Materials, Interfaces, and Photon Confinement in Dye-Sensitized Solar

14582

J. Phys. Chem. B 2010, 114, 14582–14591

Materials, Interfaces, and Photon Confinement in Dye-Sensitized Solar Cells† Byunghong Lee,‡ Dae-Kue Hwang,‡ Peijun Guo,‡ Shu-Te Ho,‡,§ D. B. Buchholtz,‡ Chiu-Yen Wang,§ and R. P. H. Chang*,‡ Materials Research Institute, Department of Materials Science and Engineering, Northwestern UniVersity, EVanston, Illinois 60208, and Department of Materials Science and Engineering, National Tsing Hua UniVersity, Hsinchu 30013, Taiwan ReceiVed: March 15, 2010; ReVised Manuscript ReceiVed: June 5, 2010

A series of experiments have been carried out to study the effects of materials quality, surface and interfacial modification, and photon confinement on standard dye-sensitized solar cells. For these studies, both physical and optical characterization of the materials has been performed in detail. In addition, DC and AC impedance measurements along with kinetic charge-transport modeling of experimental results have yielded information on how to systematically optimize the cell efficiency. The same kinetic model has been used to interpret the results of a series of experiments on interfacial modification studies using fluorine etching in combination with TiCl4 surface treatment. By using specially designed photonic crystals to confine the photons in the cells, it is shown that the best cell efficiency can be further increased by about 13%. 1. Introduction Dye-sensitized solar cells (DSSCs) are good examples of where the quality of the nanomaterials and their interfacial properties are important to device performance. Numerous publications and review articles on DSSC have appeared in the literature during the past two decades.1-4 Prototype cells for manufacturing are actively under development.5-9 Most papers have dealt with certain aspects of the DSSC fabrication or performance. In this work, we report a systematic optimization of key parameters and processing steps in a typical DSSC to produce high cell efficiency. Our effort is guided by DC electrical measurements and AC impedance measurements and modeling.10-12 From our studies, we note that the quality and purity of our starting materials, their surface properties, and interfacial materials compatibility play significant roles in determining the efficiency of the cell. In addition, we demonstrate the importance of improving the photon confinement of the cell. In this article, we start with a discussion of the principle of cell operation, followed by materials preparation, processing, device fabrication and assembly, measurements, and device modeling. A discussion of the experimental results will include detailed studies of interfacial properties via controlled experiments and the optimization of photon confinement of a typical cell. The paper will conclude with a summary of our findings and suggestions for further research and development. 2. DSSC Operational Model The principle of operation of a DSSC is well documented in the literature.13-15 Photons, absorbed by the dye molecules which are coated around the interconnecting TiO2 nanoparticles, generate excitons. Electrons and holes are formed upon the separation of the excitons. The electrons are conducted through a interconnecting semiconducting TiO2 nanoparticle thin film †

Part of the “Michael R. Wasielewski Festschrift”. * To whom correspondence should be addressed. ‡ Northwestern University. § National Tsing Hua University.

Figure 1. (a) Simplified schematic diagram of the principle of operation of a dye-sensitized solar cell (DSSC). (b) A simple equivalent circuit of a DSSC. (c) Typical current-voltage curve for a solar cell based on N719 under AM 1.5 simulated sunlight (100 mW cm-2).

to the anode electrode, while the holes are transported away through the redox electrolyte of iodide/tri-iodide in contact with the platinum catalyst at the cathode electrode of the cell (see illustration in Figure 1a). Under steady illumination, the DC cell response has been modeled by a simple circuit, as illustrated in Figure 1b. Device parameters of special interest to cell performance in this model include open-circuit voltage (Voc), short-circuit current density (Jsc), and the fill factor (FF) (see Figure 1c for an ideal J-V curve). Using these basic parameters, we can optimize the DSSC efficiency, η, which is defined as η ) JscVocFF/Ps, where Ps is incident light power density. To maximize cell efficiency, the circuit in Figure 1b tells us that we need to have the series resistance, Rs, be as low as possible while maintaining the shunt resistance, Rsh, as high as possible. However, the dominant effect comes from lowering the value of Rs. This simple and ideal circuit guides us to quickly study and optimize the cell operation in our experiments, as will be discussed below. In order to have a more in-depth understanding of the cell operation, models that include charge-transport kinetics have

10.1021/jp102359r  2010 American Chemical Society Published on Web 07/15/2010

Materials, Interfaces, and Photon Confinement in DSSCs

J. Phys. Chem. B, Vol. 114, No. 45, 2010 14583

1 1 + iωCp1 rp1

Z1 )

where, rp1 and Cp1 represent the resistance and capacitance at the Pt surface, respectively. Z3 represents the contribution from the finite Warburg impedance describing the diffusion of tri-iodide ions in the electrolyte.17,21

Z3 ) RD

Figure 2. (a) A representative electrical equivalent of a typical DSSC. (b) A schematic Nyquist plot of a DSSC with R0, R1, R2, and R3 and each peak frequency ωZ1, ωZ2, and ωZ3, respectively.

been developed to interpret measured data. In this study, we adopt the work of Adachi,16 who has shown the equivalence of the models developed by Kern et al.17 and Bisquert et al.18,19 In this model, a small-amplitude oscillation is used to perturb the DC operation of the cell under steady light illumination. From this perturbation, the complex impedance of the cell can be measured and analyzed. Through numerical fitting of the experimental data with this model calculation, it is then possible to determine the values of the parameters and decipher their contribution to the cell operation. Clearly, the more precise the description of a model, the easier it is to identify which parameters to optimize to achieve higher cell efficiency. Figure 1a gives the schematic diagram of our DSSC construction. The light enters from the anode electrode (left) made of frosted fluorine-doped tin oxide (FTO), which has a sheet resistance of RFTO ) 8Ω/0.20 On top of this electrode, a thin porous film (∼10 micrometers thick) of interconnected TiO2 nanoparticules is formed and covered with interpenetrating dye molecules and iodine electrolyte. The cathode electrode (right) is made of platinum-coated FTO substrate. The platinum film serves as a catalyst to enhance the redox activity of the electrolyte. Indium contacts are used with both FTO electrodes for the extraction of charges to the outside world. The complex impedance of the cell is the sum of each of the components given in Figure 2a. Z0, Z1, Z2, and Z3 are respectively the contact impedance (which is usually real, Zo ) Ro); the Pt-catalyzed counter electrode impedance; the complex impedance, which represents the interface among the semiconductor (TiO2), dye molecule, and iodine electrolyte; and the diffusion of the triiodide-ion-related (Warburg) impedance. Thus, the total Z is equal to the sum of Z0 +Z1 + Z2 + Z3. An idealized plot of the real part of Z, Z′(ω), versus the imaginary part of Z, Z′′(ω), as a function of the modulation frequency for a set of operating parameters is given in Figure 2b. The AC impedance measurements provide us with the information on the internal properties of the cell device. The measured impedance values (discussed below) are intimately connected to materials quality and interfacial properties, as well as the charge-transport kinetics of the cell. We have set Zo ) Ro since it represents the resistance from the FTO and the metal ohmic contact, which only has a real part of the impedance. Following Adachi, Z1 represents the impedance of the electron transfer at the Pt counter electrode, and it can be simply described by a RC circuit

1

tanh

iω (D1 /δ2)

iω (D1 /δ2)

where

RD )

kBT 2 2

m q AVC*D1δ

D1 and δ represent the diffusion coefficient of I3- and the thickness of the liquid film, respectively. The number of electrons transferred in each reaction, m, is 2 in this case. Aυ and C* are the Avogadro number and the concentration of I3in the bulk, kB is the Boltzmann constant, T is the temperature, and q is the charge. The most important contribution to Z comes from Z2, which connects incident photon to charge generation and recombination and the kinetics of charge transport. The basic premise for modeling Z2 is to consider the kinetics surrounding the TiO2 semiconductor (see Figure 3). The dye molecules inject electrons into the TiO2 conduction band as a result of photon absorption and exciton breaking on their surfaces. Some of the electrons in the conduction band, however, will decay at a rate of k1 into surface trap states of the TiO2. In addition, the charges in the trap state may be excited into the conduction band at a rate of k2. At the same time, they may decay and recombine with holes in the iodine electrolyte at a rate of kr. This model ignores all other reaction rates as being too slow in comparison. From this simple model, one formulates the transport equations for the conduction band electrons and the electrons in trap states with added terms describing injection and diffusion. We adopt here the equations used by Adachi et al. for model calculations in this work to compare with our experimental data. The tansport equations (for the conduction band electrons and for the electrons in the trap state) that describe injection, diffusion, collection, trapping, detrapping, and recombination of electrons in the TiO2 of the DSSC are given as follows16,17

∂2n(x, t) ∂n(x, t) - k1n(x, t) + k2N(x, t) + G(x, t) ) Dcb ∂t ∂x2 ∂N(x, t) ) -k2N(x, t) - krN(x, t)2 + k1n ∂t where n can express the excess electron density in the conduction band of the TiO2 under illumination, N expresses the excess electron density of the trap sites, Dcb represents the diffusion coefficient of an electron in the conduction band, and the function G provides the generation rate of electrons injected

14584

J. Phys. Chem. B, Vol. 114, No. 45, 2010

Lee et al. By defining

ωd )

Deff L2

and

γL ) Figure 3. A simplified kinetic model for calculating Z2 in a DSSC.

into the TiO2 due to the sensitization by the Ru dye.17 For the case of an externally applied harmonically modulated AC voltage, the excess conduction band electron density n(x,t) will be of the form16

n(x, t) ) ns(x) + ∆n(x)eiωt N(x, t) ) Ns(x) + ∆N(x)ei(ωt+φ) where ns and Ns are the steady-state electron densities in the conduction band and in the trap state, respectively. Here, ∆n and ∆N are the amplitudes of the modulated component of the conduction band and trap state electron density, respectively. Defining

k2 k1

Deff ) Dcb

keff ) 2Nskr γ2 )

keff iω + Deff Deff

and using the following boundary conditions

qDeff

∆I ) ) ∆J ( ∂∆n ∂x ) A ∂∆n )0 ∂x

at x ) 0

at x ) L

where L is the total cell thickness, and the impedance Z2 as obtained by Kern et al. is11,16

Z2 ) -S

1 qA

1

Deffγ

1 + e2γL 2γL 1 1-e keff

where

S)

kB T qns

1 keff

ωk iω + ωd ωd

the equivalent impedance Z2 of Bisquert16 is obtained as follows

(

Z2 ) Rω

Rw )

1 (ωk /ωd)(1 + iω/ωk)

kBT L L ) Con , Deff q2An Deff s

)

1/2

coth[(ωk /ωd)(1 + iω/ωk)]1/2 Rk )

ωd × Rω ωk ) Con

1 Lkeff

with ωk ) keff. The latter is the recombination rate constant, which is estimated to be equal to the peak frequency of the central arc, ωmax. We also have the relationship Rk ) (ωd/ ωk) · Rw. Therefore, Z2 becomes a function depending on Rw and Rk. Rw indicates the electron-transport resistance, and Rk indicates the charge-transfer resistance related to recombination of electrons at the TiO2/electrolyte interface. Using the above model for Z, we can fit our data obtained from the electrochemical impedance spectroscopy. Taking the maxima values of ωZ1, ωZ2, and Z3 and the circle diameters along the Z′ axis and using the experimental parameters (L,δ), it is then possible to extract important information from our cell measurements, such as the impedance of the platinum electrode, the diffusion coefficient of the tri-iodide in the electrolyte, the diffusion coefficient of electrons in the conduction band of the TiO2, transition rates for recombination, and so forth. The above model was used by Adachi et al. to study the effects on cell efficiency as a function of cell thickness, electrolyte composition, TiO2 materials, and the intensity of the incident light.16 In a similar manner, we have used the above model as a guide (via curve fitting) to optimize our cell performance through a series of iterative fabrication and measurement processes. Taking our optimized processing recipe as the standard cell fabrication procedure, we report in this paper the following studies: (1) an investigation of the importance of interfacial effects on cell performance and (2) how to increase cell efficiency by improving photon management. 3. Cell Preparation, Measurements, and Results In fabricating the cells, we selected the best materials that we could obtain to optimize the cell performance. This required us to synthesize our own TiO2 anatase nanoparticles (NPs), to purify purchased chemicals, and mixed electrolytes following the best literature recipe.22,23 In addition, we have developed our own processing techniques after many months of iterative studies through measurements and model calculations. 3.1. Materials Preparation and Characterization. A twostep autoclaving technique was applied to obtain the high-purity



Figure 4. Illustrations showing our fabrication procedure for DSSCs using highly crystalline TiO2 nanoparticles.

Figure 5. X-ray diffraction patterns of hydrothermally grown TiO2 nanoparticles before and after sintering at 500 °C. The inset shows the SEM images of nanopaticles after sintering.

anatase TiO2 NPs. A 30 wt % commercially available TiO2 powder (P25, Degussa) that consisted of ∼30% rutile and 70% anatase in the crystalline phase (see Figure 4a) was hydrothermally treated with 10 N NaOH in an autoclave at 130 °C for 20 h (Figure 4b), followed by repeated washing with 0.1 N HNO3 to reach a pH value of ∼1.5, as described in the literature.24 The pure Anatase colloidal TiO2 NPs were obtained by autoclaving the low-pH titanate suspension at 240 °C for 12 h (Figure 4c). Anatase TiO2 NPs were investigated by using a field emission scanning electron microscope (SEM, S4800, Hitachi) and JEOL-2010 TEM (JEOL, Japan) equipped with an energy-dispersive spectrometer (EDS) to investigate the TiO2 NPs and determine the compositions of the samples at 200 kV. The TiO2 NPs were also characterized by X-ray diffraction (D/ Max-A, Rigaku) measurements. Figure 5 shows an X-ray scan of the anatase TiO2 NPs before and after sintering, as described below. The inset shows a typical SEM photo of the TiO2 NPs used in the experiment. Figure 6 provides the high-resolution transmission microscopic images and selected area electron diffraction pattern of the TiO2 NPs. Figure 6a shows the lowmagnification TEM image of TiO2 NPs. Figure 6b provides the corresponding electron diffraction pattern taken from the TiO2 NPs, showing that the NPs are crystalline anatase TiO2. Figure 6c gives the high-resolution TEM (HRTEM) image of NPs,

Figure 6. (a) TEM, (b) SEAD pattern, (c) HRTEM bright field images, and (d) EDX analysis of TiO2 nanocrystals prepared by a two-step autoclaving technique.

indicating that the TiO2 NPs are single-crystalline. Lattices of 0.325 and 0.325 nm corresponding to (110) and (1-10) planes of the tetragonal TiO2 along the [001] zone axis, respectively, can be resolved in the HRTEM image. The HRTEM image indicates that the TiO2 NPs are single-crystalline. The inset is the corresponding inverse fast Fourier transform (IFFT) image selected area in panel (c). In Figure 6d, the EDS taken from TiO2 NPs shows that the NPs are composed of Ti and O. The Cu peak comes from the grid. From these measurements, we concluded that our starting materials were pure anatase. A paste of anatase TiO2 powder was made by stirring the mixture of 0.5 g of anatase TiO2 NPs, 100 µL of Triton X-100, and 0.2 g of polyethylene glycol (PEG, Fluka, Mw ) 20 000) into 3 mL of acetic acid (0.1 M). The TiO2 paste was spread on a SnO2/F-coated glass substrate (Pilkington, TEC 8 glass, 8Ω/0, 2.3 mm thick) by the doctor-blade technique to give a flat and smooth surface using an adhesive tape spacer (Figure

14586


4d). The film thickness was governed by the height of the adhesive tape. The exact thickness of the TiO2 film was determined by a surface profiler (TENCOR. P-10). Finally, the TiO2-coated electrode was gradually calcined to remove the polymer under an air flow at 150 °C for 15 min, at 320 °C for 10 min, and at 500 °C for 30 min, leaving a pure anatase TiO2 NP film. The post-treatment with TiCl4 aqueous solution has been applied to freshly sintered TiO2 NP electrodes. An aqueous stock solution of 2 M TiCl4 was diluted to 0.05 M in deionized water. Sintered electrodes were immersed into this solution and stored in an oven at 60 °C for 1 h in a closed vessel. After flushing with deionized water and drying, the electrodes were sintered again at 500 °C for 30 min. 3.2. Cell Fabrication. Figure 4 illustrates the sequence of steps that were used in the fabrication of our solar cells. For photosensitization studies, the calcined TiO2 NP electrode was immersed in the ethanol solution containing purified 3 × 10-4 M cis-di(thiocynato)-N,N′-bis(2,2′-bipyridyl-4-caboxylic acid4′-tetrabutylammonium carboxylate)ruthenium(II) (N719, Solaronix) for 18 h at room temperature (Figure 4e).25 The dyeadsorbed TiO2 electrodes were rinsed with ethanol and dried under a nitrogen flow. The liquid electrolyte was prepared by dissolving 0.6 M 1- butyl-3-methylimidazolium iodide (BMII), 0.03 M iodine, 0.1 M guanidinium thiocyanate, and 0.5 M 4-tertbutylpyridine in acetonitrile and valeronitrile (85:15 v/v). The cathode electrode was produced by coating F/SnO2 glass with a thin layer of a 5 mM solution of H2PtCl6 in isopropanol and was heated at 400 °C for 20 min. The two electrodes were sealed together with thermal melt polymer film (24 µm thick, DuPont) (Figure 4f). The typical active area of the cell was about 0.3 cm2. The exact area of each photoanode was calibrated by an optical scanner under a resolution of 600 dots per inch (dpi) (Figure 4g). 3.3. Cell Measurements. The conversion efficiency and J-V curve was measured by a calibrated solar simulator at Northwestern with AM 1.5 G, 100 mW/cm2. The electrochemical impedance results were obtained under the same light illumination with an impedance analyzer (Solartron 1260), and a potentiostat (Solartron 1287) when a device was applied at its Voc. An additional alternative sinusoidal voltage amplitude of 10 mVrms was also applied between an anode and cathode of a device over the frequency range of 0.05-100k Hz at 10 points per decade. As an example of how we optimize our cell efficiency, we studied the cell efficiency as a function of the TiO2 film thicknesses. In Figure 7a, we give plots of Voc, Jsc, FF, and η as functions of film thicknesses. From these measurements, we can conclude that with our cell architecture and fabrication procedures, the optimal TiO2 film thickness is around 11.5 µm. Using this thickness, we then varied other cell parameters to optimize the overall cell performance. In Figure 7b, we plot cell efficiency per TiO2 NP as a function of the NP film thickness. The efficiency per NP increases almost linear with decreasing film thickness until about 2 µm. In Figure 8, we show plots of data from AC impedance measurements of the cell with different TiO2 NP film thicknesses (from Figure 7) with best-fit model curves. In all of these cases, the total cell thicknesses (L) were the same. Thus, for very thin films, the electrolyte spacing is much larger. There are several features worth noting from our modeling. (1) With very thin TiO2 layers, the contribution from Z3 dominates that of Z2 and Z1 in the low-frequency region. (2) For thicker TiO2 layers, the contribution from Z2 is more pronounced, and we can see the presence of three distinctive circles. (3) The value for the

Lee et al.

Figure 7. (a) Relationship of DSSC device parameters as a function of film thickness. (b) Efficiency per TiO2 NP as a function of film thickness.

Figure 8. AC impedance measurement plots of cells with different TiO2 NP film thicknesses. The solid curves are from a best fit of the model used for our experimental data.

diffusion coefficient in the electrolyte, D1, is the highest for the thinnest TiO2 layer. This also implies longer mean free paths and lower recombination events and thus higher efficiency. 3.4. Interfacial Modification. In our cell preparation, it is not clear whether or not we have optimized the channel widths among the TiO2 NPs to allow the dye molecules and the iodide to freely penetrate all of the way into the depth of the TiO2 film. It is clear from Figure 7 that the maximum efficiency of



Figure 9. SEM images of pre and post fluorine-treated films.

our cell peaks at a thickness of about 11.5 µm. This implies indirectly that (1) both the dye and the iodide may have difficulties penetrating all of the way into the TiO2 film with increasing film thickness or (2) due to TiO2 surface traps, the charges are being recombined and lost on their way to the anode electrodes when the films are too thick. To see if we can improve the former situation, we tried to widen the channels in our nanoporous films by plasma etching. Along with the etching process, we also wanted to learn what other effects this might have on the cell performance since we were also modifying the interfacial property between the TiO2 surface and the dye/electrolyte layers. The etching process was performed under the following conditions. The nanoporous films were placed in a custom-built cylindrical microwave plasma reactor. Microwaves at 2.45 GHz and 600 W were launched into an 11.5 cm diameter cylindrical waveguide from a rectangular waveguide via a mode converter designed to maximize the TE01 and exclude the TE11 transmission modes. The waveguide was coupled to a horizontal 11.5 cm diameter cylindrical vacuum chamber via a quartz window. The reaction gas was introduced at a position close to the quartz window and traveled down the axial direction to the turbomolecular mechanical pump stack. The samples to be treated were placed between the gas inlet and pump stack, near the radial center of the reactor, at 45° from the vertical such that the surface to be treated faced the quartz window. The reaction gas used was 45 sccm of CF4 and 10 sccm of O2. This gas mixture is known to generate active fluorine atoms in the plasma for etching

Figure 11. AC impedance measurements of cells with different surface treatments as discussed in the text. The solid curves are from a model calculation whose parameters are listed in Table 1.

purposes.26-28 The treatment was carried out at 2 mTorr for 10 min intervals. Figure 9 shows SEM images of pre and post fluorine-treated films. It is clear from these images that fluorine effectively opens the channels among the nanoparticles of the TiO2 film. We studied a series of four cells, each with a layer of 8.6 µm thick TiO2 NP film. Each of the four samples were treated differently prior to dye and electrolyte infiltration. Cell (a) was treated with high-temperature sintering only, cell (b) with TiCl4 coating, followed by high-temperature sintering, cell (c) with same treatment as cell (b), but followed by fluorine etching, and cell (d) with the TiO2 NP film etched by fluorine first and followed by TiCl4 treatment and then high-temperature sintering. Figures 10a and 11 give, respectively, the J-V and the impedance measurements of the four cells. Using the kinetic model discussed in section 2 and extrapolating the parameters for the best fit to each of the measured impedance curves in Figure 11, we summarize the results in Table 1 for the electrical data for all four cells. Cell (a) can be considered as the control

Figure 10. (a) J-V plots of cells with different surface treatments as discussed in the text and (b) simple illustrations of different surface treatments (see text for discussion).

14588


Lee et al.

TABLE 1: Parameters for the Best Fit of the Impedance Data for Each of the Four Cells Measured in Figure 11 Deff keff ns D1 JSC Rtotal (10-5 cm2 s-1) (Hz) Rk (Ω) Rw (Ω) Rd (Ω) (1018 cm-3) (10-7 cm2 s-1) VOC (V) (mA/cm2) FF (%) EFF (%) (Ω) (a) TiO2 (b) TiO2/TiCl4 (c) TiO2/TiCl4/CF4 (d) TiO2/CF4/TiCl4

4.79 4.40 1.18 2.20

12.6 12.6 5.01 10.0

12.1 10.6 12.8 12.2

2.3 2.3 4 4

8.6 11.9 11.2 12.1

for the other three cells. There is no treatment to the TiO2 NP film in this case. Cell (b) with TiCl4 treatment has the lowest series resistance (i.e., Rtotal ) R0 + R1 + R2 + R3) and therefore the highest efficiency. The contribution to this increase of cell efficiency comes primarily from the 26% increase in Jsc and the about 4% increase from Voc. We also notice from our model that there is no change in the values of Rw and keff, that is, no change in the transport properties in the TiO2 film. As convincingly documented in recent literature, such a change can be attributed to the improved band alignment between the TiO2 and dye molecule layer to facilitate electron injection from the later to the former.29,30 The difference between cells (c) and (d) is the order in which the TiO2 NP film is being treated. In cell (c), we first treat the TiO2 surface by TiCl4 and then by fluorine etching, while for cell (d), we reversed the sequence of treatment. From our calibrated fluorine etching procedure to enlarge the TiO2 channel widths, we determined that about 10% of the TiO2 film material was removed (by pre- and post-sample weighing). Figure 10b gives an illustration of cell (c) treatment with TiCl4 coating first followed by a plasma fluorine etching process. From impedance analysis, we observe that cell (c) behaves quite differently from the rest in the group. Here, we notice that the value of Deff is reduced by about 70% compared to that of cell (b), for example. In addition, the value of keff is reduced by about 60%. On the other hand, the value of ns has about doubled. We now use our model to explain this interesting behavior of cell (c). In Figure 10b(c), we illustrate the possible structure of the TiO2 thin film as a result of fluorine etching following the SEM picture in Figure 9. The TiO2 NPs are now less connected compared to the case of cell (a), Figure 10b(a). While this situation should improve the dye molecule attachments to the NPs and allow further penetration of the molecules into the TiO2 film, it does impede the charge transport, as indicated by a large decrease in the values of Deff and keff in Table 1. We conclude that the value of ns is increased due to poor charge transport in the TiO2 film. This sequence of treatment, however, increases both the Voc and Jsc, giving an efficiency higher than that of cell (a). For the case of cell (d), we opened channels in the TiO2 film first and then coat the layer with TiCl4 to see if one can improve the cell performance compared to that of cell (a). In this case, as we can see from Table 1, cell (d) performs much better than cell (a), with increased values both for Voc ≈ 0.888 (v) and Jsc ≈ 13.56 mA/cm2. In fact, cell (d) may perform better than cell (b) if we add 10% TiO2 NPs to this cell by increasing the TiO2 film thickness. We conclude from this work that it is not easy to precisely adjust the spacing of the interparticle channels and maintain the ideal interface between the TiO2 NP film and the dye/electrolyte layers. It requires very careful processing steps, and we are currently performing this research.

3.77 4.74 9.75 5.74

7.5 5.6 5.7 7.7

0.794 0.825 0.864 0.888

12.63 15.96 13.22 13.56

71.4 69.6 72.3 71.1

7.16 9.17 8.25 8.56

65.7 63.5 79.3 67.7

the absorbing film is one way to improve the conversion efficiency of the solar cell. There are two different approaches to photon trapping,31 (1) the conventional method of using geometrical optics, such as a mirror, or (2) the use of wave optics, representing a new approach. Of course, one may also choose to use a combination of the two approaches. In this work, we demonstrate the effective use of photonic crystals for photon management in our cells. Photonic crystals are dielectric materials that are structured in one, two, or three dimensions on length scales corresponding to wavelengths of the electromagnetic spectrum for which one would like to control reflection or transmission; the result is a sometimes multidimensional diffraction grating that can reflect, transmit, and diffract light for various wavelengths. By properly designing the crystal structure, one can select the band gap(s) of the crystal for the purpose of reflecting and diffracting light as described below. In earlier work for DSSCs, a diffuse scattering layer of large TiO2 colloids32 was typically incorporated into the cell to increase photon scattering and efficiency.33 On the other hand, some of the most successful approaches developed to improve silicon photovoltaic cells were based on the realization of coherent scattering processes such as highly reflecting distributed Bragg reflectors,34 surface gratings,35 or a combination of both,36 as well as 3D photonic crystal (PhC).37-40 However, these concepts cannot be easily realized in DSSCs. Here, we take the most efficient cells and measure how much of the incoming light is actually being absorbed by the N719 dye as photons move across the solar cell. In Figure 12, we plot the quantum efficiency of the N719 cell35 versus the wavelength. We see that for this dye, the absorption spectrum is quite wide, extending from 400 to 700 nm. In the same figure, we plot the percentage of light that was transmitted (as measured by an UV/vis/NIR spectrometer) through one of our typical cells with a thickness of around 11.5 µm. In this plot, we observe that there still remains a large portion of the spectrum that was not being absorbed by our N719 dye molecules. Light with longer wavelengths beyond 750 nm have no effect on the N719 dye molecules. To capture the lost photons from our cells, we have applied a simple recycling process as described in Figure 13. Here, we simply attach a reflector to the back of the cell (the cathode electrode) by either using a Ag mirror (Figure 13a) or a stack of the photonic crystals (Figure 13b).

4. Photon Management One way to enhance the cell efficiency without altering the architecture of the conventional DSSC design is to ensure that all of the photons that enter the cell are efficiently transformed into useful charges. Reflecting unabsorbed photons back into

Figure 12. Quantum efficiency measurement of the N719 dye and the transmittance measurement of DSSCs using the same dye.



Figure 13. (a) Schematic of a DSSC with a typical geometric optic concept of reflection to trap light. (b) Schematic of a DSSC with wave optics (3D PhC) to trap light with reflection and diffraction.

Figure 14. (a) SEM micrograph of a typical ZnO PhC (or inverse opal) layer, showing both the top view (left) and the side (cross sectional) view. (b) The broad reflection spectrum on our Ag mirror along with the reflection spectra of two inverse ZnO PhCs with hole sizes of 375 and 410 nm.

Figure 13 illustrates the two approaches to light trapping; Figure 13a the use of geometric optics (a Ag mirror reflector) and Figure 13b the wave optics (a 3-D photonic crystal). First, the Ag film can offer a high reflection property, as shown in Figure 14b. Similarly, a photonic crystal can reflect light incident from any angle for frequencies and polarizations within the photonic band gap. Second, wave-optics-based devices such as surface grating and 3D PhC can be designed to diffract incoming beams into highly oblique angles according to Bragg’s law.41 The diffraction induced by an interface improves light trapping by increasing the distance that light must travel to return to the front surface of the cell. Furthermore, if the angle of the diffracted beam is greater than the critical angle, it will also be internally reflected back into the solar cell.42 The PhC can efficiently reflect and diffract transmitted radiation of the dye-sensitized cell in a specific wavelength range determined by the spectral width of its photonic band gap. The peak positions can be related to the sphere diameter and the effective refractive index of the medium using Bragg’s law, λmax)2neffd111, where d111 is the 111 lattice spacing. In our experiments, we have designed our photonic crystals in such a way that they reflect efficiently in the wavelengths region of interest (see Figure 12). Figure 14a is a SEM micrograph of a

Figure 15. (a) J-V characteristics of a 11 µm thick TiO2 film DSSC with different reflection configurations, a Ag film and 3 different PhCs. (b) Nyquist plots for each of the J-V curves in (a); the solid curves are from model calculations. See Table 2 for parameters used in the model calculations.

ZnO PhC (or inverse opal) layer, showing both the top view (left) and the side (cross sectional) view. Figure 14b also shows the broad reflection spectrum of a Ag mirror along with the reflection of two inverse ZnO PhCs with hole sizes of 375 and 410 nm, respectively. We note that these two PhCs when combined can reflect and diffract waves in the spectral range as prescribed in Figure 12. This implies that we can recycle the photons back into the DSSC for further absorption and processing. 4.1. Photonic Crystal Preparation and Measurements. The opal templates were prepared using a vertical deposition technique detailed elsewhere.43,44 Glass substrates were placed in vials with deionized water containing between 0.1 and 0.3 vol % polystyrene (PS) spheres in suspension, depending on sphere size and intended opal film thickness. The water was then slowly evaporated at 50 °C in a drying oven. This method yields polystyrene (PS) opal films with controllable thicknesses between 20 and 100 layers and large single-crystalline domains, oriented with the (111) planes of the face-centered cubic (fcc) structure parallel to the substrate surface. ZnO was infiltrated into the opal templates by atomic layer deposition (ALD)45 using diethyl zinc (DEtZn) and water (H2O) as precursors. To avoid deformation or melting of the PS structures, the growth temperature was kept at 85 °C, below the glass transition temperature of PS. The chamber pressure during growth was kept at 10 Torr, and a relatively slow flow of N2 carrier gas

14590


Lee et al.

TABLE 2: Parameters for the Best Fit of the Impedance Data for Each of the Configurations in Figure 15 Deff keff Con ns D1 JSC (10-5 cm2 s-1) (Hz) Rk(Ω) Rw(Ω) (Ω cms-1) Rd (Ω) (1019 cm-3) (10-6 cm2 s-1) VOC (V) (mA/cm2) FF (%) EFF (%) 3D photonic non crystal with metal with PhC (375 nm) with PhC (410 nm) with PhC (375/410 nm)

3.50

5.01

8.5

1.9

0.053

7.6

1.02

0.8

0.784

20.4

68.4

10.9

2.56 2.67

5.01 5.01

8.5 8.2

2.6 2.4

0.053 0.051

6 6.2

1.02 1.06

1.0 0.9

0.785 0.790

21.6 21.9

67.9 67.5

11.5 11.7

3.37

6.31

8.2

2.4

0.065

5.5

0.84

1.0

0.786

22.4

67.9

12.0

3.37

6.31

8.2

2.4

0.065

5.1

0.84

1.1

0.786

23.6

67.4

12.5

was maintained in the chamber. One ALD reaction cycle consisted of a 1 s exposure to DEtZn, followed by a 20 s N2 purge and a 1 s exposure to H2O vapor, followed another 20 s N2 purge. The PS spheres were then removed by firing the structure in air at 500 °C for 1 h, leaving an ordered fcc array of air holes in the ZnO layer. The optical property of our PhC was measured by a UV/vis/ NIR spectrometer (Perkin-Elmer LAMBDA 1050) with a wavelength range of 185-3300 nm. The sample mount had two settings for measuring diffuse (scattered) reflectivity and total (diffuse and specular) reflectivity. Specularly reflected light was tightly reflected around the angle of incidence (mirror-like reflection). Diffusely reflected light resulted from roughness, defects, and so forth and was collected by an integrating sphere. The inverse opals were mounted to the diffuse reflectance accessory sphere such that the [111] direction was pointing almost directly inward. Specular reflectivity spectra were taken by subtracting the diffuse reflectivity from the total reflectivity. 4.2. Measurements and Modeling of DSSCs with PhCs. Cells with structures illustrated in Figure 13 were fabricated, and their J-V and AC impedance measurements were performed. Figure 15a shows a series of J-V curves demonstrating an increase in Jsc values for the case of Ag mirror reflection and PhC reflections. The corresponding impedance measurements are plotted in Figure 15b, and the solid curves represent the best fits from our modeling calculations. The fitting parameters are tabulated in Table 2. It should be noted from these measurements that (1) there is no change in the Voc, fill factor, and the transport properties of the cells; (2) however, there is a systematic increase of Jsc (along with the reduction of the internal cell resistance) with the optimization of reflectors used from a Ag mirror to PhC reflectors, leading to an increase in cell efficiency by about 13%; and (3) there is an increase in the value of the diffusion coefficient D1 of the electrolyte along with the increase in cell efficiency. We demonstrated here that recycling of photons by PhC is an effective way to increase the cell efficiency. 5. Summary and Conclusion In this work, we have systematically optimized the conventional dye-sensitized solar cell efficiency by improving the quality of the TiO2 NPs, the composition of the electrolyte, the purity of the N719 dye molecules, the interfacial properties between various layers in the cell, the choice ohmic contacts, and finally the use of optimal PhCs for light confinement. With the above optimization procedure, we have achieved cell efficiency to about 12.5%. From this study, we have also identified some of the important limiting factors in the current cell design. To maximize photon absorption and surface chemical reaction, one wants the smallest size of the NP in a given volume. In order to absorb all of the incident photons by the dye molecules in the NP layer, one would want to have an

as thick as possible NP layer. However, the smaller the NP and thicker the NP layer, the harder it is for the dye molecules and the electrolyte to infiltrate into the layer to make an efficient cell. In addition to geometrical constraints, there are demands on interfacial chemical and electronic compatibility requirements for constructing efficient cells. This has been a big challenge in the field of DSSC design. We have shown that effective surface passivation will increase Jsc and Voc and enhance cell performance. More detailed studies are required to understand the related mechanisms Finally, because of the nonoptimal design of the conventional cell, we have shown that a large percentage of the photon was not captured by our dye molecules. Specially designed PhCs were used to confine a portion of the photons which otherwise would have escaped from the cells. Of course, part of this loss can be remedied if one uses more spectral appropriate dye molecules, which have higher absorption coefficients in the longer wavelengths regime. On the basis of what we have learned, we are now developing a new architectural design for the DSSC. Acknowledgment. The authors acknowledge support for this collaborative research from DOE-DE-FG02-08ER46536 for B.L. and D.B.B., the DOE-Energy Frontier Research Center, ANSER, DE-SC0001059 for P.G.;, the NSF ECCS-0823345 for D.-K.H., the NSF DMR 0706439 for S.-T.H., the NSF DMR 0843962 for R.P.H.C., the NSF-MRSEC DMR-0520513 for use of facilities, and the National Science Council of Taiwan for S.T.H. and C.-Y.W. References and Notes (1) Gra¨tzel, M. Dye-sensitized solar cells. J. Photochem. Photobiol., C 2003, 4 (2), 145–153. (2) Thavasi, V.; Renugopalakrishnan, V.; Jose, R.; Ramakrishna, S. Controlled electron injection and transport at materials interfaces in dye sensitized solar cells. Mater. Sci. Eng. R-Rep. 2009, 63 (3), 81–99. (3) Gratzel, M. Photoelectrochemical cells. Nature 2001, 414 (6861), 338–344. (4) Frank, A. J.; Kopidakis, N.; van de Lagemaat, J. Electrons in nanostructured TiO2 solar cells: transport, recombination and photovoltaic properties. Coord. Chem. ReV. 2004, 248 (13-14), 1165–1179. (5) Goncalves, L. M.; Bermudez, V. D.; Ribeiro, H. A.; Mendes, A. M. Dye-sensitized solar cells: A safe bet for the future. Energy EnViron. Sci. 2008, 1 (6), 655–667. (6) Chen, C. Y.; Wang, M. K.; Li, J. Y.; Pootrakulchote, N.; Alibabaei, L.; Ngoc-le, C. H.; Decoppet, J. D.; Tsai, J. H.; Gratzel, C.; Wu, C. G.; Zakeeruddin, S. M.; Gratzel, M. Highly Efficient Light-Harvesting Ruthenium Sensitizer for Thin-Film Dye-Sensitized Solar Cells. ACS Nano 2009, 3 (10), 3103–3109. (7) Toivola, M.; Halme, J.; Miettunen, K.; Aitola, K.; Lund, P. D. Nanostructured dye solar cells on flexible substrates - Review. Int. J. Energy Res. 2009, 33 (13), 1145–1160. (8) Sastrawan, R.; Beier, J.; Belledin, U.; Hemming, S.; Hinsch, A.; Kern, R.; Vetter, C.; Petrat, F. M.; Prodi-Schwab, A.; Lechner, P.; Hoffmann, W. New interdigital design for large area dye solar modules using a leadfree glass frit sealing. Prog. PhotoVoltaics 2006, 14 (8), 697–709. (9) Lee, B. H.; Song, M. Y.; Jang, S.-Y.; Jo, S. M.; Kwak, S.-Y.; Kim, D. Y. Charge Transport Characteristics of High Efficiency Dye-Sensitized

Materials, Interfaces, and Photon Confinement in DSSCs Solar Cells Based on Electrospun TiO2 Nanorod Photoelectrodes. J. Phys. Chem. C 2009, 113 (51), 21453–21457. (10) Fabregat-Santiago, F.; Garcia-Belmonte, G.; Bisquert, J.; Zaban, A.; Salvador, P. Decoupling of Transport, Charge Storage, and Interfacial Charge Transfer in the Nanocrystalline TiO2/Electrolyte System by Impedance Methods. J. Phys. Chem. B 2001, 106 (2), 334–339. (11) Wang, Q.; Moser, J.-E.; Gratzel, M. Electrochemical Impedance Spectroscopic Analysis of Dye-Sensitized Solar Cells. J. Phys. Chem. B 2005, 109 (31), 14945–14953. (12) Fabregat-Santiago, F.; Bisquert, J.; Palomares, E.; Otero, L.; Kuang, D.; Zakeeruddin, S. M.; Gratzel, M. Correlation between Photovoltaic Performance and Impedance Spectroscopy of Dye-Sensitized Solar Cells Based on Ionic Liquids. J. Phys. Chem. C 2007, 111 (17), 6550–6560. (13) Duncan, W. R.; Prezhdo, O. V. Theoretical studies of photoinduced electron transfer in dye-sensitized TiO2. Annu. ReV. Phys. Chem. 2007, 58, 143–184. (14) Luther, J. F. a. J. Modeling of Photovoltage and Photocurrent in Dye-Sensitized Titanium Dioxide Solar Cell. J. Phys. Chem. B 2001, 105 (21), 4895–4903. (15) Cahen, D.; Hodes, G.; Gratzel, M.; Guillemoles, J. F.; Riess, I. Nature of photovoltaic action in dye-sensitized solar cells. J. Phys. Chem. B 2000, 104, 2053–2059. (16) Adachi, M.; Sakamoto, M.; Jiu, J.; Ogata, Y.; Isoda, S. Determination of Parameters of Electron Transport in Dye-Sensitized Solar Cells Using Electrochemical Impedance Spectroscopy. J. Phys. Chem. B 2006, 110 (28), 13872–13880. (17) Kern, R.; Sastrawan, R.; Ferber, J.; Stangl, R.; Luther, J. Modeling and interpretation of electrical impedance spectra of dye solar cells operated under open-circuit conditions. Electrochim. Acta 2002, 47 (26), 4213–4225. (18) Bisquert, J.; Garcia-Belmonte, G.; Fabregat-Santiago, F.; Bueno, P. R. Theoretical models for ac impedance of finite diffusion layers exhibiting low frequency dispersion. J. Electroanal. Chem. 1999, 475 (2), 152–163. (19) Bisquert, J. Theory of the Impedance of Electron Diffusion and Recombination in a Thin Layer. J. Phys. Chem. B 2002, 106 (2), 325–333. (20) Han, L. Y.; Koide, N.; Chiba, Y.; Islam, A.; Komiya, R.; Fuke, N.; Fukui, A.; Yamanaka, R. Improvement of efficiency of dye-sensitized solar cells by reduction of internal resistance. Appl. Phys. Lett. 2005, 86 (21), 213501. (21) Hauch, A.; Georg, A. Diffusion in the electrolyte and charge-transfer reaction at the platinum electrode in dye-sensitized solar cells. Electrochim. Acta 2001, 46 (22), 3457–3466. (22) 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. Nanocrystalline dye-sensitized solar cells having maximum performance. Prog. PhotoVoltaics 2007, 15 (1), 1–18. (23) Ito, S.; Nazeeruddin, M. K.; Liska, P.; Comte, P.; Charvet, R.; Pe´chy, P.; Jirousek, M.; Kay, A.; Zakeeruddin, S. M.; Gra¨tzel, M. Photovoltaic characterization of dye-sensitized solar cells: effect of device masking on conversion efficiency. Prog. PhotoVoltaics 2006, 14 (7), 589– 601. (24) Tsai, C.-C.; Teng, H. Structural Features of Nanotubes Synthesized from NaOH Treatment on TiO2 with Different Post-Treatments. Chem. Mater. 2005, 18 (2), 367–373. (25) Park, N.-G.; Kim, K. Transparent solar cells based on dye-sensitized nanocrystalline semiconductors. Phys. Status Solidi A 2008, 205 (8), 1895– 1904. (26) Sapieha, S.; Wrobel, A. M.; Wertheimer, M. R. Plasma-assisted etching of paper. Plasma Chem. Plasma Process. 1988, 8 (3), 331–346. (27) Rauf, S. Model for photoresist trim etch in inductively coupled CF4/O2 plasma. J. Vac. Sci. Technol., B 2004, 22 (1), 202–211.

J. Phys. Chem. B, Vol. 114, No. 45, 2010 14591 (28) Fracassi, F.; Dagostino, R. Chemistry of Titanium Dry Etching in Fluorinatated and Chlorinated Gases. Pure Appl. Chem. 1992, 64 (5), 703– 707. (29) Palomares, E.; Clifford, J. N.; Haque, S. A.; Lutz, T.; Durrant, J. R. Control of charge recombination dynamics in dye sensitized solar cells by the use of conformally deposited metal oxide blocking layers. J. Am. Chem. Soc. 2003, 125 (2), 475–482. (30) Sommeling, P. M.; O’Regan, B. C.; Haswell, R. R.; Smit, H. J. P.; Bakker, N. J.; Smits, J. J. T.; Kroon, J. M.; van Roosmalen, J. A. M. Influence of a TiCl4 post-treatment on nanocrystalline TiO2 films in dyesensitized solar cells. J. Phys. Chem. B 2006, 110 (39), 19191–19197. (31) Bermel, P.; Luo, C.; Zeng, L.; Kimerling, L. C.; Joannopoulos, J. D. Improving thin-film crystalline silicon solar cell efficiencies with photonic crystals. Opt. Express 2007, 15 (25), 16986–17000. (32) Kay, A.; Gratzel, M. Low cost photovoltaic modules based on dye sensitized nanocrystalline titanium dioxide and carbon powder. Sol. Energy Mater. Sol. Cells 1996, 44 (1), 99–117. (33) Ito, S.; Zakeeruddin, S. M.; Humphry-Baker, R.; Liska, P.; Charvet, R.; Comte, P.; Nazeeruddin, M. K.; Pechy, P.; Takata, M.; Miura, H.; Uchida, S.; Gratzel, M. High-efficiency organic-dye-sensitized solar cells controlled by nanocrystalline-TiO2 electrode thickness. AdV. Mater. 2006, 18 (9), 1202–1207. (34) Johnson, D. C.; Ballard, I.; Barnham, K. W. J.; Bishnell, D. B.; Connolly, J. P.; Lynch, M. C.; Tibbits, T. N. D.; Ekins-Daukes, N. J.; Mazzer, M.; Airey, R.; Hill, G.; Roberts, J. S. In AdVances in Bragg stack quantum well solar cells, International Conference on Physics, Chemistry and Engineering of Solar Cells, Badajoz, Spain, May 13-15; Elsevier Science BV: Badajoz, Spain, 2004; pp 169-179. (35) Nishimura, S.; Abrams, N.; Lewis, B. A.; Halaoui, L. I.; Mallouk, T. E.; Benkstein, K. D.; van de Lagemaat, J.; Frank, A. J. Standing Wave Enhancement of Red Absorbance and Photocurrent in Dye-Sensitized Titanium Dioxide Photoelectrodes Coupled to Photonic Crystals. J. Am. Chem. Soc. 2003, 125 (20), 6306–6310. (36) Rogers, J. A.; Meier, M.; Dodabalapur, A. Using printing and molding techniques to produce distributed feedback and Bragg reflector resonators for plastic lasers. Appl. Phys. Lett. 1998, 73 (13), 1766–1768. (37) Takahashi, S.; Suzuki, K.; Okano, M.; Imada, M.; Nakamori, T.; Ota, Y.; Ishizaki, K.; Noda, S. Direct creation of three-dimensional photonic crystals by a top-down approach. Nat. Mater. 2009, 8 (9), 721–725. (38) Noda, S.; Tomoda, K.; Yamamoto, N.; Chutinan, A. Full ThreeDimensional Photonic Bandgap Crystals at Near-Infrared Wavelengths. Science 2000, 289 (5479), 604–606. (39) Bielawny, A.; Rockstuhl, C.; Lederer, F.; Wehrspohn, R. B. Intermediate reflectors for enhanced topcell performance in photovoltaicthinfilm tandem cells. Opt. Express 2009, 17 (10), 8439–8446. (40) Wijnhoven, J. E. G. J.; Vos, W. L. Preparation of Photonic Crystals Made of Air Spheres in Titania. Science 1998, 281 (5378), 802–804. (41) Vos, W. L.; Sprik, R.; van Blaaderen, A.; Imhof, A.; Lagendijk, A.; Wegdam, G. H. Strong effects of photonic band structures on the diffraction of colloidal crystals. Phys. ReV. B 1996, 53, 16231. ¨ pping, J.; Miclea, P. T.; Wehrspohn, R. B.; (42) Bielawny, A.; U Rockstuhl, C.; Lederer, F.; Peters, M.; Steidl, L.; Zentel, R.; Lee, S.-M.; Knez, M.; Lambertz, A.; Carius, R. 3D photonic crystal intermediate reflector for micromorph thin-film tandem solar cell. Phys. Status Solidi A 2008, 205 (12), 2796–2810. (43) Jiang, P.; Bertone, J. F.; Hwang, K. S.; Colvin, V. L. Single-Crystal Colloidal Multilayers of Controlled Thickness. Chem. Mater. 1999, 11 (8), 2132–2140. (44) Hwang, D.-K.; Noh, H.; Cao, H.; Chang, R. P. H. Photonic bandgap engineering with inverse opal multistacks of different refractive index contrasts. Appl. Phys. Lett. 2009, 95 (9), 09110/1–09110/3. (45) Ott, A. W.; Chang, R. P. H. Atomic layer-controlled growth of transparent conducting ZnO on plastic substrates. Mater. Chem. Phys. 1999, 58 (2), 132–138.

JP102359R

Materials, Interfaces, and Photon Confinement in Dye-Sensitized Solar

Recommend Documents