Removal of phosphate with a polyacrylonitrile composite functionalized by a metal organic framework-enhanced layered double hydroxide

Phosphorus (P) is an essential mineral element for many different organisms, but it causes eutrophication problems when present in excessive amounts in surface water bodies. Further degradation of water quality and the death of aquatic life follow the unrestrained algal bloom and aquatic plant growth caused by excessive enrichment with minerals and nutrients, particularly P and nitrogen. P, mainly in the form of phosphate, enters natural water bodies through agricultural runoff, sewage discharge, and industrial wastewater, such as from slaughterhouses (Bunce et al. 2018). Thus, phosphates are regarded as pollutants whose concentrations need to be diminished by water treatment.

Technologies for phosphate removal include biological wastewater treatment processes using microorganisms or algae-based systems, chemical precipitation (e.g., struvite), and physicochemical methods such as adsorption, ion exchange, and membrane filtration. The most common treatment option for phosphate removal is chemical precipitation, which is reliable and controllable but requires the addition of large-quantity chemicals and the post-treatment of sludge (Cheng et al. 2009). The adsorption technique is also prevalent because it can treat waste streams with low concentrations at high efficiency and is relatively inexpensive and environmentally friendly (sludge-free) (Shraddha et al. 2018).

Numerous adsorbents have been studied and employed for phosphate treatment, such as zeolites, silica materials, metal oxides and hydroxides, clay minerals, activated carbon and biochar, and polymers (Bacelo et al. 2020). The use of layered double hydroxides (LDHs) for anion adsorption has received increasing attention in recent years due to their anion exchangeability. LDHs consist of metal ions, hydroxyl groups, and certain intercalated anions with a common formula of [M³⁺_xM²⁺_1−x(OH)₂]Aⁿ⁻_x/n·yH₂O, where M³⁺ and M²⁺ represent the metal cations, Aⁿ⁻ is the anion, and x is located in the range of 0.18 − 0.33 (Iftekhar et al. 2018). LDHs have been studied as adsorbents for the treatment of phosphates in wastewater, either in their pristine form or after being modified and combined with other compounds (Mandel et al. 2013; Yan et al. 2018). At the same time, another emerging functional material, the metal organic framework (MOF), has been developed as an advanced solid adsorbent for phosphate treatment. Researchers have applied MOFs based on iron (Nehra et al. 2019), zirconium (Lin et al. 2015), zeolitic imidazolate frameworks (ZIFs) (Shams et al. 2016), and many other compounds for the adsorption of phosphates. Additionally, designs of MOF/layered double hydroxide (LDH) composites, whether MOF derived from LDH or vice versa, have been utilized for separation purposes. A partially converted LDH-ZIF-8 membrane has also been used for gas separation (Liu et al. 2015) and Mg–Al-LDH-supported Cu-MOF has been fabricated for the removal of methyl orange dye (Chakraborty & Acharya 2018). Based on findings from MOF/LDH studies, the combination of LDH and MOF might have great promise for adsorption applications.

To the best of our knowledge, few reported studies have investigated the adsorption of phosphate by MOF/LDH composite materials. The hypothesis that MOF/LDH composite is effective for the adsorption of phosphates led to the present study, which investigated MOF/LDH integration and phosphate adsorption. Furthermore, the usability of the adsorbent in practical applications was considered because the particle sizes of the aforementioned MOF/LDH materials were determined to lie between 0.2 and 1 μm (Liu et al. 2015; Chakraborty & Acharya 2018). Such small particles lead to an extremely high-pressure drop in columns and carryover in stirred tanks, which makes their use not feasible. To ensure the usability of the adsorbent in practical applications, a polymer-based composite was prepared using polyacrylonitrile (PAN) as a binding polymer with an as-synthesized adsorbent powder. PAN was selected for its porous character and durability, which make it suitable to function as a carrier of the adsorbent powder. Moreover, the form of the polymer composite is favorable for large-scale adsorption processes. The composite has low-pressure drop and is stable at typical application temperatures (15–35 °C). Further, PAN is known to be chemically stable and has good mechanical durability. The gelation of the PAN-based composite (MOF/LDH/PAN) is accomplished in water after heating at a mild temperature, which demands less chemical and energy input (Xu et al. 2018).

The chemicals utilized in the present study were obtained from Sigma-Aldrich and used without further purification. The chemicals that were used included sodium hydroxide (NaOH), hydrochloric acid (HCl), nitric acid (HNO₃), trimesic acid (benzene-1,3,5-tricarboxylic acid, C₆H₃(CO₂H)₃, H₃BTC), ferrous chloride tetrahydrate (FeCl₂·4H₂O), dimethylformamide (DMF), methanol (CH₃OH), aluminum nitrate nonahydrate (Al(NO₃)₃·9H₂O), zinc nitrate hexahydrate (Zn(NO₃)₂·6H₂O), Tween 80, PAN (150,000 MW, (C₃H₃N)_n), and monopotassium phosphate (KH₂PO₄).

The synthesis procedure of MIL-100(Fe) was adapted from Guesh et al. (2017) with modifications (Guesh et al. 2017). First, 1.676 g H₃BTC was dissolved in a 22.8 mL NaOH (1 M) solution (Solution 1) and 2.26 g ferrous chloride tetrahydrate was dissolved in 97.2 mL water (Solution 2). Solution 1 was then added dropwise into Solution 2 while stirring. The mixture was stirred for 24 h at room temperature. Centrifugation was applied to separate the solid from the liquid. The collected solid was further purified using DMF and methanol successively for 24 h. The washed compound was then dried overnight in a vacuum oven at 80 °C.

The as-synthesized MIL-100(Fe) (500 mg) was mixed with aqueous solutions of 0.5 M Al(NO₃)₃·9H₂O (8.5 mL) and 1 M Zn(NO₃)₂·6H₂O (8.5 mL). The mixture was then sonicated for 30 min. NaOH (1 M) solution was dripped into the mixture until the pH became 10. The resultant substance was aged overnight at 60 °C while undergoing magnetic stirring (about 18 h). The product was then centrifuged and washed thoroughly with water. Finally, the product was freeze-dried for 48 h.

The method for the synthesis of MOF/LDH/PAN composite was adapted from previously published studies (Xu et al. 2018) and amended according to the needs of the present study. First, 3.6 g MOF/LDH was mixed with 42 mL DMF and 1 mL Tween 80 by stirring at 60 °C for 2 h. Then, 2.4 g PAN powder was gradually added. Heating and stirring continued for another 2 h. The mixture was pumped into deionized water through a 0.7 mm needle while being stirred. The gelled composites were then left to age in water for 24 h. Finally, the composites were rinsed with more than 2 L warm water. The synthesized products were then stored hydrated in water without drying.

Various techniques were utilized for the characterization of the as-synthesized products, MOF/LDH powder, and MOF/LDH/PAN composite. Phase identification was done via X-ray diffraction (XRD, PANalytical X-ray diffractometer) with Co Kα irradiation at λ = 1.79 Å, and 2θ of 5°–80°. Surface functional groups were identified by Fourier transform infrared spectroscopy (FTIR, Bruker Vertex 70 model) in a spectra range of 4,000−400 cm⁻¹. The morphology was imaged using scanning electron microscopy (SEM). The applied SEM devices were the JEOL JSM-7900F and Hitachi SU3500. A cross-section sample of the composite was prepared via razor blade cutting in a liquid nitrogen environment. Transmission electron microscopy (TEM) was conducted using a Hitachi HT-7700. The specific surface area and pore size distribution were determined via N₂ adsorption–desorption isotherm analysis (Tristar^® II Plus). The particle size of the MOF/LDH powder was investigated with a Malvern Zetasizer (Nano ZEN350).

The MOF/LDH/PAN composite was tested for moisture and ash contents. It was centrifuged at 3,000 rpm for 10 min to remove the excess water content. The moisture content was measured based on centrifuged composite samples dried overnight at 105 °C. For ash content determination, the standard test method for ash content in plastics, ASTM D5630-01, was used. The composite dimensions were analyzed using a Raman imaging microscope (Thermo Scientific DXR3xi). The chemical stability against acidic solutions was investigated by immersing the composite samples in HNO₃ solutions with various pH levels (1, 1.5, 2, 2.5, and 3) and shaking them mildly (50 rpm) for 5 days at room temperature. The weight losses of the samples were then recorded.

Batch adsorption equilibrium experiments were performed for the MIL-100(Fe), LDH, MOF/LDH powder, and MOF/LDH/PAN composite. The adsorption of phosphate on the inert PAN polymer was not studied. Sorption kinetics was studied with MOF/LDH and MOF/LDH/PAN to see the effect of the matrix and particle size on the diffusion rates. The synthetic solution (KH₂PO₄) used for adsorption was controlled to have a background ionic strength (I) of 20 mM through the addition of 1 M NaCl solution. The batch studies were conducted at 15, 25, and 35 °C for 48 h, with intermediate handshaking and four-time pH adjustments to maintain a uniform mild acidic condition (pH 5.9−6.1) during the adsorption process. The kinetic studies were conducted in a 500 mL KH₂PO₄ solution with intense motor mixing at 800 rpm at 25 °C, with pH adjustment and small-volume sampling at pre-defined time intervals.

Desorption tests were performed for the MOF/LDH powder in batch experiments with NaOH solutions of various concentrations (0.1–2 M). The whole process was accomplished without drying or changing the container. Adsorption was first done in a 200 mL stock solution (approximately 50 mg P/L) for 0.1 g MOF/LDH powder for 24 h with mild rotation. The solid and liquid were then separated through centrifugation, decanting, and washing with water. NaOH solutions were added to regenerate the adsorbent and the volume of the regeneration solution was adjusted to 20 mL. Samples were collected after 24 h rotation. The desorption rate was calculated by dividing the mass of the desorbed P by the mass of the adsorbed P. The percentages of the released Zn, Al, and Fe were based on the mass of the adsorbent.

The concentrations of the tested solutions and samples were examined via inductively coupled plasma optical emission spectrometry (ICP-OES, Thermo iCAP 6300 series). Thus, all the concentrations in the present study are shown herein as the mass of P.

All the batch tests were conducted in triplicate. Each data point was plotted separately in figures instead of applying error bars because there is experimental uncertainty also in the equilibrium liquid phase concentration values. It improved the accuracy, especially in the nonlinear fitting of the isotherm plots and gave a better understanding of the quality of the data.

The as-synthesized MOF/LDH/PAN composite was used as the filling material in the column process to study its performance in phosphate recovery. The column had an inner diameter of 2.5 cm and was filled with the composite to pack a 10 cm bed. Thus, the bed volume (BV) was 49.09 cm³ , where BV refers to the volume of the inner-column space packed with the adsorbent bed. The whole process was conducted at room temperature (in this case, 25 °C). In the loading step, the influent solution (approximately 50 mg P/L) was fed into the column with the downward flow at a flow rate of 2 BV/h (1.636 mL/min) for 100 BVs. Effluent samples were taken for each BV. The regeneration was started immediately after the loading step; 0.1 M NaOH was used as the desorbing agent, with the upward flow at a flow rate of 5 BV/h (4.091 mL/min) for 30 BVs. One sample was taken every two bed volumes. The samples were then analyzed with ICP-OES.

The N₂ adsorption–desorption isotherm of the MOF/LDH was shown as Type IV with an H3 hysteresis loop, as can be seen in Figure S5. The pore size was 9.28 nm according to the pore size analysis using the Barrett, Joyner, and Halenda (BJH) method. As the BJH method underestimates the pore size by 20–30% for narrow mesopores with a < ∼10 nm diameter, the amended pore size was 11.6−13.3 nm (Thommes et al. 2015). The BET-specific surface area of the MOF/LDH was 136 m²/g. It is improved by 42.45% compared to that of the LDH, which was 95.47 m²/g. Obviously, this is due to the high specific surface area of the MIL-100(Fe) content of the MOF/LDH. The average diameter of the MOF/LDH particle samples was 0.566 μm.

The moisture content of the swollen composite sample was 77.60% (wet weight basis). The ash contents of the MOF/LDH powder and the MOF/LDH/PAN composite were 72.02 and 34.06% (dry weight basis), respectively. Thus, the dry weight-based MOF/LDH content of the MOF/LDH/PAN composite was 47.29% and the wet weight-based MOF/LDH content was 10.59%.

The mass losses of the composite samples treated with HNO₃ solutions in the durability test are shown in Table 1. At a pH of below 2, the composite lost 4.59−6.35% of its total mass. The mass loss was negligible at pH 2.5 and no mass loss was found at pH 3, showing that the composite is acid resistant and stable above pH 2.

The Freundlich model predicts an infinite uptake capacity for further C_e and results in an infinite slope (dq/dC) toward extremely diluted conditions. Thus, the Freundlich model cannot be considered a preferred model to explain the adsorption process, although it seems to fit the measured data well. The maximum Langmuir adsorption capacity was found to be 60.86 mg P/g. However, the Langmuir isotherm model is not suitable for the isotherm data of the MOF/LDH in this study because, as can be seen in Figure 4, it underestimates the uptake capacity at a higher concentration range.

Fitting of the Fowler–Guggenheim model gave a higher R² than for the Langmuir isotherm. It should be noted that there are three parameters in the Fowler–Guggenheim model but only two parameters in the Langmuir isotherm. The improvement in fit – both quantitatively (R²) and qualitatively (the shape of the calculated isotherm) – is so significant that introducing an additional adjustable parameter is justified. The Langmuir isotherm levels off too early, whereas the Fowler–Guggenheim isotherm correctly predicts monotonically increasing adsorption. It is therefore concluded that the interaction between adsorbate molecules is not a negligible factor. The negative value of χ (−5.38) indicated repulsive interactions between the adsorbed phosphate ions. The maximum adsorption capacity was 74.96 mg P/g.

Equations (4) and (5) were integrated numerically with initial conditions c_P = c_0,P and q_P(r) = 0 using a variable step: the use of a variable-order integrator (ode15s in Matlab). The diffusion coefficients of phosphate in the powder and composite beads and the mass fraction of LDH-MOF in the beads were estimated using a simplex-type direct search algorithm (fminsearch in Matlab).

The reasonably good agreement of the Fickian diffusion model with the PAN composite experiment data indicates that intraparticle diffusion is the rate-limiting factor in the adsorption process of the composite (Figure 6). A 2.21 × 10⁻¹² m²/s diffusion coefficient is in the range typical for porous adsorbents. However, in the case of the powder, the intraparticle diffusion model failed to explain the measured data. The best-fit value for the diffusion coefficient is 7.88 × 10⁻¹⁸ m²/s, which is unreasonably small and cannot be regarded as physically meaningful. The most plausible explanation for this is that the small particles (R_p = 0.28 × 10⁻⁶ m) have moved with the liquid in the stirred tank. This results in a low slip velocity at the phase boundary and, thus, high liquid film mass transfer resistance.

While the kinetics of phosphate sorption into the powder material (MOF/LDH) is higher than in the MOF/LDH/PAN composite, the foreseeable high-pressure drop in columns and carryover in stirred tanks make the use of the powder in large-scale applications not feasible. On the other hand, the polymer composite (MOF/LDH/PAN) containing functional adsorbent can be packed in a column for a large volume to achieve a high column capacity.

The desorption of phosphate from the MOF/LDH powder was done using NaOH aqueous solutions with various concentrations in batch experiments. The results are shown in Table 3. The release of Fe was negligible, which indicates that the MOF component of the composite is stable at least until 1.0 mol/L of NaOH. Release of Al and Zn was not significant until 0.2 mol/L of NaOH, indicating that the LDH component was stable at low NaOH concentrations. Although the desorption rate of P reached 71.30% with 0.5 M NaOH, the severe Zn release at and above 0.5 mol/L NaOH was a decisive factor in rejecting the use of high concentrations of NaOH. The desorption rate of P under 0.1 NaOH had a minimal difference from that under 0.2 M NaOH. To retain the adsorbent material and ensure that it would have good stability in cyclic separation, 0.1 M NaOH was chosen as the regeneration media in the later column tests.

The cumulative concentration of P in the effluent reached 1.99 mg P/L at the breakthrough point of 38.0 BVs. More detailed data are shown in Figures S7–9. At the breakthrough point, the uptake of P in the fixed bed was 18.92 mg P/g, and the removal efficiency from the fed 1.86 L phosphate solution with a 55.51 mg P/L concentration was 96.42%. As adsorption on the MOF/LDH/PAN composite is clearly mass transfer limited, even better performance is expected when the synthesis procedure is modified to produce smaller particles.

As shown in Table 4, the column adsorption performance of the current composite is very good compared to that of the other materials used in previous studies. The studies listed in Table 4 have shown large differences in feed concentration, flow rates, and not all their data are directly comparable. The higher the feed concentration, the earlier the bed is saturated and the earlier the breakthrough is expected to occur. Similarly, the higher the flow rate (in units of BV/h), the higher the mass transfer resistance and the earlier the breakthrough.

The Al₂O₃/calcium alginate composite material (Ai et al. 2022) was studied with a flow rate and particle size comparable to those in the present study. The amount of P removed per unit mass of adsorbent at breakthrough point was eight times larger with the MOF/LDH/PAN composite than with the Al₂O₃/calcium alginate. In another study with a comparable particle size (Jung et al. 2017), modified biochar calcium alginate beads were able to purify 25 BV of feed when the feed concentration was approximately one-third that in the present study. The flow rate, however, was higher than in the present study.

Among the materials that have been studied for application in the column adsorption of phosphate (Wu et al. 2007; Woumfo et al. 2015; Jung et al. 2017; Ramirez et al. 2018; Ye et al. 2019; Ai et al. 2022), the MOF/LDH/PAN composite in the present study possesses the highest total P uptake capacity (36.93 mg P/g).

The regeneration efficiency of the adsorbent determines the usability of the composite for practical applications. In the present study, desorption was conducted with the chosen desorbing agent, 0.1 M NaOH. As shown in Figure 7(b), the desorbing agent led to an asymmetrical desorption curve for the effluent P concentrations. In the beginning, the sharp peak that formed followed by a long-tail decline demonstrated that much of the desorption occurred during the first 6 BVs of regeneration. About 60% regeneration efficiency had already been reached with a 9.7 BV desorbing agent, but 70% efficiency consumed a 27.2 BV desorbing agent (Figure S10). Nuryadin and Imai reported the column regeneration of phosphate-saturated amorphous Zr/MgFe-LDH composite with 0.1 N NaOH as a desorbing agent, in which 91.70% desorption efficiency was achieved with 127.1 BV of desorbing agent, and 98.80% was finally achieved with 1,016.9 BV (8 h operation) (Nuryadin & Imai 2021).

Overall, the MOF/LDH/PAN composite is a usable and promising material in the column adsorption process of phosphate.

In the present study, an LDH-based adsorbent material was synthesized with an MIL-100(Fe) as a precursor. The MIL-100(Fe)-enhanced MOF/LDH adsorbent powder was then combined with PAN to form MOF/LDH/PAN composite beads working as a functional material. The composite, which is in the form of a polymer bead, is a favorable adsorbent form to use in adsorption processes such as adsorption columns. XRD and FTIR confirmed MIL-100(Fe) loading in the adsorbent material. SEM images clearly showed the porous structure of the composite beads and the well-dispersed adsorbent powder inside the composite. The phosphate adsorption behavior of the MOF/LDH powder was reasonably well described by the Fowler–Guggenheim isotherm model. The kinetic behavior of the composite was described by the Fickian intraparticle diffusion model. The diffusion coefficients are in the range typical for polymeric adsorbents, but a large particle size leads to high mass transfer resistance. The uptake capacity of the powder adsorbent reached 74.96 mg P/g, while the composite had a good phosphate removal ability but adsorbed less per unit mass due to the polymer structure. Nevertheless, the polymer composite form can be considered a practical solution for adsorption-based water treatment applications in tank and column processes. The composite-packed fixed bed (49.09 mL) treated 38.0 BVs (1.86 L) of the influent solution with 55.51 mg P/L phosphate. An 18.92 mg P/g uptake with 96.42% removal efficiency before the breakthrough point was achieved in the column adsorption study. The phosphate-loaded composite bed was regenerated with 0.1 M NaOH up to 70% efficiency within 30 BVs.

Overall, the MOF/LDH/PAN composite is potentially applicable to remove and/or recover phosphates in large-scale column adsorption processes where powder materials cannot be used due to high-pressure drops. However, the particle size of the composite must be reduced and the weight fraction of the active material (now 10–20%) must be increased by modifying the synthesis process. In addition, further studies should focus on simplifying the complicated synthesis procedures, reducing the use of, and increasing the percentage of the reuse of DMF and Tween 80, which would lead to a greener and more sustainable route.

This work was supported by European Regional Development Fund project ID A73961.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.