Ultrafiltration treatment of wastewater contained heavy metals complexed with palygorskite

Abstract Palygorskite was applied in complexation-ultrafiltration treatment of heavy metals in wastewater under different pH and ionic strength. The results indicated that the rejection of heavy metals increased significantly with pH value, and decreased slightly with an increase of ionic strength of Na+ and Cl–. A certain concentration of NaCl significantly reduced the rejection rate of Cu2+. The rejection of Cu2+, Zn2+ and Cd2+ could reach over 86.8%, 93.6% and 93.7% at pH of 7 and 0.1 mol/L NaCl. The rejection of heavy metals was severely affected by low molecular weight competing complexing agents and the effect of sodium tartrate was greater than triethanolamine. In the presence of sodium tartrate, the rejection of Cu2+, Zn2+ and Cd2+ could arrive over 81.4%, 57.6% and 60.5% at pH of 7 in 20 min. Palygorskite was offered a potential complexing agent for the removal of heavy metals in wastewater at the complexation-ultrafiltration process.


INTRODUCTION
Over the past decade, researchers have increasingly focused on so-called the modifi ed technique of micellar enhanced ultrafi ltration (MEUF) and it is found that by adding an appropriate surfactant in solution, the effi ciency of ultrafi ltration can be further improved 1, 2 . At present, MEUF is seen as one of the potential techniques for the surfactant-based removal of heavy metal ions from wastewater 3, 4 . It is especially important to select the appropriate surfactant, which depend s on the nature of the fomite to be eliminated in MEUF process. Surfactant molecules rejoin each other, toward critical micelle concentration (CMC) 5, 6 . These micelles are highly ionic (anionic) on which metal ion (cationic) adsorbs, therefore, true solution of heavy metals ions is converted to colloidal solution 7, 8 . The permeate passes through membrane pores along with the traces of surfactant monomers and non-solubilized metal ions. This colloidal-sized metal complex is big enough in size to retain on ultra membranes of suitable molecular weight cutoff (MWCO) 9-11 . The treatment of wastewater containing kinds of metal ions such as Co 2+ , Pb 2+ , Cu 2+ , Cd 2+ , Zn 2+ , Ni 2+ and Hg 2+ has been successfully implemented by complexation-ultrafi ltration process 12-14 . The natural character of a water-soluble polymer directly determines the operating conditions and the effi ciency of combination with target heavy metal ions, and its selection is of signifi cance in complexation-ultrafi ltration process 15, 16 . A copolymer with abundant carboxyl functional groups, the copolymer of acrylic acid and maleic acid (PMA-100), can be used as a complexing agent in the complexation-ultrafi ltration process but has not been deeply studied as water-soluble polymer in counterpart 17, 18 . Many studies also have reported that some living biomaterials have the potential to adsorb heavy metal ions from the environment, for instance, fungus, bacteria and algae 19,20 . Nevertheless, organic materials can easily cause exogenous pollution in wastewater. It is crucial for wastewater treatment to develop inorganic materials, for example, palygorskite, which increases the effi ciency of complexing in complexation-ultrafi ltration process 21, 22 . Palygorskite is an aquiferous chain-layer magnesium-aluminum silicate, which has a special structure and characters. According to the mineralogical classifi cation, it belongs to the sepiolite of the group. The ideal structure formula is [(OH 2 ) 4 (Mg,Al, Fe) 5 (OH) · 2Si 8 O 20 ] · 4H 2 O with Mg preferentially located in octahedral sites, but some of Mg 2+ cations have been already replaced by Al 3+ and Fe 3+ cations. The palygorskite is famous for their ion--exchange properties, but palygorskite has advanced to incorporate the formation of nanoscale crystals with precisely tailored physical-chemical properties recently. Hence, the potential effectiveness was improved signifi cantly in an assortment of biomedical, chemical and environmental applications 23- 25 .
The threat of a high concentration of toxic heavy metal to human health and the water environment is diffi cult to estimate. It is proceeding more effi ciently, the nanoparticles for complexation with heavy metal ions have to retain following specifi c characteristics: (1) chemical and mechanical stability, (2) the high affi nity toward the target metal ions and the low affi nity toward nontarget ions 26 , (3) low toxicity, (4) low cost and the possibility of regeneration. The key process is the nanoparticle separation, which could be achieved by low-pressure membrane processes in a manner resembling polymer enhanced ultrafi ltration (PEUF) 27-30 . The membrane characteristics like MWCO, physical-chemical properties (material type, hydrophilicity and hydrophobicity, pore size, charge, etc.), and fouling resistance, as well as energy consumption, play an important role in the nanoparticle-enhanced ultrafi ltration (NEUF) 31-33 .
The current study concentrated on elucidating the relative abilities of nanoscale materials to bind metal ions, and subsequently, to be removed from water by ultrafi ltration 34-36 . In this study, separation of micelles metal complex was investigated using ultrafi ltration membranes with MWCO ranging from 6 to 100 kDa. Effects of pH, ionic strength and competitive complexing agent in the treatment of model wastewater containing Cd 2+ , Cu 2+ and Zn 2+ by heavy metals with palygorskite nanomaterials on complexation-ultrafi ltration were investigated. The binding capacity of palygorskite dendrimers was evaluated at several experimental conditions. The discussion reveals some light on the fundamental physical-chemical infl uencing factors governing nanoparticle-enhanced fi ltration and its potential for practical application.

Materials
All the chemicals were of maximum purity and used as received without any further purifi cation. Palygorskite is provided by Lanzhou Kunlun environmental protection & Technology Co., Ltd., from Tianshui of Gansu province. It is broken, crushed and grinded, and then 200 mesh sieves. Palygorskite rock was from Tianshui, Gansu province, China. The constitution of palygorskite is supplied by the specifi cation of the product. The main composition were SiO 2 63.47%, MgO 20%, Al 2 O 3 18.01%, Fe 2 O 3 5.16% and K 2 O 3.64%. The micrographs obtained the form of palygorskite bundles from scanning electron microscopy (SEM) (Fig. 1). These individual fi bers often form long bundles that are 5 to 10 μm long and fi bers shown as individual well separated fi bers or as elongated bundles of many fi bers aggregation. It contains impurity minerals such as palygorskite, quartz and montmorillonite. The preparation of simulated heavy metal wastewater is called proper CdCl 2 , CuCl 2 , ZnCl 2 adding a certain amount of water, and the concentration of Cd 2+ , Cu 2+ , Zn 2+ is 10, 40 and 30 mg/L, respectively.

Apparatus
Th e ultrafi ltration experiment fo r laboratory scale was carried out with the apparatus shown in Fig. 2 and considerable mass transfer is achievable which membrane module is operated with refl ux 37 . The me mbrane module Table 1. Main parameters of hollow fi ber ultrafi ltration membrane   Table 1 38, 39 . Key ultrafi ltration process consists of the co ncentrate tank with feed solution, membrane module and permeate tank, which has been connected with tubing and pressure control valves, manometer, a fl ow meter with a digital panel meter/display, a peristaltic pump, a stirrer for mixing and a power supply.

Experimental procedure
I n this work, the amount of complexing agent u sed was 5 g/L of palygorskite with an overall concentration at a pH of 7. The concentrate tank immitted 30 L initial feed which was circulated through the apparatus. Operative pressure of 0.1 mpa and fl ow rate of 25 L/h were controlled in the whole process. The thermostatic water batch was utilized to keep the temperature at 293 K. I n the total recirculation phase, both the permeation and retentate stream were returned to the concentrate tank so that the concentration of feed was constant. The sample of permeate was provided for analysis. The permeate fl ux was measured by weighing the per-meate volume produced in a certain quantity of time and the concentration of the permeation was analyzed by atomic absorption spectroscopy.

Statistical analysis
Membrane fl ux is the seepage volume of unit membrane area per unit time under certain temperature and pressure. Membrane fl ux is a parameter that represents the permeability of ultrafi ltration membrane, and is related to factors such as pore size, internal structure and viscosity of slurry. The membrane fl ux was measured by measuring the quality of osmotic fl uid per unit time. The quality of osmotic fl uid was measured by electronic balance, and the sampling time was determined by the stopwatch. The calculation formula is as follows: ( 1) where: J V -ultrafi ltration membrane permeation fl ux, L/m 2 •h; V -The volume of permeation fl uid, L; S -ultrafi ltration membrane area, m 2 ; t -running time, h.
The rejection rate R, which measures the membrane's separating capabilities is defi ned as: ( 2) where: c 0 -the concentration of solute upstream of the membrane. c p -the concentration of solute downstream from the membrane.
The membrane specifi c fl ux (defi ned as the ratio of the pure water fl ux of the cleaned membranes (J c ) to that of the virgin membrane (J 0 )).
(3) S tatistical analysis was performed based on STATIS-TICA. The data were analyzed through one-way analysis of variance (ANOVA) to determine the effect of membrane molecular weight cutoff and rejection coeffi cients. Besides, the Duncan's multiple comparison tests were performed to determine the statistical signifi cance of the differences in different membrane molecular weight cutoff and permeate fl ux. The F is the ratio of the mean square between groups and means square within groups. The larger the ratio is, the greater the difference between groups is. The P is the calculated confi dence interval of test statistic F. The a, b, c and d indicate whether there is a signifi cant difference between groups.

Eff ect of pH on rejection coeffi cients of heavy metal
R ejection coeffi cient of Cu 2+ increases strongly with the increase of pH (Fig. 3). At 30 min, it was found that the rejection of Cu 2+ was signifi cantly different in the fi ve kinds of pH (F 4,10 = 178.72, p < 0.001). The rejection coeffi cient of several pH were 61.9%, 69.3%, 83.3%, 87.5%, and 90.0%, respectively. There was no signifi cant difference in the rejection coeffi cients of Cu 2+ between pH 6 and pH 7. T hen the testing showed that the rejection of Cu 2+ was signifi cantly different in the fi ve kinds of pH at 60 min (F 4,10 = 26.50, p < 0.001). The rejection coeffi cient of fi ve pH were 70.8%, 78.4%, 90.1%, 93.4%, and 94.9%, respectively. As well as, there was no signifi cant difference in the rejection coeffi cients of Cu 2+ among pH 5, pH 6 and pH 7. At 90 min, the rejection coeffi cient of several pH were 76.4%, 85.1%, 93.0%, 96.2%, and 97.2%, respectively. It was indicated that the rejection of Cu 2+ was signifi cantly different in the fi ve kinds of pH (F 4,10 = 48.28, p < 0.001). There was no signifi cant difference in the rejection coeffi cients of Cu 2+ between pH 6 and pH 7. The results are the same as pH 5 and pH 6. At 120 min, it was concluded that the rejection of Cu 2+ was signifi cantly different in the 5 kinds of pH (F 4,10 = 35.74, p < 0.001). The rejection coeffi cient of fi ve pH were 78.6%, 87.3%, 94.1%, 97.9%, and 98.9%, respectively. There was no signifi cant difference in the rejection coeffi cients of Cu 2+ between pH 6 and pH 7. There was no signifi cant difference in Cu 2+ rejection between pH 5 and pH 6. At 150 min, the rejection coeffi cient of fi ve pH were 80.5%, 89.0%, 94.2%, 98.0%, and 99.4%, respectively. It was seen that the rejection of Cu 2+ was signifi cantly different in the fi ve kinds of pH (F 4,10 = 28.37, p < 0.001). Coincidentally, the rejection coeffi cient of Zn 2+ also increases strongly with the increase of pH (Fig. 4). At 30 min, it was indicated that the rejection of Zn 2+ was significantly different in the five kinds of pH (F 4,10 = 157.57, p < 0.001). The rejection coeffi cient of fi ve pH were 68.2%, 77.0%, 92.1%, 95.7%, and 96.0%, respectively. There was no signifi cant difference in the rejection coeffi cients of Zn 2+ between pH 6 and pH 7. Then at 60 min, it was found that the rejection of Zn 2+ was signifi cantly different in the fi ve kinds of pH (F 4,10 = 60.43, p < 0.001). The rejection coeffi cient of fi ve pH were 74.7%, 84.3%, 95.1%, 97.5%, and 97.8%, respectively. There was no signifi cant difference in the rejection coeffi cients of Zn 2+ among pH 5, pH 6 and pH 7. At 90 min, it was concluded that the rejection of Zn 2+ was signifi cantly different in the fi ve kinds of pH (F 4,10 = 29.09, p < 0.001). The rejection coeffi cient of fi ve pH were 79.1%, 88.9%, 95.7%, 97.6%, and 98.0%, respectively. There was no signifi cant difference in the pH (F 4,10 = 8.73, p < 0.01). The rejection coeffi cient of fi ve pH were 87.8%, 91.5%, 95.0%, 97.6% and 98.7%, respectively. There was no signifi cant difference in the rejection coeffi cients of Cd 2+ among pH 5, pH 6 and pH 7. At 120 min, it was indicated that the rejection of Cd 2+ was signifi cantly different in the fi ve kinds of pH, (F 4,10 = 39.07, p < 0.001). The rejection coeffi cient of fi ve pH were 90.7%, 93.8%, 96.6%, 98.4%, and 99.1%, respectively. There was no signifi cant difference in the rejection coeffi cients of Cd 2+ between pH 6 and pH 7. At 150 min, it was found that the rejection of Cd 2+ was signifi cantly different in the fi ve kinds of pH (F 4,10 = 4.73, p < 0.05). The rejection coeffi cient of fi ve pH were 91.3%, 94.2%, 97.0%, 98.5%, and 99.1%, respectively. There was no signifi cant difference in the rejection coeffi cients of Cd 2+ among pH 4, pH 5, pH 6 and pH 7.
In conclusion, variance analysis and multiple comparisons of the heavy metal concentrations of Cu 2+ , Zn 2+ , and Cd 2+ in fi ve pH were performed at the operating time of 120, 90, and 120 min, respectively. There were signifi cantly different in the concentration of heavy metal Cu 2+ in the permeation between different pH (F 4,10 = 4058.71, p < 0.001). There were signifi cantly different in the concentration of heavy metal Zn 2+ in permeate between different pH (F 4,10 = 335.63, p < 0.001). There were signifi cantly different in the concentration of heavy metal Cd 2+ in the permeation between different pH (F 4,10 = 56.24, p < 0.001). Under fi ve pH conditions, the stable permeate concentration of heavy metals Cu 2+ , Zn 2+ and Cd 2+ increased with decreasing pH (Fig. 6).
In palygorskite and heavy metal aqueous solution systems, because the pH of the surface of the palygorskite was higher than the pH value of the heavy metal solution, it was preferred that hydrolysis and precipitation occurred fi rst on the surface of the palygorskite. In addition, the heavy metal ion hydroxide colloidal particles had a positive charge, and the surface of the palygorskite was negatively charged. The action of positive and negative colloidal particles promotes the adhesion of the colloidal particles on the surface of the palygorskite. Therefore, the hydrolytic precipitation of colloidal silica and the interaction of colloidal particles on palygorskite were the main mechanisms for removing heavy metals from palygorskite. The increase of pH value in the aqueous solution system was benefi cial to the hydrolytic precipitation of heavy metals and the mutual adhesion of colloidal particles. rejection coeffi cients of Zn 2+ among pH 5, pH 6 and pH 7. In the following, it was shown that the rejection of Zn 2+ was signifi cantly different in the fi ve kinds of pH, at 120 min (F 4,10 = 146.48, p < 0.001). The rejection coeffi cient of fi ve pH were 82.3%, 92.9%, 96.3%, 97.6% and 98.0%, respectively. There was no signifi cant difference in the rejection coeffi cients of Zn 2+ among pH 5, pH 6 and pH 7. At 150 min, it was seen that the rejection of Zn 2+ was signifi cantly different in the fi ve kinds of pH (F 4,10 = 17.23, p < 0.001). The rejection coeffi cient of fi ve pH were 83.4%, 93.8%, 96.4%, 97.7% and 98.0%, respectively. There was no signifi cant difference in the rejection coeffi cients of Zn 2+ among pH 4, pH 5, pH 6 and pH 7.  Fina lly, the rejection coeffi cient of Cd 2+ also increases strongly with the increase of pH (Fig. 5). It was concluded that the rejection of Cd 2+ was signifi cantly different in the fi ve kinds of pH, at 30 min (F 4,10 = 57.69, p < 0.001). The rejection coeffi cient of fi ve pH were 76.7%, 83.0%, 89.3%, 93.1%, and 95.6%, respectively. There was no signifi cant difference in the rejection coeffi cients of Cd 2+ between pH 6 and pH 7. At 60 min, it was seen that the rejection of Cd 2+ was signifi cantly different in the fi ve kinds of pH (F 4,10 = 19.46, p < 0.001). The rejection coeffi cient of fi ve pH were 83.8%, 88.9%, 93.6%, 96.6%, and 98.1%, respectively. There was no signifi cant difference in the rejection coeffi cients of Cd 2+ between pH 6 and pH 7. At 90 min, it was shown that the rejection of Cd 2+ was signifi cantly different in the fi ve kinds of

Eff ect of ionic strength on rejection coeffi cients of heavy metal
The effect of ionic strength on the rejection coeffi cient was investigated at different times (30, 60, 90, 120, and 150 min) while the other conditions such as pH, metal initial concentration, pore membrane and transmembrane pressure were constant during the test.
Rejection coeffi cient of Cu 2+ increases slightly with the increase of the concentrations of NaCl (Fig. 7). At 30 min, it was concluded that the rejection of Cu 2+ was signifi cantly different in the different concentrations of NaCl (F 2,6 = 4.08, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl was 89.9%, 88.4%, and 86.8%, respectively. Then, it was found that the rejection of Cu 2+ was signifi cantly different in the different concentrations of NaCl at 60 min (F 2,6 = 8.92, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl were 94.9%, 93.5% and 92.0%, respectively. At 90 min, it was shown that the rejection of Cu 2+ was signifi cantly different in the different concentrations of NaCl (F 2,6 = 7.76, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl was 97.1%, 94.5% and 95.3%, respectively. It was found that the rejection of Cu 2+ was signifi cantly different in the different concentrations of NaCl at 120 min (F 2,6 = 16.42, p < 0.01). The rejection coeffi cient of three different concentrations of NaCl was 98.9%, 98.7% and 97.6%, respectively. It was indicated that the rejection of Cu 2+ was signifi cantly different in the different concentrations of NaCl at 150 min (F 2,6 = 9.68, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl were 99.4%, 99.0% and 98.0%, respectively. At fi ve time points, there was no difference in Cu 2+ interception rate between 0.01 moL/L sodium chloride and 0.1 moL/L sodium chloride rejection. different concentrations of NaCl at 60 min (F 2,6 = 0.86, p > 0.05). The rejection coeffi cient of three different concentrations of NaCl was 97.8%, 97.4% and 96.9%, respectively. At 90 min, it was shown that the rejection of Zn 2+ was signifi cantly different in the different concentrations of NaCl (F 2,6 = 2.56, p > 0.05). The rejection coeffi cient of three different concentrations of NaCl were 98.0%, 97.6% and 97.0%, respectively. Then, it was found that the rejection of Zn 2+ was signifi cantly different in the different concentrations of NaCl at 120 min (F 2,6 = 9.83, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl were 98.1%, 97.6% and 97.0%, respectively. It was concluded that the rejection of Zn 2+ was signifi cantly different in the different concentrations of NaCl at 150 min (F 2,6 = 5.09, p > 0.05). The rejection coeffi cient of three different concentrations of NaCl were 98.0%, 97.6% and 97.0%, respectively.  Coincidentally, the rejection coeffi cient of Zn 2+ also increases slightly with the increase of the concentrations of NaCl (Fig. 8). It was indicated that the rejection of Zn 2+ was signifi cantly different in the different concentrations of NaCl at 30 min (F 2,6 = 9.08, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl were 96.0%, 95.0% and 93.7%, respectively. It was seen that the rejection of Zn 2+ was signifi cantly different in the At last, the rejection coeffi cient of Cd 2+ also increases slightly with the increase of the concentrations of NaCl (Fig. 9). It was concluded that the rejection of Cd 2+ was signifi cantly different in the different concentrations of NaCl at 30 min (F 2,6 = 5.37, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl were 95.6%, 95.1% and 93.7%, respectively. It was indicated that the rejection of Cd 2+ was signifi cantly different in the different concentrations of NaCl at 60 min (F 2,6 = 1.57, p > 0.05). The rejection coeffi cient of three different concentrations of NaCl were 98.1%, 97.6% and 96.9%, respectively. It was seen that the rejection of Cd 2+ was signifi cantly different in the different concentrations of NaCl at 90 min (F 2,6 = 2.48, p > 0.05). The rejection coeffi cient of three different concentrations of NaCl was 98.7%, 98.3% and 97.7%, respectively. It was found that the rejection of Cd 2+ was signifi cantly different in the different concentrations of NaCl at 120 min (F 2,6 = 10.80, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl was 99.1%, 98.6% and 98.0%, respectively. It was shown that the rejection of Cd 2+ was signifi cantly different in the different concentrations of NaCl, at 150 min (F 2,6 = 6.19, p < 0.05). The rejection coeffi cient of three different concentrations of NaCl was 99.1%, 98.6% and 98.0%, respectively. At 30 and 120 min, there was no signifi cant difference in the rejection coeffi cients of Cd 2+ between 0.01 and 0.1 mol/L 120 min. The rejection coeffi cient increased signifi cantly in 20 min. The rejection coeffi cient of three competitive complexing agents was 87.84%, 84.40% and 81.40%, respectively. The differences between the 140 and 20 min rejection coeffi cient for Cu 2+ at different competitive complexing agents were 11.01%, 13.31% and 15.31%, respectively. The retention coeffi cient of three competitive complexing agents was stable at 98.85%, 97.99% and 96.91%, respectively, and the Cu 2+ concentration in the permeate was stable at 0.46, 0.80 and 1.24 mg/L, respectively.  Conclusions there were signifi cantly different in the concentration of heavy metal Cu 2+ in the permeation between different concentrations of ionic strength (F 2,6 = 220.17, p < 0.001). There were signifi cantly different in the concentration of heavy metal Zn 2+ in permeate between different concentrations of ionic strength (F 2,6 = 155.43, p < 0.001). There were signifi cantly different in the concentration of heavy metal Cd 2+ in the permeation between different concentrations of ionic strength (F 2,6 = 252.26, p < 0.001). Under there different concentrations of ionic strength, the stable permeate concentration of heavy metals Cu 2+ , Zn 2+ and Cd 2+ decreased with decreasing ionic strength (Fig. 10). The ionic strength will change the thickness of the double electric layer on the surface of the complexing agent and promote the agglomeration of the complexing agent in the form of particles; it will also compete with the heavy metal ions for ion exchange and affect the complexation effect.

Eff ect of competitive complexing agent on rejection coeffi cients of heavy metal
Rejection coeffi cient of Cu 2+ increased slightly with the adding of sodium tartaric and triethanolamine (Fig. 11). The permeate concentration of Cu 2+ decreased with complexation time, and it changes little after about Rejection coeffi cient of Zn 2+ increased strongly with the adding of sodium tartaric and triethanolamine (Fig. 12). The permeate concentration of Zn 2+ decreased with complexation time, and it changes little after about 100 min. The rejection coeffi cient increased signifi cantly in 20 min. The rejection coeffi cient of three competitive complexing agents was 93.21%, 89.31% and 57.62%, respectively. The differences between the 100 and 20 min rejection coeffi cient for Zn 2+ at different competitive complexing agents were 4.82%, 5.38% and 33.66%, respectively. The retention coeffi cient of three competitive complexing agents was stable at 98.03%, 94.69% and 91.28%, respectively, and the Zn 2+ concentration in the permeate was stable at 0.59, 1.59 and 2.62 mg/L, respectively.
Coincidentally, the rejection coeffi cient of Cd 2+ increased strongly with the adding of sodium tartaric and triethanolamine (Fig. 13). The permeate concentration of sodium chloride. For the rejection coeffi cients, there was no signifi cant difference in the rejection of Cd 2+ between the concentrations of NaCl at the 60 and 90 min.
Jawor et al. evaluate the removal of cadmium ions from water by nanoparticle-enhanced ultrafi ltration using polymer and zeolite n anoparticles, when the ionic strength is relatively low, the rejection coeffi cients increase almost linearly with the ionic strength concentration (Jawor and Hoek 2010). Canizares et al. examined the treatment of metal plating wastewater by MEUF attaining complete removal of Cd 2+ and Pb 2+ using PAA as different pH. It can be seen lead rejection is greater than cadmium for the majority of pH values due to the higher affi nity of lead for PAA previously reported ( Cañizares et al. 2008). M oreover, the similar results of ultrafi ltration treatment of wastewater contained heavy metals have been reported in Table. 2. The palygorskite could improve the rejection of heavy metals by comparing with other studies and signifi cantly offer potential material for ultrafi ltration treatment of wastewater.

CONCLUSIONS
With the increase of pH value, the rejection of three heavy metals increased signifi cantly. In the pH range of 3-5, the pH value increased by one unit, the Cu 2+ rejection rate increased 8.91% on average, and the pH value was 6-7. The Cu 2+ rejection rate did not increase signifi cantly; the pH value ranged from 3 to 4, the pH increased by one unit, and the rejection of heavy metal Zn 2+ and Cd 2+ increased by 9.42% and 4.40% on average. At pH values of 5-7, the pH value increased, and the interception rates of heavy metals Zn 2+ and Cd 2+ increased slightly. As the concentration of NaCl increased, the rejection of three heavy metals decreased Cd 2+ decreased with complexation time, and it changes little after about 120 min. The rejection coeffi cient increases signifi cantly in 20 min. The rejection coeffi cient of three competitive complexing agents was 92.78%, 86.79% and 60.51%, respectively. The differences between the 120 and 20 min rejection coeffi cient for Cd 2+ at different competitive complexing agents were 6.34%, 9.98% and 34.24%, respectively. The retention coeffi cient of three competitive complexing agents was stable at 99.12%, 96.77% and 94.75%, respectively, and the Cd 2+ concentration in the permeate was stable at 0.09, 0.32 and 0.53 mg/L, respectively.  Analysis of variance and multiple comparisons of heavy metal concentrations in the wastewater of heavy metals Cu 2+ , Zn 2+ and Cd 2+ under different competing complexing agents were performed, at the system operating time was 120, 90 and 120 min, respectively. It was shown that there was a signifi cant difference in the concentration of heavy metal Cu 2+ in the permeate between different competing complexing agents (F 2,6 = 425.89, p < 0.001). There was a signifi cant difference in the concentration of heavy metal Zn 2+ in the permeation between different competing complexing agents (F 2,6 = 68.96, p < 0.001). There was a signifi cant difference in the concentration of heavy metal Cd 2+ in the permeation between different competing complexing agents (F 2,6 = 624.39, p < 0.001). The wastewater concentrations of the three heavy metal permeates varied similarly with competing complexing agents. The order of the concentration of the wastewater was sodium tartrate>triethanolamine>control (Fig. 14).
slightly. 0.01 moL/L NaCl had no signifi cant effect on the interception rate of three heavy metals, 0.1 moL/L NaCl signifi cantly reduced the retention rate of heavy metal Cu 2+ , and had little effect on the rejection of Zn 2+ and Cd 2+ . Low-molecular-weight competitive complexing agents have serious effects on the rejection rate of three heavy metals. The effect of sodium tartrate on three heavy metals is greater than that of triethanolamine, and the effect on Zn 2+ and Cd 2+ is much greater than that of Cu 2+ . The effect of triethanolamine on Zn 2+ and Cd 2+ is much greater than that of Cu 2+ . Under the condition of stable concentrations of three heavy metal permeates, Cu 2+ , Zn 2+ , and Cd 2+ decreased by 0.86%, 3.36%, and 2.35%, respectively, in the presence of triethanolamine. Under the conditions of sodium tartrate, Cu 2+ , Zn 2+ , and Cd 2+ decreased by 5.07%, 6.87%, and 4.37%, respectively. The order of the impact of competing complexing agents on ions was Zn 2+ > Cd 2+ > Cu 2+ . Compared to other complexing agents, the rejection coeffi cient of palygorskite decreased because the organic complexing agent had almost no rejection coeffi cient for Cu 2+ , Zn 2+ and Cd 2+ . However, considering the simple fabrication and easy handle of palygorskite, they are very promising to be used as a high-performance complexing agent for heavy metal removal.