Plasmon-enhanced photocurrent in quasi-solid-state dye-sensitized solar cells by the inclusion of gold/silica core–shell nanoparticles in a TiO2 photoanode†

Direct evidence of the effects of the localized surface plasmon resonance (LSPR) of gold nanoparticles (Au NPs) in TiO2 photoanodes on the performance enhancement in quasi-solid-state dye-sensitized solar cells (DSCs) is reported by comparing gold/silica core–shell nanoparticles (Au@SiO2 NPs) and hollow silica nanoparticles with the same shell size of the core–shell nanoparticles. The Au nanoparticles were shelled by a thin SiO2 layer to produce the core–shell structure, and the SiO2 hollow spheres were made by dissolving the Au cores of the gold/silica core–shell nanoparticles. Therefore, the size and morphology of the SiO2 hollow spheres were the same as the Au@SiO2 NPs. The energy conversion efficiency was improved nearly 36% upon incorporating the Au nanoparticles, mostly due to the increase in Jsc, while Voc and FF were unchanged. The improvement was mostly contributed by the LSPR of the Au@SiO2 NPs, whereas the other parameters, such as the electron lifetime and electron diffusion coefficient, were nearly unchanged. Therefore, LSPR is an effective tool in improving the photocurrent and consequently the performance of DSCs.


Introduction
Dye-sensitized solar cells (DSCs) have currently received much attention due to several advantages, such as the low fabrication cost and high power conversion efficiency greater than 12% under 1 sun illumination condition. [1][2][3][4] The most widely studied DSC is comprised of an electrolyte sandwiched between two electrodes coated on a transparent conducting glass, such as uorine-doped tin oxide (FTO) glass; a photoanode and counter electrode. The photoanode consists of a mesoporous semiconductor such as a TiO 2 layer with sensitizers, whereas a typical counter electrode is made of a reduction catalyst such as platinum coated onto FTO. Upon light illumination, dye sensitizers adsorbed to the surface of the mesoporous TiO 2 layer generate electrons, which are subsequently injected into the TiO 2 layer for electricity production. Therefore, light absorption by sensitizers on the photoanodes plays a major role in determining the overall energy conversion efficiency of DSCs. A large body of research has been conducted to enhance the light harvesting efficiency in TiO 2 photoanodes. In this context, the development of more efficient dye sensitizers, including organic dyes with a higher extinction coefficient 5,6 and energy-relay dyes, 7 and their effective utilization methods, such as cocktail dye 8,9 and selective dye adsorption 10,11 concepts, have prevailed. In addition, the introduction of a scattering layer and inverse opal nanostructures are also common. [12][13][14] The localized surface plasmon resonance (LSPR) phenomena of metal nanoparticles has also been investigated to enhance light harvesting efficiency. [15][16][17] The LSPR, which refers to the resonance between the electromagnetic eld and free-electron oscillation, amplies the electromagnetic eld near the metal nanoparticles, resulting in plasmon enhanced light absorption of dye sensitizers in DSCs. [18][19][20][21][22][23] The Hupp group rst reported plasmon enhanced light harvesting in DSCs using silver † Electronic supplementary information (ESI) available: Device characterization with incident-modulated photovoltage spectroscopy (IMVS) and incident-modulated photocurrent spectroscopy (IMPS). See DOI: 10.1039/c3ta11712j ‡ These authors contributed equally to this work. nanoparticles on at TiO 2 lm, demonstrating considerable potential to increase the photocurrent. 24,25 The photocurrent and power conversion efficiency of the DSC increased nearly 6-7-fold upon incorporating silver nanoparticles into a dye monolayer on the at TiO 2 lm (J sc from 14.6 to 85.7 mA cm À2 , from 0.007 to 0.045%). Recently, enhanced charge carrier generation in solidstate DSCs was demonstrated by the LSPR effects of Au NPs coated on a mesoporous TiO 2 photoanode 26 and a hexagonal array of Ag nanodome-structured counter electrode. 27 Direct effects of LSPR by metallic nanoparticles on the performance of DSCs may not be readily evaluated in common I À /I 3 À redox couples which dissolve metallic nanoparticles, such as gold, by the following reaction: 28 One way to avoid the dissolution problem of metallic nanoparticles is to create a shell with an inert material, such as SiO 2 . In this study, Au@SiO 2 NPs were synthesized in a solution process and mixed with a TiO 2 paste to fabricate Au@SiO 2 NPs incorporated mesoporous TiO 2 photoanodes. In addition, we used polyethylene glycol (PEG) based electrolyte to achieve stability of the Au core, inhibiting interaction between the Au core of Au@SiO 2 NPs and I À /I 3 À ions in the electrolyte. However, the properties of the core-shell nanoparticle-incorporated photoanodes were affected by both the metal cores and shell. Thus, the effects of the shell material on the cell performance must be considered. Until recently, even though a number of researches have been presented to improve the cell performance by utilizing the LSPR effects with metal/TiO 2 or SiO 2 core-shell structures, the quantitative analyses of the LSPR by metal cores and other changes in the photoanodes by shells are still difficult to separately evaluate. [24][25][26] This problem may be solved by comparing the results of the same photoanode structures with and without core metal: core-shell and hollow shell. The hollow shell structure can be prepared by dissolving the core metal of the core-shell sphere by the dissolution reaction with I À /I 3 À , which readily diffuses through a shell layer, such as SiO 2 . 29 In this context, SiO 2 hollow spheres-incorporated TiO 2 photoanode was fabricated to quantify the effects of the LSPR clearly by the Au nanoparticles without disturbing the shell properties or structure. The TiO 2 photoanodes incorporating SiO 2 hollow spheres have the same morphology as the initial TiO 2 photoanodes incorporating Au@SiO 2 NPs, which is helpful for accurate comparison of photoanodes with and without LSPR. Through this novel approach, the LSPR effect in DSCs can be independently demonstrated with the effect of SiO 2 shells, such as the charge injection problem from dyes into the SiO 2 shell and the change of morphology and resistance.

Synthesis of gold nanoparticles
Au NPs were prepared using a seed-mediated method. 30 First, 15 nm-diameter Au NP seeds were synthesized via citrate reduction. In a typical procedure, 10 ml of a 1 mM gold(III) chloride trihydrate aqueous solution was reuxed at boiling temperature under vigorous stirring, followed by the quick injection of 1 ml of a 39 mM sodium citrate solution. Aer 15 min, the heating was stopped, and the reaction contents were cool to room temperature. To make larger nanoparticles, a 2 ml seed particle solution was added to 100 ml of a 0.5 mM gold(III) chloride trihydrate aqueous solution containing 0.03 M CTAB and 1 mM ascorbic acid. The solution was reacted for 4 hours, and the product was collected using centrifugation (9000 rpm, at room temperature for 10 min).

Synthesis of gold/silica core-shell nanoparticles
Au@SiO 2 NPs were synthesized by a modied procedure previously reported by Obare et al. 29 In this method, growth of silica was performed aer surfactant substitution with a silane coupling group, (3-mercaptopropyl)trimethoxysilane (MPTMS). MPTMS in ethanol was added to the Au NPs solution. Aer three hours, an aqueous sodium silicate solution was added and reacted for three additional days. The contents were puried several times by precipitation using centrifugation and were redispersed in ethanol.

Paste preparation
As-prepared Au@SiO 2 NPs dispersed in ethanol were added and well mixed with the commercial titanium dioxide (TiO 2 ) paste with an average size of 20 nm (DSL 18NR-T, Dyesol). In order to achieve the same thickness of photoanodes aer sintering, excess ethanol from the paste was evaporated using nitrogen to produce a homogenous concentration of paste materials.

Device fabrication
For the formation of an electron blocking layer between the FTO substrate and oxidized species in the electrolyte, 0.1 M of Ti(IV) bis(ethyl acetoacetato)diisopropoxide in a 1-butanol solution was spin-coated on FTO glass (TEC 8, Pilkington) followed by sintering at 500 C. TiO 2 photoanodes were fabricated on the blocking layer with TiO 2 paste using a doctor blade method followed by sintering at 500 C for 15 min. Subsequently, TiO 2 nanostructure-coated FTO substrates were dipped into 40 mM TiCl 4 in H 2 O solution at 70 C for 30 min and sintered at 500 C for 15 min. TiO 2 photoanodes were dipped into the 0.3 mM N719 dye (cis-bis(isothiocyanato)bis(2,2 0 -bipyridyl-4,4 0 -dicarboxylato) ruthenium(II) bis-tetrabutylammonium, Dyesol) in an acetonitrile and tert-butanol solution (1 : 1 v/v) at 30 C for 18 hours and then rinsed with acetonitrile and dried using a stream of nitrogen. A Pt counter electrode was prepared by thermal decomposition of 0.01 M H 2 PtCl 6 in an isopropyl alcohol solution on the FTO substrate followed by sintering at 500 C for 30 min. Aer loading the dyes onto the TiO 2 electrodes, Surlyn (25 mm, Solaronix) was attached to the TiO 2 photoanode as a spacer. The polymer electrolyte was spread on the spacer gap, and the Pt counter electrode was placed on top.

Solar cell characterization
The thickness of TiO 2 lms was characterized with a surface proler (alpha-step IQ, Tencor) and eld emission scanning electron spectroscopy (JSM-6701F, JEOL). Absorption properties of photoanodes were measured by UV-Vis spectroscopy (V-670 UV-Vis spectrophotometer, Jasco) with an integrating sphere. Current-voltage characterization of the DSCs was performed with a Keithley 2400 digital source meter and solar simulator equipped with a 300 W Xenon arc-lamp (Newport) under a 1 sun illumination (AM 1.5, 100 mW cm À2 ). The light intensity was calibrated by a silicon solar cell (PV measurement). In addition, the quantum efficiency of DSCs was analyzed by an incident photon to current efficiency (IPCE) (PV measurements, Inc.) as a function of wavelength. The charge transfer resistance and electron lifetime in the photoanodes were characterized by electrochemical impedance spectroscopy (EIS) using an IM6 (Zahner) under dark conditions with a bias potential of À0.54 V. The frequency was in the range of 1 MHz to 0.1 Hz, and the amplitude was xed to 10 mV. The obtained spectra were tted and analyzed using Z-View soware with equivalent circuits.
The electron diffusion coefficient and electron lifetime of the photoanodes were evaluated by intensity-modulated photocurrent spectroscopy (IMPS) under short-circuit conditions and intensity-modulated photovoltage spectroscopy (IMVS) under open-circuit conditions as a function of light intensity using a controlled intensity modulated photo spectroscopy (CIMPS) system (Zahner) and a white light source (Zahner). The detailed measurement conditions are described elsewhere. 32 Even though the Au NPs were protected by the SiO 2 shell, the Au cores were dissolved by contact with I À /I 3 À ions penetrating the thin silica shell in a few hours. Therefore, the SiO 2 shell was treated with TiCl 4 to block the penetration of I À /I 3 À ions and consequently improve the stability of the Au core nanoparticles against dissolution. In addition, a poly(ethylene glycol) dimethylether (PEGDME, M w : 500)-based polymer electrolyte, instead of typical acetonitrile-based liquid electrolytes, was used for quasi-solid-state DSCs in order to retard the possible penetration of the I À /I 3 À ions through the SiO 2 shell. Based on our experimental results, a SiO 2 shell thinner than 8 nm hardly protected the Au core from the electrolyte contact, even though high M w PEGDME was applied as a viscous solvent for quasisolid electrolytes. Experimental data suggests that 10 nm was the minimum thickness of the SiO 2 shell necessary to protect the Au core from the dissolution.

Results and discussion
The localized surface plasmon resonance (LSPR) effects of Au@SiO 2 NPs by varying the Au core size were characterized using UV-vis spectroscopy, as shown in Fig. 1d. The absorption peak appears at 537, 547, and 565 nm for Au@SiO 2 NPs with the size of the Au core/SiO 2 shell 30/12, 50/11, and 160/10 nm, respectively. This shi in the absorption band is attributed to the change in the oscillation frequency of LSPR caused by varying the average diameter of the Au NPs. The absorption band of LSPR shis to a longer wavelength by increasing the size of Au NPs as a result of the decrease in the oscillation frequency. The coupling between the LSPR of the Au NPs and the absorption of dyes is one of the key factors for the enhanced performance of DSCs using Au NPs. In this case, the absorption peak difference or coupling wavelength mismatch of Au NPs (160 nm) with respect to the N719 dyes was $40 nm, as shown in Fig. 1(d). Additionally, when larger Au NPs were incorporated into the TiO 2 photoanodes, Mie scattering which is a long range effect should occur and it can be mixed with the effect of LSPR.
Alternatively, as the size of the Au NPs increases, the eld enhancement is more widely developed, leading to increased light harvesting of the dyes. Therefore, 50/11 nm Au@SiO 2 NPs were chosen for the fabrication of photoanodes to investigate the effects of LSPR in DSCs considering their coupling wavelength mismatch, the change in surface area, and the near-eld enhancement effects. Fig. 2a shows the schematic illustration of a Au@SiO 2 NP-incorporated TiO 2 photoanode. The LSPR of Au@SiO 2 NPs in TiO 2 photoanodes was observed by the reddish photoanode (inset photograph) and by the UV-vis spectrum, as shown in Fig. 2b. Fig. 2c shows Au@SiO 2 NPs in the photoanodes surrounded by TiO 2 NPs using scanning electron microscopy (SEM). Noticeably, the shapes of the Au@SiO 2 NPs were unchanged aer sintering at 500 C.
The photocurrent-voltage characteristics of DSCs with TiO 2 photoanodes incorporating Au@SiO 2 NPs are represented in Fig. 3. The lm thickness of photoanodes with and without Au@SiO 2 NPs was adjusted to 2 mm to more clearly characterize the effects of LSPR, which is thinner than a conventional TiO 2 layer (Fig. S1 †). In order to optimize the incorporation of Au@SiO 2 NPs for DSC performance, the concentration of Au@SiO 2 NPs in the TiO 2 paste was varied from 0.25 to 1.5 wt%. The short circuit current density (J sc ) was increased upon the incorporation of Au@SiO 2 NPs, while the open-circuit voltage (V oc ) and ll factor (FF) remained nearly unchanged. The J sc and power conversion efficiency (PCE) of DSCs with the addition of 1.0 wt% of Au@SiO 2 NPs into the TiO 2 layer were increased to 5.67 mA cm À2 and 2.66%, respectively, with respect to the same thickness reference TiO 2 photoanode without Au@SiO 2 NPs (4.35 mA cm À2 , 1.94%). However, at concentrations greater than 1.0 wt% Au@SiO 2 NPs, the J sc (5.44 mA cm À2 ) and PCE (2.53%) were slightly decreased, as shown in Fig. 3b and Table 1. The inclusion of Au@SiO 2 NPs in the photoanode may have possible side effects. First, the Au@SiO 2 NPs could inhibit the light absorption of dyes in the photoanodes while the Au@SiO 2 NPs in the photoanodes absorb the incident light as well as dyes but without converting photons to charges. 26 On that account, the light harvesting efficiency may slightly decrease when the concentration of Au@SiO 2 NPs in the photoanode becomes higher than the critical point. Secondly, Au@SiO 2 NPs with a size of $70 nm decrease the total amount of dye loading in the photoanodes due to the smaller surface area relative to 20 nm TiO 2 NPs. Finally, it is difficult to inject electrons from the excited dyes into the insulator SiO 2 shell. These side effects of the inclusion of Au@SiO 2 NPs may result in a decrease in photocurrent and consequently the photovoltaic performance of DSCs to a small extent. However, the overall energy conversion efficiency increased from 1.94 to 2.66%, which was a nearly 30% improvement, suggesting that the positive effects of LSPR are signicant. Therefore, the performance was further characterized in the following sections. In order to evaluate the quantitative effects of LSPR from Au cores excluding the SiO 2 shell effects, the Au@SiO 2 NP-incorporated photoanodes with and without the Au cores were compared. Experimentally, TiO 2 photoanodes incorporating SiO 2 hollow spheres with the same size of Au@SiO 2 NP shell but without the Au core were  introduced by dipping Au@SiO 2 NP-incorporated photoanodes in an I À /I 3 À liquid electrolyte for a few hours, which has the same morphology and thickness of the photoanode with Au@SiO 2 NPs. As expected, the SiO 2 hollow sphere was formed without a change in morphology due to the exclusive dissolution reaction eqn (1) of Au with I À /I 3 À ions, which was easily demonstrated by the disappearance of the reddish color and characterized by scanning electron microscopy, as shown by the inset photographs of Fig. 4 and S2(a). † Fig. 5a shows the decreased light absorption of N719 dyes upon incorporating the SiO 2 hollow spheres rather than Au@SiO 2 NPs, indicating the effects of the presence of Au cores. The amount of dyes adsorbed on the TiO 2 surface were characterized and this result shows that the dyes loaded in SiO 2 hollow sphere incorporated photoanodes are nearly the same as the amount in the Au@SiO 2 NPs incorporated photoanode (Fig. S2(c) †). The only difference between these photoanodes was the Au cores, suggesting that the enhanced light absorption is primarily attributed to the LSPR effects of the Au core. In order to conrm the LSPR effect more distinctly, we characterized absorption spectrum of the photoanodes additionally with N749 dyes (green dye) which absorb longer wavelength of light (>600 nm) compare to Au NPs ($530 nm), as shown in Fig. 5a. Through the distinguished peaks of LSPR and light absorption of N749 dyes, the effect of LSPR on the enhanced light absorption of dyes was clearly veried. Moreover, almost the same reectance of Au@SiO 2 NPs and SiO 2 hollow spheres incorporated photoanodes were characterized by UV-vis spectroscopy with an integrating sphere and both photoanodes show only $2% off-specular reection by scattered light (Fig. S3 †). This provides convincing evidence that absorption enhancement was mainly induced by LSPR which is near-eld effect and not Mie scattering which is far-eld effect. Fig. 5b shows the photocurrent-voltage (J-V) characteristics of DSCs that are consistent with the results obtained from the difference in the light absorption, as shown in Fig. 5a. DSCs based on the TiO 2 photoanode incorporating Au@SiO 2 NPs exhibited $28% greater J sc and PCE than those with the hollow SiO 2 spheres. The increase of J sc agrees well with the incident photon-to-current efficiency (IPCE) results, and the difference obtained by subtracting IPCE values increased at the same wavelength as the absorption band of LSPR (Fig. 5c). Moreover, the photoanode incorporating SiO 2 hollow spheres shows similar J sc of 4.4 mA cm À2 and PCE of 1.97% with respect to the reference TiO 2 photoanode, as summarized in Table 1. For the photoanodes incorporating SiO 2 hollow spheres, only the Au cores were removed from the Au@SiO 2 NPs photoanode, while   the surface area and morphology of the TiO 2 photoanode were unchanged. Thus, the small change of the photovoltaic performance between the reference and SiO 2 hollow sphere photoanodes suggested that the effects of the changes in surface area and morphology upon incorporating Au@SiO 2 NPs on the cell performance were nearly negligible. The performance enhancement upon incorporating Au@SiO 2 NPs into a photoanode is mostly due to the LSPR effects of the Au nanoparticles. Electrochemical impedance spectroscopy (EIS) was performed in dark conditions with a bias potential of À0.54 V (Fig. 6) to characterize the cell performance, and the performance parameters were obtained by tting with the general transmission model of DSCs. 33 In Fig. 6a, the Nyquist plots show two semicircles. The rst semicircle at high frequency was attributed to the charge transfer resistance at the Pt counter electrode-polymer electrolyte interface (R Pt ), and the second semicircle at mid-frequency was associated with the electron recombination resistance (R rec ) and capacitance (C) at the TiO 2polymer electrolyte interface. For Bode plots, the characteristic frequency peak in the mid-frequencies was unchanged (Fig. 6b), indicating nearly the same electron lifetimes for the two samples. The values are listed in the inset table of Fig. 6b.
The roles of the Au@SiO 2 NPs in the electron lifetime and the electron diffusion coefficient in the TiO 2 photoanodes were also evaluated with incident-modulated photovoltage spectroscopy (IMVS) and incident-modulated photocurrent spectroscopy (IMPS) as a function of the light intensity given in Fig. 7a. In accordance with the EIS measurements, the electron lifetime upon the inclusion of the Au@SiO 2 NPs was unchanged compared to that of the photoanode with the hollow SiO 2 spheres. Furthermore, the electron diffusion coefficient was also unchanged, and thus the diffusion lengths (L n , L n ¼ (D n s n ) 1/2 ) derived from these values were almost same between the photoanodes containing Au@SiO 2 NPs with and without the  Au core nanoparticles. In addition, the transient photocurrent and the diffusion coefficients of electrolyte (D[I 3 À ]) were measured, as shown in Fig. 7b and c. Nearly the same values of Au@SiO 2 NPs (2.36 Â 10 À7 cm 2 s À1 ) and hollow SiO 2 spheres (2.38 Â 10 À7 cm 2 s À1 ) incorporated photoanodes were evaluated. These results reveal that despite the enhanced light absorption of photoanodes by LSPR, the Au NPs had no inuence on the electrochemical properties in the photoanodes and electrolyte due to the presence of the insulating layer inhibiting interaction between Au cores and electrolyte.

Conclusions
Au@SiO 2 NPs incorporated into a conventional mesoporous TiO 2 photoanode resulted in a signicant increase in the energy conversion efficiency (up to 36% from 1.94 to 2.66% with a 2 mm-thick photoanode under 1 sun illumination condition) in quasi-solid state DSCs, mostly due to the enhanced photocurrent density from 4.35 to 5.67 mA cm À2 by the LSPR effects of the Au NPs. In addition, the LSPR effects were directly observed by comparing results between the Au@SiO 2 NPs-and SiO 2 hollow spheres-incorporated TiO 2 photoanodes, where the hollow spheres were obtained by dissolving the Au core with I À / I 3 À ions and had same morphologies as Au@SiO 2 NPs. The inuences from LSPR of the Au core in optical, electrochemical, and photovoltaic properties of the photoanodes were characterized by UV-vis spectroscopy and EIS measurements separate from the effect of SiO 2 shell and morphology change. From this, we demonstrated that the incorporation of the Au@SiO 2 NPs enhanced the light harvesting efficiency of dye molecules without changing the electron lifetime and diffusion coefficient of the TiO 2 photoanodes and were very effective in improving the power conversion efficiency of DSCs.