CONTROLLABLE CORE SIZE OF Au@TiO 2 THROUGH Al(NO 3 ) 3 ADDITION AND ITS EFFECTS ON DSSC PERFORMANCE

It is known that plasmonic nanoparticles in dye-sensitized solar cells (DSSC) could enhance eﬃciency through improvement in light absorbance and electron dynamics. Herein we investigated various sizes of AuNP through spontaneous Al(NO 3 ) 3 addition. Core-Shell Au@TiO 2 was prepared with various Al(NO 3 ) 3 concentrations of 0.25 mM, 0.5 mM, 0.75 mM and 1 mM. The Au@TiO 2 volume fraction of 1% was further added to TiO 2 photoanode. Based on the particle size analyzer (PSA) char-acteristics, the synthesized AuNP’s size was within a range of 34.62 nm – 139.5 nm. The highest eﬃciency of DSSC was obtained for the sample with the largest AuNP ’s diameter, i.e., 0.0313%, which is about three times higher than pristine DSSC. The increase in eﬃciency was in accord with Metallic Nanoparticle Boundary Element Method (MNPBEM) simulation, UV-vis spectroscopy, and Incident Photon to Current Conversion Eﬃciency (IPCE) analysis largest Au core diameter contributes to the strong absorbance and hence the short circuit current.


INTRODUCTION
Several ways can be done to improve the efficiency of Dye-Sensitized Solar Cells (DSSC) by adding metal in a core-shell structure into the photoanode. It is known that by incorporating core-shell of metal encapsulated by oxide, the localized surface plasmons resonance (LSPR) becomes feasible and increases the absorption of the dye molecules. Hence improvement in electron collection and device performance can be obtained. There are metals commonly used for promoting LSPR in the visible region, i.e., gold (Au) [1,2] , silver (Ag) [3] , copper (Cu) [4] , and bismuth (Bi) [5] , among which Au is the most excellent metal exhibiting strong LSPR effect. The way LSPR works in DSSC is as follows, the metallic core concentrates light. At the same time, the shell ought to provide optimal spacing, chemical stability, good electrical conductivity, and good photovoltaic properties.
Therefore, the morphology of the core, the material used, and the thickness of the shell need to be carefully designed and controlled to obtain optimal enhancement [6,7] . Typically, the enhancement effect of metal@dielectric core-shell nanoparticles is attributed to the enhanced excitation rate of dye sensitizers due to LSPR. Even though the metal nanoparticles can be protected and hindered from charge recombination, one may expect two beneficial mechanisms involved in the efficiency enhancement for using TiO 2 semiconductor as the shell. First, the hot electrons generated inside the resonant plasmonic cores can contribute directly to the photocurrent. Second, the metal cores can undergo charge equilibrium with the surrounding semiconductor and modify the Fermi level of TiO 2 , resulting in an improved cell potential [1] .
Extensive studies focusing on the prevention of recombination and improving charge transfer efficiency through incorporation of metals such as copper [8,9] , gold [10] , silver [11,12] , palladium [13] , or platinum [14] in DSSC photoanode have been reported. The noble metals offer advantages due to their resonance in the visible region and stability, where Cu is unstable against surface oxides, while Pd and Pt show very weak resonance [15] . Au and Ag have quite strong resonance among these metals due to the enhanced extinction cross-section and the tunability of plasmon resonance by the surrounding. In comparison to Ag, Au possesses many benefits, particularly it has high stability either chemically or biologically, high sensitivity by the high surfaceto-volume ratio, easy probe, and many more [16] .
Intending to modify the core-shell morphology, this research focused on changing the core size without changing the amount of Au loading at a constant shell thickness. For this purpose, Al(NO 3 ) 3 was employed since it has been known that it can be used as a template for AuNP growth, either being added at a constant concentration [17] or being varied in proportional to the amount of AuNP loading [18] . The latter has shown that the growth of AuNP can be controlled by manipulating the amount of reducing agent of HAuCl 4 . As the function of the reducing agent is limited, the mechanism is then controlled by the van der Waals attractive force leading to the integration of small AuNPs into aggregated AuNPs [19] . With these structures, we systematically studied the influence of Au core size on the plasmonic enhancement effect in DSSCs. Additionally, we report the effect of metallic core size using a simulation of metallic nanoparticles embedded in a dielectric medium using the boundary element method (BEM) approach. Finally, solar cell performance is investigated and analyzed concerning the controlled nanoscale light-matter interaction.

PREVIOUS RESEARCHES
Many studies integrate AuNP coated with a thin shell either of silica or TiO 2 into the TiO 2 mesoporous paste. The intense absorption of AuNP due to its surface plasmon resonance depends on its size and shape [20,21] . The control of size was mostly adjusted by using a sophisticated method [2] or simply by varying the volume fraction of Au and TiO 2 loading [22] . The use of Al(NO 3 ) 3 to control the growth of Au was firstly reported by [18] and explained thoroughly by [17] . By employing sacrificial hydrothermal reaction, the following reactions occur [23,24] : According to the free Gibbs energy formation (Δ ) of the above reactions, the formation of AuNP is easier due to its more negative energy. The AuNP will be formed heterogeneously on the grain boundary sites of Al. As reported in [17] , the size of AuNP was adjusted by increasing the concentration of HAuCl 4 . In this research, the number of heterogeneous nucleation sites was varied. Hence, the further growth and coarsening of the nuclei will be affected, too, through the Ostwald ripening phenomenon.

Synthesis of Au@TiO 2 Core-Shell Photoanodes
TiO 2 photoanode powder was fabricated using the coprecipitation method and calcined at 600oC for 5 hours to get the anatase phase. AuNP with various diameters size was synthesized through the addition of different concentrations of Al(NO 3 ), i.e. 0.25 mM, 0,5 mM, 0,75 mM and 1 mM. Au@TiO 2 core-shell powder was obtained by mixing the AuNP with Ti4+ solution through a combination of TTIP (Titanium Isopropoxide) and TEOA (Triethanolamine) in a nitrogen atmosphere at a molarity ratio of 2:1 (TEOA: TTIP). The washing process of Au@TiO 2 took two weeks using equates. Subsequently, the powder of Au@TiO 2 photoanodes was calcined for 48 hours at 80o C.

Characterization of Synthesized Au@TiO 2 Core-Shell Photoanodes
The particle size distribution of AuNP was carried out using a particle size analyzer (PSA) (Malvern Zetasizer). The crystal structure of photoanodes powder was characterized using X-ray diffraction (XRD) (Philips XPert MPD 30 mA, 40 kV) with Cu Kα radiation ( = 0.15406 nm) with 2 scanned within the range of 15°-90°. Absorption of photoanode powder in visible light (200 nm -800 nm) was characterized using UV-Vis spectrophotometer Perkin-Elmer Lambda 750 UV / Vis / NIR. Raman spectra were obtained using a Renishaw inVia Raman Microscope (Wotton-under-Edge, UK) operating with NIR diode laser emitting at 785 nm. Spectra were recorded with an accumulation of one scan, 10 s exposure. To find out the morphology and core-shell structure, the sample was characterized using a field-emission scanning electron microscope (FE-SEM) JEOL JIB 4610F. Plasmonics simulation was undertaken using MATLAB equipped with MNPBEM (Metallic Nanoparticles Boundary Element Method) toolbox [25] . The toolbox works in principle for homogeneous dielectric bodies inside homogeneous dielectric surroundings, in which abrupt interfaces separate the bodies. In this paper, we simulated various Au diameters inside air surroundings. Photocurrent -voltage curves were measured under 100 ∕ 2 using a solar simulator connected to PXI-E1073 and SMU PXI-4130 and LabView Software. The energy conversion efficiency ( ) was calculated using Eq. 3.
Jsc is short circuit current density, Voc is open-circuit voltage, FF is filling factor, and Pin is the power input. Jmax is the current density at maximum power output, and Vmax is the voltage at maximum power output. The fill factor can be obtained by the Eq. 4.
Incident photon to current conversion efficiency (IPCE) in the visible light spectrum was obtained by adjusting the wavelength of the incident light, i.e., a series connection of halogen lamp (GR-150 Halogen Flood Light 150W) and monochromator (CT-10T, JASCO). The value of incident light power was measured by using an optical power meter (Thorlab S-120C).

Fabrication of DSSC
Photoanodes powder weighed 0.25 grams was made into pasta by dissolution in 87.5 µL in distilled water, 125 µL of CH3COOH, and 12.5 µL of Triton X-100. The coating of photoanode paste into FTO glass was undertaken by using the doctor blade method. Subsequently, the coated FTO glass was heated at 225oC for 2 minutes using a hot plate. The coated FTO glass was then soaked in a ruthenium complex N-719 dye solution for 12 hours. Finally, DSSC components with sandwich structures were made by arranging FTO glass coated with dyed-photoanode, an electrolyte solution, and a platinum-plated FTO glass as a counter electrode.
The sample used in this research is a concrete cylinder with a diameter of 100 x 200 mm. The materials used in this research are normal concrete which is made with ordinary portland cement, silica fume, and fly ash, with Console SS-8 as High Range Water Reducer (HRWR). The typical chemical composition of ordinary portland cement powder, silica fume, and fly ash is given in Tables 1 and Table 2 .   . These results are related to the formation mechanism of the core-shell Au@TiO 2 , i.e., the Au core predominantly grew in the (111) crystal planes. However, the peak intensity remains as the amount of Al(NO 3 ) 3 increases, while the TiO 2 grew profoundly in the (101) crystal planes, and it seems any changes in Au core do not alter the TiO 2 peaks.

RESULTS AND DISCUSSION
Raman spectroscopy within the range of 100-1000 cm-1 was undertaken on pure TiO 2 and Au@TiO 2 with one mM Al(NO 3 ) 3 samples to evaluate the phases present in the prepared samples, and the results are shown in Fig 2 . Typical Raman peaks associated with anatase phase TiO 2 are detected [12] . Most of the peak intensities decreased and were shifted to a higher wavenumber   as in Au@TiO 2 . For instance, a band at about 144 cm-1 that corresponds to the external vibration of the − bond shifts towards a higher wavenumber. This indicates an interaction between the Au and TiO 2 , leading to an increase in the − bond force constant due to reducing the − bond length [26,27] . Raman spectra of Au@TiO 2 may indicate that the Au nanoparticle is in the form of Au@TiO 2 , and there is no such phase transformation, rather than an electronic environment change in the surroundings [28] .
PSA characterization was undertaken for AuNPs only and Au@TiO 2 core-shell. The measurement of AuNP can be done directly because it was in the form of a liquid colloid. The samples had to be dissolved in 10 ml aquabidest per 5 mg powder for other  Table 1 . It is seen that the addition of Al(NO 3 ) 3 results in the further growth of AuNP . All samples studied typically show two distinct peaks of core-shell diameter due to the faster reduction process of , leading to the formation of a heterogeneous population of AuNP [29] . Similar results were obtained from FE-SEM images (Fig 3 ).
According to UV-Vis characterization (Fig 4 ), all samples studied show weak surface plasmon resonance response ( 520 to 600 nm) as indicative of the thick TiO 2 shells for individual core-shell Au@TiO 2 [30] . All samples reveal a strong peak belongs to TiO 2 at 320 nm, which remains unchanged with the addition of Al(NO 3 ) 3 . Except for sample with one mM Al(NO 3 ) 3 shows three sluggish peaks 420 nm, 630 nm, and 700 nm, attributed to a strong interaction between Au and TiO 2 , resonance absorption, and interband electronic transition, respectively [31] . Similar to Yao et al. [31] , Au cores having a diameter less than 100 nm do not show any significant variation of optical absorption properties induced by Au cores. This may imply that small Au cores do not respond to the incident light.
UV-Vis characterization can also indicate the quality of the core-shell structure, whether the shell may or not cover the whole core. Since the absorption produced by the TiO 2 maintains the constant characteristics, this may suggest that the optical properties of the TiO 2 shell are not affected by the size of Au cores. The diminution of absorption intensity led by the electronic transition between Au cores and the TiO 2 shells for a sample having a core diameter less than 100 nm may indicate that the interaction between Au cores and TiO 2 shells is weakened as the shells fully cover the cores. The redshift of resonance absorption peak observed for samples with core diameter >100 nm may indicate that The TiO 2 shells coat Au cores. This redshift arises because TiO 2 , as the dielectric environment of Au has a higher refractive index, i.e., 2.5. As the TiO 2 shells coat Au cores, the absorption induced by the interband transition may split into several distinct peaks, as summarized in [30] .
It is known that incorporating Au nanoparticles into the TiO 2 can reduce the bandgap energy of the TiO 2 [12] . The reduction is known due to the free electron properties exhibiting a downward shift in the conduction band and an upward shift in the valence band [32,33] . The optical bandgap of the Au@TiO 2 core-shell nanoparticles ( Figure 5 ) was in the range of 2.49-2.75 eV, which was substantially lower than that of TiO 2 ( 3.20 eV) [34] and was likely attributed to defect-induced narrow band gap [35][36][37] . A pronounced drop in bandgap energy of sample with the largest Au core diameter can be elucidated through Raman spectroscopy results shown in Figure 2 .
Incident photon-to-current conversion efficiency measurement was performed to investigate the spectral response of plasmonenhanced DSSC (Figure 6 ). The dashed lines indicate the absorption spectra of N719 dye molecules. In general, the spectra consist of an additional peak located at 420-470 nm (marked with *). This peak is likely related to modifying electronic states Au-TiO 2 due to a hetero-junction-induced charge transfer interaction [30] . The presence of this peak is more obvious for the samples with core diameter >100 nm arising from the enhanced near-field effects of AuNPs. Figure 7 shows the scattering spectra from the MNPBEM simulation. All spectra have a relatively broad resonance around 520 nm, caused by the collective plasma oscillations of the free electron gas. Figure 7 (inset) also gives a closer look at how the cross-section depends on the size of the particle. It is clear that the larger the Au diameter, the larger the cross-section obtained. Accordingly, the surface charges and real part of the induced electric field became larger, as seen in Fig 8 . It is noted that the current simulation used air as the surroundings of Au, which has refractive indices of 1. By changing the surroundings with TiO 2 , one may obtain a redshift of the marked peak at 520 nm to 550 nm [38] .
The oscillating surface charges strongly amplify the electric field of incident light in metal nanoparticles. Light absorption increases in proportion to the square of the electric field amplitude. In this research, Au nanoparticles in DSSC improved absorption of photoelectrode and generation of photocurrent, Au nanoparticles themselves are known as strong visible light absorbance. Figure 9 illustrates the behavior of the J-V curves of the studied samples. The individual results are tabulated in Table 1 . The DSSC containing no Au@TiO 2 core-shell has the lowest Jsc and Voc. Hence its efficiency is also the lowest. For DSSC with Au@TiO 2 inclusion, the Jsc increases as the Au core diameter increases, whereas the Voc is sluggishly changed with the Au core diameter. Such an observation reveals that the mechanism responsible for the efficiency enhancement of Au@TiO 2 addition   Zhang et al. [39] The addition of Au@TiO 2 on TiO 2 induces an interfacial charge transfer process. Also, it behaves as a scattering element for plasmonic scattering to trap the light and near-field coupled with dye molecules [39] , thereby increasing the Jsc value. This agrees with the MNPBEM simulation, which shows that the larger the Au core size, the higher the scattering cross-section ( Figure 7 ).
It is clear from Table 2 that Voc of Au@TiO 2 modified photoanodes is varied along with the change in Au core size. The increase in Voc has been attributed to the photo-charging effect due to the stored electrons in the Au core, leading to upward shifting of the TiO 2 Fermi level and increasing the Voc. An earlier study on the effect of Au size on TiO 2 photoanode showed that the smaller the metal core, the greater the Voc obtained [1,40] . However, the change in Voc is not as significant as that of the Jsc, which could be due to the interplay between boosting the charge transfer-plasmonic scattering and modification of the Fermi level.

CONCLUSION
The diameter of the Au core in Au@TiO 2 core-shell can be controlled by varying (NO3)3 amounts. Cell efficiency was improved with the Au@TiO 2 core-shell, and the increase is in line with the increase in Au core size due to the change in the scattering cross-section. The efficiency was enhanced by about twice compared to that without Au nanoparticles. The enhanced performance was largely attributed to the charge transfer process and also behaved as a scattering element for the plasmonic scattering of the Au@TiO 2 core-shell nanoparticles. At the same time, the change in Fermi level due to electrons' storage capability in the Au core was minor due to the presence of defects induced by − bonds.