Electrospun Nanofiber Mats with Embedded Non-Sintered TiO₂ for Dye-Sensitized Solar Cells (DSSCs)

Abstract

TiO₂ is a semiconductor that is commonly used in dye-sensitized solar cells (DSSCs). However, the necessity of sintering the TiO₂ layer is usually problematic due to the desired temperatures of typically 500 °C in cells that are prepared on polymeric or textile electrodes. This is why textile-based DSSCs often use metal fibers or metallic woven fabrics as front electrodes on which the TiO₂ is coated. Alternatively, several research groups investigate the possibilities to reduce the necessary sintering temperatures by chemical or other pre-treatments of the TiO₂. Here, we report on a simple method to avoid the sintering step by using a nanofiber mat as a matrix embedding TiO₂ nanoparticles. The TiO₂ layer can be dyed with natural dyes, resulting in a similar bathochromic shift of the UV/Vis spectrum, as it is known from sintered TiO₂ on glass substrates, which indicates an equivalent chemical bonding. Our results indicate a new possibility for producing textile-based DSSCs with TiO₂, even on textile fabrics that are not high-temperature resistant.

1. Introduction

Dye-sensitized solar cells (DSSCs) offer the possibility to harvest solar energy, based on non-toxic and inexpensive materials. They include a semiconducting layer for which TiO₂ is often used as a thin film or nanostructure, partly blended with other materials to increase the DSSC efficiencies [1,2,3]. This TiO₂ layer is coated with a monolayer of a dye, e.g., the well-known N712 or N719, including ruthenium complexes or natural dyes, often based on anthocyanins. Dyeing is necessary, since TiO₂ has a bandgap of 3.2 eV in case of the usually applied anatase modification [4] and could thus only be excited by UV light, while the applied dye must absorb most of the visible spectrum or even in the infrared to use most of the solar photons impinging on the solar cell [5]. This means that the energy distance between the lowest unoccupied molecular orbital (LUMO) and the highest occupied molecular orbital (HOMO) is necessarily much lower than the TiO₂ band gap. In addition, the LUMO energy level must be higher than the lower edge of the conduction band of TiO₂ to allow for the excited dye electrons to travel through the semiconductor. Alternatively, some research groups try to reduce the band gap of TiO₂ by different means [6,7] or develop other semiconductors with lower band gaps [8,9,10] for possible application in DSSCs.

However, one problem of using TiO₂ is related to the sintering step that is necessary to fix the ceramic and to form a porous surface on which a large number of dye molecules can be adsorbed. This is why several research groups aim at finding low-temperature sintering processes.

Ji et al., e.g., report on adding a TiO₂ gel to a TiO₂ nanoparticle paste, only enabling sintering at 150 °C and resulting in an increase from 0.02% for the pure TiO₂ nanoparticle paste to 2.76% with additional TiO₂ gel [11]. Holliman et al. used colloidal TiO₂ films to reduce the sintering temperature to 120 °C [12]. Nguyen et al. investigated surface-disordered (“blue”) TiO₂ for low-temperature sintering at 120 °C [13]. Diverse strategies were investigated specifically for the application on textile fabrics [14], from low-temperature sintering [15] to temperature-resistant metal wires or carbon yarns as electrodes in a textile structure [16,17].

In a former investigation by our group, we found another possibility to completely avoid the sintering step: By dip-coating a stainless steel nonwoven with fine pores into a TiO₂ solution, drying was sufficient to allow for dyeing the TiO₂ in the same way as after sintering [18,19]. However, such a stainless steel nonwoven does not reveal the typical textile properties, like a soft touch or a low bending stiffness, but behaves more like a steel sheet, despite the textile production process.

This is why we report here on investigations of a common cotton fabric as an example of a macroscopic textile and a nanofiber mat with fine pores as possible matrices to embed TiO₂ from an aqueous solution. Nanofiber mats can be produced by electrospinning [20,21,22], typically using water-soluble [23] or waterproof polymers [24], partly with integrated ceramics, metals, or other materials [25,26,27]. Polyacrylonitrile (PAN), lignin, and other materials are of special interest since they can be stabilized and carbonized after electrospinning, in this way creating conductive carbon nanofiber mats that can be used in a broad variety of applications [28,29,30]. Conductivity is necessary to enable the use of nanofiber mat as an electrode in a DSSC. While this property can generally be added by dip-coating a nanofiber mat in a conductive polymer [31], here the pores should be filled with TiO₂, making carbon a good choice for the nanofiber matrix under examination.

2. Materials and Methods

The spinning solution for the production of nanofiber mats contained 16% PAN that was dissolved in DMSO (min. 99.9%, purchased from S3 Chemicals, Bad Oeynhausen, Germany).

Nanofiber mats were produced on a polypropylene nonwoven as substrate while using the needleless electrospinning machine “Nanospider Lab” (Elmarco Ltd., Liberec, Czech Republic). The following spinning parameters were used for the production: high voltage 70 kV, nozzle diameter 0.8 mm, carriage speed 100 mm/s, bottom electrode/substrate distance 240 mm, ground electrode/substrate distance 50 mm, temperature in the chamber 22 °C, and relative humidity in the chamber 32%. Spinning was carried out for 30 min.

Parts of the nanofiber mats were stabilized in a muffle furnace B150 (Nabertherm, Lilienthal, Germany), approaching a typical stabilization temperature of 280 °C at a heating rate of 1 K/min, followed by isothermal treatment at this temperature for 1 h.

A furnace (Carbolite Gero, Neuhausen, Germany) was used for the carbonization process to approach a temperature of 500 °C with a heating rate of 10 K/min in a nitrogen flow of 150 mL/min (STP), followed by isothermal treatment for 1 h.

For comparison, a cotton fabric (thickness 0.19 mm, areal weight 84.49 g/m²) was used.

For the TiO₂ solution, 15 g TiO₂ (P25 Titanium IV oxide nanopowder, 21 nm primary particle size, Sigma Aldrich, Saint Louis, MO, USA) were mixed with 125 mL distilled water and then stirred at room temperature for 15 min. Some samples (raw, stabilized, and carbonized PAN) were dip-coated into the TiO₂ solution for 5 min at room temperature and then dried.

Forest fruit tea (Mayfair, Wilken Tee GmbH, Fulda, Germany) was mixed in a ratio of 2.5 g tea:30 g distilled water for the tea solution. This tea contains anthocyanins, which are known to show a bathochromic shift when binding to TiO₂, making the chemical binding visible by a color change with respect to the dye solution [19]. The dye was extracted for 30 min. The TiO₂-coated and TiO₂-free samples were then dyed for 30 min and then dried at room temperature. Some of the samples were rinsed with water after drying to investigate the water-fastness of the dye and the TiO₂.

The DSSCs were prepared from the TiO₂ coated carbon nanofiber mats as front electrodes and conductive plastic films containing PEDOT:PSS and graphite as counter electrodes, with non-conductive glass plates supporting both supported. The electrodes were pressed together and fixed with adhesive tape. Two drops of iodine/triiodide electrolyte (from Man Solar, Petten, The Netherlands) were added on both sides of the electrodes and spread into the cells due to capillary forces. The efficient area of the cells was 6 cm².

The samples were investigated by a digital microscope VHX-600D (Keyence, Neu-Isenburg, Germany), while using a nominal magnification of 200× and a scanning electron microscope (SEM) 1450VPSE (Zeiss, Oberkochen, Germany) with energy dispersive X-ray spectroscopy (EDX) for more detailed examinations of the nanofiber mat surfaces after sputtering 20 nm gold on them to avoid the charging of the samples.

The spectra of the samples were measured using a spectrophotometer sph900 (ColorLite GmbH, Katlenburg-Lindau, Germany).

Current-voltage curves of the DSSCs prepared with dyed, TiO₂ dip-coated carbon nanofiber mats were measured with a Keithley 2450 sourcemeter, while the samples were illuminated with a solar simulator LS0500 with AM 1.5 G spectrum (LOT-Quantum Design GmbH, Darmstadt, Germany).

3. Results and Discussion

Firstly, the cotton fabric and the PAN nanofiber fiber mat (without TiO₂) were dyed and dried at room temperature. Afterwards, half of the samples was rinsed with water and dried again. Figure 1 depicts some typical results.

After dyeing, both of the fabrics show the typical reddish color of the anthocyanins in the fruit tea. Rinsing with water reduces the color, in the case of the PAN nanofiber mat completely (Figure 1d), while the cotton fabric still shows a light reddish color (Figure 1b). Former experiments with natural dyes revealed that cotton can indeed be dyed, even without mordant [32]; however, the strong reduction of the color intensity suggests that after thorough washing not much color remains. What is more important is the finding that the raw PAN nanofiber mat does not allow the dye molecules to adhere.

Next, both of the fabrics were dip-coated in TiO₂, as described above, and afterwards dyed. Again, half of the samples were rinsed with water after the first drying step. Figure 2 depicts the results. The difference between the mostly reddish color of the TiO₂ coated cotton with some more lilac areas and the completely lilac color of the TiO₂ coated PAN nanofiber mat indicates that more TiO₂ is embedded in the nanofiber mat, while the dye molecules are not chemically bound to TiO₂ in case of cotton, but physically bound between the fibers [19]. Washing the fabrics in both cases leads to a slight blue-shift of the color, indicating that the physically bound (red) dye molecules are more strongly washed off than the chemically bound ones. Especially, on the PAN nanofiber mat, the bluish color indicates that TiO₂ is still embedded in the fabric in spite of washing. This is in strong contrast to former experiments on conductive glass or other textile fabrics on which an un-sintered TiO₂ layer could not be dyed and it was even washed off partly during dyeing [18,19].

Parts of the PAN nanofiber mats were also stabilized and carbonized at 500 °C, and afterwards dip-coated again in TiO₂ and dyed. Figure 3 depicts the results.

While the color of the dyed TiO₂ on stabilized PAN (Figure 3a) is even more intense and more blueish when compared to raw PAN (Figure 2c), the TiO₂ layer on the carbonized nanofiber mat (Figure 3b) does not appear to penetrate completely into the mat, but partly stayed on top of the surface, which suggests that the dip-coating process has to be further improved for carbon nanofiber mats by modifying the coating time, the velocity at which the fabric is retracted from the TiO₂ solution, and the solid content in the solution, to avoid an unnecessary excess layer of TiO₂.

SEM images are depicted in Figure 4 for a more detailed examination of the TiO₂ layer on the nanofiber mats. On the stabilized and the carbonized PAN nanofiber mats (Figure 4c,d), a relatively even TiO₂ layer is visible, which clearly covers the nanofibers of the mat. This finding is consistent with the observation of an unnecessarily thick semiconducting layer in Figure 3b. Obviously, the TiO₂ layer should also be reduced on the stabilized PAN nanofiber mat, if this material will be used for any application. On the raw PAN nanofiber mat (Figure 4a), the nanofibers are clearly visible, embedding the TiO₂ nanoparticles. Interestingly, the surface of the PAN/TiO₂ composite looks unchanged after rinsing with water (Figure 4b), which suggests that the TiO₂ is really rigidly fixed inside the nanofiber mat.

Figure 5 additionally underlines this, showing an SEM image of the surface after rinsing with water and dyeing, i.e., after two treatments with water. While the SEM image again shows a homogeneous distribution of the TiO₂ nanoparticles over the nanofiber mat, the EDX map does not even depict the fiber that is visible in the SEM image, revealing that the majority of the TiO₂ is stored inside the nanofiber mat.

The remission spectra of the samples depicted in Figure 1 and Figure 2 are given in Figure 6 to more quantitatively investigate the dyeing process. For the stabilized and carbonized samples, these investigations were not performed due to the changing color of the fabric itself, which is superposed on the color of the dyed TiO₂ and reduces the reliability of the corresponding spectra.

In general, the minima in the remission spectra correspond to the maxima in the correlated absorption spectra. A perfectly white fabric can be expected to show a high remission near 100% over the complete wavelength range, while a completely black one shows a remission of 0%.

When comparing the cotton and PAN fabrics dyed with tea (black lines) and washed afterwards (red lines), the fabrics become significantly lighter in both cases after washing, which indicates that the dye molecules are partly (in case of cotton) or nearly completely (in case of PAN) washed out. This is clearly different for TiO₂ dip-coated fabrics. For cotton (Figure 6a), the dyeing of the TiO₂ coated sample (green line) leads to a slight shift of the remission minimum (absorption maximum) to larger wavelengths, as expected for the desired chemical bond. A small further shift is visible after washing this sample (blue line), indicating that the physically bound (reddish) dye molecules were washed off to a larger extent than the TiO₂ with the chemically bound (blueish) dye molecules. For PAN, the absolute increase in remission is larger, showing that more dye is washed off, which fits to Figure 2 showing the significant reduction in color intensity and which can be related to washing off excess dye that cannot be stored inside the nanofiber mat.

However, more interesting is the clearly increased bathochromic shift of the spectra measured on PAN with TiO₂, as compared to the spectra measured on cotton. This can also be recognized in Figure 7, which shows the wavelengths of the absorption maxima (equivalent to remission minima) of the spectra depicted in Figure 5. The grey rectangle indicates the wavelength range of the bathochromic shift, as measured in a former examination for TiO₂ on conductive glass with otherwise identical materials and equipment [33].

The raw cotton and PAN show extremal wavelengths near the values that are expected at the lower border of the grey rectangle, correlated to the spectrum of the dye solution.

In all cases, the TiO₂ coating before dyeing leads to a wavelength upshift that is larger for the washed samples, since the not chemically bound dye molecules are easily washed off. On the PAN/TiO₂ nanofiber mats, this bathochromic shift is even larger than measured in the previous experiment, where a TiO₂ layer on conductive glass was dyed, suggesting a good chemical bonding of the dye to the TiO₂ in the recent experiment.

Finally, the TiO₂ dip-coated and dyed carbon nanofiber mats were introduced as front electrodes into DSSCs. Figure 8 depicts the resulting current-voltage curves.

While the open-circuit voltages around 0.4 V are typical for anthocyanin-based dyes, the short-circuit currents that are achieved here are significantly lower than for semi-textile DSSCs with nanofiber mats as counter electrodes [31] or for DSSCs with the same foil-based counter electrodes as used here and commercially available TiO₂ coated conductive glass as front electrodes [33]. Typically efficiencies around 0.03% and 0.6% [34] are reached while using anthocyanin-based dyes without further optimization of the electrodes etc., while, in our tests we find short circuit current densities of J_SC ~0.001 mA/cm², open circuit voltages of V_OC ~0.4 V, fill factors of ~0.25 (identical to a straight line) and thus an efficiency of only 0.0001%, which is two to three orders of magnitude lower than the typical literature values.

This can be attributed to the relatively low carbonization temperature that results in only a low conductivity of the carbon nanofiber mat. More specifically, commercial FTO glasses that are used for DSSC experiments have typical sheet resistances of 15 Ohm, while the average sheet resistance of the carbon nanofiber mats used here was approx. 800 Ohm, i.e., by a factor of more than 50 higher. This large resistance significantly reduced the possible current flow along the carbon nanofiber mat. There are two possible paths to solve this problem in the future: On the one hand, carbonization at higher temperatures should be optimized to avoid the possible breaking of the nanofiber mats during their use in DSSCs; it is well-known from the literature that the conductivity can be strongly increased by higher carbonization temperatures. On the other hand, it is possible to add a highly conductive textile on the back of the nanofiber mat to support the in-plane conductivity.

Nevertheless, for the first general proof-of-concept as to whether nanofiber mats can be used to embed TiO₂ without a sintering step, tests with the more brittle nanofiber mats carbonized at higher temperatures, possibly leading to short circuits inside the DSSCs and thus requiring further research of the carbonization process, were avoided. Optimizing the interplay of electrospinning, carbonization, and dip-coating the nanofiber mats will be carried out in the near future.

It should be mentioned that, while only TiO₂–as one of the most interesting semiconductors in DSSC production–was tested here, this proof-of-concept should be extended to other semiconductors or generally ceramics. Especially, ZnO, which is also used in DSSCs, or BaTiO₃, which is interesting to be applied as a dielectric in capacitors, will be investigated in the near future. Other ceramics include Al₂O₃, which is isolating but nevertheless shows a high thermal conductivity, making a network of Al₂O₃ nanoparticles without binder interesting for heat conduction; WO₃, which can also be used in DSSCs, similar to TiO₂ and ZnO [35]; or, Nb₂O₅, which was also shown to work as a front electrode in DSSCs [36].

For these and other applications, it would be supportive to use a nanofiber mat as a matrix in which the ceramics are embedded, instead of applying them with a binder, which automatically means that they are separated from the neighboring nanoparticles, and thus destroys the possible percolation pathways. Thus, it is necessary to further investigate the possibility of using nanofiber mats as matrices for these and other ceramics and possibly further nanoparticles.

4. Conclusions

The embedding of TiO₂ in PAN nanofiber mats was found to the enable dyeing of the semiconductor without the necessity of sintering the TiO₂ layer, as it would be indispensable on glass or macroscopic textile surfaces. This allows for the use of TiO₂ on fabrics that are not heat-resistant enough to allow for sintering at the typical temperature of 500 °C.

Integration into a highly-efficient textile-based DSSC requires further optimization of the dip-coating process to avoid excess TiO₂ on the top of the carbon nanofiber mat surface and of the carbonization step to produce highly conductive, flexible carbon nanofiber mats are necessary.