Effect of Natural Zeolite Loading in Electrospun PAN/PVP Nanofiber Membranes on the Adsorption of Cationic Dye (Methylene Blue)

Materials

Polyacrylonitrile (PAN, Mw=150,000 g mol−1) and polyvinylpyrrolidone (PVP, Mw≈1,300,000 g mol−1) were obtained from Sigma-Aldrich (Singapore). N,N-Dimethylformamide (DMF, Merck, Germany) was used as the common solvent. Natural zeolite (clinoptilolite) was supplied by PT Paragon Perdana Mining and mined from Tanggamus, Lampung Province, Indonesia. The as-received zeolite was ground, sieved through a 200-mesh screen, and dried at 80 °C for 6 h before use. Methylene blue (MB, C.I. 52015, Merck, Germany) was selected as a representative cationic dye pollutant. Deionized water was used throughout. All chemicals were used as received.

Preparation of PAN/PVP/Zeolite nanofiber membranes

A base polymer solution was prepared by dissolving 0.8 g of PAN and 0.5 g of PVP in 10 mL of DMF at 60 °C under magnetic stirring (600 rpm, 4 h) until a transparent, viscous, and homogeneous dope was obtained. Four separate batches were then prepared by incorporating 0, 0.1, 0.3, and 0.5 g of sieved natural-zeolite powder into individual aliquots of the base solution, followed by an additional 2 h of stirring and 30 min of ultrasonication to promote uniform dispersion. The corresponding samples are designated PAN/PVP, PAN/PVP/Ze1, PAN/PVP/Ze3, and PAN/PVP/Ze5, where the numerical suffix indicates the zeolite mass in tenths of a gram (Table 1).

Electrospinning was carried out on a digital spinning platform (ILMI-N101, Integrated Laboratory of Materials and Instrumentation, Bandung, Indonesia). Each solution was loaded into a 10 mL plastic syringe fitted with a 21-gauge stainless-steel needle and deposited onto a rotating drum collector under the following fixed conditions: applied voltage 9 kV, feed rate 0.5 mL h−1, needle-to-collector distance 10 cm, and collection time 16 h, at ambient temperature and humidity. The as-spun membranes were peeled from the collector and dried at 60 °C. All samples were subsequently subjected to a post-spin heat treatment at 200 °C for 2 h in a tube furnace under N2 atmosphere at a heating rate of 5 °C min−1 to promote partial cyclization of the PAN backbone while preserving the zeolite framework and the overall fibrous architecture. The electrospinning parameters are identical to those employed in our earlier study of PAN/PVP/Nb2O5 nanofiber adsorbents to enable a direct comparison between the two filler systems.

Materials characterization

Fiber morphology was examined using a scanning electron microscope (SEM, JEOL JSM-6510) operated at 15 kV, after sputter-coating the samples with a thin layer of Au. The elemental composition was obtained from energy-dispersive X-ray spectroscopy (EDS) integrated with the SEM. Mean fiber diameters and their standard deviations were determined from SEM micrographs by measuring at least 100 randomly selected individual fibers per sample using ImageJ software. Chemical bonding was investigated by Fourier-transform infrared spectroscopy (FTIR, Shimadzu IRSpirit) in attenuated-total-reflectance (ATR) mode over 4000–500 cm−1. Crystallographic phase analysis was performed by X-ray diffraction (XRD, Rigaku SmartLab SE Basic) using Cu Kα radiation (λ=1.5406 Å) over a 2θ range of 10–80°. UV-visible absorption spectra were recorded on a UV-Vis spectrophotometer over 400–800 nm.

Adsorption studies

The dye-removal performance of the nanofiber membranes was evaluated using methylene blue as a representative cationic dye. In each experiment, 0.05 g of membrane was immersed in 100 mL of a 10 ppm MB solution contained in a 100 mL glass beaker. A 5 mL aliquot was withdrawn immediately before adding the membrane to determine the initial concentration, C0. The adsorption was conducted in the dark at room temperature with magnetic stirring at 400 rpm. Aliquots were withdrawn at predetermined intervals (10, 20, 30, 60, 120, 180, 240, and 300 min), and the residual MB concentration was determined from UV-Vis absorbance at 663 nm against a calibration curve built from 2, 5, 6, 8, and 10 ppm standards. The adsorption capacity at time t, qt (mg g−1), was calculated from the mass-balance expression

The removal efficiency is given by

Kinetic modelling

The kinetic behaviour of MB adsorption on the PAN/PVP/Zeolite nanofiber membranes was analysed using the linearized pseudo-first-order (PFO) and pseudo-second-order (PSO) models, expressed as

Morphology and elemental composition

Figure 1 shows SEM micrographs of the four electrospun samples at three magnification levels (1000×, 3000×, and 10,000×), together with the fiber-diameter distributions obtained from ImageJ analysis of at least 100 fibers per sample. All four samples produced continuous, bead-free, randomly oriented nanofibers with a uniform three-dimensional network topology, indicating that the addition of natural zeolite up to 0.5 g does not destabilize the electrospinning jet. The mean fiber diameter increases monotonically with zeolite loading, from 626±57 nm for PAN/PVP to 725±68 nm for PAN/PVP/Ze1, 794±69 nm for PAN/PVP/Ze3, and 902±68 nm for PAN/PVP/Ze5. The narrow and comparable standard deviations (57–69 nm) across the four samples confirm that the diameter distributions remain similarly unimodal, while the steady shift of the mean reflects the increase in dope viscosity produced by the progressive incorporation of solid filler into the PAN/PVP/DMF solution. At 10,000× magnification, zeolite particles appear as nodular features that are embedded in or partially protrude from the polymer fibers rather than being deposited as loose aggregates on the surface, demonstrating that the spinning conditions successfully encapsulate the filler inside the fiber matrix [28].

The EDS spectra and quantitative elemental analysis of the four samples are shown in Figure 2, and the corresponding atomic-percent composition is summarized in the bar chart of Figure 3a. The pristine PAN/PVP membrane (Figure 2a) contains only C (22.99 wt%), N (44.70 wt%), and O (31.93 wt%), consistent with the nominal PAN/PVP chemistry, together with a negligible Al contribution (0.37 wt%) that can be attributed to EDS background. With the incorporation of natural zeolite, both Si and Al appear and increase monotonically across the loading series: Si rises from 4.38 wt% (Ze1, Figure 2b) to 9.54 wt% (Ze3, Figure 2c) and 10.68 wt% (Ze5, Figure 2d), while Al rises from 1.38 to 2.66 and 2.90 wt% over the same series. The same monotonic trend is clearly visible in atomic percent in Figure 3a. The Si/Al ratio remains in the range 3.2–3.7 across all three loadings, in good agreement with the stoichiometry expected for clinoptilolite and confirming that the crystalline framework of the aluminosilicate is transferred intact from the raw mineral to the nanofiber matrix [29,30]. Together, the SEM and EDS results demonstrate that the nanofiber architecture is preserved, the filler is genuinely embedded rather than merely deposited, and the effective zeolite content can be tuned proportionally to the mass of precursor added to the dope.

Structural and chemical characterization

The X-ray diffraction patterns of the four samples together with the natural-zeolite reference are presented in Figure 3b. All four membranes exhibit a broad halo centered at 2θ≈17∘, which is assigned to the (100) reflection of the semi-crystalline PAN backbone superimposed on the amorphous contribution of PVP. On top of this polymer halo, progressively sharper diffraction features attributable to clinoptilolite emerge as the zeolite content is increased. In PAN/PVP/Ze1, low-intensity reflections are already visible at 2θ≈9.8∘ and 11.2^∘, which are characteristic of the (020) and (200) planes of clinoptilolite [31,32,33]. In PAN/PVP/Ze3, a strong new reflection appears at 2θ≈22.6∘, and in PAN/PVP/Ze5 this peak sharpens further and is accompanied by an additional reflection at 2θ≈29.6∘, which is another diagnostic clinoptilolite feature. The progressive emergence and intensification of these reflections without the appearance of any secondary or shifted phase demonstrate that the crystalline aluminosilicate framework is preserved through electrospinning and the subsequent 200 °C heat treatment under N2, and that the effective loading of clinoptilolite in the nanofiber matrix scales monotonically with the precursor mass.

The FTIR spectra in Figure 3c confirm the compositional integrity of the polymer matrix and the progressive incorporation of the zeolite filler. All samples show the characteristic bands of PAN/PVP: a broad O–H/N–H envelope near 3400 cm−1, aliphatic C–H stretching around 2930 cm−1, the sharp C≡N stretch of the PAN nitrile at ∼2242 cm−1, the C=O stretch of the PVP pyrrolidone ring at ∼1659 cm−1, CH2 bending around 1450 cm−1, and C–N/C–O backbone vibrations near 1288 cm−1. These polymer-associated bands remain essentially unchanged in both position and relative intensity across all Ze-loaded samples, indicating that neither the compositing step nor the 200 °C thermal stabilization significantly disrupts the PAN/PVP chemistry. A new and growing absorption band associated with the asymmetric stretching vibration of Si–O–Si / Si–O–Al bonds emerges in the ∼1000–1100 cm−1 region [32,33]. Its position shifts progressively to lower wavenumber as the zeolite content increases, from 1067 cm−1 in PAN/PVP/Ze1 to 1042 cm−1 in PAN/PVP/Ze3 and 1040 cm−1 in PAN/PVP/Ze5, while the corresponding band depth deepens from ∼96% to ∼86% transmittance. A second, weaker Si–O band centered around 520–540 cm−1 follows the same trend. The red-shift and intensification of the aluminosilicate fingerprint, together with the unchanged polymer bands, provide clear spectroscopic evidence that clinoptilolite is successfully embedded in the PAN/PVP matrix and that its effective loading increases monotonically from Ze1 to Ze5, in full agreement with the XRD and EDS trends.

Methylene blue adsorption kinetics

The time-dependent UV-Vis absorption spectra of 10 ppm MB solutions exposed to each membrane, together with the derived adsorption kinetics, are shown in Figure 4. The characteristic MB absorption peak at 663 nm decreases monotonically with contact time for all four samples (Figure 4a–d), confirming progressive dye uptake. The rate and extent of the decrease, however, depend strongly on the zeolite content. For pristine PAN/PVP, the absorbance decreases only slowly and retains a clearly visible peak at 300 min, corresponding to ∼58.7% removal and an experimental adsorption capacity of 8.2 mg g−1. The introduction of even a modest amount of zeolite dramatically accelerates the uptake: PAN/PVP/Ze1 reaches 91.7% removal (12.8 mg g−1), PAN/PVP/Ze3 reaches 96.5% removal (13.5 mg g−1), and PAN/PVP/Ze5 reaches essentially complete removal (∼100%, 14.0 mg g−1) within the same 300 min. The time-resolved adsorption-capacity curves in Figure 4e confirm this monotonic enhancement and show a classical two-stage profile for all samples: a fast initial uptake within the first 30–60 min, during which the majority of the accessible surface sites are rapidly populated, followed by a gradual approach to equilibrium as the remaining sites are filled.

The linearized PFO and PSO plots are shown in Figure 4f and Figure 4g, and the corresponding fitted parameters are collected in Table 2. The PSO model describes the data consistently better than the PFO model across the entire loading series: the PSO coefficient of determination is uniformly high (R2≥0.984 for all samples, with R2=0.996 for PAN/PVP, PAN/PVP/Ze1, and 0.993 for PAN/PVP/Ze3), whereas the PFO fit degrades noticeably at low loading, reaching R2=0.885 for the pristine PAN/PVP control. Furthermore, the PSO-fitted qe values are in close agreement with the experimentally observed adsorption capacities at 300 min, while the PFO-fitted values systematically underestimate them by roughly a factor of two. This combined evidence, namely the higher R2 and closer agreement with the experimental plateau, identifies the PSO model as the dominant kinetic regime and suggests that MB adsorption on the PAN/PVP/Zeolite nanofiber membranes is rate-limited by a chemisorption-type surface interaction rather than by pure film diffusion. The PFO rate constant k1 increases from 0.010 to 0.021 min−1 as the zeolite content increases from 0 to 0.5 g, reflecting the faster uptake observed in Figure 4e. The PSO rate constant k2 decreases from 0.007 to 0.001 g mg−1 min−1 over the same series; this decrease does not indicate a slowdown of the real kinetics but arises from the inverse qe2 dependence of the PSO expression, which mathematically couples a rising qe to a declining k2. We note that the PSO-fitted qe values for PAN/PVP/Ze3 and PAN/PVP/Ze5 (16.08 and 16.69 mg g−1, respectively) exceed the instantaneous mass-balance ceiling C0V/m≈13.96 mg g−1 at the measured initial concentration; this is a known characteristic of the linearized PSO model, whose fitted qe is an asymptotic extrapolation of the t/qt vs t regression rather than a direct experimental quantity, and should be interpreted as the equilibrium capacity predicted by the model if the experiment were run to infinite time.

The mechanistic picture that emerges from the combined SEM/EDS/XRD/FTIR/kinetic results is internally consistent. Pristine PAN/PVP already exhibits moderate MB uptake thanks to the polar nitrile groups on the PAN backbone, which can engage in dipolar and π–π interactions with the aromatic framework of the cationic dye. Adding natural zeolite introduces a much larger density of high-affinity sites for MB+: the negatively charged aluminosilicate framework provides electrostatic attraction for cations, the exchangeable Na+, K+, and Ca2+ cations in the clinoptilolite channels enable direct cation-exchange uptake, and the internal microporosity further contributes through physical pore filling. Because the effective zeolite content in the fiber rises monotonically with the precursor mass (as shown by the Si/Al EDS values in Figures 2 and 3a, the intensifying clinoptilolite XRD reflections in Figure 3b, and the deepening Si–O–Si/Si–O–Al FTIR band in Figure 3c), the density of these high-affinity sites increases in the same order, explaining the PAN/PVP < Ze1 < Ze3 < Ze5 ranking of both the initial uptake rate and the equilibrium adsorption capacity. PAN/PVP/Ze5 removes essentially all MB from solution within 300 min at 10 ppm, demonstrating the practical effectiveness of the composite under conditions representative of dilute textile effluents.