Zhaosheng
Xue
ab,
Wei
Zhang
a,
Xiong
Yin
ac,
Yueming
Cheng
a,
Long
Wang
a and
Bin
Liu
*ad
aDepartment of Chemical and Biomolecular Engineering, National University of Singapore, Singapore 117576. E-mail: cheliub@nus.edu.sg
bNUS Graduate School for Integrative Sciences & Engineering (NGS), National University of Singapore, Singapore 117456
cNational Center for Nanoscience and Technology, P. R. China
dInstitute of Materials Research and Engineering, 3 Research Link, Singapore 117602
First published on 28th May 2012
To optimize the conversion efficiency of plastic dye-sensitized solar cells fabricated by the electrophoretic deposition technique, anatase TiO2 nanoparticles of various sizes from 10 nm to 27 nm have been synthesized via a simple hydrothermal process. The obtained TiO2 nanoparticles have been characterized by X-ray diffraction and high resolution transmission electron microscopy, which confirmed that the synthesized nanoparticles are in the pure anatase phase. Rigid devices based on D149-sensitized TiO2 particles with a size of 19 nm showed the highest conversion efficiency of 7.0% among the four different devices, which was measured under illumination of AM 1.5G, 100 mWcm−2. The effect of the particle size on the photovoltaic performance of DSSCs has been systemically studied using photoelectrochemical characterizations, including intensity modulated photocurrent spectroscopy and intensity modulated photovoltage spectroscopy. The good photovoltaic performance for 19 nm TiO2 is ascribed to the good dye loading, an efficient electron transport and the high charge collection efficiency in the photoanode. Moreover, plastic DSSCs based on 19 nm TiO2 presented a conversion efficiency of 6.0% (AM 1.5G, 100 mWcm−2) under optimized conditions, showing about a 20% enhancement in the conversion efficiency as compared to that based on commercial Degussa P25 TiO2 (5.2%). These results demonstrate that optimization of the TiO2 nanoparticle size for devices fabricated using the EPD technique is an alternative method to achieve highly efficient plastic dye-sensitized solar cells.
The low temperature requirement for polymer substrates has been satisfied by numerous methods, namely chemical deposition,3,4 binder-free coating,5–7 hydrothermal synthesis,8,9 compression 10,11 and transfer.12
Electrophoretic deposition (EPD) is a widely used method to produce thin films for applications in DSSCs 13–18 and other areas.19–21 In a typical EPD process, the particles to be deposited are suspended in a solvent to form a stable colloid. Two electrodes are inserted into the colloidal suspension and a potential difference is applied across the electrodes. Charged particles in the suspension are attracted to the electrodes via electrostatic attraction, thereby forming uniform thin films. The EPD process has several advantages over other methods in the low temperature fabrication of DSSCs. Such advantages include, but are not limited to, the uniformity of the films formed and the ease of scale-up for mass manufacturing. In addition, EPD can be done without any organic binders and additives and hence there is no need for a sintering process. Of particular importance is that the process only takes a few minutes22 and the film thickness can be controlled easily by tuning the operating voltage, current, concentration of the particles in the colloidal suspension and the deposition time. Though EPD is a common technique used in low temperature DSSC fabrication, the influence of the nanoparticle size on the device performance has, to our knowledge, not been well studied. We note that similar nanoparticle size-efficiency studies have been done for conventional high temperature DSSCs.23–26 However, the high temperature sintering process introduces physical changes, such as in the pore size distribution, the porosity and the grain size of the particles in relation to the films, which may render the studies irrelevant to low temperature DSSCs. As EPD can potentially be used for the large scale production of DSSCs in future commercialization, it is important to optimize the efficiency of DSSCs fabricated using EPD by studying the effect of the nanoparticle size on the device performance. In this report, the D149 dye was selected as the sensitizer because it has a large light absorption coefficient, high power conversion efficiency in traditional DSSCs 27,28 and can be easily synthesized in a large quantity without resource limitations (unlike Ru-based dyes). The chemical structure and UV absorption of the D149 dye on TiO2 is shown in Fig. 1. In the present work, TiO2 nanoparticles of various sizes were synthesized and used in the low temperature fabrication of DSSC via EPD. The effect of the nanoparticle size on the device performance was systematically studied. It was found that an average size of 19 nm has the best device performance of 7.0% on a rigid glass substrate and 6.0% on a flexible polymer substrate. This is superior to P25-based devices, which produce efficiencies of 5.5% and 5.2%, respectively, under the same conditions. This study will assist the optimization of DSSC efficiency for devices made from non-sintered TiO2 films.
![]() | ||
Fig. 1 The chemical structure and UV absorption spectrum of D149 when it is adsorbed on a thin film of TiO2 from a solution of acetonitrile/tert-butylalcohol (v/v = 1![]() ![]() |
Samples | Crystallite size a | Particle size b | Peptizing agent | Peptization temperature | Peptization time | Post treatment |
---|---|---|---|---|---|---|
a According to the Scherrer equation b According to TEM | ||||||
A | 10 nm | 11 ± 2 nm | 1.2 mL Conc. HNO3 | 200 °C | 12 h | 1.2 mL |
B | 14 nm | 15 ± 2 nm | 1.2 mL Conc. HNO3 | 250 °C | 15 h | 1.2 mL |
C | 19 nm | 20 ± 3 nm | 3.6 mL Conc. HNO3 | 250 °C | 12 h | 3.6 mL |
D | 27 nm | 29 ± 4 nm | 22 mL 20 wt% TEAH solution | 200 °C | 5 h | 5.0 mL |
3.6 g (0.06 mol) of CH3COOH was added to 17.6 g of titanium tetraisopropoxide and the mixture was stirred at room temperature for 15 min. The mixture was then poured into 90 mL of deionized water while stirring rapidly to prevent aggregation. The resulting milky white suspension was stirred for 1 h at room temperature. Varying amounts of concentrated HNO3 or tetraethyl ammonium hydroxide (TEAH) were added to prepare TiO2 nanoparticles of different sizes. After that, the suspension was heated to 80 °C and allowed to react for another 75 min. Upon cooling to room temperature, the total volume was made up to 110 mL by adding deionized water. The suspension was transferred to a stainless steel autoclave and heated at varying temperatures and times to yield nanoparticles of different sizes. Upon cooling, varying amounts of concentrated HNO3 were added as a post-treatment step. The mixture was sonicated for 30 min before centrifugation. Deionized water was added to the sediment and the sonication/centrifuge process was repeated as a washing step. The washing process was repeated until the pH of the resulting supernatant reached 7.0. The washing process was then repeated with ethanol 3 times before drying the nanoparticles at 70 °C. The TiO2 flakes were subsequently made into powders by grinding with a mortar and pestle.
The rigid Pt counter electrode was prepared by spin-coating a 30 mM H2PtCl6 solution in isopropanol on a FTO glass substrate, followed by thermal decomposition at 400 °C for 15 min. The flexible Pt counter electrode was fabricated by sputtering Pt (20 mA for 120 s) on the PEN–ITO substrate using a JEOL JFC-1600 Auto Fine Coater. DSSCs were fabricated by sandwiching a TiO2 photoanode and a Pt counter electrode with the electrolyte in a 25 μm thick spacer. The electrolyte used had a composition of 0.5 M tetrabutylammonium iodide, 0.001 M LiClO4, 0.5 M 4-tert-butylpyridine and 0.1 M I2 in 3-methyoxypropionitrile. The active area of the cell was 0.16 cm2, which was controlled by a metal mask.
![]() | (1) |
where τ is the crystallite size (in Å), λ is the wavelength of radiation used (1.54056 Å for Cu–Kα1), β is the width of the peak at half maximum intensity (in radians) and θ is the diffracted angle at maximum intensity (in radians). The crystallite sizes calculated from the Scherrer equation generally agree with that observed by TEM. The crystallite size of the nanoparticles determined via the Scherrer equation and the nanoparticle size determined using TEM are summarized in Table 1.
![]() | ||
Fig. 2 TEM images of the synthesized TiO2 nanoparticles. (A) ∼10 nm, (B) ∼14 nm, (C) ∼19 nm and (D) ∼27 nm. The scale bars represents 50 nm. |
![]() | ||
Fig. 3 XRD spectra of TiO2 nanoparticles. (A) ∼27 nm, (B) ∼19 nm, (C) ∼14 nm and (D) ∼10 nm. The peaks of anatase are labelled. No change was observed after compression at 1 ton cm−2. |
The synthesized nanoparticles, after treatment with acid, have zeta potentials of ∼ +25 mV in the alcohol mixture (n-butanol:
isopropanol
:
ethanol v/v/v = 4
:
2
:
1) used in the EPD process. The charge present on the TiO2 nanoparticles is pH dependent and, under acidic conditions, positive charges are induced.31 Due to the positive charges present on the nanoparticle surface, the colloidal suspensions are relatively stable during the EPD process. As EPD utilizes electrostatic attraction for particle deposition, the positive charge favors their direct utilization in the EPD process without any further additives or surface modifications, which are common required for particles lacking surface charges.32
The films made from EPD followed by a compression process were mechanically stable and robust. The film, when formed on a flexible plastic substrate, can be bent easily without mechanically destroying the film. This is unlike unprocessed EPD films that need to be handled delicately in order to preserve the mechanical integrity of the film. To study the difference in film morphology brought about by compression, the films before and after compression were viewed by FESEM. Fig. 4A shows clearly that the unpressed EPD film fabricated from 19 nm TiO2 nanoparticles had visible micropores and the interparticle connectivity was poor. After compression, however, it can be clearly seen in Fig. 4B that the interparticle connection has been improved and less interparticle pores are observed. The FESEM images indicate that compression treatment can reduce local interparticle pores and improve interparticle connectivity. In addition, the film thickness was reduced by the compression process, indicating that void spaces within the uncompressed film have been significantly reduced and the nanoparticles within the film are in closer contact after compression. The films fabricated using the other nanoparticles showed the same observations as well. In addition to the FESEM studies, the effect of compression on the crystal phase of the particles was also studied. XRD did not show any change in crystal phase and crystallite size in the pressed samples.
![]() | ||
Fig. 4 FESEM images of (A) as prepared EPD films from 19 nm TiO2 nanoparticles. (B) After compression. The scale bars represents 100 nm. |
The devices have a sandwiched structure in which the D149 sensitized photoanode and the Pt counter electrode are clipped together. A 25 μm spacer (Solaronix) was used to prevent the short circuiting of the cell. A few drops of the electrolyte, containing 0.5 M tetrabutylammonium iodide, 0.001 M LiClO4, 0.5 M 4-tert-butylpyridine and 0.1 M I2 in 3-methyoxypropionitrile, was introduced to complete the cell assembly.
The effect of the TiO2 film thickness on the device performance was first investigated for the nanoparticles. The post compression thickness was controlled and set to be ∼4, ∼8, ∼11 and ∼15 μm. The changes in the open circuit voltage (Voc), the short-circuit current density (Jsc), the fill factor (FF) and the overall efficiency (η) with the TiO2 film thickness are shown in the ESI (Tables S1–S4†).
The best performing devices fabricated using different TiO2 nanoparticles have different TiO2 thicknesses. For 10 nm and 14 nm nanoparticles, their corresponding devices performed optimally at a post compression film thickness of ∼8 μm. For the larger 19 nm and 27 nm nanoparticles, the optimal film thickness was ∼11 μm. The J–V curves of the best performing devices for each individual set of nanoparticles are shown in Fig. 5 and the corresponding photovoltaic parameters are shown in Table 2. For the optimized devices, the Voc of the devices increased from 0.77 to 0.85 V when the nanoparticle size was increased from 10 nm to 27 nm. The Jsc of the devices increased from 10.6 to 14.3 mA cm−2 when the nanoparticle size was increased from 10 nm to 19 nm. When the size of the nanoparticles was further increased to 27 nm, the Jsc decreased to 9.2 mA cm−2. The FF of the devices was between 0.63 to 0.68. The best performing devices fabricated using 10, 14, 19 and 27 nm nanoparticles yielded efficiencies of 5.3%, 5.4%, 7.0% and 5.3%, respectively.
![]() | ||
Fig. 5 J–V curves for the best performing DSSCs fabricated on rigid glass substrates by the EPD technique with compression post treatment. |
Particle size (nm) | Film thickness (μm) | Zeta potential a (mV) | Photovoltaic properties b | Dye loading c (×10−7 mol cm−2) | |||
---|---|---|---|---|---|---|---|
V oc (V) | J sc (mA cm−2) | FF | ECE (%) | ||||
a Measured in a mixture of ethanol, isopropanol and butanol (v/v/v = 1![]() ![]() ![]() ![]() |
|||||||
10 | 8 | 26.1 | 0.77 | 10.6 | 0.64 | 4.8 | 2.5 |
14 | 8 | 24.7 | 0.79 | 10.7 | 0.63 | 5.4 | 1.4 |
19 | 11 | 24.0 | 0.79 | 14.3 | 0.65 | 7.0 | 1.3 |
27 | 11 | 24.9 | 0.85 | 9.2 | 0.68 | 5.3 | 0.9 |
The Voc of a device is determined by a number of factors,33 namely the Quasi Fermi level of the semiconductor, the HOMO level of the redox couple (Eredox), the light harvesting and electron injection of the sensitizer and the frequency of recombination events within the cell. Since the semiconductor, sensitizer and electrolyte used in this study are the same for all devices, the Voc trend cannot be attributed to Eredox and the identity of the dye. This leaves the recombination rate as the likely reason for such an observed Voc trend.
To understand the observed trend for the Voc of the devices made from different nanoparticles, intensity-modulated photovoltage spectroscopy (IMVS) was performed. The electron lifetime (τn) was calculated from the equation τn = 1/(2πfn,min), where fn,min is the characteristic frequency at the minimum of the IMVS imaginary component.34 As shown in Fig. 6A, the electron lifetime increased from 5.7 ms to 10 ms when the nanoparticle size was increased from 10 nm to 27 nm. As the electron lifetime calculated from the IMVS results is an indication of the number of recombination events, the longer electron lifetime for DSSCs with 27 nm nanoparticles as compared to that with 10 nm nanoparticles indicated that devices fabricated with larger particles undergo less recombination events.24 The IMVS results indicate that electron recombination has been reduced when larger nanoparticles are used and this contributes to a higher Voc for the corresponding devices.
![]() | ||
Fig. 6 (A) Electron lifetime, (B) electron transport time and (C) charge collection efficiency for thickness-optimized DSSCs fabricated with different sizes of TiO2 nanoparticles. |
To check the contribution of dye loading on Jsc for each device, the desorbed dye solutions were measured with UV absorption spectroscopy and the results are summarized in Table 2. It was found that the nanoparticle size influences the dye loading significantly in our devices. For the thickness-optimized devices, the dye loading decreases from 2.5 to 0.9 × 10−7 M cm−2 when the nanoparticle size is increased from 10 nm to 27 nm. Though the dye loading is related to the device's ability to absorb incident light, the Jsc trend of the devices differs from the dye loading trend. This indicates that the dye loading of the devices is not the sole determinant of the Jsc and the overall device conversion efficiency.
In addition to the dye loading, the electron transport within the TiO2 network could also play a part in determining the Jsc of the devices. To understand the electron transport within the photoanode fabricated from the differently sized TiO2 nanoparticles at their optimal thicknesses, intensity modulated photocurrent spectroscopy (IMPS) was performed on the devices fabricated using the synthesized nanoparticles. The electron transport time (τd) was calculated from the equation τd = 1/(2πfd,min), where fd,min is the characteristic frequency at the minimum of the IMPS imaginary component.34 As shown in Fig. 6B, the electron transport time decreases from 3.2 to 1.2 ms when the nanoparticle size is increased from 10 nm to 27 nm. The electron transport time is a measure of the average time taken for the collection of injected electrons and a faster electron time is associated with a higher photocurrent as it indicates that electrons hop across the TiO2 network and are collected at the photoanode at a faster rate.35
From the calculated τd and τn, the charge collection efficiency can be calculated using the equation: ηc = 1 − τd/τn.36 The calculated values of ηc are shown in Fig. 6C. The charge collection efficiency of the devices increased from 0.46 to 0.88 when the TiO2 nanoparticle size is increased from 10 to 27 nm. The improvement of the charge collection efficiency with the nanoparticle size can be explained using the electron trapping/detrapping model.37–39 Larger nanoparticles have both a lower surface area and a smaller number of grain boundaries, which lead to less electron trapping and is beneficial for a high Jsc and Voc.35–37 Although the charge collection efficiency data predicts that the 27 nm TiO2-based devices should have the highest photocurrent, this is not the case for the Jsc in these devices and this indicates that the charge collection efficiency alone is not indicative of the overall device performance.
From the above studies, it is clear that the device performance is dependent on the optimization of the charge collection efficiency and the dye loading. Small nanoparticles have a higher dye loading for a fixed film thickness and, hence, thinner films can be used for devices fabricated with 10 and 14 nm TiO2. However, devices fabricated with small nanoparticles exhibit poor charge transport properties and this results in a poor Jsc. An increasing particle size decreases the dye loading and increases the film thickness required for optimal performance. However, devices fabricated with larger particles show improved charge collection efficiencies. Since the films used in this study are not sintered, the thickness for which a crack-free film can be fabricated is limited and the requirement for thick crack-free films cannot be fulfilled when the large 27 nm nanoparticles are used. In our results, a moderate nanoparticle size of 19 nm shows the best efficiency. This is ascribed to the 19 nm-based devices' relatively high dye loading, as well as the good charge collection efficiency.
For comparison purposes, a device was fabricated with commercially available Degussa P25 TiO2 nanoparticles. Using the same fabrication method, P25 was deposited via EPD and used as a photoanode in DSSCs. The change in the photovolatic parameters with the film thickness is shown in Table S5† in the ESI. The optimal post compression thickness is found to be ∼ 11 μm. FESEM images of the film before and after compression are shown in Fig. S1† in the ESI.
Flexible DSSCs were fabricated with 19 nm and P25 nanoparticles. The post compression thickness of the TiO2 films was controlled and set to be ∼ 11 μm, which is the optimal thickness for both 19 nm and P25 devices. DSSCs with 19 nm nanoparticles as the photoanode have a Voc of 0.80 V, a Jsc of 10.3 mAcm−2, a FF of 0.62 and an overall efficiency (η) of 5.2%. P25-based devices have a Voc of 0.79 V, a Jsc of 9.32 mAcm−2, a FF of 0.60 and an overall efficiency (η) of 4.5%. When compared to the devices on rigid glass substrates, the efficiency is lower. This is mainly due to the drop in the photocurrent and FF, which is caused by the lower light transmittance and higher electrical resistance in the plastic substrate when compared to glass substrates.40,41
To compensate for the loss of light transmission, large submicron TiO2 particles (∼200 nm–300 nm) were subsequently synthesized41 and utilized as a scattering layer to offset the light loss caused by the plastic substrate. Based on an identical deposition method, a 2 μm scattering layer was deposited on top of the nano-sized TiO2 particles. The J–V curves of these devices are shown in Fig. 7. As such, the efficiency of the devices is significantly improved. The 19 nm nanoparticle-based devices with the scattering layer have a Voc of 0.80 V, a Jsc of 13.8 mAcm−2, a FF of 0.54 and an overall efficiency (η) of 6.0%. The P25-based device with the scattering layer has a Voc of 0.78 V, a Jsc of 11.5 mAcm−2, a FF of 0.56 and an overall efficiency (η) of 5.2%. The light scattering effect of the large submicron sized nanoparticles is well known to increase the light harvesting and hence the Jsc of the devices. 41,42 This leads to an overall increase in the device efficiency as compared to the devices without the scattering layer. As with the devices on the rigid substrates, the performance of the 19 nm nanoparticle-based devices is superior to that of P25. Upon optimizing the particle size for DSSCs fabricated via EPD, the use of a blocking layer to limit recombination,43,44 UV–O3 treatment to reduce the organic contaminants and an anti-reflection film11 for higher light utilization is expected to further improve the device efficiency.
![]() | ||
Fig. 7 J–V curves for flexible DSSCs fabricated on plastic substrates. 19 nm particles or P25 were used as the mesoporous layer and large particles (∼200–300 nm) were used as light scattering layers. The inset shows a typical TiO2 film, formed by EPD and compression on ITO–PEN, sensitized with the D149 dye. |
Footnote |
† Electronic Supplementary Information (ESI) available. See DOI: 10.1039/c2ra20542d/ |
This journal is © The Royal Society of Chemistry 2012 |