Multidimensional nanoarchitectures for improved indoor light harvesting in dye-sensitized solar cells

Multidimensional nanoarchitectures for improved indoor light harvesting in dye-sensitized solar cells Javier Castillo-Seoane a , Lidia Contreras-Bernal a,b,* , Antonio J. Riquelme c,** , Samuel Fauvel c , Yann Kervella c , Jorge Gil-Rostra a , Gabriel Lozano a , Angel Barranco a , Renaud Demadrille c , Juan R. S´ anchez-Valencia a , Ana Borr´ as a a Nanotechnology on Surfaces and Plasma Lab, Institute of Materials Science of Seville, Spanish National Research Council, University of Seville, C. Am´ erico Vespucio 49, Seville, Spain b Dpto. Ingeniería y Ciencia de los Materiales y del Transporte, Escuela Polit´ ecnica Superior Universidad de Sevilla, c/ Virgen de ´ Africa 7, Seville, E-41011, Spain c Universit´ e Grenoble Alpes, CEA, CNRS, INP, IRIG-SyMMES, F-38000, Grenoble, France ARTICLE INFO Keywords: Dye-sensitized solar cells Indoor energy harvesters 1D-nanostructured electrodes Plasma and vacuum techniques Light scattering ABSTRACT Dye Sensitized Solar Cells (DSSCs) have recently gained renewed interest for their potential in indoor light harvesting and powering wireless devices. However, to fully exploit their potential, crucial aspects require further attention, in particular, the improvement of spectral compatibility and low-light harvesting mechanisms, as well as the development of efficient photoanodes through high-yield scalable methods. In this article, we propose the use of nanocomposite photoanodes integrating mesoporous TiO 2 nanoparticles, ITO nanotubes (NT), and anatase TiO 2 shells (ITO@TiO 2 NT) prepared by step-by-step method relying on mild temperature conditions and avoiding toxic precursors. These photoanodes outperform previous attempts to implement low-dimensional ITO and ITO@TiO 2 nanowires and nanotubes for outdoor light conversion, demonstrating a power conversion efficiency under low artificial light intensity of 24 % for at 0.014 mW cm −2 , a 166 % increase compared to the conventional architectures. Advanced microstructural, optical, and electrochemical characterizations have revealed that the strong scattering effect of the light in the visible range coupled with enhanced charge collection at low-intensity illumination are the essential mechanisms responsible for such enhanced energy conversion. Remarkably, our devices retain up to 90 % of the normal incidence efficiency even under glancing illumination, while conventional reference devices retain only 30 %. 1. Introduction Harnessing light energy in indoor environments holds great promise for advancing the Internet of Things (IoT) by powering low-consumption (microwatt level) and portable electronic devices [1,2]. Among the various photovoltaic approaches, dye-sensitized solar cells (DSSCs) have re-emerged as key players due to their affordability, robustness and durability, aesthetic appearance including multi-color and semitransparency [3–5], and direct integration with light management systems [6–8]. Such characteristics also make DSSC technology attractive for its implementation in building integrated photovoltaics and dim light harvesting [9,10]. In recent years, progress has been made to improve the power conversion efficiency (PCE) under low-intensity light and also under illumination from diffuse or glancing light sources [11–13]. This progress has been facilitated by the development of specialized photosensitizers designed to optimize spectral compatibility with artificial light sources [14,15]. In addition, a new generation of electrodes, including both photoanodes and counter-electrodes, has been introduced to improve charge transport within the redox electrolyte. This advance addresses the most limiting mechanism affecting the photovoltaic performance of dye-sensitized solar cells operating under low-intensity light conditions [16]. Under such conditions, the number of excited dye molecules is reduced, and therefore, the effect of the electron recombination with the oxidized species in both the electrolyte and/or the oxidized dye molecules (reaction mechanisms 5 and 6 in Scheme 1a) can seriously affect the photovoltaic (PV) response of the * Corresponding author. Nanotechnology on Surfaces and Plasma Lab, Institute of Materials Science of Seville, Spanish National Research Council, University of Seville, C. Am´ erico Vespucio 49, Seville, Spain. ** Corresponding author. E-mail addresses: [email protected] (L. Contreras-Bernal), [email protected] (A.J. Riquelme). Contents lists available at ScienceDirect Materials Today Energy journal homepage: www.journals.elsevier.com/materials-today-energy/ https://doi.org/10.1016/j.mtener.2025.101851 Received 26 November 2024; Received in revised form 17 January 2025; Accepted 22 February 2025 Materials Today Energy 49 (2025) 101851 Available online 24 February 2025 2468-6069/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/bync-nd/4.0/ ). device [17]. In other words, the kinetic competition of the dye regeneration by the electrolyte redox reaction and the charge recombination is more pronounced under indoor lighting than outdoor lighting conditions (see Scheme 1a) [18]. It is, therefore, necessary to regulate the number of redox species of the electrolyte to minimize the recombination processes at low-intensity conditions. This problem, together with the effect of light intensity and its corrosive nature, has led to the replacement of standard iodide/triiodide electrolytes with alternatives based on cobalt and copper. In this context, Freitag et al. have recently published an unprecedented PCE of 38 % under fluorescent lighting at 1000 lux (303.6 μ W cm −2 ) using a copper (II/I) electrolyte [19]. Achieving higher efficiency, therefore, depends on improving the absorption of the incident light and optimizing charge carrier separation and transport. These factors are closely linked to the nanostructure of the photoanode. Among the preferred photoanode materials, TiO 2 mesoporous oxide films have gained widespread use due to their potential in both indoor and outdoor applications. The internal surface area and porosity of the mesoporous film have a significant effect on dye chemisorption, which determines the photocurrent generation. In addition, the morphology and structure of the photoanode have a critical effect on the charge transfer kinetics. To overcome the problem of slow electron diffusion across TiO 2 nanoparticles boundaries, researchers have turned to one-dimensional (1D) nanostructured materials such as nanotubes (NT), nanowires (NW), and nanorods (NR) as a substitute for nanoparticles [20]. These 1D nanomaterials provide directional electron transport pathways and reduce charge recombination due to their low grain boundary defects [20–24]. In addition, one-dimensional arrays typically exhibit a more pronounced light-scattering effect compared to mesoporous layers, thereby increasing light-harvesting efficiency. To the best of our knowledge, these nanostructured working electrodes in DSSCs have been tested mainly in outdoor lighting environments (AM 1.5G) [6,20–23]. However, taking such properties into account, 1D nanoelectrodes are also promising candidates under low light intensities by increasing the number of excited dye molecules and thus reducing recombination mechanisms. One of the main limitations of 1D nanomaterials is their inherently lower surface area compared to mesoporous nanoparticles [25]. To overcome this drawback, we propose here an alternative structure for the photoelectrode of DSSCs that integrates one-dimensional nanostructures (nanotubes and core@shell nanotubes) and mesoporous TiO 2 layers. These nanotubes made of a transparent conductive oxide (TCO) - specifically indium-doped tin oxide (ITO) - are also combined with a thin conformal shell of highly texturized anatase TiO 2 . Core@shell nanostructures offer several advantages, by facilitating electron transport and suppressing electron recombination. In the case of ITO@TiO 2 nanostructures, the electric field created between the layers enhances the injection of electrons from the TiO 2 to the ITO during DSSC operation, reducing electron recombination from the TiO 2 to the electrolyte [26]. Although the implementation of low-dimensional semiconducting nanostructures (mostly ZnO and TiO 2 NW, NR, and NT) has been extensively reported for both dye-sensitized and organometallic halide perovskite solar cells, the heteroarchitecture proposed here has hardly been attempted so far [20,21], without any previous examples of its implementation in DSSCs. In this article, nanotubes are prepared using a fully scalable vacuum and plasma-based technology that is easily extendable to alternative combinations of TCOs and semiconducting oxides and is compatible with large area and roll-to-roll processing. The nanotubes are formed using a soft-template process that uses single crystalline organic nanowires as vacuum-processable 1D templates [27]. This method offers compatibility with a wide range of processable substrates, including organic and polymeric supports. It is carried out at mild temperatures, requires low power consumption, is solvent-free, and avoids corrosive precursors. The method has been successfully applied to the development of hybrid, organic, semiconducting, metallic, and piezoelectric nanowires, nanotubes, and nanotrees, with amorphous or highly texturized crystalline structures [20,27–30]. The optical and transport properties of such nanoarchitectures increase the PCE under low artificial light intensity of the resulting DSSC, reaching 24 % at 0.014 mWcm −2 , a 166 % increase over the corresponding mesoporous DSSC. Indeed, our 1D nanoelectrodes lead to DSSCs with higher PV response at indoor illumination sources than those DSSCs prepared using Scheme 1. (a) Simplified scheme of the electronic bands and main recombination mechanisms of dye sensitized solar cell. Solid and dashed arrows represent processes in favor or against the charge collection, respectively. 1: Excitation of electrons from the dye molecule (Dye*) by photons; 2: Electrons injection into conduction band (CB) of the mesoporous TiO 2 layer leaving the dye molecule in an oxidized state (Dye + ); 3: Dye regeneration by receiving electrons from reducing species (Red, I − ) of the electrolyte; 4: Oxidizing species (Ox, I − 3 ) regeneration of the electrolyte by receiving electrons from platinum counter-electrode; 5: Injected electrons into CB of TiO 2 layer back into the dye; 6: Injected electrons into CB of TiO 2 layer back into the electrolyte; 7: Radiative recombination from LUMO to HOMO of the dye. (b) Synthesis process of nanostructured photoelectrodes of ITO and ITO@TiO 2 nanotubes embedded in a m-TiO 2 scaffold. J. Castillo-Seoane et al. Materials Today Energy 49 (2025) 101851 2 commercial photoanode with a thicker mesoporous layer. Finally, we will analyze the role of the light scattering effects on the performance of the cells under artificial light operation at different angles of incidence. 2. Experimental section 2.1. Synthesis of core@shell nanoelectrodes ITO nanoelectrodes were fabricated through a vacuum and plasma multistep procedure involving vacuum and plasma techniques, as previously detailed by our group (See Scheme S1) [27]. The process comprised the following sequential steps: (1) Magnetron sputtering to deposit a “seed” layer, utilizing an ITO thin film; (2) Vacuum thermal sublimation to grow metal-free phthalocyanine H 2 Pc single crystalline organic nanowires (working as soft-template) on the ITO thin film; (3) Magnetron sputtering for the deposition of a conformal ITO shell on the organic nanowires; and (4) Air annealing to eliminate the organic core, resulting in ITO-supported nanotubes. Throughout these steps, the fabrication process adhered to mild conditions, maintaining a pressure between 0.5-2.0 ×10 −2 mbar. Substrate temperatures of 350 ◦C were applied for steps (1), (3), and (4), while a temperature of 210 ◦C was employed for step (2). ITO deposition involved radiofrequency magnetron sputtering of a 3” ITO disc target (SnO 2 10 wt% doped In 2 O 3 , Kurt J. Lesker Company), using Ar as a plasma gas, a power of 150 W, and a distance to the samples-holder of 15 cm. Subsequently, a TiO 2 conformal shell was deposited at room temperature using plasma-enhanced chemical vapor deposition (PECVD) with a 2.45 GHz microwave Electron-Cyclotron Resonance (ECR) SLAN-II plasma source. The applied power was 480 W, and O 2 served as the plasma gas at a pressure of 1.0 ×10 −2 mbar. The organometallic precursor for this deposition was titanium (IV) isopropoxide (TTiP, Sigma Aldrich), at 40 ◦C with an O 2 bubbler as carrier gas. As a result, supported ITO@TiO 2 nanotubes. 2.2. Fabrication of dye-sensitized solar cells FTO TEC 15 drilled (Solaronix) and ITO (XopGlass, resistance <10 Ω/square; thickness 1.1 mm) have been used as substrates for counterelectrode and working electrode, respectively. Both types of substrates were cleaned with 15 min bath sonication using, successively, Hellmanex© solution, deionized water, acetone and isopropanol. The substrates were then treated in a UV-O 3 chamber for 15 min. After that, a platinum solution (Solaronix, Platisol) was spread over the conductive side of the counter-electrodes for subsequent annealing at 400 ◦C for 5 min. On the other hand, 8 μ m thickness of TiO 2 colloidal paste (20 nm, GreatCell) was deposited on ITO substrates by screen-printing technique (PET 1500 90/230-40 mesh) for preparing the reference electrodes. The resulting TiO 2 layer (referred to as m-TiO 2 ) was sintered at 450 ◦C for 30 min. In the case of 1D nanostructured working electrodes, the m-TiO 2 film was deposited in the same conditions. The active area of electrodes was 0.25 cm 2 and thickness was around 8 μ m. The mesoporous layer was treated with TiCl 4 bath (40 mM aqueous solution) for 20 min at 70 ◦C. The electrodes were then rinsed with distilled water and absolute ethanol. A mixture of acetonitrile and tert-butanol (1:1 % V) containing 0.2 mM of YKP-88 and 2 mM of chenodeoxycholic acid was used as dye solution. The working electrodes were immersed into the dye solution overnight in dark and at room temperature under mild agitation. After dye sensitization, the samples were washed with ethanol and dried under air. The liquid-based electrolyte solution was prepared in acetonitrile using iodide/iodine as redox pair: 0.03 M I 2 (Sigma-Aldrich), 1M BMII (Solvionic), 0.05 M LiI (Sigma-Aldrich), 0.5 M TBP (Sigma-Aldrich) and 0.1 M GuSCN (Sigma-Aldrich) [31]. Later, the working electrode and counter-electrode were sandwiched together using a thin thermoplastic film (Surlyn, 60 μ m) at 105 ◦C under pressure. The assembled device was filled with the electrolyte solution through the hole by vacuum. Finally, the hole was sealed with a piece of thermoplastic film and a coverglass. 2.3. Characterization The morphological characterization was carried out by SEM using a Hitachi S4800 microscope at 2 kV. The SEM images were treated with ImageJ (free available software) to measure the length and diameter of nanostructures. TEM images, SAED pattern and EDXS profiles were obtained using a scanning TEM microscope, TALOS F200S from FEI company, working at 200 kV with 0.25 nm resolution. To perform these measurements, the samples were scratched on top of a holey carbon TEM copper grid. For XRD measurements, a Panalytical X’PERT PRO model operating in the θ-2θ configuration with Cu K α (wavelength of 1.5418 Å) radiation as the excitation source was used. The UV–vis–NIR spectrophotometer used was an Agilent Technologies Cary 5000 Uv–Vis–NIR equipped with an integrating sphere to acquire absorptance A, diffuse and total transmittance (T D and T T ), diffuse and total reflectance (R D and R T ). The electrical resistance values have been calculated from the slope of the sample’s I-V response before and after calcination in air at 450 ◦C. Electrical measurements were taken using a 4-point probe setup with the Ossila Four-Point Probe equipment, placing the 4 probes at five different locations on the sample surface and performing a voltage sweep from -1V to 1V. Samples prepared on fused silica substrates were used to evaluate the conductivity of these ITO nanostructures. Dye loading has been carried by immersing the photoanodes in a solution of 0.1M aqueous NaOH and THF in a 1:1 %V for 3 h. After detachment, the dyes are extracted with dichloromethane and the analysis is performed by UV–vis spectroscopy. The photovoltaic response of DSSCs was studied by recording the current density-voltage curves with a Keithley 2400 at scan rate of 100 mV s −1 in reverse scan and using a mask of 0.36 cm 2 to delimit the area of the solar cell. All devices, both reference and nanostructured, were characterized following the same procedure. This approach ensures a reliable comparison of the obtained results, considering that the parameters used to record the current density-voltage curves may lead to an overestimation of the photovoltaic performance [32,33]. For that, different light sources were used: 1) solar simulator (ABET-Sunlite) with AM 1.5G filter at 100 mW⋅cm−2; 2) a cool-white laboratory LED of 6500K (lamp 1); 3) warm-white LEDs of 2700K with different luminous fluxes (806 lm, 60W (Philips) and 400 lm, 40W (ledare IKEA) (lamp 2 y lamp 3, respectively), and a cool-white compact fluorescent lamp of 6400 K, 18W and 1010 lm (ExtraStar) (lamp 4). The light intensity was measured using a Solar Light’s Model PMA2100 Dual-Input Data Logging Radiometer with a visible light photometer detector. The lux meter LM-3000 has been used for an approximate estimation of the illuminance level. The photovoltaic response under indoor light illumination has been obtained in accordance with the indications given in the reference, 34 i.e., the light source was stabilized for more than 30 min, and the characterization carried out inside a black box to avoid any external light. It should be noted that the photovoltaic efficiency was calculated by taking the light intensity recorded by the radiometer. For current density-voltage curves at different angles, we used a holder that allows the sample to be tilted with respect to the light source. Impedance spectroscopy (IS) was carried out at open circuit conditions using a white 6500K LED (lamp 1) as the illumination source. The Fermi level was fixed by the illumination intensity. The IS was measured in the range from 10 6 to 10 −1 Hz with a perturbation of 20 mV. The frequency response of the DSSCs was analyzed by a response analyzer module (Autolab, PGSTAT302N/- FRA2). The IS data was generated by NOVA 2.1.3 software and then analyzed using Z-view equivalent circuit modelling software. 3. Results and discussion 3.1. Fabrication of core@shell nanostructures. Microstructural and optical characterization The synthesis of the nanocomposite photoelectrodes was approached through a step-by-step procedure based on the combination of organic J. Castillo-Seoane et al. Materials Today Energy 49 (2025) 101851 3 nanowire (ONW) soft-template techniques (see Scheme S1 in Supporting Information) enabled by the vacuum and plasma one-reactor deposition of functional shells and screen-printing of mesoporous nanoparticles (Scheme 1b). The main steps of the method can be described as: 1) fabrication of the ITO thin film (~150 nm) “seed” layer by magnetron sputtering on a commercially available ITO-glass substrate to increase the roughness, thus the density of nucleation sites on the substrate; 2) vacuum thermal sublimation of metal free phthalocyanine organic nanowires on the as-grown ITO surface; 3) conformal deposition of a conductive ITO shell on the organic nanowires used as template; 4) organic core removal by annealing at 350 ◦C under air conditions to form the ITO NTs array. After the annealing step, the NT appear partially vertically aligned with lengths between 3.0 μ m and 7.0 μ m, with an average diameter of 142 ±19 nm (see Fig. 1a and c). Fig. S1, in Supporting Information, shows the statistical analysis of the NT diameters. For the synthesis of core@shell NT, a TiO 2 film is grown on the ITO NT by plasma-enhanced chemical vapor deposition (PECVD), as shown in route B of Scheme 1b [20,30,35]. The experimental conditions reported in Ref. [36] (see also Experimental section) were chosen to ensure the formation of the anatase phase at relatively mild temperatures (below 300 ◦C). The SEM images in Fig. 1b and d presents the core@shell ITO@TiO 2 with an average diameter of 330 nm (i.e., the thickness of the TiO 2 shell is ~95 nm, see also Fig. S1). The thickness of the shell is confirmed by the TEM image as well as by the EDX profile of a single ITO@TiO 2 NT (Fig. 2a and c). These results also support the high conformality of the TiO 2 shell growth, which exhibits a columnar structure characteristic of highly textured anatase layers [20]. The interplanar distances of the TiO 2 shell crystals are consistent with the anatase phase (see Fig. 2b). These results agree with the XRD analysis carried out on ITO and ITO@TiO 2 NT samples. In both cases, the formation of the ITO crystalline structure is evidenced by the presence of (2 1 1), (2 2 2) and (4 0 0) diffraction peaks (blue squares in Fig. 2d). Moreover, the main anatase diffraction peak at 25.4◦(orange triangle) is also observed for the ITO@TiO 2 samples despite the low shell thickness and the lower diffraction capacity of TiO 2 compared to ITO. To facilitate the observation of this small peak, a zoom of the corresponding area is shown as an inset in the lower left of Fig. 2d). As mentioned above, the 1D nanoelectrodes have been combined with TiO 2 nanoparticles (m-TiO 2 ) to form hetero-structured photoelectrodes. Thus, the final step in the synthesis of the nanocomposite photoelectrodes was the screen-printing deposition of TiO 2 nanoparticles (20 nm of nanoparticle size) (routes A and B of Scheme 1b for ITO NT and ITO@TiO 2 NT, respectively). Specifically, a total thickness of 8 μ m was deposited to ensure a homogeneous coverage of the 1D nanoelectrodes. The cross-sectional SEM images in Fig. 1e and f shows the complete coverage of the 1D nanostructures (ITO and ITO@TiO 2 NT in Fig. 1e and f, respectively) embedded in the m-TiO 2 layer which forms a matrix occupying all the free space between the nanotubes with a slight decrease in the density of the nanoparticles at the interface with the substrate. Fig. S2 shows a detailed cross-section view obtained by Focused Ion Beam (FIB), where it is possible to observe the close contact between the m-TiO 2 nanoparticles and the supported nanotubes. Note that the XRD pattern corresponding to the ITO@TiO 2 NT/mTiO 2 sample (blue line in Fig. 2d), although dominated by the (101) peak of the anatase phase due to the m-TiO 2 , shows the peaks corresponding to the ITO. It is also worth mentioning that the embedding in the m-TiO 2 nanoparticles does not reduce the ITO conductivity. Thus, Fig. S2 includes the comparison of the UV–vis–NIR transmittance spectra and XRD diffractograms for the ITO NT sample before as deposited and after the annealing in air at 450 ◦C. These results indicate that although the transmittance in the NIR range is slightly higher for the annealed sample, which might be related to a decrease in conductivity, the crystalline structure is not affected by the annealing treatment. Moreover, the fourprobe conductivity measurement carried out shows a reduction in the resistance after the annealing (R =79 ±8 Ω for as-deposited, compared with R =67.9 ±0.4 Ω after annealing). We hypothesize that such reduction is related to improving the interface adhesion between the nanotubes and the ITO seed layer employed to form the ONW soft templates and compensates for the likely oxidation of the nanostructures under the annealing treatment inherent to the fabrication of the mesoporous TiO 2 . Fig. 1. Normal-view SEM images of (a) ITO and (b) ITO@TiO 2 nanotubes. Cross-sectional SEM images of ITO NTs (c), ITO@TiO 2 NTs (d), ITO NTs +mTiO 2 (e) and ITO@TiO 2 NTs +m-TiO 2 (f). J. Castillo-Seoane et al. Materials Today Energy 49 (2025) 101851 4 Fig. S3 summarizes the optical analysis of the ITO NT (a), ITO@TiO 2 NT (b), and ITO@TiO 2 NT/m-TiO 2 (c) samples, including transmittance, reflectance, and absorptance in the UV–vis–NIR region for total and diffuse components. Focusing on the visible range (400 nm <λ <700 nm), significant light scattering is observed even for the sample with infused nanoparticles. Indeed, the diffuse component coincides with the total transmittance in a large part of the visible range for all three cases. This coincidence of the diffuse and total transmittance leads to high Haze factors due to strong scattering effects in the visible range, as shown in Fig. 3 [37]. It is worth noting that the diffuse contribution of heterostructured systems (i.e. ITO@TiO 2 NT ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 ) is significantly more pronounced compared to supported ITO nanotubes. Additionally, this contribution is greater than that of photoanodes without one-dimensional nanostructures (m-TiO 2 reference sample), and is nearly null for ITO TF sample. In addition, the total reflectance of the ITO NT and ITO@TiO 2 NT is almost identical to the diffuse component in the visible region. To further demonstrate the strong dispersive character of the nanostructured electrodes, the angular distribution of the light transmittance and reflectance components of an ITO NT sample is also compared with that of ITO thin film (ITO TF) sample (Fig. S4). The results in Fig. S4d show a decrease in the ballistic transmittance (180◦) as shorter the wavelength, which is being reduced to approximately 15 % (λ =470 nm) for the ITO NT sample. The ballistic component of transmitted light for the ITO TF (see Fig. S4b) is very high (ca. 90 % for the three wavelengths. A zoom-in of these polar representations (Figs. S4c and e) reveals that while in the ITO TF, the light is exclusively transmitted at a ballistic angle very close to 180◦, the ITO NT sample presents a highly diffuse light pattern with the typical scattering distribution as a function of the measurement angle. The effect of such scattering effect on the performance of DSSCs will be analyzed in the following sections, considering that most of the dyes for the DSSCs have the maximum light absorption at short wavelengths in the visible region of the electromagnetic spectrum [38]. 3.2. Nanotubes of ITO and ITO@TiO 2 for dye sensitized solar cells Transparent Indium Tin Oxide electrodes are one of the most widely used in third-generation solar devices such as perovskite, organic, or dye-sensitized solar cells [39,40]. As mentioned above, introducing 1D nanostructured electrodes into photovoltaic devices offers improved properties and functionalities. This section describes the impact of the nanocomposite electrodes as photoanodes in DSSCs, with particular emphasis on indoor and low-light performance. We have focused on two photoelectrodes: the ITO NT and the ITO@TiO 2 NT both embedded in the m-TiO 2 layer to enhance dye absorption (hereafter referred to as ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 , respectively). These electrodes correspond to the micrographs in Fig. 1e and f. To evaluate the impact of the nanostructured electrodes on the PV performance, we have also used commercially available ITO glass substrates coated with the same anatase TiO 2 mesoporous framework as a reference. The nanostructured electrodes (and the reference samples) were sensitized with the organic dye YKP-88, using a co-absorbent (chenodeoxycholic acid) in a ratio of 1:1 ratio. YKP-88 was chosen on its excellent performance and stability (chemical structure in Fig. S5) [41]. Fig. S6 shows the absorptance spectra of ITO@TiO 2 NT/m-TiO 2 electrode and reference samples sensitized with the dye, where it can be noted that the corresponding to Fig. 2. (a) Representative TEM micrograph of ITO@TiO 2 nanotube. (b) HRTEM image of TiO 2 shell. (c) EDXS profile of the ITO@TiO 2 nanotube cross-section. (d) XRD indexed diffractograms of ITO and ITO@TiO 2 nanotubes, and ITO@- TiO 2 nanotubes embedded into a mesoporous scaffold of TiO 2 . Fig. 3. Haze factor obtained from the UV–Vis–NIR transmittance spectra for ITO thin film, ITO and ITO@TiO 2 nanotubes, ITO NT/m-TiO 2 , ITO@TiO 2 NT/ m-TiO 2 , and reference m-TiO 2 samples. J. Castillo-Seoane et al. Materials Today Energy 49 (2025) 101851 5 the nanostructured photoanode presents slightly stronger absorptance in all the visible range. Although in the range below 600 nm the increase in the absorptance is not very relevant (see zoom in the corresponding range in Fig. S6), this is because both spectra appear saturated (near 100 % absorptance) and changes related with scattering phenomena are not appreciable through this characterization. The photon diffusion length (L d ) of the layer decreases with scattering, especially in highly absorbing media [42,43], leading to light trapping effects and enhancing solar device performance. While the dye loading is similar in both systems (see Table S1), the nanostructured layers exhibit significantly higher scattering (Fig. 3 and S3), shortening L d and increasing photon absorption by dye molecules through multiple scattering. This indicated that a greater number of dye molecules are excited (reaction mechanism 1 in Scheme 1a) when the nanostructures are integrated into the photoanode. Similar enhanced dye absorptances have been previously reported due to the enhanced optical path in highly scattering media [44–46]. To complete the DSSC, we used a platinum counter electrode on FTO glass (fluorine-doped tin oxide) and an electrolyte based on the iodide/ tri-iodide redox pair in acetonitrile as previously used [31]. It is essential to emphasize here that the main objective of this study is to highlight the advantages of the hierarchical nanostructured photoanodes and to explore the mechanisms underlying its enhanced performance under low-intensity light. Accordingly, the choice of electrolyte and counter electrode was based on the procedures established within our group for fabricating DSSCs that provide reliable comparison with standard reference cells. We anticipate that the combination of these hierarchical photoanodes with optimized electrolytes, counter electrodes, and dyes for indoor light harvesting will significantly improve the overall efficiency of the cells [47,48]. 3.3. DSSCs based on ITO and ITO@TiO 2 NT for indoor lighting DSSC is one of the most promising technologies for wireless powering the IoT and wireless sensor networks (WSN). One of the main reasons for this is its good performance in low light-intensity conditions due to its high open-circuit voltage for diffuse and low-intensity light conditions, as well as its wide angular compared to other PV technologies [11, 32,47,48]. This section describes the effect of ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 electrodes on the PV performance of DSSCs operating in indoor environments with artificial light. We also analyze the effect of the angle of incidence of light on the PCE of these devices. It is worth noting that, for this study, we have obtained the current density-voltage curves (JV curves) of DSSCs over a wide range of low light intensities. Please note that the first standard test for indoor cell characterization, IEC TS 62607-7-2:2023, has been recently reported [49]. However, this standard is still in its early implementation; therefore, benchmarking PV performance is not straightforward. Herein, we have evaluated the performance of the cells under real scenarios, with such an objective, we have adhered to recommended light intensities for offices or schools (~0.1 mW cm −2 (>500 lux)) as well as home lighting (0.04–0.014 mW cm −2 (250-100 lux, respectively)) for consistency (see Fig. S5 for the emission spectra of the tested light sources) [50–53]. Fig. 4a shows the JV curves of the champion DSSCs obtained under this variety of illumination intensities (0.1, 0.04 and 0.014 mW cm −2 , corresponding to 650, 230 and 98 lux, respectively) from a white LED of 6500K (hereafter called Lamp 1, see spectra in Fig. S5) while Table 1, Fig. 4b–and S7 show the corresponding PV parameters statistics of each type of DSSCs and illumination intensity using Lamp 1. An increase in the PV efficiency of the 1D nanostructure-based DSSCs is clearly observed as the light intensity decreases (see Fig. 4b). Here, the TiO 2 shell provides better open-circuit voltage (V OC ) than the ITO NT/mTiO 2 but at the cost of losing reduced short-circuit photocurrent (J SC ). Indeed, at 0.014 mW cm −2 the highest PCE is obtained for the ITO NT/ m-TiO 2 -based DSSC, reaching 22.6 %, followed by the ITO@TiO 2 NT/mFig. 4. Photovoltaic parameters of DSSCs based on ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 , and reference photoanodes under indoor light illumination. (a) Current density-voltage curves of the champion cells of DSSCs measured at different light intensities (red, green, and blue lines correspond to reference, ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 , respectively). (b) Photovoltaic efficiency statistic data obtained from the current density-voltage curves at different light intensities. (c) Normalized efficiency obtained from the current density-voltage curves that were recorded at different angles of incidence of light at 0.014 mW cm −2 . (a–c) These data have been acquired under artificial illumination sources using a cool-white LED (Lamp 1), following the indications reported in Ref. [34]. The measure has been carried out at reverse scan rate, at 100 mV s −1 , and using a mask of 0.36 cm 2 , slightly larger than the active area of the device, following good practices of JV characterization of DSSCs [54]. J. Castillo-Seoane et al. Materials Today Energy 49 (2025) 101851 6 TiO 2 and the reference samples (efficiencies of 15.2 % and 8.5 %, respectively). These results are in line with those published by Cao et al. These authors showed an enhancement of the PV response at low light intensities for DSSC by regulating the electron recombination rate with oxidized species by decreasing the iodine content in the electrolyte [55]. Our results also align with the work published by Sasidharan et al. for DSSCs fabricated by combining conventional m-TiO 2 with ZnO microflower. Such a nanostructured system achieved an 85 % increase in PCE under indoor illumination (1000 lux) due to improved charge collection efficiency resulting from shorter transport time and longer diffusion length [56]. In the case of our 1D nanostructure-based DSSCs, the increased number of excited dye molecules could lead to a lower electron recombination rate with oxidized species in both the electrolyte and/or the oxidized dye molecules. In fact, the positive effect of 1D ITO nanostructures becomes even more evident when comparing their PV performance with that of a DSSC containing a commercial photoanode (from Solaronix) (Fig. 4b). It is noteworthy that although the Solaronix sample contains a thicker mesoporous layer of TiO 2 on FTO glass (13 μ m) and FTO as substrate, it does not exceed the efficiency of ITO NT/m-TiO 2 - based DSSC in the light intensity range of 0.04–0.014 mW cm −2 . Table 2 compares the performance of DSSC based on ITO NT/m-TiO 2 with reported works in the literature for DSSC working under low artificial light intensity. It should be emphasized that this represents an approximate comparison of photovoltaic parameters, as there is no consensus regarding the dye and counter electrode used, the scan rate, or the recording direction of the JV curves. For this reason, Table 2 lists DSSC prepared with electrolyte based on iodide/tri-iodide redox pair as a point of similarity with the devices analyzed in this work and under 1000 lux of light intensity (the best efficiency published to date under low illumination conditions are also shown). Based on this comparison, ITO NT/m-TiO 2 -based DSSCs are positioned at average values of efficiency, slightly lower values for V OC and FF (likely due to the use of ITO as substrate), and it highlights the substantial increase of J SC . Fig. 4c shows the normalized efficiencies obtained from JV curves measured at different illumination angles, from normal to glancing (~85◦), using the indoor Lamp 1. Fig. S8 shows the angular dependence of J SC , V OC , and FF. From this experiment, it can be seen that the efficiency of the reference samples decreases to 30 % when comparing normal incidence (note that Fig. 4c is normalized to this PCE at 0◦) with an almost perpendicular incidence illumination with respect to the light source. Surprisingly, ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 -based DSSCs retain 90 % and 80 % of efficiency, respectively, under the same glancing illumination conditions. This behavior is mainly attributed to the reduced angular dependence of the photocurrent in the nanostructured DSSCs (Fig. S8). This result highlights the positive impact of the scattering phenomenon from ITO and ITO@TiO 2 NT in the photoanodes on the PV performance of DSSCs operating at low-light intensities and under realistic ambient diffuse indoor illumination. That means, the reference sample exhibits the most significant decrease, primarily due to its lower diffusive component of transmittance compared to that exhibited by the nanostructured photoanodes (see Fig. 3). The effect of different interior lamps on the PCE of the DSSCs is also studied in Fig. S9. Specifically, we have recorded the JV curves of the DSSCs using a cool-white fluorescent lamp (Lamp 2) and warm-white 2700K LEDs with different luminous fluxes (Lamp 3 and Lamp 4) as illumination sources (see the emission spectra in Fig. S5). By setting the distance to the samples to adjust the light intensity to 0.04 mW cm −2 for all the illumination sources, the PV efficiency of all the samples remains almost unchanged despite the spectral irradiance difference between the sources. This is due to the absorption spectrum of the YKP-88 dye being well aligned with the emission of standard indoor incandescent lamps, such as those used in this work (Fig. S10), causing the number of excited dye molecules generated to be similar regardless of the indoor lamp used. Table 1 Photovoltaic parameters statistic of DSSCs based on ITO NT and ITO@TiO 2 NT embedded in a mesoporous scaffold of TiO 2 , and reference samples obtained from JV curves recorded under indoor light illumination. The photovoltaic parameters have been extracted from current-voltage curves measured at different light intensities (using Lamp 1) in reverse scan (from 1V to −0.2 V) using a mask of 0.36 cm 2 and scan rate of 100 mV s −1 . Reference ITO NT /m-TiO 2 ITO@TiO 2 NT /m-TiO 2 Reference ITO NT /m-TiO 2 ITO@TiO 2 NT /m-TiO 2 Reference ITO NT /m-TiO 2 ITO@TiO 2 NT /m-TiO 2 0.1 mW cm −2 (650 lux) 0.04 mW cm −2 (230 lux) 0.014 mW cm −2 (98 lux) J sc ( μ A⋅cm ¡2 )38.3 ±15.8 47.6 ±1.4 35.8 ±5.7 15.8 ±6.7 27.8 ±2.3 19.6 ±5.6 5.6 ±2.2 17.6 ±1.5 11.8 ±3.4 V oc (mV) 515 ±45 364 ±4 411 ±51 460 ±55 337 ±13 358 ±74 392 ±53 286 ±21 318.9 ±53 FF (%) 61 ±3 67 ±1 67 ±4 54 ±1 58 ±5 57 ±3 56 ±6 63 ±5 58 ±11 PCE (%) 11.9 ±4.4 11.4 ±0.5 9.7 ±1.1 9.5 ±2.9 13.5 ±0.5 9.8 ±2.3 8.5 ±1.3 22.6 ±1.7 15.2 ±3.9 Table 2 Comparison of the performance and experimental conditions at low light intensity of the devices reported in this work with different examples in the literature. Working electrode Sensitizer electrolytesolvent Counterelectrode Light source Light intensity ( μ w⋅cm −2 /Lux) V OC (v) J SC ( μ A⋅cm −2 ) FF (%) PCE (%) Year Ref FTO/Yb-doped TiO 2 NP þAu NPs N719 I − /I 3 − - Acetonitrile Pt White LED 327/1000 0.59 153 50 14 2023 [57] FTO/TiO 2 -ZF N719 I − /I 3 − - Acetonitrile Pt Fluorescent light -/1000 0.545 68.86 70 12 2020 [56] FTO/mTiO 2 (14 μ m) N719 I − /I 3 − - 3-MPN PVPPt Solar simulator AM1.5 Class A with neutral density filter 500/- 0.65 1210 77 12 2012 [55] FTO/m-TiO2 (14 μ m) CCOD-2 I − /I 3 − - Acetonitrile Pt Fluorescent light -/2500 0.692 424 76 28 2022 [58] FTO/m-TiO 2 (14 μ m) TPATThC 6 I − /I 3 − - DESs Pt Fluorescent light 370/1200 0.523 79 72 8 2024 [59] ITO NT/m-TiO 2 (8 μ m) YKP-88 I − /I 3 − - Acetonitrile Pt White LED 177/1300 0.532 690 62 13 –This work FTO/mTiO 2 (8 μ m) XY1:L1 Cu(tmby) 2 PEDOT Fluorescent light 303.6/1000 0.995 147 78 38 2023 [19] J. Castillo-Seoane et al. Materials Today Energy 49 (2025) 101851 7 3.4. Charge collection and light harvesting mechanisms in multidimensional nanostructured photoelectrodes As a first step to unravel the main factors involved in the enhanced performance of the nanostructured photoelectrodes, we evaluated the PV performance under a solar simulator at 1-sun of light intensity. Fig. S11 shows the best JV curves for each configuration. The average of PV parameters is summarized in Table S1. As can be seen, the highest efficiencies are obtained for the reference samples, closely followed by the ITO nanostructure samples. It should be noted that the efficiencies obtained for our reference align with those reported in the literature despite the use of ITO as a substrate (see Table 2) [55–57]. The decrease in efficiency for the 1D nanostructures is mainly due to the reduced J SC , although a slight decrease in the V OC is also observed. Focusing on the 1D nanostructure-based DSSCs, the anatase shell in ITO@TiO 2 NT/m-TiO 2 has a positive effect on the V OC and the fill factor (FF) but at the cost of losing the J SC . The positive effect of the shell can be explained by a lower rate of electronic recombination and/or conduction band downshift [60,61]. Thus, from highest to lowest efficiency, the devices are ordered as follows: reference >ITO@TiO 2 NT/m-TiO 2 >ITO NT/m-TiO 2 . It should be emphasized that the 1D nanostructures barely influence the dye loading at the photoanode (see Table S1), which is mainly dominated by the thickness of the m-TiO 2 layer and is slightly higher for the reference device. The electrochemical analysis of the different configurations was carried out using impedance spectroscopy (IS). IS curves were recorded at open-circuit potential over a wide range of light intensities using a white LED as the light source (Lamp 1). This allows the devices to be probed at different positions of the Fermi level in the m-TiO 2 layer. It is important to stress that the range of light intensities employed included 0.1, 0.04, and 0.014 mW cm −2 , matching the intensities used for the JV curves presented in Fig. 4. For DSSCs based on liquid electrolytes, the impedance curve is usually characterized by two signals: the one closest to the origin, the high-frequency (HF) signal, is due to charge transfer at the platinum counter-electrode [62–64]. The second signal is the result of the parallel combination of (1) the charge transfer or recombination resistance (R rec ) and (2) the chemical capacitance of electron accumulation in the semiconductor film (in this case the m-TiO 2 layer) (C μ ) [65–67]. These signals appear as arcs in the Nyquist plot (imaginary part vs. real part) (see Figs. S12a and b). Both signals are connected by a small 45◦transmission line, which is related to the transport resistance (R trans ) in the metal oxide and is more relevant at low light intensities (Fig. S12b) [66]. Here, the impedance spectra were fitted using the transmission line equivalent circuit developed by Bisquert and coworkers (Fig. S12c) [68,69]. Fitting using this model allows for the extraction of the transport and recombination resistances. Paying attention to the parameters R rec and C μ , these evolve with the applied bias according to equations [70,71]. Rrec ∼exp(−βqV kBT)(1) C μ ∼exp( α qV kBT)(2) where V is the applied voltage, k B is the Boltzmann constant, q is the electron unit charge, T is the absolute temperature, β is the transfer parameter, and α describes the exponential distribution of intra-band states below the TiO 2 conduction band [72–74]. Regarding the voltage dependence of C μ (Fig. S13), all the structures show the characteristic voltage dependence of DSSCs, where the exponential tendency of C μ (Table S2) is consistent with that reported in the literature for mesoporous frameworks ( α ~0.15–0.35) [31,60,75]. Upon comparing the exponential dependence region of the three devices, a band shift of ~50 mV can be observed for the ITO@TiO 2 NT/m-TiO 2 device [56,60,76]. Therefore, a voltage correction is needed to ensure that the comparison of the parameters of the devices is made with the same electron density in the semiconductor [69,73,77]. This can be calculated from the chemical capacitance depicted in Fig. S13 using this numerical expression [77]. n(VOC)=kBT α qC0exp( α qVOC kBT)=kBT α qC μ (3) This is shown in Fig. 5, which depicts the relevant parameters extracted from impedance as a function of the total charge density. The recombination resistance shows the exponential dependence on the open-circuit voltage described in Equation (1). Table S2 reports the β obtained for each configuration according to Equation (1). In all cases, the β parameter is within the typical values for DSSC (β ~0.5–0.8) regardless of the nanostructured photoanode. It is also observed that the recombination resistance follows the trend: reference >ITO@TiO 2 NT/ m-TiO 2 >ITO NT/m-TiO 2 , which is in good agreement with the V oc reported in Table S1. Finally, focusing on the transport resistance in Fig. 5b, both nanostructured electrodes show a plateau at low light intensities (i.e., at low values of total charge density) rather than the classic exponential behavior of DSSCs (see the reference device) [68]. It cannot be excluded that this abnormal behavior can be related to numerical inaccuracy of the fit, given the short length of the transport line in the nanostructured devices spectra. Thus, this transport resistance must be considered as an estimation. However, it should be noted that the presence of the aforementioned transport line in the mesoporous device and its absence in the nanostructured devices is tangible evidence of the improved transport properties of these nanostructures. Also, the DSSC based on ITO NT/m-TiO 2 electrode has the lowest transport resistance at the same total charge density values. By combining the estimated transport and recombination resistances, the ratio of the small-perturbation diffusion length (the average distance that an electronic carrier can travel before recombining) to the thickness of the active layer (Ln/d) can be estimated to provide a qualitative assessment of the charge collection [69]. This ratio can be calculated, according to the IS model proposed by Bisquert and coworkers, as follows Ln d= Rrec Rtrans √(4) When the recombination resistance is much larger than the transport resistance, this ratio has values much larger than 1, which implies good electron collection. On the other hand, when transport resistance dominates over the recombination resistance, this ratio can reach values lower than 1, implying poor charge collection. Thus, this ratio can be used as a qualitative tool to compare the charge collection efficiency of different devices under various conditions. As shown in Fig. 5c, this ratio increases sharply for the 1D nanostructured samples at low charge conditions. That is, the DSSCs fabricated with the ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 photoanodes collect the charge more efficiently than the reference sample under low light-intensity conditions, which is in line with the JV curves recorded under indoor light illumination (Fig. 4). In contrast, at high light intensities (close to V OC under 1-sun illumination), the behavior reverses: the reference DSSC shows a higher charge collection efficiency compared to those based on ITO NT/ m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 . This difference accounts for the lower J SC observed in 1D nanostructure-based DSSCs compared to the reference in the JV curves recorded under 1-sun illumination (Fig. S11). It is worth emphasizing that the power conversion efficiency results from a highly complex interplay between photon absorption, electronhole pair generation, and the transport and collection of these charges. In fact, the development of models that allow simultaneous study of the optical and transport behavior of photovoltaic devices represents a significant challenge in the field [78,79]. This is particularly relevant in complex systems integrating nanostructures like those developed herein. Nonetheless, the photoanodes characterization in J. Castillo-Seoane et al. Materials Today Energy 49 (2025) 101851 8 Fig. 3, S3 and S4 proves that the incorporation of nanostructures improves the optical performance of the system by allowing an enhanced light trapping and wide angular response for improved photon absorption. At the same time, the transport models in Fig. 5 qualitatively indicate that the nanostructures enhance charge collection under low illumination conditions. Considering these results, the enhanced efficiency encountered for the multiscale nanostructured photoanodes at low-light intensity can be explained as follows: Firstly, considering that ITO nanotubes-based devices show higher J SC at low illumination intensity (see Fig. S7b and Table 1) compared to the reference. Secondly, a smoother decrease of the photocurrent density with decreasing light intensity is also observed in the case of nanostructured photoanodes (Fig. S7b). Thirdly, we also found that the ITO NT/m-TiO 2 samples maintain a good FF value over the whole illumination range (Figure S7 c). The increase in the J SC for DSSCs based on 1D nanostructures at low-light intensity is explained by, on the one hand, the strong scattering effect observed for the ITO NT and ITO@TiO 2 NT shown in Fig. 3, S3 and. This increase in the scattering is a widely reported photon management approach in DSSCs to enhance the photocurrent by increasing the photon path length [44,45]. On the other hand, it is also explained by the superior charge collection efficiency of the ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 -based DSSCs (see Fig. 5). A proposed schematic representation of the electron transport mechanism and the scattering effect is shown in Fig. 5d). 4. Conclusion In this work, 1D nanostructured electrodes such as nanotubes of ITO and ITO@TiO 2, both embedded in a mesoporous TiO 2 framework, have been incorporated into DSSCs as photoanodes. These 1D electrodes are fabricated by a proprietary soft-template method using vacuum and plasma techniques. The 1D nanoelectrodes exhibit pronounced scattering effects, which are visible as a dominant diffuse component of the transmittance and reflectance over a large part of the visible range, in contrast to conventional reference electrodes. Furthermore, this scattering occurs at short wavelengths in the visible range, where most DSSC dyes present maximum light absorption. ITO NT/m-TiO 2 and ITO@TiO 2 NT/m-TiO 2 electrodes implemented in DSSCs show the highest efficiency at low-light intensities, such as those recommended for home lighting (0.04–0.014 mW cm −2 ). The most remarkable PV improvement is found for the ITO NT/m-TiO 2 samples due to their higher J sc and better FF under these illumination conditions. Specifically, a 166 % higher efficiency is achieved compared to reference photoanodes. This phenomenon can be explained by the synergy between two effects: (1) a higher fraction of incident light is scattered by 1D nanostructured photoanodes due to their highly diffusive component at short wavelengths in the visible range. Such a feature extends the photon path, which translates to a higher number of excited dye molecules; and (2) they also provide a longer charge diffusion length at low-light intensity, which results in a significantly greater charge collection under these conditions. Thanks to these excellent properties, DSSCs based on 1D nanoelectrodes can retain up to 90 % of their initial efficiency under glancing illumination from an indoor light source. In comparison, reference DSSCs lose up to 70 % of their efficiency. Therefore, the results obtained in this work suggest that the incorporation of the ITO nanotubes into the photoanodes of DSSCs promotes the regeneration of charges in the electrolyte and in the dye at low light intensities, thus (caption on next column) Fig. 5. Impedance spectroscopy characterization of DSSCs based on ITO NT/mTiO 2 and ITO@TiO 2 NT/m-TiO 2 , and reference photoanodes. Recombination resistance (a) and transport resistance (b) extracted from Nyquist plots fitting to the equivalent circuit shown in Fig. S10 as a function of the total charge density calculated using Equation (3). (c) Ratio between small-perturbation diffusion length (L n ) extracted from impedance spectra and the active layer thickness calculated using Equation 4. d) electron transport mechanism in the conventional and nanostructured photoelectrodes. J. Castillo-Seoane et al. Materials Today Energy 49 (2025) 101851 9