Nanorod-Based Dye-Sensitized Solar Cells with Improved Charge

Nanorod-Based Dye-Sensitized Solar Cells with Improved Charge

(Parte 1 de 2)

DOI: 10.1002/adma.200701819

Nanorod-Based Dye-Sensitized Solar Cells with Improved Charge Collection Efficiency**

By Soon Hyung Kang, Sang-Hyun Choi, Moon-Sung Kang, Jae-Yup Kim, Hyun-Sik Kim, Taeghwan Hyeon, and Yung-Eun Sung*

Dye-sensitized solar cells (DSSCs) have been widely investigated on account of their special features, such as low cost and high energy conversion efficiency.[1] To attain higher efficiency (> 1 %), more detailed studies will be needed for the each part of the DSSC (e.g., photoanode (an electron transporting layer), dye (a photosensitizer), electrolyte (an electron transfer mediator), and counter electrode (a catalyst of electron transfer)). In particular, in the part of the photoanode, the insufficient electron diffusion coefficient in tradi- tional electrodes composed of nanometer-sized TiO2 particles (i.e., several orders of magnitude smaller than those in bulk single crystal TiO2) is believed to limit the power conversion efficiency.[2] This is because electron collection is determined by trapping/detrapping events along the site of the electron traps (defects, surface states, grain boundaries, self trapping etc.).[3–4] This point has promoted research toward photoanodes comprising nanoporous materials of various shapes (e.g., rods, tubes, wires, and tetra pads) with the enhanced electron transport properties due to the features of highly decreased intercrystalline contacts and stretched grown structure with the specified directionality.[5–6]

In particular, nanorods (NRs) are believed to have exceptional electron transport properties and have been considered as alternatives to nanoparticle (NPs). Recently, the Adachi group reported the application for DSSCs using hydrothermal synthesized single crystalline TiO2 NRs with a diameter of 20∼30 nm and the length of > 100 nm.[7] A power conversion a fill factor of 0.7) was obtained for the DSSC comprised of

TiO2 anatase NRs 16 lm in thickness. Highly efficient performance was achieved owing to the increased rate of electron transport resulted from the high crystalline anatase nanorod. However, the extent of dye adsorption related to the lightharvesting yield is poor due to the relatively low specific surface area.

Herein, in order to develop an efficient device, NRs were synthesized from the necking of truncated NPs applying the synthesis of an “oriented attachment approach”.[8] It was reported that the driving force for the oriented attachment between truncated spherical units is the joining of each unit by eliminating impure molecules from the spherical titanium dioxide surfaces.[9] The dye adsorption ability of the synthesized NRs might be complemented from the truncated NPs units with a large specific surface area, which can influence the high dye loading ability. Furthermore, from the point that prevailed features of TiO2 NRs in the DSSC such as the decreased intercrystalline contacts between grain boundaries and stretched grown structure to the specific directionality may improve the charge collection efficiency by the favorable electron transport rate, a comparison of electron transport and charge recombination characteristics between NP-and NR-based photoanodes in DSSCs was examined by the stepped light-induced transient measurements of photocurrent and voltage (SLIM-PCV) under front side illumination. Then, truncated NPs with a particle size of about 6 nm, as a control sample, were also prepared using the same synthesis tool.

Scheme 1 describes the TiO2 photoanodes consisting of the NPs and NRs in the configuration of DSSCs. NPs and NRs

[*] Prof. Y.-E. Sung, S. H. Kang, J.-Y. Kim, H.-S. Kim

School of Chemical and Biological Engineering and Interdisciplinary Program in Nano Science and Technology Seoul National University Seoul, 151-742 (Korea) E-mail:

S.-H. Choi, Prof. T. Hyeon School of Chemical and Biological Engineering Seoul National University Seoul, 151-742 (Korea)

Dr. M.-S. Kang Samsung Advanced Institute of Technology Energy and Environmental Lab Gyeonggi-Do, 446-712 (Korea)

[**] This work was supported in part by KOSEF (Contract No. R01-2004- 0-10143-0), and in part by the Research Center for Energy Conversion and Storage (Contract N0. R11-2002-102-0-0). Supporting Information is available online from Wiley InterScience or from the authors

Dye nanorodnanoparticle

Dye e- Dye nanorod

Dye e-Dye nanorodnanoparticle Dye nanoparticle

Dye e-Dye

Scheme 1. Simple representation of TiO2 NP- and NR-based photoanodes on the FTO coating glass substrate.

films deposited on a rigid substrate have different structural properties. Contrary to the TiO2 film composed of NPs, the

TiO2 film composed of irregular NRs display the possibility of a directed electron pathway along the long axis of the NRs in a randomly intercalated 3D structure.

Figure 1 shows high-resolution transmission electron microscope (HR-TEM) images of the NPs (Fig 1a and c) and NRs (Fig. 1b and d). In the condition where the synthesized NPs and NRs were well-dispersed, the shape and the extent of specific directionality were determined by the acidity of the precursor used.[10] This means that in a neutral solution (1 M NaCl, pH 6.6), the formation of spherical anatase NPs was preferred due to the absence of induced polarity (Fig. 1a), while in a weak acidic solution (1 M CH3COOH, pH 2.4), anisotropic NRs with a high aspect ratio (< 20) toward the

<001> lattice of the anatase NRs were synthesized (Fig. 1b). The HR-TEM image (Fig. 1c) of the 6 nm-sized anatase NPs clearly shows a <101> lattice plane, while the HR-TEM images (Fig. 1d) of the short anatase NRs with an average particle size of 4 20 nm revealed zigzag surface features with a <101> direction originating from the oriented attachment approach reported previously.[1,12]. In other words, truncated NRs from the necking of spherical NPs derived from the <001> and <101> lattice plane of different crystalline titanium dioxide NPs were formed, as shown by the white circle. Highly crystalline anatase peaks after thermal treatment (530 °C, 1h under air ambient) were also confirmed by selected area electron diffraction and high-power X-ray diffraction analyses, data shown in Supporting Information

(Figure S1 and S2). From the Scherrer equation to the anatase <101> peak at 2h = 25.3°,[13] the average crystallite sizes of the annealed NPs and NRs were approximately 1.2 nm and 12.5 nm, respectively. This suggests that thermal treatment causes the conglomeration of NPs and NRs to minimize the surface energy, leading to the growth of nanometer-sized materials.[14] This event causes a decrease in the actual specific surface area, reducing the amount of dye molecules adsorbed on the titanium dioxide surface.

Brunauer-Emmett-Teller (BET) measurement was performed to understand the 3D distribution of NPs and NRs in the thin film structure and determine the specific surface area, porosity, and surface roughness factor. Annealed NPs and NRs under the same annealing condition were used. The specific surface area of the NPs and NRs were approximately 67.3 m2 g–1 and 93.6 m2 g–1, respectively. The porosities (P)o f the NPs and NRs calculated by referring to this formula;[15]

P=V p/(q–1 +V p), where Vp is the specific cumulative pore volume (cm3 g–1) and q–1 is the inverse of the density of ana- large surface area and similar porosity to the NPs film. An estimation of the roughness factor (R) per unit film thickness of both films could be induced using the related formula;[16]

R = q(1-P)S, where q is the density (g cm–3) of anatase TiO2, P is the porosity (%) of the films, and S is the specific surface area (m2 g–1). The calculated roughness factors (lm–1) were approximately 115.6 and 169.9 for the NP and NR films, respectively, showing that the NR films have higher dye adsorption ability. Approximately one and a half time higher roughness factor in the NRs film leads to an increase in the charge harvesting efficiency from a larger amount of dye adsorption.

The surface morphology of NP- and NR-based TiO2 film indicated that the NR-based film was composed of the connected

NPs, forming the network of NRs. Therefore, the NP-like morphology for NR-based film was shown in Figure S3, measured by field emission scanning electron microscopy (FE-SEM).

In general, Jsc of a DSSC is determined by the light harvesting efficiency of a cell, charge injection yield, and charge transport characteristics. In the situation that the light harvesting efficiency induced by a large roughness factor can con-

tribute to improving Jsc, it is necessary to examine the effect of the electron transport and charge recombination character- istics, which are regarded as outstanding properties in a DSSC composed of a stretched NR film, excluding the other contributing factors. As a result, the same dye adsorbed films are prepared to compare the electron transport and charge recombination characteristics of both films. Assuming that the charge injection yield approaches unity (1) under the same light harvesting environment,[17] the factor determining Jsc in this device become the characteristics of electron transport and charge recombination. Figure 2 shows the J-V characteristics (Fig. 2a) and incident photon to current efficiency (IPCE) (Fig. 2b) of the NP-based and NR-based DSSCs with in the same number of dye molecules adsorbed (n)

Figure 1. HR-TEM images of a) TiO2 nanoparticles, and b) TiO2 nanorods with anatase structure synthesized by oriented attachment. c,d)

Magnified views of (a) and (b), respectively.

(n = 2.8 1016). In order to compare the quantitative value of dye molecules adsorbed in both films, the extent of the desorbed dye molecules in a basic 1 M NaOH solution was examined by UV-VIS spectroscopy. Using an extinction coefficient of e = 3748 cm–1 M–1 at 535 nm (N719 dye), the number of adsorbed dye molecules on the both films should be con- trolled by the thickness of the TiO2 film and the concentration of the TiO2 sol. The thicknesses of the NP and NR films with the same number of adsorbed dye molecules measured using an Alpha-Step 200 apparatus (Tencor Instruments) were approximately 5.47 lm and 4.87 lm, respectively. NP-based

DSSC showed a Voc of 0.68 V, a Jsc of 6.9 mA cm–2, a fill factor of 0.71, and an efficiency of 3.36 %, while the NR-based

DSSC showed a Voc of 0.7 V, a Jsc of 1.7 mA cm–2, a fill factor of 0.6, and an efficiency of 4.95 % under 1 sun conditions.

The significantly enhanced Jsc in the NR-based DSSC was confirmed by the result (5.5 % at 520 nm) of the IPCE func- tioned by the wavelength of the x-axis. Furthermore, from the measurement of the J-V curves in the dark state, which show the extent of charge recombination, it was demonstrated that the active side reaction (i.e., charge recombination) in the NP-based DSSC was processed, and the dark current onset of NP-based DSSC was shifted to a lower potential range

(0.4 V) compared with that (0.47 V) of the NR-based DSSC and its intensity was also higher. This result is supported by the measurements of the electron lifetime. In the case of a thicker TiO2 NR film (7.1 lm), a maximum conversion efficiency of 6.2 % was achieved, with an IPCE value of 6.5 % at 520 nm. Moreover, the TiO2 NP film (7.5 lm) shows a conversion efficiency of 4.3 % with an IPCE value of 53.6 %, as shown in Figure S4.

Figure 3 shows Jsc values of the NP- and NR-based DSSCs as a function of the amount of adsorbed dye (n). The photo- current density increased linearly with increasing amount of adsorbed dye molecules in both NP and NR films, except that the slope of Jsc with n was larger for the NR films than that for the NP films. It is evident that an understanding of the charge transport/transfer phenomenon in a nano-dimension- ally intercalated TiO2 film can realize the design of highly efficient DSSCs.

Figure 4 shows the electron diffusion coefficients (D) and lifetimes (s) of the NP-and NR-based DSSCs as a function of

Jsc, as analyzed by SLIM-PCV. The D and s values were determined by the photocurrent and photovoltage transients in- duced by a stepwise change in the laser light intensity controlled with a function generator.[18–20] The D value was obtained by a time constant (sc) determined by fitting a decay of the photocurrent transient with exp(-t/sc) and the TiO2 film thickness (X) using the equation, D = X2/(2.77sc). The s value was also determined by fitting a decay of photovoltage transi- ent with exp(-t/s). The same NP and NR films as those used in the J-V tests were utilized to examine the characteristics of electron transport and charge recombination, neglecting the other factors. Figure 4a shows the D values of the NP-and NR-based DSSCs. The D values of both samples showed a power-law dependence on the light intensity,[21] indicating an aspect controlled by similar mechanistic factors. The D value of the NR film was slightly higher than that of the NP film, showing that NR films are more favorable property for electron transport than the NP films. Considering the condition

Nanoparticle Nanorod

/ nm

Illumination state

Dark state Potential / V sc / m

A c

Nanoparticle Nanorod λ/ nm

Illumination state

Dark state Potential / V sc / m

A c


Illumination state

Dark state Potential / V sc / m

A c

Figure 2. J–V characteristics of NP- and NR-based dye-sensitized solar cells. The illumination intensity of 100 mW cm–2 with AM 1.5 and active area of 0.25 cm2 were applied.

15 NP NR

Jsc / mA

Figure 3. Photocurrent density (Jsc) values of dye-sensitized NP and NR cells as a function of the number of adsorbed dye molecules. Lines were used for linear fits of data.

where small TiO2 NPs (< 10 nm) strongly bind to each other to decrease the unstable surface energy during thermal treat- ment, this minor improvement may be the reduction of electron loss in the grain boundaries resulting from the necking of NPs[2] as well as by the increase in average crystallite size. From the parallel slope of both samples, similar traps were distributed over the whole range of the film because the slope indicates a steeper trap-state distribution.[23] Figure 4b shows the values of s as a function of Jsc. Over the light intensity range, the s values of the NR films become an order of magni- tude higher than those of the NP films. The lower lifetime of the NP films might be due to the combined effect of the downward band-edge shift confirmed by the relatively low Voc value measured from the J-V test and the increased charge re- combination rate supported by the dark current measurements. This is mainly illustrated by the effects of the surface state, which is a factor that forms intraband gap states and enormous electron loss between the grain boundaries experiencing tens of thousands trapping/detrapping events during their transit through the film.[24,25] Therefore, the average electron diffusion length (L) of the NR film was much higher than that of the NP film due to the increase in the D and s val- ues according to Einstein’s relationship L =( Ds)–0.5.[26] The average L values of the NP and NR film were estimated to be approximately 7.4 lm and 1.3 lm, respectively. This means that the electrons of the NP film spend a large fraction of their transit time at the intraband trap sites (mostly positioned -0.13 eV below the Fermi level),[27] while the electrons of the NR film also experience the trapping/detrapping events in a less obstructed environment, as shown in Scheme 1. The phenomenon shows that the more efficient DSSCs will be real- ized in a thicker TiO2 nanorod layer than that in the NP film, because both the enhanced electron transport through the ge- ometry effect and the increased electron lifetime by the suppression of charge recombination can contribute to the increase in electron diffusion length. In briefly, the high efficiency of TiO2 NR-based DSSCs was outlined by the coupled factors of the slightly improved electron diffusion coefficient in the situation of a shorter electron transporting layer and the dramatically increased electron lifetimes by retarding charge recombination between the photoinjected electrons and cations of the dye molecules or redox electrolyte. This means that more electrons surviving from the back-reac- tion can contribute to the improvement of Jsc in a favorable situation for electron transport along the photoanode layer, leading to the enhancement of the conversion efficiency. In conclusion, nanorod-based DSSC showed improved performance with a Voc of 0.68 V, a Jsc of 15.3 mA cm–2, a fill factor of 0.6, and an efficiency of 6.2 %, compared with

(Parte 1 de 2)