A polydimethylsiloxane/montmorillonite/filter paper composite as a self-standing and easily separable adsorbent for methylene blue removal from aqueous solutions

The world’s water resources face severe threats from the growing population, sudden climate changes, and significant global water pollution, all contributing to a rapidly rising demand for clean water (Sivakumar and Lee 2022; Boukind et al. 2022). Wastewater reclamation has therefore become a critical alternative source (Silva 2023). Synthetic dyes, discharged mainly from textile, pigment, plastic, leather, printing, and cosmetics industries, are among the most important pollutants (Tan et al. 2015; Kyzas et al. 2013). Annually, global dye production reaches around 700 000 tons, with 10–15% discarded as wastewater (Tan et al. 2015; Oladoye et al. 2022). Thus, these dyes have the potential to seriously contaminate water if they are not removed before being released into the environment (Katheresan et al. 2018).

One of the extensively used dyes is methylene blue (MB), widely employed in significant quantities by the textile, plastic, food, cosmetics, and pharmaceutical industries, and also for biological staining procedures (Sivakumar and Lee 2022; Khan et al. 2022; Oladoye et al. 2022). Eliminating this cationic dye is difficult because its intricate aromatic structure makes it resistant to biodegradation (Boukind et al. 2022). A variety of technologies are employed for dye removal, including membrane separation (Zhang et al. 2017; Fang et al. 2020), flocculation-coagulation (Ihaddaden et al. 2022), electrochemical techniques (Mawazi 2015), aerobic or anaerobic treatment (Ong et al. 2005), photodegradation (Zhou et al. 2010; Kaya-Özkiper et al. 2021), liquid–liquid extraction (Pandit and Basu 2004), and adsorption (Almeida et al. 2009; Leodopoulos et al. 2015; Kaya-Özkiper et al. 2022a, b). Most of these technologies effectively remove dyes, however, they require trained operators, and consume a lot of energy and chemicals (Fito et al. 2023), and may generate secondary pollutants, all of which increase operating costs. For instance, conventional chemical routes such as coagulation/flocculation produce large volumes of metal‑rich sludge (Lin et al. 2023; Lucas et al. 2025), whereas many advanced‑oxidation processes form toxic intermediates that can be even more harmful than the original dye and therefore demand additional post‑treatment (Abhisek et al. 2025).

Among these advanced methods, adsorption stands out because of its low cost, simple design, and environmental friendliness. Another advantage is the high flexibility provided by the wide variety of available adsorbents. Adsorption involves the adhesion of adsorbate molecules to the surface of an adsorbent, and its effectiveness depends on selecting the right material offering a high number of suitable adsorption sites (Fito et al. 2023). It is an inexpensive and straightforward method that can produce high-quality effluent without generating harmful substances, effectively eliminating various organic and inorganic pollutants. Consequently, adsorption is widely applicable in pollution control (França et al. 2022).

A variety of adsorbents, including carbon-based materials such as activated carbon or graphene-based nanocomposites (Snik et al. 2022, 2025), zeolites, metal-organic frameworks, industrial waste-based materials, metal oxides, biosorbents, and clay minerals, are utilized for wastewater treatment (Kyzas et al. 2013; França et al. 2022). Considering that most reported adsorbents are powders, recovery and reuse of powdered adsorbents remain a major bottleneck because conventional separation methods such as filtration, centrifugation, and sedimentation are inefficient, energy‑intensive, and require extra post‑processing, prompting researchers to seek better alternatives (Kaya-Özkiper et al. 2022b; Nasiri et al. 2022; Satyam and Patra 2024). Thus, several approaches were reported in the literature for the production of easily separable adsorbents. For example, Kaya-Özkiper et al. produced sepiolite- and metakaolin-based alkali-activated monoliths having varying porosities for their utilization in MB removal from water (Kaya-Özkiper et al. 2022b). An increase in porosity of the monoliths resulted in enhanced MB removal capacities at the expense of decreased compressive strength. The sepiolite-based alkali-activated monolith, having 40% porosity, provided an MB uptake capacity of 10.3 mg g⁻¹, a promising uptake value with the advantage of an easily recoverable adsorbent. Another method is the production of adsorbents that can be separated from the adsorption environment by magnetic separation. For example, Fan et al. synthesized β-cyclodextrin–chitosan modified Fe₃O₄ nanoparticles as an adsorbent for MB removal (Fan et al. 2012). The resulting adsorbent provided a MB uptake of 2.48 g g⁻¹, with the advantage of simple and eased magnetic separation.

An alternative approach is the use of chemically modified filter paper (FP) as an easy-to-separate adsorbent. FP provides many advantages, including a highly optimized industrial production process, a fibrous and porous structure, and being biogenic (Böhm et al. 2014; Postulka et al. 2019). Paper is basically a nonwoven mat of cellulose fibers, which are renewable and highly abundant (Al-Shemy et al. 2022). For example, cotton linter paper contains about 80% cellulose (Sczostak 2009). Cellulose is produced in quantities estimated at approximately 10¹² tons annually on Earth, making it the most prevalent biopolymer (Haldhar et al. 2018). Cellulose is not only cost-effective and biodegradable but also exhibits excellent thermal and chemical stability (Paul et al. 2016). Since paper is mainly composed of cellulose, it possesses many of its advantages: It is easily modifiable, non-toxic, lightweight, and has strong mechanical properties (Park et al. 2016). The hydroxy groups on the surface of cellulose fibers facilitate chemical functionalization (Boukind et al. 2022). These attributes, combined with its affordability and biodegradability, highlight cellulose’s and thus paper’s potential as a highly promising material for dye adsorption and removal.

Despite its advantages, pure cellulose has limited adsorption capacity without modification (Boukind et al. 2022). Moreover, significant efforts have been directed towards improving the mechanical strength of paper in wet state (Paul et al. 2016). Only few approaches have been reported in the literature focusing on the chemical modification of ordinary FP for its utilization as a dye removal adsorbent. For instance, Li and colleagues coated cetyl trimethyl ammonium bromide modified bentonite (with loadings varying between 10 to 40 wt%) onto FP by polyacrylic acid as a binder for its utilization as a simple membrane for the adsorption of MB and orange II (Li et al. 2018). Another study illustrating the use of modified FP as a filtration material for MB removal was reported by Shi and colleagues, where particles of a Zr-based metal-organic framework, UiO-66, were immobilized on carboxymethylated FP via the solvothermal method (Shi et al. 2021). Gopalakrishnan et al. reported the growth of flower-like structured MoS₂ nanosheets on graphene dipped FP by hydrothermal method and the resulting adsorbent was tested for its MB removal performance from aqueous environment (Gopalakrishnan et al. 2020). Qian and coworkers modified FP with sodium alginate as a mechanically strong MB removal adsorbent (Qian et al. 2018). Few further examples show the successful utilization of modified FP-based adsorbents for the removal of other dyes such as prussian blue (Rego et al. 2023), as well as other contaminants such as tetracycline (Cao et al. 2022), phosphate (Arshadi et al. 2018), and heavy metals (d’Halluin et al. 2017).

One alternative method for modifying the structure of paper is the attachment of inorganic particles on paper surfaces through polymer chains (Koşak Söz et al. 2018). Attaching inorganic particles onto inorganic surfaces (unlike organic paper surfaces) through polymer chains by a highly practical and reproducible method was demonstrated by the studies of Botter et al. (Botter et al. 1992), Soares et al. (Soares et al. 1996), and Krumpfer and McCarthy (Krumpfer and McCarthy 2011). In these studies, inorganic surfaces were hydrophobized by trimethylsilyl-terminated polydimethylsiloxane (PDMS) chains, also known as non-reactive silicone oil, through the formation of covalent bonds to various inorganic surfaces (silicon wafers containing hydroxy groups, metal oxides, and glass surfaces) (Botter et al. 1992; Soares et al. 1996; Krumpfer and McCarthy 2011). Krumpfer and McCarthy proposed two possible reaction mechanisms to explain the covalent bonding between PDMS chains and surface hydroxy groups: i) Hydrolysis of PDMS followed by condensation with –OH groups on the surface, and ii) direct (acid catalyzed) silanolysis of PDMS by a surface silanol (Krumpfer and McCarthy 2011).

In our previous studies, we proved for the first time that this surface modification approach, which was only proposed for coating inorganic surfaces, could also be used to modify organic FP surfaces, since these surfaces are inherently hydrophilic and also slightly acidic. Thus, without the need for a metal catalyst, expensive starting materials, or multi-step approaches, we were able to use PDMS chains as a binder between the paper substrate surface and the inorganic particles on it by using an easy-to-perform method (Koşak Söz et al. 2018, 2020; Söz et al. 2021; Koşak Söz 2021, 2022; Evren et al. 2024). Further publications on surface modifications using trimethylsilyl-terminated PDMS-based coatings are already available in the literature. However, in these studies, the polymerization of the coating was based on the well-known hydrosilylation method in the presence of a Pt or Pb catalyst (Mai et al. 2019; Ni et al. 2019; Eduok et al. 2021; Bian et al. 2021; Zhang et al. 2021). Among these studies, those by Mai et al. and Eduok et al. involved the modification of cotton fabrics, which can be considered as woven alternatives to paper, by coating them with PDMS and inorganic particles other than montmorillonite to prepare superhydrophobic membranes for oil/water separation. Zhang et al. reported a multilayer coating comprising various components, including trimethylsilyl-terminated PDMS and MMT, for preparing fire-resistant cotton fabric. Nevertheless, the catalyst-free and PDMS-mediated immobilization of MMT onto filter paper for dye adsorption has not been explored in any of the aforementioned studies. Such an approach is not only novel, but also circumvents a critical limitation of conventional Pt- or Pb-catalyzed hydrosilylation routes, wherein residual catalytic species may persist within the coating matrix. Mertoğlu-Elmas and Çınar, for example, could reveal high levels of Pb in the coating layer of paper and cardboard used in the packaging industry through the use of coupled plasma optical emission spectroscopy (Elmas and Çınar 2018).

The main aim of this study was to evaluate FP as a support for powdered adsorbents, with the goal of eliminating the separation issues with powder sorbents, while also elucidating the interactions among FP, PDMS, and MMT using a range of complementary characterization techniques. We used our previous experience in modifying organic paper surfaces by our above-mentioned method to attach MMT, a cost-effective and high-performing MB removal adsorbent in powder form (Almeida et al. 2009), onto FP by PDMS chains with a novel, catalyst-free, and thus environment-friendly approach for its utilization as a self-standing adsorbent for MB removal. To the best of our knowledge, there is no report in the literature on the catalyst-free, PDMS-mediated immobilization of MMT onto FP for dye adsorption applications. Results demonstrate that upon the attachment of only 2.2 wt% of MMT onto FP via PDMS functionalization, the MB removal efficiency almost doubled as compared to pristine FP. Characterization results illustrate that the integrity of the paper-based composite is intact even upon MB removal measurements lasting 15 days. Furthermore, characterization results indicate the presence of potential covalent bonding between PDMS chains and the FP substrate. These findings highlight the potential of paper-based composites in sustainable adsorbent technologies, providing significant environmental and economic benefits.

Contact angle measurements showed that the coated sites of the PDMS/MMT/FP samples, like the pristine FP, absorb water droplets within seconds, so that no contact angle values could be measured, indicating that the samples are highly hydrophilic. Such an immediate adsorption behaviour was also observed for the PDMS/FP sample, which may seem unexpected at first sight. However, the reason for the hydrophilic behaviour of the PDMS/FP sample is the nature of the paper substrate, which is a porous non-woven material composed of cellulose fibers. The pores on the surface of the FP have diameters of up to several tens of microns, which cannot be completely covered by the application of a dilute PDMS solution of 0.7 wt% PDMS in solvent. Therefore, water droplets could still be adsorbed by the paper substrate through the capillary effect (Söz 2022).

Figure 1 presents the XRD patterns of FP, MMT, and PDMS/MMT/FP samples, along with control samples; PDMS/MMT, and PDMS/FP. The positions of the MMT peaks observed at 2θ values of 12.5, 20.1, and 35.1°, and the more intense features at 9.1 and 26.9° are consistent with earlier reports (Wilson et al. 2006; Harish Kumar et al. 2021). After the addition of PDMS to MMT there is no significant change observed in the XRD spectrum of PDMS/MMT control sample, except for a slight decrease in intensity of the peaks at 9.1° and 26.9°. This indicates that the MMT particles cannot be fully separated into their single sheets with the proposed method.

Bild vergrößern

Native cellulose, including cotton-derived filter paper, possesses a semi-crystalline structure comprising ordered crystalline domains interspersed with less ordered amorphous regions. The XRD pattern of FP exhibits characteristic cellulose peaks at 15.2° and 16.6° (corresponding to the 1–10 and 110 peaks, respectively (French 2014)) along with additional reflections at 23.0° (the 200 peak), and 34.8° (the 004 peak) (French 2014; Wang et al. 2017; Lizundia et al. 2019; Kandathil et al. 2019; Alula et al. 2020). Upon coating the FP with PDMS, the intensity of the aforementioned low intensity cellulose peaks decreased, and a similar effect was also observed for the PDMS/MMT/FP composite. Notably, no peaks corresponding to MMT were detected in the XRD spectrum of the PDMS/MMT/FP sample, which is attributed to the relatively low loading of MMT particles on the FP.

FTIR spectra of MMT, PDMS, PDMS/MMT, FP, PDMS/FP, and PDMS/MMT/FP in the regions of 3650–2700 and 1700–600 cm⁻¹ are presented in Fig. 2a and b, respectively. MMT exhibits a characteristic Si–O in-plane stretching feature at 1027.4 cm⁻¹ (Ahmed et al. 2018; Caccamo et al. 2020). PDMS showed distinctive IR bands at 686.2 cm⁻¹ related to Si–O–Si vibrations (Al-Oweini and El-Rassy 2009; Ruan et al. 2023) and at 787.9 cm⁻¹ belonging to Si–CH₃ (Johnson et al. 2013). A band at 1008.6 cm⁻¹ and a band appearing as a shoulder at 1074.5 cm⁻¹ was attributed to Si–O–Si stretching vibrations in PDMS (Meškinis et al. 2007). Bands at 1258.4 cm⁻¹ and 2962.4 cm⁻¹ were assigned to the symmetric CH₃ deformations of PDMS (Si–CH₃) (Ferreira et al. 2013; Johnson et al. 2013), and C–H stretching (Ferreira et al. 2013), respectively.

Bild vergrößern

The FTIR spectrum of the PDMS/MMT control sample (Fig. 2a and b) demonstrated a combination of characteristic bands from both PDMS and MMT. Upon incorporating these two components, the band intensities of PDMS weakened and blue-shifted. Specifically, the C–H stretching band at 2962.4 cm⁻¹ shifted to 2963.7 cm⁻¹, the Si–CH₃ deformation band of PDMS at 1258.4 cm⁻¹ shifted to 1261.1 cm⁻¹. The observed blue shift of the Si–CH₃ deformation band from 1258.4 cm⁻¹ to 1261.1 cm⁻¹ may indicate restricted chain mobility and strengthened Si–C bond environments, potentially due to interfacial interactions between the PDMS matrix and the montmorillonite surface. Such interactions can lead to enhanced bond stiffness or reduced vibrational freedom, as also supported by the harmonic oscillator model, given below in Eq. 10:

where ν̄ is the wavenumber (cm⁻¹), c is the speed of light (cm s⁻¹), k is the force constant of the bond (N m⁻¹), and μ is the reduced mass of the bonded atoms (Nakamoto 2008; Smith 2011). This relationship indicates that even slight shifts in wavenumber can reflect changes in bond strength or interfacial interactions. In our case, a blue shift, which corresponds to an increase in wavenumber, typically signifies a strengthening of the bond or restriction of vibrational freedom due to interfacial effects. Similarly, the Si–CH₃ band at 787.9 cm⁻¹ shifted to 796.8 cm⁻¹, and the Si–O–Si vibration band at 686.2 cm⁻¹ shifted to 692.5 cm⁻¹. Moreover, a broad band centered at 1011.9 cm⁻¹ is emerging because of the overlap of the Si–O in-plane stretching band of MMT (1027.4 cm⁻¹) and Si–O–Si stretching band of PDMS (1008.6 cm⁻¹). In addition, a broad shoulder at 1027 cm⁻¹ is observed, which indicates a slight red shift of the Si–O stretching vibration originally associated with MMT. This red shift suggests a modest weakening or softening of the Si–O bonds within the MMT structure, potentially caused by interfacial interactions with the surrounding PDMS matrix. Such interactions, such as partial coordination, dipolar interactions, or physical adsorption, may modify the local electronic environment of the silicate layers, thereby reducing the effective bond stiffness and resulting in the observed downward shift in vibrational frequency. The occurrence of both blue and red shifts in the characteristic vibrational bands strongly indicates alterations in the local bonding environment, arising from specific interfacial interactions between PDMS chains and MMT particles.

The spectra of FP, PDMS/FP, and PDMS/MMT/FP displayed similar patterns, which are dominated by the features of FP. These FP-based samples contain characteristic –OH bands of cellulose at 3332.5 and 3370.1 cm⁻¹, along with asymmetric and symmetric CH₂ stretching vibrations at around 2899 and 2870 cm⁻¹ (Fig. 2a) (Proniewicz et al. 2001; Schwanninger et al. 2004; Yi et al. 2022; Evren et al. 2024). The bands at 1029.9 and 1054.0 cm⁻¹ were attributed to C–OH stretches of the secondary and primary alcohol groups in the cellulose backbone, respectively (Fig. 2b). The presence of PDMS even after Soxhlet extraction could be evidenced by the presence of weak bands at 2969.1 cm⁻¹ and 2940.1 cm⁻¹, with the latter most likely to be associated with the presence of the Si–CH₃ band of the PDMS. It is noteworthy that the bands at 2969.1 and 2968.1 cm⁻¹ observed in PDMS/MMT/FP and PDMS/FP, respectively, were not present in pristine FP. However, the appearance of a possible Si–O–C band, which would point out the covalent interactions between PDMS chains and cellulose and/or MMT particles, could not be detected due to the overlapping of the Si–O–Si and C–O–C bands in the 1000–1100 cm⁻¹ region and due to low amount of coating with respect to the paper substrate underneath.

To further comment on the possible interactions between FP and PDMS, XPS analysis was performed in the Si 2p, C 1s, and O 1s regions for FP, PDMS/FP, and powder PDMS/MCC control sample. In the C 1s region, the pristine FP displayed four peaks at 284.6, 286.3, 287.9, and 289.2 eV corresponding to C–C and C–H, C–O, C = O and/or C–O–C, and O = C–O, respectively (Fig. 3a) (Chen et al. 2021; Hu et al. 2021). The O 1s region of FP (Fig. 3b) was deconvoluted into three peaks at 531.1, 532.7, and 534.5 eV, which correspond to oxygen atoms in glycosidic linkages between two D-glucose residues (C–O–C), oxygen atoms within the cellulose backbone, and hydroxyl oxygen atoms on the glucose residue ring structure, respectively (Hu et al. 2021).

Bild vergrößern

For the C 1s region of the PDMS/FP sample, the deconvoluted peaks at 283.8, 285.8, and 287.2 eV were assigned to C–Si or C–H (Kim et al. 2018; Jankauskaite et al. 2020), C–O and C=O (Sharma et al. 2007; He et al. 2020), and/or C–O–C bondings (Sharma et al. 2007), respectively (Fig. 3c). The oxidized C peaks might be a result of oxidation of the PDMS surface or sample contamination (Malecha et al. 2010). In the O 1s region of the PDMS/FP sample (Fig. 3d), two peaks at 532.0 and 534.0 eV were identified as Si–O bonds, found in siloxane units and higher oxidation states, respectively (Malecha et al. 2010; Ohkubo et al. 2018). The Si 2p spectrum of PDMS/FP (Fig. 3e) revealed two deconvoluted components: a peak at 101.2 eV, corresponding to silicon bound to carbon atoms, and a peak at 102.5 eV, indicating the silicon bound to oxygen atom found in Si–O–Si backbone (O’Hare et al. 2004; Morent et al. 2007; Malecha et al. 2010). Since XPS analysis is limited to the first about 10–30 atomic layers of a surface, depending on the material (Atzei et al. 2016), it is well-suited for analyzing surface properties within < 10 nm (Krishna and Philip 2022). As our coating thickness exceeds 10 nm, the analysis should be confined mainly to the PDMS-coating layer. However, it should be noted that, due to the roughness of the paper substrate’s surface, achieving a uniform coating thickness throughout the entire sample surface is not feasible. At this stage, it is important to point out that there are studies in the literature revealing the freshly emerged Si–O–C bonds at around 101.5 to 102.5 eV after reacting softwood kraft lignin and aminopropyl/methyl silsesquioxane (Bemerw Kassaun and Fatehi 2024), hexamethyldisiloxane vapor and diamond-like carbon (Meškinis et al. 2007, 2020), or silicones and acrylates (Ye et al. 2017). However, the expected Si–O–C formation at the interface between the cellulose fibers of the FP and the PDMS coating in our study could not be revealed by the XPS measurements on the PDMS/FP (paper-based composite) as a third peak in its Si 2p spectrum. This is most likely due to the insufficient depth of the X-ray analysis and the low number of Si–O–C bridges with respect to those of the Si–O–Si components.

To address this limitation and better reveal the interfacial chemistry, a simplified model powder compound, PDMS/MCC, was investigated. The preparation conditions were held similar to the PDMS/MMT/FP composite. As shown in Fig. 3f, the C 1 s spectrum of the PDMS/MCC model compound exhibits four distinct peaks. The signals at 284.3, 286.2, 287.6, and 288.3 eV are assigned to C–Si/C–C/C–H, C–O/C=O, C–O–C, and O–C=O linkages, respectively. The data are in good agreement with the data obtained in the C 1 s region for pristine FP (Fig. 3a). In Fig. 3g, the O 1 s region of PDMS/MCC was resolved into three peaks at 530.7, 532.6, and 534.5 eV. The data are consistent with the O 1s spectrum of pristine FP shown in Fig. 3b. Slight changes may arise from the contribution of PDMS.

As shown in Fig. 3h, the position of the non-deconvoluted Si 2p peak of the PDMS/MCC is blue-shifted by 0.5 eV relative to the non-deconvoluted Si 2p peak of PDMS/FP (Fig. 3e). This shift toward higher binding energy in PDMS/MCC suggests that a new and more intense feature dominates the spectrum. Therefore, the Si 2p region of this model compound was deconvoluted into three distinct peaks. The peaks at 101.0 eV and 102.2 eV are assigned to Si–C and Si–O–Si linkages, respectively, similar to the features observed for PDMS/FP (Fig. 3e). Notably, the intermediate peak at 101.6 eV may be attributed to Si–O–C bonds, indicating the possible formation of covalent linkages between PDMS and cellulose in this simplified system. These observations are in agreement with previous studies (Ye et al. 2017; Bemerw Kassaun and Fatehi 2024). While not conclusive, this feature may be regarded as supporting spectroscopic evidence for interfacial interactions that could not be resolved in the PDMS-coated FP sample due to the limited surface sensitivity of XPS and the relatively large coating thickness of PDMS on FP. Furthermore, as evidenced by the relatively low intensity of the Si 2p spectrum of PDMS/MCC and the strong resemblance between the C 1 s and O 1 s spectra of PDMS/MCC and pristine FP, it can be concluded that the amount of PDMS remaining on the MCC after Soxhlet extraction is relatively low. This reduced coating thickness of PDMS on MCC was considered advantageous for resolving Si–O–C bonds.

Importantly, these results align well with our earlier Raman spectroscopic studies on MCC/PDMS model compounds, in which the formation of Si–O–C bonds was evidenced by new peaks at 1161 and 806 cm⁻¹, corresponding to asymmetric and symmetric Si–O–C stretching vibrations, respectively (Koşak Söz et al. 2018). These vibrational signals provided compelling confirmation of covalent linkages between PDMS chains and cellulose substrates. We would like to note that the Raman spectra obtained for PDMS/MCC and PDMS/MMT are presented in Fig. S1; however, these spectra display broad, featureless backgrounds. This is attributed to the limited spectral resolution of the conventional Raman spectrometer used in the present study, whereas our previous study (Koşak Söz et al. 2018) employed a confocal Raman spectrometer. Therefore, the present XPS findings, together with our earlier reported Raman data, mutually reinforce the presence of Si–O–C bonding at the PDMS–cellulose interface, validating the interfacial reactivity proposed in this study even when such bonds are not readily detectable in thicker coated systems.

Next, to determine the mass loss patterns of the samples and the coating thickness of the paper-based composites, TGA was performed on pristine FP, PDMS/FP, and PDMS/MMT/FP, along with MMT and PDMS/MMT. Three measurements were performed for each sample to determine the coating thickness, and their average values were calculated. Representative TGA curves showing mass loss (%) as a function of temperature are presented in Fig. S2–S6. The residual masses of the coating and the sheet were calculated, and using the specified densities of the coating materials, the coating thickness was determined (Koşak Söz et al. 2018). The total applied coatings (PDMS + MMT) on the composite paper sheets were theoretically found to be 4.0 ± 0.7 g m⁻², which is approximately 1 µm thickness. Furthermore, calculations show that the loading of MMT in the PDMS/MMT/FP composite is 2.2 ± 0.3 wt%.

Figure 4 depicts typical SEM images of cotton linter paper sheets before (FP) and after the coating application (PDMS/MMT/FP) at various magnifications. Figure 4a–c reveal the pristine FP surface, which is composed of fibrillar domains, the diameters of those range from several nanometers to micrometers. However, the surface topography of FP changed significantly upon the coating process, indicated by the dispersion of MMT particles of various sizes across the surface in the composite (Fig. 4d–f). SEM images revealed the successful incorporation of the inorganic MMT particles on the FP substrate thanks to the practical spray-coating approach.

Bild vergrößern

TEM micrographs (Fig. 5a–d) illustrate the dispersion state of montmorillonite (MMT) particles within the PDMS matrix and its morphology. Thin, individually exfoliated nanosheets are visible across the lacey carbon support (Fig. 5a,b), while other regions display laminated clay layers (Fig. 5c,d). The coexistence of isolated sheets and stacked domains indicates a mixed morphology, pointing out that MMT is only partially exfoliated after incorporation into PDMS (Li and Peng 2018).

Bild vergrößern

To complement the morphological observations, BET analysis was conducted to quantitatively determine the specific surface area (SSA) of the individual components and the resulting composites. The SSA values of MMT, Whatman FP paper, and the PDMS/MMT/FP composite were determined as 203.4, 1.3, and 1.8 m² g⁻¹, respectively. These findings are consistent with previous reports on the surface characterization of MMT-based materials (Biglari et al. 2021), paper substrates (Karchangi et al. 2024), and cellulose-derived materials (Gustafsson and Mihranyan 2016). The observed increase in SSA from FP to the PDMS/MMT/FP composite reflects the contribution of the MMT filler within the composite matrix. This is in line with the established role of BET analysis in tracking surface area changes upon composite formation or post-adsorption modifications (Nazari et al. 2022).

Next, the adsorption performances of FP, PDMS/FP, and PDMS/MMT/FP were investigated based on the effect of contact time at a constant initial concentration of MB (C₀ = 5 mg L⁻¹) using 6 cm × 6 cm paper samples (Fig. 6a and b). Equilibrium was reached within approximately 4 h for all adsorbents. At equilibrium, FP, PDMS/FP, and PDMS/MMT/FP provided MB removal efficiencies of 51.3, 49.3, and 90.0%, respectively (Fig. 6a), at identical testing conditions. The almost equivalent MB removal performance of FP and PDMS/FP indicates negligible adsorption on PDMS, whereas incorporating only 2.2 wt% MMT in the composite significantly enhanced performance, nearly doubling MB removal efficiency. The uptake values (q_t) with increasing contact time are provided in Fig. S7a. It should be noted that the uptake calculation for these adsorbents is challenging because it considers the entire mass of the adsorbent (please see Eq. 1), which is 0.32 g for each sample, despite MMT (only 2.2 wt%, ~ 6.4 mg) being the primary contributor to the performance increase. For PDMS/MMT/FP, a q_e of 1.40 mg g⁻¹ was achieved compared to 0.76 mg g⁻¹ for pristine FP (Table 2), highlighting the significant impact of MMT. Besides, while powdered adsorbents often achieve higher uptake capacities, self-standing materials like these composite papers or other reported monoliths provide easier separation, though at the cost of reduced capacity (Kaya-Özkiper et al. 2022b). Table 3 illustrates this trade‑off: most paper‑based adsorbents and monoliths show lower MB capacities (apart from the sodium‑alginate‑modified FP) while adsorbents such as hydrogel beads, functionalized graphene, and activated carbons deliver much higher uptakes.

Bild vergrößern

Adsorption data in Fig. 6a up to 4 h were analyzed using both pseudo-first-order (Fig. S8) and pseudo-second-order (Fig. S9) kinetic models. Kinetic parameters are detailed in Table 2. The pseudo-second-order plots (Fig. S9) for the samples demonstrated strong agreement, with high correlation coefficients (R² = 0.999). This suggests that the adsorption process is primarily controlled by chemical adsorption, implying the potential exchange and/or sharing of electrons between the adsorbate and the adsorbent (Kaya-Özkiper et al. 2022a). Adsorption data were further analyzed by intraparticle diffusion model (Fig. S10 and Table S1, provided in the Supplementary Materials). In the initial time window (first 60 min of adsorption) the q_t vs t^0.5 line for both PDMS/MMT/FP and pristine filter paper (FP) passes almost through the origin, signifying that external film resistance is negligible. Over this period, intraparticle diffusion appears to be the predominant rate-controlling mechanism, with the composite exhibiting a higher diffusion rate (k_dif = 0.137 mg g⁻¹ min^−0.5) compared to FP (k_dif of 0.081 mg g⁻¹ min^−0.5). After 60 min the behaviours diverge: PDMS/MMT/FP enters a second intraparticle-diffusion stage characterised by a lower slope (k_dif = 0.017 mg g⁻¹ min^−0.5) and a positive intercept (C = 1.10 mg g⁻¹), indicating that diffusion inside the MMT-rich pore network continues. FP, in contrast, shows an almost horizontal line after 60 min of contact time (slope ≈ 0), signifying that adsorption has effectively plateaued.

Next, the MB percent removal (Fig. 6b) and MB uptake (Fig. S7b) of both PDMS/MMT/FP and FP at different initial MB concentrations were evaluated. At the maximum concentration studied (17 mg L⁻¹), the MB uptake was 2.69 mg g⁻¹ for the composite. Figure 6b demonstrates that an increase in the initial MB concentration decreases the MB removal efficiency (%). This is attributed to the high number of adsorption sites that are available at low concentrations, leading to an enhanced interaction between the adsorbent and adsorbate, resulting in a higher MB removal at low concentrations. As the initial concentration of MB increases, adsorption sites start to become more occupied, hindering further adsorption (Kaya-Özkiper et al. 2022a).

The equilibrium data shown in Fig. 6b and Fig. S7b for both the FP and PDMS/MMT/FP samples were fitted to Langmuir and Freundlich isotherm models (Fig. S11, S12). Because MMT is the main capacity‑enhancing component of the composite, powdered MMT was also modeled for comparison (Fig. S13). The resulting R² values reveal that the Langmuir isotherm provides the best fit for PDMS/MMT/FP, whereas pristine FP is well described by both models. Powdered MMT, however, follows the Langmuir isotherm almost perfectly, which accounts for the composite’s Langmuir‑like behavior even though pristine FP conforms to either model. This finding confirms that adsorption proceeds via monolayer coverage, with negligible interactions between sorbed molecules, indicating a surface composed of homogeneous binding sites with equal sorption energies (Kaya-Özkiper et al. 2022a). Fit parameters are provided in Table 4. To examine the influence of pH on adsorption performance, powdered MMT was used as the model adsorbent (Fig. S14). Uptake declined at pH values below 5 and above 7, with the decrease being more pronounced under acidic than under basic conditions. The sharper decline at low pH suggests that deprotonation of surface functional groups (such as –OH) is suppressed, preventing the formation of the negatively charged surface needed to attract the cationic MB dye.

The significant increase in MB adsorption with the incorporation of only 6.4 mg MMT per 0.32 g of paper-based composite (PDMS/MMT/FP) highlights the synergistic combination of MMT’s ideal characteristics as an adsorbent and FP’s effectiveness as a substrate. FP serves as an efficient platform for dispersing MMT particles thanks to its nonwoven, randomly oriented cellulose fibers, which allow water and/or aqueous liquids to pass through and reach the adsorption sites of MMT. In addition to its physical properties, FP has hydroxy groups and deprotonated hydroxy groups on its surface, enabling it to interact with MB, contributing also to the MB adsorption alongside dispersing MMT particles, as evidenced by the good correlation of FP’s kinetic data with pseudo-second-order adsorption kinetics. To investigate the surface-charge characteristics of the PDMS/MMT/FP composite, the point of zero charge of the as-prepared sample was determined as shown in Fig. 7. According to the figure, the PDMS/MMT/FP composite is expected to be readily protonated at pH values below 3.84, well below the pH conditions of our adsorption tests (7.83 for the composite). Therefore, under the experimental conditions, the composite surface is predominantly negatively charged, which accounts for its favorable affinity toward the cationic dye MB.

Bild vergrößern

To assess the MB desorption for regeneration, we immersed spent adsorbents in ethanol and in 0.1 M HCl. Ethanol was largely ineffective (Fig. S15a–c), whereas 0.1 M HCl removed more dye without significantly damaging the composite (FTIR-confirmed; Fig. S16) as shown in Fig. S15d-f. Subsequently, the reusability of PDMS/MMT/FP composites was investigated through multiple adsorption–desorption cycles using MB solutions at initial concentrations of 12 and 17 mg L⁻¹, followed by desorption in 0.1 M HCl (Fig. S17a and b). Initially, the composites exhibited removal efficiencies of 68.7 and 52.2% at 12 and 17 mg L⁻¹ MB concentrations, respectively. However, after one regeneration cycle, these efficiencies dropped significantly to 15.8 and 15.3%, respectively. As regeneration cycles continued, a gradual decrease in adsorption capacity and removal efficiency was observed, eventually approaching a stable value (Fig. S17). This drop in performance indicates the loss of active adsorption sites or possible structural alterations in the composite during repeated regeneration processes.

To test whether the PDMS/MMT/FP composite was intact upon MB removal tests, SEM images were obtained on a composite that was subjected to MB adsorption test conditions lasting 15 days. As shown in Fig. 8a–c, the SEM images confirm that MMT particles remained on the FP surface even after an adsorption test lasting 15 days, which was performed under constant shaking at 150 rpm. EDX analysis further supports this (Table S2, Fig. S18 (as provided in the Supplementary Materials), Fig. 8e), showing a comparable aluminum content caused by MMT in the post-adsorption sample (0.78 wt%) to that before adsorption (0.95 wt%) (Table S2).

Bild vergrößern

To further assess the structural integrity and possible chemical changes following extended exposure to dye solution, FTIR and XRD analyses were performed on the PDMS/MMT/FP composite after 15 days of adsorption (Fig. 9). The FTIR spectrum of the used composite retained the characteristic absorption bands of cellulose-based FP. Notably, a slight blue shift was observed in the –OH stretching vibration band, shifting from 3332.5 cm⁻¹ (as-prepared composite) to 3337.4 cm⁻¹ after MB adsorption. It would suggest that the OH groups play an active part in MB adsorption (Elmorsi et al. 2019). The presence of a band at 2966.7 cm⁻¹ confirms the continued presence of the Si–CH₃ stretching vibration associated with PDMS, indicating that the polymer matrix remains chemically intact after prolonged immersion. The diffraction pattern of the spent PDMS/MMT/FP composite remained largely unchanged compared to the as-prepared sample in Fig. 1, indicating that the overall structural framework of the composite was well preserved throughout the adsorption process. SEM FTIR, and XRD data show that the composite formed by our proposed practical spray method remains intact even after harsh conditions, making it a robust and efficient adsorbent.

Bild vergrößern

This study presents a novel, simple, and catalyst-free process for attaching MMT to FP substrates via PDMS chains, resulting in self-standing, mechanically robust paper-based composites. Contact angle measurements confirmed that both pristine FP and the PDMS/MMT/FP composite absorb water within seconds, underscoring their hydrophilicity. XRD showed that the structure of FP remains intact after MMT incorporation via PDMS, while TGA indicated a combined PDMS and MMT coating of 4.0 ± 0.7 g m⁻² and a MMT loading of 2.2 ± 0.3 wt% in the composite. Spectroscopic data suggest interactions among MMT, PDMS, and FP cellulose fibers, including the possible formation of Si–O–C bonds between PDMS and cellulose. Furthermore, SEM images showed that MMT particles of various sizes are uniformly dispersed across the composite surface. Under identical test conditions (100 mL of 5 mg L⁻¹ MB solution, 6 cm × 6 cm sheets, ≈ 0.32 g adsorbent), pristine FP removed 51.3% of MB, whereas the PDMS/MMT/FP composite achieved 90.0% removal. This nearly two-fold improvement obtained with only 2.2 wt% MMT highlights FP’s effectiveness as a substrate. The composite closely followed pseudo-second order kinetics indicating chemisorption and fitted better to the Langmuir isotherm. Characterization of fresh and spent adsorbents, point‑of‑zero‑charge measurements, and pH‑dependent performance tests all indicate that the cationic dye adsorbs primarily through interactions with –OH groups on both the FP substrate and the incorporated MMT. Even after continuous shaking at 150 rpm for 15 days in MB solution, the PDMS/MMT/FP composite remained intact, evidenced by SEM images showing persistent MMT particles, and EDX analysis detecting comparable Al levels in the as-prepared (0.95 wt%) and post-adsorption (0.78 wt%) samples. These findings demonstrate that the environmentally friendly, catalyst-free route developed here offers a simple way to link MMT to FP via PDMS, yielding self-standing adsorbent sheets that can be readily separated from water without filtration or centrifugation. The methodology opens avenues for future studies aimed at incorporating a broader range and higher loadings of inorganic materials to achieve even greater enhancements in MB removal performances. Nevertheless, we acknowledge that employing THF and prolonged Soxhlet extraction, methods not fully aligned with green-chemistry principles, was necessary to ensure coating quality and purity (Koşak Söz et al. 2018). Thus, identifying greener alternatives will be a priority in future work. Another future work will extend the adsorbent to multi-component systems, including anionic dyes and mixed heavy-metal/dye solutions, to validate its practical application.