Introduction

In spite of the loss of fame following the advent of perovskite solar cells, dye sensitized solar cells (DSSCs), which can be manufactured from eco-friendly, low-cost materials, still deserve attention due to the careful materials engineering and coupling they are the object of, asking scientists of different fields to join efforts to shine light on basic phenomena. Major efforts in DSSCs are still devoted to cell components engineering1,2, including the seek for wide band gap semiconducting metal oxides (MOx) as electron transport materials (ETM) alternative to TiO21,2,3,4,5,6, and to the development of new photosensitizers enabling the exploitation of a wider portion of the solar spectrum7. Investigations should be then focused not only to enhance the overall functional performances of the solar energy converting devices, which are so far limited, but especially to unveil fundamental phenomena and processes.

In the field of ETMs to be used as photoanodes in DSSCs, SnO2 is an interesting candidate, assuming that a material-by-design approach is adopted in the preparation of nanostructures featuring proper optoelectronic characteristics, such as light management, charge transport and energy alignment with the chosen sensitizer. Indeed, over the years several attempts to optimize the aforementioned parameters by means of material shaping8,9,10,11,12,13, capping/mixture with other oxides14,15,16,17,18,19, or doping20,21 have been reported. Most of these studies dealt with the use of SnO2 nanoparticles (SnO2 NPs) of size spanning from about 6 to 140 nm, and were focused on understanding the role of light-matter interaction and charge transport through tin dioxide14,15,17,20,21,22,23.

In the field of photosensitizers, phthalocyanines (Pcs) play a prominent role due to their excellent photo- and electrochemical stability and unique light harvesting capability in the red/NIR spectral regions, allowing the exploitation of a relevant portion of the solar photon flux that cannot be captured by other dyes24. Such properties can be finely modulated by proper synthetic modifications that are relatively simple to perform25,26. The optoelectronic properties of Pcs can indeed be effectively tuned by varying the organic substituents on the macrocyclic core, as this approach is successful in the modulation of the HOMO-LUMO energy levels and of the electron density distribution. Metallation of the Pc core, in particular coordination of Zn(II) that enables the existence of long-lived singlet excited states, is another strategy that proved of great utility for DSSCs applications24. Because of this remarkable versatility, Pcs are appealing candidates as light harvesters in DSSCs, as witnessed by an impressive number of literature examples dealing with their use in combination with TiO2 photoanodes27,28,29,30,31,32. These studies also highlighted some issues seriously affecting the application of Pcs in TiO2-based DSSCs, among which their aggregation on the titania surface33, leading to the formation of never-ending dye layers on the ETM, jeopardizing the device functional performances due to enhanced exciton recombination34. Unsymmetrically substituted Pcs bearing bulky, electron-donating groups on three of the four isoindole subunits which constitute the Pc macrocycle, and one or more electron-withdrawing groups able to interact with MOx on the forth one have been designed to alleviate this issue27,28. However, the addition of co-adsorbing agents (e.g. deoxycholic acid), competing with the sensitizer for anchoring sites on the TiO2 surface and hindering interactions between Pc units, remains instrumental even with this favorable molecular pattern31,35.

With these limitations in mind and following our continuous efforts to develop both ETMs and light harvesters for DSSCs, with particular focus on SnO2 and ZnO as alternative materials for photoanodes36,37,38, and on dyes capable of efficiently replacing the most commonly used Ru(II) complexes39,40,41,42, we herein investigate the unprecedented use of Zn(II) complexes of unsymmetrically substituted push-pull Pcs (ZnPcs) as photosensitizers in DSSCs featuring SnO2 NPs as ETM.

Results and Discussion

The most successful ZnPcs sensitizers reported so far are characterized by the concurrent presence of electron-rich aryloxy groups –OAr rotated with respect to the macrocycle plane and -COOH binding groups, according to a structural pattern essentially developed by a trial and error approach24,27,28. Computational techniques were also applied to support a rational molecular design of unsymmetrical push-pull ZnPcs, strongly advocating the use of -NH2, -NHR, -NR2, NHPh or -NPh2 donor groups instead of –OAr to improve DSSC functional performances43,44,45. In order to test the validity of these predictions, we recently made available complex BI54 (Fig. 1), the first example of unsymmetrical ZnPc bearing -NPh2 donor groups.

Figure 1
figure 1

Peripherally substituted ZnPcs BI55 and BI54 bearing diphenylamino donors.

Full size image

While the measured optoelectronic parameters of BI54 were in good agreement with computational models and potentially useful for DSSC applications, the photovoltaic performance of the dye when TiO2 NPs were used as ETM disproved theoretical predictions46. This inconsistency was ascribed, among others, to the problematic adsorption of BI54 on TiO2 NPs and to the possible shift of the CB of TiO2 leading to a decreased charge injection efficiency from the excited state of the sensitizer. It was thus interesting to investigate the photovoltaic behavior of BI54 in combination with a different ETM. On the other hand, the binding mode of unsymmetrical ZnPcs on MOx and their photovoltaic behavior are strongly influenced by the interposition of a linker between the Pc core and the anchoring group33,47,48,49. We thus synthesized BI55 (Fig. 1, see Supplementary Materials for details), an analog of BI54 where the Pc core and the –COOH anchoring group are connected by an aryl ring, and tested the two ZnPcs in parallel as photosensitizers in DSSCs fabricated with SnO2 NPs as ETM.

The energy levels of the HOMO-LUMO orbitals of BI55 were assessed by electrochemical measurements combined with the analysis of the optical properties (Figures S1 and Table S1 in the Supplementary Materials). With HOMO and LUMO positioned at about −5.5 eV and −3.8 eV, respectively, and a typical UV-Vis absorption spectrum characterized by strong bands centered at 353 nm (Soret band) and 718 nm (Q band), the new dye satisfies the thermodynamic requirements for an effective use as light harvester in DSSCs, analogously to BI5446.

Thanks to a favorable band alignment (sketched in Fig. 2), the use of SnO2 should provide for a better driving force as for electron injection than that with TiO2, given that the nanostructured MOx is properly designed for photogenerated charge transport and light scattering. SnO2 is also expected to offer enhanced stability under UV exposure as compared with TiO2 thanks to the wider band gap50. Finally, SnO2 is fit for coupling with NIR absorbing molecules, thus appearing the ideal candidate to be tested with both ZnPcs.

Figure 2
figure 2

Energy band alignment of TiO2, SnO2 and BI54, BI55 dyes. Commercial Ru(II) based dye N719 is reported for comparison purposes. Energy position of HOMO and LUMO of BI55 and BI54 have been obtained by electrochemical and optical analyses as described in the text.

Full size image

SnO2 nanoparticles used as ETM were synthesized through a wet chemical approach using methanol as a solvent. In order to check the complete removal of methanol, as well as any residual surface hydroxyl groups, Fourier transform infrared spectroscopy was applied. The spectra pertaining to the material after the drying procedure in oven and to the metal oxide after the calcination are reported in the Supplementary Materials (Figure S2). SEM analysis of SnO2 photoanodes (Fig. 3) evidences a homogenous deposition of the scattering layer over a wide area (Fig. 3a) and shows that the metal oxide is constituted of uniform spherical nanoparticles whose size is about 20 nm (Fig. 3b), as previously verified by XRD36.

Figure 3
figure 3

SEM images of SnO2 nanoparticles based photoanodes. Scale bar: (a) 20 μm; (b) 200 nm.

Full size image

Estimation of the active area per mm2 of the SnO2 photoelectrode has been performed by using the commercial Ru(II) based dye N719 as a molecular probe, considering a molecular area as high as 1.6 nm2 for the dye51 and is reported in Fig. 4a, together with the data pertaining to TiO2 photoelectrode, used as a benchmark.

Figure 4
figure 4

(a) Calculation of the active area per mm2 featured by the SnO2 (black markers) and TiO2 (green markers) photoanodes. (b) Dye loading for the photoelectrodes based on SnO2 (black markers) and TiO2 (green markers) NPs.

Full size image

An average value of about 1.25 m2/mm2 has been calculated for the SnO2 NPs (TiO2 photoanodes, composed of both transparent and scattering layers, i.e. with particles featuring diameters from 20 nm to 150–250 nm on average, featured a value as high as 1.84 m2/mm2). Active area is a critical parameter since it relates to dye loading capability, hence to the current that can be photogenerated in the solar energy converting device. Dye loading has been also evaluated by detaching the adsorbed dye N719 and found be about 2.45 × 10−9 mol/mm2 for SnO2 photoanodes and 1.91 × 10−9 mol/mm2 for TiO2-based photoelectrodes (Fig. 4b). This is a relevant finding: it has indeed been previously reported a significantly lower dye loading for SnO2, ascribed to the more acidic nature of the SnO2 surface (anatase TiO2 presents the isoelectric point at pH 6.2, while for SnO2 it is at pH 4–5)52. Usually the lower photocurrent featured by SnO2-based DSSCs is then attributed to this difference in dye loading. However, in the present study, SnO2 shows a dye uptake capability slightly higher than TiO2 and, as we will see, this is reflected in the short circuit photocurrent density of the devices exploiting the two oxides as ETMs.

The functional parameters of various DSSCs based on SnO2 as ETM and dyes BI54 or BI55 as light harvester are reported in Table 1, together with data obtained using the commercially available dye N719.

Table 1 Functional parameters recorded for the dye sensitized solar cells presented in this work.
Full size table

As previously mentioned, ZnPcs are usually subjected to aggregation due to π-π stacking, which leads to the formation of adsorbed multi-layers onto the MOx surface, detrimental for functional performances. We have recently demonstrated46 that also BI54 suffers of this issue on TiO2 photoelectrodes and we verified the same in case of SnO2. Indeed, solar converting device sensitized with dye BI54 in the absence of co-adsorbing agents show very low functional performances (Entry 3 in Table 1 and Figure S1 in the Supplementary Materials). Deoxycholic acid (DCA) was thus added to the sensitizing mixture (50 mM), which resulted in enhancing all the functional parameters: JSC in increased of more than seven times, VOC about 2.5 times and an increase of 26 times is recorded as for the photoconversion efficiency (entry 4 in Table 1). Thus, the addition of co-adsorbing agents acting as molecular spacers is confirmed as critical in order to effectively keep the dye molecules separated on the MOx surface and to avoid the formation of never-ending layers atop the ETM surface.

In order to assess the capability of SnO2 NPs as ETM in DSSCs with respect to the more commonly used TiO2, the behavior of devices with photoanodes sensitized with two different dyes, N719 and BI54, was investigated.

Figure 5a shows the J-V characteristics in dark and under simulated solar light of DSSCs sensitized with N719, where the ETM is either TiO2 (green lines) or SnO2 (black lines). Functional parameters are reported in Table 1, entries 1 and 2).

Figure 5
figure 5

Comparison of the performance of TiO2 and SnO2 as ETM for DSSCs. (a) and (c) = J-V characteristics under simulated solar light (solid lines) and in dark (dashed lines); (b) and (d) = electron lifetime. (a) and (b): N719 is used as light harvester. (c) and (d): BI54 is used as light harvester. Black line: SnO2; green and pink lines: TiO2.

Full size image

When the commercial Ru-based dye N719 is used as light harvester, a higher VOC value is recorded for the TiO2-based device (740 mV) compared with the solar cell working with SnO2 as ETM (510 mV), as expected, while comparable short circuit current is observed (11.61 vs. 11.44 mA/cm2 for TiO2 and SnO2 photoanodes, respectively). Electron lifetime (Fig. 5b) is found to be slightly longer for the cell based on SnO2.

It is worth noting that the performances of the present bare (i.e. without any surface treatment aimed at reducing back recombination between the MOx CB and the electrolyte) SnO2-based DSSCs sensitized with N719 are much higher than those previously reported for other bare SnO2 photoanodes, as it can be seen in Table 2.

Table 2 Functional parameters of SnO2-based DSSCs reported in literature. ‡ Sizes inferred by SEM images reported in the supporting information of ref. 20.
Full size table

Open circuit photovoltage value observed in the present work is the highest recorded for SnO2-based DSSCs where neither TiO2 buffer layer is applied to the conductive substrate nor TiCl4 post treatment is used to cap the SnO2 ETM. This outcome is of extreme relevance: it indicates, indeed, that the synthesized SnO2 NPs feature an appropriate charge transport skill, with reduced exciton recombination, as compared with previously investigated photoanodes based on the same MOx.

When dye BI54 is applied as light harvester (Fig. 6c,d), the observed open circuit photovoltage is still higher for TiO2 (540 mV) than for SnO2 (180 mV), for which, however, it is reduced beyond the expectations. On the contrary, short circuit current density is almost two times higher when SnO2 is applied as ETM (2.605 mA/cm2 vs. 1.376 mA/cm2 recorded for TiO2-based DSSC). One of the advantages of SnO2 over TiO2 is the higher electron mobility: mobility reported for SnO2 (μe ≈ 250 cm2V−1s−1 for SnO2 single crystal53 and μe ≈ 125 cm2V−1s−1 for SnO2 nanostructures54) is order of magnitude higher than the one reported for TiO2 (μe < 1 cm2V−1s−1 for TiO2 single crystal55). However, this higher mobility is of no help with Ru-based light harvesters, as widely shown in literature, while it is capable of spurring the photogenerated current when a light harvester with specifically designed feature is applied.

Figure 6
figure 6

Comparison between DSSC sensitized with BI55 (orange line and markers) and BI54 (black line and markers). (a) J-V characteristics (solid lines: simulated solar light; dotted line: dark); (b): electron lifetime; (c) recombination resistance; (d) chemical capacitance.

Full size image

Electron lifetime appears slightly longer for the TiO2-based device (Fig. 6c). The Reader should however be aware that a longer electron lifetime may also be correlated to the presence of charges trapped in undesired surface states, which are of no use in converting the solar light.

Analysis of J-V characteristics in dark provides useful information on the origin of the seemingly inconsistent behavior of the two MOx. We should herein remind that VOC should ideally match the difference in energy between the conduction band (CB) of the ETM and the standard redox potential of the electrolyte couple (the iodine/triiodine in the systems analyzed here). Such view well explains the difference in VOC observed for TiO2- and SnO2-based DSSCs exploiting dye N719 as a dye (see the energy scheme reported in Fig. 2), but it is no longer satisfactory when BI54 is instead used. J-V characteristics in dark (reported as dotted lines for both MOx and both dyes in Fig. 6a,c) show an increased dark current for SnO2-based devices, irrespectively of the used dye. The difference in dark current is moderate for devices sensitized with dye N719 (Fig. 6a): for both cells dark current onset is around 350 mV. This difference is however more pronounced in DSSCs sensitized with dye BI54 (Fig. 6c), for which the onset moved below 200 mV. The dark current in solar cells comes from charge recombination: an enhancement of dark current is then associated to a faster recombination22,which is usually reflected in lower open circuit voltage values, as it is in the present cases. Given that the two devices have exactly the same structure but for the sensitizer, this effect is to be correlated with the nature of the ZnPc light harvester, which apparently is inhibiting the charge transport more than dye N719. The differences in the onset of dark current recorded for dye N719 and BI54 can be ascribed to the skill of N719 in partially blocking the recombination at the FTO||electrolyte interface56, which is not featured by the ZnPc BI54. One of the reasons why SnO2 is investigated as alternative ETM to TiO2 in excitonic solar cells is, as mentioned, the higher electron mobility featured by this MOx. However, it should not be neglected that this higher electron mobility might also result in enhanced recombination, leading to a more pronounced electron-hole recombination.

The most relevant photovoltaic parameters of SnO2-based DSSCs sensitized with ZnPCs BI55 or BI54 are summarized in Fig. 6 (functional parameters are reported in Table 1, entries 5 and 6).

Interestingly the shapes of J-V characteristics are quite similar and suggest no difference between the resistances of the two devices (shunt resistance, RSH, and series resistance, RS). Since the values for VOC only differ for 67 mV (260 mV and 327 mV for sensitizer BI55 and sensitizer BI54, respectively), a different adsorption mode of the dyes to the surface of oxide might be responsible of the difference in term of JSC. The Fermi level EF and the electric field that assists the injection of the charges are possibly similar for the two devices and the density of the charge is lower in the case of BI55.

Also in this case, dark J-V characteristics bring useful information to explain the difference in performances. The device sensitized with ZnPc BI55 shows an enhanced dark current, which, accordingly to what mentioned above, suggests a reduced skill of the dye in blocking the recombination at the FTO||electrolyte interface. By equaling the short circuit currents57, i.e. the charge injection (orange markers in Fig. 6a), it is possible to observe that the DSSCs work in the same way. This strongly suggests that the limiting factor in the BI55-sensitized device is the charge injection from the LUMO to the SnO2 CB. The EIS analysis applied to these two cells confirms this hypothesis (Fig. 6c,d). The two devices show no differences in RREC (Fig. 6c), while the trends of Cμ (Fig. 6d) suggests a downward of the CB for the oxide sensitized with BI54, as previously reported46. Indeed, the slope of Cμ is different when the two ZnPcs are applied: Cμ pertaining to the device sensitized with ZnPc BI55 is less steep as compared with that observed in case of BI54, indicating a beneficial and better capping of the SnO2 trap states by this latter. We have reported similar observations when TiO2 is applied as ETM and comparison is made between the dyes N719 and BI5446. Also in that case, poor injection was found responsible for the lower functional performances.

Conclusions

We presented a study on the use of two unsymmetrically substituted ZnPcs as light harvesters in DSSCs exploiting either TiO2 or SnO2 nanoparticles as ETM.

Investigated molecules have been especially designed to fulfil the requirements identified by computational models as for energy level alignment between their LUMO and the MOx CB, good charge injection and absorption range extended to the infrared portion of the solar spectrum.

On the other hand, SnO2 nanoparticles developed in the present study have shown very nice performances when the commercial Ru(II)-based dye N719 is applied as sensitizer as compared with previously reported tin dioxide photoanodes, featuring a photogenerated current density comparable with that observed with TiO2 photoandoes and very good open circuit photovoltage.

Despite the excellent preconditions for both components, device functional performances proved to be below the expectations (less than 1% as for photoconversion efficiency). Main reasons behind this outcome have been identified in the poor charge injection from the excited ZnPcs to the CB of SnO2 and low capability of ZnPcs in contributing to block the recombination at the FTO||electrolyte interface. As in the case of TiO2-based DSSCs, our findings also confirm the ease of aggregation of ZnPcs on the MOx surface its negative impact on the device performance. We will thus focus our efforts on improving the performance of SnO2-based devices through specific modifications of the substitution pattern of the phthalocyanine ring, according to a strategy recently demonstrated in the case of perovskite solar cells featuring ZnPcs as hole transporting materials58.

While hypotheses built up by molecular modelling are useful hints for the design of new photosensitizers, this work confirms that a thorough analysis of the results obtained through experimental tests remains crucial for further structural modifications and refinements. Indeed, many parameters play a relevant role in the working mechanism of a solar energy converting devices and unexpected factors may reveal more critical than expected.

Methods

Dye-sensitized solar cells

SnO2 nanoparticles synthesis

In a round bottomed flask 1.2 ml SnCl4 were dissolved in 100 ml of methanol. After the fumes had disappeared, 4 ml NH4OH (30% in water) were added dropwise in about 10 minutes. As soon as NH4OH was added, a white flocculate appeared in the reaction mixture. Mixture was let reacting for about 20 minutes at room temperature, and then the solvent was slowly evaporated at 80 °C in an oven (6–7 h). The raw product was then annealed at 450 °C for 2 h under air atmosphere. All chemicals were purchased from Sigma-Aldrich and used without any further purification.

SnO2 paste preparation

SnO2 paste was prepared by mixing 0.8 g of SnO2 nanoparticles powder with ethyl cellulose (0.7 g) and α-terpineol (1.5 ml), in ethanol/water dispersion medium (8/3.5, V/V). The mixture was kept under vigorous stirring overnight to obtain a homogenous paste suitable for tape casting.

Photoanode preparation

TiO2 pastes (Dyesol 20 nm anatase nanoparticles paste 18NR-T and WER2-O, composed of anatase nanoparticles with diameter between 150 and 250 nm) were purchased from Dyesol. The redox electrolyte was composed of 0.1 M LiI, 0.05 M I2, 0.6 M 1,2-dimethyl-3-n-propylimidazolium iodide, and 0.5 M 4-tert-butylpyridine dissolved in acetonitrile.

The TiO2 and SnO2 paste were deposited onto FTO glass (sheet resistance 10Ω /☐) by tape casting technique. The paste was dried 15 min at ambient conditions and then fired on hot plate for 6 min (125–130 °C) for the TiO2 based photoanodes and 10 min at ambient conditions and then fired on hot plate for 45 min (100 °C) for SnO2 based photoanodes. In the case of TiO2 paste two layers of transparent TiO2 (18NR-T) and one of scattering TiO2 (WER2-O) were deposited.

All the samples were finally annealed at 450 °C for 30 min under ambient atmosphere. Photoanode thickness was evaluated by stylus profilometry and found about 15 μm.

The sensitization of photoanodes was performed for 5 hours in dark condition using solutions containing 0.5 mM N719 in ethanol and 0.17 mM BI54 or BI55 in dichloromethane.

Device assembly

DSSCs were fabricated using dye sensitized oxide photoanodes and platinized FTO glass as a counter electrode (5 nm Pt thin film deposited by sputtering) with 60 μm-thick plastic spacers (Surlin from Solaronix). The I3−/I− redox couple electrolyte was composed of 0.1 M LiI, 0.05 M I2, 0.6 M 1, 2-dimethyl-3-n-propylimidazolium iodide, and 0.5 M 4-tert-butylpyridine dissolved in acetonitrile.

Scanning electron microscopy analysis was carried out in a LEO 1525 FE-SEM.

Dye loading quantification and metal oxide surface area evaluation were performed by spectrophotometry, detaching the absorbed dye N719 by using a 0.1 M NaOH aqueous solution and recording the UV-Vis absorption spectra in a T80 spectrophotometer (PG Instruments) after building up a calibration curve with standard solutions; quartz cuvettes were used (1 cm optical path).

The current–voltage (I-V) characteristics of the fabricated cells were measured without masking by using a Keithley 2400 SourceMeter under simulated sunlight using an ABET2000 solar simulator at AM 1.5 G (100 mW cm−2) calibrated using a reference silicon cell. The electrochemical impedance spectroscopy (EIS) was carried out in dark conditions using a SOLARTRON 1260 A Impedance/Gain-Phase Analyzer, with an AC signal 20 mV in amplitude, in the frequency range between 100 mHz and 300 kHz. An external bias was applied from 0 to 100 mV above the VOC.

Electron lifetime of the photogenerated charges present in the CB of the dyed semiconductor is calculated from open circuit voltage decay using the following equation:59

$${\tau }_{n}=-\frac{{k}_{B}T}{e}{(\frac{d{V}_{OC}}{dt})}^{-1}$$

where kBT is the thermal energy, e is the positive elementary charge.