Initial Performance of Dye Solar Cells on Stainless Steel Substrates

Feb 19, 2008 - The suitability of stainless steel for dye solar cell substrate was investigated with respect to performance and stability using photov...
0 downloads 17 Views 251KB Size
Download PDF
J. Phys. Chem. C 2008, 112, 4011-4017

4011

Initial Performance of Dye Solar Cells on Stainless Steel Substrates Kati Miettunen,* Janne Halme, Minna Toivola, and Peter Lund Laboratory of AdVanced Energy Systems, Department of Engineering Physics and Mathematics, Helsinki UniVersity of Technology, P.O. BOX 5100, FIN-02015 TKK, Finland ReceiVed: NoVember 29, 2007; In Final Form: December 27, 2007

The suitability of stainless steel for dye solar cell substrate was investigated with respect to performance and stability using photovoltaic characterization, electrochemical impedance spectroscopy (EIS), open circuit voltage decay (OCVD), and substrate polarization measurements. Stainless steel was employed both as photoelectrode and as counter electrode substrate gaining initial cell efficiencies of 4.7% and 3.5%, respectively. The leakage current from the stainless steel substrate was found to be very low. The effect of the stainless steel substrate on the performance of the other cell components was also examined. The traditional data analysis based on external cell voltage was shown to be inadequate and even misleading. Here, the voltage over a single cell component was determined computationally on the basis of EIS measurements as a function of cell current; through this approach, we found that the stainless steel counter electrode did not have any impact on the photoelectrode whereas the stainless steel photoelectrode substrate decreased the effective electron lifetime and the recombination resistance of the dyed TiO2 film.

1. Introduction Nanostructured dye solar cells (DSC) are relatively new photovoltaic devices. Their simple manufacturing method may lead to lower fabrication costs compared with conventional solar cells. Traditionally, DSCs are deposited on TCO glass sheets which have been evaluated to be the most expensive single component in DSCs, responsible for over 30% of the material costs.1 The preparation of DSCs on flexible metal substrates is one way to reduce the costs and also to enhance roll-to-roll mass production. Full flexibility requires that the other substrate should be flexible as well, for which a transparent conductive oxide (TCO) coated plastic sheet would be suitable. Stainless steel (StS) has been used both as counter electrode (CE)2-4 and photoelectrode (PE) substrate5-8 since it withstands temperatures required for thermal platinization of the CE and sintering of the PE. On the other hand, in the case of a StS PE cell configuration, light has to enter the cell through the counter electrode causing additional losses such as light absorption in the electrolyte and the counter electrode catalyst layer (Figure S1, Supporting Information). The suitability of uncoated StS as a conducting substrate in DSCs with respect both to performance and to stability is not fully clear. ITO and SiOx/ITO coatings have been reported to have a positive impact on the cell performance, especially on the current, and it has been suggested that the back reaction from the StS substrate could be notable.6 The stability of DSCs with a stainless steel substrate has not been as good as that of glass based DSCs.3,9 Unlike many other metals, StS substrates have withstood electrolyte in soaking tests without any signs of corrosion.2-4 Besides corrosion, metals might cause surface contamination of the TiO2 layer. 400 ppm or more iron oxide mixed in the porous TiO2 layer has been reported to decrease the photocurrent significantly10 which raises questions on the chemical compatibility of StS and TiO2. Accordingly, it has been recommended that metal objects excluding titanium should * Corresponding author. E-mail address: [email protected].

be kept away from the cell, although direct evidence on surface contamination through this route has not been presented.11 This study aimed to determine the performance of the StS based solar cells in such a way that the degradation effects could be reliably disaggregated from their characteristic performance in stability studies to follow. A detailed analysis of the initial state of StS PE and CE cells was performed in comparison to glass cells. The goal was to experimentally decouple the effects on the PE and CE performance, when StS is used either as the PE or CE substrate, and for the PE to distinguish quantitatively the rate of electron recombination via the TiO2 film and directly via the substrate electrolyte interface. To facilitate systematic comparison with the standard glass based DSC, glass was used instead of plastic as the other substrate in the StS cells. The primary hypothesis to be tested in terms of electrochemical performance was that the PEs have similar performance in the StS CE and glass cells, as well as the CEs in the StS PE and glass cells owing to their identical materials and preparation. Additionally, the nanostructured TiO2 photoelectrode films were expected to function similarly in the StS PE and glass cells for the same reason. 2. Experimental Methods 2.1. Samples. The studied photoelectrode substrates were stainless steel 304 (1.25 mm, Outokumpu Ltd.) and fluorinedoped tin oxide (FTO) coated glass (2.5 mm, Pilkington TEC15, sheet resistance 15 Ω/sq, Hartford Glass Company, Inc.). Before use the substrates were washed with mild detergent and rinsed in tab water followed by 3 min in an ultrasonic bath first in ethanol and then in acetone. Porous TiO2 layers with size of 4 mm × 8 mm were deposited on the photoelectrode substrates by distributing commercial titania paste (STI) using tape as a frame. The layers were heated at 110 °C for about 10 min after the deposition of each layer and finally sintered at 450 °C for 30 min. The thickness of the TiO2 films was measured to be approximately 15 µm using a Dektak 6M stylus profiler (Veeco Instruments). The photoelec-

10.1021/jp7112957 CCC: $40.75 © 2008 American Chemical Society Published on Web 02/19/2008

4012 J. Phys. Chem. C, Vol. 112, No. 10, 2008 trodes were sensitized for 16 h in a dye solution consisting of 0.32 mM cis-bis(isothiocyanato)bis(2,2′-bipyridyl-4,4′-dicarboxylato)-ruthenium(II) bis-tetrabutylammonium (Solaronix SA) in ethanol (99.5 wt %). The counter electrodes were prepared using the thermal platinization method.12 A few drops of 5 mM tetrachloroplatinate PtCl4 (Sigma-Aldrich) dissolved in 2-propanol were spread on the substrates followed by firing at 385 °C for 15 min. A 25 µm thick Surlyn ionomer resin film 1702 (DuPont) with 5 mm × 14 mm aperture was employed as the spacer between the electrodes. Liquid electrolyte consisting of 0.5 M LiI, 0.03 M I2, and 0.5 M 4-tert-butylpyridine in 3-methoxypropionitrile was inserted through filling channels drilled through the glass. The channels were closed with a 40 µm thick Surlyn 1601 foil (DuPont) and a thin cover glass. Before use, the spacer and encapsulation resin films went through similar ultrasonic baths as the substrates. Copper tapes were employed as current collector contacts and Electrolube conductive silver paint was applied on the edge of the conductive tape and the substrate to reduce resistance. In addition to the complete solar cells, substrate-counter electrode (SU-CE) cells were prepared using both stainless steel and glass substrates. These cells are similar to solar cells except for the photoelectrode which is replaced with a substrate. In order to make the substrates resemble the photoelectrode substrate, they were thermally treated the same way as the PEs at 450 °C for 30 min. Thermal treatment13 and dyeing14 of the substrates have been reported to change their electrochemical properties. Three different kinds of samples were prepared (Figure S2, Supporting Information): bare substrates, dyed substrates, and bare substrates coated with an undyed porous TiO2 film. In order to reach good comparability of the data, the SU-CE cells were made simultaneously with the complete solar cells and all dyed components were sensitized at the same time during which the undyed substrates for SU-CE cells were kept in ethanol. The results were confirmed by repeating the preparation and the measurement for other series of samples. All of the presented data is from one series of cells. 2.2. Measurements. Photovoltaic measurements were performed using a solar simulator with halogen lamps providing 1000 W/m2 AM 1.5 G equivalent light intensity determined by a calibrated silicon reference cell with spectral filter to mimic typical DSC response. The solar cells were placed on an aluminum plate cooled to 25 °C with Peltier elements. The IV curves were measured using a Keithley 2420 SourceMeter. Both illumination from the PE side (here after referred to as PE light) and from the CE side (CE light) were used. The ready-made solar cells were provided with black masks with a rectangular 6 mm × 10 mm aperture. There was also a black tape behind the cells in the measurement. According to the literature,11 using a mask with a slightly larger aperture size compared with the active area of the cell leads to the most reliable results. In open circuit voltage decay (OCVD) measurements, the cells were illuminated using a red LED (λpeak ) 639 nm) while keeping the cells at the open circuit voltage. After the VOC had stabilized, the light was turned off, and the decay of the open circuit voltage was then recorded in 50 ms intervals using an Agilent 34970A data logger with 1 MΩ ( 2% input impedance and response time less than 40 ms. The measurements were performed in a black box to minimize stray light. Electrochemical impedance spectroscopy (EIS) was performed over the frequency range 100 mHz-100 kHz with Zahner Elektrik’s IM6 Impedance Measurement unit. Measurements were taken in the dark in potentiostatic mode using 10

Miettunen et al. TABLE 1: Average Initial Solar Cell Performance Characteristics of 3-4 Samples and Their Standard Deviations StS PE StS CE glass

VOC (mV)

iSC (mA/cm2)

FF (%)

η (%)

-664 ( 10 -672 ( 16 -663 ( 8

11.7 ( 0.5 13.7 ( 0.3 13.8 ( 0.4

57 ( 2 36 ( 3 57 ( 1

4.4 ( 0.2 3.3 ( 0.3 5.2 ( 0.1

mV amplitude. To examine the stability of the measurement, the frequency range was measured twice starting from 100 kHz to 100 mHz and then back to 100 kHz. ZView2 software was employed for the equivalent circuit analysis. Steady-state IV measurements of the SU-CE cells were performed using the same equipment as for the EIS. To avoid hysteresis, the IV curves in the voltage range -0.5-0.5 V were measured in 20 mV intervals with 30 s stabilization time for each voltage point. 3. Results 3.1. Initial Photovoltaic Performance. The short circuit current of the StS CE cells was similar to those of glass cells (Table 1, Figure S3, Supporting Information) and in good agreement with literature.2-4 This is expected since in these cells the PEs are identical and the illumination is from the same direction. However, we acknowledge that multiple15 and back reflection of light within the cell differ between StS and glass substrates which poses some uncertainty to the comparison of the iSC values. The cells with StS PE illustrated somewhat lower currents as can be expected because of the reversed light direction (Table 1). Others have needed an additional thin blocking SiOx/ITO underlayer between the StS substrate and the dyed TiO2 to reach as high iSC as reported here6,7 and the highest iSC without such a layer has been about 40% lower.5 Besides the optical losses in the counter electrode and the electrolyte layer, the reversed light causes the electrons to be generated on average further from the photoelectrode substrate which may lower the electron collection efficiency. The glass cells illuminated from the CE side gave similar iSC as the StS PE cells which indicates that the lower photocurrent in the StS PE cells is caused by the optical effects. The open circuit voltage (VOC) was approximately equal in all the cells (Table 1). Replacing glass PE substrate with StS lowered the efficiency only 15% (Table 1). Even though the iSC and VOC were higher with the StS CE compared with the StS PE, the StS CE cells had a lower overall efficiency due to a low FF (Table 1), in accordance with the literature data of thermally treated StS CEs.2,3 The low FF is due to the high charge-transfer resistance at the StS CE as confirmed by the EIS results discussed later in this paper. Contamination of the TiO2 film by iron has been suggested to affect the iSC negatively.10 However, here, iSC was equal in the StS CE and glass cells. Also, the StS PE and glass cells with CE light had similar iSC. These observations refer that iron contamination does not have a significant effect on the initial cell performance. This was confirmed with a separate test in which glass photoelectrodes were purposely kept in direct contact with a StS substrate during their sintering. This treatment influenced neither the photovoltaic performance nor the EIS characteristics of the cells. However, the position of the StS, which was different in the contamination test compared with StS PE, might have an effect on the contamination. Contamination near the collecting substrate/TiO2 layer contact might be more important than on the other side of the film.

Dye Solar Cells on Stainless Steel

J. Phys. Chem. C, Vol. 112, No. 10, 2008 4013

3.2. Electrochemical Impedance Spectroscopy. In this section, the computational method for separation of the single electrode performance is demonstrated followed by the analysis of the photoelectrodes and the counter electrodes. The equivalent circuit model used here was similar to that presented by Fabregat-Santiago et al.16-18 and it is discussed in detail in Supporting Information. 3.2.1. Determination of the Polarization CurVe of a Single Impedance Component. Two electrode EIS measurements can be used to determine the differential resistance of a cell component (x) as a function of cell current and voltage, Rx(Icell) and Rx(Vcell) respectively, given that the frequency response of this component can be distinguished in the EIS spectrum. Except in the case of purely ohmic resistances, Rx is voltage dependent. Thus, quantitative comparison of Rx(Vx) between different types of cells should be done at equal potential drop Vx over the component. The measured cell voltage is

Vcell ) Vx + ∆V

(1)

where ∆V is the voltage loss in the other components. Vx is needed for the determination of the polarization curve of the component and can be measured with the help of an appropriate internal reference electrode.19,20 In case of two electrode measurements, there are two possibilities for the determination of Vx. One can try to determine ∆V in order to make a correction in the measured Vcell. Unfortunately, this voltage compensation is straightforward only in case of ohmic losses that account for only a small part of ∆V in DSCs. Here, we use another approach based on Rx(Icell) measurements suggested recently by FabregatSantiago et al.17 Contrary to Rx(Vx), Rx(Ix) can be measured with normal two electrode measurement given that

Ix ) Icell ) I

(2)

that is, all of the external current passes through the component in question. Since the differential resistance Rx(I) represents the derivative of the polarization curve Vx(I) of the component x, Vx(I) can be obtained by integrating Rx(I) over a certain current range:

Vx(I) )

∫II Rx(I) dI + Vx,0

Figure 1. Charge-transfer resistance at the photoelectrode/electrolyte interface RPE as a function of cell voltage for different cell types in the dark. Average data for 3-4 cells. Error bars indicate the standard deviation and are shown when larger than the marker size.

(3)

0

where (Vx,0, I0) is a known reference point. With Vx(I) determined, Rx(Vx) can be readily calculated from the Rx(I) data. In the analysis of DSCs, RPE(VPE) and RCE(VCE) are typically of primary interest. In practice, the reference points VPE,0(i0) and VCE,0(i0) can be obtained from complete cell polarization curve at the point where current is close to zero since there VPE equals Vcell while VCE is zero. 3.2.2. Photoelectrode Charge-Transfer Resistance. The aim here is to analyze the impact of the StS on the photoelectrode as it is used as the PE or CE substrate. The substrate/electrolyte interface and the TiO2/electrolyte interface are effectively connected in parallel. Thus, in practice only one constant phase element/resistor pair (here after marked as CPEPE/RPE) can be experimentally detected. Figure 1 shows the RPE as a function of the external cell voltage for the different types of cells, displaying typical exponential voltage dependence.16 From -0.5 V to -0.7 V, the StS CE cells have higher RPE compared with the glass cells, which at first sight suggests differences in the electrochemical properties of the equivalently prepared PEs in these cells. The StS PE cells in contrast illustrate similar RPE as the glass cells at the -0.7 V while being lower at other

Figure 2. RPE of StS PE, StS CE, and glass cells as a function of cell current density in the dark. The average linear fits are drawn for each solar cell type. The fits corresponding to the StS CE and glass cells are overlapping.

potentials. However, as the RPE is plotted against the cell current, the situation is practically the opposite (Figure 2). First, the StS CE cells and the glass cells follow the same curve. Second, the StS PE cells match with the glass cells in the low currents but differ toward higher currents. If the components were similar, their Rx(i) curves should also be equal. Note that according to eq 3, Rx(i) defines the Vx(i) only within the unknown constant Vx,0, and thus equal Rx(i) is a necessary but not in general a sufficient condition for equal Vx(i). Therefore, while Figure 2 suggests that the PEs of the StS CE and glass cells might be similar and the StS PE different, this needs to be confirmed by calculation of the photoelectrode polarization curve VPE(i) taking into account the appropriate reference point VPE,0(i0). To calculate the RPE, a linear fit was made to match the logarithmic values of RPE against i (Figure 2). Data at -0.2 V corresponding to the lowest absolute current points in Figure 2 were excluded since they did not follow the same trend in all the cases. The fitted equation was

log (RPE) ) A log(i) + B

(4)

where A and B are fitted constants. By using eqs 3 and 4, VPE(i) was obtained as

VPE(i) )

i1+A - i1+A 0 × 10B + V0 1+A

(5)

It was mentioned previously that a logical starting point for integration would be the open circuit state. Here, -0.3 V was

4014 J. Phys. Chem. C, Vol. 112, No. 10, 2008

Miettunen et al.

Figure 3. Calculated RPE as a function of voltage at the photoelectrode. Average data for 3-4 cells. Error bars indicate the standard deviation and are shown when larger than the marker size. The StS CE and glass cell curves are overlapping.

employed as V0, since it corresponded to the lowest current in which the measured data and the fitted line had a good correspondence. From -0.2 V to -0.3 V, the measured Vcell was assumed to be equal to VPE. The error of this assumption was estimated to be insignificant since toward the open circuit state the cell resistance and thus voltage is dominated by the PE. The RPE as a function of VPE (Figure 3) confirm that the electrochemical performance of the glass photoelectrodes were corresponding in the StS CE and glass cells while the StS PEs differed. The same observation can be made from the calculated polarization curves of the photoelectrode recombination reaction in the dark (Figure S4, Supporting Information). The StS PE cells have significantly lower recombination resistance and higher absolute currents than the glass PEs. 3.2.3. Photoelectrode Capacitance. Constant phase elements (CPE) were chosen instead of pure capacitors to describe the double layer charging at the photoelectrode/electrolyte interface since they corresponded better with the measured data and are commonly used in similar cases. The equivalent capacitance CPE of the photoelectrode/electrolyte interface CPEPE was determined from the EIS data as18

CPE )

1 RPEωPE

ωPE ) (QRPE)

(7)

Q is the prefactor and β the exponent in the constant phase element impedance function:

ZCPE )

1 Q(jω)β

τeff )

(8)

where ω is the angular frequency and j the imaginary unit. The voltage dependence of RPE and CPE of the glass PEs is qualitatively consistent with literature.16 The CPE of all the cell types follow the same trend from -0.5 V to -0.7 V (Figure S5, Supporting Information) where the CPE is dominated by the chemical capacitance of the TiO2 film.16 This indicates that in this voltage region, the energetic distribution and density of electron trap states in the TiO2 are equal in all studied cell types.21 The capacitance of the glass PEs were in good correspondence also from -0.2 V to -0.3 V whereas the StS PEs gave much higher CPE. This difference is presently unknown

1 2π fmin

(9)

Since τeff is the inverse of the characteristic angular frequency, it can also be calculated from the fitted equivalent circuit parameters using eq 7 with τeff ) ωPE-1. The latter method was used here. In the OCVD measurements, τeff is obtained as23

(6)

where ωPE is the characteristic angular frequency calculated from the fitting parameters as18 -1/β

but may be either due to the difference in the capacitance of the substrate/electrolyte interface or density of low-energy trap states in the TiO2. 3.2.4. Counter Electrode Performance. As the charge-transfer resistance at the counter electrode/electrolyte interface RCE of the different cell types is compared as a function of the cell voltage (Figure S6, Supporting Information), it can be noted that the StS CEs show much larger charge-transfer resistance than the glass CEs which appear to have similar performance regardless of the photoelectrode type. As the RCE is plotted against current density (Figure S7, Supporting Information), the StS CEs show still considerably larger RCE than the glass CEs. Interestingly, the RCE of the StS PE and glass cells show a slight difference. The fact that RCE decreases as a function of voltage is qualitatively consistent with overpotential dependent electrode kinetics. The polarization curves of the CEs could not be calculated. First, the amount of the data points was too small which causes problems in the reliable determination of the RCE(i) function. Second, the measured points are far from the open circuit state which causes problems in the definition of the reference point VCE,0(i0). However, we note that the definition of the CE polarization should be considerably easier under illumination where RCE can be determined with good accuracy near open circuit state with VCE,0(i0 ) 0) ) 0. 3.3. Effective Electron Lifetime. An important performance characteristic of the DSC photoelectrode is the effective electron lifetime τeff. Both EIS and open circuit voltage decay (OCVD) measurements can be used for determining τeff. In the EIS spectrum, τeff corresponds to the frequency fmin at the minimum of the PE imaginary impedance peak:22

τeff ) -

( )

kBT dVOC e dt

-1

(10)

where kB is the Boltzmann coefficient, T is temperature, and e is the elementary charge. Since in the OCVD measurement the solar cell is at the open circuit state, the cell voltage equals to VPE. The transport resistance in the TiO2 layer does not affect the OCVD measurements since there is no current flow through the cell. In the case of EIS, the transport resistance is ruled out by definition, since VPE is estimated on the basis of the recombination resistance data. Figure 4 shows τeff determined both by the EIS and the OCVD methods as a function of VPE. It can be noted that τeff measured by OCVD and EIS were in good correspondence. The StS CE did not have a marked impact on τeff, as expected. Compared with the glass PE cells, which consist of the glass and StS CE cells, the StS PE cells had significantly lower electron lifetimes at PE in the voltages more negative than -0.2 V whereas in the less negative voltages than -0.2 V the lifetimes were equal. For all of the cells, τeff measured by EIS was about equal from -0.6 V to -0.7 V. Note that the τeff values of the glass PE cells settle among the highest reported for the respective PE type.23,24

Dye Solar Cells on Stainless Steel

Figure 4. Effective electron lifetimes at the photoelectrode measured with EIS in the dark and with OCVD. Dots correspond to the voltage decay measurements and symbols to EIS measurements (StS CE, black; StS PE, gray; glass, light gray).

The shapes of the glass PE curves are in good correspondence with literature.23-26 The glass PE τeff curves compose of three parts: the slopes from 0 V to -0.3 V, another from -0.5 V to -0.7 V, and the rather flat part in between. The most recent interpretation is that the slope at the most negative voltages corresponds to the recombination via TiO2 conduction band and the slope at the less negative voltages via substrate.25 In this model, recombination via TiO2 surface traps is considered to be insignificant. Another interpretation ignores the substrate effects completely and explains the response in the less negative voltages with electron transfer via surface traps.26 Both models are supported by good correspondence with experimental data.25,26 The former theory is, however, better supported by recent experimental results: First, with a recombination blocking layer at the substrate, the slope in the more negative voltages continues to the less negative voltages.25 Our experiments where the glass substrate was equipped with a 4 nm thick compact TiO2 blocking layer by atomic layer deposition confirm this result (data not shown). Second, in the less negative voltages, the charge-transfer resistance and capacitance at PE measured by EIS match with those of the substrate.16 Also, at voltages more negative than -0.5 V the EIS data display transmission line features implying dominance of the TiO2 film in the recombination reaction. Since the StS PEs had notably shorter electron lifetime already at -0.5 V where τeff is dominated by recombination from the TiO2 layer, it appears that the StS PE substrate affected the TiO2 layer. Attributing the data near 0 V to the substrate recombination,25 the recombination from the StS can be concluded to be similar with glass. Finally, it is worth commenting on the disagreement of the EIS and OCVD methods for τeff values below 0.1 s (Figure 4). Similar inconsistency has been reported also by others23 and is most likely due to the transient nature of the OCVD experiment, for two possible reasons. First, eq 10 used in the OCVD analysis is valid only under the assumption of quasi-equilibrium between trapped and conduction band electrons, 23 and this condition may not be met in reality during a fast voltage decay corresponding to low apparent τeff values. Second, in the EIS the cell was in the dark, whereas in the OCVD the first τeff data were obtained shortly after turning of the light. τeff has been reported to be lower under illumination than in the dark,27-29 and it is thus possible that the lowest τeff values reflect the transition from the illuminated to dark conditions and are not exactly characteristic to the dark conditions. The reliability of the OCVD has been previously examined in comparison to intensity modulated

J. Phys. Chem. C, Vol. 112, No. 10, 2008 4015

Figure 5. Average polarization curves of the StS SU-CE cells. Error bars indicate the standard deviation. The left branch corresponds to the reduction of the I3-.

Figure 6. Average polarization curves of the glass SU-CE cells. Error bars indicate the standard deviation. The left branch corresponds to the reduction of the I3-.

photocurrent spectroscopy (IMVS) with variable results.23,25,27 It should, however, be noted that the IMVS is performed under illumination which again raises questions in quantitative comparability of the data. 3.5. Recombination from the Substrate. Figures 5 and 6 show steady-state polarization curves of the SU-CE cells. Because of the low current densities, the charge-transfer resistance at the CE and the series resistance do not significantly influence the data in Figures 5 and 6 meaning that Vcell in practice equals the potential difference in the substrate/ electrolyte interface. The leakage current corresponding to the electron recombination reaction, that is, reduction of I3- to Iis represented in the negative voltages. The positive voltages correspond instead to the reversed reaction, that is, oxidization of I- to I3-. The leakage current was almost 1 order of magnitude smaller from the bare StS substrates compared to the glass substrates. For thermally treated glass substrates, the values reported elsewhere are one to two decades higher than here.13,30 In complete DSCs the substrate is covered with a monolayer of adsorbed dye which is why the SU-CE cells with dyed substrates illustrate best the actual recombination from substrate in a solar cell. Dye adsorption reduces the recombination current notably in the case of glass substrates (Figure 6), in agreement with previous results.14 In the case of StS substrates, the dye had an effect only at high positive voltages indicating that the dye attaches to the StS substrate, but cannot further suppress the already low recombination current. The dyed StS and glass substrates gave almost identical curves in the negative polarization and good correspondence also in the positive polarization which implies that the dye is governing the recombination. The

4016 J. Phys. Chem. C, Vol. 112, No. 10, 2008 main conclusion from the bare and dyed StS SU-CE cells is that the recombination from the StS substrate does not appear to be a problem contrary to what has been suggested previously.6 The rather good symmetry of the polarization curves with the bare substrates agrees with literature13,14,25 as does also the notable asymmetry with the dyed substrates.14 The fact that the current was not as high in the highest positive voltages compared to glass might refer to a smaller amount of adsorbed dye. The asymmetry with the dye has been explained with the catalytic effect of the dye to the charge transfer from the electrolyte to the substrate, but not to the reverse reaction.14 The porous TiO2 layer increased the recombination on average by 2 orders of magnitude (Figures 5 and 6). As the absolute current through the TiO2 coated substrate was significantly larger than that through the bare substrate, it can be concluded that the TiO2 layer was governing the features of the polarization curve of the TiO2 coated substrates. Hence, the TiO2 SU-CE cell data was calculated according to the area of the TiO2 layer (32 mm2) whereas the other SU-CE cell data was examined according to the area of the substrate in contact with the electrolyte (70 mm2). The porous TiO2 layer has been previously shown both to increase the current,14 which was the case here, and to decrease it.13 This difference may be explained by the much higher recombination current from the substrate in ref 13 compared with our case while the leakage current through the TiO2 film was about equal. Hence, the blocking effect of the TiO2 film in ref 13. The polarization curves of the TiO2 SUCE cells with StS and glass substrates show large differences (Figures 5 and 6). The StS TiO2 SU-CE cell curves were asymmetric with higher currents at the negative polarization, whereas in the glass TiO2 SU-CE cells the behavior was opposite. This difference suggests again that the StS substrate had affected the TiO2 layer. 4. Discussion The differences in the photovoltaic measurements of the StS CE solar cells compared with the glass cells, namely, the lower fill factor and thus lower efficiency of the StS CE cells, could be explained with the poorer catalytic properties of the StS CE. Both EIS and effective electron lifetime measurements gave similar characteristics for the PE in the StS CE and glass solar cells which indicates that the StS CE substrate did not have an impact on the PE, at least in the initial stage. However, longterm studies are still required before the suitability of the StS as the counter electrode substrate can be fully confirmed. The main observation in the comparison of the glass and StS PE solar cells was that although the photovoltaic performances were rather equal when measured with CE light, the other measurement techniques revealed notable differences in the voltages between -0.4 V and -0.5 V. These differences can be attributed to the dyed TiO2 film, since the data at -0.5 V corresponded to the TiO2 film in case of all the studied photoelectrodes as explained in the equivalent circuit analysis section. In other words, the dyed TiO2 layer appears to be affected by the StS PE substrate. Comparison of the solar cell and SU-CE cell polarization curves measured in the dark (Figures 7 and 8) clarifies this question. For glass photoelectrodes in Figure 8, the addition of the dye suppressed the recombination current through the TiO2 layer. The curve of the glass PE solar cells is at the more negative voltages close to the uncoated TiO2 and in the less negative voltage closer to the dyed substrate. It appears that the dye has a potential dependent recombination blocking effect on the TiO2 film.

Miettunen et al.

Figure 7. Average polarization curves of the StS PE solar cells and SU-CE cells with either dyed or undyed porous TiO2 coated StS substrate measured in the dark. The error bars indicate the standard deviation and are shown only when exceeding the marker size.

Figure 8. Average polarization curves of the glass solar cells and SUCE cells with either dyed or undyed porous TiO2 coated glass substrate measured in the dark. The error bars indicate the standard deviation and are shown only when exceeding the marker size.

Also, in the case of StS PE cells, the dye decreased the current through the TiO2 layer but the effect was not that large compared with the glass cells, especially at low potentials. In fact, the StS PE solar cell curve has a similar slope as the one of TiO2 coated substrate (Figure 7). As the solar cell curves illustrate such large differences, it can be concluded that the StS PE substrate had an impact on the TiO2 layer and/or the dye. The SU-CE cell results indicate, however, that the substrate did not have an effect on the dye adsorbed onto it. Combining these observations, we can hypothesize that the StS PE substrate interacts with the TiO2 layer but not with the dye and the dye in the StS PE solar cells merely covers a certain area of the StS-affected TiO2 surface and blocks the current from those parts while the current leakage through the other recombination pathways stays the same. The mechanism which causes the StS PE to affect the TiO2 layer is not yet known, but it could be linked with the same effect which eventually leads to cell degradation.3,9 5. Conclusions The aim of this study was to examine the suitability of the StS substrates in DSCs with respect both to performance and to stability. It was shown that relatively high efficiencies can be obtained with DSC deposited on stainless steel substrate even without any underlayer coatings. Recombination from the StS substrate is equal to or less than that from the glass substrates. The tests performed indicate that there is no notable contamination of the porous TiO2 layer at least during the cell preparation.

Dye Solar Cells on Stainless Steel The presented EIS data analysis method based on the cell current enables quantitative comparison of single electrode performance between different types of cells. Most importantly, the polarization curve of the PE recombination reaction can be determined from complete standard solar cells without additional reference electrodes. Here, similarity of the StS CE and glass cell PEs could be confirmed as well as the difference between the StS PE and the glass cells. Conventional EIS analysis based on the cell voltage would not only have been insufficient in certifying these effects but also have led to erroneous conclusions especially in the case of the StS CE cells. These observations stress the importance of proper analysis method. The results indicate that the StS substrate in the StS PE cells has an effect on the electrochemical function of the TiO2 layer. Systematic stability studies with help of the presented methods are required to find out whether this effect can be linked to a cell degradation over time or to a difference in stable physical properties, for example, at the substrate/TiO2 film interface. Acknowledgment. This study was partly funded by the Finnish Funding Agency for Technology and Innovation (Tekes) and the Nordic Energy Research Program. K.M. is grateful for the scholarship of the Graduate School of Energy Technology. We thank Mr. Olli Himanen for the assistance in the preparation of the OCVD equipment and Planar Inc. for the atomic layer deposition of the blocking layers. Supporting Information Available: The structure of the solar cells (Figure S1), the structure of the SU-CE cells (Figure S2), typical current-voltage curves of the solar cells in 1 sun illumination (Figure S3), calculated polarization curves of the photoelectrode recombination reaction in the dark (Figure S4), equivalent capacitances of the photoelectrode CPEs of individual cells (Figure S5), RCE as a function of cell voltage in the dark (Figure S6), RCE as a function of cell voltage in the dark (Figure S7), and the detailed equivalent circuit analysis. This information is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Kroon, J. M.; Bakker, N. J.; Smit, H. J. P.; Liska, P.; Thampi, K. R.; Wang, P.; Zakeeruddin, S. M.; Gra¨tzel, M.; Hinsch, A.; Hore, S.; Wu¨rfel, U.; Sastrawan, R.; Durrant, J. R.; Palomares, E.; Pettersson, H.; Gruszecki, T.; Walter, J.; Skupien, K.; Tulloch, G. E. Prog. PhotoVolt: Res. Appl. 2007, 15, 1-18. (2) Toivola, M.; Ahlskog, F.; Lund, P. Sol. Energy Mater. Sol. Cells 2006, 90, 2881-2893. (3) Ma, T.; Fang, X.; Akiyama M.; Inoue K.; Noma H.; Abe E. J. Electroanal. Chem. 2004, 574, 77-83.

J. Phys. Chem. C, Vol. 112, No. 10, 2008 4017 (4) Fang, X.; Ma, T.; Akiyama M.; Guan, G.; Tsunematsu, S.; Abe, E. Thin Solid Films 2005, 472, 242-245. (5) Kang, M. G.; Park, N.-G.; Ryu, K. S.; Chang, S. H.; Kim, K.-J. Chem. Lett. 2005, 34, 804-805. (6) Kang, M. G.; Park, N.-G.; Ryu, K. S.; Chang, S. H.; Kim, K.-J. Sol. Energy Mater. Sol. Cells 2006, 90, 574-581. (7) Jun, Y.; Kim, J.; Kang, M. G. Sol. Energy Mater. Sol. Cells 2007, 91, 779-784. (8) Onoda, K.; Ngamsinlapasathian, S.; Fujieda, T.; Yoshikawa, S. Sol. Energy Mater. Sol. Cells 2007, 91, 1176-1181. (9) Miettunen, K.; Toivola, M.; Halme, J.; Armentia, J.; Vahermaa, P.; Lund, P. Proceedings of 22nd European PhotoVoltaic Solar Energy Conference 3-7 September 2007, Milan, Italy, 512-515. (10) Kay, A., Ph.D. Thesis, Ecole Polytechnique Fe´de´rale de Lausanne, Switzerland, 1994. (11) Ito, S.; Nazeeruddin, K.; Liska, P.; Comte, P.; Charvet, R.; Pe´chy, P.; Jirousek, M.; Kay, A.; Zakeeruddin, S.; Gra¨tzel, M. Prog. PhotoVolt: Res. Appl. 2006, 14, 581-601. (12) Papageorgiou, N.; Maier, W. F.; Gra¨tzel, M. J. Electrochem. Soc. 1997, 144, 876-884. (13) Cameron, P. J.; Peter, L. M.; Hore, S. J. Phys. Chem. B 2005, 109, 930-936. (14) Sastrawan, R.; Renz, J.; Prahl, C.; Beier, J.; Hinsch, A.; Kern, R. J. Photochem. Photobiol. A: Chem. 2006, 178, 33-40. (15) Park, J.; Koo, H.-J.; Yoo, B.; Yoo, K.; Kim, K.; Choi, W.; Park, N.-G. Sol. Energy Mater. Sol. Cells 2007, 91, 1749-1754. (16) Fabregat-Santiago, F.; Bisquert, J.; Garcia-Belmonte, G.; Boschloo, G.; Hagfeldt, A. Sol. Energy Mater. Sol. Cells 2005, 87, 117-131. (17) Fabregat-Santiago, F.; Bisquert, J.; Palomares, E.; Otero, L.; Kuang, D.; Zakeeruddin, S. M.; Gra¨tzel M. J. Phys. Chem. C 2007, 111, 65506560. (18) Bisquert, J.; Garcia-Belmonte, G.; Fabregat-Santiago, F.; Ferriols, N. S.; Bogdanoff, P.; Pereira, E. J. Phys. Chem. B 2000, 104, 2287-2298. (19) Hoshikawa, T.; Yamada, M; Kikuchi, R.; Eguchi, K. J. Ele. Chem. 2005, 577, 339-348. (20) Lobato, K.; Peter, L. M.; Wu¨rfel, U. J. Phys. Chem. B 2006, 110, 16201-16204. (21) Bisquert, J. Phys. Chem. Chem. Phys. 2003, 5, 5360-5364. (22) Kern, R.; Sastrawan, R.; Ferber, J.; Stangl, R.; Luther, J. Electrochim. Acta 2002, 47, 4213-4225. (23) Zaban, A.; Greenshtein, M.; Bisquert, J. ChemPhysChem 2003, 4, 859-864. (24) Quintana, M.; Edvinsson, T.; Hagfeldt, A.; Boschloo, G. J. Phys. Chem. C 2007, 111, 1035-1041. (25) Cameron, P. J.; Peter, L. M. J. Phys. Chem. B 2005, 109, 73927398. (26) Bisquert, J.; Zaban, A.; Greenshtein, M.; Mora-Sero´, I. J. Am. Chem. Soc. 2004, 126, 13550-13559. (27) Boschloo, G.; Hagfeldt, A. J. Phys. Chem. B 2005, 109, 1209312098. (28) Montanari, I.; Nelson, J.; Durrant, J. R. J. Phys. Chem. B 2002, 106, 12203-12210. (29) Bauer, C.; Boschloo, G.; Hagfeldt, A. J. Phys. Chem. B 2002, 106, 12693-12704. (30) Cameron, P. J.; Peter, L. M. J. Phys. Chem. B 2003, 107, 1439414400.