Selective adsorption of anionic dyes from aqueous solution by nickel (II) oxide

Dye wastewater is continuously generated by industries such as textile, papermaking, carpets, rubbers, paint, printing, pharmaceutical, food, cosmetic industries and so on. It is estimated that approximately 7 × 10⁵ tons of dyestuff are produced per year (Rafatullah et al. 2010), and 10–15% of the dyes are discharged as waste during the operations (Zhu et al. 2011), which results in water pollution. Additionally, the widespread dyes in water are not only toxic to aquatic organisms, but are also carcinogenic and cause mutagenicity to humans (Ma et al. 2014).

For the sake of removing organic dyes from aqueous solutions, various techniques have been used such as photocatalysis, biological processes, ultrafiltration, electrochemical degradation, advanced oxidation processes and so on (Kıranşan et al. 2014; Liu et al. 2016; Wei et al. 2017). Among the various physico-chemistry processes, the adsorption process is favorable in dye water pollution control because of its low cost, insensitivity to toxic substances, simplicity and ease of handling (Gürses et al. 2006; Gupta et al. 2012; Karaca et al. 2013). Traditional porous adsorbents, including zeolites (Wang & Peng 2010), activated carbon (Riverautrilla et al. 2011), natural clays (Srinivasan 2011), and polymer-based porous materials (Pourfarzolla 2014) are commonly used to treat water pollution. However, these traditional porous materials have drawbacks, such as low adsorption capacity, slow adsorption kinetics, and low selectivity (Zheng et al. 2017). In order to overcome these drawbacks, the new materials are still desirable (Zhang et al. 2017).

Lately, due to their excellent performance as advanced materials, metal oxide has been widely studied for dye removal (Li et al. 2015), but the adsorption capacities are limited. Farrokhi et al. (2014) reported the adsorption of reactive black 5 by ZnO on magnetite nanoparticles with a capacity of 22.1 mg/g. Singh & Bahadur (2013) studied Fe₃O₄, ZnO and TiO₂ for the removal of various dyes. However, the adsorption capacities are still limited. The studies on highly effective metal oxide materials are still scarce.

NiO is a significant and promising transition metal oxide which has the advantages of high thermal and chemical stability, low cost and environmental compatibility (Zheng et al. 2017). Very recently, NiO has been employed in water treatment and is synthetized by various chemical and physicochemical technologies (Hu et al. 2017). It has been widely used in many fields such as gas sensors (Tian et al. 2016), supercapacitors (Cheng et al. 2015), catalysis (Ye et al. 2016) and lithium-ion batteries (Jadhav et al. 2016), and NiO particles have positive surface charge at circumneutral pH (pH = 6–8) which makes it suitable to adsorb anionic dyes from aqueous solution (Zheng et al. 2017). Hence, NiO also shows excellent performance of adsorption for anionic dyes, such as AO7, ID and CR. It is imperative to design highly efficient NiO adsorbents for dye removal from water. Herein, we report a facile synthesis of NiO through a low-cost calcination treatment approach, which makes the adsorption process for barely removing anionic dyes with efficiency, simple design, wide adaptability, and easy operation become promising.

Additionally, the as-obtained NiO was utilized to adsorb six kinds of organic dyes and the adsorption mechanism was investigated. The X-ray diffraction, scanning electron microscopy, Brunauer-Emmett-Teller, X-ray photoelectron spectroscopy, and Fourier transform infrared spectroscopy analysis were used to study the adsorption mechanism. The kinetic data were fitted using the pseudo-first-order, pseudo-second-order, and intra-particle diffusion models. The effects of pH and temperature on adsorption as well as the thermodynamic and adsorption equilibrium were also studied.

All reagents and chemicals were of analytical reagent grade and were used as received without further purification. Nickel nitrate hexahydrate (Ni(NO)₂·6H₂O, ≥98.5%), oxalic acid (H₂C₂O₄·2H₂O, OxA, ≥99.5%), hydrochloric acid (HCl, 36.0–38.0%), sodium hydroxide, AO7, ID, crystal violet, CR, methylene blue, rhodamine B from Chengdu Kelong (Chengdu, China) chemical reagent factory were of analytic purity.

Nickel oxalate adsorbents were prepared by mixing 0.1 mol/L nickel nitrate with 0.1 mol/L OxA in a beaker. The synthesized nickel oxalate adsorbents were dried in the oven at 80 °C, and ground into powder by agate mortar. The prepared materials were used as precursors for pyrolytic treatment under the desired temperature in a tube furnace (OTF-1200XΦ50, Hefei KEJING Materials Technology Co., Ltd, Hefei, China) at a heating rate of 2 °C·min⁻¹ for 2 h in an air atmosphere to obtain NiO_x (Ayodele et al. 2014).

In order to investigate the adsorption capacity of NiO_x towards dyes, 0.02 g of NiO_x was immersed in a cone bottle (100 mL) containing 0.1 mmol/L AO7, ID, crystal violet, CR, methylene blue and rhodamine B, respectively. The water used for the preparation of the reagents and reaction system was purified by a Milli-Q (18.25 MΩ·cm). The mixture was stirred under a constant rate for 90 min under the desired temperature and then the solution was filtered by the micro PES membrane (pore size: 0.22 μm). The samples were withdrawn from the reaction solution and the residual samples concentration in the filtered solution was determined using UV-visible spectrophotometer. Control experiments were conducted and excluded the interaction between the filter and dyes.

After saturation, regeneration experiments were carried out. First, the NiO_x was washed by pumping water through the column in order to remove unadsorbed adsorbate traces. Then, the adsorbate was eluted with ethanol. The aforementioned procedure was repeated for eight consecutive regeneration cycles.

A sequence of adsorption experiments was performed by taking equal amounts of AO7, ID and CR which have similar chemical structure so as to evaluate the adsorption selectivity of NiO_x. The maximum absorbance in the UV spectrum can be obtained at the wavelength of 489 nm for AO7 and CR, 484 and 610 nm for AO7 and ID and 504 and 610 nm for ID and CR. For the sake of discussing the changes of absorbance in UV spectrum and visible color of the mixed dye solutions before and after adsorption by NiO, the mixed dye adsorption experiments were carried out by contacting 0.02 g of NiO_x with 100 mL mixed dye solution containing 0.05 mmol/L AO7 + 0.05 mmol/L CR, or 0.05 mmol/L AO7 + 0.05 mmol/L ID, or 0.05 mmol/L ID + 0.05 mmol/L CR at 25 °C for 90 min. The absorbance within the range of 300–700 nm was recorded for the mixed dye solutions before and after the adsorption.

The pH values of solutions were measured by the pH-meter (FiveEasy Plus, METTLER TOLEDO, Shanghai). A UV-visible spectrophotometer (UV-1800, Shanghai MAPADA Instrument Co., Ltd, Shanghai, China) was used for analyzing the dye concentrations by measuring the maximum absorbance at wavelengths of 484 nm for AO7, 610 nm for ID, 590 nm for crystal violet, 497 nm for CR, 664 nm for methylene blue and 554 nm for rhodamine B. After filtration with a cellulose ester membrane of 0.45 μm pore size and freeze vacuum drying in turn, the solid samples of residual NiO_x after reaction (reaction time 90 min) were prepared. The surface properties of the materials before and after adsorption were characterized by X-ray diffraction (XRD, EMPYREAN, The Netherlands), scanning electron microscopy (SEM, JSM-7500F, Japan), Brunauer-Emmett-Teller (BET, Micromeritic ASAP2460, USA), X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, USA), and Fourier transform infrared spectroscopy (FT-IR, Nicolet 6700, USA).

The AO7 and ID adsorption effect of the NiO_x synthesis at different calcination temperatures is shown in Figure 1. Obviously, the adsorption activity first increased and then decreased with increasing the calcination temperature of supports. Specifically, the samples at 250, 300, 350, 400 °C have a large adsorption, but the sample at 200 °C has little adsorption. Consequently, when the calcination temperature is 300 °C, the synthetic materials show great catalytic property with low energy consumption and present a good application prospect.

NiO_x exhibited different adsorptive performances towards different dyes, as shown in Figure 2. For AO7, ID and CR, NiO_x shows better removal efficiency. However, for crystal violet, methylene blue and rhodamine B, the removal efficiency is less than 1%.

The uptake of dyes by the NiO_x may be a complex process, with a variety of intermolecular forces possibly contributing to interactions between a penetrant and the polymer, including pi–pi interactions between aromatic groups, dipole–dipole interactions and, for ionic species, charge–charge and charge–dipole interactions (Satilmis & Budd 2016).

From Figure 2 we can see that NiO_x shows selectivity adsorptive capacities towards AO7, ID, crystal violet, CR, methylene blue and rhodamine B. As we know, AO7, ID and CR are anionic dyes which contain negatively charged sulfonic (–SO₃-) groups, while no –SO₃- exists in crystal violet, methylene blue and rhodamine B. Bandara et al. (1999) proposed that azo dye adsorption on titanium, iron, and aluminum oxides occured via the –SO₃- of the dye. Bauer et al. (1999) and Kyriakos et al. (2005) concluded that the adsorption of AO7 on the TiO₂ surface occurs via the –SO₃- of AO7 through the formation of a bidentate inner sphere surface complex. Zhang et al. (2016a) suggested that acid fuchsin adsorption on ZnO was the result of the electrostatic attraction between Zn²⁺ and –SO₃- in acid fuchsin. In view of the aforementioned reviews, this study investigated AO7 and ID adsorption on NiO_x and the electrostatic attraction may also occur between Ni²⁺ and –SO₃-. Hence, the XRD patterns, SEM images, XPS spectra, and FT-IR spectra analysis were used to analyse the original NiO_x and the NiO_x-AO7 and NiO_x-ID which were the NiO_x after adsorbing AO7 and ID.

XRD patterns for the three samples (NiO_x, NiO_x-AO7 and NiO_x-ID) are shown in Figure 3. All the observed patterns show five peaks located at 37.2, 43.3, 62.8, 75.4 and 79.4°, which are indexed to the (111), (200), (220), (311), and (222) lattice planes of NiO (JCPDS No. 47–1049) with a face-centered-cubic (fcc) structure (Lv et al. 2015). No other peaks are observed, which indicates that the NiO_x is NiO, and pure crystalline NiO can be gained by calcining the precursor (Zheng et al. 2017). The obtained three XRD patterns indicate the formation of NiO and show no other diffraction peaks. Consequently, no crystal change occurs between the NiO and the other dyes during the adsorption process.

To characterize the textural properties of the NiO, the BET specific surface area and pore-size distribution of the NiO were determined by N₂ adsorption-desorption at –196 °C. From Figure 4(a), we can see that a high BET surface area (S_BET, 251.82 m²/g) and a total pore volume (V, 0.38 cm³/g) are obtained in virgin NiO. As shown in Figure 4(b), in accordance with the classification of the International Union of Pure and Applied Chemistry, the obtained NiO (average pore diameter 5.78 nm) is normally classified to mesopore (2–50 nm) (Sing 1985).

To investigate the morphologies and structures of NiO, NiO-AO7 and NiO-ID composites, SEM images of them were characterized. According to the SEM findings, NiO is successfully synthesized. As shown in Figure 5, it can be seen that NiO have rough surfaces, which resulted in large specific surface areas and active adsorption sites (Zhou et al. 2017). These results are consistent with the analysis obtained from the BET analysis. These SEM images can reflect the crystallite evolutions of samples. It can be seen that the shape of NiO has no obvious change and the crystal could maintain their morphologies after adsorption, indicating that the framework is stable.

XPS was used to further investigate the detailed chemical composition and chemical states of the NiO, NiO-AO7 and NiO-ID samples. The survey spectra (Figure 6(a)) display that NiO, NiO-AO7 and NiO-ID are composed of C, O, S and Ni. The high-resolution O1s spectra of NiO before and after AO7 and ID adsorption are shown in Figure 6(b). The surface hydroxyl of the NiO can be verified by inspecting the O spectrum. For the virgin NiO, a peak at 529 eV is found, which is related to lattice oxygen. A dominant peak at 530.9 eV is ascribed to surface hydroxyl groups (i.e. Ni–OH) (Yang et al. 2015a; Wei et al. 2017). The O1s spectra after AO7 and ID adsorption show that the peak of COO appears, which further confirms the higher extent of the adsorption process. The C1s spectra are shown in Figure 6(c), a dominant peak at 284.6 eV can be ascribed to C–C bonds and relates to the sp2 hybridized graphite-like carbon atom in grapheme (Tang et al. 2017). It is noteworthy that there are C–C, C–O and COO signals on the surface of the virgin NiO, which is because the original NiO was gained by calcining nickel oxalate. Moreover, the intensity of the C–C, C–O and COO signals are higher in NiO-AO7 and NiO-ID samples as compared with that in the NiO, which indicates that the AO7 and ID have been adsorped. In Figure 6(d) and 6(e), the sulfur and nitrogen are detected in the NiO-AO7 and NiO-ID but not in the virgin NiO, demonstrating the existence of AO7 and ID in the final NiO samples. With respect to the S2p orbitals, only one peak at 167.8 eV can be seen, it could be confirmed that the sulphur is bound to the functionalized carbons in the form of sulphonic acid groups (–SO₃-) (Landwehr et al. 2017). The Ni2p spectra are shown in Figure 6(f). It is obvious that compared to the NiO, the spectra of NiO-AO7 and NiO-ID are seldom changed and are near to the peak of NiO, which indicates that the framework is stable. These results are consistent with the analysis obtained from the SEM analysis. The high-resolution Ni2p core-level spectrum of the virgin NiO can be assigned to the Ni2p_3/2 (850–865 eV) and Ni2p_1/2 (870–885 eV) spin–orbit levels. The region of Ni 2p_1/2 contain one peak centered at ca. 872.1 eV, a broad satellite peak of Ni 2p_1/2 at ca. 879.0 eV, two other peaks for Ni 2p_3/2 at ca. 853.6 and 855.4 eV and the satellite peaks of Ni 2p_3/2 centered at 860.5 eV, respectively (Wu et al. 2016; Zheng et al. 2017). The stronger peaks at 853.6 and 872.1 eV correspond to Ni²⁺ in Ni–O bonds (Chen et al. 2015; Mishra et al. 2018). In the reaction system, it is probable that the –SO₃- groups react with the oxygen-containing substance in the adsorbent or the lattice oxygen to generate a positive hexavalent sulfide. Thereby, the above results further confirmed that –SO₃- groups were firmly bonded to the NiO.

The FT-IR spectroscopy was performed to further study the adsorption mechanism. FT-IR spectra were characterized before and after AO7 and ID adsorption. As shown in Figure 7, it can be inferred that the AO7 and ID molecules are adsorbed on the surface of NiO. For AO7 and ID adsorption, some of the characteristic peaks of AO7 and ID (marked with an imaginary line) are found in the AO7-loaded and ID-loaded NiO which are absent in the pristine NiO, confirming the existence of AO7 and ID in the final solid (Chen et al. 2013; Zhang et al. 2016a). Evidently the peaks of NiO have a significant change after adsorption. As for the spectra of AO7 and ID, the bands at 1,123 and 1,037 cm⁻¹ are due to the coupling between the benzene mode and vs (SO₃) (Zhang et al. 2007; Jin et al. 2014). After adsorption experiments, some FT-IR peaks of AO7 and ID nearly disappear. However, the strong peaks at 1,037 and 1,218 cm⁻¹ suggest AO7 and ID adsorption onto the NiO. It should be noted that the appearance of bands at 1,366 cm⁻¹ for NiO-AO7 and 1,261 and 1,505 cm⁻¹ for NiO-ID may show that the AO7 and ID molecule were adsorbed on the NiO surface via the two oxygen atoms of the sulfonate group of the dye (Bourikas et al. 2005). In particular, two new peaks at 1,210 cm^–1 for NiO-AO7 and 1,190 cm^–1 for NiO-ID are found after adsorption. These results suggest the electrostatic attraction between Ni²⁺ and –SO₃- in AO7 and ID (Bourikas et al. 2005; Zhang et al. 2016a).

At the pH range of 4.0–10.0, the effects of initial pH on the adsorption capacities of AO7 and ID onto NiO are shown in Figure 8. The adsorption capacities of AO7 and ID changed slightly with pH increasing from 4.0 to 10.0, which shows that the change of pH has little effect on the adsorption performance. Especially at circumneutral pH (pH = 6–8), the removal rates are high and basically unchanged, which proves that NiO particles have a positive surface charge at neutral pH and makes it suitable to adsorb anionic dyes from aqueous solution (Zheng et al. 2017). Zheng et al. (2017) reported that the CR can be adsorbed by flower-like NiO, whose surface hydroxyl groups would be easily protonated, and then the NiO surface have a net positive charge. There is no doubt that the solution pH plays a crucial role in that adsorption process. However, the above-mentioned results reveal that the pH has no effect on the adsorption activity.

On the basis of the above results, the adsorption mechanism of AO7 and ID onto the NiO surface is proposed to be electrostatic adsorption. When the virgin NiO dispersed in aqueous suspension, the NiO surface had Ni²⁺. The NiO surface Ni²⁺ has a positive charge. As anionic dyes, the structures of AO7 and ID contain an –SO₃- group. Therefore, the negatively charged –SO₃- group of AO7 and ID would be attracted to the positively charged surface of the NiO particles through electrostatic interaction. In addition, the solution pH plays no role in that adsorption process.

So as to investigate the effect of contact time and adsorption kinetics for dye adsorption, 0.1 g of NiO was immersed in a 500 mL beaker containing 0.1 mmol/L AO7 or ID. The effects of contact time on the adsorption capacities of AO7 and ID by NiO are shown in Figure 9(a). Within 30 min, the adsorption capacity increased rapidly. It is observed that the adsorption equilibrium can be nearly achieved within 120 min for AO7 and ID. AO7 and ID nearly reached the saturated adsorption within 90 min.

In order to further analyze the adsorption kinetics for removing AO7 and ID by NiO, the pseudo-first-order, pseudo-second-order, and intra-particle diffusion models were applied to fit the experimental data. The pseudo-first-order model assumes that the adsorption rate is proportional to the difference between q_e and q_t (Zhang et al. 2014). For the pseudo-second-order equation, it is considered that the sorption process is controlled by a chemical adsorption mechanism involving electron sharing or electron transfer between adsorbent and adsorbate (Zhang et al. 2016b). The intraparticle diffusion models which were often used to identify the possibility of intra-particle diffusion resistance affecting adsorption process is also studied (Zhang et al. 2013).

The experimental data are fitted to the kinetics models, and the resultant parameters are listed in Table 1. According to the analysis of correlation coefficients as shown in Table 1, R² of the pseudo-second-order model are all much closer to 1.0 than other kinetics models, indicating that the pseudo-second-order equation is the most suitable one to describe the adsorption kinetics of NiO for AO7 and ID. It indicates that chemisorption is involved in adsorption.

In this study, the thermodynamic parameters of the adsorption process are obtained from experiments at various temperatures (35–60 °C). The relationship between temperature and adsorption capacities of AO7 and ID on NiO is shown in Figure 9(b). For AO7, the adsorption capacities decrease slightly with the increasing temperature, suggesting that adsorption of AO7 onto the NiO became less favorable at higher temperatures. However, for ID, the adsorption capacities increase slightly with the increasing temperature, suggesting that adsorption of ID onto the NiO became more favorable at higher temperatures (Kamari et al. 2009).

The values of ΔG, ΔH and ΔS are tabulated in Table 2. The negative values of ΔG for AO7 and ID indicate that the adsorption on the NiO is a spontaneous. For AO7, the negative values of ΔH show that the sorption process is exothermic for sorbent and the negative values of ΔS suggest a decrease of the randomness at the solid–liquid interface (Dragan et al. 2017). However, for ID, the positive values of ΔH show that the sorption process is endothermic for sorbent and the positive values of ΔS indicate the increase of the randomness at the solid–liquid interface.

For investigation on the adsorption mechanism, adsorption isotherms for AO7 and ID have been measured, since it is fundamental to describe the interactive behavior between the adsorbates and adsorbents. The dye adsorption isotherms of NiO are shown in Figure 10(a). Two adsorption isotherm models, Langmuir (6) and Freundlich (7), were used to further describe the adsorption equilibrium. The Langmuir isotherm is mainly applied to describe the monolayer adsorption that occurred on the homogeneous surface of adsorbent (Li et al. 2013). The Freundlich model is applied to describe heterogeneous surface adsorption and multilayer adsorption under various non-ideal conditions (Li et al. 2013).

The calculated parameters and correlation coefficients for the adsorption of AO7 and ID onto NiO are summarized in Table 3. According to Table 3, the R² of the linear form for the Langmuir model are much higher than the Freundlich model in each investigated dye system. The isothermal adsorption behaviors of AO7 and ID on NiO all follow the Langmuir equation, indicating that a monolayer adsorption of the dyes takes place at a homogeneous surface with approximately identical energy.

Since reusability of a spent adsorbent is one of the most crucial factors when assessing the cost-effectiveness of the overall processes, the regeneration of exhausted NiO was carried out. During this study, spent NiO was chemically regenerated with ethanol. Ethanol, a green solvent, can be separated from dyes after elution by the distillation process and eluted dyes can be utilized for various applications (Yusuf et al. 2017). Figure 10(b) shows the percentage of AO7 adsorbed onto NiO in each of three consecutive adsorption-regeneration cycles. In the first adsorption cycle, the removal of AO7 was 71.5%. After three consecutive cycles of adsorption-regeneration, the adsorption of AO7 was reduced to 54.4%. The results showed 23.9% loss in NiO adsorption performance for AO7 after three consecutive regeneration cycles.

Figure 11 shows the changes of absorbance in UV spectrum and visible color of the mixed dye solutions before and after adsorption by NiO. As shown in Figure 11(a), only one obvious absorption band located at 489 nm can be observed for the mixture of AO7 and CR solutions. After 90 min, the absorbance at 489 decreased significantly and the removal rate reached 85.9%. Figure 11(b) and 11(c) show the characteristic absorption bands located at 484 and 610 nm for the mixture of AO7 and ID solutions and 504 and 610 nm for the mixture of ID and CR solutions, respectively. Similarly, for Figure 11(b) and 11(c), the characteristic absorption band at 610 nm of ID after adsorption was nearly absent. The absorbance at 484 nm for AO7 also decreased significantly and the removal rate reached 80.8% (Figure 11(b)). It is a remarkable fact that the characteristic absorption band at 504 nm of CR after adsorption also disappeared. These results agree well with the changes in the visible color after adsorption, especially the mixture of ID and CR solutions (Figure 11(c)), whose color changes from purple to colorless. All the results mentioned above demonstrate the high effectiveness and no selectivity of NiO for removing AO7, ID and CR in the mixed dye solutions which are anionic dyes that contain a negatively charged sulfonic group (–SO₃-).

In the dyeing process, a considerable number of inorganics have been utilized as mordants or additives, which have a significant influence on wastewater treatment (Liu et al. 2016). To investigate the realistic conditions, the effect of various water matrices (i.e. running water and surface water) on the removal of AO7 and ID was studied using the NiO, as shown in Figure 12. It is observed that the surface water was used as received without further purification. As can be seen in Figure 12, the removal efficiency of AO7 (ID) in deionized water was 86.55% (83.51%), while the removal efficiencies in the running water and surface water were 75.73% (76.69%) and 68.74% (71.30%) (within 90 min reaction), respectively.

It is universally acknowledged that there are inorganic species (e.g. bicarbonate, sulfate and chloride) in the running water and surface water for the cationic sites on the surface of the NiO. This process can limit effective adsorption of AO7 and ID. Nevertheless, even though the removal rate for AO7 (ID) dropped in the surface water sample, it also had a higher removal rate of 68.74% (71.30%), which suggests that the inorganic species could slightly reduce but not inhibit the removal rate in natural waters. Thus, the promising result demonstrates that NiO has a great adsorption potential for actual wastewater.

In this study, the adsorption performances of NiO for dyes (AO7 and ID) were evaluated. NiO could show adsorptive removal capacity towards anionic dyes (AO7 and ID) which have –SO₃- structure. The results indicated that the excellent adsorption capacities towards AO7 and ID by NiO may be the electrostatic attraction between Ni²⁺ and –SO₃-. Furthermore, the results indicate that the temperature shows obvious function to adsorption properties, but the solution pH plays no role in the removal of dyes. The adsorption capacity of AO7 and ID is 178.57 and 227.27 mg/g respectively under ambient temperature (25 ± 1 °C). The adsorption process is spontaneous and exothermic for AO7, but endothermic for ID. Finally, NiO shows no adsorption selectivity for the binary anionic dye system and has a good adsorption potential for actual wastewater. In summary, the new and facile preparation method and pronounced adsorption properties enable NiO to be promising in the practical removal of anionic dyes from effluents and it is a promising strategy for selectivity of anionic dyes from mixed dyes.

Appreciation and acknowledgment are given to the National Natural Science Foundation of China (No. 51508353) and the Chengdu Science and Technology Department (grant No. 2015–HM01–00536-SF). In addition, we are grateful for the characterization examination test provided by ‘ceshigo’ (www.ceshigo.com).