Invent of a simultaneous adsorption and separation process based on dynamic membrane for treatment Zn(II), Ni(II) and, Co(II) industrial wastewater

2.1 Material

Chitosan (average M_w200,000), polyvinyl alcohol (average M_w72,000) from Sigma–Aldrich and Acetic acid (glacial) 100% was obtained from Merck. Sargassum glaucescens was collected from the Oman Sea on the coast of Chabahar, Iran. Nonwoven polyester fabric was purchased from Baftineh, Iran. Bovine bones were prepared and cleaned of meat. Industrial wastewater was collected from shahrekord industrial park, Iran. The wastewater details are given in Table 1.

2.2 Preparation of dynamic particle and DM support

According to Fig. 1a SGN was washed twice with water. It was dried at 80 °C for 20 h in the oven and crushed with a high-energy planetary ball mill (Fara Pajouhesh, Isfahan FP2, Iran) at 600 rpm for 2 h. The bovine bones were boiled in water four times at 120 °C for 3 $\mathrm{h}$ , 4 times and all the fat of bovine bones was removed. The bones were washed and crushed in the range of 2–5 $\mathrm{m}\mathrm{m}$ . Bovine bone was calcined to 850 °Cfor 6 $\mathrm{h}$ in the furnace, and then hydroxyapatite was crushed to the particle size range of 10–20 $\mu \mathrm{m}$ using a kitchen mill (Pars Khazar, GR-1.2.3P, Iran) for 20 $\mathrm{m}\mathrm{i}\mathrm{n}$ .

Close

Fig. 1

Schematic of DM preparation steps (a), Schematic of the DM filtration process and DM cake after 60 min (b), the perspective drawing of membrane module (c) and, strategy of experiments and methodology (d).

Export to PPT

Close

Fig. 1

Schematic of DM preparation steps (a), Schematic of the DM filtration process and DM cake after 60 min (b), the perspective drawing of membrane module (c) and, strategy of experiments and methodology (d).

Export to PPT

The Cs (5 $\mathrm{w}\mathrm{t}.\%$ ) and PVA (10 wt%) solution were dissolved in acetic acid (2%, v\v) and were stirred for 18 h at 45 °C. Then, Cs and PVA were blended at a ratio of 7:3 on the magnetic heater stirrer for 24 $\mathrm{h}$ at 45 °C. The polyester web was installed on the collector at 50 °C The Cs/PVA solution was electrospanned on the polyester web at a high DC voltage of 20 KV, with a tip-collector distance of 17 $\mathrm{c}\mathrm{m}$ and a flow rate of 0.6 $\mathrm{m}\mathrm{l}{\mathrm{h}}^{-1}$ . In order to heat-cross-linking, the Cs/PVA nanofiber membrane was placed in the oven at 110 °C for 10 $\mathrm{h}$ .

2.3 Design process

A plug reactor was designed from a plastic pipe with a length of 16 $\mathrm{m}$ and an internal diameter of 12 mm. This long pipe was wrapped around a cylinder with a diameter of 100 mm. The input of the plug reactor was connected to a wastewater tank with a pump (Water Pump, 12 Volts). According to Fig. 1b, the effluent was mixed with the adsorbent (SGN and hydroxyapatite powder) in a circular flow inside the plug reactor.

According to Fig. 1c the membrane module made of two polyethylene pieces. The DM support (Cs/PVA nanofiber @ polyester) was installed between polyethylene pieces (DM support diameter = 20 mm). Four small vibrators (vibration DC, 5 Volt, 1500 round per minute, Taiwan) were installed in the piece. This system was equipped with an air compressor to fluid flowing through the DM (Positive pressure TMP). A valve was installed at the down of the polyethylene tube for rapid evacuation.

2.4 Measurement and methods

The heavy metal ions concentration was determined with atomic absorption spectrometry (Varian Atomic Absorption 240, USA). Turbidity was measured with a turbidity meter (ET266020, Lovibond Corporation, Germany). Nanofiber and nanoparticles morphology were studied using scanning electron microscope (FEI Quanta 200 SEM, USA). Functional groups of DM elements were determined with FT-IR spectra in the range of 400–4000 ${\mathrm{c}\mathrm{m}}^{-1}$ (Nicolet 8700, Thermo Fisher Scientific, Japan). The structure of hydroxyapatite was analyzed using X-ray diffraction (Philips PW3040). The specific surface area of dynamic membrane was determined using the Brunauer–Emmert–Teller method (Costech International Sorptometer 1042, Italy). In order to determine the stability of Cs/PVA nanofiber membranes, deswelling experiments were done on the platform shaker. The pieces of Cs/PVA nanofiber membrane ( $10mm\times 10mm$ ) were immersed in 50 $\mathrm{m}\mathrm{L}$ of deionized water at 25–45 °C for 24 h Degree of deswelling was calculated from Eq. (1):

2.5 Adsorption and separation experiments

2000 $\mathrm{m}\mathrm{l}$ of wastewater including Co(II), Zn(II) and Ni(II) ions was pumped in the plug reactor. The wastewater and adsorbents contacted during a circular flow in the plug reactor; Co(II), Zn(II) and Ni(II) ions were adsorbed onto the hydroxyapatite powder and SGN. This mixture (wastewater and adsorbents) was introduced into the tube of DM module, in several cycles (Each charge = 400 $\mathrm{m}\mathrm{l}$ ). The hydroxyapatite powder and SGN were remained on the DM support and the DM cake was formed.

Fig. 1d shows the strategy of experiments and methodology. Experiments were done in three Sections: i) Adsorption: optimizing parameters affecting in the adsorption process in the plug reactor: The effect of pH 4–9 at SGN = 6 $\mathrm{g}$ , hydroxyapatite powder = 4 $\mathrm{g}$ and, Cs/PVA nanofiber@ polyester web = 10 $\mathrm{m}\mathrm{g}$ with retention time of 40 min. The effect of adsorbent mass SGN = 2–10 $\mathrm{g}$ , hydroxyapatite powder = 1–5 $\mathrm{g}$ (Fig. S4 c) at optimized pH and retention time = 40 min. The effect of retention time 10–120 min at optimized adsorbent mass and pH. Experiments were performed at constant room temperature 28 °C to make the process cost-effective. This section was independent of the next two section.

ii) Separation: optimizing mass of SGN and hydroxyapatite powder and, interval between two vibration shocks at constant TMP (0.2 $\mathrm{b}\mathrm{a}\mathrm{r}$ ) with standard response surface methodology (RSM) design called CCD at the Design-Expert 11 software. Table S1 shows the ranges as well as coded and un-coded levels of the variables considered in this study.

iii) Optimization of TMP: experiment of filtration at different TMP (Fig. 3a) at optimized conditions.

Close

Fig. 2

Characterization of DM: The SEM image and diameter distribution of Cs/PVA nanofiber membranes (a), SGN (b) and hydroxyapatite powder (c), FT-IR spectra of Cs/PVA nanofiber membranes, SGN and hydroxyapatite (d) and deswelling test of nanofiber membranes (e), X-ray diffraction profile of hydroxyapatite (f).

Export to PPT

Close

Fig. 3

TMP profiles (a), Instantaneous membrane flux profiles (b), TMP effect on the removal efficiency of turbidity (e) and DM resistance profiles (f).

Export to PPT

In the output of plug reactor, the removal efficiency of heavy metal ions (R%) and adsorption capacity ( ${\mathrm{q}}_{\mathrm{e}}$ ) were calculated using Eq. (2) and Eq. (3), respectively:

2.6 DM resistance and fouling index

In the cake filtration model, membrane fouling led to the deposit of foulant material on the external surface of the membrane, which increased resistance (Choi et al., 2009) DM resistance was calculated using Eq. (5) (Tong et al., 2011):

Harouna et al. (Mahamadou Harouna et al., 2019) developed the first order derivative equation by finite difference methods and conducted a direct comparison of the membrane fouling under different conditions. The fouling index ( $m^{-1}$ ) was calculated using Eq. (6) (Xiong et al., 2016):

3.1 Characterization of DM

The SEM of Cs/PVA in Fig. 2a showed that electrospanning produced a homogeneous and microporous nanofiber. The nanofibers have an average diameter and surface area of 95.5 $\mathrm{n}\mathrm{m}$ and 172.46 ${\mathrm{m}}^2{\mathrm{g}}^{-1}$ , respectively (Fig. S1 a). The Cs/PVA nanofiber membrane put in the oven at 110 °C for 10 h to cross-linking and nanofiber membrane stability in the aqueous solution was increased. Cross-linking leads to a decrease in polar $\mathrm{O}-\mathrm{H}$ groups (Fig. S2), and water resistance increases (Tian et al., 2019). Cross-linked and non-cross-linked membrane stability was compared in Fig. 2e. For non-cross-linked membranes in the range of 10–20 $\mathrm{h}$ at 25–45 °C the degree of deswelling reached 60 $\%$ , while under this condition, deswelling degree of cross-linked membrane decreased by only 20%. According to previous results (Esmaeili and Beni, 2014), the high stability degree of the Cs/PVA nanofiber membranes in the aqueous medium with 350 $\%$ of the average deswelling degree indicates that the heat cross-linked is a more suitable method than cross linking by glutaraldehyde vapor with 93 $\%$ of average deswelling degree (Aliabadi et al., 2014).

The SGN was obtained with average size of 50–120 $\mathrm{n}\mathrm{m}$ (Fig. 2b). In previous work (Akbar Esmaeili and Aghababai Beni, 2015), SGN were produced within 1 $\mathrm{h}$ in the planetary ball mill with range of 30–300 nm and a specific surface area of 11.25 ${\mathrm{m}}^2{\mathrm{g}}^{-1}$ , but in this work, planetary ball mill worked for 2 $\mathrm{h}$ ; the specific surface area of SGN with BET analysis was obtained as 12.99 ${\mathrm{m}}^2{\mathrm{g}}^{-1}$ (Fig. S1 b) and according to Fig. 2b, this nanoparticle average diameter was observed 95.6 nm. Also, the SEM of hydroxyapatite powder in Fig. 2c shown, the particle diameter is about 10–20 $\mu \mathrm{m}$ . There are two reasons that SGN are nano-size but hydroxyapatite particles are not nano-size: SGN precipitate in the effluent after a short time; but nano-hydroxyapatite is very stable in the aqueous phase and does not precipitate easily; also, to create proper porosity in the dynamic membrane, the presence of a micrometer-sized particle is necessary to prevent membrane fouling.

According to FT-IR results in Fig. 2d, on the hydroxyapatite spectrum, free $\mathrm{O}-\mathrm{H}$ bands are indicative at 3580 ${\mathrm{c}\mathrm{m}}^{-1}$ . The absorption bands from 947 to 1080 ${\mathrm{c}\mathrm{m}}^{-1}$ were assigned to the phosphate esters stretching $\mathrm{R}-\mathrm{P}$ with two strong bands at 965 and 1048 ${\mathrm{c}\mathrm{m}}^{-1}$ , while Googerdchian et al. (Googerdchian et al., 2018) reported that the phosphate is visible from 460 to 1100 ${\mathrm{c}\mathrm{m}}^{-1}$ . In the Cs/PVA, SGN and hydroxyapatite spectrums, the vibrational band of from 2900 ${\mathrm{c}\mathrm{m}}^{-1}$ was attributed to $\mathrm{C}-\mathrm{H}$ stretching of the alkyl groups. The broad and strong vibration around 3200 to 3400 ${\mathrm{c}\mathrm{m}}^{-1}$ is indicative of $\mathrm{O}-\mathrm{H}$ stretching in H-bonded type of vibration and the strong peak at 1390 ${\mathrm{c}\mathrm{m}}^{-1}$ was assigned to the ${\mathrm{C}\mathrm{H}}_2$ and ${\mathrm{C}\mathrm{H}}_3$ stretching vibrations (Fan et al., 2019). The SGN for broad band at 3400 cm⁻¹ was assigned to O—H stretching vibrations and other at 2900 cm⁻¹ was C—H stretching. The stretching vibrations of carboxylate O—C—O related to 2889 cm⁻¹. The C⚌O signal related to 1566 cm⁻¹ SGN and hydroxyapatite spectrums, respectively. The absorption 415 cm⁻¹ related to C—OH vibration and O—C—O (carboxylate group) stretching vibration for SGN. The bands measured at 1311, 1080 and 1029 cm⁻¹ related to C—C—H and O—C—H, to C—O stretching vibrations. The bands C—O and C—C vibrations of rings. The band 946 cm⁻¹ in SGN indicated of bond presence by the C—O stretching vibration hydroxyapatite spectrums. XRD pattern of hydroxyapatite was shown in Fig. 2f. It can be seen that the prepared hydroxyapatite belongs to the hexagonal structural type of hydroxyapatite with space group ${\mathrm{P}6}_3/\mathrm{m}$ . According to Fig. S1 c, the water contact angle test of Cs/PVA nanofiber @ polyester web and DM (Cake layer) after separation process shown DM is hydrophilic.

3.2 Adsorption studies in the plug reactor

Adsorption of heavy metal ions onto adsorbents involves a combination of various mechanisms such as electrostatic adsorption, complexation, ion exchange, covalent forces, Van der Waals forces and surface adsorption (Esmaeili and Aghababai Beni, 2018). The pH is very important in the adsorption process with effects on the adsorbents-ion reactions (Beni, 2021). According to Fig. S3a, at pH = 6 adsorption efficiency was more than 95%, while at pH 3 and 9 the adsorption efficiency decreases. When the pH of the neutral metal ion solution was alkaline, the ions were converted to insoluble hydroxide (Googerdchian et al., 2012).

The stability of the membrane support layer was evaluated at acid and alkaline conditions. The pieces of Cs/PVA nanofiber@ polyester web (DM support) were contacted with a 200 ml wastewater sample for 40 min at pH 4, 6 and 8. According to Fig. S3b, at pH = 6 the stability of Cs/PVA nanofiber@ polyester was highest; these results showed that most heavy metal ions are adsorbed by SGN and hydroxyapatite (DM cake) compared to CS/PVA nanofibers. In the previous study, the optimum pH for the removal of Ni(II) and Co(II) by SGN and Cs/PVA nanofiber (Esmaeili and Aghababai Beni, 2018) at pH = 6 was optimized. Googerdchian et al. (Googerdchian et al., 2018) reported pH = 3 is suitable for the Pb(II) adsorption with nano-hydroxyapatite. At low pH the concentration of ${\mathrm{H}}^{+}$ ions is high and causes protons to bind to functional groups in the adsorbent active sites and to interfere with the binding of metal ions and adsorbents (Beni and Esmaeili, 2020). The adsorbent mass and composition ratio were tested according to the pattern in Fig. S3c. The results showed that with increasing the adsorbent mass: the removal efficiency of heavy metal ions increased but the adsorbent capacity decreased. Previous results showed the SGN capacity for Ni(II) and Co(II) adsorption is about 7–9 and 2–3 $\mathrm{m}\mathrm{g}{\mathrm{g}}^{-1}$ , respectively (Esmaeili and Aghababai Beni, 2015). Googerdchian et al. (Googerdchian et al., 2018) synthesized nanohydroxyapatite from bovine bone for Pb(II) adsorption with 180 $\mathrm{m}\mathrm{g}{\mathrm{L}}^{-1}$ of concentration and adsorption capacity reported 200 $\mathrm{m}\mathrm{g}{\mathrm{g}}^{-1}$ . Therefore, by using 6 $\mathrm{g}$ of SGN and 5 $\mathrm{g}$ hydroxyapatite powder, more than 90% of the heavy metal ions were removed. The SGN and hydroxyapatite are widely used in the wastewater treatment due to their low cost, high adsorption capacity and high removal efficiency (Beni and Esmaeili, 2020).

As shown in Fig. S3d, the adsorption process was very fast. At 10 $\mathrm{m}\mathrm{i}\mathrm{n}$ of retention time, the removal efficiency was about 60%, at 30 and 90 $\mathrm{m}\mathrm{i}\mathrm{n}$ the removal efficiency was 94% and 98%, respectively. The system was received to pseudo-equilibrium after 40 $\mathrm{m}\mathrm{i}\mathrm{n}$ .

3.3 Separation process in the DM module

3.4 Effect of vibration on DM performance

Experimental condition was set in optimal state, DM filtration process was performed for 28 $\mathrm{m}\mathrm{i}\mathrm{n}$ without vibration and DM resistance was calculated. When the vibrator turned off, according to Fig. 4a-b, the instantaneous membrane flux was decreased, resistance increased and removal efficiency of turbidity was improved because DM was made with dense layers. Roebuck and Tremblay (Roebuck and Tremblay, 2016) synthesized highly porous 3D network of mono-dispersed particles without compressed; so, the membrane resistance was not increased. Turn on vibrator prevented DM layer condensation and reduced fouling index from 1.040 to 0.798 $m^{-1}$ . The cross-flow can be used to remove the attached sludge from the surface of DM support (Liu et al., 2009), whereas, in this research, the DM backwashed after 56 $\mathrm{m}\mathrm{i}\mathrm{n}$ filtration with 500 ml water at pressure of 12 bar for 30 $\mathrm{s}$ (Gasemloo et al., 2019). The vibrator placed DM particles in the best orientation; the large DM particles were placed close to the support layer (Fig. 4c). In vibrator “off” state, DM support layer lost its quality, after backwashing because nanoparticles were imprisoned between polyester base and PVA/Cs nanofibers. It should be noted that a continuous activation of the vibrator was inappropriate because it prolongs the formation of DM layer, also the removal efficiency of turbidity reduced severely, so according to the optimization results, Fig. S5 selected 5 $\mathrm{s}$ “on” and 5 $\mathrm{s}$ “off” pattern by programmable switch for the vibrator.

Close

Fig. 4

Schematic diagrams of DM filtration in vibrator “on” and “off” mode (a), the effect of turning off the vibrator on DM resistance (b) and removal efficiency of turbidity (c).

Export to PPT