Titanium dioxide/graphene oxide blending into polyethersulfone hollow fiber membranes improves biocompatibility and middle molecular weight separation for bioartificial kidney and hemodialysis applications

Abstract

Hollow fiber membranes (HFMs) are critical components in hemodialysis and bioartificial kidney (BAK) applications, with ongoing research focused on optimizing biomaterials for improved performance. In this study, polyethersulfone (PES) HFMs were modified by incorporating titanium dioxide (TiO₂) and graphene oxide (GO) during the spinning process. This approach leverages the non-toxicity, hydrophilicity, and dispersion stability of TiO₂ alongside the large surface area of GO to enhance membrane properties. Characterization and performance evaluations demonstrated that TiO₂/GO-doped PES HFMs exhibit superior biocompatibility and hemocompatibility compared to plain PES, TiO₂/PES, and GO/PES membranes. Confocal microscopy revealed improved HEK293 cell attachment and proliferation, corroborated by MTT assays showing higher cell viability and flow cytometry indicating no cytotoxic effects. Hemocompatibility tests confirmed negligible hemolysis and anti-inflammatory properties, making the membranes suitable for blood-contacting applications. Furthermore, separation performance analyses highlighted TG(0.5/1.5) as the optimal composition, offering a balance of enhanced toxin removal and cell compatibility. These findings establish TiO₂/GO-doped PES HFMs as promising candidates for BAK and hemodialysis, combining excellent biocompatibility, hemocompatibility, and separation efficiency.

---

1. Introduction

Hollow fiber membranes (HFMs) are a class of semipermeable membranes distinguished by their superior properties, including a high surface area-to-volume ratio, compact structure, high packing density, and inherent self-supporting nature. These unique characteristics make HFMs highly suitable for a wide range of commercial applications, especially in biomedical and separation processes involving gases and water.¹ Their remarkable versatility has driven widespread use in the development of bioreactors for large scale cell expansion, and bioartificial organs, including bioartificial kidneys, livers, and pancreas. Furthermore, HFMs play critical roles in blood oxygenators, plasmapheresis systems, and blood purification technologies such as hemodialysis and hemodiafiltration. The integration of HFMs in biomedical and tissue engineering solutions offers immense potential for revolutionizing life-saving bioartificial devices, significantly enhancing therapeutic outcomes and patient care.²

The main focus of this work is to utilize a HFM based biomaterial for hemodialysis and BAK applications. Polymers which are widely used for hemodialysis and BAK applications are polysulfone (Psf) and polyethersulfone (PES). They offer excellent membrane formation ability, mechanical strength, thermal stability, and chemical inertness, which makes them suitable for withstanding sterilizing environments.^3–6 However, these polymers lack hydrophilicity, biocompatibility and activate inflammatory response when they come in contact with blood and cells. Thus, various additives such as graphene oxide (GO), titanium dioxide (TiO₂), silicon dioxide (SiO₂), and MXene into polymeric HFMs have been explored to enhance hydrophilicity and impart biocompatibility.^7,8

In the present study, we have modified PES polymeric HFMs by incorporating TiO₂ and GO nanomaterials to enhance the adhesion and proliferation of kidney cells and separation of toxins to utilize this biomaterial for potential bioartificial kidney applications. This work is an extension to our previous work where TiO₂ was incorporated into PES HFMs to improve the biocompatibility and separation performance.⁹ TiO₂ was selected due to the excellent properties like high hydrophilicity, antibacterial properties, availability, high catalytic activity, low toxicity, high surface area and low cost.^10,11 On the other hand, carbon-based nanomaterials like graphene oxide (GO) and functionalized carbon nanotubes have high aspect ratio, low density, super hydrophilicity, high strength and stiffness. The high hydrophilicity of GO arises from the oxy functional groups (hydroxyl, epoxy, carbonyl, carboxyl) present in it.¹² GO being amphiphilic in nature has a wide distribution of polarity from base to edge from hydrophobic to hydrophilic. Thus, despite having oxy functional groups (hydroxyl, epoxy, carbonyl, carboxyl) present at the edge, the overall hydrophilicity of GO is inadequate. Various efforts have been made to increase the hydrophilicity of GO by preparing sulfonated GO, and addition of TiO₂, ZnO, and SiO₂.¹³ Additionally, GO nanosheets serve as a suitable host for the dispersion and stabilization of inorganic nanoparticles like TiO₂, preventing their agglomeration. In addition, to enhance the hydrophilicity of PES and Psf, TiO₂ and GO also help in enhancing the surface roughness, cell attachment and biocompatibility.^14,15 Previously, it was reported that the GO supports differentiation of stem cells and graphene-based nanomaterials (graphene, graphene oxide and reduced graphene oxide) are used in neural, bone, cartilage, cardiac, skeletal muscle and skin tissues.¹² Furthermore, TiO₂ nanomaterials are well known for tissue regeneration owing to their high tensile strength, flexibility and corrosion resistance, and support the attachment of different cell types, including osteoblasts, cardiomyocytes, pluripotent mesenchymal stem cells and fibroblasts.¹⁶

In this work, HFMs were fabricated by incorporating TiO₂ and GO into PES. The motivation of utilizing both the nanomaterials together i.e., TiO₂/GO in a polymeric membrane, was based on the hypothesis that these additives will not only increase hydrophilicity and biocompatibility of the HFMs but also promote better distribution of TiO₂ NPs in the polymeric matrix with low aggregation and agglomeration.^17,18 These HFMs were expected to show enhanced separation performance, hydrophilicity and biocompatibility as compared to plain PES to be utilized in hemodialysis and bioartificial kidney applications.

2. Materials and methods

2.1. Materials

TiO₂, PES (Ultrason® E 6020 P), and N-methyl-2-pyrrolidone (NMP) were procured from MERCK (Germany), BASF (Germany), and SD Fine-Chem Ltd (India), respectively. Graphite powder (natural, −325 mesh, purity ∼ 99%) was procured from Alfa Aesar (India). Sulfuric acid (H₂SO₄), orthophosphoric acid (H₃PO₄), potassium permanganate (KMnO₄), sodium nitrate (NaNO₃), hydrogen peroxide (H₂O₂), and hydrochloric acid (HCl) were procured from Sigma-Aldrich (United States). HEK293 cells were procured from the National Centre for Cell Science (NCCS, India). Dulbecco's modified Eagle's medium (DMEM) (Himedia, India) containing 2 mM L-glutamine (Sigma-Aldrich, United States), supplemented with penicillin (100 units per mL), streptomycin (100 μg mL⁻¹) (Gibco, Invitrogen, United States), and 10% fetal bovine serum (Himedia, India) was used for the cell culture. Bovine serum albumin (BSA) was purchased from HiMedia, India. BD Vacutainer PLUS plastic plasma tubes (143 USP units sodium heparin coated) were procured from Becton, Dickinson and Company, USA. Glutaraldehyde, paraformaldehyde and dimethylsulfoxide (DMSO) were procured from Sigma Aldrich, India. The micro BCA protein kit was purchased from ThermoFisher Scientific, USA. MTT dye, Triton X-100, and polyethylene glycol (PEG) of different molecular weights (1 kDa, 6 kDa, 10 kDa, 20 kDa, 35 kDa) were purchased from HiMedia, India. Fluorescein isothiocyanate-phalloidin (FITC-phalloidin) and 4′,6-diamidino-2-phenylindole (DAPI) were procured from ThermoFisher Scientific, USA. Urea and creatinine were procured from Loba Chemie, India, and lysozyme and indoxyl sulfate from Sigma Aldrich, India.

2.2. Cell line maintenance

The human embryonic kidney cell line (HEK293) was cultured in DMEM media supplemented with 10% FBS, 1% L-glutamine, and 1% penicillin–streptomycin antibiotic solution at 37 °C, and 5% CO₂ (Thermofisher scientific CO₂ incubator, USA).

2.3. Synthesis of graphene oxide

Graphene oxide was synthesized by a modified Hummer's method as per the literature protocol¹⁹ detailed in the ESI.†

2.4. Fabrication of TiO₂/GO–PES HFMs

TiO₂ and GO were dispersed in NMP solvent at varying concentrations (0%, TiO₂/GO (0.5/1.5), TiO₂/GO (1/1), TiO₂/GO (1.5/0.5), 2% TiO₂, and 2% GO) using a bath sonicator (Branson Sonifier 450 CT, USA). A dilute PES solution in NMP was first introduced to the nanomaterial dispersion, followed by the gradual addition of the remaining PES to achieve a final concentration of 20 wt% PES in the dope solution. The mixture was stirred for at least 12 hours to ensure uniform blending. The degassed dope solution, prepared by removing entrapped air bubbles, was then utilized for hollow fiber membrane (HFM) fabrication via the dry-wet spinning phase separation process. During this process, the polymer solution was extruded through a coaxial nozzle (spinneret). The process parameters and schematic of HFM fabrication are shown in Table 1 and Fig. 1, respectively. The resulting fibers were soaked in water for 24 hours to eliminate residual solvent and subsequently taken ahead for further analysis.

Table 1 The process parameters for HFM fabrication with varying TiO₂ and GO concentration

Ambient temperature (°C)		25
Relative humidity (%)		50–60

---

	Samples	TiO2	GO	PES	NMP
Dope solution composition in NMP (wt%)	P	0	0	20	80
	TG(0.5/1.5)	0.5	1.5	20	78
	TG(1/1)	1	1	20	78
	TG(1.5/0.5)	1.5	0.5	20	78
	2T	2	0	20	78
	2G	0	2	20	78

---

Bore solution composition		DI water
Dope solution temperature (°C)		25
Bore solution temperature (°C)		25
Dope flow rate (mL min^−1)		3
Bore flow rate (mL min^−1)		3
Air gap (cm)		30
Coagulation bath composition		DI water
Rinse bath composition		DI water
Coagulation bath temperature (°C)		30

---

Fig. 1 Schematic representation of the preparation of TiO₂/GO-doped PES-based HFMs.

2.5. Physicochemical characterization of TiO₂ and GO nanomaterials

Morphological studies of TiO₂ and GO were carried out using scanning electron microscopy (SEM) and transmission electron microscopy (TEM).

For SEM, TiO₂ and GO powder were coated with iridium by sputter coating using an Auto fine coater JFC-1600 (JEOL, Japan). Samples were observed under an electron microscope at 10 kV (JSM-7600F, Jeol).

For TEM, TiO₂ and GO powder were suspended in ethanol and sonicated for about 10–15 min and then drop cast on a copper grid for analysis under TEM.

XRD analysis was performed to determine the crystalline properties of TiO₂ and d-spacing of the prepared GO powder (Smart Lab, Rigaku, United States). Cu-Kα radiation (k = 1.54178 Å) was used for determining the XRD patterns in a 2θ range of 10°–90° and a scan rate of 8° per min.

FTIR spectra of GO was performed to identify different functional groups using the FTIR spectrometer (Bruker, Tensor 27, United States) in the attenuated total reflectance mode using germanium (Ge) crystals, for the wavenumber range 400–4000 cm⁻¹.

2.6. Physicochemical characterization of HFMs

Physicochemical characterization of the HFMs was performed using scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), X-Ray diffraction (XRD), contact angle measurement, and atomic force microscopy (AFM) to determine the morphology, elemental analysis, crystalline behaviour, hydrophilicity, and surface roughness of the HFMs.

2.6.1. Analysis of HFM morphology. The morphology of the fabricated HFMs was analysed by freeze-fracturing the membranes in liquid nitrogen to preserve their structure. The samples were then mounted on copper sheets using carbon tape and sputter-coated with iridium using an Auto Fine Coater JFC-1600 (JEOL, Japan). Morphological examination was conducted with a field emission scanning electron microscope (FE-SEM, JSM-7600F, JEOL) operated at 10 kV.

2.6.2. Analysis of surface crystallinity. XRD analysis of the HFMs was performed to determine the presence of TiO₂ and GO in it. Cu-Kα radiation (k = 1.54178 Å) was used for determining the XRD patterns in a 2θ range of 10°–90° and a scan rate of 8° per min (Smart Lab, Rigaku, United States).

2.6.3. Contact angle measurement. The hydrophilicity of the prepared HFMs was assessed by measuring their surface wetting characteristics using a Digidrop GBX contact angle analyzer (GBX Instruments, Romans, France). HFMs were cut with a sharp razor blade, flattened, and affixed to a glass slide using double-sided adhesive tape. A 3 μL drop of deionized water was applied to five different locations on the surface at varying time intervals, and the corresponding contact angles were measured.

2.6.4. Surface topography of the HFMs. The surface roughness of the HFMs was evaluated using atomic force microscopy (AFM) (MFP-3D-BIO, Asylum Research). Tapping mode (air) was used to scan a 5 μm × 5 μm area using a silicon nitride probe cantilever (spring constant ∼40 N m⁻¹). Samples were cut, flattened and mounted on a glass slide. Average surface roughness (R_a) and root mean square surface roughness (R_Rms) were measured to evaluate the surface roughness.

2.7. Biocompatibility evaluation of the HFMs

Biocompatibility of the HFMs was evaluated as kidney cell attachment, proliferation, and cytocompatibility. The study was performed on days 4 and 8, after seeding cells on HFMs as explained in ref. 9.

2.7.1. Confocal laser scanning microscopy study. Cells on the HFMs were washed with PBS, fixed with 4% paraformaldehyde, and permeabilized using 0.1% Triton™ X-100. F-actin filaments were stained with FITC-phalloidin (1 : 100) for 4 h at 4 °C, and nuclei were stained with DAPI (1 : 100) for 10 min and further washed with PBS. The samples were mounted on glass slides, and cell morphology was imaged using a confocal microscope (Zeiss LSM 780) at 20× magnification.

2.7.2. MTT cell viability assay. An MTT assay was used to evaluate cell viability by measuring the reduction of yellow MTT to purple formazan by metabolically active cells. MTT solution (5 mg mL⁻¹) was prepared in PBS and filtered through a 0.22 μm filter. A volume of 200 μL of MTT solution (Himedia, India) was added to each well containing HFM samples and incubated at 37 °C for 4 hours. Subsequently, 1 mL of dimethyl sulfoxide (DMSO) was added to dissolve the formazan crystals, followed by shaking (Eppendorf thermomixer compact 5350 mixer) at 300 rpm for 30 minutes. The optical density (OD) of the resulting solution was measured at 570 nm using an ELISA plate reader (Molecular Devices, Spectramax® iD3). All experiments were performed in triplicate, and the results are presented as mean ± standard deviation.^9,20

2.7.3. Fluorescence activated cell sorting study. The protocol for HEK293 cell attachment was followed as described in ref. 9. HFMs were transferred to 1.5 ml centrifuge tubes, and 300 μl of trypsin was added. After 10 minutes, 600 μl of media was added to neutralize the trypsin, and the tubes were shaken for 30 minutes. HFMs were removed, and the remaining solution containing cells was centrifuged to pellet the cells. The supernatant was discarded, and the cells were resuspended in 500 μl PBS. The samples were then stained with propidium iodide (PI) dye for a live–dead assay in flow cytometry.

2.7.4. Live dead staining by calcein AM and PI dye. The HFMs with cells seeded on the surface were stained with 2 μM calcein AM dye and incubated for 30 min in a humidified atmosphere of 5% CO₂ and 37 °C. Dye was removed followed by washing with PBS. A further 2 μM of PI dye was added and incubated for 10 min followed by washing with PBS and observed under confocal microscope (Yokogawa CSU-X1) at 10× magnification.

2.8. Hemocompatibility of the HFMs

After obtaining written permission from the institute ethical committee, human blood was withdrawn from the antecubital area of the arm of a healthy volunteer in 10 ml EDTA coated Vacutainer® Plus plastic plasma tubes for hemolysis, and complement activation experiment. Informed consent was taken from the volunteers prior to blood collection for the experiments.

2.8.1. Hemolysis. Hemolysis testing was performed to evaluate the blood compatibility of HFMs, with hemolysis under 5% considered non-toxic a per ASTM F-756-08 standards. Human blood was centrifuged at 166 × g for 15 minutes to separate platelet-rich plasma (discarded) and packed erythrocytes. The erythrocytes were washed three times with normal saline solution (NSS, 0.9% NaCl) and adjusted to 50% hematocrit with NSS. HFMs, pre-washed with NSS, were incubated in 0.5 ml blood solution at 37 °C for 60 minutes. NSS and distilled water served as negative and positive controls, respectively. After incubation, samples and controls were centrifuged at 1000 × g for 5 minutes, and the absorbance of the supernatant was measured at 542 nm using a plate reader. The hemolysis ratio (HR) was calculated using the following equation:
AS: absorbance of sample supernatant, AP: absorbance of positive control (DI water), AN: absorbance of negative control (NSS).

2.8.2. Complement activation. Plasma was separated from human blood collected in a Vacutainer® tube by centrifugation at 1500 × g for 15 minutes. HFMs were pre-washed with NSS and incubated in the plasma at 37 °C for 30 minutes. The recovered plasma was diluted 1 : 10 using the specimen diluent provided in the MicroVue™ SC5b-9 Plus EIA kit. The terminal complement complex (TCC, SC5b-9) concentration was determined using the ELISA kit protocol (Quidel Corporation, San Diego, USA).^9,21–23

2.9. Separation performance of HFMs

The separation performance of HFMs was evaluated in the form of pure water permeability (PWP), molecular weight cut-off (MWCO), and bovine serum albumin (BSA) solute rejection, and toxin removal.

2.9.1. Pure water permeability (PWP). Pure water permeability (PWP) measures the ultrafiltration coefficient (K_UF) of the HFMs as volumetric flow rate per surface area per transmembrane pressure (ml m⁻² h⁻¹ mmHg⁻¹).⁷ During testing, DI water was passed through the lumen of the HFM samples at 60 ml min⁻¹ and 0.5 bar for 2 hours, with the permeate collected every 15 minutes. K_UF was calculated using the equation:
where Q is the permeate flow rate, n is the number of fibers, D_i is the inner diameter of the fiber, L is the effective length of the fiber, and ΔP is the transmembrane pressure. The average of six readings per sample and three samples per run was reported as mean ± S.D.

2.9.2. Molecular weight cut-off (MWCO) and bovine serum albumin (BSA) solute rejection. Hollow fiber membrane (HFM) samples were tested for molecular weight cut-off (MWCO) by evaluating the rejection of polyethylene glycol (PEG) with molecular weights of 1 kDa, 6 kDa, 20 kDa, 35 kDa, and 100 kDa. PEG solution (150 ppm) was passed through the lumen of the HFMs for 30 minutes at a flow rate of 60 ml min⁻¹ and a transmembrane pressure of 50 kPa. Percent rejection (%R_exp) was measured in the experiment using the equation
Here, C_p and C_F denote the concentration of permeate and feed, respectively.

PEG concentration was estimated using the colorimetric method, and the detailed protocol is explained in ref. 9.

To assess protein rejection, the BSA rejection of HFMs was measured similarly using a 1 g L⁻¹ BSA solution (66.5 kDa, comparable to human serum albumin). After 30 minutes, permeate collected from the shell side was analysed using the Micro BCA Protein Assay Kit (Pierce Biotechnology, USA). The same equation was used to calculate % BSA rejection.

2.9.3. Hemodialysis experiment for toxin removal. A mixture containing urea (1500 mg L⁻¹), creatinine (100 mg L⁻¹), lysozyme (40 mg L⁻¹), and indoxyl sulfate (40 mg L⁻¹) in buffer solution was circulated through the lumen side of the dialyzer module at a flow rate of 100 mL min⁻¹, while the dialysate was perfused through the shell side at 200 mL min⁻¹ under a transmembrane pressure (TMP) of 150 mmHg. The dialysis experiment was conducted over a period of 4 hours in accordance with ISO 8637:2004 standards. The concentrations of urea and creatinine were quantified by an ISO-certified pathology lab, whereas lysozyme and indoxyl sulfate concentrations were determined using a spectrophotometer (SpectraMax® iD3 microplate reader, molecular devices) at 280 nm and 270 nm, respectively. The detailed procedure is explained in our previous work.⁹

3. Results and discussion

3.1. Physicochemical characterization of TiO₂, and GO

Physicochemical characterization of TiO₂ and GO nanomaterials was performed using SEM, TEM, XRD, and FTIR to determine their morphology, size, structure, crystalline phase, and functional groups.

3.1.1. Morphological study of the TiO₂ and GO by FESEM and TEM. The morphology of TiO₂ and GO is illustrated in Fig. 2. SEM and TEM images of TiO₂ nanomaterials reveal a spherical structure. HRTEM images showed atomic arrangement of TiO₂ nanomaterials with a lattice spacing of 0.35 nm, corresponding to the (101) crystal plane of anatase TiO₂. The average particle size of TiO₂, as measured from TEM images, was ∼86.73 nm. SEM and TEM images of GO exhibit a sheet-like structure with a crumpled morphology.²⁴

Fig. 2 (a) SEM; (b) TEM micrograph; (c) high resolution TEM micrograph; (d) electron diffraction pattern of TiO₂; (e) particle size distribution of the TiO₂ nanomaterial, calculated using b, TEM micrograph. SEM and TEM micrograph representing the spherical structure of TiO₂ having particle size ∼86.73 nm with standard deviation of 1.32. (f) SEM and (g) TEM micrographs of graphene oxide (GO) showing the sheet-like structure with a crumpled morphology. (h) XRD pattern of GO and graphite, (i) XRD pattern of the TiO₂ nanomaterial. (j) FTIR of GO nanosheets. The XRD plot of graphene oxide shows the diffraction peak at 11.26° whereas for graphite the peak is at 26.45°. A change in peak was observed due to the formation of graphene oxide by oxidation of graphite. The XRD pattern of the TiO₂ nanoparticles shows maximum intensity peaks of anatase phase (JCPDS card no. 21-1272). The FTIR spectra of graphene oxide (GO) show oxy-functional groups in it, like carboxyl, epoxy, hydroxyl, and carbonyl.

3.1.2. XRD analysis of GO and TiO₂. The XRD results of graphene oxide (GO), presented in Fig. 2, exhibit a prominent peak at 11.26°, corresponding to an interlayer spacing of 7.84 Å. In comparison, the XRD pattern of graphite shows a characteristic peak at 26.45°, indicating an interlayer spacing of 3.36 Å. The shift in the diffraction peak from 26.45° for graphite to 11.26° for GO reflects a significant increase in interlayer spacing, attributed to the oxidation and intercalation of oxygen-containing functional groups within the graphene oxide layers.^25,26

The XRD pattern of TiO₂, presented in Fig. 2, exhibits prominent diffraction peaks at 25.30°, 36.95°, 37.81°, 38.57°, 48.04°, 53.88°, 55.06°, 62.68°, 68.77°, 70.29°, 75.06°, and 82.68°. The sharp and well-defined peaks confirm the crystalline nature of the TiO₂ nanoparticles. Notably, the strong peaks at 25.30° and 48.04° are characteristic of the anatase phase of TiO₂ as per JCPDS card no. 21-1272.^9,27–29

3.1.3. FTIR analysis of GO. FTIR spectra of GO shown in Fig. 2 confirm the presence of various functional groups in it, like carbonyl (C=O), carboxyl (–COOH), epoxy (–C–O–C), hydroxyl (–OH), and aromatic (C=C). The peak at 3672.26–3274.95 cm⁻¹ depicts the peak for hydroxyl bonds, 1720.41 cm⁻¹ for C=O bonds, 1616.26 cm⁻¹ for C=C bonds, 1384.81 cm⁻¹ for C–OH bonds, and 1224.73, 1051.15 and 540.05 cm⁻¹ for C–O–C bonds.²⁵

3.2. Physicochemical characterization of HFMs

3.2.1. Morphology studies by SEM. The morphology of the HFMs was observed using SEM. Fig. 3 shows the overview, cross-section and inner surface of the HFMs. We observed concentric HFMs as shown in the overview and the cross-sectional side-view of HFMs having a finger-like structure. The inner surface of the HFMs shows the porous nature of the membrane.

Fig. 3 (a)–(f) FEGSEM images of PES (P), TiO₂/GO (TG(0.5/1.5), TG(1/1), TG(1.5/0.5)), 2% TiO₂ (2T), and 2% GO (2G) PES-based HFMs showing the over-view, cross-sectional view, and inner pore structure of the HFMs. The overview shows the concentric nature of the HFMs, the cross section shows the large finger-like structure and the inner surface shows the porous structure.

3.2.2. Elemental mapping of HFMs. Elemental mapping of the hollow fiber membranes (HFMs), as shown in Fig. 4, confirms the successful incorporation of TiO₂ into the HFMs (TG(0.5/1.5), TG(1/1), TG(1.5/0.5), and 2T) during the spinning process. The mapping results reveal a uniform distribution of titanium throughout the membrane matrix. Elemental composition data presented in Table 2 further demonstrate that the atomic percentage of titanium increases with higher TiO₂ concentrations during spinning, following the sequence TG(0.5/1.5), TG(1/1), 2T, and TG(1.5/0.5). The high atomic percentage of titanium in TG(1.5/0.5) could be attributed to localized aggregation of TiO₂ particles, as evidenced by elemental mapping. This slight clumping may have led to deviations from the trend.

Fig. 4 Elemental mapping of HFMs showing successful incorporation of TiO₂ in TiO₂/GO and TiO₂ incorporated PES HFMs.

Table 2 Elemental composition (atomic%) of HFMs

S. no.	HFM sample name	Carbon (%)	Oxygen (%)	Sulfur (%)	Titanium (%)
1	P	75.16	17.43	7.41	0.00
2	TG(0.5/1.5)	73.87	19.51	6.16	0.45
3	TG(1/1)	72.45	20.40	6.26	0.89
4	TG(1.5/0.5)	70.84	19.99	7.41	1.77
5	2T	69.93	23.96	4.92	1.19
6	2G	74.21	19.56	6.22	0.01

---

3.2.3. XRD plot of the HFMs. The XRD analysis of the HFMs confirms the presence of both TiO₂ and GO, as shown in Fig. 5. The diffraction patterns reveal that the characteristic peaks of TiO₂ become increasingly prominent with higher TiO₂ concentrations. Also, sharp and intense peaks indicate high crystallinity of TiO₂. In contrast, the GO peak exhibits significantly lower diffraction intensity, appearing as a subtle hump. This peak becomes more discernible with increased GO content, particularly in the TG(0.5/1.5) and 2G HFMs. The observed diffraction patterns substantiate the successful incorporation of TiO₂ and GO into the HFM matrix, corroborating the successful formation of TiO₂/GO incorporated HFMs.

Fig. 5 (a) XRD plot, and (b) hydrophilicity evaluation of HFMs. The XRD plot of the HFMs shows the successful incorporation of TiO₂ and GO in the TiO₂/GO PES HFMs. Prominent peaks of TiO₂ are observed in TiO₂ incorporated HFMs (TG(0.5/1.5), TG(1/1), TG(1.5/0.5, and 2T). A small peak of GO is shown specifically for 2G and TG(0.5/1.5) HFMs. Contact angle results depict increased hydrophilicity as the concentration of TiO₂ is increased in the HFMs. (*) denotes p < 0.05, (**) denotes p < 0.01, and (***) denotes p < 0.001.

3.2.4. Hydrophilicity evaluation by contact angle. The contact angle analysis performed on the inner surface of the HFMs evaluates hydrophilicity, with lower contact angle values indicating enhanced hydrophilic behaviour, while higher values reflect hydrophobic characteristics. As shown in Fig. 5, the pristine membrane (P) exhibits a contact angle of 74.06 ± 0.93°, which progressively decreases with the incorporation of TiO₂ nanoparticles: 69.22 ± 1.30° for TG(0.5/1.5), 66.16 ± 0.95° for TG(1/1), 64.82 ± 0.69° for TG(1.5/0.5), and 59.57 ± 0.67° for 2T. The contact angle for the 2G membranes measures 69.22 ± 2.78°. These results clearly indicate that the hydrophilicity of the HFMs increases with higher concentrations of TiO₂, attributed to the intrinsic hydrophilic nature of the TiO₂ nanoparticles, thereby enhancing the membrane's surface wettability. Additionally, incorporation of GO also enhances the hydrophilicity confirmed from the low contact angle of 2G as compared to pristine PES (P).

3.2.5. Roughness of the HFMs by AFM. The AFM results shown in Fig. 6 represent the surface roughness of the HFMs showing that the TiO₂ and GO incorporated HFMs exhibit high surface roughness R_a and R_Rms (TG(0.5/1.5), TG(1/1), TG(1.5/0.5), 2T, and 2G) as compared to plain PES (P) HFMs. This enhanced surface roughness could be due to formation of nanoscale bumps on the outer surface by non-solvent induced phase separation, mediated due to diffusion of this hydrophilic additive towards the non-solvent side (water). Also, inorganic fillers like TiO₂ and 2D nanomaterials, such as GO, when incorporated within a polymer such as PES, interfere with the polymeric arrangement, and can lead to non-uniform phase separation with some surface protrusion leading to enhanced surface roughness.

Fig. 6 (a)–(f) AFM images of the HFMs. The modified HFMs exhibit increased surface roughness compared to plain PES (P) which is favourable for cell attachment.

3.3. Biocompatibility evaluation of HFMs

Biocompatibility of the HFMs was evaluated with human embryonic kidney HEK293 cell lines by measuring cytocompatibility as cell attachment and proliferation using confocal microscopy, cellular viability using MTT cell viability assay, live dead assay by flow cytometry and staining with calcein AM and PI dye.

3.3.1. Confocal microscopy study. Confocal micrographs (Fig. 7) demonstrate the attachment and proliferation of HEK293 cells on the surface of the HFMs. The results indicate that TiO₂/GO (TG) incorporated HFMs exhibit significantly enhanced cell attachment compared to pristine PES (P), 2% TiO₂ PES (2T), and 2% GO PES (2G) membranes. The surface of the TiO₂/GO incorporated HFMs displays dense cell coverage, indicating a more favourable substrate for cellular adhesion and growth. The synergistic incorporation of both GO and TiO₂ in the HFMs contributes to improved hydrophilicity, and surface biocompatibility, promoting superior cell attachment and proliferation of HEK293 cells relative to pristine PES and individual TiO₂ or GO incorporated membranes. Incorporation of TiO₂/GO into PES enhances spheroid formation, driven by the presence of TiO₂, while GO promotes cell spreading and attachment on the HFM surface.^9,19,30,31 Additional confocal micrographs are provided in the ESI† to corroborate the results.

Fig. 7 Attachment and proliferation of the HEK293 cell line, imaged with a confocal microscope. The confocal images of the HFMs display the attachment and proliferation of HEK293 cell lines on the HFMs (plain PES (P), TG(0.5/1.5), TG(1/1), TG(1.5/0.5), 2T, 2G). TiO₂/GO incorporated HFMs favour a higher density of cell attachment along with spheroid formation. Spheroid formation is observed prominently in the lower set of confocal images for TG(0.5/1.5).

3.3.2. MTT cell viability assay. The MTT cell viability assay measures the metabolic activity of viable cells, where mitochondrial dehydrogenase enzymes reduce the water-soluble yellow MTT reagent to insoluble purple formazan crystals. This reduction occurs only in metabolically active cells, making the amount of formazan produced directly proportional to the number of viable cells. The results of MTT assay shown in Fig. 8 demonstrate that the modified HFMs enhance cell viability, as indicated by increased formazan formation. Among the tested samples, modified HFMs illustrate significantly higher cell viability as compared to pristine PES (P) HFMs. TG(0.5/1.5), TG(1/1), and 2T exhibited absorbance above 0.85, indicating a significantly higher number of metabolically active viable cells. This result correlates well with confocal microscopy observations, confirming superior biocompatibility and enhanced cell attachment on these HFMs.

Fig. 8 (a) MTT cell viability assay, (b) live dead assay by flow cytometry, and (c) live dead staining with calcein AM and PI. The MTT cell viability assay shows the presence of a significantly higher number of viable cells on the modified HFMs as compared to plain PES (P). The live dead assay shows no adverse effect due to the presence of TiO₂ and GO in the HFMs. Live dead staining by calcein AM and PI dye further confirms the presence of a significantly higher number of live cells (green) on modified HFMs as compared to plain PES (P). (*) denotes p < 0.05, (**) denotes p < 0.01, and (***) denotes p < 0.001.

3.3.3. Live dead assay by flow cytometry. The FACS (fluorescence-activated cell sorting) analysis quantifies the percentage of live and dead HEK293 cells on the HFM surfaces, providing a direct assessment of cytocompatibility, as shown in Fig. 8. The low proportion of dead cells observed confirms that the cells attached to the HFMs remain healthy, indicating that the HFM scaffold offers a conducive environment for cellular growth. These results demonstrate the suitability of HFMs for biomedical applications, including bioartificial kidney (BAK) systems. Furthermore, the incorporation of TiO₂ and GO into the HFMs does not adversely affect cell viability, further supporting the biocompatibility of the composite material.

3.3.4. Live dead imaging by calcein AM and PI dye. Confocal micrographs in Fig. 8, obtained using calcein AM/PI staining, reveal a significantly higher density of viable HEK293 cells (green fluorescence) on the modified HFMs, with minimal red fluorescence corresponding to PI-stained non-viable cells. These results confirm the enhanced cytocompatibility of the modified membranes and highlight their potential suitability for bioartificial kidney applications.

3.4. Hemocompatibility of the HFMs

Hemocompatibility of the HFMs is explained using hemolysis and complement activation to show their suitability for blood contact application.

3.4.1. Hemolysis. A hemolysis experiment was conducted to evaluate the blood compatibility of the membrane material. Hemolysis refers to the rupture of red blood cells when blood comes into contact with the membrane surface. According to ASTM F-756-08 standards, the acceptable hemolysis limit for biomaterials is less than 5%. The results shown in Fig. 9 demonstrated that all HFM samples exhibited hemolysis levels within the permissible range, confirming the absence of the hemolysis limit, depicting their suitability for blood-contact applications. The hemolysis values for our HFMs fall within the range indicating the absence of any hemolytic effect as shown in Table S1 in the ESI† as per ASTM F-756-08 standards.

Fig. 9 (a) Hemolytic study, and (b) complement activation of HFMs. HFMs are hemocompatible with almost negligible hemolysis, making it a suitable biomaterial for blood contacting applications. Complement activation results show a significantly lower value of SC5b9 as compared to plain PES. Therefore, it can act as a potential biomaterial for blood contacting applications.

3.4.2. Complement activation. The activation of the complement system through various pathways results in the formation of the terminal complement complex (TCC, SC5b-9) by assembly of C5 through C9. Hemodialysis membranes are known to activate the complement system predominantly via the alternative pathway. TCC formation serves as an important indicator of membrane biocompatibility and the SC5b-9 marker is commonly used to assess disease activity and clinical manifestations. Elevated SC5b-9 levels indicate increased disease activity, whereas lower SC5b-9 levels reflect improved membrane biocompatibility and performance. From Fig. 9, it was observed that the modified HFMs (TG(0.5/1.5), TG(1/1), TG(1.5/0.5), 2T, 2G) showed significantly lower TCC as compared to pristine PES (P). This shows that the respective HFMs are suitable biomaterials for blood contacting applications.

3.5. Separation performance of the HFMs

Separation performance was tested using pure water permeability (PWP), molecular weight cut-off (MWCO), BSA rejection and toxin removal of urea, creatinine, lysozyme and indoxyl sulfate.

3.5.1. Pure water flux. The pure water permeability (PWP) experiment was conducted to determine the ultrafiltration coefficient (K_UF) of HFMs, defined as the volumetric flow rate per unit surface area per unit transmembrane pressure (ml m⁻² h⁻¹ mmHg⁻¹). A higher K_UF value is desirable for filtration applications. The K_UF increased progressively from 42.49 ± 2.79 ml m⁻² h⁻¹ mmHg⁻¹ for plain PES (P) to 133.81 ± 15.81 ml m⁻² h⁻¹ mmHg⁻¹ for TG(0.5/1.5), 147.07 ± 4.14 ml m⁻² h⁻¹ mmHg⁻¹ for TG(1/1), 149.7 ± 9.59 ml m⁻² h⁻¹ mmHg⁻¹ for TG(1.5/0.5), and 152.86 ± 5.01 ml m⁻² h⁻¹ mmHg⁻¹ for 2T. In contrast, 2G HFMs showed a K_UF of 111.81 ± 8.56 ml m⁻² h⁻¹ mmHg⁻¹. The results indicate improved ultrafiltration coefficient with increasing TiO₂ content, attributed to enhanced hydrophilicity. TiO₂/GO incorporated HFMs, TiO₂ PES, and GO PES HFMs demonstrated significantly higher K_UF values compared to plain PES (P) and commercial Hemoflow F6 membranes (20.87 ± 0.43 ml m⁻² h⁻¹ mmHg⁻¹). Fig. 10 illustrates the trend in K_UF values, following the order: Commercial < P < 2G < TG(0.5/1.5) < TG(1/1) < TG(1.5/0.5) < 2T.

Fig. 10 (a) Ultrafiltration coefficient, (b) MWCO, and (c) BSA rejection of HFMs. The ultrafiltration coefficient results showed significantly higher values in modified HFMs as compared to plain PES (P) and commercial HFMs. The MWCO experiment showing % PEG rejection of HFMs shows removal of low and middle MW solutes while retaining proteins in it. BSA rejection shows that the HFMs prevent protein loss from them, a desirable feature for hemodialysis and BAK application.

3.5.2. Molecular weight cut-off (MWCO) and bovine serum albumin (BSA) solute rejection. The molecular weight cut-off (MWCO) is defined as the molecular weight of a solute at which 90% retention occurs. Fig. 10 shows that the HFMs such as P, TG(1/1), 2T, and 2G exhibited MWCO values around 35 kDa, while TG(0.5/1.5) and TG(1.5/0.5) demonstrated rejection above 50 kDa. Despite these high MWCOs, the HFMs effectively retained essential proteins. To validate protein retention, BSA rejection was performed, as bovine serum albumin (∼66.5 kDa) closely resembles human serum albumin in molecular weight. Membranes suitable for hemodialysis must prevent albumin loss, and the developed HFMs demonstrated efficient albumin rejection, a desirable property for hemodialysis applications. Based on the MWCO and BSA rejection results, the HFMs demonstrate suitability for the removal of low and middle-molecular-weight solutes while effectively retaining essential proteins, making them appropriate for hemodialysis applications, which is further verified from the toxin removal test shown in Table 3.

Table 3 Separation % of uremic toxins tested on our HFMs. Urea clearance (UC), creatinine clearance (CC), lysozyme clearance (LC), indoxyl sulfate clearance (ISC), and albumin rejection (AR) are represented in the table. TG(0.5/1.5) showed the best results of toxin removal as compared to other HFMs

S. no.	HFMs	Urea clearance (UC)%	Creatinine clearance (CC)%	Lysozyme clearance (LC)%	Indoxyl sulfate clearance (ISC)%	Albumin rejection (AR)%
1	P	99.59 ± 3.5	94.56 ± 1.4	28.70 ± 0.1	15.38 ± 0.1	90.26 ± 0.2
2	TG(0.5/1.5)	98.22 ± 3.1	92.69 ± 1.8	40.47 ± 0.1	41.71 ± 0.4	95.86 ± 0.1
3	TG(1/1)	96.57 ± 3.3	96.85 ± 0.4	34.6 ± 0.5	30.11 ± 1.0	95.34 ± 0.2
4	TG(1.5/0.5)	97.94 ± 1.1	97.25 ± 0.2	35.42 ± 0.2	22.44 ± 0.5	97.92 ± 0.2
5	2T	97.94 ± 1.1	97.25 ± 0.2	35.35 ± 0.2	27.22 ± 0.4	97.92 ± 0.2
6	2G	97.94 ± 1.0	96.78 ± 0.2	30.02 ± 0.4	25.02 ± 0.4	97.22 ± 0.1

---

3.5.3. Uremic toxin removal efficiency. The results of the hemodialysis experiment demonstrated efficient removal of urea and creatinine, as well as effective clearance of lysozyme (a middle-molecular-weight solute) and indoxyl sulfate (protein-bound toxin). Table 3 presents the clearance rates of urea, creatinine, lysozyme, and indoxyl sulfate using the dialyzer module incorporating the developed HFMs, highlighting their suitability for advanced hemodialysis applications. From the results, TG(0.5/1.5) showed the best result of toxin removal of low MW, middle MW and protein bound toxins as compared to other HFMs.

Overall, TG(0.5/1.5) demonstrates promising potential for bioartificial kidney and hemodialysis applications due to its superior performance in key functional properties. It exhibits enhanced cell attachment and proliferation of kidney cells, effectively supports both cell attachment and spheroid formation, and promotes the presence of metabolically active viable cells. Additionally, it shows improved toxin separation efficiency. This balanced combination of cell proliferation support and separation capability makes TG(0.5/1.5) a suitable candidate for BAK and hemodialysis applications.

Table 4 represents the separation performance of previously reported literature and our current work. Our HFM shows the efficient removal of low molecular weight (urea, creatinine), middle molecular weight (lysozyme) and protein bound toxins (indoxyl sulfate).

Table 4 Comparative table summarizing the separation performance of our HFMs with previously reported literature. Here UC, CC, ISC and LC represent urea clearance, creatinine clearance, indoxyl sulfate clearance, and lysozyme clearance, respectively

S. no.	HFMs	Feed	Time (h)	Dialysis performance (%)				Ref.
				UC	CC	ISC	LC
1	Polysulfone (PSf)/iron oxide (Fe2O3)	Simulated	4	82	—	—	46.7	32
2	Dual layer Cu-BTC/PDA/PAN nanofiber	Simulated	4	92.8	82.3	—	—	33
3	Double layer MMM	Simulated	4	—		30	—	34
4	MMM-OIF	Simulated	4		62	—	—	35
5	DLHF-N-PMMA	Simulated	6	39.2	—	—	97.2	36
6	Poly(ether sulfone)–TPGS–graphene oxide	Goat blood		46.4	52.2	—	11.2	37
7		Simulated		91.2	93.3	—	28.3
8	PES/TPGS/nanozeolite	Goat blood	3	34.7	32.4	—	—	38
9	Commercial Hemoflow F60S	Goat blood	3	14.2	5.9	—	—
10	PES/GO	Simulated	3	86.3	89.3	—	—	19
11	PES/GO	Simulated	0.7	—	90.2	—	—	39
12	TiO2/PES	Simulated	4	97.9	97.3	27.2	35.4	9
13	TiO 2 /GO PES	Simulated	4	98.2	92.7	41.7	40.5	This work

---

4. Conclusions

We developed TiO₂/GO-doped PES-based hollow fiber membranes (HFMs) with the hypothesis that incorporating these nanomaterials would synergistically enhance the properties of HFMs due to both TiO₂ and GO, widely recognized for their cost-effectiveness, non-toxicity, high hydrophilicity, mechanical strength, and stiffness. Comprehensive characterization and experimental analyses demonstrated that TiO₂/GO PES-based HFMs exhibit superior suitability as a substrate for hemodialysis and bioartificial kidney (BAK) applications. Modified HFMs showed enhanced biocompatibility, supporting the adhesion, proliferation, and growth of HEK-293 cells. Confocal microscopy revealed extensive cell attachment, spreading, and spheroid formation on the TiO₂/GO membranes. MTT assay confirmed a higher number of metabolically active viable cells, and further live/dead assays by flow cytometry and staining with calcein AM and PI indicated no cytotoxic effects on HEK293 cell lines and significantly high live cells on modified HFMs. Hemocompatibility tests demonstrated non-hemolytic behaviour, confirming their safety for blood-contacting applications. Complement activation assays further indicated anti-inflammatory properties upon blood contact. Separation performance assessments revealed superior toxin removal efficiency for the TG(0.5/1.5) composition, highlighting a balanced combination of cell proliferation support and toxin clearance. Based on these findings, TG(0.5/1.5) demonstrates superior biocompatibility, anti-inflammatory properties, and enhanced functional performance, making it a suitable candidate for utilization in hemodialysis and bioartificial kidney applications.

Data availability

The data supporting the findings of this study are available from the authors upon reasonable request.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors acknowledge the Sophisticated Analytical Instrument Facility, Department of Chemical Engineering, Department of Physics, Department of Bioscience and Bioengineering, Microfactory, Tata Center, IIT Bombay, India, for characterization and instrumentation facilities, required for carrying out this study. One author (N. P.) acknowledges a research fellowship grant from PMRF, Government of India. The authors acknowledge research grants received from DST, India and Sine IIT Bombay, India.

References

1. C. Wojciechowski, M. Wasyłeczko, D. Lewińska and A. Chwojnowski, Molecules, 2024, 29, 2637.
2. N. Diban and D. Stamatialis, J. Chem. Technol. Biotechnol., 2014, 89, 633–643.
3. T. V. Plisko, A. V. Bildyukevich, L. Zhao, W. Huang, V. V. Volkov and Z. Huang, Fibers, 2021, 9, 28.
4. M. Safarpour, V. Vatanpour and A. Khataee, Desalination, 2016, 393, 65–78.
5. T. Laurenţiu, N. Valentin, B. Narcis, P.-L. Mirela, M. Emilian and C. Dana-Alexandra, J. Eng. Stud. Res., 2018, 22(4) DOI:10.29081/jesr.v22i4.254.
6. Z.-Q. Shi, H.-F. Ji, H.-C. Yu, X.-L. Huang, W.-F. Zhao, S.-D. Sun and C.-S. Zhao, Colloid Polym. Sci., 2016, 294, 441–453.
7. G. J. Dahe, R. S. Teotia, S. S. Kadam and J. R. Bellare, Biomaterials, 2011, 32, 352–365.
8. N. A. Alenazi, M. A. Hussein, K. A. Alamry and A. M. Asiri, Des. Monomers Polym., 2017, 20, 532–546.
9. N. Pandey, N. Kar and J. Bellare, ACS Appl. Bio Mater., 2025, acsabm.4c01309.
10. Y. C. Chuang, C. C. Chang, F. Yang, M. Simon and M. Rafailovich, Mater. Sci. Eng., C, 2021, 118, 111366.
11. L. Xue, P. Long, H. Wei and Y. Liang, Adv. Mater. Res., 2011, 299–300, 600–603.
12. S. R. Shin, Y. Li, H. Jang, P. Khoshakhlagh, A. Nasajpour, Y. S. Zhang and A. Tamayol, Adv. Drug Delivery Rev., 2017, 105, 255–274.
13. W. Zhou, Q. Liu, N. Xu, Q. Wang, L. Fan and Q. Dong, Membranes, 2023, 13(4), 377.
14. Y. Hou, K. Cai, J. Li, X. Chen, M. Lai, Y. Hu, Z. Luo, X. Ding and D. Xu, Int. J. Nanomed., 2013, 8, 3619–3630.
15. M. Vercellino, G. Ceccarelli, F. Cristofaro, M. Balli, F. Bertoglio, G. Bruni, L. Benedetti, M. A. Avanzini, M. Imbriani and L. Visai, Nanomaterials, 2016, 6(7), 124.
16. L. Zhao, J. Chang and W. Zhai, J. Biomater. Appl., 2005, 19, 237–252.
17. M. Safarpour, V. Vatanpour and A. Khataee, Desalination, 2016, 393, 65–78.
18. C. P. M. D. Oliveira, V. R. Moreira, Y. A. R. Lebron, C. K. B. D. Vasconcelos, K. Koch, M. M. Viana, J. E. Drewes and M. C. S. Amaral, Chemosphere, 2022, 306, 135591.
19. A. Modi, S. K. Verma and J. Bellare, J. Colloid Interface Sci., 2018, 514, 750–759.
20. N. Kar, D. Gupta and J. Bellare, Toxicol. Rep., 2021, 8, 1054–1066.
21. R. Teotia, S. K. Verma, D. Kalita, A. K. Singh, G. Dahe and J. Bellare, J. Mater. Sci., 2017, 52, 12513–12523.
22. S. K. Verma, A. Modi, A. Dravid and J. Bellare, RSC Adv., 2018, 8, 29078–29088.
23. S. K. Verma, A. Modi and J. Bellare, Colloids Surf., B, 2019, 181, 890–895.
24. B. Moongraksathum, P.-T. Hsu and Y.-W. Chen, J. Sol–Gel Sci. Technol., 2016, 78, 647–659.
25. R. K. Singh, R. Kumar and D. P. Singh, RSC Adv., 2016, 6, 64993–65011.
26. G. Surekha, K. V. Krishnaiah, N. Ravi and R. P. Suvarna, J. Phys.: Conf. Ser., 2020, 1495(1) DOI:10.1088/1742-6596/1495/1/012012.
27. T. Theivasanthi and M. Alagar, arXiv, 2013, preprint, arXiv:1307.1091 DOI:10.48550/arXiv.1307.1091.
28. W. Li, R. Liang, A. Hu, Z. Huang and Y. N. Zhou, RSC Adv., 2014, 4, 36959–36966.
29. G. Nagaraj, D. Brundha, C. Chandraleka, M. Arulpriya, V. Kowsalya, S. Sangavi, R. Jayalakshmi, S. Tamilarasu and R. Murugan, SN Appl. Sci., 2020, 2, 734.
30. A. Modi, S. K. Verma and J. Bellare, J. Colloid Interface Sci., 2017, 504, 86–100.
31. S. K. Verma, A. Modi and J. Bellare, Med. Devices Sens., 2021, 4, e10128.
32. N. Said, M. N. Z. Abidin, H. Hasbullah, A. F. Ismail, P. S. Goh, M. H. D. Othman, M. S. Abdullah, B. C. Ng, S. H. S. A. Kadir and F. Kamal, J. Appl. Polym. Sci., 2019, 136, 48234.
33. W. Li, S. Chao, Y. Li, F. Bai, Y. Teng, X. Li, L. Li and C. Wang, J. Membr. Sci., 2022, 642, 119964.
34. D. Pavlenko, E. Van Geffen, M. J. Van Steenbergen, G. Glorieux, R. Vanholder, K. G. F. Gerritsen and D. Stamatialis, Sci. Rep., 2016, 6, 34429.
35. O. E. M. Ter Beek, M. K. Van Gelder, C. Lokhorst, D. H. M. Hazenbrink, B. H. Lentferink, K. G. F. Gerritsen and D. Stamatialis, Acta Biomater., 2021, 123, 244–253.
36. M. N. Z. Abidin, P. S. Goh, N. Said, A. F. Ismail, M. H. D. Othman, M. S. Abdullah, B. C. Ng, H. Hasbullah, S. H. Sheikh Abdul Kadir, F. Kamal and S. Mansur, Sep. Purif. Technol., 2020, 236, 116216.
37. A. Modi, S. K. Verma and J. Bellare, ACS Appl. Bio Mater., 2020, 3, 1589–1597.
38. S. K. Verma, A. Modi, A. K. Singh, R. Teotia and J. Bellare, J. Biomed. Mater. Res., 2018, 106, 1286–1298.
39. M. Z. Fahmi, M. Wathoniyyah, M. Khasanah, Y. Rahardjo, S. Wafiroh and A. Abdulloh, RSC Adv., 2018, 8, 931–937.

---

Footnote

† Electronic supplementary information (ESI) available: Synthesis protocol of graphene oxide; synthesis protocol of the TiO₂/GO nanocomposite; physicochemical characterization of the TiO₂/GO nanocomposite; confocal micrographs of modified HFMs; hemolytic limit. See DOI: https://doi.org/10.1039/d5tb00229j

---