High performing PbS Quantum Dot Sensitized Solar Cells exceeding 4% efficiency: The role of metal precursor in the electron injection and charge separation

Here we report the preparation of high performance Quantum Dot Solar Cells (QDSCs) based on PbS/CdS co-sensitized nanoporous TiO2 electrodes. QDs were directly grown on the TiO2 mesostructure by Succesive Ionic Layer Absorption and Reaction (SILAR) technique. This method is characterized by a fast deposition rate which involves random crystal growth and poor control of the defect states and lattice mismatch in the QDs limiting the quality of the electrodes for photovoltaic applications. In this work we


Introduction
Since the early nineties, sensitized solar cells 1 have attracted a great deal of attention as low cost alternative for photovoltaic devices.Over the last two decades, an intensive effort has been carried out around the world in order to increase the efficiencies of these devices with a current record efficiency of 12%. 2 In spite of this remarkable performance, there is still room for further improvement.One of the strategies to improve the efficiency of these devices entails shifting the light absorption threshold into the near IR region, where the solar photon flux density is maximum.Inorganic semiconductor materials appear as ideal candidates as sensitizers, operating in the near IR region, since the bulk band gap of some of these semiconductors (as PbS) perfectly matches the spectral range of interest for light harvesting.
In the last few years, the use of inorganic semiconductors as alternative sensitizers for Dye Sensitized Solar Cells (DSCs) has experienced an impressive enhancement, [3][4][5][6][7][8][9] reflected in the continuous growth of the number of scientific papers published.The efficiencies obtained with these sensitizers remain below those reported for molecular dyes.4][5][6][7][8][9] The recent reports of perovskite semiconductor on nanoporous matrix with outstanding efficiencies, around 10%, 13,14 will undoubtedly enhance the interest in inorganic semiconductors as light harvesting materials in nanostructured devices.
One of the fundamental differences between inorganic semiconductors and dyes lays on the synthetic versatility of the former ones.Inorganic semiconductors can be prepared following many different synthetic routes and the growth method has a significant influence on their properties. 15,16 s an example, colloidal chemistry offers a precise control over the crystalline properties and size of the semiconductor material. 11low a critical size, these materials exhibit quantum confinement effects with band gap controlled by particle size.The solar cells using these materials as sensitizers are termed Quantum Dot Sensitized solar Cells (QDSCs).Colloidal QDs can be directly attached to the nanoporous structure of a sensitized solar cell [17][18][19] or using a bifunctional linker. 18,20,21 Ufortunately, the QD loading of sensitized electrodes with colloidal QDs has generally showed to be insufficient to harvest all the incident light. 15wever, the photovoltaic efficiency of solar cells with colloidal QDs is significantly enhanced when inverted type-I CdS/CdSe core/shell QDs are employed. 227][28] The advantage of higher QD loading is partially balanced by the lower control over the QD growth conditions.Compared to colloidal QDs, lower crystalline quality is obtained for QDs directly grown on nanostructured electrodes with a broad distribution of QD sizes together with the development of grain boundaries. 15In these conditions, it is expected that the growth method and growth conditions dramatically affects the final photovoltaic performance when directly grown QDs are employed.It has been showed that CdS/CdSe QDSCs grown by CBD systematically exhibit higher open circuit voltage, V oc , compared to QDs grown by SILAR. 16Here we show that the selection of the metallic precursors to grow PbS/CdS QDs by SILAR has a dramatic effect on the final QDSC performance.
PbS is a particularly interesting semiconductor material, with a bulk band gap in the IR region, 0.41 eV. 29Its band gap can be tuned by reducing the size of the PbS nanoparticles to the quantum confinement region.Promising results on multiple carrier generation have been reported with colloidal PbS QDs. 30 By tuning the band gap of PbS, a double objective can be attained: i) better match with the optimum absorption band gap 31 and ii) a correct band alignment in order to inject photoexcited electrons into TiO 2 conduction band (CB). 32With a band gap in the near IR region, it is expected that PbS QDSCs exhibit high photocurrents, J sc .In fact, J sc larger than 20 mA/cm 2 has been observed with PbS colloidal QDs in thin film colloidal QD solar cells (Schottky 33 and Depleted Heterojunction Solar cells 34 ).However, these high current densities have not been obtained in the case of QDSCs until the present study where we report a photocurrent as high as 22.3 mA/cm 2 .In order to attain high photocurrent with PbS QDSCs some important obstacles needed to be solved.The most important one was relied to the solar cell stability, since PbS is not stable in contact with polyionide or polysulfide electrolytes. 35We solved this problem by coating PbS with CdS, obtaining a stable behavior of the heterostructured PbS/CdS absorber with polysulfide electrolyte. 36,37 Onthe other hand, other authors have showed that the nanostructured electrode for PbS/CdS QDSCs is not fully optimized.The group of Qing Wang showed that it is possible to increase PbS/CdS photocurrents, up to 17.4 mA/cm 2 , by using nanostructured SnO 2 instead of TiO 2 . 38Alternatively, Qingbo Meng's group showed that a hierarchical pore distribution of the TiO 2 nanostructured photoanode led to a significant increase of the solar cell performance, J sc =18.8 mA/cm 2 and efficiency η=3.82%. 39Here we show that further improvement, J sc =22.3 mA/cm 2 and η=4.20%, can be obtained by optimizing the crystal growth of PbS/CdS, using different precursor salts.We have systematically analyzed the prepared samples to unveil the physical origin of the increase of efficiency.Compared to electrodes sensitized with nitrate precursors, those sensitized with acetate precursors exhibit higher injection rate (from the sensitizer to TiO 2 ) due to lower internal recombination in the QDs as consequence of the different contribution of the QD surface states in each case.

Experimental Section
Device preparation.Glass with a transparent and conductive SnO 2 :F (FTO) layer, Pilkington, 15/sq resistance, were used as substrate.FTO was coated by a compact layer of 150 nm TiO 2 deposited by spray pyrolisis of titanium(IV)bis(acetoacetonato) di(isopropanoxylate) and sintered at 450ºC for 30 minutes.The mesoporous photoelectrodes were prepared using a double layer film of interconnected titania nanoparticules deposited over compact TiO 2 by doctor blade technique.The mesoporous film for QDSC preparation consists of a 9 µm-thick transparent layer of TiO 2 (DSL 18-NRT, 20 nm average particle size) and a 5 µm-thick layer of scattering particles (DSL, WERO-4, 300-400 nm particle size distribution).For light absorption characterization films with a single layer of transparent paste (3 µm-thick) and no scattering layer were prepared in order to avoid the light scattering effect in absorption measurements.The films were sintered 30 minutes at 450 ºC to obtain a good electrical contact between nanoparticles.Film thickness was measured by profilometry (Dektack 6 from Veeco).
The Successive Ionic Layer Adsorption and reaction (SILAR) technique was used to grow double layer PbS/CdS films.This technique is well described in previous works. 26,28,37 SLAR involves the crystal growth "layer by layer" by sequentially dipping the substrates into the ionic precursor solutions.A 0.02 M methanolic solution was used as lead source for PbS deposition and a 0.05 M methanolic solution as cadmium precursor for CdS.The sulfide precursors were 0.02 M and 0.05 M solutions of Na 2 S × 9 H 2 O in methanol/water (1:1, V/V) for Pb 2+ and Cd 2+ ions, respectively.
After the each dipping step in a precursor solution, the electrodes were dipped in a solution without precursor in order to rinse the precursor excess.The sequence of dipping processes metallic precursor-rinse -sulphur precursor-rinse constitutes a SILAR cycle.The duration of each deep in precursor or rinse solution was 1 min.The amount of deposited material increases with the number of SILAR cycles.Two types of precursors for Cd and Pb were employed, CH 3 COO -(acetate, hereafter Ac) and NO 3 -(nitrate, hereafter N) salts.High magnification TEM pictures of the electrodes sensitized with different precursors are compiled as Supporting Information SI1.In all the experiments, the SILAR process was carried out by SILAR equipment from ISTest at room temperature under air atmosphere.After sensitization, all the samples analyzed in this study, have been coated with 2 SILAR cycles of ZnS, by alternately dipping into 0.1M Zn(CH 3 COO) 2 and 0.1M Na 2 S × 9H 2 O solutions for 1 min/dip, rinsing with Milli-Q ultrapure water between dips.
The device preparation was carried out by sandwiching the working electrode (sensitized photoanode) with the Cu 2 S counter electrode and using polysulfide electrolyte and scotch tape as spacer (50 μm thick).Polysulfide electrolyte was 1 M Na 2 S, 1 M S, and 0.1 M NaOH solution in Milli-Q ultrapure water under nitrogen bubbling mixture and the Cu 2 S counter electrodes were prepared by immersing brass in HCl solution at 70ºC for 5 min and subsequently dipping into polysulfide solution for 1 min, resulting in a porous Cu 2 S electrode. 40The geometric area of the cells was 0.196 cm 2 .In Fig. 1(a) a cartoon of the analyzed QDSC structure is depicted.
For each studied device configuration, at least two identical cells, but generally more, have been produced in order to check the reproducibility of the analyzed devices.
The dispersion is lower than 15%.Some examples of reproducibility are showed in Supporting Information, SI2.
Characterization of TiO 2 sensitized electrodes.The absorption spectra were recorded by a Cary 500 UV-VIS Varian photospectroscometer. TEM measurements were carried out with a JEM-2100 Electron Microscope (JEOL) operated at 200kV.

Photoelectrochemical characterization. The Incident Photon to Current Efficiency
(IPCE) measurements were performed employing a 150 W Xe lamp coupled with a monochromator controlled by a computer; the photocurrent was measured using an optical power meter 70310 from Oriel Instruments, using a Si photodiode to calibrate the system.Current potential (J-V) curves and impedance spectroscopy (IS) measurements were obtained using a FRA equipped PGSTAT-30 from Autolab.The cells were illuminated using a solar simulator Sun2000 from ABET Technologies at AM 1.5 G, where the light intensity was adjusted with an NREL-calibrated Si solar cell with a KG-5 filter to 1 sun intensity (100 mW/cm 2 ).J-V curves under illumination were performed using mask and with no antireflective layer.Impedance spectroscopy measurements were carried out in dark conditions at different forward bias, applying a 20 mV AC sinusoidal signal over the constant applied bias with the frequency ranging between 400 kHz and 0.1 Hz.
Ultrafast characterization.The principle and setup of the lens-free heterodyne detection transient grating (TG) technique have been reported in detail in previous papers. 41,42 riefly, in TG characterization a pump probe beam indices over the sample though a transmission grating surface.Sample can be excited by the optical interference pattern, that photoexcites electron-hole pairs.Changes in the reflection coefficient, dependent on charge density, photoinduced with this pattern are registered allowing to monitor the carrier density. 43In this experiment, the laser source was a titanium/sapphire laser (CPA-2010, Clark-MXR Inc.) with a wavelength of 775 nm, a repetition rate of 1 kHz, and a pulse width of 150 fs.The light was separated into two beams.One of them was used as a probe pulse.The other light beam was used to pump an optical parametric amplifier (OPA) (a TOPAS from Quantronix) to generate light pulses with a wavelength tunable from 290 nm to 3 µm.It was used as a pump light in the TG measurement.In this study, the pump pulse wavelength was 520 nm and the probe pulse wavelength was 775 nm.

Surface photovoltage spectroscopy (SPV).
For recording SPV spectrums the measurements were performed in parallel-plate capacitor arrangement. 44Light comes over the sample through a transparent conductive oxide electrode that works as one of the plates del capacitor, while the FTO substrate is the other contact acting as the second capacitor plate.In order to ensure that no current is flowing between the two plates a insulator mica sheet is placed between sample and contact.SPV spectra were measured under high vacuum by using a halogen lamp with a quartz-prism monochromator for the excitation and a chopper for modulation (modulation frequency 8 Hz, signal detected with a lock-in amplifier).For the time-resolved and surface photovoltage spectroscopy 4-µm-thick measurements, nanocrystalline transparent TiO 2 thin films were prepared, without scattering layer.

Results and Discussion
Combining Pb and Cd, nitrate and acetate salts, four different combinations of PbS/CdS solar cells can be produced.Fig. 1 shows the results obtained with sensitized electrodes produced after 2 and 5 cycles of PbS and CdS, respectively.The number of SILAR cycles has been optimized after a preliminary study on the solar cell performance with the number of SILAR cycles and metallic precursor, see Supporting Information, SI3 and SI4.The solar cell performance, Fig. 1(b), is clearly affected by the metallic precursor employed for the growth of both PbS and CdS.The solar cell parameters extracted from the J-V curves plotted on Fig. 1(b) are shown in Table 1.The enhancement in solar cell performance is particularly dramatic when Cd (Ac) is used as cadmium precursor, leading to a remarkable photocurrent, with values higher than 20 mA/cm 2 when Pb (Ac) is also employed as lead precursor.The efficiency of the cell using acetate precursors for both materials attains efficiencies as high as 4.20%.Similar efficiency, 4.24%, but with lower photocurrent values has been observed when only one SILAR cycle with Pb (Ac) was applied, see SI4.This extraordinarily high photocurrent can be reproducibly obtained, see Supporting Information, SI2.These large photocurrents are due to the extension of the light absorption to the red and near IR (NIR) obtained with the PbS sensitization, as evidenced by the IPCE spectra showed in Fig. 1(c).In a previous study, we showed that that a fluorine treatment on the bare TiO 2 electrode also enhances the PbS deposition kinetics, 36 increasing the photocurrent, when the fluorine treatment was applied to TiO 2 .However, the recombination rate in PbS/CdS QDSCs increases with the amount of PbS deposited material, 37   In order to unveil the origin of this enhanced performance with acetate precursors, cells plotted in Fig. 2(a) and (b) have been characterized by impedance spectroscopy and analyzed with the standard models for QDSCs. 18,28,37,45,46 Thecompared values obtained for chemical capacitance, Cµ, and recombination resistance, R rec , are depicted in Fig. 2(c) and (d), respectively.The similar values obtained for Cµ, indicate that the nature of the precursor does not affect the position of the TiO 2 conduction band. 46Additionally, the recombination resistance, R rec , is very similar for both precursors, Fig. 2(d).This result suggests that there is no change in the recombination rate of electrons in the TiO 2 with acceptor species in the QDs and/or the electrolyte, since these are the recombination processes susceptible to be monitored by IS. 46,47 Moreover, this result is in good agreement with the similar V oc obtained for both cells (Figure 2b and Table 2).Consequently, both solar cells are similar in terms of light absorption and IS characterization, although their photovoltaic behavior is clearly different, particularly in terms of J sc .This means that the origin of the difference relies on a process which is not accessible by these techniques: electron/hole photoinjection after photocarrier generation.Ultrafast processes, as photoinjection, have been characterized by the transient grating (TG) technique. 41,42 n parallel with the experiments presented in Figure 2, we prepared samples for ultrafast characterization with the same optical density, Fig. 3(a).
From the absorption spectra for both sets of samples we can assume that films have similar density of generated photocarriers.The TG signal of the semiconductor QDs on the fast time scale used in this study (less than ns) is proportional to the change in the refractive index, Δn(t), of the sample due to photoexcitation, which can be approximately determined by 41,42 where the first and second term represent the changes in refractive index induced by photoexcited electrons and holes, respectively.N e (t) and N h (t) are the photoexcited electron and hole densities, respectively.m e and m h are the effective masses of electrons and holes, respectively, and A is a proportionality constant.The exact contribution of electrons and holes to Δn(t) depends inversely on their effective mass.According to the Drude theory, we can consider that only free photoexcited electrons and holes are responsible for the population grating signals.For bulk PbS, the effective masses of electrons and holes are 0.09m 0 and 0.09m 0 (m 0 is the electron rest mass), 48 respectively, so both the photoexcited electron and hole carrier densities in the PbS QDs contribute to the signal and it not possible to discriminate the origin between electrons and holes.It is known that the effective mass of electrons for TiO 2 is about 30 m 0 , which is about three orders larger than that for PbS.Therefore, the TG signal due to the injected electrons in TiO 2 can be ignored.
We found that the TG response for PbS and PbS/CdS sensitized electrodes can be accurately fitted with a double exponential decay plus an offset, as shown in eq. ( 2), where A 1 , A 2 and A 3 are constants, and τ 1 and τ 2 are the time constants of the two decay processes.The fitting curves using eq. 2 are represented with solid black line in Fig. 3(b) and the values obtained from the fitting are indicated in Table 3.Here, the constant term A 3 corresponds to the slowest decay process, in which the decay time (in the order of ns) is much longer compared to the time scale of 400 ps measured in the present study.The three different decay processes have a weight on the total decay process defined as , where i=1, 2 and 3 is the process that is weighted.The obtained weights for the different processes are also indicated in Table 3.
Table 3. Fitting parameters and corresponding errors of TG responses shown in Fig. 3(b) according to eq. ( 2).The weight of each decay process in percentage is also indicated.The dependence of the TG response on the pump intensity was measured and it was found that the dependence of the maximum signal intensity on the pump intensity was linear.Additionally, the waveforms of the different responses perfectly overlapped when they were normalized.These results indicate that the time constants were independent on the pump intensity, and many-body recombination processes such as Auger recombination could be neglected.Therefore, it is reasonable to assume that the decay processes of photoexcited electrons and holes in the PbS QDs are due to onebody recombination processes such as trapping and/or transfer.As shown in Table 3, the decay times τ 1 and τ 2 of the two fast decay processes are about few ps and a few tens to hundreds ps, respectively.τ 1 and τ 2 decrease as PbS (N) > PbS (Ac) > PbS (N)/CdS (Ac) > PbS (Ac)/CdS (Ac).Lower times are detected when acetate precursor is used, but results are not totally conclusive when the fitting error is considered, see Table 3.On the other hand, the fitting error for A 1 , A 2 and A 3 is significantly lower compared to that for τ 1 and τ 2 , allowing a more conclusive analysis of these data.It is worth noting that the weight of the three decay processes, A 1 , A 2 and A 3 , also change systematically as showed in Table 3.A 1 and A 2 decrease as PbS (N) > PbS (Ac) > PbS (N)/CdS (Ac) > PbS (Ac)/CdS(Ac), while A 3 follows the opposite trend.One possible effect of the CdS coating on PbS is to reduce the surface defects of PbS QDs. 37It is clear that the surface defects could greatly affect photoexcited electron and hole dynamics and concomitantly, the photovoltaic properties.In order to achieve high IPCE and photocurrents, surface defects should not be significantly reduced. 34Therefore, the first two decay processes could be assigned to electron and hole trapping processes, and the last slow decay process corresponds to electron/hole injection.A 1 and A 2 correspond to trapped electron and hole concentration since these parameters significantly decrease after CdS coating, especially for PbS (Ac)/CdS(Ac).A 3 can be assigned to injection of electron/hole, i.e., the relative electron injection efficiency, dramatically increases after CdS coating.With this interpretation, compared to PbS(N), lower concentration of surface defects exist for PbS (Ac), from the different weight of the trapping/injection processes, see Table 3..These results are in good agreement with the J sc values observed in Fig. 2(b).Therefore, we can conclude that the different weights observed in Table 3 can be attributed to different properties of the PbS surface states (nature, density, energy level...), with a dramatic impact on the photoinjection of carriers from PbS QDs.An increase of the injection efficiency is detected when Pb (Ac) is used as precursor instead of Pb (N).In addition, after CdS deposition, the injection efficiency increases for both lead precursors, indicating that the CdS capping passivates the PbS surface states.A similar beneficial effect has been described for the ZnS capping used in QDSCs, which leads to a significant enhancement of the cell photocurrents. 19,49 ote that all the cells analyzed in Fig. 3 have also been capped with ZnS as described in the experimental section.
In order to validate the conclusions extracted from the ultrafast characterization by transent grating technique, surface photovoltage (SPV) measurements have been carried out.SPV is particularly sensitive to characterize charge separation. 50,51 5][56] In SPV, a voltage as a function of wavelength, for a modulated light beam, is recorded.In order to observe a SPV signal, two different processes must take place: i) chage generation and ii) charge separation. 50,51 ig. 4 shows the typical modulated SPV spectra obtained for the analyzed PbS/CdS samples, with Ac precursor for both Pb and Cd depositions and 1 and 5 SILAR cycles, respectively.The modulated light beam is connected to a Lock-in amplifier that measures the voltage originated by the charge separation of the photogenerated charge.Fig. 4 shows the in-phase and phase-shifted signals in linear and logarithmic scale.The in-phase signals were positive over the whole spectral range, providing evidence for preferential modulated separation of photo-generated electrons towards the internal interface. 44,50,52 I the linear scale Fig Consequently, SPV measurements provide evidence of the presence of surface states.Samples PbS(Ac) with 3 cycles and PbS(N) with 4 cycles can be properly compared with respect to the onset wavelengths and SPV amplitudes.The SPV amplitudes of these samples were 4.5 and 2.7 mV in the maximum and 0.007 and 0.011 mV at 1900 nm, respectively.Therefore, normalizing both signals at their maximum amplitudes, the SPV signals related to modulated charge separation from defect states at 1900 nm were more pronounced by a factor of 2.5 for the nitrate sample compared to the acetate one.This analysis indicates a high presence of traps at the band gap for nitrate samples, in good agreement with the results derived from TG characterization.
The presence of trap states in the characterized samples and the higher density of traps for PbS (N) is also discussed in term of the SPV phase in SI6.
Despites the hypothesis of surface states is extensively used to justify the results obtained for QDSCs, it is commonly not justified with experimental results.In this manuscript we relate experimentally by first time the relation between surface states and solar cell performance in QDSCs, that we have analyzed with TG and SPV.The exact mechanism to explain the different density of surface states caused by the different precursors is being currently investigated, nevertheless some hypothesis could be formulated.SILAR technique consists in crystal growth by the successive adsorption and reaction of ionic layers.So the species present in the ion adsorption equilibrium at the solid-liquid interfaces have a key role in the growth process and the final properties of the QDs films (crystal size, crystallinity and deposition rates, among others).The metal ions adsorbed on the surface are stabilized by the counter ions.In this sense, the strength of the coordination bond between the metal and the counter ion has a direct influence in the crystal deposition and in the surface defects distribution.The coordination strength can be measured in terms of stability constant K n . 57Strong coordination makes difficult the substitution of the counter ion by the sulfide, increasing consequently the number of crystal defects.On the other hand, weaker coordination increases deposition rate but reduces the stability of the surface cadmium atoms.A compromise needs to be found for an optimal performance.When comparing nitrate and acetate precursors, the last one presents lower coordination strength.In absolute term acetate exhibits a moderate coordination strength of carboxyl group that seems suitable to obtain well crystallized film and higher growth rate, as was previously demonstrated by Sagawa et al. for CdS films. 57her factor that could affect the crystal growth is the pH of the metal precursor

Conclusions
PbS/CdS QDSCs with reproducible conversion eficiencies higher than 4% have been fabricated using Pb and Cd acetate precursors in the SILAR QD growth processes.
Unprecedented photocurrents for sensitized solar cells, including dye sensitized, higher than 20 mA/cm 2 have been obtained.Compared to nitrate precursors, the use of acetate precursors leads to faster deposition rates of both PbS and CdS.We have showed that besides this effect, the precursor plays a more intricate role on the final solar cell performance.Cells prepared with the same amount of light absorbing material exhibit higher performance when acetate precursors are employed.From impedance spectroscopy characterization, we ruled out any effect of the metallic precursor on the position of the TiO 2 conduction band and on the recombination rate of electrons in the

Figure 1 .
Figure 1.(a) Schematic diagram of the PbS/CdS sensitized solar cells.Solid arrows indicate electron photoexcitation and injection, and dotted arrows internal recombination (before injection), blue arrows indicate the regenerative reaction of QDs with the redox couple and the reverse reaction at the counter electrode.(b) Currentvoltage curves under 1 sun illumination and (c) IPCE of the analyzed devices.QDs were in situ grown by SILAR technique with 2 cycles for PbS and 5 cycles for CdS.(Ac) and (N) in the legend refers to the metallic salt source employed for PbS and CdS deposition by the SILAR process, acetate and nitrate, respectively.(d) Absorption spectra of electrodes produced with CdS grown with cadmium acetate salt as cadmium precursor.Identical color code is used in (b), (c) and (d).
and a judicious balance between photocurrent and photovoltage is compulsory to optimize the conversion efficiency.In order to further investigate the role of the metallic precursor in PbS/CdS QDSCs, solar cells with electrodes sensitized by different number of SILAR cycles but same amount of deposited material, see Fig.2(a), have been prepared.The corresponding J-V curves are plotted in Fig.2(b) and the photovoltaic parameters of these devices are summarized in Table2.Higher J sc is obtained for the cell prepared with Pb (Ac) precursor, suggesting an additional role of the precursor material on the final cell performance besides the change of the deposition rate.

Figure 2 .
Figure 2. Study of the role of PbS salt precursor.PbS/CdS samples were prepared using acetate and nitrate as lead source varying the SILAR conditions to obtain films with comparable absorption properties.The notation (X/Y) is referred to the number of SILAR cycles for PbS (X) and CdS (Y), respectively.(a) absorption spectra, (b) J-V curves, (c) chemical capacitance, Cµ, and (d) recombination resistance, R rec , as a function of applied voltage, V app .

Fig. 3 .
Fig. 3. (a) Absorption spectra for Pb/CdS co-sensitized and PbS single sensitized TiO 2 photoanodes employing acetate (Ac) and nitrate (N) as lead precursor for PbS growth, while only (Ac) precursor was used for CdS growth.(X/Y) is referred to the number of SILAR cycles for PbS (X) and CdS (Y), respectively.(b) TG responses measured under N 2 of the electrodes plotted in (a), solid line is the result of the fitting using eq.(2).
Fig.4(b), the absolute SPV signals can be also analyzed at the longer wavelengths.The

Fig. 4 .
Fig. 4. Surface photovoltage spectra of the sample PbS (Ac) / CdS (Ac) with 1 and 5 cycles respectively for the in-phase (filled circles) and phase-shifted (open circles) SPV signals in (a) the linear scale and in (b) the logarithmic scale for the absolute signals.The line shows the intensity spectrum of the halogen lamp.

Fig. 5 .
Fig. 5. (a) Spectra of the surface photovoltage amplitude of PbS (Ac)/CdS (on the top) and PbS (N)/CdS (below) samples for 1, 2, 3, and 4 cycles of PbS deposition by SILAR (circles, triangles, squares, and stars, respectively).5 SILAR cycles and Cd (Ac) has been used as cadmium precursor in all the cases for CdS deposition.The lines show the intensity spectrum of the halogen lamp.(b) Values of the surface photovoltage amplitude at 700 nm, (on the top) and of the wavelength related to the onset of the surface photovoltage due to absorption and charge separation in the PbS quantum dots (below) for the PbS (acetate)/CdS (circles) and PbS (nitrate)/CdS (triangles) as a function of the number of SILAR deposition cycles.

TiO 2
to accepting species in QDs and/or electrolyte.TG measurements indicate a stronger weight of injection in comparison with trapping in surface states for samples prepared with acetate precursors.SPV measurements have confirmed that the metallic precursor employed in the QD growth of PbS and CdS has a dramatic effect on the density of surface states.A lower density of traps is detected when the acetate precursor is employed for electrode sensitization.The exact mechanism to explain the different density of surface states caused by the different precursors is being currently investigated.This work unambiguously unveils the dramatic role that surface states play on the QDSC performance and paves the way to further improve the QDSCs efficiencies by the appropriated treatment of surface states.

Table 1 .
Photovoltaic parameters of the analyzed sensitized solar cells under 1 sun illumination (shortcircuit current, j sc , photovoltage, V oc , fill factor, FF, and conversion efficiency,).Photoanodes were sensitized by SILAR using 2 cycles for PbS and 5 cycles for CdS.

Table 2 .
Photovoltaic parameters of the solar cells plotted in Fig.2under 1 sun illumination.