Understanding Alternative Solar Cell Concepts The Application of Intensity Modulated Photo Spectroscopy in Combination with EIS Practical Course 1 Dr. Michael Multerer Kronach Impedance Days 212
KIT 212 CIMPS 1 General Light induced processes like photosynthesis are of fundamental significance in the world. In technique, the photo-electric effect occurring at illuminated interfaces like semiconductor junction barriers is of great importance. Here, the electric potential and the current flowing at the interface site, are depending on the illumination. Light intensity with photo-voltage on the one hand and with photo-current on the other hand are building two typical force-response couples for a dynamic transfer function analysis. Like the analysis of the impedance transfer function in the EIS, the Intensity Modulated Photo Spectroscopy can help understanding the complex systems, in particular when photo-electric effects are playing a significant role. Aim of the experiment CIMPS is based on the idea of the dynamic analysis of the photo-effect by means of small-signal linear sinusoidal modulation technique. The CIMPS experiment shall demonstrate, how the two frequency domain transfer functions and photo-voltage efficiency = photo-voltage / intensity photo-current efficiency = photo-current / intensity can be determined by the application of modulated light with swept frequency on a photo-electric active object. The meaning and the general behaviour of both transfer functions shall be explained, together with the corresponding EIS. The importance of a stable, controlled intensity together with high modulation linearity during the whole experiment shall be emphasized (the reason for the name CIMPS = Controlled Intensity Modulated Photo Spectroscopy) and the principle of the CIMPS feedback will be explained. The typical dynamic response of an ideal semiconductor barrier junction will be discussed. Starting from that, the effects, limiting the photo-efficiency of alternative technique solar cells can be identified in the transfer function and their appearance and properties can be interpreted. Test procedures One of three different experimental set-ups will be demonstrated dependent on the availability. The first one is based on a photosensitive semi-conducting iron oxide (hematite, Fe 2 O 3 ) layer, the second on dye sensitized solar cells based on TiO 2 or ZnO, and the third one on a solar cell based on organic polymers. For the first experiment, an electrochemical cell with a light inlet at the spiral Ptcounter electrode (Photo-Electro-Chemical Cell = PECC ) is equipped with an Ag- AgCl reference electrode and an.5 mol/l NaNO 2 aqueous electrolyte. A square Fe metal sheet ( 2 cm 2 ) which was tempered some minutes at 6 C to form a hematite-type oxide layer, acts as working electrode sample. A potentiostatic EIS (KHz to.1hz, starting at 1KHz, 5mV amplitude) is recorded at the OCP under darkness. Then, the 455nm-LED source with a light intensity of about 5 to
KIT 212 CIMPS 2 W/cm 2 is switched on at the window of the PECC. Another EIS is recorded at the same potential after the settling of the photo-current under light. An AC modulation of some % is now superimposed to the light and the photo-current efficiency spectrum is recorded (1KHz to.1hz, starting at Hz). The DC-photocurrent at the end of the run is now used as set current for the PECC and the potentiostat is switched to galvanostatic mode. This procedure shall ensure, that the following photo-voltage efficiency spectrum is recorded under the same system conditions as the spectra before. The second set-up uses a PECC-2 appropriate for the DSSC. This cell has light entrance and outlet both from the counter electrode as well as the rear side enabling the illumination of the working electrode to reach the Oxide-layer through the ITOcovered glass carrier. Alternatively completely assembled thin-film cells may be used. The DSSC components are fabricated either by the Material Science Department of the Erlangen University, the Freiburger Material Forschungsanstalt (FMF) or the LNQE together with the Inst. f. Phys. Chemie of the Hannover University. A sketch of the set-up of the cell and the dye sensitized solar cell (DSSC) is shown in figure 1. Figure 1: Scheme of cell and DSSC (Grätzel type) The DSSC consists of nanoporous TiO 2, with a thickness of about 6 µm impregnated with a.3 mol/l solution of N719 dye in ethanol, a solution of.5 mol/l LiI and.5 mol/l I 2 in acetonitrile as electrolyte, and a transparent Indium-Tin-Oxide (ITO) coating. Alternatively a complete DSSC, built from TiO 2 -nanorods on Ti-foil will be used. In the third set-up an Organic Solar Cell from the bulk-hetero-junction type is mounted on the site of the PECC. The working principle of this OSC is sketched in figure 2. A PEDOT:PSS [poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate)] film as hole-conductor is covered with a polymer photon absorber P3HT [poly(3- hexylthiophene)] blended with the fullerene based electron-acceptor PCBM [(phenyl- (6, 6_)-C61)-butyric acid methyl ester]. Excitons created by the light in the P3HT backbone dissociate and pass their electrons to the PCBM. Electrons and holes follow their individual work-function staircase across the different contact materials to reach selectively the appropriate electrodes. The physical structure is depicted in figure 3.
KIT 212 CIMPS 3 Figure 2: Working principle of the OSC under test. Left hand side: bulk hetero-junction photoactive layer material blend. Right hand side: energy scheme. light substrate Glass-Cr-Al-Cr-(P3HT-PCBM-BLEND)-PEDOT:PSS-Au Figure 3: Structure of the OSC under test. Left hand side: photograph of the illuminated side. Right hand side: schematic of the different layers.
KIT 212 CIMPS 4 Test procedure using the DSSC In figure 4, the impedance measurements expected at open circuit potential (OCP) considering different illumination densities at a wavelength of λ= 53 nm are shown. impedance / Ω 1M K 1K 1K 1 dark 1 W/m 2, 53nm 5 W/m 2, 53nm 55 W/m 2, 53nm 1m 1 1K 1M frequency / Hz 9 75 6 45 3 15 phase / Figure 4: Impedance spectra at OCP and λ= 53 nm as a function of the illumination intensity At OCP one finds a decreasing low frequency impedance resulting from an increased forward biasing of the photo-junction with increasing illumination intensity. Therefore, this type of experiment is suitable to get access to the photo-junction impedance and suggests that the recombination of photo-generated charge is the major mechanism of dissipation. In figure 5, the impedance spectra expected at fixed potential (darkness OCP) as a function of different illumination densities is plotted. The dominant contributions in the observed spectra is the polarization of the electrochemical double layer accompanied by a small forward biasing of the photo-junction. From this observation you can conclude that this type of experiment is well suited to determine especially the chemical properties, i. e. the double layer impedance and therefore offers an access, especially to this parameter. 9 impedance / Ω 1M K 1K 1K Φ [W/m 2 ] 1 5 55 I [µa] 9.24 51.3 5 75 6 45 3 phase / 15 1m m 1 3 1 1K 1K K frequency / Hz Figure 5: EIS at fixed potential (darkness OCP) as a function of different illumination densities
KIT 212 CIMPS 5 In figure 6, the dynamic photo-current measurements (IMPS) expected at fixed potential (darkness OCP) under different illumination intensities are depicted. After the mostly accepted theories [1-4], the transfer function should then be dominated by the time constant of the photo-electron diffusion. However, it should not be neglected, that the dynamic behavior is also determined by the serial contributions of the impedance network affecting the way from the photocurrent generating process ( current source ) [5] to the electrode. 1µ photocurrent / AW -1 m 2 3µ 1µ 3n n 3n Φ [W/m 2 ] 1 5 55 I [µa] 9.24 51.3 5 8 6 4 2 phase / 1m m 1 3 1 1K 1K frequency / Hz Figure 6: Dynamic photo-current measurements (IMPS) at fixed potential (darkness OCP) under different illumination intensities The dynamic photo-voltage measurements (IMVS) at fixed current (under the system conditions described above, the same current, which is observed in the IMPS experiments is used as set-current), expected under different illumination intensities are shown in figure 7. At constant current, the dynamic behaviour of the photo voltage is mainly determined by the photocurrent generating process and the impedance of the photoactive layer acting as a shunt impedance in parallel to the current source.
KIT 212 CIMPS 6 15 photovoltage / VW -1 m 2 m 1m 1µ Φ [W/m 2 ] 1 5 55 I [µa] 9.24 51.3 5 5 phase / 1m m 1 3 1 3 1K frequency / Hz Figure 7: Dynamic photo-voltage measurements (IMVS) at fixed potential (corresponding to the IMPS measurements) under different illumination intensities Following the literature, another regime for measuring the photovoltage spectrum should be used: at the OCP under illumination (no outer current flowing), all photoelectrons created must be consumed inside the cell by back-reactions. Following the mostly accepted theories, the dominating time constant then should be determined by the recombination kinetic of the photoelectrons. However, similar to the situation for the photocurrent, it should not be neglected, that the dynamic behaviour is also affected by impedance network parts effective in parallel to the current source. Generally one can conclude that considering a common model, each of the presented electrochemical techniques emphasizes a certain part of the model. Consequently, the assignment of the components within the model is facilitated, provided that the stability of the system during the experiments is ensured. Especially, the illumination conditions have to be controlled properly. This is the task of the CIMPS set-up presented. For instance in figure 8, the Nyquist representation of a photo-voltage- (red) as well as the corresponding photo-current (blue) spectrum of the series was simulated by the equivalent circuit depicted on the bottom side of the figure.
KIT 212 CIMPS 7 photo-voltage ' /mvw -1 m 2 2 2 3 4 25 2 photo-current '' /µaw -1 m 2-2 -4-6 15 5-5 - -15 photo-voltage '' /mvw -1 m 2-2 2 4 6 8 photo-current ' /µaw -1 m 2 Figure 8: Upper side: Nyquist representation of a photo-voltage- (red) as well as the corresponding photo-current (blue) spectrum; lower side: Corresponding equivalent circuit used to model the data; 1: Photo current source, 2: Forward resistance, 3: Diffusion impedance, 4: Chemical capacity, 5: Charge transfer resistance, 6: Double layer capacity and 7: Electrolyte resistance
KIT 212 CIMPS 8 Conclusion for EIS, IMPS/IMVS application on DSSC The important dynamic behavior of one component in the network, what is the photocurrent source, reflects the time constants of the photo-induced charge carrier lifetime and the diffusion mobility. It is claimed, that the photo-induced charge can reach the electrodes directly with a drift speed of typically meters/s. Due to the usually porous character of the oxide (needed to provide high dye loading for effective light absorption), the electrons have to go long ways to the anode by diffusion through the porous system with a low effective mobility. Electron diffusion is in kinetic competition with recombination at the electrolyte-oxide-interface. Therefore one is interested on the time constants found in IMVS at OC conditions and in IMPS at short circuit conditions. It is assumed, that the first one is associated with the recombination t r and the second one t c is associated to the diffusion kinetics of the photoelectrons. The term 1 - t c / t r determines the efficiency. In literature often a very simple access to this parameters is proposed. It is based on the assumption, that the overall photocurrent response at short circuit conditions is dominated by t c, and the overall photo-voltage response at OCP conditions is dominated by t r. Such a procedure neglects the influence of other contributions to the overall transfer function. Bay and West [5] proposed therefore an alternative DSSC model, which directly takes into account the porous structure of the photo anode material. 1. P.E. Jongh, D. Vanmaekelbergh, J. Phys. Chem. B 11 (1997) 2716-2722 2. L. Dloczik, O. Ileperuma, I. Lauermann, L.M. Peter, E.A. Ponomarev, G. Redmond, N.J. Shaw, I. Uhlendorf, J. Phys. Chem. B 11 (1997) 1281-1289 3. G. Schlichthoerl, N.G. Park, A.J. Frank, J. Phys. Chem. B 14 (2) 244-252 4. R. Kern, R. Sastrawan, J. Ferber, R. Stangl, J. Luther, Electrochim. Acta 47 (22) 4213-4255 5. L. Bay, K. West, Solar Energy Materials & Solar Cells, 87 613-628 (25).
KIT 212 CIMPS 9 Test procedure using the OSC The charge transfer- and migration kinetics of a bulk-heterojunction OSC is more complex than that of an oxide-based DSSC. Excitons-recombination through different intermediates, excitons diffusion to the electron-acceptor and their charge separation, photo-electron and hole migration to the electrodes appear on a time scale, usually not accessible by IMPS/IMVS. The simple straight-forward strategy to assign the IMPS dominating high frequency time constant to the rate determining charge diffusion/migration process and the IMVS dominating low frequency time constant to the loss determining recombination process does therefore make not much sense. A much more promising way is to use EIS in combination with IMPS/IMVS in a similar way like in fuel cell and battery research in order to understand the complete network by means of a full model simulation and analysis. This procedure is very successful in fuel cell and battery research since long, but nevertheless it suffers there from the ambiguity of impedance models: different models may lead to identical spectra. Additional evaluation methods are necessary in order to assign the resulting data unequivocally. Fortunately the situation regarding ambiguity can be solved easily in solar cell research. If one is able to explain the triplet of EIS, IMPS and IMVS data with one common dynamic model, the interpretation is already unequivocal with high probability. The reason is, that every individual method sees both common but also different aspects from the object properties, which are invisible for the other methods. For example, the position within the network description and the parameter values of the element (photocurrent source), representing the photoelectric process, are invisible for EIS, but strongly significant for IMPS/IMVS. Another example is, that shunt current paths, appearing in parallel to the object, have no influence on the IMPS, and series impedance contributions have no influence on the IMVS spectrum, while both affect the EIS. At the beginning of the experiment, a potentiostatic EIS (KHz to 1Hz, starting at 1KHz, 5mV amplitude) is recorded under light at short circuit conditions. For the illumination, a 455nm-LED source with an intensity of about W/cm 2 is used. An AC modulation of some % is now superimposed to the light and the photo-current efficiency spectrum is recorded (5KHz to 1Hz, starting at 1KHz). The DC-photocurrent at the end of the run is now used as set current for the PECC and the potentiostat is switched to galvanostatic mode. This procedure shall ensure, that the following photo-voltage efficiency spectrum is recorded under the same system conditions as the spectra before. The phenomenology of the measured spectra (figure 9 left) is then compared with spectra, expected from ideal solar cells (photo-diodes). The spectra are processed similar to the analysis of impedance spectra in the SIM software. A model, developed for the OSC is used for the TRIFIT-procedure. This procedure is able to fit the three different data sets from EIS, IMPS and IMVS to one common fit optimum by means of a model, which must be able to determine all three transfer functions. A scheme of TRIFIT is sketched in figure 9 right. Finally, the model will be discussed and an assignment of components and processes will be tried (figure 1).
KIT 212 CIMPS 1 Z / Ω PC eff. / AW -1 PC eff. / VW -1 phi / o 9 Multiple Function Complex Nonlinear Least Squares Fit Gradient calculation ZCV skalar deviation sum 1_osc_pz 2_osc_pc 3_osc_pv 75 6 Measured spectra: Z: impedance C: photo-current V: photo-voltage Model Deviations Parameter variation ZCV(t,ω,P n ) fit loop 1 45 Z ( ω i ) Σ ( Z(ω i ) - ZCV(,ω i ) ) 2 C ( ω j ) Σ ( C(ω j ) - ZCV(1,ω j ) ) 2 Σ 1m 3 V ( ω k ) Σ ( V(ω k ) - ZCV(2,ω k ) ) 2 µ 15 2 5 3 1K 3K 1K 3K frequency / Hz Figure 9: Right hand side: experimental results from EIS, IMPS and IMVS of the OSC. Left hand side: scheme of the Thales SIM TRIFIT CNLS joined fitting procedure for three different kinds of spectral transfer functions: impedance, dynamic photocurrent and dynamic photovoltage. 7 4 N 5 User 1 #1 2 3 6 Photo-active layer: 1 Photocurrent source, breakpoint frequencies 2 blocked hole diffusion in the porous layer 3 electrostatic capacity Polymer anode: 4 finite hole diffusion in the anode layer 5 electrostatic capacity Shunt loss 6 shunt resistance Series loss 7 contact resistance 1 21.85 µa 8.53 MHz 9.18 MHz 2 739.3 KΩ s -1/2 7.47 Ks -1 3 85.5 nf 4 1.62 KΩ s -1/2 118.4 Ks -1 5 23.9 nf 6 3.261 KΩ 7 9.435 Ω Figure 1: Assignment of the bulk heterojunction OSC phases and processes to a dynamic equivalent circuit model.
KIT 212 CIMPS 11 2K impedance / Ω phase / o 135 photocurrent / AW -1 m 2 1m 3µ phase / o 135 3m m photovoltage / VW -1 m 2 phase / o 135 1K 5 9 µ 3µ 9 3m 1m 9 2 5 2 5 3 1K 3K 1K 3K frequency / Hz 45 1µ 3µ 1µ 2 5 3 1K 3K 1K 3K frequency / Hz 45 3m 1m 3µ 2 5 3 1K 3K 1K 3K frequency / Hz Figure 8: from left to right measured EIS, IMPS and IMVS spectra of the OSC under test (symbols) and TRIFIT fitting results after the model from figure 1 (solid lines). 45 Conclusion for the OSC EIS, IMPS and IMVS spectra of organic solar cells can be evaluated successfully with the strategy popular in fuel cell and battery research.