1 Introduction

One of the biggest challenges of the twenty-first century is the supply of adequate clean water for drinking water, industrial use, and sanitation. In addition to the fact that clean water is indispensable for vital activities, the rapid increase in the world population and industrialization also force natural resources. A wide variety of industries such as cosmetics, pulp and paper, and textiles, in addition to high water consumption for their processes, also cause the production of highly polluted wastewater (Ağtaş et al., 2020).

When various industries are compared, the textile industry stands out as one of the industries that produce the highest amount of wastewater. It has been reported that more than half (54%) of dye waste discharged into the environment worldwide originates from the textile industries. In addition, it is known that the dyes used in the textile industry are quite harmful to aquatic life and the environment. In this context, owing to the efficient treatment and reuse of textile wastewater, it is possible to reduce both the pollution load given to the environment and the clean water used by the textile industry (Agtaş et al., 2021).

Many traditional wastewater treatment methods such as biological treatment, flocculation-coagulation, adsorption, and advanced oxidation are widely used in the treatment of textile wastewater. However, traditional treatment methods cannot be sustainable in textile wastewater treatment due to reasons such as the content of textile wastewater affecting microorganisms or the need for high amounts of chemicals for effective treatment (Li et al., 2019). Considering the tightening of legal discharge standards day by day, membrane technologies come to the fore for textile wastewater treatment. In addition, membranes have high efficiency in treatment, need low space, and are highly effective in chemical and water recovery, making membranes an important option for textile wastewater treatment (Keskin et al., 2021a).

Photocatalysis, one of the advanced oxidation techniques, is used to decompose dye compounds using hydroxyl radicals and convert them into components such as mineral acids, water, and carbon dioxide. Various semiconductor materials such as ZnO (Aftab et al., 2022; Qi et al., 2017), CeO2 (Iqbal et al., 2022), ZnS (Hu et al., 2005), quantum dots (Korkut et al., 2023), metallic doped TiO2 (Ahmadpour et al., 2024), g-C3N4 (Qamar et al., 2023), vanadium oxide (V2O5) (Naseem & Durrani, 2021), and tin oxide (SnO2) (Ahmad et al., 2023) have been employed for the photocatalytic oxidation of organic substance. Among these materials, titania (TiO2) stands out with its low cost, high photocatalytic activity, and inert and chemical stability (Das & Basu, 2015). Titania particles must be removed from the effluent after the process has been carried out. However, since the particles are quite small in size, they are difficult to recover. Therefore, titania can be immobilized on various support materials such as cellulose fibers, various adsorbents, glass, and carbon (Alinsafi et al., 2007). In a study carried out to degrade textile dyes by photocatalytic method and to reduce their toxicity, immobilized titania was used. In this pilot-scale study, Acid Blue 25 dye was chosen as the main pollutant. In light of the data obtained, it has been reported that the target dye can compose and its toxicity can be reduced (Mahmoodi & Arami, 2009). In another study in which iron-doped titania was positioned on graphene oxide, the photocatalytic performance was investigated using Rhodamine B. As a result of the experiments, it was stated that the best Rhodamine B removal was found to be 91% after 120 min. It was underlined that a significant total organic carbon (TOC) and chemical oxygen demand (COD) removal was achieved in studies with real textile wastewater (Isari et al., 2018). Because of their porous surface, low cost, mechanical and chemical resistance, and tubular construction, TiO2/HNT nanocomposites (Fig. 1) have a lot of potential for eliminating water pollutants (Du & Zheng, 2014; Jiang et al., 2015; Mishra & Mukhopadhyay, 2019; Wang et al., 2011; Zheng et al., 2015). Dyes used in photocatalytic tests are mostly used in the textile and dye industry and are removed by adsorption and/or degradation (Abid et al., 2023).

Fluidized bed application is an application that has attracted attention in recent years in studies related to the coating of TiO2 on support materials. In this application, the balls of support material coated with TiO2 are usually “vented” in the reactor with the flow of water or gas supplied from the side or bottom of the reactor, allowing the bed to flow in the reactor. Thus, the system in which TiO2 is applied in suspension is imitated and the advantages of using TiO2 in a supported manner are utilized. In the studies on the applications of TiO2 in the fluidized bed reactor, such as quartz sand, sintered glass, which are less adsorptive, or different types of activated carbon which are high-capacity adsorptive materials have been used (Baek et al., 2013; Haarstrick et al., 1996; Kanki et al., 2005; Pozzo et al., 2011).

Fig. 1
figure 1

Schematic structure of TiO2/HNT

Full size image

To the best of our knowledge, there is no study in the literature using TiO2/HNT nanoparticles in UV-C photocatalytic reaction and directly with real textile wastewater. In the scope of this study, photocatalytic treatment and nanofiltration membranes and configurations have been tested for the treatment of cotton textile industry wastewaters which have been dyed at high temperature and alkaline conditions. In these experiments, real textile wastewater and photocatalytic reactor effluent were used. Then, NF membranes were tested to provide high flux, high dye, and divalent salt removal in real wastewater.

2 Material and Methods

This study consists of three main parts. These are UV photocatalytic pretreatment, NF membrane production, and NF application with real textile wastewater. In order to reveal the difference in pretreatment, a NF experiment set was conducted without pretreatment. The details of the methods used are explained in this section.

2.1 Wastewater Characteristics

Two samples of textile wastewater (WW) were used in this research. Heat recovery tank WW and printing WW obtained from a cotton textile factory in Turkey. The characteristics of wastewater is given in Table 1. Filtration tests were performed under 9 bar for 1 h constantly.

Table 1 Characterization of real textile wastewaters
Full size table

2.2 Reagents and Chemicals

HNT (halloysite nanotube) was supplied from EsanNANO Group (Turkey) as 5 μm average in size. Detailed information about HNTs was given in our previous study (Keskin et al., 2021b). HNT was used in both fluidized bed photoreactor and membrane production. Hydrochloric acid (HCl, 36.5–38%) was purchased from Sigma-Aldrich, Ltd. Titanium (IV) isopropoxide (TTIP, Ti[OCH(CH3)2]4) was purchased from Alfa Aesar.

For the production of ultrafiltration membrane, polysulfone (PSf, Udel® P-3500) was purchased from Solvay Specialty Polymers, USA. N-Methyl 2-pyrrolidinone (NMP) which was a solvent was bought from Ashland (USA).

To fabricate thin films, disodium-3–3′-disulfonate-4–4′-dichloro-diphenyl sulfone (s-DCDPS) was provided from Akron Polymer Systems, USA. N, N-dimethylacetamide (DMAc) and toluene were obtained from Sigma-Aldrich. 3-aminophenol (m-AP) was bought from Sigma-Aldrich and dried before use. Potassium carbonate and isopropyl alcohol were purchased from Sigma-Aldrich, Ltd. Also disodium 3–3′-disulfonate-bis[4-(3-aminophenoxy) phenyl] sulfone (s-DADPS) was synthesized by using these chemicals. Trimesoyl chloride (or 1,3,5-benzenetricarbonyl trichloride, TMC, 98%), anhydrous piperazine (PIP, ≥ 99.0%), and anhydrous cyclohexane (≥ 99.0%) were purchased from Sigma-Aldrich, Ltd. Reactive Orange (Mw: 617.54 g/mol) was purchased from Sigma-Aldrich, Ltd. Setazol Red reactive dye (Mw: 1469.98 g/mol) was kindly donated by Setas Chemicals, Tekirdag, Turkey. All experiments were conducted with ultrapure water.

2.3 Fluidized Bed Photoreactor and NF Hybrid System

2.3.1 Preparation and Characterization of TiO2/HNT Photocatalyst by Hydrothermal Method

The HNT nanotubes were dried at 110 °C for about 12 h. In this way, the water molecules between the layers are irreversibly removed. A one-step hydrothermal method was used to coat the HNT with TiO2 nanoparticles (Wang et al., 2011). Using titanium (IV) isopropoxide (1.73 mL) (TTIP, Ti[OCH(CH3)2]4, Alpha Aesar) as the initiator alkoxide, it was dissolved in 20 mL isopropanol (IPA, C3H8O, Aldrich) and stirred at room temperature for 2 h. On the other hand, HNT (1 gr) was added to 20 mL isopropanol, and the particles were dispersed in the solvent for 30 min by applying 40% power with the help of probe sonication (Qsonica Ultrasonic Processor—maximum power 700 W). Both mixtures were brought together in a Teflon container and adjusted to pH 4 with the aid of 6 M hydrochloric acid. As it is known, when the surface of the TiO2 catalyst is less than 6.3, the surface is positively charged (Evgenidou et al., 2005). However, the surface of HNT is negatively charged when pH < 2.4. Thus, positively charged particles can be easily absorbed onto HNT and decomposed effectively (Abdullayev et al., 2009). In this study, it is aimed to work at pH 4 as suggested in the literature (Wang et al., 2011). Then, the lid of the Teflon container was tightly closed, placed in a stainless steel autoclave, and kept in the oven set at 150 °C for 24 h. Afterwards, it was cooled to room temperature, removed from the steel autoclave, and washed 3 times with 5 mL of water (first by vortexing and then centrifuging at 4200 rpm for 5 min). TiO2/HNT composites in the form of white solid obtained were dried in an oven set at 110 °C for 12 s after grinding in agate mortar until they became fine powder.

The structure and morphology of raw HNTs and HNTs coated with TiO2 (TiO2/HNTs) photocatalysts were characterized by field-emission scanning electron microscopy (FE-SEM)(FEG-SEM Zeiss Leo Supra 35VP) and high-resolution transmission electron microscopy (HRTEM) (JEOL ARM 200 CF microscope with aberration-corrected probe and accelerating voltage of 200 kV).

2.3.2 Photocatalytic Degradation Experiments

The UV reactor assembly for the fluidized bed photoreactor system was built in Turkey by a company experienced in reactor preparation. In order to ensure maximum efficiency in the reactor, an internally illuminated model was prepared, which is more suitable for the fluidized bed system (Braham & Harris, 2009). The schematic of the designed reactor can be seen in Fig. 2c. The ready state of the reactor is also given in Fig. 2a, b. In the interior of the reactor, there is a quartz sheet in which the lamp to be used to activate the photocatalyst and to provide illumination from the inside is placed. The lamp used in the reactor is a UV-C lamp with a power of 42 W and a luminous intensity of 110 μW/cm2 at 1 m (LightTech, Austria). According to the data provided by the manufacturer, the lamp emits peak UV radiation at 253.7 nm. By giving dry air to the system, both the TiO2/HNT photocatalysts were aerated and the solution was mixed, and the oxygen needed in the photocatalytic reaction was provided to the reactor.

Fig. 2
figure 2

Images of the fluidized bed photoreactor (a, b), detailed schematic figure of reactor (c)

Full size image

Reactive orange (RO16, Mw: 617.54 g/mol) and setazol red (SR120, Mw: 1469.98 g/mol) chemical dyes were used as model reactive dyes for raw HNT and TiO2/HNT (2 g/L) adsorption and photocatalytic activity tests. Their chemical structures are given in Fig. 3. The initial concentrations of RO16 and SR120 were 100 mg/L and 200 mg/L. The solutions of dyes were prepared using ultrapure water. Prior to the photocatalytic activity test, the sample was stirred in the dark for 24 h to reach an adsorption–desorption equilibrium. An adsorbent-free experimental set was utilized for the direct UV light activity test. After stirring for 24 h, the UV light was turned on in the adsorbent-containing experimental sets. At some time intervals (about 1 h), dye concentrations were evaluated using a UV–vis spectrophotometer.

Fig. 3
figure 3

Chemical structure of RO16 and SR120

Full size image

2.4 Fabrication of NF Membranes

As the support layer, a polysulfone (PSf)-based porous membrane was used for the interfacial polymerization (IP) and fabricated as indicated in our previous study (Ormanci-Acar et al., 2020).

2.4.1 Fabrication of NF Membrane Containing HNT

First of all, for preparing an aqueous solution, 2% w/v PIP was added and dissolved in distilled water and 0.2% w/v TMC was dissolved in anhydrous hexane to get an organic phase solution at ambient temperature. HNTs were added into the aqueous phase and dispersed by probe sonication for 0.02% HNT TFN membranes. In the fabrication of 0.04% cycloHNT TFN membrane, HNT was added into TMC containing cyclohexane phase that has a higher polarity than hexane, a typical solvent used in TFC fabrication by interfacial polymerization, and thus provided a better dispersion before adding TMC. This process was carried out for 2 h at 90 watts before adding PIP. For 2 min, a side of the PSf support membrane was soaked in the aqueous phase solution by using a Plexiglas frame. The excess aqueous solution was decanted. Then, the same side of the support layer saturated with the aqueous-phase solution was soaked with the organic-phase solution for 60 s. The membrane was kept in an oven at 70 °C for 5 min to make the membrane surface more stable, after removing the excess organic solution by decanting. As a final step, membranes were rinsed with distilled water and stored in distilled water at 4 °C before characterization tests.

2.4.2 Fabrication of NF Membrane with Containing s-DADPS

For fabricating 80% s-DADPS TFC NF membrane, 0.4% w/v PIP and 1.6% w/v s-DADPS were mixed and dissolved in distilled water. To prepare the organic phase solution, 0.2% w/v TMC was dissolved in anhydrous hexane at ambient temperature. A side of the PSf support was put in the aqueous phase solution for 2 min. The excess aqueous solution was decanted. Then, the same side of the support layer saturated with the aqueous-phase solution was put in the organic-phase solution for 60 s. The membrane was kept in an oven at 70 °C for 5 min to make the membrane surface more stable after removing the excess organic solution. As a final step, membranes were thoroughly rinsed with distilled water and stored in distilled water at 4 °C before characterization tests.

2.4.3 Fabrication of NF Membrane with Containing s-DADPS and HNT

To fabricate 80% s-DADPS 0.01% HNT TFN NF membrane, 0.4% w/v PIP and 1.6% w/v s-DADPS were put and dissolved in distilled water. For preparing the organic phase solution at room temperature, 0.2% w/v TMC was dissolved in anhydrous hexane. HNTs were added into the aqueous phase and dispersed by probe sonication for 2 h at 90 watts before adding PIP. One side of the PSf support membrane was soaked in the aqueous phase solution for 2 min. The excess aqueous solution was decanted. Then, the same side of the support layer saturated with the aqueous-phase solution was soaked with the organic-phase solution for 60 s. After removing the excess organic solution by decanting, the membrane was kept in an oven at 70 °C for 5 min to make the membrane surface more stable. As a final step, membranes were thoroughly rinsed with distilled water and stored in distilled water at 4 °C before characterization tests.

2.5 Characterization of NF Membranes

Membrane surface properties were explained by roughness (Zeiss optic profilometer, USA), contact angle (KSV Attension Theta, USA), zeta potential, SEM, and Fourier-transform infrared spectroscopy (FT-IR). The roughness of membrane surfaces was measured by 3D Optical Surface Profilers (Zygo New view 7100, Middlefield, CT) with at least three membranes. In the sessile drop technique used in the contact angle measurement, a drop of distilled water was dripped onto the dry membrane surface. After the drop was kept on the surface for 5 s, 10 measurements were recorded at 25 °C with 1-s intervals. The measurement was repeated by taking samples from 5 different parts of the membrane. Results are reported with the mean and standard deviation of all measurements. Zeta potential analyses (SurPASS, Anton Paar GmbH) were done by circulating 500 mL of 1.0 mM KCl feed solution at pH 8 through an adjustable gap cell containing membrane samples. Helmholtz-Smoulchowski equation was used for zeta potential calculations (Paar, 2017). A piece of membrane coupon (1×1 cm2) was used for SEM analysis on the surface and cross-section layer. The samples were naturally air-dried and the dried samples were coated in 4.5 nm gold prior to SEM analysis. FT-IR spectroscopy (Spectrum 100, PerkinElmer, USA) was used to chemically analyze the membrane structure.

2.6 Experimental Set-up

Figure 4 shows the configurations for the treatment of wastewater.

Fig. 4
figure 4

NF alone membrane treatment (a), configuration of PC/NF treatment (b) for two real textile wastewater

Full size image

Membrane performance tests for water permeability and real wastewater removal were determined using the Sterlitech (HP4750 models) dead-end filtration system. Before filtration experiments, membranes were compacted for 30 min at 12 bars for NF membranes to obtain stable membrane performance. After compaction, pure water was filtered through the membrane under 9 bar for NF membranes. Pure water flux was calculated according to Eq. 1. COD was measured based on SM 5220-C and the conductivity levels of the samples were determined by conductivity meter (Hach Lange). TOC was measured by Shimadzu TOC V-CPH. Removal rates of COD, conductivity, and TOC were calculated by Eq.2. A UV–vis spectrophotometer (DR5000, Hach Lange, USA) was used to absorbance at wavelengths of 436 nm, 525 nm, and 620 nm (denoted as A436, A525, A620), and Pt–Co respectively. Chroma is denoted as color number (CN) and calculated in Eq. 3.

$$PWF (L {m}^{-2}{h}^{-1})=\frac{{m}_{water}}{{t}_{elapsed}*{\rho }_{water}*{A}_{membrane}}$$
(1)
$$R(\%)=\left(1-\frac{{C}_{permeate}}{{C}_{feed}}\right)x100$$
(2)
$$CN=\frac{{A}_{436}^{2}+{A}_{525}^{2}+{A}_{620}^{2}}{{A}_{436}+{A}_{525}+{A}_{620}}$$
(3)

3 Results and Discussion

Two wastewater matrices with different compositions were initially tested and compared in terms of dye and organic matter degradation and conductivity rejection by NF following PC pretreatment. In the PC step, TiO2/HNT nanoparticle was synthesized, characterized, and used as a photocatalyst in a fluidized bed photoreactor. In the NF process, novel membranes including variable monomers and nanoparticles were fabricated and described morphological properties.

3.1 Characterization and Photocatalytic Activity Test of TiO2/HNT

The structure and morphology of raw HNT and TiO2/HNT nanoparticles were characterized by FE-SEM. As shown in Fig. 5b, the surface of HNT has been successfully coated with TiO2 nanoparticles using the hydrothermal method. After coating with TiO2 nanoparticles, it is obvious that the dispersion is homogeneous and the cylindrical structure of HNT nanotubes is preserved.

Fig. 5
figure 5

SEM images of HNT (a), and TiO2/HNT (b)

Full size image

The morphological structures of TiO2/HNT nanocomposites were characterized by HRTEM. The HRTEM images of the nanocomposite structures are shown in Fig. 6. The size of the hollow HNT nanotubes was found to be about 1.2 µm, and the inner and outer diameters of the tubes were found to be about 10 and 50 nm, respectively. The TiO2 nanoparticles attached to the surface of the HNT nanotubes by hydrogen bonding were found to be dispersed throughout the HNT tubes, and the nanoparticle size was found to vary between 6 and 15 nm.

Fig. 6
figure 6

HRTEM and mapping images of TiO2/HNT

Full size image

The adsorption and photocatalytic degradation figures of RO16 (100 ppm) and SR120 (200 ppm) chemical dyes on raw HNT and TiO2/HNT (2 g/L) are shown in Fig. 7. Considering the photocatalytic degradation values, when the photocatalyst was HNTs, 73.98% of the 173 ppm SR120 concentration remaining after adsorption was degraded in 12 h and 83.81% in 16 h as a result of UV interaction. On the other hand, when TiO2/HNTs were used as a photocatalyst, 97.9% of the 175.9 ppm concentration of SR120 remaining after adsorption was degraded in 12 h and almost 100% in 16 h as a result of interaction with UV. In the UV control experiment, 46.5% of the SR120 was degraded. As a result, TiO2/HNT material provided 100% removal of SR120 when photocatalyzed by UV. While the adsorption of HNT photocatalyst was 37.37% in 24 h, this value was found to be 35% when TiO2/HNTs were used. In terms of photocatalytic degradation, only HNT degraded 51.52% of the 62.1 ppm RO16 concentration remaining after adsorption at the end of 7 h as a result of UV photodegradation. Besides, almost 100% of the 62.5 ppm concentration of RO16 remaining after TiO2/HNT adsorption was degraded as a result of UV treatment for 7 h. In the UV control set, 12.41% of the RO16 was degraded after 7 h photodegradation. Consequently, TiO2/HNT material provided 100% removal of RO16 when photocatalyzed by UV. According to Du et al., after 4 h of UV irradiation, a TiO2/HNT composite made using the sol–gel technique degraded 81.6% of methylene blue (Du & Zheng, 2014).

Fig. 7
figure 7

Adsorption and photocatalytic degradation of HNTs and TiO2/HNT for setazol red 120 (SR120) and reactive orange 16 (RO16)

Full size image

The photochemical processes occur when TiO2 nanoparticles are irradiated with UV light in the presence of HNTs (Mishra & Mukhopadhyay, 2019). HNTs act as electrical insulators, preventing charge transfer during UV irradiation and contributing to the efficient movement and separation of electrons (e−) and holes (h+) on the TiO2 surface. Additionally, HNTs enhance dye degradation by bringing dye molecules closer to TiO2 through electrostatic forces. Upon UV irradiation, a photoelectron is generated, creating a hole in the valence band and an electron in the conduction band of TiO2/HNTs. These photogenerated electrons reduce Ti+4/HNTs cations, while holes oxidize O2−* anions, leading to the production of OH* radicals. These radicals act as potent oxidizing agents, degrading organic molecules adsorbed on the TiO2/HNT surface and present in the surrounding environment. Simultaneously, O2 voids are created on the surface, filled by water molecules, resulting in increased surface hydrophilicity. The efficiency of organic molecule breakdown depends on their stability and structure. The electrons in the conduction band react with O2, generating superoxide radicals (O2 − *), which accelerate the oxidation process and prevent electron/hole recombination. The formation of highly reactive hydroxyl radicals (OH*) through the reaction of H + with O2 − * contributes to pollutant degradation (Abid & A. Ben Haj Amara, M. Bechelany, 2023).

3.2 Characterization of Membranes

The characterization results of the fabricated membranes are given in Table 2. The membranes utilized were also a part of our group’s earlier research; nevertheless, to aid in comprehension, the membrane characterization outcomes were also provided within the parameters of this research, with pertinent group articles cited for each outcome.

Table 2 Characterization results of membranes
Full size table

Optic profilometry was used to determine the roughness of all membrane surfaces. Analysis was repeated with at least three different membrane samples. It is expected that the membrane surface roughness will increase with the expansion of the surface area. Rougher surfaces provide an increase in the hydrophilicity, and an enhancement of the water flux of the membrane (Freger, 2003; Hirose et al., 1996). As presented in Table 2, according to the roughness results, it was understood that %0.04 cycloHNT TFN has the roughest surface. Hydrophilicity is also an important parameter that gives an idea about flux.

In order to determine the membrane hydrophilicity, contact angle measurements were made for all membranes produced. The average results of contact angle measurements are given in Table 2. The contact angle of flat and smooth surfaces is higher than the ridge-and-valley surfaces (Wang et al., 2014). Supporting this knowledge, the %0.04 cycloHNT TFN had the highest contact angle.

Besides, the molecular weight of s-DADPS (673 g/mol) is considerably higher than that of PIP (86.14 g/mol). It is thought that this causes the transport of amine monomers to the organic phase (the phase containing TMC monomer) to be slower (Akbari et al., 2016; Song et al., 2005). Thus, since the IP takes place more slowly, a thinner and flawless layer can be obtained. If the improvement in the contact angle is considered in general, it can be said that this increase in hydrophilicity increases with the amount of carboxylic groups released as a result of the partial hydrolysis of the acyl chloride agent of the TMC monomer (Wei et al., 2013). Membranes containing s-DADPS monomer have lower both roughness and contact angle values. However, as can be seen in Table 2, HNT-added membranes became rougher and more hydrophilic (Ormanci-Acar et al., 2018).

As the negative charge of the produced membrane surface increases, the dye removal efficiency also increases. It is anticipated that membrane fouling from textile wastewaters containing negatively charged ions might be effectively prevented due to the higher negative surface charge of TFN membranes. The surface charge of the membranes was determined by zeta potential analysis. Zeta potential analysis is an important analysis that effectively and reliably determines the surface charge of membranes. Due to the unique structure of the NF membrane, the separation capacity varies depending on the steric effect due to small pore diameters and the charge on the pore surface (Donnan, 1924; Seidel et al., 2001). Positron annihilation lifetime spectroscopy measurements reveal that the porosity of commercial NF membranes varies from 1.3% (membrane active layer polyamide) to 9.7% (membrane active layer polyethersulfone) (Boussu et al., 2007). The surface porosity can also be found by calculating the ratio of the area of the pores to the total area, which is 11.7% for NF270 and 17.1% for NF90 (Hilal et al., 2005). Since it is known that NF membranes generally have a negative charge (Schaep et al., 1998), Table 2 shows that the produced membranes also have negative surface charges at pH = 8–8,4. Carboxylic and sulfonic acid groups from TMC and s-DADPS are effective in making the surface more negative. These functional groups are hydrolyzed to their protons at natural pH and become more negative as the pH rises. Many studies say that pH change has an effect on the membrane load, depending on the separation of functional groups (Bellona et al., 2004; Childress & Elimelech, 1996; Schaep et al., 1998; Tiraferri & Elimelech, 2012). Hu et al. (2016) stated that the -SO3H groups in the sulfonated monomer increase the negative groups of the membrane surface (Hu et al., 2016). The -SO3H groups in the s-DADPS monomer employed in this research exhibit the same effect.

FT-IR was used to determine the functional groups of NF membranes. The results obtained are given in Fig. 8. The IR bands at 1322 cm−1, 1292 cm−1, 1238 cm−1, and 1148 cm−1 show typical polysulfone polymer characteristics (Misdan et al., 2013). The peak at 1639 cm−1 shows the interfacial polymerization, i.e., the C = O band. In addition, the jump of the C-N amide group located in the peak at 1584 cm−1 provides evidence for successful interfacial polymerization (Veerababu et al., 2014). Symmetric and asymmetric stretching of S = O bonds emerges in the peak at 1311 cm−1 (Akbari et al., 2016). The peaks at 1013 cm−1 and 1079 cm−1 show the jumping of the sulfonate groups of the s-DADPS monomer (Akbari et al., 2016). 1003.74, 908.11, and 748.17 cm−1 bands indicate the presence of HNT (Theng et al., 1982). The 1003.74 cm−1 peak shows the Si–O-Si vibrations of HNT (Bordeepong et al., 2011). The vibration at 912 cm−1 represents the hydroxyl groups responsible for the presence of HNT. The peak at 748 cm−1 is indicative of the Al–OH surface hydroxyl groups of HNT (Farmer, 1974; Frost, 1995).

Fig. 8
figure 8

FT-IR spectrum of fabricated NF membranes

Full size image

The surface morphologies of whole NF membranes were examined by SEM and the images obtained are given in Fig. 9. When the surface images are examined, it is noteworthy that the surface of the bare PIP (0%) membrane is less smooth. As the HNT ratio increases, wavy structures form on the surface, making the membrane surface rougher when Fig. 9c and d are compared. This can be verified with RMS values 0.20 and 0.30 of 0.02% HNT and 0.04% cycloHNT respectively. It was determined that ring-like structures were formed in the membrane containing 80% S-DADPS (Fig. 9b). With the addition of 0.01% HNT, these rings became denser, and a rougher surface was obtained, in line with Ghanbari and Mozia et al. work (Ghanbari et al., 2015; Mozia et al., 2019) (Fig. 9e).

Fig. 9
figure 9

SEM images of NF membranes (A bare PIP TFC, B %80 s-DADPS TFC, C %0.02 HNT TFN, D %0.04 cycloHNT TFN, E %80 s-DADPS %0.01HNT TFN)

Full size image

3.3 Performance Results of PC/NF and Alone NF Processes

3.3.1 Heat Recovery Wastewater

In this case, two processes were applied for the treatment of the WW from the heat recovery process. Firstly, raw wastewater was fed to a UV photocatalytic reactor (PC) as a pretreatment step. Then, the effluent from the PC was used as a feed solution for TFC and TFN NF membranes produced in this research. In the second process, raw wastewater was fed to alone NF membranes directly. The WW fluxes and the removal efficiencies of COD, conductivity, and color were compared to determine the effects of pretreatment. The treatment performance and effluent characteristics of the PC process are shown in Table 3. COD removal was achieved > 90% in wastewater completely decolorized with PC. However, a significant amount of COD concentration remained in the effluent. HNT can be utilized as support to reduce TiO2 agglomeration and raise the specific surface area (and subsequently the absorbance and catalytic activity), as several groups have shown (Wang et al., 2011). In Alinsafi et al.’s (2007) study, the COD removal reached a maximum of 90% by UV photocatalysis in textile WW (Alinsafi et al., 2007). After PC, the pH value of WW decreased to 3.58. The representation of the dye degradation mechanism by hydroxyl radicals (OH•) is articulated as follows (Fig. 10).

Table 3 Characterization of PC effluent and the removal efficiency of PC
Full size table
Fig. 10
figure 10

Mechanism of dye degradation by OH radical (Uma et al., 2019)

Full size image

Furthermore, when the pH value was above 10, the average reaction rate increased. This is because when the solution is sufficiently alkaline (pH values more than 10), the increased concentration of OH dot radicals increases the degradation rate. The findings align with Wei and Wan’s research on the photocatalytic oxidation of phenol, emphasizing that a pH above 10 facilitates a more rapid degradation of organic compounds (Wei & Wan, 1991). The efficacy of chemical oxygen demand (COD) exhibited a discernible decline as the initial concentration increased. This phenomenon can be attributed to the heightened concentration of dye and COD, resulting in an intensified coloration of the solution. Consequently, the diminished path length of photons entering the solution contributes to a reduced number of photons reaching the catalyst surface (Castillo-Suárez et al., 2023; Uma et al., 2019).

The NF flux performance analysis made with PC effluent is given in Fig. 11. The s-DADPS monomer causes macrovoid formation during cross-linking because its molecular weight is higher than that of PIP. In addition, since the s-DADPS monomer contains sodium sulfonate (NaSO3) ionic groups, the electrostatic surface of the membranes is negatively charged. The electrostatic surface changing and macrovoid spaces forming provide a more hydrophilic membrane and enhance salt and dye removal efficiency.

Fig. 11
figure 11

The flux of WW from heat recovery process produced by NF permeate with and without PC pretreatment

Full size image

The color removal efficiencies were 100% for all types of membranes. There was no difference between the two processes. COD and conductivity levels were used to characterize the content of organic and ionic substances in WW, respectively, and to reflect the degree of water pollution. Table 4 shows that the rejections to COD in the heat recovery WW were enhanced by the PC pretreatment. There was no difference in COD removal with these NF membranes in terms of concentration in the effluent and removal efficiency. This proved the positive contribution of PC pretreatment in terms of effluent concentrations.

Table 4 COD levels of permeate and removal rates for heat recovery WW
Full size table

The results show that PC pretreatment led to a tenfold decrease in COD values in the effluent (Table 4). Also, conductivity values were lower in the pretreated effluent compared to the non-pretreated effluent (Table 5). Notably, the bare PIP TFC membrane achieved a 95–96% success rate in conductivity removal. Furthermore, PC pretreatment of the bare PIP TFC membrane increased the flux from 18.3 to 27.6 L/m2.h (Fig. 11).

Table 5 Conductivity levels of permeate and removal rates for heat recovery WW
Full size table

3.3.2 Printing Wastewater

The other WW was obtained from the green printing process in the cotton textile industry (Table 6). The COD removal efficiency was succeeded at > 96% rate in a similar heat recovery WW case. After PC pretreatment, pH was declined heavily to 2.60.

Table 6 Performance results after photocatalytic reactor pretreatment
Full size table

The flux values of NF membranes are portrayed in Fig. 12. Sulfonation is one of the most frequently used and one of the most important chemical modification processes used for polymeric materials. The sulfonation process can be defined as the covalent bonding of the sulfonic (-SO3H) group to the polymer chain. The sulfone group that enters the polymer structure gives the material very unusual and beneficial properties compared to its premodification properties such as good wettability, water solubility, ion exchange and transfer properties, and membrane properties. Considering this, the high flux obtained in TFC and TFN membranes containing 80% S-DADPS shows that the additive is in place. It also gave similar results with the flux values in the literature (Xie et al., 2012). It is thought that the sulphonic groups contained in the S-DADPS monomer increase the water passage through the membrane. On the other hand, it is thought that a thinner layer is formed because the S-DADPS monomer, which has a higher molecular weight, passes into the organic phase more slowly (Xie et al., 2012).

Fig. 12
figure 12

The flux of wastewater from green printing process produced by NF permeate with and without photocatalytic pretreatment

Full size image

It was observed that PC pretreatment slightly reduces the flux (Fig. 12). However, the surface of HNT is negatively charged when the pH is above 2.4. Thus, positively charged particles can be easily absorbed onto HNT and decomposed effectively (Isari et al., 2018). According to the results of TFN NF membrane tests, the introduction of HNT in the active layer reduced flux values but did not affect removal efficiencies significantly as it is compatible with the literature (Tables 7 and 8) (Mozia et al., 2019).

Table 7 COD levels of permeate and removal efficiencies for printing WW
Full size table
Table 8 Conductivity levels of permeate and removal efficiencies for printing WW
Full size table

When Table 7 is examined, it was understood that NF membranes were quite successful in organic matter removal. Again, 80% S-DADPS TFC membrane had the highest values among other membranes with 98% and 86%. It was followed by high-efficiency HNT-doped membranes. It was seen that both flux and efficiency improvement were achieved. The “steric effect” and the “charge effect” are both involved in the NF membrane separation mechanism. It is achievable to reject charged solutes that are larger than the membrane pores. Multivalent negative ions are generally separated by NF membranes with a greater rejection degree than monovalent ions. This could make it conceivable to extract ions from wastewater and divide them according to their ionic valences (Fievet et al., 2002). It can be said that the moderate performance of doped membranes in conductivity removal is due to the fact that textile wastewater contains monovalent salts rather than divalent salts (Table 8).

Color removal in wastewater during the photocatalysis process is the result of decreased concentration of the dyes as well as the cleavage of the -N = N- bond, which determines the colors of the dyes. Removal of color does not mean that organic matter has been completely removed from the treated solutions. The decomposition of dyes can lead to smaller organic molecules that do not provide coloring, but organic matter is still present in the solution (Ormancı-Acar et al., 2022). Since the dye substances are purified in the photocatalytic reactor, there is no decolorization load left for the NF process.

Considering the overall results of the study, the optimum PC + NF combination in terms of treatment and filtration performances is compared in Table 9. In heat recovery WW treatment, although the highest flux was obtained with %80 s-DADPS TFC NF recorded as approximately 71.4 L/m2.h, the best treatment performance was obtained by bare PIP TFC NF membrane. In NF treatment in printing WW, the 80% s-DADPS TFC NF membrane provided a flux of 42.7 L/m2.h. The removal efficiency of organic matter was 98.1%, while the conductivity was 73.1%. Compared to other studies using TiO2, almost two times higher results were obtained in terms of treatment performance.

Table 9 Comparison with the literature
Full size table

4 Conclusion

The use of TiO2/HNT in real wastewater with UV photocatalysis was the first in the literature with this study. TiO2/HNT, selected as a photocatalyst, showed that it has a higher degradation potential than raw HNT in characterization tests. In tests conducted with real wastewater, it provided complete decolorization. The results obtained in this study showed that photocatalytic treatment is totally efficient in removing the dye substances in real wastewater sources that have different compositions.

In heat recovery wastewater, the combination of UV photocatalysis and nanofiltration allows the production of water of higher quality than the individual processes with global removals in COD higher than 99% for all fabricated novel membranes. Almost 100% COD removal was achieved. Conductivity removal performance remained in the 50% band with NF membranes after pretreatment. Bare PIP membrane was > 95% effective in conductivity removal. It can be said that the moderate performance of doped membranes in conductivity removal is due to the fact that textile wastewater contains monovalent salts rather than divalent salts. Steady flux increased with pretreatment in bare and noval TFN membranes at varying degrees.

As a result of PC + NF and NF studies performed on the printing wastewater sample, > 98% COD removal was achieved. It has been determined that PC pretreatment does not have a dominant contribution to conductivity removal. But, treated wastewater quality was significantly enhanced in terms of COD and color.

If color removal was the basic goal of this work, the increase in COD removal is an additional positive factor, as it would improve the efficiency of a NF process. No pH adjustment is necessary and wastewater at high pH can be treated directly after suspended solids removal.