Efficient adsorption of perfluorooctane sulfonate and perfluorobutane sulfonic acid from aqueous solution onto amine-functionalized diatomite granules

Abstract

Per- and polyfluoroalkyl substances (PFAS), often referred to as 'forever chemicals,' are hazardous, bio-accumulative compounds that pose serious risks to both human health and the environment. PFAS are characterized by the presence of strong carbon–fluorine (C-F) bonds, which render them highly resistant to degradation and environmentally persistent. To address these challenges, (3-aminopropyl) triethoxysilane (APTES)-modified diatomaceous earth granules (APTES-DE) were investigated as a potential alternative adsorbent. FTIR, SEM-EDX, XRD, XPS, and BET analyses were used to characterize the APTES-modified diatomaceous earth granules. The adsorption performance of granules synthesized under varying formulation conditions was evaluated against two commonly detected PFAS compounds in contaminated water, perfluorobutane sulfonic acid (PFBS) and perfluorooctane sulfonate (PFOS). The effectiveness of APTES-modified diatomaceous earth (APTES-DE) is evaluated by investigating the adsorptive removal of model long-chain and short-chain PFAS compounds, PFBS and PFOS, under various experimental conditions. The maximum removal efficiency of PFAS reached 92.43% for PFOS and 89.90% for PFBS. The adsorption isotherms of both PFOS and PFBS on APTES-DE were best described by the Freundlich models, indicating the multilayer adsorption on heterogeneous surfaces. Furthermore, regeneration experiments demonstrated that the adsorbent maintained high reusability over five successive sorption-desorption cycles, with only slight declines in adsorption performance of 4.56% for PFOS and 3.62% for PFBS. Adsorption experiments indicate that electrostatic interactions, hydrophobic interactions and hydrogen bonding were the primary mechanisms driving PFAS removal. The lower adsorption efficiency observed for the short-chain PFAS compound, PFBS, compared to PFOS, suggests a positive correlation between adsorption performance and the length of the C–F chain.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-41388-4.

Introduction

A class of emerging contaminants recognized for their remarkable chemical and thermal stability is perfluoroalkyl and polyfluoroalkyl compounds (PFAS)¹. Due to their distinctive properties, PFAS are widely utilized in Consumer products such as paper goods, textiles, and furniture surface coatings². The stable chemical structure, high electronegativity, and small atomic size of fluorine render PFAS highly persistent in the environment, with long biological half-lives and significant accumulation potential³. Owing to their diverse applications and resistance to degradation, these compounds are extensively distributed in the environment⁴.In addition to their widespread prevalence in the environment, PFAS are hazardous and bioaccumulative, which urges remedial measures. These substances have been associated with several harmful health outcomes, such as chronic illnesses and decreased immunization efficiency⁵. Furthermore, they have been known to cause several serious health hazards affecting humans, including cancer, fertility issues, and premature birth⁵. The average concentration of PFAS in wastewater originating from industries involved in the production of aqueous film-forming foams and waterproofing chemicals was found to exceed 1000 ng/L³.

Treatment of PFAS-contaminated waters is typically not possible using conventional treatment techniques used in treatment plants, such as filtration, sedimentation, and coagulation and flocculation⁶. Although there has been evidence of success in the laboratory or demonstration setting in breaking down per- and polyfluoroalkyl substances (PFAS) using particular destruction techniques, including sonolysis and electrochemical treatment, these technologies are energy-intensive and better suited for concentrated, small-volume waste streams^6,7. Granular activated carbon (GAC) is a suitable and viable solution for removing PFAS from contaminated waters⁸. However, GAC has some drawbacks, including a decreased ability to remove short-chain PFAS, reduced efficiency due to background components such as anions and naturally occurring organic matter⁹, and Inefficient adsorption behavior that triggers rapid breakthrough¹⁰. Moreover, non-renewable raw materials are required to produce the most widely used GAC, which is made from bituminous coal. Thus, it is imperative to investigate more environmentally friendly and long-lasting adsorbents for PFAS cleanup, with an emphasis on faster kinetics and enhanced performance from renewable sources. At an equilibrium concentration of 1 µg L⁻¹, ion-exchange polymers exhibited the highest sorption capacity, followed by non-ion-exchange polymers, while granular activated carbon showed the lowest performance. In contrast, at a much lower equilibrium concentration of 100 ng L⁻¹, non-ion-exchange polymers demonstrated superior adsorption capacity compared to the other adsorbents⁸. In a reported study, granular activated carbon (GAC) demonstrated a markedly higher maximum Langmuir sorption capacity for PFOS (123.5 µmol g^-1) compared with PFBS (48.3 µmol g^-1), reflecting the stronger interaction of the longer-chain PFOS with the sorbent surface. The preferential uptake of PFOS over PFBS was primarily attributed to its greater hydrophobicity and enhanced surface affinity associated with the longer perfluorinated carbon chain¹¹.Selecting a suitable adsorbent is a crucial step in developing an effective adsorption process. Various adsorbents, including activated carbon, chitosan composites, metal-organic frameworks (MOFs), carbon nanotubes, porous silica materials, and graphene oxide, have been widely reported for the removal of PFAS¹². Despite their widespread use, conventional sorbents exhibit several limitations, including low affinity for short-chain PFAS, slow adsorption kinetics, and a lack of selectivity. Consequently, significant research efforts have focused on increasing the adsorption capacity, regeneration ability, and selectivity of carbonaceous materials for PFAS removal. Among these, amine-functionalized adsorbents have demonstrated improved performance due to the presence of positively charged active sites formed by amine groups, which facilitate strong electrostatic interactions with negatively charged PFAS species¹³. (3-Aminopropyl) triethoxysilane (APTES), an organosilane compound, is known for its ability to form strong interactions with hydroxyl and oxygen-containing functional groups present on silicate and biomass-based materials¹⁴. Recent investigations highlight the potential of APTES-modified silica materials as effective adsorbents for the remediation of aqueous solutions, achieving significant removal of mercury (Hg) and cadmium (Cd)¹⁵. Hydrophobic interactions have primarily driven the removal of PFAS compounds such as PFOA and PFOS, with amine-functionalized adsorbents often exhibiting high adsorption capacities¹⁶.

Diatomaceous earth (DE), a naturally occurring, porous siliceous material, has shown potential for removing various contaminants, including bisphenol-A, a common contaminant of emerging concern (CEC), as well as heavy metals and chemically modified dyes. Despite this, there is limited data on its effectiveness in removing PFAS from water. To bridge this knowledge gap, this study focuses on the removal of PFAS (PFBS and PFOS) from aqueous solutions through a fixed-bed reactor using APTES-functionalized diatomaceous earth granules as the adsorbent. To date, no comprehensive work has reported on the adsorption behavior of PFAS onto APTES-functionalized diatomaceous earth granules, highlighting the novelty of the present study. Among physical wastewater treatment methods, adsorption is favoured for its ease of application, efficiency, and ecological friendliness¹⁷.

Diatomaceous earth (DE) is a siliceous biogenic material derived from the fossilized remains of diatoms, microscopic aquatic plants. DE is a naturally occurring amorphous material distinguished by its unique physico-chemical properties, including high thermal conductivity, high porosity, fine particle size, permeability, excellent adsorption capacity, and a large surface area that ranges from 10 to 30 m²/g^18,19. Application of DE based adsorbents for removing pharmaceutical compounds is not well-documented, despite reports of their effectiveness in eliminating bisphenol-A, a common contaminant of emerging concern (CEC), as well as heavy metals and dyes, particularly when chemically modified. Due to its natural abundance, DE is a cost-effective adsorbent, and its porous structure further enhances its capacity to trap a wide range of contaminants. Composed primarily of amorphous silica (SiO₂·nH₂O), DE exhibits a highly porous silicon-oxygen structure, with void spaces accounting for 80-90% of its volume.²⁰. However, the inherent properties of perfluorooctanesulfonic acid (PFOS) limit the effectiveness of unmodified DE as an adsorbent. PFOS exhibits unique amphiphilic behavior, featuring a hydrophilic polar functional group and a fluorinated hydrophobic tail. Although raw diatomite is highly porous, it offers limited reactive sites for chemical interactions. Nevertheless, DE can be cost-effectively modified to enhance its adsorption capacity for PFOS. For instance, APTES is used as a surface modifier to introduce hydrophobic characteristics suitable for PFAS adsorption. Previous studies have shown that surface modification markedly improves PFOS adsorption performance, with modified diatomaceous earth and zeolite achieving adsorption capacities of 14.1 and 25.5 mg g^-1, respectively, compared to significantly lower capacities for their unmodified materials (4.72 and 0.39 mg g^-1, respectively)²⁰.The present study introduces a straightforward synthetic approach for developing an efficient DE-based silica adsorbent for PFAS removal. In this study, APTES-functionalized diatomaceous earth granules were employed for adsorption of two PFAS species, namely PFBS (short-chain) and PFOS (long-chain), from aqueous solutions. To synthesize APTES-DE, ground waste was functionalized using 3-aminopropyltriethoxysilane (APTES). APTES functionalization is commonly used with inorganic materials such as silica and aluminosilicates. Grafting APTES aims to enhance the adsorption performance of the modified material. Moreover, the role of this modification in improving PFAS adsorption is well understood, as it facilitates strong interactions between PFAS molecules and the surface -NH₂ groups of DE-APTES. The most available data on removal efficiency is primarily reported from batch adsorption experiments, which hampers accurate prediction of system performance, especially during scale-up. In contrast, continuous adsorption systems are preferred for full-scale applications due to their operational simplicity, ability to handle large volumes of wastewater, and scalability. Adsorbent-packed fixed-bed columns are widely used to investigate the dynamic behavior and practical applicability of adsorbent-adsorbate systems in continuous flow operations. The selection and application of adsorbents in these columns are influenced by several factors, including adsorption efficiency, pollutant removal performance, and operational challenges such as clogging and channelling.

Despite increasing interest in developing adsorption-based technologies for PFAS removal, diatomaceous earth modified with APTES has not been extensively or systematically investigated as an effective adsorbent. Significant research gaps persist in understanding its adsorption behaviour toward both long-chain (PFOS) and short-chain (PFBS) PFAS, specifically the contribution of grafted amine groups in modulating PFAS surface interactions, and the material’s long-term regeneration and reuse potential across multiple adsorption–desorption cycles. Furthermore, most previous investigations have primarily employed batch experiments, with limited exploration under fixed-bed column conditions that better reflect real-world treatment systems. To bridge these knowledge gaps, the present work introduces a cost-effective, amine-functionalized diatomaceous earth adsorbent. It systematically evaluates its adsorption capacity, interaction mechanisms, and reusability for PFAS removal from water.

In this study, laboratory-scale fixed-bed adsorption columns packed with APTES-modified diatomaceous earth (DE) granules were investigated to assess their efficiency in PFAS removal. The resulting breakthrough curves are analyzed, and the experimental results are evaluated using a range of established adsorption models to evaluate system dynamics and predictive accuracy. The findings from this investigation may contribute to the development of more efficient strategies for eliminating PFAS from contaminated water.

Materials and methods

Synthesis of amine-modified diatomite (APTES-DE) granules

DE was procured from Sigma-Aldrich (analytical grade). To facilitate granulation and enable easier spray application 40% silica solution was prepared by diluting colloidal silica in water. A minimal amount of this binder was used to ensure effective agglomeration. Manual agglomeration of diatomaceous earth powder was carried out in a plastic container by uniformly spreading approximately 10 g of the powder, followed by the addition of a 10% binder solution prepared by diluting 40% silica solution. The formed granules were pre-dried in an oven at 50 °C for 24 hours and subsequently sintered at 750 °C for 10 minutes in a muffle furnace to enhance structural integrity. Surface silylation of sintered DE granules was performed by first drying the granules for 12 hours at 120 °C to remove surface moisture and enhance the availability of silanol groups. After cooling in a desiccator, approximately 1 g of dried granules was added to a solution of 4 mL of APTES dissolved in anhydrous toluene (50 mL). The mixture was ultrasonically dispersed for 30 minutes to ensure uniform contact, followed by refluxing at 120 °C for 20 hours under continuous stirring (400 rpm) in a nitrogen atmosphere, with a calcium chloride drying tube attached to maintain anhydrous conditions²¹. Upon completion, the solid product was separated by filtration, thoroughly washed with fresh toluene multiple times to remove unreacted silane, and then cured overnight at 90 °C in a drying oven to promote covalent bonding and remove residual solvents.

Amine-modified diatomite granules (APTES-DE) characterization

SEM, a primary technique for material characterization, was used to provide detailed insights into the morphological features of the samples at the nanoscale. For SEM analysis, a microtome or sharp razor blade was used to extract a small section of the sample from the bulk material. The elemental composition of the adsorbent surface was further analyzed using EDAX coupled with SEM (Philips, Netherlands; EDAX XL-30), operated at an accelerating voltage of 15-25 kV. The crystalline structure of the material was investigated by XRD using a Rigaku diffractometer (Cu-Kα, λ = 1.54 Å) at a scan rate of 5°/min over a 2θ range of 10°–80°. XPS was used to examine the elemental composition and chemical states on the surface of APTES-modified diatomaceous earth granules, using a Thermo Fisher Scientific K-Alpha Plus instrument with an Al Kα X-ray source. Additionally, FTIR spectroscopy (Perkin Elmer, Spectrum GX) was conducted to investigate changes in surface functional groups of the adsorbent materials. The pore size distribution and specific surface area (SSA) were determined using nitrogen adsorption–desorption measurements at 77 K with an ASAP 2020 instrument (Micromeritics, USA)¹⁷.

pH_PZC determination

The pH_PZC of APTES-DE was determined using the salt addition method, providing insights into its surface charge behavior²². NaCl solutions with pH values from 2 to 10 (adjusted using 0.1 M HCl/NaOH) were prepared and mixed with the adsorbent (1 g/L). The mixtures were equilibrated at 30 °C for 24 h, after which the final pH values were recorded.²³.

Batch adsorption

At 25 °C, batch adsorption tests were performed with 0.4 g of APTES-treated diatomite in 80 mL of PFAS solution (PFOS and PFBS) (25–100 mg/L) within 100 mL Erlenmeyer flasks. The initial pH (~8.0) corresponded to the natural solution pH. Although PFAS typically occurs in the environment at ng/L-µg/L levels, for example, up to 9700 µg/L has been reported in point-source groundwater, the sorption capacity of the materials was evaluated under standard experimental conditions at higher concentrations. The mixture was agitated at 120 revolutions per minute in a thermostatic shaker for the optimal duration. After the adsorption process, the remaining PFAS concentration was analyzed. Using an 11 μm filter paper, the adsorbent was removed from the mixture.

This aspect is crucial for adsorbents that may exhibit higher sorption capacities in concentrated media compared to environmentally relevant concentrations. Organoclays and other organically modified materials have previously been reported as effective PFOS adsorbents^24,25. Even with relatively low adsorbent dosages, such materials can facilitate the removal of substantial amounts of PFOS. Analysis was done using an Agilent high-performance liquid chromatography system, equipped with a mass-selective detector, a binary pump, and an autosampler. A column, C18 (XBridge of dimension- 4.6 × 250 mm, Agilent, USA) have been used. The mobile phases consisted of methanol (A) and 10 mM ammonium acetate (B), applied in programmed ratios. Separate gradient elution programs were used for PFOS and PFBS due to differences in hydrophobicity. For PFOS, the methanol content was increased from 60% to 95%, while PFBS was analysed using a higher initial aqueous phase, with methanol increased from 30% to 80%. Methanol (A) and 10 mM ammonium acetate (B) served as the mobile phases, and the gradient programs are detailed in Suppl. Table S1 and S2. For each 10 μL injection, the column temperature, set at 40 ± 0.5 °C, and the flow rate of 0.4 mL/min. The percentage removal percentage (R%) and adsorption capacity (Qe) were calculated using the equation 1 and 2.

1

2

where, C_o and C_f indicate the initial and final concentrations of PFAS (in mg/L) within the solution; V is the solution volume (L), M is the adsorbent mass (g), and Qe (mg g^-1) the equilibrium adsorption capacity for PFAS onto the adsorbent

Adsorption isotherm and kinetic modeling

Non-linear adsorption isotherm model

Several adsorption isotherms were employed to study the interaction between PFAS and the adsorbent surfaces. The non-linear isotherm models, shown by equations. (3) & (4), include the Freundlich and Langmuir.

3

4

Where n and K_F are the Freundlich constants, K_L is the affinity constant (L mg^-1), K_T (L mg^-1) is the binding constant, and q_me denotes the maximum adsorption capacity(mgg^-1,).

Non-linear adsorption Kinetic models

Understanding whether adsorption follows physisorption or chemisorption, along with the factors influencing the adsorption rate, requires a thorough knowledge of kinetics²⁶. A nonlinear equation, as in equations 5 and 6 is used for fitting PFO and PSO kinetic models.

5

6

In the kinetic models, K₁ (min^-1) describe the PFO rate constant, q_t (mgg^-1) refers to the amount of adsorption at time t, K₂ (g mg^-1 min^-1) denotes the PSO rate constant and q_e (mgg^-1) the equilibrium adsorption.

Experimental setup for fixed-bed adsorption

The experimental setup consisted of a perspex column with a height of 25 cm and an interior diameter of 2.5 cm. To prevent the loss of adsorbent during filtration, glass wool was inserted at both ends of the fixed-bed column to secure the adsorbent and provide support. A continuous downward influent flow was achieved in the column using a peristaltic pump. At defined time intervals, effluent samples were collected for analysis. To optimize adsorption performance, variations in bed height and adsorbent particle size were systematically investigated. At the beginning of each experiment, PFAS molecules are effectively adsorbed within the column, resulting in an effluent concentration close to zero. As the adsorption sites gradually become occupied, the effluent concentration begins to increase, marking the breakthrough point, typically defined as 5% of the influent concentration. Beyond this point, the concentration in the outflow continues to rise until it approaches the exhaustion point, where the inlet and outlet PFAS concentrations become nearly equal (95% of the influent concentration). Concentration values measured over time were used to construct breakthrough curves (BTCs), which were then analyzed under varying operational conditions, as summarized in Table 3.

Table 3.

Calculated results for the non-linear pseudo-first order and pseudo-second order kinetic model parameters for PFBS and PFOS.

Kinetic model	Parameters	PFBS	PFOS
Kinetic-1st order	qe	1.82	3.38
	K1	0.06	0.06
	R2	0.96	0.92
Kinetic-2nd order	Qe	2.25	4.14
	K2	0.03	0.01
	R2	0.99	0.97

Breakthrough sorption capacity

To evaluate the kinetic parameters governing column adsorption dynamics, analysis was performed on breakthrough curves illustrating C_t/C_o as a function of time (t). The breakthrough point is defined at C_t/C_o = 0.1, indicating 90% removal of influent PFAS, while column exhaustion is reached at C_t/C_o = 0.95, corresponding to 95% concentration of the influent in the effluent. At C_t/C_o = 1, the column is effectively saturated, and the adsorption process is considered complete.

The area under the breakthrough curve, as defined in equation (7 and 8), is used to estimate the adsorption capacity of the column.

7

8

In this context, qtotal denotes the total amount of PFAS adsorbed (mg), C_o/C_t represents the influent-to-effluent concentration ratio, A denotes the integrated area under the breakthrough curve,Q denotes the flow rate (mL/min) and t refers the time (min). Furthermore, equation (9) is used to calculate the total amount of PFAS accumulated in the column, denoted as m_total (mg).

9

Equations (10) and (11) are used to compute column q_e, (mg/g) and the percentage removal (%), respectively.

10

11

Here, M(g) denotes the adsorbent mass and Z (cm) denotes the fixed-bed height

Results and discussions

The SEM micrographs in Figure 1 revealed that the pore widths were approximately 0.5 µm. As expected, EDS analysis confirmed that silicon and oxygen were the predominant elements in the DE. Notably, the frustule structure remained intact after APTES functionalization. However, the biologically derived silane and the porous nature of the DE contributed to particle agglomeration, resulting in the SiO₂ particles being clustered together, with the frustules partially enclosing and surrounding the pores Figure 1, (A-D). The two main constituents in the APTES modified diatomite structure of the diatomite sample were Si and O (Figure 2a & 2b). Additionally, minerals have trace quantities of K, C, Na, and Fe. Consistent with previous reports on other DE samples, trace metals such as Na, Ca, and Fe, were also detected²⁷. These elements, likely present as oxides, are attributed to the geological origin of the DE. The infrared spectrum of raw DE (Figure 3) shows broad absorption region between 3100 and 3700 cm^-1, with distinct peaks at 3502, 3592, 3634, 3682, and 3741 cm^-1, indicate O-H stretching vibrations, likely originating from surface silanol hydroxyl groups or adsorbed water molecules. Furthermore, the two absorption bands at 2930 and 2855 cm^-1 are attributed to the stretching vibrations of C–H groups²⁸.The prominent absorption bands observed in APTES-DE at around 2931 and 2887 cm^-1 correspond to the asymmetric and symmetric stretching vibrations of CH₂ groups, respectively²⁷.Additionally, the characteristic asymmetric and symmetric stretching vibrations of methylene groups are observed in the regions 2935–2915 and 2865–2845 cm^-1, respectively, and the methyne (–CH–) stretching vibration is identified in the range of 2900–2880 cm^-1 in both raw DE and APTES-DE spectra²⁹. The partial structural organization of altered SiO₂ toward a cristobalite-like phase, together with Si–O bending deformation modes, is evidenced by weak absorption bands observed in the range of approximately 690 and 530 cm^-1 respectively in both raw DE and APTES-modified DE³⁰.The absorption band observed at around 1030 cm^-1 is assigned to the Si–O–Si stretching vibrations of the siloxane network in APTES-modified diatomaceous earth (APTES-DE)³⁰. The absorption bands observed at approximately 1550 and 1650 cm^-1 in the FTIR spectrum of APTES-modified DE are attributed to N–H bending vibrations of secondary amine structures, confirming the successful grafting of amine functionalities from APTES onto the DE surface²⁹.XRD analysis (Figure 4a) indicated that APTES-DE was primarily composed of distinct diffraction peaks characteristic of crystalline SiO₂, indicating its high crystallinity. The pattern confirms the composition and crystallite phase purity of both APTES-DE. Diffraction peaks at 22.0°, 28.4°, 31.5°, and 36.1° for the purified diatomite correspond to the (101), (111), (102), and (200) planes of cristobalite-type crystalline SiO₂ (JCPDS 39-1425; a = 4.973 Å, b = 4.973 Å, c = 6.924 Å)³¹. The pH_PZC of raw diatomaceous earth was 6.5, reflecting its siliceous nature. Following APTES modification, the pH_PZC increased to about 8.2, attributable to the incorporation of protonable amine groups on the surface. This upward shift shows the modified adsorbent surface becomes positively charged at neutral pH, thereby strengthening electrostatic interactions with negatively charged PFAS molecules. Figure 4b. Shows the pH_PZC for raw and APTES-DE, demonstrating a shift from 6.5 to 8.2 after functionalization.The BET surface area and pore volume of DE granules and APTES-DE are summarized in Table 1. The DE exhibited a pore volume of 0.56 cm³/g and a surface area of 34.62 cm³/g. In contrast, APTES-DE exhibited a reduced pore volume of 0.24 cm³/g and a lower surface area of 18.24 m²/g. The decrease in specific surface area (SSA) after amine functionalization is typically attributed to partial pore blockage and coverage of the internal surface by grafted amine groups, which hinders N₂ adsorption during BET analysis²⁰.

Fig. 1.

Morphological properties of APTES-DE granules. (A-B) Scanning electron microscopy of the synthesized APTES-DE granules at 10.0 μm (C-D) Scanning electron microscopy of the synthesized APTES-DE granules at 100.0μm.

Fig. 2.

Elemental composition of APTES-DE granules. (a) EDAX analysis of APTES-DE granules before PFAS adsorption. (b) EDAX analysis of APTES-DE granules after PFAS adsorption.

Fig. 3.

FTIR spectra of Raw diatomaceous earth (DE) and APTES-DE granules in the wavenumber range of 500-4000 cm^-1.

Fig. 4.

(a) XRD spectrum of APTES-DE granules. (b) pHpzc of raw diatomite and APTES-modified diatomite.

Table 1.

The BET-specific surface area and pore volumes of raw diatomaceous earth (DE) granules and APTES-modified diatomaceous earth granules (APTES-DE).

Adsorbent	BET-specific surface area (m2/g)	Pore volumes (cm3/g)
Raw diatomaceous earth (DE)	34.62	0.56
APTES-modified diatomaceous earth granules (APTES-DE)	18.24	0.24

The survey scan spectrum for X-ray photoelectron spectroscopy (XPS) of APTES-DE powder confirms the presence of C 1s, O 1s, N 1s, Si 2p, and Si 2s peaks, indicating successful surface modification (Figure 5a). Deconvolution of the N 1s spectrum revealed two distinct components: a peak centered at approximately 399.2 eV, attributed to neutral amine (–NH₂) groups, and a second peak at around 401.7 eV, corresponding to protonated amine species (–NH₃⁺) introduced by APTES³² (Figure 5b) . An increased intensity in the C 1s signal at approximately 284.6 eV reflects the presence of alkyl chains from the APTES molecule²¹. The Si 2p (100eV) and O 1s (530 eV) peaks are characteristic of the silica matrix of diatomaceous earth¹⁴. These spectral features collectively confirm the effective grafting of amine groups onto the DE surface.

Fig. 5.

(a) XPS survey spectra of APTES-DE granules (b) N1s spectra of APTES-DE granules.

Adsorption isotherm models

The interplay between the structural characteristics of the adsorbent and the physicochemical properties of the adsorbate governs the adsorption of PFAS (PFOS and PFBS) in aqueous solutions. In this study, PFOS and PFBS concentrations were maintained within the range of 25-200 mg/L, well below their critical micelle concentration (CMC) of approximately 500 mg/L in distilled water. This ensured that adsorption occurred predominantly as monomeric species rather than micellar aggregates, allowing a clearer evaluation of adsorbent-adsorbate interactions. At such sub-CMC levels, PFOS molecules are expected to remain free in solution rather than forming aggregates. The Langmuir and the Freundlich isotherm models, which facilitated the evaluation of adsorption capacity, surface heterogeneity, and adsorbate-adsorbent interactions. In the Freundlich isotherm, the parameter n represents adsorption intensity; its value of 0 < n < 10 confirms an effective and favorable adsorbate-adsorbent interactions, often linked to multilayer adsorption on heterogeneous surfaces²⁰(Figure 6a and 6b). In addition, the Langmuir model provided a better fit to the adsorption data, and the Langmuir constants (K_L) were less than 1, indicating favorable adsorption. Table 2 summarizes the calculated parameters of the isotherm models for both PFOS and PFBS. Langmuir modeling revealed maximum adsorption capacities of 26.47 mg g^-1 for PFOS and 17.97 mg g^-1 for PFBS.

Fig. 6.

(a) Non-linear Langmuir and Freundlich and isotherm fitting curves for the adsorption of PFBS on APTES-modified diatomaceous earth granules. (b) Non-linear Langmuir and Freundlich and isotherm fitting curves for the adsorption of PFOS on APTES modified diatomaceous earth granules.

Table 2.

Adsorption isotherm parameters for PFBS and PFOS adsorption using an isotherm model.

Isotherm models	Model equation	Parameters	PFBS	PFOS
Langmuir isotherm		Qmax (mg g−1)	17.97	26.47
		KL (L mg−1)	0.11	0.04
		R2	0.95	0.90
Freundlich isotherm		1/n	0.3	0.40
		KF (mg g−1) (L mg−1)1/n	4.46	3.66
		R2	0.97	0.94

Kinetic studies

Kinetic data were analyzed through PFO and PSO models to provide insights into the adsorption mechanism and the processes governing the rate of adsorption. For PFOS, the experimental data fitted well to PSO kinetic model (R² = 0.97), indicating that its adsorption likely involved a combination of physisorption and chemisorption (Figure 7a and 7b). Similarly , PFBS adsorption on APTES-DEalso showed a stronger affinity to the PSO model (R² = 0.99), consistent with kinetic behaviors reported for PFBS adsorption on other adsorbents (Figure 7b). Table 3 summarizes calculated results for the non-linear pseudo-first order and pseudo-second order kinetic model parameters for PFBS and PFOS.

Fig. 7.

(a) Non-linear kinetic adsorption fitting for the adsorption of PFBS on APTES-modified diatomaceous earth granules. (b) Non-linear kinetic adsorption fitting for the adsorption of PFOS on APTES-modified diatomaceous earth granules.

Fixed-bed adsorption study

Effect of bed height

Column filled adsorbent performance under dynamic flow conditions was assessed using an influent PFAS solution with an initial concentration of 25 mg/L and a controlled flow rate of 2 mL/min. Breakthrough curves (BTCs) were generated for columns with varying bed heights (3, 5, and 7 cm). As expected, a rise in bed height corresponded to a greater amount of adsorbent within the column. Among the tested configurations, the 7 cm bed height exhibited the slowest breakthrough, indicating more efficient adsorption (Suppl. Fig. 1 and Suppl. Fig. 2.). Experimental results showed PFAS removal efficiencies of 82.4%, 90.2%, and 92.3% with corresponding bed heights of 3, 5, and 7 cm, respectively. In contrast, lower bed heights led to earlier breakthrough and faster exhaustion of the adsorption column. A substantial increase in bed height consistently enhanced column performance, resulting in lower PFAS concentrations in the effluent. The delayed exhaustion at higher bed heights allowed the adsorbent to be used for longer durations without replacement. In contrast, decreasing the bed height resulted in fast adsorbent saturation, which reduced overall PFAS removal performance. Supp l. Fig. 1 & Suppl. Fig. 2 presents the breakthrough behavior for the fixed-bed adsorption of PFBS (short-chain) and PFOS (long-chain) using APTES-modified diatomaceous earth (APTES-DE), highlighting the effects of bed height and column performance. The experimental data reveal that none of the systems reached full bed saturation (C_t/C_o = 1), indicating that complete breakthrough was not achieved within the tested operational time. A delayed breakthrough and steeper curve were observed for PFOS, reflecting its stronger interaction with the adsorbent and higher adsorption capacity. In contrast, PFBS exhibited an earlier breakthrough with a more gradual slope, suggesting weaker binding affinity and more rapid saturation primarily due to its shorter carbon chain. This behavior emphasizes the influence of adsorption mechanisms: PFOS removal is largely driven by hydrophobic interactions, whereas PFBS relies more on electrostatic attraction, which is more sensitive to variations in pH and surface charge. Overall, the breakthrough profiles demonstrate that APTES-DE demonstrates higher effectiveness for long-chain PFAS removal, and optimizing variables, including the flow rate and bed height, is essential for improving adsorption efficiency, particularly for short-chain compounds like PFBS. Studies have shown that in anion exchange and GAC adsorbents, short-chain PFASs are displaced by long-chain PFASs, leading to poor removal of the shorter-chain compounds³³. Furthermore, the removal efficiencies of shorter-chain PFASs in wastewater are consistently reported to be low³⁴.

Effect of flow rate

At the pilot or industrial scale, flow rate plays a crucial role in assessing the performance of treatment processes operating under continuous conditions. As illustrated by the breakthrough curves in Suppl. Fig. 3 and Suppl. Fig. 4, the effect of varying feed flow rates (1, 1.5, and 2 mL/min) was investigated while keeping the initial concentrations of PFOS and PFBS at 25 mg/L and maintaining the bed height constant at 7 cm. The breakthrough curves indicate that adsorption was initially rapid and consistent at all flow rates due to the abundance of available binding sites capable of capturing PFAS molecules. As these sites gradually became occupied, the adsorption efficiency declined. An increase in flow rate resulted in steeper breakthrough curves, reflecting reduced breakthrough times and lower PFAS removal efficiency. Suppl. Fig. 3 and Suppl. Fig. 4 illustrates the breakthrough behavior of PFBS (short-chain) and PFOS (long-chain) in fixed-bed adsorption using APTES-DE, demonstrating the influence of bed height on column performance. In all experiments, the bed depth and PFAS concentration were maintained at 7 cm and 25 mg/L, respectively. At the lowest flow rate (1 mL/min), the breakthrough curves showed a more gradual breakthrough compared to those obtained at 1.5, and 2 mL/min. In contrast, at the highest flow rate (9 mL/min), the BTC displayed a sharp initial rise during column operation. Based on the experimental results, PFAS removal efficiencies of 85.3%, 90.7%, and 80.7% were achieved at feed flow rates of 1, 1.5, and 2 mL/min, respectively. Higher influent flow rates at equilibrium resulted in elevated unadsorbed PFAS levels, owing to shortened adsorbent-adsorbate contact time. The highest flow rate (2 mL/min) significantly limited this interaction, leading to incomplete adsorption and the development of steep breakthrough curves during the initial reaction phases. In contrast, lower flow rates (1 and 1.5 mL/min) allowed for longer residence times, enhancing PFAS capture. Over time, the breakthrough curve slope decreased, indicating diminished adsorption capacity as available active sites became depleted and reached equilibrium. The extended residence time at lower flow rates also facilitated greater accumulation of PFAS on the adsorbent surface. On the other hand, higher flow rates shortened the contact time between PFAS and the column, preventing the system from reaching adsorption equilibrium. Consequently, the breakthrough curve exhibited a sudden and sharp rise³⁵.

Adsorption models

A comprehensive understanding of adsorption column dynamics is indispensable for full-scale design, as breakthrough curve analyses derived from experimental studies form the basis for predicting operational lifespan. In this study, the experimental breakthrough data were correlated with the Thomas, Adams-Bohart, and Yoon-Nelson models to evaluate the column adsorption dynamics and predict key operational parameters.

Thomas model

In dynamic adsorption studies, the Thomas model is commonly used to analyze column effectiveness under different operating conditions, including variations in bed height, flow rate, and feed concentration^36,37. The model operates under the assumption of Langmuir kinetics while neglecting axial dispersion³⁸, is expressed as equation 12: 

12

where m represents the adsorbent mass (g), t is the operation time (min), K_TH denotes the Thomas rate constant (mL·mg-1·min-1), Q denotes the flow rate (mL·min^-1),qo denotes the equilibrium adsorption capacity (mg·g^-1), and C_o and C_t (mg·L^-1) are the PFAS concentrations in the influent and at time t, respectively

The breakthrough curves and corresponding parameters (k_TH, q₀, and R₂) were obtained through Analysis of linear regression using the Thomas model equation and are presented in Suppl. Table S3. & Suppl. Table S4. The Thomas model produced a good fit for all of the experimental conditions, exhibiting strong correlation coefficients between 0.95 and 0.98. The results further indicated an inverse correlation between the adsorption capacity (q₀) and the Thomas rate constant (K_TH), suggesting that higher adsorption capacities were associated with lower rate constants. As the adsorbent mass and influent concentration increased, (K_TH) tended to decrease. Conversely, increasing the inflow rate led to an increase in (K_TH). This behavior can be explained by the fact that the mass transfer driving force becomes more pronounced at higher influent concentrations, promoting greater adsorption capacity. However, lengthened interaction time of the solute with the adsorbent, at higher adsorbent dosages or column height, reduced the mass transfer rate, thereby lowering (K_TH) but increasing q₀. Conversely, at an elevated flow rate, the retention period of PFAS ions on the adsorbent surface decreased, resulting in a lower adsorption capacity. Nonetheless, this also accelerated the mass transfer rate, resulting in a higher Thomas rate constant.

Suppl. Figures 2a & 2b show the impact of varying bed height (Z = 3, 5, 7 cm) on breakthrough curves at Co = 25 mg/L and Q = 1.0 mL/min. Table S3 presents the associated parameters derived from the respective BTC. The sigmoidal profile observed at Z = 7 cm resembles findings from previous studies and is notably less steep than that at Z = 3 cm. Overall, increasing the bed height enhances the treated volume, Adsorption capacity, and efficiency in removing, as a greater amount of adsorbent provides more binding sites, reduces axial dispersion, and improves adsorbate transport within the column

The Yoon-Nelson (YN) model

The Yoon-Nelson (YN) model suggests that the probability of an adsorbate molecule being adsorbed decreases in proportion to its likelihood of passing through the adsorbent bed.

Equation (13) represents the model’s non-linear form,

13

where τ denotes the breakthrough time corresponding to 50% adsorbate removal (minutes), and K_YN refers to the Yoon-Nelson rate constant (min^-1). Additionally, the Yoon-Nelson model demonstrated a strong agreement with the experimental data, with R² values that closely matched those of the Thomas model. The (Suppl. Table S3. & S4) shows that the Yoon-Nelson rate constant (K_YN) decreases with the rise in bed height, but rises with higher influent flow rates (Suppl. Table S5. & S6)³⁶. This rise in K_YN is attributed to enhanced mass transfer forces in the liquid phase under higher flow and concentration conditions. Conversely, elevating the influent flow velocity reduced the τ, whereas a greater bed height extended τ³⁹. At higher adsorbent bed heights, a greater number of adsorption sites become available, enhancing PFAS removal, thereby extending the 50% breakthrough time (τ). In contrast, at elevated influent flow rates and concentrations, a greater number of PFAS molecules are delivered to the bed, quickly saturating the limited adsorption sites and resulting in earlier breakthrough. Similar trends in both the breakthrough time (τ) and Yoon-Nelson rate constant (K_YN) have been consistently reported in previous studies³⁵. According to the Yoon–Nelson model, decreasing the flow rate (Q) from 2 to 1 mL/min led to an increase in τ, accompanied by a corresponding decrease in K_YN. Conversely, decreasing the bed height (Z) resulted in a reduction in τ, accompanied by an increase in K_YN. The observed trend is primarily due to lower bed depths and higher flow rates, which promote faster saturation of the column³⁸.

The Adams-Bohart (AB) model

The Adams-Bohart (AB) model, generally applied for interpreting the early breakthrough region, suggests that the adsorption rate depends on both the influent adsorbate concentration and the adsorbent’s remaining capacity. This relationship is expressed in equation 14.

14

where Z symbolizes bed height (cm), v is linear flow velocity (cm/min), N_O is maximum volumetric adsorption capacity (mg/L), and K_AB is the Adams-Bohart kinetic constant (L/mg·min). Among the models evaluated, the Bohart-Adams model showed the lowest R², making it the least suitable for describing the breakthrough curve (BTC). The Adams-Bohart kinetic constant (K_AB) exhibited a positive correlation with flow rate and a negative correlation bed height, suggesting that the early phase of adsorption within the column is largely influenced by mass transfer, which regulates the migration of solute molecules from the bulk solution toward the adsorbent surface (Suppl. Table S5. & S6). The increase in the Bohart-Adams kinetic constant (K_AB) at increased flow rates is due to the greater kinetic energy of PFAS molecules. In contrast, at higher influent concentrations and greater bed heights, PFAS molecules experience more resistance due to denser packing and increased adsorbent material, which reduces their kinetic energy and leads to a lower K_AB⁴⁰_. Operating at reduced flow rates under conditions of higher influent concentration or bed height is favorable for lowering K_AB (suggesting decreased internal resistance) while simultaneously improving adsorption capacity³⁹.

Mechanisms

Aminated sorbents generally exhibit greater efficiency in removing sulfonated PFAS compared to carboxylic PFAS of similar chain lengths, primarily due to the higher hydrophobicity of sulfonated compounds^41,42. Sorption capacity and PFAS hydrophobicity both increase with the length of the perfluorocarbon chain. For instance, in competitive kinetic sorption studies involving PFAS mixtures of varying chain lengths, long-chain PFAS (C₇-C₁₄) demonstrate higher equilibrium adsorption capacities than their short-chain counterparts (C₄-C₆). Moreover, short-chain PFAS tend to desorb after initial adsorption, indicating that long-chain PFAS can displace them over time⁴². This observation implies that the enhanced retention of long-chain PFAS results from their ability to participate in both hydrophobic and electrostatic interactions, whereas the adsorption of short-chain PFAS is governed mainly by electrostatic forces. The pH_PZC of APTES-modified diatomaceous earth (DE) granules typically lies around 8.2, depending on the extent of surface functionalization and the synthesis conditions. Grafting APTES (3-aminopropyltriethoxysilane) onto the DE surface introduces amine groups (–NH₂), which can protonate under acidic conditions, imparting a positive surface charge, or deprotonate under basic conditions, resulting in a neutral or negatively charged surface. The effect of solution pH on adsorption efficiency was systematically investigated. The results revealed that APTES-DE exhibited high removal efficiencies for PFBS and PFOS within the acidic to neutral pH range (3-7). This improvement is attributed to the protonation of surface amine groups on the adsorbent, which results in positively charged NH₃⁺ or NH⁺ species that facilitate strong electrostatic interactions with negatively charged PFAS molecules. However, a marked decline in removal efficiency was observed at pH values above 7, attributable to reduced protonation of amine groups at higher pH, leading to a decrease in surface positive charge and, consequently, weaker electrostatic attraction toward PFAS. As the pH exceeds 7, the amine groups on the APTES-DE surface undergo only partial protonation. For instance, ethylenediamine, which contains two amine groups, exhibits two pKa values, and within the pH range of 6.9 to 9.9, typically only one amine group remains protonated. At pH values above 10, most of the amine functionalities on APTES-DE become deprotonated and neutral. Additionally, the deprotonation of surface carboxyl and hydroxyl groups at high pH contributes to the development of a negatively charged surface. Therefore, at pH ≤ 7, the APTES-DE surface maintains a net positive charge due to fully protonated amine groups (NH₃⁺ )²⁸, which promotes strong electrostatic attraction toward negatively charged PFAS species, particularly short-chain compounds like PFBS, resulting in enhanced adsorption performance. Furthermore, hydrogen bonding interactions are established between the amine and silanol groups on the APTES-modified surface and the oxygen atoms of the sulfonate headgroups of PFAS, thereby enhancing adsorption. Although PFAS adsorption decreases at pH values above 7, removal still occurs, indicating that mechanisms beyond electrostatic attraction, hydrophobic interactions, act synergistically with electrostatic forces, contributing substantially to the overall adsorption mechanism. Electrostatic interactions, arising from the attraction between protonated amine groups and negatively charged PFAS, govern the adsorption process, while at higher pH, reduced electrostatic interactions enhance the role of hydrophobic forces, particularly for long-chain PFAS. In contrast, short-chain PFAS, such as PFBS, rely more heavily on electrostatic attraction and therefore show a greater decline in removal efficiency as the number of positively charged surface sites diminishes with increasing pH¹³. Overall, the results suggest that PFAS adsorption onto APTES-DE is governed by a combination of electrostatic and hydrophobic interactions, with pH primarily affecting the electrostatic component. Therefore, PFAS (PFOS and PFBS) adsorption onto APTES-modified diatomaceous earth granules is governed mainly by three interactions: electrostatic attraction, hydrogen bonding, and hydrophobic interactions. The presence of amine groups (–NH₂/–NH₃⁺) on APTES-DE and Si–O–Si framework vibrations indicates that adsorption mainly occurs via electrostatic attraction between protonated amines and the sulfonate groups (–SO₃^-) of PFOS/PFBS, with additional contributions from hydrogen bonding and hydrophobic interactions, particularly stronger for PFOS due to its longer perfluorinated chain. These interactions collectively explain the enhanced affinity of the APTES modified adsorbent, as illustrated in Fig. 8.

Fig. 8.

Schematic illustration of the proposed adsorption mechanism of PFOS and PFBS onto APTES-modified diatomaceous earth (APTES-DE).

Regeneration of the APTES-DE granules

Adsorbent regeneration is critical for practical applications, as it reduces operational costs and enhances the sustainability of the treatment process. In this study, PFAS desorption from spent APTES-DE was carried out using a solvent mixture of 50% ethanol and 0.5% NH₄OH solution⁴³.In each adsorption–desorption cycle, regeneration was carried out using a desorption solvent volume of approximately 10 mL g^-1 of adsorbent, and this solvent-to-adsorbent ratio was consistently maintained across all consecutive adsorption–desorption cycles. After each desorption cycle, the regenerated APTES-DE granules were thoroughly rinsed with deionized water to remove residual ethanol–NH₄OH solution and desorbed PFAS. The desorption efficiency was then calculated using equation (15).

15

The regenerated APTES-DE was then reused in subsequent PFAS removal experiments. As illustrated in Figure 9, the regenerated adsorbent maintained high removal effectiveness during five consecutive adsorption-desorption cycles, indicating the APTES-DE material’s remarkable reusability and stability. Furthermore, the regeneration studies confirmed the high reusability of the adsorbent across five consecutive sorption-desorption cycles, as evidenced by only minor declines in removal efficiency (4.56% for PFOS and 3.62% for PFBS). These results suggest minimal loss of active sites and demonstrate the effectiveness of the regeneration process.

Fig. 9.

Removal efficiency over 5 cycles using a mixture of 50% ethanol and 0.5% NH₄OH solution as the desorbing agent.

Conclusions

APTES-modified diatomaceous earth granules were successfully synthesized as an efficient filter aid, providing an active adsorbent with enhanced surface functionality. The structural, morphological, and surface properties of the APTES-modified diatomaceous earth granules were studied using SEM-EDS, FTIR, XPS, XRD, and BET analyses. The adsorption of PFOS and PFBS was evaluated through both batch (discontinuous) and continuous column studies. The Langmuir and Freundlich models effectively explained the adsorption isotherms; the fixed-bed column breakthrough behavior was interpreted based on kinetic studies performed under batch experimental conditions. Effects of key operational parameters, including influent concentrations of PFOS and PFBS, flow rate, and bed height, in continuous studies were studied. The maximum removal efficiencies achieved were 92.43% for PFOS and 89.90% for PFBS. The adsorption behavior of both PFOS and PFBS on APTES-modified diatomaceous earth was best fitted by the Freundlich isotherm model, suggesting multilayer adsorption on a heterogeneous surface. The adsorbent demonstrated excellent reusability, with only minor declines in performance after five regeneration cycles (4.56% for PFOS, 3.62% for PFBS). Adsorption was primarily driven by combination of electrostatic, hydrophobic interactions, and hydrogen bonding with pH primarily affecting the electrostatic component and hydrophobic interactions, lowers efficiency for PFBS compared to PFOS highlights the positive correlation between adsorption performance and perfluorinated chain length. The strong affinity of PFOS toward APTES-modified diatomaceous granules was demonstrated by its rapid and high removal efficiency. In contrast, PFBS exhibited lower adsorption capacity, although its removal was also rapid.

Abbreviations

APTES

(3-aminopropyl) triethoxysilane

APTES-DE

(3-aminopropyl) triethoxysilane modified diatomaceous earth granules

DE

Diatomaceous earth

SEM

Scanning electron microscopy

EDAX

Energy-dispersive X-ray spectroscopy

FTIR

Fourier-transform infrared spectroscopy

XPS

X-ray photoelectron spectroscopy

BET

Brunauer-emmett-teller

GAC

Granular activated carbon

Kf

Freundlich equilibrium constant

b

Langmuir constant

n

Freundlich constant

K₁

Pseudo-first-order kinetic rate constant

K₂

Pseudo-second order kinetic rate constant

PSO

Pseudo-second order

PFO

Pseudo-first order

PFAS

Perfluoroalkyl and polyfluoroalkyl substances

PFBS

Perfluorobutane sulfonic acid

PFOS

Perfluorooctane sulfonate

PFOA

Perfluorooctanoic acid

Qe

Equilibrium adsorption capacity

Qm

Maximum adsorption capacity

Author contributions

Amreen Bano: Writing, Validation, Formal analysis. Bablu Prasad, Hemen Dave: Conceptualisation, data curation. Kumar Suranjit Prasad: Writing, Visualisation, Supervision.

Data availability

Data will be made available upon personal request to the corresponding author.

Declarations

Competing interest

The authors declare that they do not have any competing interests.