Characterization of modified PET tem templates and composite PET tems
The development of MOF-functionalized track-etched membranes requires precise control over the chemical architecture of the supporting surface to ensure reproducibility, homogeneity, and nanoscale functionality. In this work, we employed surface-initiated reversible addition-fragmentation chain transfer (RAFT) polymerization to achieve uniform grafting of poly(N-vinylformamide) (PNVF) onto PET track-etched membranes (TeMs). This controlled radical polymerization technique allows for fine-tuning of polymer chain length and distribution through the use of appropriate RAFT agents such as 4-cyano-4-(phenylcarbonothioylthio)pentanoic acid (CPPA), thereby offering a high level of control over the resulting surface properties.
RAFT polymerization was selected deliberately due to its ability to provide reproducible surface functionality, which is critical for subsequent covalent immobilization of MOF particles. Homogeneous grafting is particularly important in this system for two reasons: (1) to enable uniform distribution of reactive groups across the membrane surface, ensuring consistent MOF attachment, and (2) to avoid excessive polymer deposition that could lead to pore blockage in the TeM nanochannels. Such occlusion could compromise fluid transport and reduce effective surface area for uranium adsorption. Therefore, polymerization conditions were carefully optimized to maintain a balance between sufficient functional group density and preservation of membrane porosity. Additionally, oxidation of the PET surface prior to benzoyl peroxide (BP) immobilization was found to increase the number of initiator sites, thereby enhancing the overall grafting efficiency in subsequent steps.
To validate each modification stage and to investigate structure-function relationships in the resulting composite membranes, a suite of complementary characterization methods was applied, including FTIR, SEM, EDX, contact angle measurements, XPS, and nitrogen adsorption-desorption porosimetry. These techniques were critical in confirming the chemical changes on the membrane surface and understanding their impact on final sorption performance.
Poly(vinylamine) (PVAm) was chosen as a key intermediate due to its water solubility, pH responsiveness, and high density of primary amine groups. However, the direct synthesis of PVAm is hindered by the instability of its monomer, which readily undergoes imine-enamine tautomerization. As a practical alternative, we used PNVF as a precursor and converted it into PVAm via alkaline hydrolysis. UV-induced graft polymerization of the NVF monomer was conducted by systematically varying parameters such as monomer concentration, solvent, and reaction time.
We initially investigated the influence of solvent type and monomer concentration on the grafting degree. In agreement with previous reports, higher alcohols yielded greater grafting efficiency due to increased hydrophobic interactions and monomer partitioning. As illustrated in Fig. 2a, the grafting degree followed the trend: ethanol < n-propanol < n-butanol. Concurrently, we observed that increasing NVF concentration beyond 10% led to partial degradation of the PET TeM structure, with up to 50% monomer decomposition. Therefore, all subsequent experiments were performed in n-butanol using a 10% monomer concentration.
The effect of polymerization time on the grafting degree was also examined, with reaction durations ranging from 1 to 6 h (Fig. 2b). Experiments were repeated five times for each time point to ensure reproducibility. Grafting degrees achieved at 2 and 3 h were below 5%, indicating limited polymerization activity. In contrast, a 6-hour reaction yielded a significant increase in grafting (~ 20%); however, this condition compromised membrane integrity, rendering the material brittle. Notably, the high standard deviation in grafting degree after 6 h can be attributed to mechanical damage observed in some samples during post-reaction drying.
Following the optimization of NVF grafting conditions, the stepwise chemical modification of PET TeMs and Cr-MIL-101 MOFs was confirmed through Fourier-transform infrared (FTIR) spectroscopy. Figure 3a displays the FTIR spectra of pristine PET, PNVF-g-PET, and PVAm-g-PET membranes. Pristine PET exhibited characteristic absorption bands at 1716 cm (C = O stretching of ester groups), 1244 cm (C-C-O stretching), and ~ 1000 cm (aromatic C-H bending), consistent with previous reports. Upon grafting of N-vinylformamide (PNVF), new peaks appeared at 1600 cm and 1550 cm, corresponding to the C = O stretch and N-H bending of the formamide group, respectively. After alkaline hydrolysis, the 1600 cm band decreased in intensity, indicating successful conversion of amide groups into primary amines and formation of PVAm chains. The effect of hydrolysis parameters (NaOH concentration and reaction time) on amine group generation was also investigated. As shown in Fig. 3c, optimal amine density was obtained at 0.1 M NaOH for 10 min. Under harsher conditions (0.2 M or > 10 min), membrane degradation occurred, while milder conditions (0.05 M) led to insufficient hydrolysis.
In parallel, the synthesis and post-synthetic functionalization of Cr-MIL-101 MOFs were monitored using FTIR. As shown in Fig. 3b, the amino-functionalized MIL-101(Cr) exhibited characteristic vibrations at ~ 1582 cm (N-H bending), 1400-1500 cm (C-N deformation), and ~ 1653 cm (C = O of the terephthalic linker). After azide modification, a distinct and sharp absorption band appeared at 2205 cm⁻¹, confirming the successful introduction of azide groups. Broad bands near 3000 cm⁻¹ were attributed to O-H stretching from MOF ligands, while the aromatic C-H bending around 616 cm remained consistent.
Next, alkyne groups were introduced onto the PVAm-g-PET surface via DCC-mediated amidation using propiolic acid. This reaction proceeded through the formation of an O-acylurea intermediate that facilitates nucleophilic attack by the primary amines, forming stable amide bonds. Figure 4 illustrates this functionalization route. The presence of a sharp alkyne C ≡ C stretching vibration at 2200 cm (visible in Fig. 3a) confirms the success of this surface modification, preparing the substrate for the subsequent click reaction with azide-functionalized MOFs.
As illustrated in Fig. 4, the alkyne groups were covalently introduced onto the surface of PVAm-grafted PET TeMs via DCC-mediated amidation of propiolic acid. This reaction enabled selective formation of amide bonds between the carboxylic acid moieties and surface amines, resulting in stable alkyne-functionalized membranes. These reactive alkyne groups served as anchoring sites for the subsequent immobilization of azide-functionalized MIL-101(Cr) particles through copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC), a highly efficient and chemoselective "click" reaction. This strategy ensured a robust covalent attachment between the MOF and the modified membrane surface. By this way, both the structural integrity of the MOF and the accessibility of its porous architecture were preserved. The successful integration of Cr(azide)MIL101 MOFs onto the alkyne@PET TeM surface was then confirmed through detailed surface analysis by X-ray photoelectron spectroscopy (XPS). XPS is a powerful surface-sensitive technique that enables both elemental identification and chemical state analysis of atoms in the outermost layers of materials. To confirm each functionalization step and monitor the changes in surface chemistry, survey scans and high-resolution spectra were collected for all modified membrane samples.
The wide-scan XPS spectrum of pristine PET TeMs (Fig. 5a) exhibited dominant peaks corresponding to carbon (C1s) and oxygen (O1s), which are consistent with the chemical structure of polyethylene terephthalate upon grafting with poly(N-vinylformamide) (PNVF), a distinct nitrogen (N1s) peak emerged at ~ 399 eV, indicating the successful incorporation of nitrogen-bearing formamide groups (Fig. 5b). Additionally, a sulfur (S2p) signal was detected, originating from the thiocarbonyl end groups of the RAFT agent (CPPA), providing direct evidence that the grafting process was mediated via RAFT polymerization.
Following alkaline hydrolysis to convert PNVF to PVAm, the N 1s peak further increased in intensity, while C1s and O1s peaks decreased slightly, reflecting the removal of formyl (-CHO) groups and the formation of primary amines (Fig. 5c). The persistence of the sulfur peak in both grafted and hydrolyzed membranes confirmed the retention of RAFT agent fragments at the polymer chain ends, thereby substantiating the controlled grafting process.
High-resolution C1s spectra provided further chemical insights. For pristine PET (Fig. 5d), peaks at 284.8 eV (C-C), 286.4 eV (C-O), and 288.6 eV (O-C = O) were observed. After MOF immobilization, significant changes were detected in the carbon environment, with increased intensity in peaks associated with C-N and C-O functionalities, suggesting the successful incorporation of polar groups and MOF-related chemical species (Fig. 5e).
Moreover, the Cr2p core-level spectrum of the MOF-modified membrane (Fig. 5f) displayed distinct peaks at ~ 577 eV (Cr2p) and ~ 587 eV (Cr2p), characteristic of Cr(III) oxidation states in the MIL-101 framework. A shoulder near the main Cr 2p peak further corroborated the presence of Cr(III) species, validating the integrity and successful immobilization of the MOF structure on the PET TeM surface.
Surface wettability of the modified PET TeMs was systematically evaluated using contact angle (CA) measurements. These measurements were used to assess changes in hydrophilicity resulting from successive surface modifications, including polymer grafting and MOF immobilization. As shown in Fig. 6, the contact angle values decreased progressively after each functionalization step, indicating an increase in surface hydrophilicity. Specifically, PNVF-g-PET membranes exhibited variable CA values depending on the solvent used during grafting. Among the tested conditions, the lowest contact angle was observed for membranes grafted in n-butanol at the highest NVF concentration and optimal reaction time. This observation is consistent with the higher grafting degrees achieved under these conditions, leading to denser coverage of polar functional groups.
Following immobilization of Cr(azide)MIL101 MOFs onto the alkyne-functionalized PET surface, a further decrease in contact angle was recorded. This enhancement in hydrophilicity can be attributed to the inherently polar nature of the MOF structure and the additional surface roughness introduced by MOF deposition. Collectively, these results confirm that the composite membrane surface becomes increasingly hydrophilic through each modification step, which is advantageous for applications involving aqueous media.
Scanning electron microscopy (SEM) was employed to investigate the surface morphology and pore distribution of the PET TeM template following MOF immobilization. Representative images of the MOF@PET TeM composite membrane (degree of grafting: 10%, MOF loading degree: 2.4 ± 0.4%) are presented in Fig. 7. The top-view SEM images (Fig. 7a) reveals the presence of MOF crystallites distributed across the membrane surface, while the cross-sectional image (Fig. 7b) confirms that these MOF particles primarily reside on the surface rather than penetrating deeply into the porous matrix. This surface-localized distribution is advantageous for maximizing interfacial accessibility during sorption or catalytic applications. The effect of time on the degree of MOF immobilization on the membrane surface at the preliminary stage was also investigated using SEM-EDX mapping of the MOF@PET (Fig. 8a-d). Unfortunately, even after repeated experiments, it was unable to achieve a high degree of MOF loading exceeding 2.4%, despite increasing the loading time. Data on the change in the degree of MOF loading depending on time are presented in Fig. 8e. The analysis of the EDX spectra also indicates that the maximum saturation of MOF in terms of chromium content on the membrane surface is achieved after 3 days of immobilization (Fig. 8f).
It is important to emphasize that the efficiency of MOF immobilization is closely linked to the density of alkyne functional groups on the membrane, which in turn depends on the number of surface amino groups formed after hydrolysis. Each repeating unit of PNVF contains a single formamide group, which upon hydrolysis yields one primary amine group and a formate byproduct. These amines subsequently react with propiolic acid in the presence of DCC, forming stable amide linkages that introduce terminal alkyne moieties. The presence of sulfur groups observed in the XPS spectra corroborates the persistence of RAFT chain ends, while the covalently bound alkynes act as click-compatible anchor points.
Upon introduction of azide-modified MIL-101(Cr) nanoparticles, the copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) reaction facilitates a site-specific and robust covalent attachment of the MOF to the alkyne-functionalized membrane. This click reaction yields a triazole linkage, replacing the azide with an imidazole-type structure and anchoring the MOF to the polymer chain. The ligand in MIL-101(Cr) contains carboxylate groups and an azide moiety introduced at the ortho-position of aminoterephthalic acid, which react selectively with the alkyne units on the PET TeM. Consequently, the degree of grafting, and thus the density of reactive alkynes, directly impacts MOF deposition efficiency.
Brunauer-Emmett-Teller (BET) analysis was carried out to examine the porosity and pore size distribution of the prepared MOF@PET composites, as well as the powder MOF and the pristine PET template (Fig. 9). According to IUPAC classification, adsorption isotherms fall into six categories, and the isotherm obtained for MOF@PET belonged to type IV, which is typically associated with mesoporous structures in the 2-50 nm range. The hysteresis loop observed also matches the behavior usually seen in industrial mesoporous adsorbents. As expected, the unmodified PET track-etched membrane showed very low porosity, and a small surface area of only 1.72 m²/g. In contrast, the azide-modified MIL-101(Cr) powder exhibited a much higher surface area, about 649.8 m²/g, which is in good agreement with literature values. When roughly 2% of the MOF was immobilized on the PET surface, the surface area of the composite increased noticeably, from 1.72 to about 15.6 m/g, which confirms that the presence of the MOF effectively contributes to the overall porosity (Table 1).
Understanding the adsorption kinetics of U(VI) on functionalized composite membranes is essential for evaluating their practical applicability and elucidating the interaction mechanisms between the adsorbate and the active surface of the sorbent. In this study, we examined the time-dependent uptake behavior of uranium(VI) ions by MOF@PET TeM membranes, as well as the effect of pH, diffusion mechanisms, and kinetic models.
The pH of the solution is a key factor influencing uranium speciation and adsorbent surface charge, and consequently the sorption performance. In aqueous media, U(VI) predominantly exists as cationic species (e.g., UO, UOOH, and (UO)(OH)) in the pH range of 4-6. At higher pH values (> 6), both cationic and anionic uranyl species may coexist, with increasing prevalence of anionic forms such as (UO)(OH) . Therefore, the net surface charge of the adsorbent and the dominant U(VI) species must be carefully considered to maximize sorption efficiency.
As shown in Fig. 10a, the variation of surface charge of MOF@PET TeM as a function of pH reveals a point of zero charge (pHpzc) at approximately 6.5. Below this value, the surface acquires a positive charge, whereas at pH levels above 6.5, the surface becomes negatively charged. Since U(VI) species such as UO, UOOH, and (UO)(OH) are positively charged in mildly acidic to neutral pH conditions (pH 4-6), operating at pH 6.3, which is just below the pHpzc, ensures favorable electrostatic attraction while minimizing structural degradation of the MOF under more acidic conditions. This selection is also consistent with previous studies that optimized U(VI) adsorption near the pHpzc of MOF-based sorbents for enhanced selectivity and capacity.
The effect of contact time on U(VI) adsorption was evaluated at this selected pH, and the results are presented in Fig. 10b. The sorption kinetics showed a relatively slow adsorption profile, with only ~ 10% of uranyl ions being removed from solution within the first 120 min. Equilibrium was reached after approximately 96 h, and the corresponding sorption capacity (Qₑ) was calculated to be 418.0 mg/g.
To assess the specific contribution of the Cr(azide)MIL-101 MOF component to the sorption performance, control experiments were conducted using alkyne-functionalized PET membranes that underwent the same surface modification steps, excluding the MOF immobilization. As shown in Fig. 10b, these control membranes exhibited a negligible uranium uptake (~ 3.4 mg g after 168 h), more than 100 times lower than that of the MOF-grafted composite. This result clearly indicates that the MOF domains are the dominant contributors to the overall sorption capacity. Although the support was originally grafted with PVAm chains, the subsequent alkyne functionalization likely blocked or replaced most of the free amine groups that could have otherwise contributed to uranyl coordination. Thus, the residual adsorption observed in the control membrane is minimal, confirming that the high uranium uptake in the composite can be primarily attributed to the immobilized Cr(azide)MIL-101 MOF. According to previous studies, the original PET membrane exhibits an inert nature towards various types of heavy metal ions, and its influence on the sorption capacity of the final composite is minimal.
The initial uranium concentration had a pronounced effect on the adsorption process. At low concentrations (10-300 mg/L), the uptake increased very sharply because the available sites on the MOF@PET surface were far from being occupied. With further increase above 100 mg/L, the adsorption continued to grow but at a slower rate, and then gradually approached a plateau, which is consistent with saturation of active sites (Fig. 10c).
According to the well-established classification of Giles, adsorption isotherms are divided into four main classes: S-type (cooperative adsorption, uptake increases with concentration), L-type (Langmuir-like, high affinity at the beginning followed by surface saturation), H-type (very steep at low concentration, indicating extremely strong interaction between adsorbent and adsorbate), and C-type (linear behavior, uptake directly proportional to concentration). In our case, the adsorption curve belongs to the L-type, showing that U(VI) ions interact strongly with the MOF@PET surface initially, but the capacity is limited by the finite number of binding sites In addition, based on the IUPAC categorization, the adsorption isotherm also falls into Type II, which is usually characteristic of multilayer adsorption on porous materials. This means that uranium ions first form a strong monolayer on the MOF surface, and at higher concentrations, weaker multilayer adsorption can also occur.
(b) Effect of contact time on U(VI) sorption capacity of the composite sorbents (pH = 6.3, [U(VI)] = 100 ppm) and (c) effect of initial uranium ion concentration on U(VI) adsorption capacity.
To better understand the sorption dynamics and evaluate the governing mechanism, the experimental data were fitted to three commonly used kinetic models: pseudo-first-order, pseudo-second-order, and Elovich models. The pseudo-first-order model assumes that the rate of occupancy of adsorption sites is proportional to the number of unoccupied sites. The linearized form is:
where q and q are the adsorption capacities at equilibrium and time t, respectively, and k is the rate constant. As shown in Fig. 11a, the data displayed moderate linearity, and the derived kinetic parameters (from Table 2) yielded relatively low correlation coefficients (R), suggesting that this model does not adequately describe the adsorption behavior in this system.
The pseudo-second-order model is based on the assumption that chemisorption is the rate-limiting step, involving valency forces through sharing or exchange of electrons. Its linearized form is:
where is the second-order rate constant.
The plot of versus t (Fig. 11b) shows superior linearity compared to the first-order model, with higher R values and better agreement between experimental and calculated values. This suggests that U(VI) adsorption is best described by a pseudo-second-order kinetic mechanism, consistent with chemisorption dominating the uptake process.
The Elovich equation is particularly useful for systems with heterogeneous surface energies and where desorption plays a role near equilibrium. Its mathematical form is:
where is the initial adsorption rate (mg g min) and is the desorption constant (g mmol).
As shown in Fig. 11c, the Elovich model also fits the data reasonably well, capturing the initial rapid adsorption followed by slower approach to equilibrium. According to Table 1, the high R values and favorable α/β ratios indicate that this model can complementarily describe the sorption process by accounting for surface heterogeneity and energetic distribution.
Among the three kinetic models evaluated, the pseudo-second-order model exhibited the highest correlation with experimental data. This indicates that U(VI) adsorption onto MOF@PET TeM is likely governed by chemisorption, involving interactions between U(VI) ions and specific active sites, such as carboxylate or azide-derived triazole groups.
These findings are consistent with literature reports on metal-ion uptake by MOF-modified surfaces, where the concentration of available functional groups and their accessibility strongly influence the adsorption rate and mechanism. Furthermore, the Elovich model reinforces the presence of site heterogeneity, suggesting a combination of surface-driven and diffusion-mediated adsorption behaviors.
Diffusion kinetics of uranium ion sorption describes the process of uranium ion adsorption by the sorbent surface, where the rate of the process is limited by the diffusion of ions in the liquid phase to the sorbent surface and/or in the sorbent pores. Depending on the conditions, the sorption process can be limited by external diffusion (diffusion in the solution volume) or internal diffusion (diffusion in the sorbent pores). In this study, we used the Boyd and Weber-Morris model to describe the diffusion kinetics of uranium ion sorption. The intraparticle diffusion model is formulated based on the Weber-Morris approach and is used to determine the stage that limits the rate of the adsorption process (Table 2). Generally, substances present in a solution are adsorbed by mass transfer-surface, film diffusion and diffusion through pores. It is known that if the graph of Q versus t (Fig. 11d) is linear and passes through the origin, then intraparticle diffusion is the rate-controlling step. From the presented data, it is clear that for MOF@PET composites, the kinetic curve is bilinear (i.e., uranyl ion sorption is accompanied by two separate steps). The first, sharper step is associated with the diffusion of U(VI) ions in the boundary layer. The second step corresponds to the final equilibrium stage. In addition, the first linear section does not pass through the origin, indicating high initial sorption of U(VI) with diffusion through the solution to the outer surface of the adsorbent through a thick boundary layer. The corresponding intraparticle diffusion parameters are: = 0.65 mg g min and C = 411.6 mg g, confirming the strong boundary layer effect (Table 2).
The Boyd model is used to understand whether film diffusion is the rate-controlling step of the adsorption process. According to this kinetic model, the external surface around the adsorbent has a major influence on the diffusion of the adsorbate and pore diffusion controls the mass transfer when the graph is linear and passes through the origin. Figure 11e shows the graph of Boyd's model for U(VI) adsorption: it can be concluded that the graph does not pass through the origin and therefore the rate is controlled by film diffusion. The apparent Boyd rate constant was calculated as k = 0.20 min (Table 2).