III. RESULTS AND DISCUSSION

Figures 1(a) and 1(b) show the XRD patterns as-synthesized of NiS using absolute ethyl alcohol and distilled water as solvents, respectively. The space group of the as-prepared product of Fig. 1(a) is P63/mmc and the cell parameters (a = b = 3.420 Å, c = 5.300 Å) are also in good agreement with the literature data (Card No. 750613). While the space of the as-prepared product of Fig. 1(b) is R _3m and the cell parameters (a = b = 9.619 Å, c = 3.149 Å) are in good agreement with the literature data (Card No. 862281). No impurity phase is detected for the two patterns. The former is α-NiS and the latter is β-NiS which has a rhombohedral structure of the millerite phase.^{Reference Ghezelbash, Sigman and Korgel17}

FIG. 1.

XRD patterns of NiS synthesized with different solvents: (a) absolute ethyl alcohol and (b) distilled water.

Figures 2(a) and 2(b) depict the morphologies of α-NiS synthesized with absolute ethyl alcohol as solvent. They have a sphere-like shape with the diameter about 50–80 nm. The SEM images of β-NiS synthesized with distilled water present cross-like shapes, see Figs. 2(c)–2(e). It can be seen that cross-like β-NiS architectures are composed of nanorods with a shape of hexagonal prism. The nanorods are 2–5 μm in length and 1000 nm in diameter. Microstructural analysis for β-NiS was carried out by TEM, shown in Fig. 3. It can be seen that the nanorods are well single-crystallized. Unfortunately, TEM measurements could not be well performed on α-NiS, since it is easy to melt when electron beam focused on it.

FIG. 2.

SEM images of NiS synthesized with different solvents: (a) and (b) absolute ethyl alcohol as solvent; (c), (d), and (e) distilled water as solvent.

FIG. 3.

TEM (a), HRTEM (b), and electron diffraction (c) images of NiS synthesized with distilled water.

The saturated vapor pressure of alcohol (2958.72 kPa at 200 °C) is much higher than that of distilled water (1554.77 kPa at 200 °C). When alcohol and distilled water are heated, the pressure in the absolute ethyl alcohol rises faster and this has great impact on the morphology of the graininess samples. The cross-like samples synthesized with distilled water could be attributed to Ostwald ripening process.^{Reference Roosen and Carter21} In another way, many nanoparticles with different sizes appear in the solution at the early stage of reaction. With the reaction proceeding continuously, assembled nanoparticles with a relatively larger size may serve as the crystal nucleus. Then the nanoparticles with a smaller size would dissolve and recrystallize at the site of the larger nanoparticles because nanoparticles with smaller size own higher surface free energy.^{Reference Li, Chai, Wang, Xi, Liu and Qian22} Nickel sulfide crystals grow preferentially in a certain direction and then cross-like β-NiS comes into being. Note that there is a good repeatability for the same routine experiments. A controllable synthesis for two phases by selecting different solvents is realized.

The optical properties of the as-obtained α-NiS and β-NiS samples are characterized by the UV-vis spectra.^{Reference Du, Yin, Zhu, Huang, Wu, Zeng, Yan and Zhang23}Figure 4(a) shows the absorption spectrum of α-NiS particles. A peak located at about 254 nm and many weak shoulders located around 270 nm, 288 nm, and 308 nm are clearly seen. The spectra in Fig. 4(b) show a strong absorption of β-NiS particles in the ultraviolet region. There is an absorption peak at around 269 nm. The spectrum absorption is shifted to much higher energy (∼4.8 eV) compared to that of bulk NiS (∼2.1 eV).^{Reference Hu, Chen, Chen, Lin and Li24} Similar phenomena were also observed by Hu et al.^{Reference Nakamura, Fujimori, Sacchi, Fuggle, Misu, mamori, Tamura, Matoba and Anzai25} This remarkable blue-shift (about 2.7 eV) can probably be attributed to the small dimensions or well-ordered array of the NiS walls.^{Reference Hu, Chen, Chen, Lin and Li24} Bulk NiS is an inorganic semiconductor with a direct band gap. The band gaps for the two phases of NiS can be estimated by a plot of (αhν)² versus hν (α = absorbance, h = Planck's constant, and ν = frequency),^{Reference Nyari, Barvinschi, Bǎies, Vlǎzan, Barvinschi and Dekany26} shown as insets in Fig. 4. The band gap of α-NiS and β-NiS is about 0.8 eV and 1.0 eV, respectively, bigger than that of nonmetallic phase of NiS (0.2–0.3 eV).^{Reference Okamura, Naitoh, Nanba, Matoba, Nishioka, Anzai, Shimoyama, Fukui, Miura, Nakagawa, Nakagawa and Kinoshita27} Certainly, the reason may be that the band gap of the NiS is also relevant to its size and microstructure, so called blue-shift. Usually, with the dimension reduction as well as quantum size effect, blue shift appears. Although NiS nanoparticles were once reported to have a band gap of 3.53 eV measured via UV-visible spectroscopy,^{Reference Sohrabnezhad, Pourahmad, Sadjadi and Zanjanchi13} it is still black at room temperature and has a low metal/insulator transition (252 K) temperature^{Reference Larsson14} and the high temperature phase of NiS is a metal.^{Reference Larsson14} Taking our results and Hu's^{Reference Nakamura, Fujimori, Sacchi, Fuggle, Misu, mamori, Tamura, Matoba and Anzai25} into consideration as well as Okamura's report,^{Reference Okamura, Naitoh, Nanba, Matoba, Nishioka, Anzai, Shimoyama, Fukui, Miura, Nakagawa, Nakagawa and Kinoshita27} it is reasonable to infer that the band gap reported by Sohrabnezhad^{Reference Sohrabnezhad, Pourahmad, Sadjadi and Zanjanchi13} for NiS nanoparticles is not the band gap of NiS.

FIG. 4.

UV-vis absorption spectra and the band gap value which was estimated by a related curve of (αhν)² versus photon energy plotted of NiS synthesized with different solvents: (a) absolute ethyl alcohol and (b) distilled water.

DSSCs were fabricated using NiS with two different phases as CEs mentioned above and Pt as a CE for comparison. Photocurrent density (J) versus photovoltage (V) and energy conversion efficiency curves of different morphologies of NiS CEs and Pt CE in acetonitrile solutions containing 10 mM LiI, 10 mM I₂, or 0.1 M LiClO₄ as the supporting electrolyte are shown in Fig. 5. The short-circuit current densities (J _sc) are 11.4 mA cm⁻² and 9.8 mA cm⁻² for the two kinds of DSSCs with NiS as CEs, respectively. But the open circuit voltage (V _oc) is 0.68 V for both of them. Figure 5 also depicts that DSSCs using α-NiS as CE yields a conversion efficiency (η) of 5.2%, which is much better than that of β-NiS (η = 4.2%) with an increase of one quarter. For a comparison, the DDSC with Pt as CE was also fabricated and measured. J _sc = 14.7 mA cm⁻², which is larger than those of DSSCs with NiS as CEs. But the open circuit voltage for the DDSC with Pt as CE is 0.63 V, which is little smaller than those of DSSCs with NiS as CEs. However, the conversion efficiency η = 6.3% for the DSSC with Pt as CE, which is larger than those of DSSCs with NiS as CEs. After a simple calculation, the filling factors (FF) for the three DSSCs are 67.0% (α-NiS), 63.0% (β-NiS), and 65.1% (Pt), respectively. The filling factor may reflect the output characteristics of solar cells, which is an important parameter. There are many factors influencing the value of FF. Among them, the series resistance is dominated. The rise of the series resistance reduces FF and, consequently, so does the efficiency of the solar cell.

FIG. 5.

J–V curve (a) and P–V curve (b) of NiS synthesized with different solvents.

The IPCE spectra of devices with different CEs are shown in Fig. 6. The IPCE curve (λ) could be defined as

(1)$${\rm{IPCE}}\left( {\rm{\lambda }} \right) = 1240{{{J_{{\rm{sc}}}}} \over {{\rm{\lambda }}{P_{{\rm{in}}}}}}{\rm{,}}$$

where J _sc is the short circuit current density (mA cm⁻²), λ is the wave length, and P _in is the incident light power density (mW cm⁻²). The device with α-NiS exhibited a little higher IPCE (71.16%) compared to the β-NiS DSSC (65.90%). In addition, the IPCE plotted as a function of excitation wave length of Pt-coated FTO cell reaches a value of 77.85%. Moreover, the photocatalytic performances of different CEs were consistent with the value estimated from IPCE curve on the whole.^{Reference Cho, Hwang, Kim and Jang28–Reference Grätzel30}

FIG. 7.

Nyquist plots of EIS for the symmetrical cells and the equivalent circuits fabricated with (a) α-NiS and (b) β-NiS film, respectively, and sputtering Pt for comparison.

To analyze the behavior mentioned above, the EIS was carried out under I⁻/I₃⁻ electrolyte to analyze the internal resistance of α-NiS, β-NiS, and Pt in the DSSCs. Figure 7 shows the Nyquist plots for the symmetrical cells fabricated with three identical CEs (α-NiS, β-NiS, and Pt). Also, we propose an equivalent circuit for modeling DSSCs based on the results of EIS analysis used to fit the experimental EIS data. The charge transfer resistances R _ct of α-NiS, β-NiS, and Pt are 3.1 Ω, 6.148 Ω, and 1.734 Ω, respectively. The comparable charge transfer resistances of α-NiS and β-NiS to Pt play a major role in the observed conversion efficiencies. As is well known that the catalytic activity of the electrodes can be reflected from the R _ct in the electrode/electrolyte interface, which is directly proportional to the diameter of the semicircle in the high-frequency regions.^{Reference Jiang, Li and Gao31,Reference Nazeeruddin, Kay, Rodicio, Humphry-Baker, Muller, Liska, Vlachopoulos and Grätzel32} The high-frequency intercept represents the series resistance of DSSCs (R _s), which is mainly composed of three resistance elements, namely, the sheet resistance of the FTO, the resistance of ionic diffusion in the electrolyte, and the resistance at the interface between the CE and the electrolyte. The R _s of α-NiS and β-NiS CEs are 12.04 Ω and 12.77 Ω, bigger than that of Pt CEs (6.767 Ω), which means that electrical conductivity is not as good as Pt CE. The detailed photovoltaic parameters for three DSSCs are listed in Table I.

FIG. 6.

IPCE spectra of DSSCs with NiS CEs and Pt-plated FTO CE.

TABLE I.

EIS parameters of the dummy cell and properties of the DSSCs assembled with two identical CEs of α-NiS and β-NiS.

There are two reasons for DSSCs prepared with α-NiS as CE own higher conversion efficiency than β-NiS as CE. One is that the charge transfer resistance R _ct of α-NiS is smaller than that of β-NiS, the other is that the particle size of α-NiS is much smaller than nanorods of β-NiS. The smaller the size of particles, the larger the specific surface area is. With the increase of specific surface area of α-NiS, the conversion efficiency reaches a higher value. On the other hand, although nickel sulfides show a higher redox activity,^{Reference Zhang, Zhang, Cui, Feng and Zhu33} all of the DSSCs with them as CEs show lower conversion efficiencies than that with a Pt plate as CE (current value, η = 6.3%). It is known to all, the Pt plate itself also acts as a light reflector and increases the operation rate of light and could, thus, remarkably improve the efficiency of the solar cells.^{Reference Nazeeruddin, Kay, Rodicio, Humphry-Baker, Muller, Liska, Vlachopoulos and Grätzel32} However, if taking the convenience and low cost of DSSCs with NiS as CEs into consideration, the current reported DSSC with α-NiS as CE paves one route for designing and applying low-cost Pt-free CE materials in DSSC from inorganic nanostructures.