⁶ Jun et al. used selective laser melting for producing 3D fractal mono Teepole antennas. This fractal structure offered improved matching and radiation patterns of the antenna with lesser material (75% reduction).²⁷ Another interesting application of polymer-based 3D-printed fractals is in the field of photocatalytic activity. Using plasma grafting technology, a hybrid polymer material was developed by immobilizing catalyst nanoparticles on printed fractals.²⁸ Apparently, 3D printing can expedite the process of fractal design, utilizing less material, and providing high accuracy.

Commercially available PET and PEN are widely used transparent polymer substrates for flexible DSSCs.^29,30 However, it has standard shapes and sizes with limited possibilities for geometrical manipulations. There is a research gap in exploiting the beneficial role of the point-wise additive material consolidation typical of 3D printing to build fractal-structured polymer substrates, employing a variety of transparent filaments for substate fabrication. For example, P51 Polycarbonate (PC) is strong, tough, and transparent, and can sustain softening up to 96°C temperature.³¹ However, it is highly brittle and difficult to use in flexible DSSCs. Polyethylene terephthalate glycol (PETG) is another transparent filament that offers high flexibility but easily softens at moderately elevated temperatures during curing.³² However, these transparent filaments can be processed by additive manufacturing to develop intricate customized substrates, replacing PET, PEN, and glass substrates for use in DSSCs.

In the current research, FFF is used to process a transparent filament and print customized substrates for photoanode and counter electrode in DSSCs. Fractal pattern texture-based substrates (FS) are printed to experimentally evaluate the enhancement of the dye adsorbent capability of nano-crystalline oxide semiconductors, reacting sites, and contact with electrolytes. Fractal geometries are used in both photoanode and the counterelectrode, targeting enhancements in both light absorption and the number of active reaction sites for ion transfer. ITO coating on as-printed and fractal-based substrates is done using radio frequency (RF) magnetron sputtering with optimized parameters. High-order fractal pattern texture on as-printed substrates is done using laser engraving, as FFF has limitations in printing fourth-order fractals and above. Lastly, PCE and SSC density (J_SC) of as-printed electrode DSSCs are compared with FS Photo-anode DSSCs, FS counter-electrode DSSCs and FS Photo-anode counter-electrode DSSCs. The experiment results show that fractal pattern-based photoanode and counter anode outperformed in terms of Jsc, FF, and PCE, as compared with as-printed substrate-based DSSCs.

Experimental Methods

Materials and methods

Various transparent filaments are available for substrates, ranging from clear polylactic acid, acrylonitrile butadiene styrene (ABS), PETG, polymethyl methacrylate (PMMA), PC, and thermoplastic polyurethane (TPU). Natural PLA and clear PLA are two classes of transparent PLA, where clear PLA is more transparent. Clear ABS is not as transparent as PLA due to its inherent nature. PETG is a durable filament with a smoother final sample. This is suitable for printing big, clear stronger samples. PMMA is one of the transparent filaments available with high strength and impact resistance. PC is a strong and optically transparent filament, but it requires a high extrusion temperature (>300°C); otherwise, it produces translucent components. TPU which is a thermoplastic elastomers (TPE) is the only flexible transparent filament available, having good elongation and flexibility.

Flexibility and transparency being paramount, TPU was tested first for the substrates as it offers maximum flexibility with average transparency. Natural clear, flexible filament, TPE material by 3D Solutech, Amazon USA has been selected as the experimental material to print flexible substrates as it was observed to print as easily as the common polymer filaments used for 3D printing. Since the transparency was not as good as expected, post-processing by sandpaper cleaning and chemical treatment with acetone and ethanol was needed to improve the transparency. The chemical treatments cause two successive layers to merge and avoid entrapped gases or air bubbles, thereby improving transparency. Further, printing parameters such as 100% infill, slow printing speeds, and thicker first few layers were found to further improve the transparency. The CAD model of the bare substrates and the as-printed and post-processed TPE substrates are shown in Supplementary Figure S1(a), and Supplementary Figure S2, respectively.

Next, the transparent P51 PC filament from Sigma-Aldrich New Zealand was tested for printing the DSSC substrates. The printed prototype is shown in Supplementary Figure S1 (c), which has proved out to be highly brittle, exhibiting low transparency, and difficult to print. Filament breakage in the 3D printer filament channel is the major issue. In addition, printed bare substrates are too brittle to be used as flexible substrates. Lastly, polysmooth transparent polymer filaments from Polymaker, 3D printing solutions, Australia were also found to offer maximum transparency with average flexibility and had been used in the current study for all substrate printing. The polysmooth^TM material is based on polyvinyl butyral for smooth consolidation. The polysmooth^TM part needed no cumbersome post-processing after printing to achieve transparency. After printing, the substrates, isopropyl alcohol, were sprayed three times on each side and let it dry in ambient conditions for 12 h. The isopropyl alcohol helps merge the layers and remove the trapped air, eventually rendering the substrates to be more transparent. Air entrapment is possible after prolonged exposure to ambient conditions, which eventually deteriorates the performance of DSSCs. Technical information from the website of Polysmooth website clearly indicates that the optimum moisture absorption level is too low at around 2.36%. Considering this, it is expected that polymer degradation is not a major issue at this stage. Substrates are 3D printed with polysmooth filaments of 1.75 mm diameter, using the FFF method based on the CreatBot F430 commercial 3D printing system. Initially, the manufacturer’s specifications were used to try and print substrates using polysmooth on the CreatBot system. However, a few initial trials became necessary before converging on the final optimum process conditions combining the material and printing system together. Based on the observations from these initial repeated trials, the final optimized process parameters are layer thickness 0.4 mm, infill type meander style at 90° rotation, nozzle diameter 0.4 mm, layer overlap 50%, infill density 100%, printing speed 60 mm/s, printing head temperature 200°C, and support plate temperature 50°C. The slicing software used is Simplify 3d, and rectangular shapes of thickness 1 mm are printed to act as substrates.

The dimensions of the substrates are 2 cm × 2 cm for both the photoanode and counter electrode. The design is done in the SolidWorks (2020) CAD software packages. IFS is a mathematical method of constructing fractals based on the set theory. An IFS fractal is made up of a union of several copies of itself and drawn in 2D.³³ The IFS creates a virtual model of fractal-based design, which needs to be remodeled to develop a physical model using additive manufacturing. Ullah et al. did three remodeling methodologies based on binary-grid, convex/concave-hull, and line-model techniques.²⁶ In addition, the fractal pattern (Hilbert curve in the current study) design is done in Wolfram Mathematica 12.0, following the method by Li et al. who coded a different 2D fractal-based curve in Mathematica software and exported it into SolidWorks.³⁴ The CAD model developed is converted into (.STL) format and exported for printing. The CAD models of fractal pattern-based substrates (fourth-order fractal) and corresponding printed substrates are shown in Supplementary Figure S1 (b) and (d), respectively. However, higher order fractal pattern printing (sixth order or above) becomes difficult as printing dimensions transit from micro to nano range. So, a facile process is needed for developing fractal patterns on printed substrates. For this, the SolidWorks CAD model is exported in (.DXF) format to a laser engraving system (Thunder Laser, Laser Cutter-Nova Series 35, DC CO2, China). The laser engraving parameters are laser engraving speed 200 mm/s, laser power 16 W, and laser beam size 0.14 mm.

Once the substrates with the required criteria are printed and post-processed for transparency, the fractal pattern is laser engraved in the case of texture-based substrates. The next step is to develop an electrode by coating a transparent conducting material on the printed substrates. Indium-doped tin oxide (ITO) is widely used as a transparent conducting oxide material for coating different polymer substrates. Different polymer-based substrates and coating methods were reported earlier for coating transparent conductive oxides on polymer substrates. Kim et al. deposited ITO by RF magnetron sputtering on a PMMA polymer substrate. The sheet resistance of 19 Ω/sq and optical transparency of 85% using ITO thin films (150 nm) were obtained by heating the substrates.³⁵ Yang et al. compared the electrical and optical properties of deposited ITO by RF magnetron sputtering on glass, PC, and metallocene cyclo-olefin copolymers (mCOC) substrates and reported 6.35 × 10⁻⁴ Ω-cm, 5.86 × 10⁻⁴ Ω-cm and 6.72 × 10⁻⁴ Ω-cm as resistivity responses, respectively, and the transparency values in the range of 85%–90% in the visible range for all.³⁶ In this work, oxides are deposited on nonconducting polymer substrates by RF magnetron sputtering unlike conducting material where dc (DC) magnetron sputtering is used.³⁷ RF magnetron sputtering is a process where an RF electric field accelerates inert gases to bombard a target material, which is the specimen to be sputtered (ITO material in the present case). The target is sputtered all around, while the magnetic field developed constrains, the sputtered atoms to be directed toward substrates, which are placed in front of the target but outside of the generated plasma region.³⁸ The ITO sputtering target is 99.99% pure (In₂O₃/SnO₂ 90/10 wt %), has a diameter of 3 inches, and a thickness of 2 mm, and the back plate copper is purchased from Hunan Fushel Technology Limited, Hunan, China. The ITO target is deployed in an ATS 500 UK RF magnetron sputtering machine with the optimized parameters as mentioned in Table 1. The microstructure and morphology of the thin films are determined using field emission scanning electron microscopy (Hitachi S4160, Cold Field Emission, voltage 20KV). The crystallinity of the thin films is characterized by an XRD system (Empyrean, Malvern Panalytical, PIXcel^3D area detector, UK, Cu K alpha wavelength = 1.5406˚A). The sheet resistance of ITO films is measured with the four-point probe method (Ossila Four-Point Probe System, UK). The transmittance spectra of the ITO coating and light-absorption characteristics of photoanodes (as-printed and fractal) are examined by UV/VIS/NIR spectrophotometer (Jasco V-770, US).³⁹ In addition, the two-point resistance measurement uses a multimeter, as shown in Supplementary Figure S3.

Table 1.

RF Magnetron Sputtering Process Parameters for ITO Target and the Coating Responses on Printed Substrates

Parameters	Value
Base pressure	4.5 e-4 Pa
Working pressure	2 Pa
Argon gas flow	70 sccm
Sputtering power	100 W
Rotation	15 RPM
Temperature	Room temperature
Substrate target distance	10 cm
Optical transparency	78 %
Sheet resistance	28 Ω/sq

CM, centimeters; ITO, indium-doped tin oxide; Pa, pascal, RF, radio frequency; RPM, revolutions per minute; SCCM, standard cubic centimeters per minute.

Hitachi E1045 Ion Sputter system is used for the ion beam sputtering of the counter electrode using a platinum target.⁴⁰ A few trial-and-error runs were done varying the process conditions and measuring the conductivity and continuity of the coating by the four-point probe method and SEM imaging respectively. Finally, the optimum ion beam sputtering parameters are identified to be sputtering current 25 mA, working pressure 7 Pa, substrate target distance 30 mm, and argon gas flow rate 30 sccm. The sputtering is done at room temperature. The coating is done for 5 min at a rate of 10 nm/minute, depositing a coating thickness of ∼50 nm at the end. The four-point probe method is used to evaluate the sheet resistance of the deposited layer. In addition, the two-point resistance measurement is done using a multimeter, as shown in Supplementary Figure S4 for manual measurement and verification The printed substrates coated by ITO and platinum will be the anode and the cathode, respectively. Further, TiO₂ slurry with precursors is plastered over the anode surface and heated to temperatures around 150–200°C to remove the precursors and achieve the mesoporous TiO₂ structures. Post-processing is required to develop necking between TiO₂ nanoparticles, and earlier reported treatments are static compression to interconnect grains, UV irradiation to oxidize and remove organic compounds, and hydrothermal treatments to convert amorphous TiO₂ to crystalline form.⁴¹ Most of the polymer substrates cannot withstand the high-temperature sintering needed to evaporate the binder (polyethylene glycol and tert-butanol) as well as the high compression needed for necking. So, an alternative binder-free TiO₂ coating process reported recently by Poh et al.⁴² is used in this work, where the ratio of TiO₂ and ethanol is varied to obtain an effective paste for coating. Further, hydrochloric acid (HCl) is added to the TiO₂ paste to improve the connectivity of the particles.⁴³ Poh et al. developed binder-free nanostructured TiO₂ flexible DSSCs that achieve PCE of 2.12% with TiO₂ powder to ethanol of 25 wt%. In addition, suitable proportions of HCl are added to the slurry to improve the particle connectivity.⁴² The two-step preparation starts with mixing P-25 TiO₂ powder (Aeroxide^(R) P25, Evonik Industries AG, Germany) with ethanol in the weight ratio of 25% (TiO₂ to ethanol) using mortar and pestle under ambient conditions. The original size of TiO₂ particles ranges from 10 to 50 nm. A unique combination of anatase and rutile crystal structure (predominantly anatase structure) is shown by the TiO₂ powder. A few drops of 1 M HCl are added while stirring the paste to improve the viscosity. Once adequate rheology and viscosity of the paste are achieved, as shown in Supplementary Figure S5, the paste is deposited on the ITO-coated side of the substrates via a doctor blade technique over an area of 1 × 1 cm². The film is dried under ambient conditions for 10 min and subsequently at 50°C for 1 h in the oven. This heating temperature is relatively much lower than what Poh et al. used in the literature (150°C) because the printed polymer substrates start deforming in shape and cannot withstand high temperatures.

A bending test is performed after the TiO₂ coating is dried on flexible substrates. The test method is manually bending the coated sample, following the same procedure as reported by Poh et al under outward bending mode, where the opposite corners of the substrates are bent backward so that the TiO₂ film faces tension.⁴² Repeated testing and imaging showed the film to be intact even after three months. Also, this work is still at the proof-of-concept stage and once the basic functionality is ascertained, further process enhancements can be evaluated for the longevity of the film coating on the flexible substrates. The film coating area is 2 × 1 cm², and the visual appearance of the film after bending is shown in supplementary Figure S6. Once the TiO₂ film is dried, it is subsequently dye-immersed for 24 h. Ruthenizer 535 complex dye (N719, 95%) procured from Sisco Research Laboratories Pvt. Ltd, India is used. The Ruthenizer 535 complex dye is mixed with deionized water (0.5 mM of N719 in deionized water) instead of ethanol. The TiO₂ coated substrates are immersed in the dye and left for 24 h to be fully absorbed on the TiO₂ thin layer. The dye solution is 0.5 mM of N719 in deionized water, unlike the ones reported in the earlier literature where ethanol was used as a solvent.⁴⁴ The printed substrates become highly sticky and lose their shape once immersed in ethanol solution dye, which is why the deionized water solution dye is considered here.

Flexible DSSCs are assembled by sandwiching dye-soaked photoanode and Pt-coated counter electrode holding the assembly together by clips. Before assembling, a few drops of high-stability electrolytes (HSE) purchased from Sigma-Aldrich (USA) are used.⁴⁵ The assembled DSSCs are shown in Supplementary Figure S7. Photocurrent-voltage (I-V) characterization of assembled DSSCs is performed using keysight (B2911A) precision source/measure unit. The solar simulator (ABET Technologies, 10500) is used to illuminate the light of 100 mV/cm² after calibrating with a photodiode. The I-V response is measured three times, and the result describes the mean value. The performance of the DSSCs is calculated based on the formula shown below in Equation 1.

η = \frac{J_{s c} * V_{o c} * FF}{P_{i n}}

(1)

where

J_{s c}

is the measured photocurrent density based on short-circuit (based on the effective electrode area of TiO₂ film is 1 cm²),

V_{o c}

is the open-circuit voltage, FF is the fill factor, and

P_{i n}

is the light intensity of the solar simulator (100 mV/cm²).

Results and Discussions

Microstructural characterizations

Optical micrograph images of the as-printed and the laser-engraved substrates are presented in Figure 2a and b, respectively. Figure 2a shows the parallel stranded structure resulting from the fused filament tracks deposited as per the raster strategy followed. Figure 2b clearly shows the fractal structure resulting from the laser engraving superimposed on the parallel raster structure. Evidently, the laser engraving has completely changed the substrate surface structure, adding microchannels and additional deep holes at the end of the raster scan lines, together with spattered particulate forms on the high surfaces of the additional groves formed.

FIG. 2.

Optical micrograph images of (a) as-printed substrates and (b) laser-engraved substrates and SEM images of the substrate after sputter coating in the (c) as-printed and (d) after laser engraving. SEM, scanning electron microscopy.

Figure 2c shows the straight and parallel groves formed due to the deposition of the polymer filament following the raster path orientation and the resulting strands that coalesce gradually into each other resulting in the meso-structures typical of FDM. Further, as evident in Figure 2d, the laser engraving has added the fractal structure on top of the normal strand structure from the FFF of the substrate. Again, added porosity and deep holes at the ends of the laser tracks are evident in Figure 2d because of the laser engraving to produce the fractal structured topography for the substrate surface.

In addition to fractal patterns, the printed polymer substrates inherently carry infill patterns that develop during printing. It may be observed that the sputter-coated platinum layer is uniformly spread all around the electrode. There may be some role played by sputter coating in imparting surface porosity, but the laser engraving played a major role in the porosity of the surface observed on the surface topology. At the end of each linear track, the laser stops and either takes a turn or moves in a different direction. This slowing down of the laser beam without the reduction in power has resulted in deep holes at the end of each of the laser scan lines during the course of remelting by laser engraving to obtain the fractal structures. The platinum coating could partly fill some of these deep holes but, there are rabbit-hole like impressions left at places where the platinum coating is unable to cover the laser penetration fully. The laser scanning also led to spattering of the polymer particles from the deep gouges made which landed on adjacent banks on either side of the laser tracks and remained plastered as debris seen as fine particles disbursed all around the higher areas of the fractal structured surface as shown in Figure 2b and d.

The theoretical form and dimensions of the seventh-order fractal structure are presented in the supplementary data S9 C. The SEM images of the surface morphologies of ITO-coated as-printed and laser-engraved fractal pattern surfaces are shown in Figure 3a and b, respectively. The thickness of the ITO coating is < 100 nm, whereas the theoretical fractal depth is 0.1 mm. Although ITO coating covers the grooves, wall sides, and top surfaces uniformly, it does not cover the full depth of the grooves. Examining the two images in Figure 3 representing the nanostructured patterns of the coatings at the same magnification, it may be noted that the ITO coating adheres more densely to the fractal-based polymer substrates compared with the as-printed counterpart. In both cases, the images are taken at the depth of the grooves. Evidently, the fractal pattern results in a rough surface, which eventually allows for better ITO deposition per unit area. Further, the ITO-coated fractal pattern substrates showed less sheet resistance than the as-printed substrates, justifying the denser coating as evidenced by the two-point multimeter measurements done subsequently. The SEM image of the surface morphology of the photoanode coated with TiO₂ is shown in Figure 4. The TiO₂ particles are aggregated, and the interparticle connectivity and the necking phenomenon are more pronounced in this case compared with the observations reported by Poh et al.⁴² earlier for the same weight ratio of TiO₂. A plausible reason is the low-temperature (50°C) evaporation of HCl, which is responsible for partially and temporarily dissolving the surface of TiO₂ particles. After evaporation of HCl, the physical junction between nearby particles remains in aggregates, resulting in the structure as seen in Figure 4. The high-temperature heat treatment at 150°C used by Poh et al.⁴² develops porosity, but the particles are dispersed more uniformly with less agglomeration.

FIG. 3.

SEM images of the ITO coated (a) as-printed and (b) fractal pattern-based polymer substrates at a magnification of 200 KX. ITO, indium-doped tin oxide; SEM, scanning electron microscopy.

FIG. 4.

SEM images of deposited TiO₂ film with a weight ratio of 25% in ethanol. SEM, scanning electron microscopy.

Crystallinity characterization

Figure 5a represents the XRD patterns of ITO films deposited on the 3D-printed transparent substrates using Polysmooth. The first intense diffraction band is centered at about 19.4^° illustrating the amorphous structure of the polymer substrates. The four high intense diffraction peaks at about 30^°, 35^°, 51^°, and 60^° in these patterns coincide with those produced by the (2 2 2), (4 0 0), (4 4 0), and (6 2 2) crystalline planes of the cubic structure of indium oxide.^37,46 The XRD pattern of the P25 TiO₂ thin film powder is shown in Supplementary Figure S8. The diffraction patterns exhibited reflections indexed at (1 0 1) 25.40^°, (1 0 3) 37.17^°, (0 0 4) 37.93^°, (1 1 2) 38.68^°, (2 0 0) 47.05^°, (1 0 5) 54.05^°, and (2 1 1) 55.18^°, corresponding to a tetragonal anatase structure, and indexed at (1 1 0) 27.50^°, (2 1 1) 55.35^°, (1 0 1) 36.05^°, (1 1 1) 41.22^°, and (2 1 0) 44.07^° to a tetragonal rutile structure.⁴⁷ XRD spectra of the ion beam sputtered thin film coating of platinum on the 3D-printed substrate are presented in Figure 5b. The first peak is obtained at 19.2^°, corresponding to the 3D-printed polymer substrates. The intensity of the peak (1 1 1) 39.5^° is the highest among all. Other observed peak intensities are (2 0 0) 45.8^°, (2 2 0) 67.17^°, (3 1 1) 80.4^°, and (2 2 2) 87.2^°. These peak intensities match with earlier studies done by Slavcheva et al.⁴⁸ However, the peak intensities observed in the current study are less than those reported by Slavcheva et al. as the sputtering duration is 10 min, leading to a lesser coating thickness.

FIG. 5.

XRD patterns of (a) ITO films deposited and (b) platinum films based on ion-beam sputtering on 3D-printed transparent polymer substrates. ITO, indium-doped tin oxide; XRD, X-ray diffraction.

Light-absorption characteristics of photoanodes

The patterned photoanode can trap more light than the planar counterparts, as was reported in the past.^5,8 In the current research, photoanodes based on the as-printed and the fractal-based substrates are used. Light-absorption characteristics are evaluated using UV–vis spectroscopy on N719 dye-loaded photoanodes, and the resulting spectra are presented as percent absorption against wavelength in Figure 6. Following in the same lines as reported by Wooh et al, the UV–vis absorption spectra are performed in the visible range (400–800 nm) on the smooth sides, with both fractal and as-printed samples. In the as-printed states, the substrates are already light trapping to some extent due to the inherently textured material domains resulting from the raster strategies used. The fractal structures superimposed by laser engraving will further enhance the light trapping as reflected in the two graphs presented in Figure 6. In comparison, the light absorption response of the fractal-based photoanode is almost half that of the result reported by Wooh et al.⁸ One of the obvious reasons is the internal structure of the substrate. Printed substrates are porous, even after printing to 100% density, compared with glass-based or commercial polymer-based substrates. This might allow a fraction of light to be absorbed within the multilayered polymer substrate. Commercial glass and polymer substrates with around 90% transparency can absorb light better on the fractal side, whereas the printed polymer substrates with a transparency of around 40% would have much lesser light absorption despite the presence of fractal structures. All these factors contribute to the lesser absorption capability of the as-printed and fractal-based substrates compared with commercial glass or polymer-based substrates. Visual examination of light absorption by fractal-based substrates is presented in the supplementary Figure S9, which clearly depicts a significant increase in absorbance with the enhancement of light trapping capability when compared with as-printed substrates.

FIG. 6.

Light-absorption characteristics of the as-printed and the fractal TiO₂ photoanodes after N719 dye loading.

DSSCs performance characterization

The photocurrent density–voltage (J–V) characteristics of the printed substrate DSSC and its performance are evaluated based on SCC density (J_SC), open-circuit voltage (V_OC), short-circuit current (I_SC), maximum voltage (V_max), maximum current (I_max), maximum power (P_max), FF, and cell efficiency ( $η$ ) responses. The J–V response values of the DSSC are listed in Table 2, and the graphical representation of which is presented in Figure 7. The results in Table 2 or the curves in Figure 7 clearly demonstrate that the fractal-based anode and cathode (FA-FC) outperformed all other combinations of substrates in terms of the short-circuit current, FF, and PCC responses. Highly textured anodes are proven to develop high surface areas, eventually rendering large reaction sites, and trapping large amounts of light. The performance of a DSSC is highly influenced by the dye adsorbent capability of the nanocrystalline oxide semiconductors (TiO₂), which has a direct correlation with the active surface area. The 3 mL of 0.1 M NaOH aqueous solution desorbed the N719 dye from the TiO₂ film, and this solution is analyzed for the UV–Vis spectra.⁴⁹ Although the dye absorbance capability of TiO₂ obtained from the as-printed and fractal photoanode witnessed the same behavior, this result as presented in Supplementary Figure S10 is in accordance with other observations reported earlier by Wooh et al.⁸ However, the absorbance spectra obtained from UV–Vis spectroscopy, as shown in Figure 6, clearly elucidate the fractal-based photoanode absorbance capability to be higher than that of the as-printed counterparts, which explains the better performance of the fractal photoanode-based DSSCs. In addition, fractal-based counter electrodes (FC) have been proven to enhance the catalytic activity of electrolytes and charge transfer at the counter electrode and electrolytes interface. Eventually, the FA-FC-based DSSCs leveraged the benefits of texturing both the anode and the cathode and resulted in almost two times better PCE responses compared with the as-printed anode and cathode-based DSSCs (PA-PC). Further, the current results also indicated that the FA-FC-based DSSCs generate a maximum current density (J_max) of 6.71 mA/cm² and PCE of 3.904%, whereas the FA-PC-based DSSCs generate a J_max of 6.10 mA/cm² and PCE of 3.10%. This result suggests that the fractal-based photoanode performs better than the fractal-based counter electrode in terms of efficiency and current density. This has also been observed in earlier research where significantly more attention was paid to the texture-based anode than the texture-based cathode.

Table 2.

Critical Responses of the DSSCs Experimentally Tested with Varying Conditions and a Comparative Evaluation with the Corresponding Results Reported by Poh et al.⁴²

Substrates	Current study				Poh et al.⁴²	Improvement against the best result
Substrates	FA-FC	FA-PC	PA-FC	PA-PC	Best output	Improvement against the best result
J_SC (mA/cm²)	9.34 ± 0.41	8.09 ± 0.36	7.11 ± 0.31	6.32 ± 0.29	6.43	45.26%
V_oc (V)	0.84 ± 0.038	0.81 ± 0.04	0.79 ± 0.04	0.78 ± 0.04	0.72	16.67%
FF	0.50 ± 0.02	0.47 ± 0.02	0.45 ± 0.02	0.43 ± 0.02	0.46	8.70%
$η$	3.90 ± 0.18	3.10 ± 0.13	2.56 ± 0.12	2.14 ± 0.10	2.12	83.96%

DSSCs, dye-sensitized solar cells; FA-FC, fractal-based anode and cathode; PA-PC, as-printed anode and cathode-based DSSCs.

FIG. 7.

J–V curves of as-printed and fractal substrate-based anode or cathode for DSSCs, where FA represents fractal anode substrates, DSSCs, dye-sensitized solar cells; FC, fractal cathode; J–V, photocurrent density–voltage PA, as-printed anode; PC, as-printed cathode.

Finally, it is considered pertinent to compare the results from the current investigations with the reports made earlier by Poh et al.,⁴² as the electrolytes (high stability electrolyte), dye (N719), binder-free titanium oxide film formation, and the solar simulator illumination power (100 mW/cm²⁾ are all the same in both the studies. The printed polymer substrates, dye solution dissolved in DI water, and the ITO coating result 28 ohms/sq are the differences in the present study. Poh et al.⁴² reported J_sc 6.43 mA/cm², FF 0.46, and PCE 2.12%, and results from the PA-PC-based DSSCs from the current research at J_sc 6.32 mA/cm², FF 0.43, and PCE 2.14% are comparable. The DSSCs based on the as-printed polymer substrates show lower transparency (around 40%), higher sheet resistance of the electrode, use DI water-based dye, and the lower temperature sintering of TiO₂ film causes particle aggregation and consequent formation of the film with lower porosity. Despite these shortcomings, the current results from the PA-PC-based DSSCs are still comparable with the earlier reported values due to the beneficial role played by the inherently textured structure of photoanode and the counter electrode of as-printed substrates. Further, laser engraved fractal-based FA-FC DSSCs resulted in J_sc 9.34 mA/cm² and PCE 3.90%, which are 45% and 83%, higher, respectively, than the values reported by Poh et al.⁴² Evidently, these results prove that the fractal pattern-based photoanode and counter anode outperform the rest of the combinations in terms of J_sc, FF, and PCE responses.

Conclusions and Future Prospects

For the first time, transparent and flexible components are fabricated by 3D printing to be used as substrates for DSSCs. Facile laser engraving is also successfully used to build fractal-based textures on the printed substrates that serve as photoanode and counter electrode in the DSSCs. The progress with DSSCs based on 3D printed fractal-structured substrates is still in the initial stages of development, but the results are indicative of positive trends. The maximum attained PCE, and Jsc are 3.90% and 9.34 mA/cm², respectively, for FA-FC-based DSSCs, which are 82.2% and 47.7%, higher than PA-PC-based DSSCs, respectively. The fractal-based photoanode performs better in terms of J_sc and PCE than the fractal-based counter electrode. The main shortcoming is that higher-order fractal textures are found to be difficult to achieve directly by 3D printing based on the FFF processes. Porosity is another issue that causes light to be trapped between the layers of substrates, thereby reducing the overall efficiency of the cell. Beyond this, the transparency responses of the printed substrates evaluated in the current study were of concern as they stand at almost half the transparency levels possible by the glass substrates. Further, the transparency and conductivity levels of the ITO coatings obtained by RF magnetron sputtering were not on par with the commercially available polymer substrates.

For the future course, an immediate task is to further explore and optimize the ITO coating, targeting better transparency and conductivity responses. Next, the TiO₂ coating method also needs to be optimized to achieve better mechanical integrity, stability, and longevity under repeated bending conditions. Further, improvements in substrate fabrication can be explored by means of stereolithography (SLA) 3D printing process, where transparent resins can be used to build solid, homogenous, isotropic, and more transparent substrates that are free of porosity. In addition, more intricate, micro to nano-range textures can be developed in the printing process itself using a CAD model, unlike the FFF process. SLA can also allow to develop micro-scale textures, in situ, during the printing process, thereby avoiding the post-printing laser engraving step in the production process. Another interesting aspect is to explore the possible integration of arrays of micro-lenses into the printed substrates by suitably controlling the SLA process parameters for better convergence of incident light on the photoanode, thereby enhancing the cell performance. Beyond this, it is also necessary to explore alternative printable transparent substrates that are inert to essential dye solutions such as ethanol. More emphasis can also be placed on facile and versatile techniques for coating semiconductor TiO₂ that render mesoporous structures without the need for high-temperature sintering. Lastly, an alternative technique needs to be explored for developing the substrate textures, as the current laser engraving caused unintentional porosity and spattering of materials all around. For example, development of nature-inspired dendritic structure-based textures is a possible avenue to explore, targeting further enhanced throughputs of DSSCs.

Authors’ Contributions

V.K. is the scholar and the main contributor to the conceptualization, methodology, software, verification, investigation, writing original drafts, and reviewing and editing tasks. S.S. is a Professor in Mechanical Engineering at AUT and the primary supervisor of the research undertaken. His contributions include conceptualization, methodology, resources, original draft writing, review and editing, supervision, project administration, and funding acquisition tasks.

Footnotes

Author Disclosure Statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.

Funding Information

Sarat Singamneni received funding from Royal Society of New Zealand based on the Marsden Grant No. MFP AUT1901 through which some materials and in-kind support was extended to carry out the research reported here.

Supplemental Material

References

Richhariya

, Kumar

, Tekasakul

, et al. Natural dyes for dye sensitized solar cell: A review. Renewable & Sustainable Energy Reviews, 2017; 69(vember 2016):705–718; doi: 10.1016/j.rser.2016.11.198

Nogueira

, Longo

, De Paoli

M-A

. Polymers in dye sensitized solar cells: Overview and perspectives. Coordination Chemistry Reviews, 2004; 248(13–14):1455–1468; doi: 10.1016/j.ccr.2004.05.018

Baxter

. Commercialization of dye sensitized solar cells: Present status and future research needs to improve efficiency, stability, and manufacturing. J Vacuum SciTechnolo A: Vacuum, Surfaces, and Films, 2012; 30(2):20801; doi: 10.1116/1.3676433

MacDonald

, Looney

, MacKerron

, et al. Latest advances in substrates for flexible electronics. J Soc Info Display, 2007; 15(12):1075–1083; doi: 10.1002/9783527679973.ch10

Aliaghayee

, Fard

, Zandi

. Development of macro-porous silicon based dye-sensitized solar cells with improved light trapping. J Electrochem Sci Technol, 2016; 7(3):218–227; doi: 10.5229/JECST.2016.7.3.218

M-Y

, Chen

. Improved performance of dye-sensitized solar cells with laser-textured nanoporous TiO2 photoanodes. Materials Letters, 2012; 66(1):162–164; doi: 10.1016/j.matlet.2011.08.022

Lee

C-H

, Lu

M-D

, Guan

Q-Z

, et al. Thickness-controllable textured TiO2 underlayer for a flexible dye-sensitized solar cell sub-module. Mater Res Express, 2014; 1(2):25503; doi: 10.1088/2053-1591/1/2/025503

Wooh

, Yoon

, Jung

, et al. Efficient light harvesting with micropatterned 3D pyramidal photoanodes in dye‐sensitized solar cells. Adv Mater, 2013; 25(22):3111–3116; doi: 10.1002/adma.201300085

Knott

, Makarovskiy

, O’Shea

, et al. Scanning photocurrent microscopy of 3D printed light trapping structures in dye-sensitized solar cells. Sol. Energy Mater. Sol. Cells, 2018; 180:103–109; doi: 10.1016/j.solmat.2018.02.028

10.

Kong

, Xiao

, Kim

, et al. Performance improvement of dye-sensitized solar cells by surface patterning of fluorine-doped tin oxide transparent electrodes. Thin Solid Films, 2011; 519(10):3173–3176; doi: 10.1016/j.tsf.2011.01.251

11.

Tsai

C-H

, Hsu

S-Y

, Lu

C-Y

, et al. Influences of textures in Pt counter electrode on characteristics of dye-sensitized solar cells. Organic Electronics, 2012; 13(1):199–205; doi: 10.1016/j.orgel.2011.10.020

12.

Gao

, Wang

, Chang

, et al. Nitrogen doped porous graphene as counter electrode for efficient dye sensitized solar cell. Electrochimica Acta, 2016; 188:441–449; doi: 10.1016/j.electacta.2015.12.008

13.

James

, Contractor

. Study on nature-inspired fractal design-based flexible counter electrodes for dye-sensitized solar cells fabricated using additive manufacturing. Sci Rep, 2018; 8(1):17032–17012; doi: 10.1038/s41598-018-35388-2

14.

Gerhards

, Marckmann

, Tolle

, et al. Mechanically V-textured low cost multicrystalline silicon solar cells with a novel printing metallization. IEEE, 1997:43–46; doi: 10.1109/PVSC.1997.653920

15.

Nositschka

, Beneking

, Voigt

, et al. Texturisation of multicrystalline silicon wafers for solar cells by reactive ion etching through colloidal masks. Solar Energy Materials and Solar Cells, 2003; 76(2):155–166; doi: 10.1016/S0927-0248(02)00214-3

16.

Dobrzański

, Drygała

, Gołombek

, et al. Laser surface treatment of multicrystalline silicon for enhancing optical properties. J. Mater. Process. Technol. 2008; 201(1–3):291–296; doi: 10.1016/j.jmatprotec.2007.11.278

17.

Moon

D-B

, Lee

, Roh

, et al. Three-dimensional out-of-plane geometric engineering of thin films for stretchable electronics: A brief review. Thin Solid Films, 2019; 688:137435; doi: 10.1016/j.tsf.2019.137435

18.

Zhao

, Zeng

, Gu

, et al. Nature inspired fractal tree-like photobioreactor via 3D printing for CO2 capture by microaglae. Chem Eng Sci, 2019; 193:6–14; doi: 10.1016/j.ces.2018.08.057

19.

Huang

, Hwang

, Radermacher

. Review of nature-inspired heat exchanger technology. Inter J Refrig, 2017; 78:1–17; doi: 10.1016/j.ijrefrig.2017.03.006

20.

Han

, Huang

, Li

, et al. Bio-inspired networks for optoelectronic applications. Nat Commun, 2014; 5(1):5674–5677; doi: 10.1038/ncomms6674

21.

Thekkekara

, Gu

. Bioinspired fractal electrodes for solar energy storages. Sci Rep, 2017; 7(1):45585–45589; doi: 10.1038/srep45585

22.

Roe

, Bies

, Montgomery

, et al. Fractal solar panels: Optimizing aesthetic and electrical performances. PLoS One, 2020; 15(3):e0229945; doi: 10.1371/journal.pone.0229945

23.

Stergiopoulos

, Arabatzis

, Cachet

, et al. Photoelectrochemistry at SnO2 particulate fractal electrodes sensitized by a ruthenium complex: Solid-state solar cell assembling by incorporating a composite polymer electrolyte. J. Photochem. Photobiol. A, 2003; 155(1–3):163–170; doi: 10.1016/S1010-6030(02)00394-5

24.

Berenschot

, Jansen

, Tas

. Fabrication of 3D fractal structures using nanoscale anisotropic etching of single crystalline silicon. J Micromech Microeng, 2013; 23(5):55024; doi: 10.1088/0960-1317/23/5/055024

25.

Effect of gold loading and TiO2 support composition on the activity of Au/TiO2 sphotocatalysts for H2 production from ethanol–water mixtures. Journal of Catalysis, 2013:307–317; doi: 10.1016/j.jcat.2013.05.031

26.

Ullah

AMMS

, D’Addona

, Harib

, et al. Department of Mechanical Engineering, Kitami Institute of Technology. Fractals and additive manufacturing. Int J Automation Technol, 2016; 10(2):222–230; doi: 10.20965/ijat.2016.p0222

27.

Jun

, Sanz-Izquierdo

, Parker

, et al. Manufacturing considerations in the 3-D printing of fractal antennas. IEEE Trans Compon, Packag Manufact Technol, 2017; 7(11):1891–1898; doi: 10.1109/TCPMT.2017.2730366

28.

de Rancourt de Mimérand

, Li

, Guo

. Photoactive hybrid materials with fractal designs produced via 3D printing and plasma grafting technologies. ACS Appl Mater Interfaces, 2019; 11(27):24771–24781; doi: 10.1021/acsami.9b06982

29.

Wei

Y-H

, Chen

C-S

, Ma

C-CM

, et al. Electrochemical pulsed deposition of platinum nanoparticles on indium tin oxide/polyethylene terephthalate as a flexible counter electrode for dye-sensitized solar cells. Thin Solid Films, 2014; 570:277–281; doi: 10.1016/j.tsf.2014.03.060

30.

Lee

K-M

, Chiu

W-H

, Lu

M-D

, et al. Improvement on the long-term stability of flexible plastic dye-sensitized solar cells. J Power Sour, 2011; 196(20):8897–8903; doi: 10.1016/j.jpowsour.2011.06.049

31.

P51 Polycarbonate 3D Printing Filament. Available from: https://www.sigmaaldrich.com/NZ/en/product/aldrich/trd3d0062 [Last accessed: March 11, 2023].

32.

Baiju

, Murali

, Rao

, et al. Heat sink assisted elevated temperature sintering process of TiO2 on polymer substrates for producing high performance flexible dye-sensitized solar cells. CEP:PI, 2020; 149:107817; doi: 10.1016/j.cep.2020.107817

33.

Zobrist

, Sabharwal

. Progress in computer graphics. 1992.

34.

, Huang

, Le

, et al. Scale-free and small-world properties of VAF fractal networks. Fractals, 2016; 24(03):1650033; doi: 10.1142/S0218348X1650033X

35.

Kim

D-H

, Park

M-R

, Lee

G-H

. Preparation of high quality ITO films on a plastic substrate using RF magnetron sputtering. Surface and Coatings Technology, 2006; 201(3–4):927–931; doi: 10.1016/j.surfcoat.2006.01.004

36.

Yang

C-h

, Lee

S-C

, Lin

T-C

, et al. Electrical and optical properties of indium tin oxide films prepared on plastic substrates by radio frequency magnetron sputtering. Thin Solid Films, 2008; 516(8):1984–1991; doi: 10.1016/j.tsf.2007.05.093

37.

Kim

D-H

, Park

M-R

, Lee

H-J

, et al. Thickness dependence of electrical properties of ITO film deposited on a plastic substrate by RF magnetron sputtering. Appl Surfe Sci, 2006; 253(2):409–411; doi: 10.1016/j.apsusc.2005.12.097

38.

, Diao

, Wang

, et al. Effects of heat treatment on properties of ITO films prepared by rf magnetron sputtering. Vacuum, 2004; 75(2):183–188; doi: 10.1016/j.vacuum.2004.01.081

39.

Eshaghi

, Graeli

. Optical and electrical properties of Indium Tin Oxide (ITO) nanostructured thin films deposited on polycarbonate substrates “thickness effect”. Optik, 2014; 125(3):1478–1481; doi: 10.1016/j.ijleo.2013.09.011

40.

Moraes

, Saito

, Leite

, et al. Optical, electrical and electrochemical evaluation of sputtered platinum counter electrodes for dye sensitized solar cells. Appl Surf Sci, 2016; 364:229–234; doi: 10.1016/j.apsusc.2015.12.114

41.

Hsiao

P-T

, Lu

M-D

, Tung

Y-L

, et al. Influence of hydrothermal pressure during crystallization on the structure and electron-conveying ability of TiO2 colloids for dye-sensitized solar cells. J Phys Chem C, 2010; 114(37):15625–15632; doi: 10.1021/jp1061013

42.

Poh

, Ahmad

, Ting

, et al. Performances of flexible dye‐sensitized solar cells fabricated with binder‐free nanostructure TiO 2. J Mater Sci: Mater Electron, 2021; 32(9):12031–12041; doi: 10.1007/s10854-021-05833-6

43.

Weerasinghe

, Sirimanne

, Franks

, et al. Low temperature chemically sintered nano-crystalline TiO2 electrodes for flexible dye-sensitized solar cells. J Photochemistry Photobiology A: Chemistry, 2010; 213(1):30–36; doi: 10.1016/j.jphotochem.2010.04.016

44.

, Zhao

, Wang

, et al. Great improvement of photoelectric property from co-sensitization of TiO2 electrodes with CdS quantum dots and dye N719 in dye-sensitized solar cells. Mater Rese Bulletin, 2013; 48(7):2566–2570; doi: 10.1016/j.materresbull.2013.03.009

45.

Yadav

, Ravishankar

, Pescetelli

, et al. Stability of dye-sensitized solar cells under extended thermal stress. Phys Chem Chem Phys, 2017; 19(33):22546–22554; doi: 10.1039/C7CP04598K

46.

Sandoval-Paz

, Ramírez-Bon

. Indium tin oxide films deposited on polyethylene naphthalate substrates by radio frequency magnetron sputtering. Thin Solid Films, 2009; 517(8):2596–2601; doi: 10.1016/j.tsf.2008.10.016

47.

Zouzelka

, Rathousky

. Photocatalytic abatement of NOx pollutants in the air using commercial functional coating with porous morphology. Applied Catalysis B: Environmental, 2017; 217:466–476; doi: 10.1016/j.apcatb.2017.06.009

48.

Slavcheva

, Topalov

, Ganske

, et al. Influence of sputtering pressure on surface structure and oxygen reduction reaction catalytic activity of thin platinum films. Electrochimica Acta, 2010; 55(28):8992–8997; doi: 10.1016/j.electacta.2010.08.047

49.

Zan

, Xiao

, Wen

. Synthesis of TiO 2 nanorice and their improved dye sensitized solar cells performance. J Mater Sci: Mater Electron, 2017; 28(5):4107–4113; doi: 10.1007/s10854-016-6030-z