Influence of Different Cations of N 3 Dyes on Their Photovoltaic Performance and Stability

The N3 dye was modified by substituting two of its protons by potassium or sodium cations. The performance and stability of dye-sensitized solar cells incorporating the new dyes were evaluated under light soaking (1000 W · m−2) at 50◦C. Photocurrent measurements demonstrated that proton substitution by potassium cations rends the system more stable. Further characterization of the potassium-based devices was performed by electrochemical impedance spectroscopy to investigate the charge-transfer phenomena occurring at the different interfaces of the cells.


Introduction
Nowadays, a particular interest in the development of alternative energy sources arises, especially motivated not only by the need of reducing the dependency on fossil fuel resources, but also for providing the reduction of the CO 2 emissions.An attractive strategy to overcome the present energy problem is using renewable energy sources, such as the direct solar radiation, for producing clean energy.In this sense, the direct conversion of sunlight into electricity by means of photovoltaic systems makes an important contribution to this energy contend in an environmentally friendly way [1,2].In conventional solar cells, the charge separation occurs at the interface of two materials of different conduction mechanisms, exploiting the photovoltaic effect [3].More recently, a new generation of cells emerged, the dye-sensitized solar cells (DSCs).DSCs are considered very promising since they use low cost, abundant and environmentally safe raw materials, showing relatively highenergy efficiency [4].
A DSC is made of a nanoparticulated titania film coated with an adsorbed dye monolayer.This thin film is applied on a glass substrate coated with a transparent conducting oxide (TCO) that collects the photoinjected electrons.A back electrode consists of the same conducting glass coated with platinum.This serves as a catalyst for the redox reaction occurring in the electrolyte present in between the two electrodes (Figure 1).
In DSCs, the dye is a crucial component to achieve high overall photoelectric conversion efficiency.Up to now, ruthenium complexes have been widely investigated due to their advantageous spectral properties, device photostability and high conversion efficiency [5][6][7].An example of these ruthenium polypyridine complexes with better performance as sensitizer is the cis-di(thiocyanato)bis(4,4 -dicarboxilic acid-2,2 -bipyridine)ruthenium(II), commonly known as N3 [8].Since the development of the N3 dye in 1993 [8], its tetrabutylammonium (TBA) salt N719 has been used as the standard red dye because of its unmatched performances.Similarly to the N719 dye, new dyes were developed bearing in mind that the number of protons in the dye influences the open-circuit potential and the short circuit current of the DSC [9].In this work we modified the N3 dye by partially substituting its protons with different cations, namely, sodium and potassium.The modified dyes were subsequently incorporated in final DSC devices, which International Journal of Chemical Engineering were then submitted to accelerated thermal/light soaking aging tests for performance and stability evaluation.The cells were subjected to full sunlight irradiation at 50 • C for about 1000 hours, during which time their photovoltaic parameters were periodically monitored.
Electrochemical impedance spectroscopy (EIS) was used to investigate the charge transfer phenomena occurring at the different interfaces of the DSCs.This technique has been widely used for the characterization of several electrochemical systems and, in particular, to analyze internal resistances in the DSCs [10][11][12][13][14][15][16][17][18][19].In 2000 Bisquert et al. applied their models in the analysis of the mechanisms of electron recombination in nanoporous TiO 2 dye-sensitized solar cells [18].After this first approach, EIS became more and more useful, playing a crucial role in what concerns modeling and understanding the complex charge phenomena occurring in DSCs [10][11][12][13][14][15][16][17][18][19].The charge-transfer resistance at the TCO layer, the charge-transfer resistance at the counterelectrode/electrolyte interface, and the charge-transfer resistance at TiO 2 /dye/electrolyte interface can therefore be obtained by fitting the EIS results to appropriate equivalent electrical circuits [20].

TiO 2 Electrode Preparation.
A screen-printed double layer film of interconnected TiO 2 particles was used as mesoporous negative electrode.A 7-μm thick film of 20nm-sized TiO 2 particles was first printed on the fluorinedoped SnO 2 conducting glass electrode and further coated by a 5-μm thick second layer of 400-nm-sized light scattering anatase particles.The detailed preparation procedures of TiO 2 nanocrystals, pastes for screen-printing and doublelayer nanostructured TiO 2 film have been reported elsewhere [22].

Dye-Sensitized Solar Cell Fabrication.
The working electrode, described above, was sealed to the counter electrode (FTO glass-15 Ω/square-coated with a platinum solution chemically deposited at 450 • C for 15 minutes) by means of a 25 μm-thick transparent Surlyn ring (from DuPont) at 130 • C for 15 seconds.The cells were filled with an electrolyte solution through a predrilled hole in the counter electrode.The hole was then sealed with a Bynel disc and a thin glass to avoid leakage of the electrolyte.

Photovoltaic Measurements.
All photovoltaic measurements were performed under a 450 W xenon light source able to provide 1000 W • m −2 sunlight equivalent irradiation (AM 1.5).The spectral output of the lamp was matched in the region of 350-750 nm with the aid of a Schott K113 Tempax sunlight filter (Präzisions Glas & Optik GmbH, Iserlohn, Germany) so as to reduce the mismatch between the simulated and true solar spectra.Various incident light intensities were regulated with wavelength neutral wire mesh atenuators.I-V curves were obtained by applying an external bias to the cells and measuring the respective photocurrent response with a digital source meter (Keithley Instruments Inc., Ohio, USA Model 2400).The incident photon-to-current conversion efficiency (IPCE) was recorded by a data-collecting system as a function of excitation wavelength.The incident light from a 300 W xenon lamp (ILC Technology, Calif, USA) was focused through a Gemini-180 double monochromator (Jobin Yvon., UK) onto the cell under test.The monochromator output was incremented through the visible spectrum to generate the IPCE (λ) as defined by IPCE(λ) = 12400(J sc /λφ), where λ is the wavelength (nm), J sc is short-circuit photocurrent density (mA•cm −2 ), and φ is the incident radiative flux (mW•cm −2 ).Photovoltaic performance was determined for an active area of 0.158 cm 2 defined by a metal mask.

Electrochemical Impedance Measurements.
Impedance experiments were carried out with a computer-controlled potentiostat (EG&G M273) equipped with a frequency response analyzer (EG&G M1025).The frequency range was 0.005 Hz-100 kHz and the magnitude of the modulation signal was 10 mV.All the measurements were performed at room temperature in the dark at  and N719 dyes in association with electrolyte E1.The cells were labelled as device A, B and C, respectively.For simplicity, the results for one single cell of each type are presented hereafter.

Results and Discussion
Comparing the three systems under study, device C shows the best initial photovoltaic performance.In fact, it is already known by its unmatched performance.On the basis of cations' substitution in the N3 dye, the anchoring groups of the adsorbed sensitizer transfer most of its protons to the semiconductor surface, charging it positively.This change in the TiO 2 surface enhances the adsorption of the anionic ruthenium complexes and favors electron injection from the excited state of the dye into the conduction band of the semiconductor, resulting in higher photocurrents.Nevertheless, this positive shift of the Fermi level induced by surface protonation leads to a low open-circuit potential.In fact, if the sensitizer has less protons, it is expected to obtain high open-circuit potential and low photocurrents, and the other way round if the sensitizer is fully protonated.So, an optimal degree of protonation of the sensitizer should be considered to reach maximum overall conversion efficiency [9].In this study we observed that the sodium-based system (device B) presents higher values of short-circuit current, J sc , and open-circuit voltage, V oc , than cells with the potassiumbased system (device A)-Figure 2(a).However, the latter system has higher overall power conversion efficiency, η, due to a higher value of the fill factor, FF.This fact is explained by a reduction in the series resistance of the cell with potassium salt dye.
In Figure 2(b) the incident photon-to-current conversion efficiency (IPCE) for devices A, B, and C can be compared.The IPCE is plotted as a function of the excitation wavelength.In line with the better photovoltaic performance, device C shows the highest IPCE: maximum of about 67% at 530 nm.At the same wavelength, device A reaches a maximum of 54%, while device B reaches approximately 63%.
Apart from efficiency, the long-term stability is also a key issue regarding the industrial development and B and C showed a drop in efficiency of about 23% and 25%, respectively.The higher stability of device A is mainly due to an increase in the photocurrent (6%) during the first week of aging, which was then maintained at the same level for the rest of the aging period.Additionally, this system shows a quite stable evolution of the FF values, indicating constancy in series resistance during the aging process.Despite its high photovoltaic performance as a fresh cell, the N719 system was unstable, exhibiting a marked decrease in J sc and V oc soon after the first week.This may be due to desorption of dye from the TiO 2 surface.Moreover, the sodium salt containing device B does not rend the system very stable, as suggested by the strong oscillation in all its photovoltaic parameters.
The stability tests allow us to conclude that the number and type of substituting cations will strongly influence the photovoltaic performances of the devices.Considering the ionic potentials of the cations we can realize how strongly they will be electrostatically attracted to ions of opposite charge and to what extent the cations will repel other ions of like-charge.As the sodium cation has higher ionic potential than the potassium cation, we may be induced to say that the stability is improved when substituting the N3 protons by sodium cations.However, this conclusion is not straightforward since we have to considerer the DSC system with all its components.The present work proves this fact since potassium substitution renders a better stability.The positive shift of the Fermi level upon adsorption of the dye decreases the gap between the redox couple and the Fermi level.This fact will strongly influence the TiO 2 /dye/electrolyte interaction, which is not totally understood.
Bearing in mind the promising results in terms of performance and long-term stability of the potassiumbased DSC, a deeper characterization of the system was performed.In this sense, the effect of the aging process in the overall performance of device A was analyzed by means of electrochemical impedance spectroscopy.This technique allows us to determine the charge-transfer resistances at the platinum counter-electrode and at the TiO 2 /dye/electrolyte interface as well as to determine the Nernstian diffusion of I − /I − 3 ions within the electrolyte [10,20].The Bode and Nyquist (Z -imaginary part of impedance versus Z -real part of impedance) plots for the potassium-based system, before and after the aging process, are shown in Figure 4.
Several physical models have been developed in an effort to understand all the complex charge-transfer processes that take place in DSCs [15,18,23].These works employ the diffusion-recombination model to study the electronic processes taking place at the semiconductor, whereas the electron transfer phenomena at the electrolyte and contact interfaces are described by simple RC electrical arrangements.Actually, the electron transport and charge recombination in nanocrystalline TiO 2 films have been widely studied by Bisquert et al. [18], who suggested an infinite transmission line model to describe the chargetransfer phenomena occurring at the mesoscopic TiO 2 film, as presented in Figure 5.
The transmission line model assumes the TiO 2 photoanode as an interconnected network where electrons, after excitation, can diffuse toward the external circuit with a resistance R w or recombine at the TiO 2 /electrolyte interface.Assuming that the recombination phenomenon is only related to electron transfer by back reaction with the electrolyte (dark current), it can be described by a charge transfer resistance, R k , and a chemical capacitance, C μ .If L is the thickness of the mesoscopic TiO 2 film, the electron transport resistance through the semiconductor is defined as R w = r w L, while the interfacial charge recombination resistance and the chemical capacitance at the interface are, respectively, R k = r k /L and C μ = c μ L.
The other charge-transfer processes occurring in DSCs are also considered in the electrical analogue, as shown in Figure 5. Regeneration of I − 3 at the counter electrode is characterized by R CE and C CE , which represent the chargetransfer resistance and the double-layer capacitance at the platinized FTO glass, respectively.
Finally, the ZView software was used to fit the experimental data to the equivalent circuit presented in Figure 5.For better fitting, all capacitor elements were replaced by  In the Nyquist diagram, symbols correspond to the impedance data obtained experimentally in the dark under −0.75 V bias, while solid lines represent the fittings according to the equivalent circuit present in Figure 5.
constant phase elements (CPEs), non-ideal capacitances associated with a nonuniform distribution of current in the heterogeneous TiO 2 film [24].Table 1 presents some parameters useful to understand the underlying mechanisms related to the aging process.Besides R k and R w , the electron lifetime n /τ n ) can also be obtained [23].According to the diffusion-recombination model, the ratio R k /R w ( 1) clearly shows that the charge-transfer resistance associated to recombination of electrons at the TiO 2 /electrolyte interface (R k ) is much higher than the electron transport resistance through the semiconductor (R w ).This means that the cell exhibits carrier collection efficiency near unity [23].
Moreover, Table 1 shows that L n is much higher than the thickness of the mesoscopic TiO 2 film (L ≈ 12 μm) in both fresh and aged cells, confirming the excellent transport properties.The increased J sc for the aged sample can be attributed to the longer electron diffusion length compared to the fresh sample (J 0 hours sc = 12.1 mA•cm −2 and J 1000 hours sc = 12.9 mA•cm −2 ).As shown in Figure 4(a), the middle frequency peak position of the aged sample was slightly shifted to higher frequencies, revealing a decrease in the electron lifetime.The decreased electron lifetime is mainly ascribed to a major decrease in R k (in Table 1) for the aged sample compared to the fresh one.This means that electrons recombine more easily with the electrolyte upon aging.A faster electron recapture taking place in the aged device prevents the electron accumulation in the TiO 2 nanoparticles from reaching the same density as for the fresh devices.As a result, the drop of V oc was observed in aged devices (Figure 3).The same conclusions can be drawn from the significant decrease of the intermediate semicircle in Figure 4(b), which corresponds to a lower value of the charge-transfer resistance (R k ) at this interface.The increased electron diffusion coefficient was probably due to the positive shift of the conduction band energy level edge.
A decrease in the redox charge-transfer resistance at the platinum counter electrode was observed upon aging (R 0 hours CE = 9.1 Ω and R 1000 hours CE = 5.5 Ω).This explains the decrease of the left-hand side semicircle in the Nyquist diagram.This semicircle corresponds to the high-frequency peak in the Bode plot, which has moved toward larger values (meaning lower electron lifetimes and shorter redox reaction times).Additionally, an overlap between the photoelectrode impedance (middle semicircle) and that related to the Nernstian diffusion within the electrolyte (righthand semicircle) can be noticed.This was probably due to a larger overpotential for the I − /I − 3 redox reaction on the platinum electrode of the aged sample, accompanied with a fill factor problem (FF decreases from 0.67 to 0.59 upon aging).

Conclusion
Two N3-based dyes were synthesized bearing in mind that the number of protons in the sensitizer influences the open-circuit potential and the short circuit current of the DSC.Potassium and sodium cations were used to substitute two protons of the N3 dye and the resulting dyes were compared with the well-known N719 dye that contains two TBA (tetrabutylammonium) cations.Despite presenting the highest initial photovoltaic performance, the N719 system was rather unstable, exhibiting a strong decrease in J sc and V oc upon aging, which might be due to dye desorption.In addition, sodium substitution in place of TBA does not help to enhance the stability of devices, whereas substitution with potassium salt revealed better stability compared to the other two systems.In fact, this system kept close to 90% of its initial performance after 1000 hours of light soaking at 50 • C.

Figure 1 :
Figure 1: Schematic representation of a dye-sensitized solar cell.

Figure 2 (
Figure 2(a) shows the photocurrent density-voltage curves obtained under AM 1.5 simulated sunlight (1000 W • m −2 ) for DSCs prepared with [2K + (N3, 2H + )], [2Na + (N3, 2H + )],and N719 dyes in association with electrolyte E1.The cells were labelled as device A, B and C, respectively.For simplicity, the results for one single cell of each type are presented hereafter.Comparing the three systems under study, device C shows the best initial photovoltaic performance.In fact, it is already known by its unmatched performance.On the basis of cations' substitution in the N3 dye, the anchoring groups of the adsorbed sensitizer transfer most of its protons to the semiconductor surface, charging it positively.This change in the TiO 2 surface enhances the adsorption of the anionic ruthenium complexes and favors electron injection from the excited state of the dye into the conduction band of the semiconductor, resulting in higher photocurrents.Nevertheless, this positive shift of the Fermi level induced by surface protonation leads to a low open-circuit potential.In fact, if the sensitizer has less protons, it is expected to obtain high open-circuit potential and low photocurrents, and the other way round if the sensitizer is fully protonated.So, an optimal degree of protonation of the sensitizer should be considered to reach maximum overall conversion efficiency[9].In this study we observed that the sodium-based system (device B) presents higher values of short-circuit current, J sc , and open-circuit voltage, V oc , than cells with the potassiumbased system (device A)-Figure2(a).However, the latter system has higher overall power conversion efficiency, η, due to a higher value of the fill factor, FF.This fact is explained by a reduction in the series resistance of the cell with potassium salt dye.In Figure2(b) the incident photon-to-current conversion efficiency (IPCE) for devices A, B, and C can be compared.The IPCE is plotted as a function of the excitation wavelength.In line with the better photovoltaic performance, device C shows the highest IPCE: maximum of about 67% at 530 nm.At the same wavelength, device A reaches a maximum of 54%, while device B reaches approximately 63%.Apart from efficiency, the long-term stability is also a key issue regarding the industrial development and

Figure 4 :
Figure 4: Bode (a) and Nyquist (b) diagrams obtained for device A before and after 1000 hours under thermal/light soaking stress.In the Nyquist diagram, symbols correspond to the impedance data obtained experimentally in the dark under −0.75 V bias, while solid lines represent the fittings according to the equivalent circuit present in Figure5.

Figure 5 :
Figure 5: Transmission line model used to fit the EIS experimental data.

Table 1 :
Parameters determined by fitting the EIS experimental data of device A to the equivalent circuit as shown in Figure5.