Competitive adsorption of pharmaceuticals and dissolved organic matter from hospital effluent on amine-functionalized periodic mesoporous organosilicas

Abstract

The competitive adsorption of clofibric acid (CFA) and sulfamethoxazole (SMX) on amine-organosilane-modified periodic mesoporous organosilicas (PMOs) along with the coexistence of actual complex of dissolved organic matter (DOM) were systematically investigated. The effects of amine-silane units (1N-, 2N-, and 3N-silane) and the molar loading ratios of amine-silane units (10%, 25%, and 40% of organosilica precursor) were evaluated. 3 N-Silane’s loading rate at 40% provided the highest amine group density and , resulting in the highest adsorption performance of CFA and SMX. Adsorption rate and capacity of CFA were better than those of SMX, which closely correlated with the hydrogen bonding, attractive electrostatic interaction, and hydrophobic partitioning corresponding to the and values of the adsorbates. Compared to SMX, CFA’s adsorption onto 40% and 3N-magnetized PMO performed higher competitive adsorption, and solution pH had a lower effect on it. Hydrophilic-like DOM may interact with CFA and SMX only in the aqueous phase, resulting in adsorption reduction in a mixed-solute solution. Interestingly, hydrophobic-like DOM can be co-adsorbed and promote the adsorption of SMX in a mixed-solute solution through multilayer adsorption. While coexisting with the actual mixed DOM complex, the competitive adsorption capacity of CFA was reduced and that of SMX increased. However, a reduction in the CFA adsorption capacity of 3 N-magnetized PMO after regeneration by ethanal extraction was detected due to the collapse of the organosilica structure.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-17629-3.

Introduction

Nowadays, pharmaceutical personal care products (PPCPs) are often found in aquatic systems, including water supply and wastewater worldwide, as they have pseudo-persistent characteristics^1,2. Sulfamethoxazole (SMX) is a significant antibiotic that has a herbicidal effect on folate biosynthesis in plant cells and is toxic in fish cells^3,4. SMX has been found to be up to 2 µg L^− 1 in sewage treatment effluent, 1.9 µg L^− 1 in surface water, and 470 µg L^− 1 in groundwater¹. Clofibic acid (CFA) is the metabolite of the lipid regulators clofibrate, etofibrate, and etofylline clofibrate, which are used to decrease the level of cholesterol and triglycerides. The effect of CFA has been reported in causing changes in gene level involved with lipid homeostasis, both in exposed fish and in epigenetic mechanisms^5,6. Up to 41.4 µg L^− 1 CFA was detected in wastewater treatment plants⁵representing one of the most abundant pharmaceutical compounds in the environment. CFA and SMX can partially be eliminated by conventional biodegradation and advanced oxidation processes⁷. Biodegradation process can partially eliminate CFA with a reported removal efficiency of around 51%⁵; however, the performances of photodegradation and ozonation were better, as they facilitated more effective degradation of CFA to 99.3% and 58%, respectively^8–11. Nevertheless, the toxicity of the intermediates or by-products of SMX and CFA resulting from these degradation processes remains and needs further mitigation. Thus, at the end of conventional biodegradation processes, the adsorption process as tertiary treatment (especially for specific sources, such as hospital wastewater treatment facilities) should be an appropriate method to eliminate remaining PPCPs and their persistent intermediates, preventing the dispersion of their toxicity into the environment. Previous studies have reported the adsorption performance of SMX and CFA onto various hybrid organic-inorganic adsorbents, such as metal-organic frameworks (MOFs), functionalized mesoporous silicas, graphene oxides (GOs), and modified carbonaceous adsorbents (as summarized in Table S1 of supplementary information; SI; however, most focused on single-solute adsorption without investigating their competitive adsorption among coexisting pollutants and backgrounds in wastewater.

PMOs form a class of organic–inorganic hybrid materials in which organic moieties are architecturally embedded in the framework walls of inorganic channels. Organic groups of PMOs are uniformly distributed inside the framework, and they are more hydrophobic and stable in water than pure organized silicas, facilitating mechanical and hydrothermal stability, diffusion of guest molecules, and lack of pore blocking in inner pore walls (MCM-41 or SBA-15)¹². PMO materials are synthesized via a sol–gel technique using a bridged silsesquioxane, (RO)₃Si-R′-Si(RO)₃, as the organosilica source in the presence of a supramolecular assembly of surfactant molecules as the template. In contrast to pure mesoporous silicates, amine-terminated organosilanes, such as N-(2-aminoethyl)-3-aminopropyl-trimethoxysilane (AAPTS), can be mixed well due to their structural similarities and cocondensed with the bridged silsesquioxane during the sol–gel process in which the mesophase of PMO is formed¹³. In previous research, the effects of amine-terminated units (1–3 units per molecule) of added organosilanes and their loading rate as high as 40% molar ratio of organosilica precursor on their morphology have been investigated, suggesting the potential of the latter for various applications¹³. PMOs have been utilized in many applications, such as catalysis¹⁴drug hosts¹⁴protein selection¹⁵and adsorbents. Currently, only a few studies have reported the removal efficiency of PMOs on aquatic pollutants, such as heavy metals¹⁶herbicides¹⁷and PPCPs¹⁸. For example, the adsorption efficiency of ciprofloxacin onto PMOs in aquatic systems was reported to be rapid, and its adsorption capacity was reported to be high (up to 291 µg/g) due to the hydrophobicity of PMOs combined with hydrogen bonding and π-π bonding¹⁸. Fluoroalkyl and amine functionalization on PMOs was applied for the removal of short-chain perfluorobutanoic acid via non-electrostatic interactions. Moreover, regeneration by a 50%/50% methanol/water mixture containing 500 mM of NH₃⋅H₂O could provide an excellent reuse efficiency for at least five cycles¹⁹. Moreover, amine-bridged PMOs were reported to have a high Se(VI) adsorption capacity (∼175 mg/g) over a wide pH range (3–9), and the adsorbents could be regenerated by 500 mM NaCl solution¹⁶. The excellent desorption efficiency and reusability of the PMOs could promote sustainable applications, especially for pollution prevention processes.

In aquatic environments or effluents of biological treatment facilities, DOM, such as humic acids and the small molecular structures of extracellular polymeric substances, are often found as the background matrix. DOM can increase or decrease the adsorption capacity of PPCPs on adsorbent surfaces due to multilayer interaction or site competition, respectively. A previous study found that the presence of tannic acid can increase the adsorption capacity of ciprofloxacin on functionalized mesoporous silicas by multilayer adsorption between the positive charges of the amine group in ciprofloxacin and negative charges of tannic acid²⁰. In contrast, coexisting tannic acid and humic acid competitively reduced the adsorption capacity of oxytetracycline onto amino-functionalized mesoporous silica-magnetic graphene oxide nanocomposite at a low concentration range²¹. Nevertheless, the competitive and/or synergistic adsorption mechanisms between PPCPs and DOM onto PMOs’ surfaces still lack theoretical information for prediction and need further investigation.

In this study, the effects of the following three types of amine functional groups on the adsorption capacity of SMX and CFA and their mechanisms are investigated: 3-aminopropyltriethoxysilane, 3-[2-(2-aminoethylamino)ethylamino]-propylthimethoxysilane), and [2-(2-aminoethylamino)propyl]-trimethoxysilane (1N-, 2N- and 3N-silane, respectively); these groups have different functional group loading ratios ranging from 10 to 40% molar ratio (to organosilica precursor). Moreover, the effect of the addition of Fe₃O₄ nanoparticles to the core of the PMO structure was also evaluated, which aimed to enhance the solid–liquid separation without filtration or centrifugation under an applied magnetic field. Finally, to stimulate the actual conditions of the aquatic system, such as the effect of pH and the role of the actual DOM complex (hydrophobic-like DOM and hydrophilic-like DOM) from hospital wastewater effluents on SMX and CFA, competitive adsorption and related mechanisms were also investigated.

Materials and methods

Chemical reagents

Bis(troethoxysilyl)ethane (BTSE) and cetyltrimethylammonium chloride (CTAC) were purchased from Sigma-Aldrich. 3-Aminopropyltriethoxysilane, 3-[2-(2-Aminoethylamino ethylamino]-propylthimethoxysilane), and [2-(2-Aminoethylamino)propyl]-trimethoxysilane (1N-, 2N-, and 3N-silane) were purchased from Aldrich. Iron (II) sulfate heptahydrate (99%) and iron (III) chloride hexahydrate (96%) were purchased from Sigma-Aldrich and Ajax Finechem, respectively. Ammonium hydroxide (28–30%) was purchased from J.T. Baker. SMX (98%) and CFA (97%) were purchased from Sigma-Aldrich. Tryptophan purchased from Sigma-Aldrich was selected as a hydrophilic-like DOM (PHI-DOM).

Adsorbent synthesis

The pristine PMO and seven amine-functionalized PMOs varying in terms of type (amine unit) and loading ratio of organic moieties were synthesized using Burleigh’s method²²and the synthesis scheme is shown in Fig. S1. In short, the molar proportion in the mixture was (1.0–x) BTSE: x R′Si(OR)₄:0.12 CTAC:1.0 NaOH: 230 H₂O, where R′Si(OR)₄ comprised amine-terminated organosilanes (1N-, 2N-, and 3N-), and x was their molar proportion of 0.1, 0.25, and 0.4 (−10PMO, −25PMO and −40PMO), respectively. First, CTAC was dissolved in distilled water and stirred in a PTFE vessel, and drops of NaOH were added into the solution. BTSE (with/without amine-terminated organosilanes) was then added to the solution and stirred at room temperature for 12 h. After filtration of the solution, a white precipitate was collected and rinsed with distilled water and then dried at 105 °C for 12 h. Furthermore, CTAC was removed through the process of solvent extraction using 350 mL of ethanol for 1.0 g of adsorbent under reflux condition for 10 h. Functionalized PMOs were then designated as nNmPMO, where n and m represented the type (1N-, 2N-, and 3N-silane) and the loading ratio (10, 25, and 40% molar), respectively, of organo-silanes used for the functionalization.

Fe₃O₄ nanoparticle (SP) was synthesized through coprecipitation from aqueous alkaline solutions, as given in the following procedure: FeCl₃·6H₂O (0.046 mol) and FeSO₄·7H₂O (0.023 mol) were dissolved in 150 mL deionized water for 3 min. Then, 20 mL of ammonium hydroxide (25%) was quickly added to the mixture solution. After 30 min, 3 mL of oleic acid was added to the previous iron solution. The mixture was heated at 75 °C, and this temperature was maintained for 1 h. The SP was collected through magnetic separation, washed three times with deionized water and ethanol, and subsequently dried in vacuum conditions. Then, an SP embedded in amine-functionalized PMO was synthesized by adding 0.5 g SP to 250 mL of PMO synthesis mixture (as above molar ratios).

Adsorbent characterization

The crystalline structure of synthesized PMOs was characterized by powder x-ray diffractometry (XRD) equipped with Cu Kα radiation (DMAX 2200, Rigaku, Japan). The XRD patterns were collected in the range of 2θ = 0.5–6°. The surface area and porosity of synthesized PMOs were analyzed by an Autosorb-1 Quantachrome automatic volumetric sorption analyzer at 77 K. The specific surface area and pore diameter were calculated using the Brunner–Eller–Teller (BET) and Barrett–Joyner–Halenda (BJH) equations, respectively. The morphology of the synthesized PMOs was investigated using a scanning electron microscope (SEM) (JSM-5410LV, JEOL, Japan). The presence of amine functional groups was quantified by the nitrogen content measured with a CHN analyzer (LECO-CHN-2000 CHN analyzer). Surface functional groups and organic moieties in the PMOs were verified using Fourier transform infrared spectroscopy (FTIR) in the range of 400–4000 cm^− 1 (Nicolet iS50 Thermo Scientific, USA). The chemical composition of the selected PMO surfaces and their electron states were analyzed by x-ray photoelectron spectroscopy (XPS) spectrum using a DLD system operated at 150 W (Axis Ultra, Kratos, Japan). The pH at the point of zero charge (pH_pzc) of the synthesized PMOs was measured using acid–base titration at controlled ionic strength at 10 mM. The zeta potential of 3N40PMO-SP under 6.61–8.69 pH and with an ionic strength of 2 mM was analyzed using Zetasizer Ultra (Malvern, Model: ZSU5700). The Boehm titration method was used to investigate the density of the synthesized PMOs in the oxygen-containing functional groups (AOGs)²³.

Dissolved organic matter (DOM)

Effluent from the actual hospital’s wastewater treatment facility in the Bangkok metropolitan area (conventional activated sludge) was collected and fractionated for use as a background DOM to investigate the effect of a mixed DOM complex, hydrophobic-like DOM (HPO-DOM), and hydrophilic-like DOM (HPI-DOM) on CFA and SMX adsorption capacities onto PMOs. For DOM fractionation, the effluent was filtered through a glass fiber filter (GF/F, pore size 0.75 μm) and then adjusted to pH 2 using H₂SO₄. After that, the water sample was fed through resin (DAX-8) at a flow rate of 20 mL min^− 1. The filtrate from the DAX-8 resin column can be classified as HPI-DOM that was eluted from DAX-8 resin with 0.1 N of 25 mL NaOH followed by 0.01 N of 125 mL NaOH at a flow rate of 200 mL h^− 1. DOM concentration in effluent was quantified by a TOC analyzer (Shimadzu, Japan). A fluorescence excitation–emission matrix (FEEM) scan using a 1-cm quartz fluorimeter cuvette (HORIBA Aqualog^®, Japan) was applied to indicate the types of various key components of DOM in the effluent, such as humic-like, fulvic-like, and tryptophan-like, with excitation wavelength values 250–450 nm and emission wavelength values 250–620 nm in 10-nm steps.

Batch adsorption experiments

The adsorption kinetics and isotherms of seven PMOs were conducted as batch experiments in single- and mixed-solute solutions. Adsorption kinetics were studied by fixing solid/liquid ratio at 0.5 g L^− 1 and the initial concentration of CFA or SMX at 4 mg L^− 1 under controlled pH 7 at ionic strength (IS) 10 mM (by phosphate buffer). The experiments were conducted at 25 °C using 150 rpm and then filtered the mixture through a nylon filter by controlling the mixing time to 0–4 h. The concentrations of CFA and SMX were measured using high-performance liquid chromatography (HPLC) equipped with a diode array detector (DAD). The C18 column (4 × 250 mm, 5 μm) was used in HPLC-DAD using an elution gradient of water (A) and acetonitrile (B) at room temperature. Initially, the elution condition was a mobile phase with B 100% v/v, which was reduced to 70% in 5 min, and the flow rate was set at 1 mL min^− 1. The kinetics data were applied to linear-driving-force (LDF) and pseudo-second-order (PSO) kinetic models. Diffusion mechanisms were also analyzed using the intraparticle diffusion model by Weber and Morris. The adsorption isotherms of seven PMOs were studied by varying the initial concentration of SMX and/or CFA from 0.5 to 4 mg L^− 1 in the conditions same as those of the kinetic tests for 24 h. The isotherm data were applied to the linear and Freundlich isotherm models. The effect of pH on the adsorption capacities of SMX and CFA on selected PMO were investigated by adjusting pH at 5, 7, and 9, with 10 mM of phosphate buffer at 25 °C.

The competitive adsorption in the mixed-solute solution was determined by the coexistence of CFA and SMX in the solution, and the adsorption capacity was evaluated under the same conditions as those in the isotherm study. Moreover, the competitive adsorption of PPCPs (CFA and SMX) versus the three conditions of DOM (HPI-DOM, HPO-DOM, and mixed DOM complex) was also determined.

The regeneration of CFA-adsorbed 3N40PMO-SP was conducted with 99% ethanol extraction at a solid-to-liquid ratio of 2 g L^− 124. The mixture was stirred for 2 h, separated by filtration, and dried at 80 °C for 12 h. Then, the CFA adsorption capacity of the regenerated PMO was measured, and the reuse cycle was repeated twice. FTIR, nitrogen adsorption isotherm, and N content analysis were applied to evaluate the physicochemical properties of the regenerated 3N40PMO-SP.

Results and discussion

Material characterization

The pristine PMO displayed a typical XRD pattern for a hexagonal mesostructure, featuring one main diffraction peak (100) at 2θ = 2.01–2.50° (Fig. S2). The XRD patterns of amine-functionalized PMOs exhibited a shift of d₁₀₀ and a lower intensity than pristine PMO. Primarily, the loss of structural ordering was due to a poor match between the amorphous silicate framework and the amine moieties located inside the channels of the porous material. This loss was further combined with the effect of hydrogen bonding between the amine functional group and the silanol group on the silica structure. The physicochemical properties of all synthesized PMOs are shown in Table 1. The data plots of nitrogen adsorption isotherms of all synthesized PMOs are also illustrated in Fig. S3. The BET surface area and porosity of all PMOs revealed a high surface area (484–887 m² g^− 1), which aligned with the previous study¹³. However, the presence of amine organic ligands exhibited a decrease in surface area, which might have been related to the collapse of the porous structure of PMOs due to hydrogen bonding between amine functional groups and the silanol group on the silica structure. Interestingly, the surface area and pore volume of 3N40PMO-SP increased to 887 m² g^− 1 and 0.587 cm³ g^− 1, respectively, suggesting that the presence of SP as the core structure of PMO may promote the formation of the mesoporous structure of amine-functionalized PMO. The range of BJH pore size of all PMOs was 2.7–3.6 nm, being 2.7- to 4.0-fold larger than the molecular size of SMX and CFA (0.98 and 0.89 nm, respectively), which may result in a high possibility for diffusion into the internal pore structure of PMOs²⁵.

Table 1.

Physicochemical characteristics of synthesized pmos.

Sample	Nitrogen content		Particle size	d100	pHzpc	SBETa	Vtb	Pore sizec
	wt%	mmolN g−1	(µm)	Å		(m2 g−1)	(cm3 g−1)	(nm)
PMO	0.34	0.243	19	3.20	4.7	484	0.292	2.70
1N10PMO	0.89	0.636	44	1.30	7.9	860	0.278	2.38
2N10PMO	0.72	0.514	42	1.57	7.4	684	0.327	3.01
3N10PMO	0.89	0.636	35	1.30	7.7	618	0.206	3.02
3N25PMO	1.61	1.150	40	0.84	8.5	697	0.204	2.96
3N40PMO	1.83	1.307	38	0.80	9.4	584	0.131	3.62
3N40PMO-SP	1.39	0.993	31	–	9.2	887	0.587	2.84

^aAnalyzed using N₂ adsorption isotherm with the Brunauer-Emmett-Teller (BET) method, out gas temperature at 150 °C.

^bPore volume analyzed by BJH Method Cumulative Desorption.

^cPore size analyzed by BJH desorption method.

The nitrogen content of PMOs, quantified on an atomic (mmol_N g^− 1) basis, did not increase along with the unit of amine groups in the organosilanes (1N10PMO, 2N10PMO, and 2N10PMO), suggesting that several amine units per silane molecule can induce the aggregation of organosilane via hydrogen bonding, resulting in low dispersion with the BTSE to form an organosilicate framework¹³. Meanwhile, the increase in the amount of 3N-silane (10–40%) caused an increase in nitrogen contents in the following order: 3N40PMO > 3N25PMO > 3N10PMO. Moreover, the nitrogen content of 3N40PMO-SP (0.993 mmol_N g^− 1) (with SP) was lower than that without SP (3N40PMO) (1.307 mmol_N g^− 1), which might have been caused by the interaction of the residual SP with 3N-silane, reducing the dispersion of 3N-silane into the framework. The SEM images of all PMOs are shown in Fig. 1. The morphology of PMOs strongly depends on the synthesis conditions, such as pH, type of surfactant, and organosilanes. As shown in Fig. 1, both small and large particles, with a size range of around 5–50 μm, were mixed. The proportion of large to small particles in the amine-functionalized PMOs was greater than that in the pristine PMO.

Fig. 1.

SEM micrographs (1,000 × magnification) of the pristine PMO, amine-functionalized PMOs (1N10PMO, 2N10PMO, 3N10PMO, 3N25PMO, 3N40PMO), and SP modified PMOs (3N40PMO-SP).

The FTIR spectra of the pristine PMO, 1N40PMO, 2N40PMO, 3N40PMO, and 3N40PMO-SP are illustrated in Fig. 2. The presence of NH and NH₂ (as the stretching vibration of N─H normally found at 3360–3430 cm^− 1)²⁶ could not be identified for any amine units (1N, 2N, and 3N) with added PMOs, which might be due to the interference of a strong and broad spectrum of O─H stretching (3447–3595 cm^− 1) of adsorbed water molecules onto all synthesized PMOs. The C─H stretching bands at 2292 and 2853 cm^− 1 and the C─H deformation band at 1467 cm^− 1 could be assigned to the bridging ethylene moieties in the PMOs²⁷. The existence of organosilica precursors could be observed by the Si─C bonds corresponding to the bands at 1270–1072 cm^− 113. Moreover, the spectrum at 560 cm^− 1 of 3N40PMO-SP can be assigned to iron (II) oxide stretching, which identifies the presence of SP in the 3N40PMO-SP structure²⁸.

Fig. 2.

FTIR spectra of (a) PMO, (b) 1N40PMO, (c) 2N40PMO, (d) 3N40PMO, and (e) 3N40PMO-SP.

The surface charge density of all PMOs was determined by reporting pH_zpc (Table 1) obtained from the plot of pH initial versus pH final of the acid-base titration (as shown in Fig. S4). In the pristine PMO, pH_zpc was lowest at 4.7. After adding amine-organosilanes, the pH_zpc of functionalized PMOs increased up to 9.4, corresponding to the of the ─NH₂ group (approximately 10), and exhibited a correlation with the amine group density (as a nitrogen content value) for both the unit of the amine group (1N- to 3N-) and the loading percentage (10–40%). Moreover, the zeta potential of 3N40PMO-SP was measured as a function of the pH range from 6.61 to 8.69, adjusted by phosphate buffer at 0.002 M (shown in Fig. S5). The zeta potential exhibited positive charges (10.9–15.0 mV) for all pH, dominated by the protonated amine functional group (–NH₃⁺), and correlated with the pH_zpc of 3N40PMO-SP (at 9.2) measured using acid–base titration.

The AOGs quantified using Boehm titration (shown in Table S2) exhibited an increase in AOG density after loading 3N-silane (40%) from 47.5 to 55.0 µmol g^− 1 (comparing between PMO and 3N40PMO), which might be caused by an increase in the silanol group (Si─OH) due to the incomplete formation of bridged silsesquioxane structures. Moreover, after embedding with SP, the AOGs of 3N40PMO-SP decreased to 44.0 µmol g^− 1 (approximately 20%), compared with 3N40PMO, due to the substitution of SP into the bridged silsesquioxane structures.

In addition, the surface composition of the selected 3N40PMO-SP focused on C, O, and N was confirmed by investigating its detailed chemical and electron states using XPS (Fig. 3). The high-resolution XPS spectrum of 3N40PMO-SP on C1s could be deconvoluted to be three peaks (Fig. 3(a)) at binding energy at 288.4 (O─C═O), 286.1 (C─O, C─N, Si─O─C), and 284.6 (C─C or aliphatic C) eV^29–31. Remarkably, the peak at 288.4 eV corresponded to the bidentate carboxylate carbon of SP’s oleic acid capping agent of SP³⁰. The spectrum of O1s could be fitted by two deconvoluted peaks (Fig. 3(b)) of binding energy at 532.2 (Si─O─Si) and 529.7 (Fe─O of Fe₃O₄) eV^32,33which identified the main silicate framework and SP. The N1s spectra could be deconvoluted by two peaks (Fig. 3(c)) of binding energy at 402.6 (positively charged N such as NH₃⁺ or hydrogen-bonded N) and 399.4 (free ─NH₂) eV^34,35representing grafted amine groups.

Fig. 3.

Narrow scan XPS images of pristine 3N40PMO-SP (a–c), CFA-loaded 3N40PMO-SP (d–f), and SMX-loaded 3N40PMO-SP (g–i).

Adsorption kinetics

The adsorption kinetics of SMX and CFA on the synthesized PMOs are shown in Fig. 4. The adsorption of SMX and CFA by all PMOs was rapid at the beginning and reached an equilibrium state after approximately 1 h. The adsorption capacity of CFA at the equilibrium state of the pristine PMO was the lowest, and it increased on a 10% loading ratio of modified PMOs (Fig. 4(a)). However, the increase of the amine units in organosilanes (1N10PMO, 2N10PMO, and 3N10PMO) did not influence the CFA adsorption rates and capacities significantly, which was related to the almost constant amine group density (0.514–0.636 mmol_N/g) at the same 10% loading ratio. In contrast, the increase in the loading ratio (10 to 40%) had a more significant effect on enhancing the CFA adsorption rates and capacities (Fig. 4(b)). Moreover, the presence of SP in 3N40PMO-SP reduced CFA adsorption rates and capacities (Fig. 4(b)) compared to those detected from 3N40PMO, which might be caused by the reduction in amine group density and repulsive interaction from iron oxide on Fe₃O₄ particle, which supposed to be a negative charge at operating pH (7). Meanwhile, SMX adsorption rates and capacities for all amine-silane-modified PMOs (except 3N40PMO-SP) were more than threefold higher than those observed for pristine PMO (Fig. 3(c and d)). The presence of amine groups at the lowest density of 1N10PMO exhibited the lowest adsorption capacity, followed by 2N10PMO and 3N10PMO, respectively (Fig. 4(c)). However, the varying loading ratio of 3N-silane (10–40%) did not provide a significant change in the adsorption kinetic profile (Fig. 4(d)). Similar to the adsorption kinetic of CFA, the presence of SP in the PMO structure can reduce the adsorption capacity of SMX but increase it more than twofold for pristine PMO (Fig. 4(d)). The adsorption capacities at equilibrium of CFA on to all studied PMOs were higher than those of SMX, which might have been caused by the higher hydrophobicity ( of CFA and SMX are 2.57 and 0.89, respectively) and stronger effect of negative charge of CFA ( of CFA and SMX are 2.8 and 1.6, 5.7, respectively)³⁶.

Fig. 4.

Adsorption kinetics of CFA (a, b) and SMX (c, d) adsorbed on synthesized PMOs (pH 7.0, ionic strength 10 mM, solid/liquid ratio 0.5 g L^− 1, temperature 25 °C). Dash lines present the data from the pseudo-second-order model.

The experimental kinetic data were fitted with adsorption kinetic models, which are the LDF model and the PSO model for calculating the initial adsorption rate (), half-life () and intraparticle diffusion coefficient (), etc. The kinetic data were then analyzed using the intraparticle diffusion model by Weber and Morris to determine the diffusion mechanism and the rate-limiting step in the adsorption process. The parameters used in each model are described in SI. The physical properties of the associated adsorbents and their experimental operation parameters in the kinetic experiment are provided in Table S3. An overview of the kinetic parameters of all PMOs calculated from the LDF and PSO models is provided in Table 2. The adsorption kinetics of the CFA and SMX onto all PMOs could be well matched to the LDF and PSO models with low RRSME values. For the CFA adsorption rates, the from the PSO model of 3N40PMO was highest and remarkably reduced (2.4-fold) for 3N40PMO-SP, which might have been caused by the heterogeneous surface (especially the effect of the negative charge of iron oxides). However, the increase of exhibited a correlation with the amine group density by varying the loading rates of amine silane in the following order: 3N10PMO < 3N25PMO < 3N40PMO. SMX adsorbed on 1N10PMO showed the highest value, while the relationship of to amine group density or porosity of the rest of the PMOs could not be detected. The values of CFA were significantly higher than those of SMX, which might have been related to the stronger minus charge and high hydrophobicity of CFA than SMX.

Table 2.

Kinetic parameters of CFA and SMX adsorption on synthesized PMOs derived from linear-driving-force (LDF) and pseudo-second-order (PSO) models.

Adsorbents		LDF				PSO
					RRSME					RRSME
		mg g−1	s−1	cm2 s−1	mg g−1	mg g−1	g mg−1 min−1	mg g−1 min−1	min	mg g−1
Clofibric acid (CFA)
PMO	1.518	1.394	0.050	3.03 × 10−9	0.097	1.520	1.018	2.352	0.646	0.219
1N10PMO	3.889	3.302	0.033	1.08 × 10−8	0.450	3.892	0.154	2.328	1.672	0.675
2N10PMO	3.915	3.345	0.040	1.16 × 10−8	0.455	3.917	0.182	2.787	1.405	0.682
3N10PMO	4.160	3.555	0.033	6.64 × 10−9	0.483	4.143	0.152	2.616	1.584	0.710
3N25PMO	4.208	3.621	0.033	8.71 × 10−9	0.483	4.197	0.162	2.847	1.474	0.698
3N40PMO	5.281	4.374	0.032	7.71 × 10−9	0.747	5.314	0.107	3.022	1.758	0.910
3N40PMO-SP	3.217	2.587	0.043	6.86 × 10−9	0.362	3.194	0.122	1.249	2.558	0.687
Sulfamethoxazole (SMX)
PMO	0.352	0.315	0.061	3.64 × 10−9	0.031	0.353	3.512	0.437	0.807	0.055
1N10PMO	1.760	1.601	0.062	2.01 × 10−8	0.125	1.755	0.676	2.080	0.843	0.308
2N10PMO	2.118	1.770	0.055	1.61 × 10−8	0.237	2.122	0.259	1.168	1.817	0.430
3N10PMO	2.182	1.871	0.061	1.24 × 10−8	0.241	2.172	0.398	1.881	1.155	0.397
3N25PMO	2.257	1.842	0.053	1.41 × 10−8	0.288	2.259	0.223	1.137	1.987	0.448
3N40PMO	2.488	2.015	0.070	1.68 × 10−8	0.274	2.461	0.193	1.167	2.108	0.561
3N40PMO-SP	1.772	1.382	0.036	5.84 × 10−9	0.261	1.760	0.205	0.635	2.771	0.327

The of CFA and SMX derived from the LDF model (Table 2) were 2–3 magnitudes smaller than the diffusion coefficient in water () (6.49 and 6.25 × 10^− 6 cm² s⁻¹ for CFA and SMX, respectively)^36,37, revealing the resistance through the PMOs internal pore structure. Modification of PMOs by amine groups can enhance the diffusibility of CFA and SMX into the PMOs’ pores, as the increase of values of all amine-modified PMOs compared to the pristine PMO. The of CFA adsorbed on 1N10PMO and 2N10PMO exhibited better diffusion compared to those of 3N-modified PMOs, which might have been due to the pore blocking of three units of amine silane. In contrast, of SMX did not show a significant difference among all amine-modified PMOs.

The intraparticle diffusion model by Weber and Morris applied for all PMOs was displayed by plotting versus (Fig. S6 and S7), and the derived parameters are shown in Table S4). The plots display differentiated linearity for all PMOs, indicating that multiple adsorption steps are associated with the adsorption process. The second region of linearity represents a slower adsorption rate than the first region, which implies that interparticle diffusion controls the overall rate of adsorption. However, for all PMOs, the adsorption plots of CFA and SMX did not pass through the origin, which may indicate that the intraparticle diffusion was not the only rate-limiting step; the external mass transfer might have also played an important role in the adsorption³⁸.

Adsorption isotherms

The adsorption isotherms of CFA and SMX on the studied PMOs are illustrated in Fig. 5, and the surface loading that normalized adsorption capacity (Y axis) by surface area (µg m^− 2) is shown in Fig. 5(b and d). As shown in the adsorption kinetic results, the adsorption capacities of CFA and SMX onto 3N40PMO were the highest compared with lower nitrogen content PMOs in the following order: 3N40PMO > 3N25PMO > 3N10PMO > 2N10PMO > 1N10PMO > 3N40PMO-SP > PMO (Fig. 4(a and c)). The adsorption capacities of CFA onto all PMOs were approximately twofold higher than those of SMX. As mentioned previously, the adsorption of CFA and SMX were strongly influenced by the hydrophobicity of their molecules ( of CFA > SMX) and the potential of deprotonation at pH 7, as referred to the values (CFA < SMX), promoting the surface affinity (hydrophobic partitioning) and attractive electrostatic interaction of CFA onto PMOs, respectively. For both CFA and SMX, the adsorption capacities were related to the nitrogen content ratios of PMOs (Table 1). An increase in the amine unit in the organosilanes (1N, 2N, and 3N) significantly increased the adsorption capacities (mg g^− 1) and surface loading (µg m^− 2) of PMOs (Fig. 5), while the increase in 3 N-silane loading (10%, 25%, and 40%) only exhibited significant enhancement of adsorption capacity and surface loading of CFA and SMX at 40% 3N-silane loading. This is attributed to the increased nitrogen density, promoting hydrogen bonding between the protonated amine (─NH₃⁺) of the 3N-silane on the adsorbent surface with the deprotonated carboxylic groups (COO─) and amine groups (─NH⁻ and ─NH) of the CFA and SMX molecule, respectively³⁸. Moreover, the increase of pH_zpc of 3N40PMO (to 9.4) may promote the positive charge of the surface, enhancing the surface affinity of deprotonated CFA and SMX molecules via attractive electrostatic interaction. In addition, the presence of SP in 3N40PMO-SP may reduce the density of amine groups on the surface and interfere with iron oxide’s active surface functional groups (such as ─FeOOH), resulting in the lowest adsorption capacity of CFA and SMX compared to all amine-silane PMOs (except the pristine PMO).

Fig. 5.

Adsorption isotherms (a, c) and surface coverage (b, d) of CFA and SMX adsorbed on synthesized PMOs (pH 7.0, ionic strength 10 mM, solid/liquid ratio 0.5 g L^− 1, temperature 25 °C).

Adsorption isotherm models

Two isotherm models (linear and Freundlich isotherm models) were applied to describe the adsorption behavior of CFA and SMX onto all PMOs due to the low range of equilibrium concentration (0–4 mg L^− 1) and the appearance of all isotherms close to the linearity. A description of the two isotherm models is presented in SI. The derived isotherm parameters of CFA and SMX adsorbed onto all PMOs are reported in Table S5. The Freundlich isotherm model had a higher correlation coefficient (R² than the linear isotherm model. The linear isotherm constant () and the Freundlich isotherm constant () of CFA and SMX were in the following order: 3N40PMO > 3N25PMO > 3N10 PMO > 2N10PMO > 1N10PMO > 3N40PMO-SP > PMO, presenting their correlation with the amine group density of PMOs (except 3N40PMO-SP). The and of CFA for all PMOs were higher than those of SMX due to the higher adsorption capacities of the former. The adsorption intensity () of the Freundlich isotherm for CFA and SMX adsorbed on all PMOs was in the range of 0.646–1.515, which is mostly close to 1, representing the linearity shape of all isotherms.

Effect of pH

The adsorption of CFA and SMX onto 3N40PMO-SP was selected to evaluate the effect of pH (5, 7, and 9) on adsorption capacities by comparing the adsorption isotherms and derived parameters (Fig. S8 and Table S6). The adsorption of CFA was dependent on the pH, and it was highest at pH 5 and lowest at pH 9. At pH 5, the surface of 3N40PMO-SP had a net positive charge, while CFA was negatively deprotonated ( = 2.84), resulting in better adsorption capacity. Moreover, the effectiveness of hydrogen bonding between CFA and coexisting silanol groups (Si─OH) increased when the silanol groups were positively charged (at a low pH)³⁹. When the pH was increased to 7 and 9, the net positive charge of 3N40PMO-SP decreased to almost neutral at pH 9 (pH_zpc = 9.2), which may reduce the attractive electrostatic interaction of CFA and 3N40PMO-SP’s surface. Meanwhile, at pH 5 and 7, the adsorption capacities of SMX on 3N40PMO-SP were not significantly different, which might have been caused by the zwitterionic molecular structure of SMX at pH 5 ( = 1.6 and 5.7). At pH 7, the zwitterionic molecule of SMX was deprotonated to be anionic, which might have promoted the attractive electrostatic interaction of SMX onto 3N40PMO-SP; however, the net positive charge decreased due to the increase in pH. At pH 9, the adsorption capacity of SMX was remarkably decreased from pH 5 and 7 than that of CFA, which might indicate the significant role of the electrostatic interaction of SMX adsorbed on 3N40PMO-SP than CFA.

Active functional groups

The dominant adsorption mechanisms of CFA and SMX on PMOs can be (1) hydrophobic partitioning, (2) hydrogen bonding between active surface functional groups of PMOs (silanol and amine group) and targeted PPCPs (such as carboxylic and amine group), and (3) electrostatic interaction, which is directly related to the pH condition. CFA and SMX-loaded 3N40PMO-SP were selected as models for investigating the binding sites. A high-resolution narrow scan XPS of the N1s spectra of virgin 3N40PMO-SP and CFA or SMX loaded 3N40PMO-SP (Fig. 3) was compared to evaluate the role of amine groups of 3N40PMO-SP on these synergetic adsorption mechanisms. The deconvoluted N1s spectra of CFA and SMX-loaded 3N40PMO-SP (Fig. 3(f and i)) showed the shift (to lower binding energy) and increase of peak area ratio of hydrogen-bonded NH₂ from 402.6 to 402.2 eV and 18.0–25.6% for CFA adsorption, and from 402.6 to 401.9 eV and 18.0–18.9% for SMX adsorption, respectively, which indicated that the hydrogen bonding in amine groups plays a key role in adsorption between 3N40PMO-SP and CFA or SMX.

Competitive adsorption of CFA and SMX in a mixed solute solution

The adsorption capacities of CFA and SMX onto 3N40PMO-SP in single- and mixed-solute solutions at pH 7 are compared in Fig. 6. The adsorption capacities of CFA decreased slightly with the coexistence of SMX, while the adsorption capacities of SMX significantly decreased in the presence of CFA in the solution. The higher acidity and hydrophobicity of CFA, resulting in stronger attractive electrostatic interaction and hydrophobic partitioning, may promote a higher competitive adsorption of CFA than SMX.

Fig. 6.

Competitive adsorption isotherms of CFA (a) and SMX (b) on 3N40PMO-SP in mixed-solute solutions (ionic strength 10 mM, solid/liquid ratio 0.5 g L^− 1, temperature 25 °C).

Characterization of DOM from hospital wastewater and adsorption of surrogate DOM

The effluent was collected from the general hospital’s wastewater treatment facility (activated sludge). The total DOM concentration of the effluent was reported at 6.91 mgC L^− 1, as analyzed by a TOC analyzer. After fractionation, the concentrations of HPO-DOM and HPI-DOM were quantified at 3.73 and 2.15 mgC L^− 1 (54.9% and 31.1% of total DOM), respectively. The FEEM spectrum showed that the pristine effluent exhibited two fluorescence peaks at 280 nmEx/350 nmEm and 325 nmEx/420 nmEm, as shown in Fig. S9(a). This helped in the identification of tryptophan-like substances (HPI-DOM) and humic-like substances (HPO-DOM), respectively. After fractionation, the FEEM spectra of fractionated HPI-DOM and HPO-DOM were measured, and the tryptophan-like and humic-like structures of the dominant DOM were confirmed (as shown in Fig. S9(b and c)). Due to the high ionic strength of fractionated HPI-DOM and HPO-DOM and the convenience of initial concentration adjustment, surrogate tryptophan and humic acid were used as representatives of HPI-DOM and HPO-DOM, respectively, to study the effects of DOM on PPCPs adsorption of all PMOs.

The adsorption isotherms of tryptophan and humic acid onto all PMOs are illustrated in Fig. 7. The derived parameters of the adsorption isotherms of tryptophan and humic acid are shown in Table S7. From the preliminary adsorption kinetic test (Fig. S10 and Table S8), the adsorption of tryptophan and humic acid reached an equilibrium state within 2 h. The adsorption isotherms of tryptophan onto all PMOs exhibited extremely low adsorption capacities, around 0.010–0.097 mgC g^− 1 (Fig. 7(a)), which might have been caused by the repulsive interaction of the positive charge of protonated amino groups of tryptophan ( = 2.4 and 9.4) at pH 7. In contrast, the adsorption isotherms of humic acid in PMOs (Fig. 7(b)) demonstrated significantly higher adsorption capacities (approximately 2.3-fold) compared with those of tryptophan, providing a correlation with the amine-silane load rate of PMOs (3N10PMO, 3N25PMO, and 3N40PMO). Although varying amine units (1N, 2N, and 3N) of organosilane on PMO surfaces may enhance humic acid adsorption capacity, the effect of the number of amine units in organosilane on the adsorption capacity remains unclear.

Fig. 7.

Adsorption isotherms of tryptophan and humic acid adsorbed on synthesized PMOs (pH 7.0, ionic strength 10 mM, solid/liquid ratio 0.5 g L^− 1, temperature 25 °C).

Effects of DOM complex on CFA and SMX adsorption

The influence of the DOM complex in the effluent of hospital wastewater on CFA and SMX adsorption onto 3N40PMO-SP was investigated in this study by comparing the effects of PHO-DOM (humic acid), HPI-DOM (tryptophan), and total DOM complex (actual effluent) on CFA and SMX adsorption capacities, as illustrated in Fig. 8. The adsorption capacities of CFA decreased due to the coexistence of tryptophan (Fig. 8(a)) and humic acid (Fig. 8(c)). Although tryptophan was not significantly adsorbed by 3N40PMO-SP in the mixed solute with CFA, humic acid was adsorbed at approximately 1.2 mgC g^− 1 at an initial concentration of 6.91 mgC L^− 1. Therefore, the active site competition (probably at ─NH₂ and Si─OH) between CFA and humic acid seems to be the main reason for the decrease in CFA adsorption. However, the slight decrease in CFA adsorption by the coexistence of tryptophan might have been caused by the interaction between tryptophan and CFA via hydrogen bonding at carboxylic groups with π-π electron–donor acceptor interaction between the benzene rings of tryptophan and CFA³⁸. The plausible interaction mechanisms of CFA onto 3N40PMO under coexistence of tryptophan and humic acid were illustrated in Fig. 9.

Fig. 8.

Effects of DOM on CFA and SMX adsorption capacity on 3N40PMO-SP; tryptophan (HPI-DOM) (a, b); humic acid (HPO-DOM) (c, d); mixed DOM complex from hospital effluent (e, f).

Fig. 9.

Plausible adsorption mechanisms of CFA onto 3N40PMO-SP under coexistence with tryptophan and humic acid (at pH 7).

The adsorption of SMX onto 3N40PMO-SP in coexistence with tryptophan decreased compared with the single-solute solution, while the adsorption of tryptophan onto 3N40PMO-SP can be neglected, which might indicate the interaction between SMX and tryptophan in aqueous phase, resulting in the reduction of SMX adsorption (Fig. 8(b)). Surprisingly, the adsorption of SMX on 3N40PMO-SP significantly increased by coexisting with humic acid (Fig. 8(d)). Moreover, the adsorption capacity of humic acid seems to be higher (up to 3.16 mgC g^− 1) than those detected in coexisting with CFA (Fig. 8(d)). The results may indicate multilayer adsorption via the interaction between SMX and humic acid, which is supposed to be hydrogen bonding at carboxylic groups of humic acid and amine groups of SMX along with π-π electron–donor acceptor interaction. The proposed multilayer adsorption mechanisms among SMX, tryptophan and humic acid were summarized as shown in Fig. 10.

Fig. 10.

Plausible adsorption mechanisms of SMX onto 3N40PMO-SP with coexistence with tryptophan and/or humic acid (at pH 7).

The adsorption capacities of CFA and SMX in a mixed-solute solution coexisting with the actual mixed DOM complex of the effluent are investigated and reported in Fig. 8(e and f). Interestingly, in the actual mixed DOM complex, the reduction in CFA adsorption increased along with an increase in the equilibrium concentration of CFA and DOM adsorption (Fig. 8(e)). Apart from the active site competition of HPO-DOM, the multilayer adsorption of SMX and HPO-DOM may interfere with the accessibility of CFA to interact with the amine functional groups of 3N40PMO-SP. However, the competitive adsorption of SMX increased in the presence of the actual mixed DOM complex (Fig. 8(f)), which was enhanced by the multilayer adsorption of SMX and HPO-DOM in the actual mixed DOM complex.

Reusability of 3N40PMO-SP

After regeneration by ethanal extraction, the CFA adsorption capacities of the first and second reused 3N40PMO-SP decreased by approximately 43.1% and 74.7% (as shown in Fig. 11), respectively, compared with that of the virgin 3N40PMO-SP. The FTIR spectra of CFA-loaded 3N40PMO-SP and the regenerated 3N40PMO-SP (as shown in Fig. S11) revealed that CFA adsorbed on 3N40PMO-SP could be removed by the ethanol extraction process due to the disappearance of the peak at 1489 cm^− 1 in the ─COOH group of CFA. However, the BET surface area, pore volume, and pore size calculated from the nitrogen adsorption isotherm exhibited degradation of the porosity of the first reused 3N40PMO-SP, from 887 to 452 m² g^− 1, 0.587 to 0.292 cm³ g^− 1, and 2.84 to 2.58 nm, respectively, which might be caused by the hydrolysis of ≡ Si─O─Si ≡ bonds into ≡ Si─OH⁴⁰ during the adsorption and regeneration process. The nitrogen content of the first reused 3N40PMO-SP also decreased by about 69% (from 0.993 to 0.429 mmol_N g^− 1), confirming the loss of loaded amine–silane in the porous structure. Hence, the enhancement of organosilica structure stability for 3N40PMO-SP, especially in aqueous solution, still requires further investigation to be applied as a sustainable adsorbent.

Fig. 11.

Reusability of 3N40PMO-SP (initial concentration at 40 mg L^− 1, pH 7, and ionic strength at 10 mM).

Conclusion

In this study, the effects of adding amine units (1N-, 2N-, and 3N-silane) and loading ratio (10%, 25%, and 40%) of the amine silane precursor into the PMOs structure, including the decorating Fe₃O₄ nanoparticles as the core of the silicate structure on the physicochemical properties, adsorption kinetics, and isotherms of pharmaceuticals (CFA and SMX) in single- and mixed-solute solutions were investigated. In addition, the role of pH and coexisting actual DOM background from hospital effluent in the competitive adsorption of CFA and SMX was evaluated. The increase in amine-silane loading (to 40% of 3N-silane) can provide the highest amine group density and pH_zpc, significantly improving the adsorption rate and capacities of CFA and SMX via hydrogen bonding and attractive electrostatic interaction at the amine groups of PMOs, thus cooperating with the hydrophobic partitioning of the adsorbate. Due to the higher of SMX, the adsorption performance of SMX can be easily influenced by the solution pH than CFA. Without background DOM, CFA performed better in competitive adsorption than SMX, which correlated with the higher and lower of CFA. Interestingly, the presence of PHI-DOM (tryptophan) slightly reduced the adsorption of CFA and SMX on 3N40PMO-SP. HPO-DOM (humic acid) exhibited better performance in competitive adsorption against CFA, reducing CFA adsorption. The co-adsorption of HPO-DOM enhanced the adsorption of SMX via multilayer adsorption. Moreover, the coexistence of the actual mixed DOM complex reduced CFA adsorption onto 3N40PMO-SP while promoting the competitive adsorption of SMX. The collapse of porosity and the loss of amine functional groups from the organosilica structure of 3N40PMO-SP caused a reduction in the CFA adsorption capacity of the reused 3N40PMO-SP.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed to the study’s conceptualization and methodology. Investigation and formal analysis were performed by Surawut Siratham. Resources and funding acquisition were performed by Patiparn Punyapalakul and Khemarath Osathaphan. The first draft of the manuscript was written by Surawut Siratham and Patiparn Punyapalakul, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The 90th Anniversary of Chulalongkorn University Fund (Ratchadphisek Somphot Fund). National Research Council of Thailand (NRCT) and Chulalongkorn University (N42A680379).

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its Supplementary Information files. Should any raw data files be needed in another format they are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.