Selective Extraction of Terbium Using Functionalized Metal–Organic Framework-Based Solvent-Impregnated Mixed-Matrix Membranes

Abstract

Advancements in membrane separation techniques will expand the applications and requirements for highly specialized, inventive, efficient, and resistant separation materials. The selective separation of rare earth elements (REEs) is one of the expanding applications of membrane-based techniques, as their use is becoming more widespread. Membrane techniques are becoming increasingly desired as environmentally friendly, straightforward methods for treating wastewater and separating metals. For the separation of REEs, an innovative impregnated mixed-matrix membrane (IMMM) technique was developed in this study. It provides a selective, efficient, and reusable method that is suitable for industrial applications. Terbium was selectively adsorbed from other REEs using organophosphorus IMMM with a loading capacity of 113.2 mg/g in 3 h and was reused three times without destroying the initial membrane. Solvent impregnation is thought to offer specific chelation sites that are selective for terbium separation.

1. Introduction

The need for raw rare earth elements (REEs) has grown dramatically in recent years due to their great potential and crucial role in many contemporary industries. For example, praseodymium (Pr), neodymium (Nd), dysprosium (Dy), and terbium (Tb) are required to produce permanent magnets [1]. It is anticipated that the use of these four elements—Pr, Nd, Dy, and Tb—will increase in low-carbon technology, among other applications [2]. The demand for these elements is expected to spike in 2050, especially for electric motor and wind turbine applications [3]. When the above metals are incorporated into critical industrial applications, a high degree of purity is required. Several studies have been carried out to find an efficient method for REE separation and purification, in which the researchers focused on using non-liquid material membranes for REE separation [3]. Although separation techniques such as solvent extraction and ion exchange are convenient for industrial processes, several technological challenges exist, such as high contamination, the difficulty of separating solvents from low-concentration solutions, the extensive use of acid and solvents, the need to change to greener methods, etc. [4].

In recent research, porous materials such as activated carbon, silica gel, and zeolite were shown to have feasible commercial applications for wastewater treatment and rare earth extraction. However, their reusability is questionable [5]. Metal–organic frameworks (MOFs) are promising candidates constructed from both organic and inorganic parts, resulting in a highly porous and crystalline structure with a high surface area [6]. MOFs were applied as adsorbents for the separation of heavy metals and rare earths [7]. For example, a Cu-BTC MOF was applied for the separation of lanthanides under various pH conditions. Although this MOF showed no selectivity toward lanthanides, the adsorption capacity was high. Rare earths can also be selectively separated from other heavy metals [8]. Khalil et al. synthesized a Co-based MOF modified by diallylamine to improve the MOF’s selectivity. Although Ce exhibited some selectivity toward Eu, the MOF was not applied to a mixture of rare earths [9]. Zhang et al. synthesized a hybrid MOF from both ZIF-8 and UIO-66-NH₂ that showed an extraordinary adsorbent capacity for Nd, Gd, Eu, and Er but its selectivity remained insufficient [7]. Further attempts have been made to improve both selectivity and adsorption through MOF modification. Pei et al. synthesized MIL-101 modified by poly ionic liquids and found it to be efficient for the complete removal of La, Sm, and Nd ions, with a 99.8% removal rate but no selectivity [10]. However, MOF stability in aqueous acidic solutions is still a challenge, together with selectivity control [11]. To enhance MOF stability and recyclability in acidic aqueous solutions, membrane technology is adopted for REE separation. Hence, recent developments have been focused on using membrane technology for a wide spectrum of applications, especially individual separation and purification of REEs from wastes and ores. Hua et al. investigated nanofibrous membranes by incorporating synthesized UIO-66-(COOH)₂ into polyacrylonitrile nanofibers to separate Tb from Eu [12]. The membranes’ stability and large adsorption capacity were confirmed, but unfortunately, there was no selectivity toward Tb. In terms of increasing adsorption capacity, Yao et al. succeeded in adsorbing Gd ions onto a fabricated 2-hydroxyphosphono-acetic acid-UIO-66-polyester fabric membrane [13]. Regardless of these advancements, MOF membrane technology is still very limited and further research is urgently needed. The main problem of implementing REEs in industrial applications pertains to purity, since the high similarity in REEs’ ionic radii is the reason for the multi-element mixture existence even after applying advanced separation techniques. Thus, mixed-matrix membranes included in MOFs are a propitious technology based on combining both MOFs and polymer materials. By combining both, the membrane can be designed with appropriate permeability by adjusting the pore size, while using polymers would balance the membrane’s selectiveness with its permeation and improve its stability [14]. Zhang et al. designed an MMM MOF for the efficient removal of heavy metals from contaminated water [15]. It was synthesized by mixing a Eu-mtb MOF (cationic [Eu₇ (mtb)₅(H₂O)₁₆]·NO₃ 8DMA·18H₂O) in poly(vinylidene fluoride), which enables fast extraction of Cr₂O₇²⁻ from aqueous solutions with a high loading capacity. However, the published work on MOF-based mixed-matrix membranes is very limited. It is important to consider this technique while focusing on choosing the correct MOF filler to control pore size and the right polymer to balance permeation ability.

In this study, we established an innovative MOF-based mixed-matrix membrane impregnated with a solvent extractant named Cytop^®501 for the selective extraction of terbium from a REE mixture and heavy metal contaminations. Firstly, UiO-66 was successfully modified and incorporated into a polybenzimidazole polymer (PBI) to form the MMM. Then, Cytop^®501 was successfully impregnated to adjust the pore size for terbium selective extraction. This is an innovative methodology that can be extended for the individual separation of REEs in the future by adjusting the IMMM characteristics.

2. Experimental Section

2.1. Chemicals and Reagents

Zirconium chloride, terephthalic acid (TPA), trimellitic acid (TMA), diethylenetriamine (DETA), toluene, 36% hydrochloric acid, and dimethylformamide were purchased from Sigma-Aldrich (Tokyo, Japan), all with high purity. Additionally, Sigma-Aldrich, Japan provided all of the metal salts, terbium nitrate, dysprosium nitrate, europium nitrate, neodymium nitrate, yttrium nitrate, praseodymium nitrate, and gadolinium nitrate, all with high purity. Cytop^®501 (Solvay, Tokyo, Japan) (bis(2,4,4-trimethylpentyl)phosphinic acid) was provided by Solvay, Japan, with high purity. The polybenzimidazole polymer (PBI) was provided from Sato Light Industrial Co., LTD, Tokyo, Japan, while DMA (N,N-dimethylacetamide) was supplied by Fujifilm Wako Pure Chemical Corporation, Osaka, Japan.

2.2. Synthesis of Modified MOF

For comparison, UiO-66 was prepared by using a solvothermal method. A total of 0.4 gm of zirconium chloride, 0.28 gm of terephthalic acid, 0.2 mL of deionized water, and 50 mL of N,N-dimethylformamide were placed in a Teflon container and heated at 120 °C for 24 h using an autoclave. The product was separated by centrifugation (6000 rpm, 10 min), washed twice with DMF and ethanol, and then dried to obtain UiO-66. In contrast, UiO-66-COOH was prepared via the heating of ZrCl₄, TBA, and TMA as the linker with HCl and DMF in a Teflon beaker with a ratio of 1:1:2:76, respectively, setting the linker ratio to 25% TMA and 75% TBA for 24 h at 180 °C. Afterward, filtration was performed by decantation and washing with DMF and distilled water. The product was dried overnight in a vacuum at 80 °C. Subsequently, UiO-66-COOH-DETA was synthesized by refluxing 0.5 g UiO-66-COOH and 0.926 g DETA in 50 mL toluene for 16 h at 100 °C in an oil bath. The resultant product was collected by centrifuging and washed three times with ethanol before being dried at 80 °C overnight.

2.3. Preparation of UiO-66-COOH-DETA Impregnated with Cytop^®501

A total of 0.25 g UiO-66-COOH-DETA was soaked with 0.29 g of Cytop^®501 (Solvay, Tokyo, Japan) in 50 mL of methanol and dried in the oven at 60 °C until completely dried. The impregnation of Cytop^®501 was physically observed through the MOF’s color change to yellow. The resultant product was a dry powder with no precipitation or remnants of Cytop^®501.

2.4. Preparation of MOF-Based Mixed-Matrix Membrane

UiO-66 or UiO-66-COOH-DETA alone or impregnated with Cytop^®501 MOFs was incorporated separately with the PBI to form a mixed-matrix membrane. In total, 10–40% of the specified MOF was added to a solution of PBI in DMA and then allowed to dissolve completely through stirring overnight and applying an ultrasonic disperser. An extra DMA was added to the mixture for easy handling due to the high viscosity of the PBI. The solution was poured into a round Petri dish, and the fabrication of the mixed-matrix membrane was achieved through the stepwise evaporation of the solvent in the oven by raising the temperature to 100 °C for 4 h and then maintaining this temperature for another 14 h. The synthesized membrane was cut into circles to match the liquid membrane separation machine inlet diameter.

2.5. Characterization

Powder X-ray diffraction (XRD) analysis was performed using a MiniFlex 600 (Rigaku, Osaka, Japan) where all the samples were dried under vacuum at 80 °C overnight before any measurements were performed. Fourier transform infrared spectroscopy (FTIR) was conducted using an IRAffinity-1 (Shimadzu, Osaka, Japan) with a KBr pellet, crushing and forming a membrane as a reference. Pore size and surface area were measured by using BELSORP-Max (MicrotracBEL, Osaka, Japan) after the samples were heated under vacuum at 200 °C for 4 h prior to analysis under liquid nitrogen. A plasma atomic emission spectrometer ICPE-9820 (Shimadzu, Osaka, Japan) was used to detect the ion concentration in the feed and stripping solutions.

2.6. Adsorption Experiments

UiO-66, UiO-66-COOH, UiO-66-COOH-DETA, and UiO-66-COOH-DETA impregnated with Cytop^®501 membranes were tested with a 0.1 mM rare earth solution and/or mixed with a heavy metal solution. All the MOFs were set to 20% per membrane for the adsorption and desorption experiments. The aqueous feed solutions were prepared by dissolving metal salts in mixtures of 0.1 M HNO₃ and 0.1 M NH₄NO₃ solutions, and the pH was adjusted by mixing both solutions. The adsorption isotherm under various pH values was examined, in addition to the adsorption capacity, stripping ability and reusability. The adsorption experiments were conducted by using a liquid membrane separation machine in which a metal solution was placed. Samples were taken periodically and analyzed with ICP-AES after certain time intervals. The rate of adsorption can be calculated by dividing [M³⁺]_f by [M³⁺]₀, where [M³⁺]_f and [M³⁺]₀ are the concentration of the metal ions ion in the feed solution at each sampling interval and the initial feed concentration, respectively. Similarly, the stripping analysis was carried out after the extraction steps using diluted acid, in which samples were taken after certain time intervals and analyzed using ICP-AES to confirm the stripping of terbium. The adsorption kinetics were examined via both a pseudo-first-order dynamic model and a pseudo-second-order dynamic model with the following equations:

$$q_t=q_e\left(1-e^{-K_{1}t}\right)$$

$$t/q_t=t/q_e+1/K_2{q}_e^2$$

where q_t is the amount of terbium ions (mg/g) adsorbed at time t (h). q_e (mg/g) is the amount of terbium ions adsorbed at equilibrium. K₁ and K₂ are the equilibrium constants of the pseudo first-order and pseudo second-order dynamic equations, respectively.

3. Results and Discussion

3.1. Characteristics of MOF Adsorbents

Characterization of the adsorbents was performed by XRD and FT-IR to elucidate the modified synthesized MOFs. Figure 1a shows the XRD patterns for both the initial UiO-66 and the synthesized UiO-66-COOH-DETA. The XRD pattern is almost the same, with the only change being the intensity, indicating that the original UiO-66 structure was maintained after the functionalization of UiO-66 with COOH-DETA. These results agree with those of Ahmed et al. [16], who studied the functionalization of UiO-66 with COOH. Their paper elucidated the partial closure of the MOF’s pore size after functionalization with COOH. In this work, a similar phenomenon was observed after the functionalization of UiO-66 with COOH-DETA, as shown in Figure 2a.

Interestingly, in this study, the pore size was greater with UiO-66 than with UiO-66-COOH-DETA. UiO-66-COOH-DETA is non-porous but retains the crystal structure of UiO-66. Its non-porosity may be due to the strong interaction through hydrogen bonding resulting from modification with diethylenetriamine. Interestingly, pore size increased for the impregnated UiO-66-COOH-DETA-Cytop^®501. The pore size for UiO-66 was 4.19 nm, with a surface area of 1559 m²/g, and dropped to 0.99 nm when functionalized with COOH-DETA, with a surface area of 14.3 m²/g; it then increased to 10.3 nm when impregnated with Cytop^®501, with a surface area of 55.6 m²/g. The impregnation succeeded in opening the pores of UiO-66-COOH-DETA, which altered terbium selective separation in the adsorption experiments. SEM images of the synthesized UiO-66-COOH-DETA and after impregnation with Cytop^®501 revealed no change in the MOF’s crystal structure; however, pore size increased after impregnation, as shown in Figure 2b. Particle size was about 50 nm, and impregnation led to well-distributed open pores of the MOF. On the other hand, Figure 1b represents the successful modification of UiO-66 particles with COOH-DETA. Due to C=O, NH wagging vibration and intensity changed around 1568 cm⁻¹ and 759 cm⁻¹. The CN stretching peak at 1357 cm⁻¹ confirmed the successful functionalization of the UiO-66 surface. The new peak, around 3000 cm⁻¹, was assigned to the free NH groups after modification with DETA. The impregnation of Cytop^®501 was achieved, as explained in the Section 2. It was decided that this amount is sufficient to achieve better pore size and selectivity for terbium ion permeation. The synthesized MOFs and the membrane used for extraction are shown in Figure 3a–c.

3.2. REE Adsorption by the Prepared Membranes

Initially, the rare earth adsorption behaviors of a mixture of 0.1 mM of Dy³⁺, Eu³⁺, Gd³⁺, Nd³⁺, Pr³⁺, Tb³⁺, and Y³⁺ were examined. The mentioned rare earths were extracted using a liquid membrane separation machine with the synthesized membranes and the UiO-66 membrane as a reference. All the MOFs were fixed to 20% in the synthesized membranes. The adsorption rate depends mainly on membrane type, pH, and time. Figure 4a shows that terbium can be selectively adsorbed from other rare earths at pH 3.1 for 1 h by the UiO-66-COOH-DETA impregnated Cytop^®501 membrane, while the other membranes show no adsorption ability for any metals. While pH is a very important factor for the optimum adsorption, the experiments were carried out with pH values ranging from 3 to 6.5. However, some rare metals will precipitate after pH 5.5; thus, pH 4.5 was considered the optimum condition from our experimental results. Figure 4b presents the selective adsorption of terbium; about 94% was adsorbed from other metals at pH 4.5 for 5 h. We tested the effect of time intervals on the adsorption efficiency and found that after 5 h, the adsorption ratio does not increase much. In this regard, the optimum adsorption conditions are set to pH 4.5 and 5 h. These data prove that the mixed-matrix impregnated membrane can be promising for the selective separation of rare earths by controlling the pore size of the membrane. The selectivity of terbium adsorption by the UiO-66-COOH-DETA impregnated Cytop^®501 membrane can be attributed to the chelating ability of Cytop^®501 and the change in pore size due to Cytop^®501 impregnation, which is selective for terbium adsorption compared to that of accompanying rare earths. This is a unique result that was published for the first time and can be expanded for other rare earths in the future. Lee et al. [17] tried to functionalize Cr-MIL-101 with different groups such as ethylenediamine, diethylenetriamine, and N-(phosphonomethyl) iminodiacetic acid (PMIDA) for the separation and selectivity of rare earths. Their results confirmed the increase in Gd³⁺ ion separation compared to other rare earths due to the ability to coordinate with the Lewis basic sites in PMIDA that enhanced the binding ability. Thus, functionalization with a specific function group can increase binding with a specific metal ion. Although Cytop^®501, containing the functions of P=O and OH, could selectivity bind to terbium, in our study, Cytop^®501 did not react with the main MOF structure and was impregnated inside the pore, offering strong and free binding sites.

3.3. Selectivity Tests

Considering that rare earth metal ions exist with other contaminated metals, high adsorption and selectivity toward only terbium are needed for practical and industrial applications. Therefore, the adsorption of terbium in co-existence with other ions, such as 0.1 mM of each Fe³⁺, Al³⁺, Ni²⁺, Zn²⁺, Cu²⁺, Mg²⁺, and Ca²⁺, was tested at pH 4.5 for 5 h, similar to the optimum condition of the rare earth experiment using the UiO-66-COOH-DETA-impregnated Cytop^®501 membrane. The results in Figure 5 show that terbium was selectively adsorbed from all the contaminated metal ions, proving the superior practical application of the innovative synthesized membrane.

3.4. Adsorption Capacity and Adsorption Kinetics

The adsorption capacity was measured using a 40% UiO-66-COOH-DETA impregnated Cytop^®501 mixed-matrix membrane, where the samples were taken from the feed containing 123 mg/g of Tb³⁺ during the adsorption experiments and analyzed by using an ICP-AES device. The data, as shown in Figure 6, confirmed the fast adsorption ability of the impregnated membrane after only 1 h, and 113.2 mg/g of Tb³⁺ was adsorbed at 3 h. q_t refers to the amount of terbium ions adsorbed from the feed solution during the adsorption experiment by using a liquid membrane separation machine at time t. The kinetics of Tb³⁺ adsorption using the UiO-66-COOH-DETA solvent-impregnated Cytop^®501 mixed-matrix membrane were investigated to elucidate the kinetic effect of solvent impregnation. We calculated the equilibrium constants K₁ and K₂ for pseudo-first- and second-order kinetics, respectively. The value of K₁ is 1.35 with an R² of 0.9388, whereas the value of K₂ is 0.2224 with an R² of 0.996. Therefore, the pseudo-second-order model fitted the kinetics data with high accuracy, indicating that the solvent-impregnated mixed-matrix membrane is acting with a chelating exchange mechanism controlled by a secondary chemical reaction on the surface. After surface saturation with terbium ions, Cytop^®501 provides abundant binding sites on the surface, and the terbium ions diffuse into the pores by chemisorption with the binding sites within Cytop^®501. This mechanism allows for selective adsorption, which was reduced by the functionalization of UiO-66 with COOH-DETA and prevented the adsorption of most rare earth ions. Thus, solvent impregnation is the key to surface functionalization and overcoming the weak coordination of MOFs after functionalization.

3.5. Reusability

To confirm the reusability of the UiO-66-COOH-DETA impregnated Cytop^®501 mixed-matrix membrane, the desorption of Tb³⁺ ions was conducted using 0.2 M HCl as the stripping receiving solution, similar to the adsorption experiment, using a liquid membrane separation machine. From Figure 6, the stripping of terbium ions adsorbed by 40% impregnated membranes can be seen with the initial terbium concentration, where q is the amount of terbium stripped in the diluted hydrochloric solution after 6 h. As shown in Figure 7, the adsorbent membrane retained around 94% of its initial adsorption efficiency in the second round and achieved 91% after the third cycle, which supports the practical application of the impregnated mixed-matrix membrane for commercial purposes.

4. Conclusions

The adsorption of terbium from an aqueous solution using Cytop^®501 impregnated mixed-matrix membranes of modified UiO-66 MOFs was examined. The innovative membrane exhibited excellent selectivity toward terbium over other rare earths and contaminated metal ions due to its strong chelation ability and selective pore size that fitted the adsorption of terbium ions. The membrane showed an excellent adsorption capacity of about 113.2 mg/g of Tb³⁺ after 3 h, while desorption was possible using diluted hydrochloric acid. Its reusability was confirmed after three cycles. Solvent impregnation of the UiO-66 modified MOF succeeded in increasing the pore size and permeation of the membrane, both of which deteriorated after UiO-66 modification with COOH-DETA, proving that impregnation can alter the characteristics of MOFs and contribute to the different selectivity toward metal ions depending on the nature of the impregnated solvent. The adsorption process is very fast, taking only 5 h to adsorb 94% of terbium in the aqueous solution. This is the first time that this technique has been published for solvent impregnation of a mixed-matrix membrane. The results are outstanding, showing promise for commercial applications using terbium separation, and will be further developed in the future for the individual separation of rare earths.