Adsorption of iron(III), cobalt(II), and nickel(II) on activated carbon derived from Xanthoceras Sorbifolia Bunge hull: mechanisms, kinetics and influencing parameters

Rapid development of industrialization and urbanization processes can generate large amounts of toxic heavy metal ions that can contaminate surface and groundwater resources. Once heavy metals have been released into natural and man-made ecosystems, even at trace levels, the ions do not biodegrade and may bioaccumulate in living tissues through the food chain, resulting in the uptake of heavy metals by humans and other organisms, impacting human health and inhibiting the growth of aquatic organisms (Tabrizi & Zamani 2016). Thus, the removal of heavy metals from industrial pollution has become a critical and global environmental problem. Heavy metal pollution results from Fe(III), Co(II), and Ni(II) used in mining operations, battery manufacturing, iron and steel industrial units, etc., which are responsible for direct and indirect discharge of effluents into headstreams (Güzel et al. 2008).

Many conventional technologies, such as flocculation, membrane separation, ion exchange, adsorption, and hybrid methods combining these techniques have been applied (Tao et al. 2015; Dotto et al. 2016). Among these approaches, adsorption is one of the most promising technologies because of its high efficiency, low maintenance costs, simplicity to operate and lack of secondary sludge production. Recently, the application of carbonaceous materials as adsorbents for removing heavy metals from wastewater has become an attractive field for scientific studies. Activated carbon is considered as an eco-friendly and effective adsorbent due to its distinctive structure, high surface area, well-developed pore volume and large number of active sites for adsorbing heavy metal ions (Choi et al. 2015). Considering economic capital, it is important to produce an abundance of effective activated carbon from low-cost and renewable by-products and agricultural wastes, e.g., watermelon shells, coconut shells, palm shells, walnut shells and other agricultural residues, to serve as adsorbent materials for removing heavy metals from effluents (Daud & Ali 2004; Azevedo et al. 2007; Depci et al. 2012; Moreno-Barbosa et al. 2013).

In the northern region of China, Xanthoceras sorbifolia Bunge hull (XSBH) was used to produce activated carbon with extensive reserves but has rarely been used as a precursor for activated carbon for the following two reasons: XSBH may be useful for agricultural waste assessment and to increase the competitiveness of the carbonaceous markets in China. Activated carbon is prepared using physical or chemical activation. In comparison to physical activation, chemical activation is a procedure in which the precursor is treated with chemical agents and then pyrolysed by heating at temperatures of 450–950°C under a flow of inert gas (nitrogen or argon). Chemical activation is preferred in most cases because it depends on the biomass precursors and results in higher global yields, larger surface area and less burn-off char, and has lower energy costs with lower operating temperatures relative to physical activation (Tang et al. 2016). Additionally, chemical activation can result in certain improvements; for example, chemical activation can generate surface functional activation groups, produce new activation sites and strengthen the mechanical stability of the adsorption layers. Agricultural waste activated carbon materials possess basic functional groups containing oxygen, e.g., hydroxyl, carbonyl, phenolic, and alcoholic groups. The various functional groups on the surface of activated carbon have affinities and adsorption capacities for heavy metal ions because they can donate an electron pair to form metal-binding chelates or complexes. The adsorption mechanisms are related to, for example, surface complexation, diffusion through micropores, chemisorption, ion exchange, and electrostatic interactions, which may act alone or together (Oliveira et al. 2014). Hence, the potential usage of XSBH as an economic, abundant, renewable, and simple adsorbent for the removal of heavy metal ions from aqueous environments is thoroughly discussed in this paper.

In this work, we successfully produced activated carbon (hereafter termed XSA) from XSBH by using chemical activation with H₃PO₄, and characterized the produced XSA using nitrogen adsorption/desorption isotherms (Brunauer–Emmett–Teller (BET) equation) and scanning electron microscopy/energy dispersive X-ray spectroscopy (SEM/EDX). The prepared XSA was used to remove heavy metal ions, including Fe(III), Co(II), and Ni(II), from aqueous solutions. To date, only a few studies have been reported that compare Fe(III), Co(II), and Ni(II) heavy metal ions. The main novelty of this work is that it can be used to analyse the different adsorption capacities of XSA for the iron group ions Fe(III), Co(II), and Ni(II), which are all located in the fourth period and in group VIII of the periodic table but have different ionic radii, oxidation numbers, electronegativities and valence-electron arrangements. Many factors that influence the adsorption and desorption capacities of Fe(III), Co(II), and Ni(II) by XSA, such as the initial metal concentration, pH, adsorption temperature, adsorption time, desorption agent HNO₃ concentration, desorption temperature and sonicated desorption time, were investigated systematically. Kinetic models were obtained by fitting the experimental data to pseudo-first order, pseudo-second order, intraparticle and Elovich kinetic models, and the equilibrium isotherms were determined using the Langmuir, Freundlich, Temkin, and Dubinin–Radushkevich models. Moreover, the current research presents an exhaustive regeneration study. Finally, we addressed the adsorption mechanisms by using Fourier transform infrared spectroscopy (FTIR). The use of XSA as a low-cost and easily available adsorbent for removing Fe(III), Co(II), and Ni(II) can serve as a valuable reference for controlling the environmental pollution of water.

XSBH was collected from solid waste markets located in Chifeng (China) as a precursor for preparing the activated carbon. Stock solutions of Fe(III), Co(II), and Ni(II) were prepared using their respective nitrate and chloride salts (Fe(NO₃)₃·9H₂O, Co(NO₃)₂·6H₂O, and NiCl₂·6H₂O (analytical grade)), which were purchased (Tianjin Beilian Fine Chemicals Co., Ltd, China) and prepared in deionized distilled water, and appropriate serial concentrations of each metal ion were prepared using the stock solutions for adsorption experiments. All other commercial chemicals were of analytical reagent grade and were used without further purification. All solutions were prepared using re-ionized distilled water.

The samples of XSBH were washed thoroughly with tap water twice and then with hot deionized distilled water to remove leftover rinds, colour, and water-soluble impurities. Next, the samples were dried for 24 hours at 80°C in a hot air oven (DZF-6210, Shanghai, China) and ground to a powder (0.5–0.8 mm). The chemical activation method was used with H₃PO₄ to prepare the activated carbon from XSBH (Molina-Sabio & Rodrigues-Reinoso 2004). Dried XSBH was washed until the pH of the filtrate was near 7 according to a pH meter (PB-10, Sartorius, Germany) and dried at 100°C for 24 h to completely remove moisture. The dried mass was mixed well with H₃PO₄ (45 wt %) and magnetically stirred (500 rpm) at 50°C for 1 h. The chemical ratio was defined as the weight ratio of H₃PO₄ to the precursor. The slurry was heated to a final carbonization temperature of 500°C in a tube furnace (FSX2-12-15N, Tianjin, China) for 1 h. Samples were cooled to the room temperature under flowing nitrogen and washed sequentially with de-ionized distilled water until a neutral pH was attained. The products were then vacuum dried in an oven overnight at 120°C before crushing, grinding, and passing through 200-mesh sieves (Molina-Sabio & Rodrigues-Reinoso 2004). Yields of 34% were attained. Then, the resulting powdered carbon XSA was stored in air-tight containers for future use.

To investigate the reusability of the XSA, repeated adsorption/desorption experiments were performed. After the first batch adsorption experiment, the activated carbon was washed with re-deionized distilled water to remove the remaining acid and then dried in a vacuum (DZF-6210, Shanghai, China) at 70°C before performing the next adsorption experiments. The regenerated XSA was used in five consecutive cycles under identical conditions.

The BET specific surface area, pore structure and pore size of XSBH and XSA were measured using nitrogen adsorption/desorption isotherms at 196°C (Micromeritics ASAP 2020, USA). Morphological changes and surface analysis of the samples were recorded using SEM-EDX (Hitachi S-4800). Chemical characterization of the functional groups was detected using an FTIR spectrophotometer (Thermo Nicolet, NEXUS, TM) with a scanning range of 4,000–400 cm⁻¹.

The surface area and elemental analysis are the important properties of activated carbon. Table 1 provides a comparison of the surface porosity structure parameters and elemental analysis of XSA by using H₃PO₄, ZnCl₂ and KOH as chemical activation agent, obtained from the N₂ adsorption/desorption isotherm and EDX. From the data in Table 1, it can be observed that the BET surface area, Langmuir surface area and total pore volume of XSA using H₃PO₄ as activation agent were relatively high (compared with ZnCl₂ and KOH) and reached maximum values of 688.62 m²/g, 1,132 m²/g and 0.377 cm³/g, respectively. Meanwhile, the results represented an average pore diameter of 2.20 nm, implying that most of the mesoporous structures developed during the chemical activation process (Kırbıyık et al. 2016). Additionally, it also can be seen from the Table 1 that carbon and oxygen are the main components of XSA because carboxyl and hydroxyl groups are potential functional groups for heavy metals. The results indicated that XSA has higher carbon content and oxygen content during H₃PO₄ activation treatment (Abdolali et al. 2015). These signals were consistent with the FTIR spectral data. Based on the above-mentioned discussion, XSA can allow the formation of adequate activated sites and functional groups for chelation and complexation with heavy metals.

The values of adsorption capacity by other activated carbons from various materials in literature are presented in Table 2 for comparison. It is clear from this table that the adsorption capacity of XSA for Fe(III), Co(II), and Ni(II) is much higher than with other activated carbons.

When comparing the Fe(III), Co(II), and Ni(II) ions, the adsorption capacity of Fe(III) is maximum. This result may be explained by the following observations. (1) The ionic radii of the ions increase as follows: Fe(III) (60 pm) < Ni(II) (70 pm) < Co(II) (72 pm). Smaller ionic radii correspond to a greater affinity of the metal for the active sites. (2) Chelation may produce coordination compounds between metal ions and the functional groups –COO⁻ and –OH⁻ on the surface of XSA. The oxidation number of Fe(III) is +3, with a common coordination number of 6, and the oxidation number of Ni(II) and Co(II) is +2, with a common coordination number of 4. Larger coordination numbers result in a greater adsorption capacity. In addition, the octahedral complex (d²sp³) of Fe(III) is packed and more stable than the tetrahedron complexes (dsp²) of Co(II) and Ni(II) ions. (3) The electronegativities are 1.96 for Fe(III), 1.91 for Ni(II), and 1.88 for Co(II). It is well known that higher electronegativity corresponds to greater element reactivity. Thus, the reactivity of the metals with XSA decreased as follows: Fe(III) > Ni(II) > Co(II). (4) According to the periodic table, the arrangement of single electrons in the 3d orbitals of Fe(III), Co(II) and Ni(II) are 3d⁵, 3d⁷, and 3d⁸, respectively. Thus, the outer valence electrons of Fe(III) are arranged as 3d⁵, which is a semi-stable state, and the valence electrons of Co(II) are arranged as 3d⁷, which is less stable, and can lose two electrons, and the valence electrons of Ni(II) are arranged as 3d⁸, which is the least stable, and can lose three electrons. Thus, the metallic properties of XSA for Fe(III), Ni(II), and Co(II) adsorption gradually decrease. Based on the above discussion, ion exchange, electrostatic attraction and chelation are the most beneficial for Fe(III) adsorption on XSA, followed by Ni(II) and Co(II). In summary, Fe(III) was present at a minimum initial concentration of 640 mg/L and had a maximum adsorption capacity of 241.13 mg/g.

The pH plays an important role in controlling the adsorption of heavy metal ions from aqueous solutions because it directly impacts the abilities of the metal to compete with H⁺ and functional groups on the surface of the adsorbent (Zarandi et al. 2016). Thus, the pH was examined and was observed to vary from 2.0 to 6.3. The effects of pH on Fe(III), Co(II), and Ni(II) adsorption to XSA are shown in Figure 2(b). As shown in Figure 2(b), the adsorption increased from 120.21 mg/g to 239.12 mg/g for Fe(III), 93.01 mg/g to 184.42 mg/g for Co(II), and 40.85 mg/g to 123.16 mg/g for Ni(II) as the pH increased from 2.0 to 5.8, respectively. At higher pH values, adsorption of Fe(III), Co(II), and Ni(II) ions decreased to different degrees.

The adsorption of Fe(III), Co(II), and Ni(II) was very low at pH 2.0, which may be attributed to the fact that the high concentration of H⁺ in the highly acidic solution competes with the metal ions for the functional sites that have been protonated on the surface of XSA, resulting in electrostatic repulsion and the suppression of metal adsorption. For Fe(III) ions, functional groups such as carboxyl and hydroxyl groups begin to deprotonate by releasing H⁺ ions when the pH increases (2.0–2.8), and more negatively charged sites become available, which facilitates Fe(III) uptake. Maximum adsorption occurs at pH 2.8. At pH values higher than 2.8, hydrolysis and polymerization greatly increase, leading to the formation of gelatinous Fe(OH)₃ sediment; therefore, the adsorption of Fe(III) decreases. For Co(II) and Ni(II), the optimal pH values are 5.8 and 4.1, respectively. The surface of XSA is more negatively charged and more easily attracts positively charged metallic cations when the pH is between 2.0 and 5.8 for Co(II) and between 2.0 and 4.1 for Ni(II). Furthermore, at higher pH values, Co(II) (pH > 5.8) and Ni(II) (pH > 4.1) can be hydrolysed, generating the amphoteric hydrolysis products Co(OH)₂ and Ni(OH)₂, and K^ɵ_sp [Co(OH)₂] 1.60 × 10⁻¹⁵ > K^ɵ_sp [Ni(OH)₂] 6.48 × 10⁻¹⁶, which indicates that the solubility of these ions increased and the adsorption capacity decreased. Thus, the adsorption capacity is greater for Ni(II) than for Co(II). In summary, the adsorption capacities of the ions decrease as follows: Fe(III) > Ni(II) > Co(II).

The influences of temperature on adsorption should be investigated when considering practical applications. A correlation between the adsorption temperature and adsorption capacity was observed from 25°C to 70°C and is illustrated in Figure 2(c). As observed in the figure, the adsorption capacities of Fe(III), Co(II), and Ni(II) all increased and reached maximum adsorption capacities of 192.56 mg/g to 240.61 mg/g for Fe(III), 69.43 mg/g to 125.15 mg/g for Co(II), and 78.91 mg/g to 185.32 mg/g for Ni(II) as the temperature increased from 25°C to 50°C, which indicated that higher temperatures are conducive to the adsorption of metal ions on XSA. This result potentially occurs because physical adsorption mainly takes place when the temperature is near room temperature, resulting in a certain amount of adsorption. When the temperature increased, chemisorption was dominant, which was potentially the result of increasing the kinetic energy in the solution, which increased the probability of collisions between the metals and activated sites. In addition, increasing temperatures may produce a swelling effect within the mesoporous structure, which facilitates the uptake of metals. The general upward tendency also indicates that the adsorption process is endothermic and that net heat adsorption occurs. However, a slight decrease in the adsorption capacity at higher temperatures (>50°C) may occur because the chemical adsorption process is exothermic, spontaneous and entropy reducing; thus, as the temperature increases, the adsorption capacity would decrease. In addition, at high temperatures (>50°C), the bonds between the metals and functional groups may be destroyed and desorption is favoured. Then, additional studies were carried out using an adsorption temperature of 50°C for all of the iron group ions. Meanwhile, the same decreasing trend in adsorption capacity of Fe(III) > Ni(II) > Co(II) was observed.

It is important to determine the effects of the amount of time required to reach the adsorption equilibrium when conducting a batch of experiments. Figure 2(d) shows the influences of adsorption time on the adsorption of metal ions by XSA. As shown in Figure 2(d), the adsorption capacities of the Fe(III), Co(II) and Ni(II) ions increased rapidly from 177.81 mg/g to 236.98 mg/g for Fe(III), from 95.43 mg/g to 121.27 mg/g for Co(II), and from 120.19 mg/g to 185.29 mg/g for Ni(II), respectively, as the adsorption time increased from 20 min to 60 min. Next, the adsorption equilibrium reached a plateau for a prolonged period. This result could explain why metals were adsorbed by diffusion into the microporous structure and high vacancy active sites on the surface of XSA. In addition, coordination and chelation complexes were generated between the metals and functional groups, resulting in a sharp adsorption equilibrium. Thereafter, no further adsorption occurred with prolonged adsorption until 120 min, which may be attributed to the saturation of the functional groups and activated sites on the surface of XSA. Consequently, the optimal adsorption time is 60 min in this study for the removal of the iron group ions by XSA. Similar results of adsorption capacity, with a descending order of Fe(III) > Ni(II) > Co(II), were obtained.

The fits of these four models were checked by each linear plot of ln (q_e−q_t) versus t (pseudo-first-order model), (t/q_t) versus t (pseudo-second-order model), q_t versus t^0.5 (intraparticle diffusion), and q_t versus lnt (Elovich model), respectively. The R² and constant values for the four adsorption kinetic models were calculated and are given in Table 3. According to the calculated kinetic model parameters in Table 3 and from comparing the experimental equilibrium adsorption capacity, it was found that the values provided by the pseudo-second-order model were more similar to the estimated equilibrium adsorption capacity than those provided by the pseudo-first-order, intraparticle diffusion and Elovich models. Also, it could be observed that the R² value of the pseudo-second-order kinetic model was close to 1.0 (R² Fe(III), 0.9952, R² Co(II), 0.9957; R² Ni(II), 0.9886). These results suggested that the pseudo-second-order model better described the adsorption kinetic processes of Fe(III), Co(II) and Ni(II) onto XSA. Therefore, it is obvious that chemical adsorption should be the rate limiting step of the adsorption of iron group ions onto the prepared activated carbon XSA.

Comparisons of these isotherm models for the adsorption of the iron group ions onto XSA was performed by comparing each linear plot of C_e/q_e versus C_e (Langmuir isotherm model), lnq_e versus lnC_e (Freundlich isotherm model), q_e versus lnC_e (Temkin isotherm model), and lnq_e versus ɛ² (Dubinin–Radushkevich isotherm model). The calculated constants are listed in Table 4. From Table 4, it is clear that the Langmuir model provides the best fit relative to the other models, with maximum monolayer adsorption capacities of 249.56 mg/g for Fe(III), 126.05 mg/g for Co(II), and 187.96 mg/g for Ni(II) and R² values of 0.9914, 0.9846, and 0.9957, respectively. Interestingly, the R² values of the four isotherm models decreased in the following order: Langmuir > Freundlich > Dubinin–Radushkevich > Temkin for Fe(III); Langmuir > Freundlich > Dubinin–Radushkevich > Temkin for Co(II); and Langmuir > Dubinin–Radushkevich > Temkin > Freundlich for Ni(II). It is understood that the Langmuir model corresponded to a dominant electrostatic attraction, ion exchange, and chelation mechanism, while the order revealed that the adsorption process involved physical adsorption and complexation at the interface and the outer heterogeneous surface of XSA.

Table 4 shows the calculated R_L values, which were 0.0538 for Fe(III), 0.188 for Co(II) and 0.0743 for Ni(II), respectively. The results implied that XSA was favourable for the adsorption of iron group ions under the adsorption conditions employed in the current work.

The effects of different desorption temperatures on the desorption capacity of metal-loaded XSA are shown in Figure 3(b). The amount of desorption increased before slightly decreasing with increasing temperature. This fluctuation in desorption could be attributed to the fact that increasing temperature may improve the adsorption efficiency of the active sites and because H⁺ and the iron group ions can compete with each other for activate sites on the surface, increasing the desorption capacity when the temperature increases. In addition, the adsorption efficiency of the active sites may be weakened at a temperature above 40°C, thus producing a detrimental effect on desorption (Bas et al. 2014), which further substantiated the results of adsorption temperature. The desorption capacity decreased in the following order: Fe(III) > Ni(II) > Co(II).

The effects of different desorption sonication times on the metal-loaded XSA are shown in Figure 3(c). The desorption capacity increased during the first stage, and then no further increase occurred after desorption equilibrium and with increasing sonication time. This phenomenon corresponds to the ultrasound rules of producing holes. After a certain ultrasound reaction time, the number of holes in the solution reached saturation, the high temperature generated ultrasonic cavitation, and the high pressure increased the adsorbate energy, which was favourable for desorption (Mata et al. 2010). The desorption capacity still decreased from Fe(III) > Ni(II) > Co(II) at an optimum sonication time of 60 min.

Recovering and reusing an adsorbent would decrease processing costs and is very important for technical applications. The reusability of XSA was tested by performing five consecutive adsorption/desorption processes for each metal ion. All of the adsorption and desorption capacities of the iron group ions in the consecutive cycles are tabulated in Table 5. The data showed that XSA could be recycled up to three times for Co(II) and up to four times for Fe(III) and Ni(II) ions with very little loss of efficiency. Up to the 3rd–4th cycles, it was clear that the capacities of adsorption and desorption had excellent performance, indicating that XSA could be an economically feasible method for the removal of iron group ions from wastewater treatment.

The asymmetrical stretching vibration at 3,440 cm⁻¹ (Figure 4(a)) decreased and shifted to lower wavenumbers of 3,439, 3,430 and 3,432 cm⁻¹ after the adsorption of Fe(III), Ni(II) and Co(II) (Figure 4(b)–4(d)), respectively, especially for Fe(III) ions. The spectra of XSA after Fe(III) desorption were chosen as a model for assessing the differences due to the interactions between metal ions and functional groups on the surface of XSA (Figure 4(e)). The changes in –OH adsorption showed that the degree of hydroxyl polymerization varied from multimer to monopolymer in the XSA by binding Fe(III), Ni(II) and Co(II) ions (Figure 4(b)–4(d)), and Fe(III) had a higher binding capacity for XSA, and the asymmetrical stretching vibration was restored to 3,441 cm⁻¹ after Fe(III) desorption (Figure 4(e)). Interestingly, the characteristic adsorption band at 1,738 cm⁻¹ (Figure 4(a)), which corresponds to the asymmetric stretching vibration of the C = O bond in carboxylic acids, was diminished after adsorption of Fe(III), Ni(II) and Co(II) (Figure 4(b)–4(d)) and slightly weakened after desorption of Fe(III) at a lower wavenumber of 1,729 cm⁻¹ (Figure 4(e)). The changes in the peak at 1,624 cm⁻¹ (Figure 4(a)) correspond to the carbonyl and carboxylate groups that were reduced after adsorption of the iron group ions at 1,615, 1,618, and 1,617 cm⁻¹ (Figure 4(b)–4(d)), and were slightly regained after desorption of Fe(III) at a lower wavenumber of 1,620 cm⁻¹ (Figure 4(e)). The peak at 1,058 cm⁻¹ (Figure 4(a)) was reduced and moved to 1,053⁻¹ after adsorption for Fe(III) (Figure 4(b)), 1,056 cm⁻¹ for Ni(II) (Figure 4(c)), and 1,057 cm⁻¹ for Co(II) (Figure 4(d)). The adsorption effect for Fe(III) ions was typically the highest, which indicated that the combined ability of Fe(III)-XSA was higher than that of Co(II) and Ni(II). Then, the peak slightly decreased at 1,059 cm⁻¹ after Fe(III) desorption (Figure 4(e)), which corresponded with the results of the adsorption experiments. From these observations, it was tentatively concluded that the major adsorption sites on the surface of XSA for metal binding are the hydroxyl, carboxyl and carboxyl functional groups and that other groups (methyl, C ≡ C, methylene) hardly changed. In addition, slight changes were observed in the FTIR spectra of Fe(III) desorption, and the spectra were basically restored to their original shape. In conclusion, the coordination of Fe(III), Co(II) and Ni(II) ions with the presented functional groups and the chemical adsorption are predominant in the adsorption process of the iron group ions. Additionally, the basic structures and properties of XSA remained relatively stable during the process of metal adsorption and desorption, indicating that XSA is a good renewable adsorbent for wastewater treatment.

The most plausible explanations for the descending order of adsorption of Fe(III) > Ni(II) > Co(II) is that the ionic radius, the oxidation number, electronegativity, and the arrangement of the valence single electrons influenced the adsorption of the iron group ions by XSA, with a smaller ionic radius, higher oxidation number, greater electronegativity, and more stable outer electron arrangement resulting in preferential absorption.

In this work, a prepared XSA was tested for removing Fe(III), Co(II) and Ni(II) ions from aqueous solution. The maximum adsorption capacities were 241.13 mg/g for Fe(III), 126.05 mg/g for Co(II), and 187.96 mg/g for Ni(II), respectively. The adsorption kinetics and isotherm showed that the adsorption processes were described best by the pseudo-second-order and Langmuir models. Desorption and regeneration experiments revealed that HNO₃ was able to elute 167.23 mg/g of Fe(III), 107.34 mg/g of Co(II), and 114.22 mg/g of Ni(II). The FTIR and EDX spectra suggested the adsorption was a chemical adsorption process. Also, the ability of the metals to bind with XSA decreased in the following order: Fe(III) > Ni(II) > Co(II).

Authors Xiaotao Zhang and Yinan Hao contributed equally to this paper. We are grateful for financial support from the National Natural Science Foundation of PR China (No. 21467021) and the International Science & Technology Cooperation Program of China 2013DFA32000).