Transition metal antimonates with the general formula MSb2O6 (M = Zn, Cd, Pb, Ni, etc.) have been investigated primarily because of their interesting structure, electronic, and optical properties. These oxides crystallize in the orthorhombic crystal structure in the space group P42/mnm. NiSb2O6 has attracted attention due to its magnetic photocatalytic, as transparent conductor, and sensors properties . It finds applications as antimony metal oxide catalyst and in resistors [2-4]. Several methods have been reported for the synthesis of NiSb2O6 materials including solid state method using NiSO4 and NaSbO3 at 500 °C , solid state method using NiO and Sb2O4 at 900 °C , sol-gel [4,5], hydrothermal , solid state method using Sb2O3 and NiO at 800 °C for 72 h , solid state method using NiO and Sb2O3 at 1450 °C for 48h , microwave - assisted colloidal  and recently a solid state  methods.
MG is classified in the dyestuff industry as a triarylamine dye and used in pigment industry. MG has been used extensively in the leather, paper, silk, cotton, and jute dyeing processes. It is also used as an antifungal and anti-protozoan agent in fisheries and aquaculture industry [11, 12]. MG is a non-biodegradable dye pollutant and has now become a highly controversial compound due to the risks it poses to the consumers of treated fish, including its effects on the immune and reproduction systems. Furthermore, MG and its metabolites are known to cause mutagenic, carcinogenic, and teratogenic effects to living organisms . It should not be used for beverages, food, medicines because it causes skin irritation, blurred vision or cause interference. Its inhalation may cause irritation to the respiratory tract, and in large quantities can cause tissue damage and inflammation of kidneys . Pure and functionalized metal oxides nanopowders have found several catalytic and photocatalytic applications [15,16]. Recently, several metal oxides have been used for the degradation of MG under different conditions including: MoS2/TiO2 nanocomposite , PbCrO4 , TiO2/ZrO2 , Nix : TiO2 , TiO2 , V doped ZnO , TiO2 , Ni1-xCoxFe2O4 , Pt/TiO2/SiO2 , Sr2As2O7 , etc.
To find optimum values of parameters affect on the malachite green photodegradation conditions, experimental design method is used. In the method, a Design Of Expert
(DOE) using Central Composite Design (CCD) is applied.
Design of Expert is a piece of software designed to help with the design and interpretation of multi-factor experiments. In photocatalytic processes, the software is used to help researchers design an experiment to see how much a photocatalyst and H2O2 are used and how many minutes are needed to finalize the degradation process. The software offers a wide range of designs, including factorials, fractional factorials and composite designs. Design Expert offers computer generated D-optimal designs for cases where standard designs are not applicable, or where we wish to augment an existing design [27, 28]. A Box-Wilson Central Composite Design, commonly called a central composite design (CCD), contains an imbedded factorial or fractional factorial design with centre points that is augmented with a group of points that allow estimation of curvature. If the distance from the centre of the design space to a factorial point is ±1 unit for each factor, the distance from the centre of the design space to appoint is |α| > 1. The precise value of α depends on certain properties desired for the design and on the number of factors involved [27, 28].
In the present study, a facile one-step solid state method was explored for the synthesis of nanostructured Ln-doped NiSb2O6 powders using Ni(NO3)2, Sb2O3 and Ln2O3 raw materials at 800 °C for 8 h. To the best of our knowledge, there is no information available in the literature about doping the Ln2O3 into NiSb2O6 crystal system under the present conditions. The direct optical band gap energies of the as-prepared NiSb2O6 nanomaterials were initially estimated from ultraviolet-visible spectra. Photocatalytic performance of the synthesized NiSb2O6 nanomaterials is investigated for the degradation of MG under solar light condition. Experimental design method was used to optimize factors affecting the degradation reaction. The factors are the amount of the nanocatalyst, H2O2 and the reaction time.
All chemicals were of analytical grade, obtained from commercial sources, and used without further purification. Phase identifications were performed on a X-ray powder diffractometer D5000 (Siemens AG, Munich, Germany) using CuKα radiation. The morphologies of the obtained materials were studied by a field emission scanning electron microscope (Hitachi FE-SEM model S-4160). Absorption spectra were recorded on an Analytik Jena Specord 40 (Analytik Jena AG Analytical Instrumentation, Jena, Germany). The Software used for the design of experiment (DOE) was Design Expert 7. Measurement of the photocatalytic activity of the synthesized samples was investigated in the presence of H2O2 (30%, w/w) under visible light source. A Shimadsu, UV-Vis 1650 PC spectrophotometer was used to measure the absorbance spectra of MG in the range of 200–700 nm by a quartz cell with an optical path of 1 cm. A BEL PHS-3BW pH-meter with a combined glass-Ag/AgCl electrode was used for adjustment of test solution pH.
In a typical solid state synthetic experiment, 0.01 mmol of Ln2O3 (Eu2O3 (S1), Gd2O3 (S2), Ho2O3 (S3) and Yb2O3 (S4)), 1 mmol of Sb2O3 (Mw = 291.5 gmol−1) and 1 mmol of Ni(NO3)2.6H2O (MW = 290.7 gmol−1) were mixed in a mortar and ground until a nearly homogeneous powder was obtained. The obtained powder was added into a 25 mL crucible and treated thermally in one step at 800 ºC for 8 h. The crucible was then cooled normally in the furnace to the room temperature.
RESULTS AND DISCUSSIONS
The XRD patterns of the as-synthesized Ln-doped NiSb2O6 samples are reported in Fig. 1 (a-e) with the results of the profile matching analyses (full lines). Fig. 1 shows the XRD analyses of the obtained samples in the θ-2θ geometry with Cu-Kα radiation. The structural results done by FullProf program showed that the patterns had a main NiSb2O6 tetragonal crystal structure with space group of P42/mnm [7-9, 29, 30]. S0 is the as reported data corresponded to the pure NiSb2O6 . According to the rietveld analyses shown in Fig. 1 (b-e), it is clear that doping the lanthanide ions reduced the purity of NiSb2O6. The impurity phases are Sb2O5 , Sb2O3  and NiO . The data reveal that when the dopant ions are introduced into the crystal system, the volume of the as-synthesized samples are decreased slightly. However we know that the ionic radii of the lanthanide ions are larger than those of Ni and Sb ions, the surprising observation can be due to the lanthanide 4f contraction effect on the cell volume of the final doped products.
Some crystallographic parameters were calculated and summarized in Table 1 to study the dopant effects on the crystallographic property of the as-synthesized nanomaterials.
The lattice volume is obtained by the below formula (Table 1):
The data indicates that the intercalation of the dopant ions into NiSb2O6 crystal system has occurred into the larger crystal cavity.
The interplanar spacing (d) value is also calculated by the below formula:
By using the peak with the highest intensity at 2θ ≈ 27.56 º and the (h k l) value of (2 1 1), the above equation is changed to the following formula:
The grain size data (D) of the obtained nanomaterials is calculated by Scherrer equation (Table 1):
Where D is the entire thickness of the crystalline sample, λ is the X-ray diffraction wavelength (0.154 nm), K is the Scherrer constant (0.9), B1/2 of FWHM is the full width at half of its maximum intensity and θ is the half diffraction angle where the peak is located. As could be found from the data, the grain size of the as-synthesized doped materials has been decreased when the dopant ions are introduced into the crystal system.
Also, the dislocation density (δ) [(lines/m2)1014] value related to the number of defects in the crystal is calculated by the following relationship:
As it is indicated in Table 1, it is found that the δ value is increased when the dopant ions are introduced into the crystal system. However, the data show that doping the lanthanide ions into the crystal cavity increases the dislocation density considerably that can be due to the decreasing the nanomaterials grain sizes.
The strain ε (10–3) values are also calculated by the following formula:
Table 1 includes the variation of ε as a function of the dopant ion type in the crystal system. The data show that the value of strain for NiSb2O6 is small. But, when the lanthanide ions are doped into the crystal system, the strain value is increased. This observation is probably due to the retrograding the crystallinity degree of the obtained material.
Fig. 2 (a-d) related to S1 – S4, respectively, present the FESEM images of the as-prepared NiSb2O6 nanomaterials. As previously reported, pure NiSb2O6 has particle morphology with the average size of 40 – 50 nm. However, when the dopant ions were introduced, the morphology was changed to homogeneous sponge structure. The data reveal that doping the lanthanide ions improved the morphology and reduced the particle size of the final product. The data reveals that the crystallite sizes are 30 – 40, 20 – 30, 20 – 30 and 20 – 30 nm for S1, S2, S3 and S4, respectively.
Energy dispersive X-ray analysis
Fig. 3 (a – d) illustrates the EDX analysis spectra of S1-S4, respectively, with 0.01 mmol of the dopants ions into the crystal system studying the compositional analysis of Eu3+ or Gd3+ or Ho3+ and/or Yb3+ in NiSb2O6. The peaks corresponded to Eu or Gd or Ho or Yb and (Ni, Sb and O) atoms present in the samples are labeled. The respective energy positions and the specific X-ray lines from various elements are also indicated. Besides, the A% values of the dopants in the obtained samples and studying the capacity of NiSb2O6 to accept the ions in the crystal systems is reported. The A% values are 0.48, 0.64, 0.43 and 0.57 for Eu3+, Gd3+, Ho3+ and Yb3+ doped nanomaterials. The data reveal that when a heavier lanthanide ion is included in the crystal system, the capacity of the crystal system to accept the ions is decreased.
The UV-Vis spectra and direct optical band gap energies of the as-synthesized NiSb2O6 nanocomposites which obtained from UV-Vis absorption spectra are shown in Fig. 4. According to the results of Pascual et al., the relation between the absorption coefficient and incident photon energy can be written as (αhν)n = A(hν - Eg), where A is a constant and Eg is the direct band gap energy if n=2. The Band gap energy was evaluated from extrapolating the linear part of the curve to the energy axis. The smallest value of the direct optical band gap energies are 1.6 and 1.8 eV for S1, S2, S3 and S4, respectively.
The photocatalytic activity of the previously synthesized NiSb2O6 sample (S0) was investigated for the degradation of MG in the presence of H2O2 (30%, w/w) under visible light irradiation. The other synthesized nanomaterials were also studied for their photocatalytic performance at the optimized conditions obtained by NiSb2O6 nano-photocatalyst. To prepare 45 ppm MG dye solution, 11.2 mg of MG powder was dissolved in 250 mL of deionized water. The pH value of the obtained solution was 4. In a typical photocatalytic experiment, certain amount (g) of the as-synthesized NiSb2O6 photocatalyt was added to 70 mL of the prepared MG aqueous solution and sonicated for 10 min in a dark room to establish an adsorption/desorption equilibrium between MG molecules and the surface of the photocatalyst. Afterwards, certain volume (mL) of H2O2 was added into the mixture solution, followed by further magnetic stirring under solar light condition. When the designed time (min) was elapsed, the solution was drawn out and the photocatalyst was separated by centrifugation in order to measure the absorption spectra of MG and calculate the MG concentration using UV-Vis spectrophotometry. The mixture was kept at a constant stirring of 300 rpm at the temperature of the experiment. The photodegradation yield (%) of MG was calculated by the following formula:
where, A0 and At represent the initial absorbance of MG at 610 nm and the absorbance at time t, respectively.
Experimental design for achieving optimal conditions in MG degradation process
Researchers utilize two different approaches to obtain the optimal conditions in chemical reactions, namely one-at-a-time and experimental design methods. Recently, the experimental design method is receiving more attention. In the method, different experiment factors affect each other. This is not considered in the one-at-a time approach. Full factorial design is one of the basic designs. In this design, all possible combinations of the factors and their settings are simultaneously considered. Assume that there is k investigating variables and each variable could be set to m distinctive levels. The number of possible combinations of the factors and their settings will then be mk. In chemical systems, three levels of the factor setting is common because such designs allow the determination of all main effects and all interaction effects with small number of experiment .
Full factorial design is one of the most powerful design tools in which three levels of each factor are used to design a set of proposed runs. In this design, the experimental points are embedded at the center (central points) and on the midpoints of the edges. In full factorial method, the relation between the factors and response is theoretically modeled which causes the reproducibility of the results. So, it is possible for experiments to elucidate the results. Response surface methodology (RSM) is a mathematical and statistical method analyzing experimental design by applying an empirical model.
Design of experiments (DOE) is a targeted and planned method for finding the relationships between independent variables, their overall impact on dependent variables and obtaining maximum information from the minimum possible experiments. The Response surface methodology (RSM) using input data, offers the graphical relationship between responses and variables, and performs multiple regression analysis [27,28].
The adequacy of the applied model is checked using analysis of variance (ANOVA) which needs some replicate experiments.
The statistical test of ANOVA analyzes the variances and examines the significance of the factors and their interactions on themselves and other factors. Then, with the help of the RSM, the validated model is plotted in three dimensions and interpreted to find the best conditions for the process. ANOVA of regression parameters for the quadratic model was computed in Table 4. The Fisher’s F test in the ANOVA analysis was performed to compare either model variance or factors with residual (error) variance, where the larger F-values and the smaller P-values indicate the more significant terms of the model [27, 28]. This ratio is called an F-distribution (F-value), varying from 1 to larger values. Values far from 1, exceeding from the tabulated F-value, provide evidence against the null hypothesis, indicating the significance of the regression parts of the fitted models. Equivalently, the null hypothesis is rejected when p-value is less than a significant level. In order to obtain the significant and reliable model at 95% confidence level, the p-values for the fitted model and its corresponding terms should be smaller than 0.05. The p-value of the present regression was smaller than 0.05, showing that the model was significant at a high confidence level (95%). A further assessment of the fitted model can be carried out using the lack-of-fit test. Via this statistical test, the residual part is sub-divided into pure error and lack-of-fit. In other words, it distinguishes the random error from the systematic one, causing the lack of fitting of the model with specific order. Therefore, at the 95% confidence level, the p-values for the lack-of-fit should be greater than 0.05, which is not significant. As shown in Table 4, the outcomes of ANOVA are completely in agreement with the above statements.
Also the coefficient of determination (the R-square, adjusted–R-square) was used to express the quality of fit of polynomial model equation. In this case, R2 of variation fitting for Y% = 97 indicated a high degree of correlation between the response and the independent factors (R2 = 0.9821). Also, the high value of adjusted regression coefficient (R2-adj = 0.9643) indicated high significance of the proposed model. This means that, the difference between experimental and the predicted responses is negligible. Also the predicted R-squared value (0.8803) was reasonable. It indicates the high accuracy and reliability of the developed mode to determine the response value shown in Fig. 5.
Fig. 5 represents a plot of the predicted versus the experimental degradation efficiency. This figure shows a good agreement between the predicted and experimental degradation efficiency (R2 = 0.9821) and represents the adequacy and significance of the model. Also, Fig. 5 indicates the normal plots of the predicted versus the observed response for the degradation efficiency. As it is evident in this figure, the data points obtained consistently appear on a straight trend line, demonstrating that there is no obvious dispersal. Dispersal of residuals is also shown in Fig. 5 (a-d).
The observed data of the factorial design was fitted to a quadratic response model. Prior to the analysis, low and high factor levels were coded to -1 and +1, respectively. Equation 1 shows the relation between the factors and the yield of the reaction, Y%, based on the first order model:
Y%=15.07191+15.29258×A+1346.28964×B+1.96845×C+562.5×A×B-22.5×B×C-28.24838×A2-5996.05179×B2-0.010522×C2 (Equation 1)
Fig. 6 shows the two dimensional plots of degradation yield associated with the desirability plots for catalyst amount versus H2O2 volume factors at the constant 50 min photocatalytic reaction time. The data show that the degradation yield and desirability is high when the catalyst amount and H2O2 volume are 30 mg and 0.04 mL, respectively. In this case, the desirability is 0.93 and degradation yield is 97 %.
To illustrate the effect in the above model, the three–dimensional response surface plots of the response are shown in Fig. 7. To show the effects of the three factors on the photodegradation process, the response surface methodology (RSM) was used. Fig. 7 (a-c) represents the 3D plots related to the interaction of AC, AB and BC, respectively. The semi-curvatures of these plots indicate the interaction between the variables. In other words, at a certain reaction time, when H2O2 and catalyst amounts increase, dye removal percentage improves. This means that, the mass transfer of dye molecule enhances on the surface of the catalyst and the dye adsorption process on the catalyst reaches equilibrium state quickly. Also, by increasing the catalyst amount, further surface area of adsorbent is available for dyes molecules leads to enhance the dyes removal percentage.
Fig. 8 (a and b) illustrate the photocatalytic performance of the as-synthesized nanomaterials. Fig. 8a, presents MG degradation spectra by S0. The data confirms the high performance of the sample to degrade MG by the mentioned photocatalytic conditions. Fig. 8 b shows a comparison study of the catalytic performance among the as-synthesized nanomaterials. As could be seen from the data, a non-considerable decrease in the catalytic activity is found when the dopant ions are introduced into the crystal system.
Effect of different parameters on the photocatalytic degradation
The optimal conditions were obtained by design expert software for S0. It was found that the optimum condition was 0.04 mL H2O2, 30 mg catalyst, and 35 min reaction time. The volume and concentration of the as prepared MG solution were 70 mL and 45 ppm, respectively, for obtaining the optimum conditions.
Also, we studied the degradation of blank dye solution (without catalyst and H2O2) at the optimized condition under the visible light irradiation. It was found that the dye degradation was negligible. Besides, we investigated the degradation process at the optimized condition at the dark room. In the conditions, the degradation yield was nearly zero.
Proposed mechanism for photocatalytic degradation
Converting visible light energy to chemical energy in the photocatalysis reaction by NiSb2O6 could be similar to previously reported and extensively studied mechanism; in a way that electrons (e-) and holes (h+) could be excited under visible light irradiation to the conduction and the valence band edge, respectively. These photo-excited e- and h+ then can transfer to the surface of the photocatalyst (NiSb2O6 particles), where they react with oxidants and reductants, respectively, or recombine in the absence of e- and h+ traps. The recombination of e- and h+ could be greatly minimized in the presence of H2O2, which traps the e- and h+ to form •OH and •O2- species . The excited e- reacts with H2O2 to form HO• and OH- groups. The formed OH- group reacts with h+ and forms HO• group. Besides, H2O2 can trap the photoexcited species in another way. It reacts with h+ and forms HO2• radical and H+. The produced HO2• is decomposed to •O2- and H+. H2O as the solvent of the photodegradation reaction can react with h+ to form HO• and H+ species. The produced H+ ion reacts with the dissolved O2 and 2e- and forms the initial H2O2. The dissolved O2 can also react with e- to form •O2-. The so-formed •OH or •O2- species are used for the decomposition of organic contamination molecules such as MG to intermediates or mineralized products through oxidation reactions [35-37].
At high amounts of H2O2 volume, the photodegradation yield maybe decreased. It is because the produced highly reactive hydroxyl radical (eq. 11) may react with excess amount of H2O2 and produces hydroperoxyl radical (HO2•) (eq.12) which is less reactive and ultimately inhibits the degradation with producing O2 and H2O in (eq.14). The reaction mechanism is explained below [38, 39]:
To show the merit of the present work, we have compared NiSb2O6 nanocatalyst results with some of the previously reported catalysts for the degradation of MG (Table 5). It is clear that NiSb2O6 showed greater activity than some other heterogeneous catalysts. Besides, the data reveal that the high yield of the photodegradation was achieved by the application of visible light condition in this work.
The novelty of the present work was doping new dopant ions into NiSb2O6 crystal system by a novel method, introducing a new application of the material as a high performance nano-photocatalyst to remove water pollutant dye under visible light irradiation, and presenting several analyses to study the physical properties of the obtained materials. The present work studied the synthesis of highly crystalline NiSb2O6 nanomaterials via one-step facile solid state method. Some lanthanide ions such as Eu, Gd, Ho and Yb were doped into NiSb2O6 crystal system. The XRD patterns indicated that NiSb2O6 was crystallized well in orthorhombic crystal system under the present solid state condition. The morphology of the obtained materials was sponge. FESEM images revealed that the particle sizes of the doped materials were smaller than those of pure NiSb2O6. The photocatalytic data indicated that the obtained materials had excellent efficiency for the removal of MG from aqueous solution. It was found that the optimum condition was 0.04 mL H2O2, 30 mg catalyst and 35 min reaction time. It was found that the catalytic performance was excellent when the pH value was in the range of 4 to 10, the dye concentration was up to 60 ppm and the dye volume was up to 90 mL.
CONFLICTS OF INTEREST
The authors do not have any personal or financial conflicts of interest.