Ag Nanoparticle–Functionalized Open-Ended Freestanding TiO2 Nanotube Arrays with a Scattering Layer for Improved Energy Conversion Efficiency in Dye-Sensitized Solar Cells

Dye-sensitized solar cells (DSSCs) were fabricated using open-ended freestanding TiO2 nanotube arrays functionalized with Ag nanoparticles (NPs) in the channel to create a plasmonic effect, and then coated with large TiO2 NPs to create a scattering effect in order to improve energy conversion efficiency. Compared to closed-ended freestanding TiO2 nanotube array–based DSSCs without Ag or large TiO2 NPs, the energy conversion efficiency of closed-ended DSSCs improved by 9.21% (actual efficiency, from 5.86% to 6.40%) with Ag NPs, 6.48% (actual efficiency, from 5.86% to 6.24%) with TiO2 NPs, and 14.50% (actual efficiency, from 5.86% to 6.71%) with both Ag NPs and TiO2 NPs. By introducing Ag NPs and/or large TiO2 NPs to open-ended freestanding TiO2 nanotube array–based DSSCs, the energy conversion efficiency was improved by 9.15% (actual efficiency, from 6.12% to 6.68%) with Ag NPs and 8.17% (actual efficiency, from 6.12% to 6.62%) with TiO2 NPs, and by 15.20% (actual efficiency, from 6.12% to 7.05%) with both Ag NPs and TiO2 NPs. Moreover, compared to closed-ended freestanding TiO2 nanotube arrays, the energy conversion efficiency of open-ended freestanding TiO2 nanotube arrays increased from 6.71% to 7.05%. We demonstrate that each component—Ag NPs, TiO2 NPs, and open-ended freestanding TiO2 nanotube arrays—enhanced the energy conversion efficiency, and the use of a combination of all components in DSSCs resulted in the highest energy conversion efficiency.

To date, several approaches for improving the energy conversion efficiency of TiO 2 nanotube array-based DSSCs have been reported. Metal NPs, which can harvest light via surface plasmon resonance (SPR), have been used to enhance the energy conversion efficiency of DSSCs by introducing Au or Ag NPs into TiO 2 nanotube arrays [29][30][31][32]. Barrier layers remove TiO 2 nanotube arrays, so open-ended TiO 2 nanotube arrays, which can also be classified as arrays of columnar nanopores, have been used for DSSCs to provide increased energy conversion efficiency [33]. Moreover, the energy conversion efficiency of TiO 2 nanotube array-based DSSCs can be further increased by introducing a scattering layer to the active layer [34].
So far, TiO 2 nanotubes that make use of a scattering layer [34] or plasmonic materials [14] have been reported, but a scattering layer with plasmonic materials has not been used in TiO 2 nanotube-based DSSCs. In this study, we report the development of freestanding TiO 2 nanotube arrays filled with Ag NPs and large TiO 2 NPs, which improve the energy conversion efficiency of DSSCs. Furthermore, we compare the effects of Ag NPs and large TiO 2 NPs in open-and closed-ended freestanding TiO 2 nanotube arrays in DSSCs. The energy conversion efficiencies of the following eight types of DSSCs were compared: closed-ended freestanding TiO 2 nanotube arrays with/without Ag NPs and/or a TiO 2 scattering layer and open-ended freestanding TiO 2 nanotube arrays with/without Ag NPs and/or a TiO 2 scattering layer.

Results and Discussion
2.1. Structure of DSSCs with Freestanding TiO 2 Nanotube Arrays with Channels Containing Ag NPs Figure 1 illustrates the fabrication of DSSCs with Ag NPs and large TiO 2 NPs to enable plasmonic and scattering effects in open-ended freestanding TiO 2 nanotube array-based DSSCs. Ti plates were anodized and then annealed at 500˝C for 1 h to prepare anatase TiO 2 nanotube arrays. After carrying out secondary anodization, the TiO 2 nanotube arrays were easily detached from the Ti plates. TiO 2 nanotube arrays, once separated from the Ti plates, are termed "closed-ended freestanding TiO 2 nanotube arrays". Freestanding TiO 2 nanotube arrays have a barrier layer at the bottom that disturbs electron transport and electrolyte diffusion. This barrier layer was removed using the ion-milling method with several minutes of Ar + bombardment to yield "open-ended freestanding TiO 2 nanotube arrays". The closed-and open-ended freestanding TiO 2 nanotube arrays were transferred to fluorine-doped tin oxide (FTO) glass using TiO 2 paste and annealed to enhance the adhesion between the closed-and open-ended freestanding TiO 2 nanotube arrays and the fluorine-doped tin oxide (FTO) glass. To improve the energy conversion efficiency by the plasmonic effect, Ag NPs were embedded in the channel of freestanding TiO 2 nanotube arrays using 254 nm ultraviolet (UV) irradiation with aqueous silver nitrate. To further enhance the energy conversion efficiency, large TiO 2 NPs (400 nm) as a scattering layer were coated onto the active layer by the doctor blade method. This substrate was sandwiched with the counter electrode and filled with electrolyte. The active area of the DSSCs was 0.25 cm 2 .

Characterization of Freestanding TiO2 Nanotube Arrays with Channels Containing Ag NPs
Field emission scanning electron microscope (FE-SEM) images of freestanding TiO2 nanotube (TNT) arrays are shown in Figure 2. The top side of the freestanding TiO2 nanotube arrays had 100nm-diameter pores, as shown in Figure 2a. The bottom layer of closed-ended freestanding TiO2 nanotube arrays before ion milling lacked pores, as shown in Figure 2b. However, after ion milling, 20-nm-diameter pores were evident on the bottom layer of open-ended freestanding TiO2 nanotube arrays, as shown in Figure 2c. Open-ended TNT arrays can be prepared by chemical etching [35] or physical etching [33,36]. In the chemical etching method, the bottom layers of TNT arrays were easily removed by the etchant. However, the surface morphology and length of TNT arrays were also dissolved in etchant and TNT arrays are fragile when they are attached to a substrate because of their amorphous crystallinity. In the physical etching method, the bottom layer of TNT arrays was removed by the plasma or ion milling process, which is not simple. However, the surface morphology and length of TNT arrays are not damaged in the process and they are very stable when they are attached to a substrate because TNT arrays have the ability to change crystallinity from the amorphous to the anatase phase. After UV irradiation using a silver source, ~30 nm Ag NPs were seen in the channels of freestanding TiO2 nanotubes in high-angle annular dark-field (HAADF) images, as shown in Figure 2d. The length of the TiO2 nanotubes was ~22 µ m and the length of the scattering layer, which consisted of 400 nm TiO2 NPs, was ~10 µ m.

Characterization of Freestanding TiO 2 Nanotube Arrays with Channels Containing Ag NPs
Field emission scanning electron microscope (FE-SEM) images of freestanding TiO 2 nanotube (TNT) arrays are shown in Figure 2. The top side of the freestanding TiO 2 nanotube arrays had 100-nm-diameter pores, as shown in Figure 2a. The bottom layer of closed-ended freestanding TiO 2 nanotube arrays before ion milling lacked pores, as shown in Figure 2b. However, after ion milling, 20-nm-diameter pores were evident on the bottom layer of open-ended freestanding TiO 2 nanotube arrays, as shown in Figure 2c. Open-ended TNT arrays can be prepared by chemical etching [35] or physical etching [33,36]. In the chemical etching method, the bottom layers of TNT arrays were easily removed by the etchant. However, the surface morphology and length of TNT arrays were also dissolved in etchant and TNT arrays are fragile when they are attached to a substrate because of their amorphous crystallinity. In the physical etching method, the bottom layer of TNT arrays was removed by the plasma or ion milling process, which is not simple. However, the surface morphology and length of TNT arrays are not damaged in the process and they are very stable when they are attached to a substrate because TNT arrays have the ability to change crystallinity from the amorphous to the anatase phase. After UV irradiation using a silver source,~30 nm Ag NPs were seen in the channels of freestanding TiO 2 nanotubes in high-angle annular dark-field (HAADF) images, as shown in Figure 2d. The length of the TiO 2 nanotubes was~22 µm and the length of the scattering layer, which consisted of 400 nm TiO 2 NPs, was~10 µm. The ultraviolet-visible (UV-vis) spectrum of Ag NPs in the channels of freestanding TiO2 nanotubes is shown in Figure 3. A broad absorption peak centered at 402 nm was observed. The value is different from what it would be in the general solution phase, which is a 420 nm UV absorbance from 30 nm Ag NPs. This discrepancy may stem from different synthesis and measurement conditions used in this study [37][38][39]; the Ag NPs were synthesized by UV irradiation (at 254 nm) without adding a stabilizer and the Ag NPs were measured under dry conditions. The absorption band of Ag NPs is matched with the dye. cis-diisothiocyanato-bis(2,2'-bipyridyl-4,4'-dicarboxylato) ruthenium(II) bis(tetrabutylammonium) (N719) has two visible absorption bands, 390 nm and 531 nm, [40] that were affected by the plasmon band. Moreover, the shell of Ag NPs was prepared with TiCl4 to prevent the trapping of electrons by Ag NPs and to enable better electron transport in the channel of the TiO2 nanotube arrays. The ultraviolet-visible (UV-vis) spectrum of Ag NPs in the channels of freestanding TiO 2 nanotubes is shown in Figure 3. A broad absorption peak centered at 402 nm was observed. The value is different from what it would be in the general solution phase, which is a 420 nm UV absorbance from 30 nm Ag NPs. This discrepancy may stem from different synthesis and measurement conditions used in this study [37][38][39]; the Ag NPs were synthesized by UV irradiation (at 254 nm) without adding a stabilizer and the Ag NPs were measured under dry conditions. The absorption band of Ag NPs is matched with the dye. cis-diisothiocyanato-bis(2,2'-bipyridyl-4,4'-dicarboxylato) ruthenium(II) bis(tetrabutylammonium) (N719) has two visible absorption bands, 390 nm and 531 nm, [40] that were affected by the plasmon band. Moreover, the shell of Ag NPs was prepared with TiCl 4 to prevent the trapping of electrons by Ag NPs and to enable better electron transport in the channel of the TiO 2 nanotube arrays.

DSSCs with Closed-Ended Freestanding TiO2 Nanotube Arrays with Channels Containing Ag NPs and Large TiO2 NPs
The photocurrent-voltage curves of DSSCs fabricated using closed-ended freestanding TiO2 nanotube arrays measured under air mass 1.5 illumination (100 mW/cm 2 ) are shown in Figure 4 and Table 1. Four types of closed-ended freestanding TiO2 nanotube array-based DSSCs were fabricated in order to assess the effect of each component on the energy conversion efficiency: closed-ended freestanding TiO2 nanotube array-based DSSCs without Ag or large TiO2 NPs (a), with Ag NPs (b), with large TiO2 NPs (c), and with both Ag NPs and large TiO2 NPs (d). The open-circuit voltage (Voc), short-circuit current (Jsc), fill factor (ff), and energy conversion efficiency (η) values are shown in Table  1. The energy conversion efficiency of DSSCs based on closed-ended freestanding TiO2 nanotube arrays lacking NPs was 5.86%. The energy conversion efficiencies of DSSCs based on closed-ended freestanding TiO2 nanotube arrays with Ag NPs, with large TiO2 NPs, and with both Ag NPs and large TiO2 NPs were 6.40%, 6.24%, and 6.71%, respectively. The introduction of Ag NPs increased the energy conversion efficiency significantly, by 9.21% (actual efficiency change, from 5.86% to 6.40%) compared to closed-ended freestanding TiO2 nanotube array-based DSSCs without Ag and large TiO2 NPs, because of increased light harvesting by the plasmonic effect. The introduction of large TiO2 NPs also increased the energy conversion efficiency significantly, by 6.48% (actual efficiency, from 5.86% to 6.24%), owing to increased light harvesting by the scattering effect. Moreover, the introduction of both Ag NPs and large TiO2 NPs increased the energy conversion efficiency significantly, by 14.50% (actual efficiency, from 5.86% to 6.71%), because of increased light harvesting resulting from both the plasmonic and scattering effects.

DSSCs with Closed-Ended Freestanding TiO 2 Nanotube Arrays with Channels Containing Ag NPs and Large TiO 2 NPs
The photocurrent-voltage curves of DSSCs fabricated using closed-ended freestanding TiO 2 nanotube arrays measured under air mass 1.5 illumination (100 mW/cm 2 ) are shown in Figure 4 and Table 1.  Table 1. The energy conversion efficiency of DSSCs based on closed-ended freestanding TiO 2 nanotube arrays lacking NPs was 5.86%. The energy conversion efficiencies of DSSCs based on closed-ended freestanding TiO 2 nanotube arrays with Ag NPs, with large TiO 2 NPs, and with both Ag NPs and large TiO 2 NPs were 6.40%, 6.24%, and 6.71%, respectively. The introduction of Ag NPs increased the energy conversion efficiency significantly, by 9.21% (actual efficiency change, from 5.86% to 6.40%) compared to closed-ended freestanding TiO 2 nanotube array-based DSSCs without Ag and large TiO 2 NPs, because of increased light harvesting by the plasmonic effect. The introduction of large TiO 2 NPs also increased the energy conversion efficiency significantly, by 6.48% (actual efficiency, from 5.86% to 6.24%), owing to increased light harvesting by the scattering effect. Moreover, the introduction of both Ag NPs and large TiO 2 NPs increased the energy conversion efficiency significantly, by 14.50% (actual efficiency, from 5.86% to 6.71%), because of increased light harvesting resulting from both the plasmonic and scattering effects.

DSSCs with Closed-Ended Freestanding TiO2 Nanotube Arrays with Channels Containing Ag NPs and Large TiO2 NPs
The photocurrent-voltage curves of DSSCs fabricated using closed-ended freestanding TiO2 nanotube arrays measured under air mass 1.5 illumination (100 mW/cm 2 ) are shown in Figure 4 and Table 1.  Table  1. The energy conversion efficiency of DSSCs based on closed-ended freestanding TiO2 nanotube arrays lacking NPs was 5.86%. The energy conversion efficiencies of DSSCs based on closed-ended freestanding TiO2 nanotube arrays with Ag NPs, with large TiO2 NPs, and with both Ag NPs and large TiO2 NPs were 6.40%, 6.24%, and 6.71%, respectively. The introduction of Ag NPs increased the energy conversion efficiency significantly, by 9.21% (actual efficiency change, from 5.86% to 6.40%) compared to closed-ended freestanding TiO2 nanotube array-based DSSCs without Ag and large TiO2 NPs, because of increased light harvesting by the plasmonic effect. The introduction of large TiO2 NPs also increased the energy conversion efficiency significantly, by 6.48% (actual efficiency, from 5.86% to 6.24%), owing to increased light harvesting by the scattering effect. Moreover, the introduction of both Ag NPs and large TiO2 NPs increased the energy conversion efficiency significantly, by 14.50% (actual efficiency, from 5.86% to 6.71%), because of increased light harvesting resulting from both the plasmonic and scattering effects.   The photocurrent-voltage curves of DSSCs fabricated using open-ended freestanding TiO 2 nanotube arrays are shown in Figure 5 and Table 2 Table 2. The energy conversion efficiency of DSSCs based on open-ended freestanding TiO 2 nanotube arrays lacking NPs was 6.12%. The energy conversion efficiencies of DSSCs based on open-ended freestanding TiO 2 nanotube arrays with Ag NPs, with large TiO 2 NPs, and with both Ag NPs and large TiO 2 NPs were 6.68%, 6.62%, and 7.05%, respectively. The introduction of Ag NPs, large TiO 2 NPs, and both increased the energy conversion efficiency by 9.15%, 8.17%, and 15.20%, respectively. Compared to closed-ended freestanding TiO 2 nanotube arrays, the energy conversion efficiency of DSSCs based on open-ended freestanding TiO 2 nanotube arrays was 5.07% (6.71%-7.05%) higher due to enhanced electron transport and electrolyte diffusion [33].

DSSCs with Open-Ended Freestanding TiO2 Nanotube Arrays with Channels Containing Ag NPs and Large TiO2 NPs
The photocurrent-voltage curves of DSSCs fabricated using open-ended freestanding TiO2 nanotube arrays are shown in Figure 5 and Table 2 Table 2. The energy conversion efficiency of DSSCs based on open-ended freestanding TiO2 nanotube arrays lacking NPs was 6.12%. The energy conversion efficiencies of DSSCs based on open-ended freestanding TiO2 nanotube arrays with Ag NPs, with large TiO2 NPs, and with both Ag NPs and large TiO2 NPs were 6.68%, 6.62%, and 7.05%, respectively. The introduction of Ag NPs, large TiO2 NPs, and both increased the energy conversion efficiency by 9.15%, 8.17%, and 15.20%, respectively. Compared to closed-ended freestanding TiO2 nanotube arrays, the energy conversion efficiency of DSSCs based on open-ended freestanding TiO2 nanotube arrays was 5.07% (6.71%-7.05%) higher due to enhanced electron transport and electrolyte diffusion [33].
Although TiO2 nanotube array-based DSSCs have great potential, as far as we know, the theoretical maximum improvement by Ag NPs or a TiO2 scattering layer of TiO2 nanotube-based DSSCs has not yet been reported. However, the open-ended TiO2 nanotube-based devices exhibited an increase in one-sun efficiency from 5.3% to 9.1%, corresponding to a 70% increase, which is a much higher increase than we achieved [35]. We believe that there is ample room to improve efficiency by combining each parameter in an optimal condition based on theoretical studies.   Although TiO 2 nanotube array-based DSSCs have great potential, as far as we know, the theoretical maximum improvement by Ag NPs or a TiO 2 scattering layer of TiO 2 nanotube-based DSSCs has not yet been reported. However, the open-ended TiO 2 nanotube-based devices exhibited an increase in one-sun efficiency from 5.3% to 9.1%, corresponding to a 70% increase, which is a much higher increase than we achieved [35]. We believe that there is ample room to improve efficiency by combining each parameter in an optimal condition based on theoretical studies.

Preparation of Freestanding TiO 2 Nanotube Arrays
TiO 2 nanotube arrays were prepared by anodization of thin Ti plates. Ti anodization was carried out in an electrolyte composed of 0.8 wt % NH 4 F and 2 vol % H 2 O in ethylene glycol at 25˝C and at a constant voltage of 60 V direct current (DC) for 2 h. After being anodized, the Ti plates were annealed at 500˝C for 1 h under ambient conditions to improve the crystallinity of the TiO 2 nanotube arrays, and then a secondary anodization was conducted at a constant voltage of 30 V DC for 10 min. The Ti plates were immersed in 10% H 2 O 2 solution for several hours in order to detach the TiO 2 nanotube arrays from the Ti plates and produce freestanding TiO 2 nanotube arrays. The bottom of the TiO 2 nanotube arrays was removed by ion milling with Ar + bombardment for several minutes in order to prepare open-ended freestanding TiO 2 nanotube arrays.

Preparation of Ag NPs in the Channel of Freestanding TiO 2 Nanotube Arrays
A TiO 2 blocking layer was formed on FTO glass by spin-coating with 5 wt % titanium di-isopropoxide bis(acetylacetonate) in butanol, followed by annealing at 500˝C for 1 h under ambient conditions to induce crystallinity. A TiO 2 paste was coated onto the TiO 2 blocking/FTO glass using the doctor blade method, and closed-and open-ended freestanding TiO 2 nanotube arrays were then introduced onto the TiO 2 paste. The substrate was annealed at 500˝C for 1 h under ambient conditions to enhance the adhesion between the TiO 2 NPs and freestanding TiO 2 nanotube arrays. The substrate was dipped in 0.3 mM AgNO 3 aqueous solution and exposed to 254 nm UV irradiation. The substrate was treated with 10 mM TiCl 4 solution at 50˝C for 30 min and then annealed at 500˝C for 1 h.

Fabrication of DSSCs with Freestanding TiO 2 Nanotube Arrays with Channels Containing Ag NPs
A substrate that consisted of freestanding TiO 2 nanotube arrays with channels containing Ag NPs was coated with~400 nm TiO 2 NPs for scattering and then annealed at 500˝C for 1 h under ambient conditions to induce crystallinity and adhesion. The substrate was immersed in a dye solution at 50˝C for 8 h to function as a working electrode. The working electrode was sandwiched with a counter electrode, Pt on FTO glass, by a 60-µm-thick hot-melt sheet and filled with electrolyte solution composed of 0.7 M 1-butyl-3-methyl-imidazolium iodide (BMII), 0.03 M I 2 , 0.1 M guanidium thiocyanate (GSCN), and 0.5 M 4-tertbutylpyridine (TBP) in a mixture of acetonitrile and valeronitrile (85:15, v/v).

Instruments
The morphology, thickness, size, and presence of Ag NPs in the channels of freestanding TiO 2 nanotube arrays were confirmed using a field emission scanning electron microscope (FE-SEM, JSM-6330F, JEOL Inc., Tokyo, Japan) and the high angular annular dark field (HAADF) technique with a scanning transmission electron microscope (TEM, JEM-2200FS, JEOL Inc., Tokyo, Japan). The current density´voltage (J´V) characteristics and the incident photon-to-current conversion efficiency (IPCE) of the DSSCs were measured using an electrometer (Keithley 2400, Keithley Instruments, Inc., Cleveland, OH, USA) under AM 1.5 illumination (100 mW/cm 2 ) provided by a solar simulator (1 kW xenon with AM 1.5 filter, PEC-L01, Peccell Technologies, Inc., Yokohama, Kanagawa, Japan), or using a solar cell IPCE measurement System(K3100, McScience Inc., Suwon, Korea) with reference to the calibrated diode.

Conclusions
In this study, we compared the natural consequences of altering three parameters, the plasmonic effect, the scattering effect, and open-vs. closed-ended freestanding TiO 2 nanotubes, as a basic means of exploring improvements in efficiency. We demonstrated that the plasmonic and scattering effects enhanced the energy conversion efficiency of freestanding TiO 2 nanotube arrays in DSSCs. Ag NPs were added to the channels of TiO 2 nanotube arrays by UV irradiation to induce a plasmonic effect, and large TiO 2 NPs were introduced to TiO 2 nanotube arrays to induce a scattering effect. The energy conversion efficiency of DSSCs with both Ag NPs and large TiO 2 NPs was higher than that of DSSCs without Ag NPs owing to the plasmonic effect, and it was higher than that of DSSCs without large TiO 2 NPs owing to the scattering effect. Compared to closed-ended freestanding TiO 2 nanotube arrays, open-ended freestanding TiO 2 nanotube arrays [40] exhibited enhanced energy conversion efficiency. We demonstrate that Ag NPs, TiO 2 NPs, and open-ended freestanding TiO 2 nanotube arrays enhanced the energy conversion efficiency; furthermore, the combination of all components exhibited the highest energy conversion efficiency. Our research suggests that the energy conversion efficiency of DSSCs is improved by both the plasmonic and scattering effects. This knowledge has applications in organic solar cells, hybrid solar cells, and perovskite solar cells.