1 Introduction

It has been several decades since the pressure was put on the utilization of the available energy sources, as petroleum derivatives are the commonly known examples, for meeting the population growth as well as the economic advancement. Unfortunately, the usage of the non-renewable energy sources has negative impacts on the environment and the living organisms. Necessity is the mother of invention; this has attracted economists and policy makers for finding clean energy sources like photovoltaics [1,2,3]. The use of solar energy technology would benefit in producing photon-to-electron energy from the most abundant energy source; the sun, and this technology does not require intensive maintenance like other energy sources. The solar energy can be harvested in different ways to produce energy, like the conversion of the sun light into electricity or using this light for splitting water to produce green hydrogen [4, 5].

Dye sensitized solar cells (DSSCs), also known as Grätzel cells, were demonstrated in 1991 as a favourable low-cost solution to silicon solar cells. These cells were fabricated by using a large band gap semiconductor, TiO2, which was sensitized with an organic light harvester; dyes. The energy levels of the solar cell’s components must be well engineered, so that upon absorbing the light, electrons can be injected from the highest occupied molecular level (HOMO) to the lowest unoccupied molecular level (LUMO) of the dye, which then injected to the conduction band (CB) of the semiconductor, while the holes are sent to the electrolyte. The electrons flow through the external circuit to reduce the electrolyte which had been oxidized upon injecting electrons for re-generating the dye. Another path for making the DSSCs is by replacing TiO2, n-type semiconductor, with a p-type one like NiO. The dyes in this case inject holes to the valence band (VB) of the semiconductor, which then diffuse to the cathode electrode [6]. Grätzel’s invention encouraged scientists worldwide for advancing this type of solar cells, as it was predicted to be a promising way of finding an economically reasonable devices for harvesting solar energy. This is to compete the conventional photovoltaic (of the first two solar cells' generations) as DSSCs can be fabricated at lower cost and from eco-friendly materials. Additionally, the technology advancement has been going on fast, meeting the lifestyle trend when utilized in portable wireless electronic devices such as laptops, music players and mobile phones. Currently, these devices live on battery supply of energy [7].

The advantages of fabricating solar cells onto flexible conductive plastic substrates stem from a relatively low-cost manner allowing them to meet the industrial requirement for easy fabrication. One the other hand, there are some challenges when constructing thin films of semiconductors on flexible substrates. These stem from the temperature-restricted deposition methods. Most DSSCs have been fabricated on rigid substrates, like conductive glass, using binders and sintering at high temperatures to enhance the adhesion of the films onto the substrate. Unfortunately, the melting points of the plastic substrates (less than 180 °C for polyethylene naphthalate (PEN) and even lower for polyethylene terephthalate (PET)) prevent the usage of high temperatures [8]. Therefore, a mechanical compression seems to be a good method for enhancing the film-substrate adhesion as well as the inter-particle contact.

The first p-type DSSCs is accounted for Lindquist and co-workers; they constructed NiO-based DSSC in 1999, opening the door for a competition with the counterpart n-type DSSCs [9]. Sooner after that, the same group constructed the first tandem dye sensitized solar cell (T-DSSC) on glass substrates, in which TiO2 was sensitized with cis-di(thiocyanato)-N,N-bis(2,2,-bipyridyl-4,4,-dicarboxylic acid)-ruthenium (II), N719 dye as the photoanode whereas erythrosine B-dyed NiO was the photocathode. It was noticed that the power conversion efficiency (PCE) of the T-DSSC was lower than TiO2-DSSC. This observation was blamed to unmatched photocurrent from the involved two photoelectrodes. The photovoltage of the T-DSSC was the sum of the individual photoelectrodes when being constructed in a separate DSSC [10]. The fabrication of T-DSSCs, by attaching both photoanode and photocathode sensitized with relevant dyes, was said to be a way of enhancing the individual DSSCs type, cathodic or anodic. The strict requirement of fabricating T-DSSCs is that the establishment of current density matching of both photoelectrodes [10].

It should be mentioned that, as the solar cells can generate electricity from the sunlight during the daytime, this may hinder its capability to produce electricity comparable to the current conventional energy sources. One of the solutions to conquer this matter is to convert the sunlight into other fuels like oxygen and hydrogen. This conversion can be done by using photoelectrochemical cell (PEC) in a process known as photo-electrolysis, and the hydrogen can be conserved to be used in fuel cells to generate electricity. Moreover, the resulted hydrogen form splitting process can be used as fuel that is not harming the environment. The production of green hydrogen as a renewable source of energy would arise several advantages, such as relative high energy per weight unit, fuel for energy generation, mitigating the use of carbon-based energy, lowering the danger of global warming [11]. The produced hydrogen would be used and utilized by chemical and petroleum industries. It is estimated that 78% of hydrogen came from natural gas reservoir and oil whereas only 4% came from water electrolysis.

There have been several materials used for water photodecomposition with different sorts of radiation [12, 13]. The use of oxide semiconductors in photochemical cells for solar-to-fuel cells was seen to offer lower fabrication cost and higher stability. The solar energy can be harnessed by the semiconductors that involved in the PEC cell [14,15,16].

The fabrication of tandem DSSC requires using two sensitized electrodes, at least, stacked together which can produced higher photovoltages compared to a single junction solar cell. TiO2 and NiO were among photoelectrodes used in DSSC. Their use in water splitting requires higher stability of these semiconductors in aqueous solutions, have conduction bands (CB) and valence band (VB) positions that are suitable for the charge transfer, also has high transparency in the visible region [17, 18].

Indium tin oxide (ITO) exhibit many advantages that advanced its ability to be utilised in various scientific fields, in photovoltaics, sensors and water splitting [19,20,21]. These are related to its low fabrication cost, being optically transparent and having outstanding electrical conductivity [22]. ITO can be deposited onto plastic or glass substrates by different techniques, and its thermal stability was seen to increase with increasing the grain size [23]. TiO2 has been the main candidate as a photoanode, used in many applications, due to its high surface area, oxidizing ability and positive CB potential, about – 100 mV against normal hydrogen electrode at pH = 0, which is above the redox potential for splitting water. Pristine TiO2 has a large band gap of around 3.2 eV which make it difficult to absorb the visible and infrared light parts of the solar radiation, which is one of its main drawbacks. Doping TiO2 to narrow its band gap would overcome this problem and enhance its absorption [24, 25]. The morphology of TiO2 can also be modified for better PEC water splitting [26,27,28,29].

In PEC cells, TiO2 absorbs the solar energy creating holes and electrons, the holes on its surface would oxidize water to produce oxygen while electrons are transferred to the photocathode, like Pt or NiO, to reduce water and form hydrogen [30,31,32]. The photocatalytic efficiency of water splitting, and hydrogen production is badly affected by the charge recombination between the semiconductor and the used catalyst, therefore, the inhibition of back electron transfer is needed and can be done by using SnO2/TiO2 core–shell as an example [33].

ZnO and Fe2O3 are another n-type semiconductor used in PEC water splitting. The former has a band gap energy like TiO2 and suffer from the same problem regarding its absorbance range of the solar spectrum. Hence it underwent similar treatments of doping and surface improvement [34, 35]. Although, Fe2O3 has a narrower band gap, of 2.1 eV, which make it favourable for water splitting, it suffers from poor charge transportation. Doping to enhance its electrical conductivity were mandatory [36].

NiO has been the most p-type semiconductor utilized in DSSC and PEC. Its VB potential is 400 mV against NHE in phosphate buffer having pH = 6.8. Sensitizers must have HOMO more positive than VB of NiO, this is to transfer the excited electrons from its VB to the dye if used for DSSC [37]. Usually, pristine NiO suffers from charge recombination, poor charge mobility and unfavourable charge transfer at the interfaces. This requires improving its properties by doping process or chemical modifying the sensitizers. These treatments would enhance its photoconversion ability and photocatalytic properties [38, 39].

In this study, a demonstration of an easy way to fabricate a T-DSSC, from all solution processable materials, on conductive plastic substrates is shown. A flexible T-DSSC is constructed using nanostructured TiO2 and NiO which were sensitized separately by N719 and coumarin 343 dyes, respectively. The electrical characterization of the flexible T-DSSC is evaluated by using current–voltage characteristics; the J–V of flexible T-DSSC was compared with two references of DSSCs constructed from TiO2-DSSC and NiO-DSSC, individually. A schematic diagram of the cell's components is shown in Fig. 1.

Fig. 1
figure 1

The schematic diagram of the flexible Tandem DSSCs construction

Full size image

The mechanism of T-DSSC takes all processes occurring in n-DSSC and p-DSSC into account. In n-DSSC, the sensitizer (the dye) is excited by the bombarding photons and excite photo-generated electrons from its highest occupied molecular level (HOMO) to the lowest unoccupied molecular level (LUMO), then inject the electrons to the conduction band (CB) of the semiconductor, the electrons diffuse through the semiconductor layer to the transparent conducting oxide (TCO) electrode while the holes are sent to the counter electrode. Similar process occurs in p-DSSC, but the holes are injected upon the light excitation to valence band (VB) of the semiconductor while electrons are sent to reduce the electrolyte and reach the anode by diffusing process. An illustration of these processes is shown in Fig. 1.

2 Experimental methods

It should be noted that all the chemicals were used as received without further purification. Both the photoanode and the photocathode were constructed using simple method at low temperature.

2.1 The photoanode preparation

The TiO2 thin layers were doctor bladed from ethanolic paste. The TiO2 paste was prepared by placing 7 g of TiO2 (P25 powder) into a ball milled jar, followed by 16 g of absolute ethanol, the mixture was ball milled for nearly 15 h at a rate of 230 r.p.m. The formed mixture was stored in a sealed glass container and kept in a fridge. A thin film of the paste was coated on ITO/PEN flexible substrates (13 Ω/cm−2, Peccell Technologies, Inc., Japan) by doctor blade method, this was done at the laboratory ambient conditions, followed by drying at room temperature for few minutes before being annealed at 150 °C, then with compress treatment at 30 kN. The same conditions of coating and annealing were done two further times but at 30 and 0 kN, respectively. The TiO2 electrodes were soaked in 1 × 10–4 M ethanolic N719 (ditetrabutyl-4,4ˋ-dicarboxylate) dye (Solaronix SA) for 10 h.

2.2 The photocathode preparation

The NiO thin layers were doctor bladed from ethanolic paste. The NiO paste was prepared by placing 7 g of NiO powder into a ball milled jar, followed by 16 g of absolute ethanol, the mixture was ball milled for 15 h at a rate of 230 r.p.m. The formed mixture was stored in a sealed glass container and kept in a fridge. A thin film of NiO was doctor bladed onto ITO/PEN flexible substrates (13 Ω/cm−2, Peccell Technologies, Inc., Japan), this was done at the laboratory ambient conditions, followed by drying at room temperature for few minutes before being annealed at 150 °C, then underwent compression treatment at 30 kN. The coating, annealing and compression were done two further times but at 0 kN. The NiO electrodes were soaked in 1 × 10–4 M ethanolic coumarin 343 dye for 10 h.

2.3 The electrolyte

The electrolyte system consisted of a conventional iodide/triiodide based redox system; 5 ml acetonitrile, 5 ml 3-methoxypropionitrile, 0.4 M LiI (0.5354 g), 0.4 M Tetra butyl ammonium iodide (1.4775 g TBAI), 0.04 M I2 (0.1015 g) and 0.3 M N-methyl benzimidazole (0.3965 g NMB).

2.4 n-type DSSCs

TiO2 electrodes sensitised with N719 dye were rinsed with ethanol absolute to remove the excess dye molecules, then cleaned the surroundings around the active area with cotton buds. A sylurn tape was placed around the active area. Finally, Pt electrode, with a hole made in it, was placed on the active photoanode. The two electrodes were ironed for few seconds before the introduction of the electrolyte through the hole with the vacuum system.

2.5 p-type DSSCs

NiO electrodes sensitised with coumarin 343 dye were rinsed with ethanol absolute to remove the excess dye molecules, then cleaned the surroundings around the active area with cotton buds. A sylurn tape was placed around the active area. Finally, Pt electrode, with a hole made in it, was placed onto the active photocathode. The two electrodes were ironed for few seconds before the introduction of the electrolyte through the hole with the vacuum system.

2.6 Tandem DSSCs

In this configuration, TiO2 and NiO electrodes were sensitised using different dyes. TiO2 was used as the photoanode whereas NiO was the photocathode. The injection of the electrolyte was through a hole made in the photoanode. TiO2 electrodes sensitised with N719 dye were rinsed with ethanol absolute to remove the excess dye molecules, then cleaned the surroundings around the active area with cotton buds. A sylurn tape was placed around the active area. NiO electrodes sensitised with coumarin 343 dye were rinsed with ethanol absolute to remove the excess dye molecules, then cleaned the surroundings around the active area with cotton buds. The two electrodes were ironed for few seconds before the introduction of the electrolyte through the hole with the vacuum system.

2.7 Treatment with Mg(OH)2

0.01 M Mg(NO3)2.6H2O (Sigma Aldrich) aqueous solution was prepared and used for electrodepositing Mg(OH)2 layer onto TiO2 layer for 120 s, before soaking the photoanode in the N719 dye for 10 h. Followed by sandwiching the sensitised photoanode with NiO sensitised photocathode and the conventional iodide electrolyte was used for facilitating the charge movement.

3 Results and discussion

Before discussing the results, it would be appropriate to explain the working mechanism of T-DSSC.

According to Fig. 2, the dye is excited upon the bombarding photons, injecting electrons to the conduction band of TiO2 which diffuse to the working electrode, while the holes are sent to the electrolyte. The open circuit voltage, Voc, of the tandem cell is determined by the energy difference between the CB of TiO2 and VB of NiO.

Fig. 2
figure 2

schematic diagram showing the working principle of T-DSSCs

Full size image

It is known that in order to obtain a good T-DSSC, the photocurrent produced by both photoelectrodes should be matched [10]. This could be done by optimising both the thickness of the semiconductor layers and the amount of the dye adsorbed onto the photoelectrodes. The performance of T-DSSC and the reference cells (individual n-type and p-type DSSCs) was evaluated by examining the current density–voltage (J–V) measurements. The key parameters for evaluating the solar cells, performance, open-circuit voltage (Voc), short-circuit current (Jsc), fill factor (FF) and the efficiency (η) are shown in Tables 1 and 2.

Table 1 The open circuit voltage (Voc) and short circuit current (Jsc) of individually constructed photoelectrodes of tandem DSSC as well as the total T-DSSC
Full size table
Table 2 Summary of the measured key parameters of all DSSCs shown in this work
Full size table

At the first stage of this work, the J–V measurements showed a promising working process with Voc of 0.832 V and Jsc of 0.701 mA/cm2, as is shown in Fig. 3, however, there is a deform in the J–V curve indicating a high series resistance, which causes low FF, this could be due to the incapability of the NiO photocathode to produce a photocurrent density identical to the photoanode counterpart.

Fig. 3
figure 3

Photocurrent density–voltage curve under illumination for T-DSSC constructed from non-diluted TiO2 paste and NiO paste

Full size image

It was noticed that the TiO2 paste was concentrated, thus giving thicker films allowing high amount of the dye molecules to be adsorbed. This would result in producing higher photocurrent than the NiO photoelectrode can produce. To overcome this problem, the TiO2 paste was diluted in ethanol, 1:7 mL (v/v). As a result, the TiO2-DSSC gave a photocurrent identical to NiO-DSSC after this treatment. Figure 4a–c show this effect whereas Table 1 summarized the findings. The NiO DSSC produced 156 mV, TiO2 DSSC produced 759 mV whereas the T-DSSC produced 942 mV. The photocurrent produced by each solar cell was identical, indicating strong match between the photocurrent produced by each single DSSC. It is assumed that the connection between the TiO2 nanoparticles occurred during the drying process of the paste, this resulted from removing water of the hydrogen bonded network of OH-covered titania particles.

Fig. 4
figure 4

Photocurrent density–voltage curve of a TiO2-based DSSC, NiO-based DSSC (b) and c flexible T-DSSC

Full size image

The dilution treatment showed an improved performance, compare Figs. 3 and 4c. It could be seen that the dilution resulted in identical photocurrents produced by individual DSSCs, as is shown in Fig. 4. This is proven by the transmittance measurements, as shown in Fig. 5, which confirmed the outstanding capability of the semiconductors' thin films to transmit light in the visible light and infrared light regions of the solar spectrum, this means high percentages of the light reaching the sensitizers. High amounts of photons would bombard the dyes and forcing them to inject the charge carriers onto the respected semiconductors.

Fig. 5
figure 5

aTransmittance measurement of the TiO2 electrode and b the NiO electrode

Full size image

It was seen that the open circuit voltage Voc of the flexible T-DSSC is 942 mV (Fig. 4c), 759 mV for n-DSSC (Fig. 4a) while 156 mV for p-DSSC (Fig. 4b), which means the Voc of the reference cells are adding together when constructing flexible T-DSSC. The short circuit current Jsc of both n-DSSC and p-DSSC are matched when utilised in the construction of the flexible T-DSSC; 0.138 mA cm−2. Furthermore, a layer of Mg(OH)2 was electrodeposited onto TiO2 photoanode for 120 s followed by soaking the photoanode in the N719 dye for 10 h before constructing the T-DSSC. This treatment showed improved Voc by 2.34% to 964 mV, however the Jsc decreased to 0.0774 mA cm−2 as in Fig. 6. It should be noted that the illumination was done from both sides of T-DSSC as well as from each side individually. The efficiency parameters are shown in Tables 2 and 3.

Fig. 6
figure 6

J–V curve of flexible T-DSSC illuminated from a TiO2 side, b NiO side and c both sides

Full size image
Table 3 The open circuit voltage (Voc) and short circuit current (Jsc) of the T-DSSC treated with Mg(OH)2
Full size table

Although, the treatment of the photoanode with Mg(OH)2 before constructing the T-DSSC resulted in an increase in the Voc, a drop by nearly 50% of the photocurrent is observed when compared without the treatment. This would be explained by looking at the illumination from individual side of the tandem cell (Fig. 6a and b). In other words, the Mg(OH)2 could have shifted the conduction band of the photoanode, which increased the Voc produced by the photoanode, thus enhancing the tandem cell Voc. The literature suffers from the lower amounts of publications regarding the fabrication of these flexible devices. Table 4 shows a list of the literature compared with the current work. The majority of the literature focused on the fabrication of such device onto rigid substrates. This was performed by using fluorine-doped tin oxide (FTO) or indium-doped tin oxide (ITO) coated onto glass substrates as the platform of the devices.

Table 4 The comparison of the results in this work with the literature
Full size table

As a matter of truth, these devices are fabricated in a simple manner showing their capabilities in converting the sun light to electricity. However, as they were fabricated onto flexible substrates, they would not stand high temperatures as they would deform and start melting around 235 °C, beside that these devices were fabricated without binders. Taking the sintering restriction and the neglection of the binders into account, poor interconnection of the semiconductors particles is the result which would increase the sheet resistance that is usually above 60 Ω/sq [44], which was observed in this work. It was reported that the fabrication of semiconductors from free-binder pastes would lead to formation of large cavities among the particles. This may influence the connection between semiconductors' particles and the electron transfer pathway [45]. Overall, this work showed that the ability of fabricating simple tandem DSSCs that can produce high voltages with no need for binders neither for high temperature sintering. This work is believed to be possibly used during water splitting by PEC cells if its performance is improved, such as by further doping the semiconductors or using perovskite absorbers like CH3NH3PbI3 or its derivatives.

4 Conclusion

A flexible T-DSSC was constructed from binder-free TiO2 and NiO pastes, and were sensitised with N719 and coumarin dyes, respectively onto ITO/PEN substrates by using doctor blading technique. It was proved that flexible T-DSSC is able to deliver a Voc much higher than a single semiconductor DSSC. The illumination from both sides of the T-DSSC delivered better results than from single side of the cell. The performance of the fabricated T-DSSC was reasonable as it was made on flexible substrates restricting the heat treatment for better inter-particle connectivity. This tandem solar cell would be good candidate for supplying energy to PEC cells during water splitting process.