Ultrasonication-assisted co-precipitation synthesis of ZnO, Fe₂O₃, and MnO₂ nanoflowers for photocatalytic degradation of methylene blue

Abstract

The wastewater contaminated with dyes, especially those used in the textile industries, poses a serious environmental problem because of the large amounts of synthetic dyes, such as methylene blue, that are not easily treated using the traditional treatment processes. The existing methods are mainly partial degradation by adsorption or photocatalysis, which demonstrates an important research gap in the complete removal of dyes. The proposed research will establish a sustainable and efficient solution through the synthesis of inorganic metal oxide nanoparticles (ZnO, Fe₂O_3, and MnO₂) with different oxidation states (+ 2, + 3, and + 4). The innovation is that the complete degradation of dye has been done with nanoflower (NF) synthesized through an ultrasonicated-assisted co-precipitation process that improves surface morphology, purity, and yield. Material characterization techniques were used to confirm the nanoflowers with a high surface area. The influence of metal ion concentration, pH, and dye concentration on methylene blue dye removal efficiency was evaluated by optimization studies. The dye removal was the highest in MnO₂ NF (90.42% adsorption, 93.48% photocatalysis), then Fe₂O₃ and ZnO. MnO₂ NF has the best removal performance for MB, while ZnO NF has the highest retention of activity after the recycling experiments. Thus, MnO₂ NF is the best material under the fresh optimized conditions, while ZnO NF has a relatively better cycling stability. The research findings are that metal oxide flowers are very efficient, sustainable, and scalable to real-life wastewater treatment, which fills the research gap and provides a solution to the environmental problems caused by dye pollutants.

Introduction

Dye wastewater refers to polluted sewage generated from the manufacturing process of textiles, dyeing, and fabrics¹. It consists of extremely high concentrations of suspended solids (350–1200 mg/L), heavy metals, minor contaminants, nutrients, dissolved chemical oxygen demand (250–1000 mg/L), and about 109 microorganisms per millilitres². Up to 300,000 tons of wastewater, incorporating artificial pigments, are discharged into worldwide purification centres and resources³. Dyes, defined as chemicals that bind with the materials that give long-lasting colour, could be classified as ionic and non-ionic dyes, whereas ionic dyes could be further categorized as anionic and cationic dyes⁴. Methylene blue (MB), for instance, is a cationic or referred to as a basic dye commonly found nationwide extensively in fabrics, textiles, and dyeing manufacturing industries associated with dyeing leather, silk, colouring paper, as well as coating for paper stock⁵. Consequently, a wide variety of photocatalysts have been developed and investigated for water treatment due to their ability to effectively degrade organic pollutants under light irradiation^6–13.

The metal ions used in the selection also demand a band gap of the visible light spectrum to facilitate the photodegradation of the dye¹⁴. An example is zinc oxide (ZnO) with a wide band gap of 3.37 eV that passes the cut, and is significantly cheaper and readily available than other metals, such as TiO₂¹⁵. Its unique characteristics include plausible excitation binding energy of approximately 60 meV, high electron mobility, high transparency, and strong room temperature photoluminescence¹⁶. The hexagonal wurtzite edifice of ZnO further expedites a wide variety of morphological nanoflowers, for example, nanorods, nanowires, nano combs, nanotubes, nanocages, nano springs, nano helices, and nanoflowers (NFs)¹⁶. Most importantly, ZnO possesses photocatalytic properties. Although adsorption often takes place for dye removal, the photocatalytic approach acts as a second line of degradation, which further helps to oxidize the chemical component in the dye to the surroundings. So the activation site will not be saturated and is capable of continuously adsorbing to remove the dye mixtures¹⁷. To compare the performance between different nuclear charges or oxidation numbers of nanoflowers, Fe₂O₃ (+3) and MnO₂ (+ 4) have been selected due to the exemplary wide band gap of 1.9 eV and 2.3 eV. Along with ZnO (+2), the three different charges of metal oxides were synthesized, studied, and discussed in the dye wastewater treatment.

Even though the conventional displacement purge adsorption technique is applied to address the disadvantages of adsorption, this technique necessitates a rise in capital cost because more adsorbents are added to eliminate the existing situation¹⁷. It is at this point that the photocatalysis process is introduced to improve and increase the rate of dye degradation¹⁸. Typically, the precipitation method might have poor regulation of the morphology and size of the end-product. However, with the help of ultrasonic treatment, the crystallinity and size of the synthesized structures could be regulated precisely. The ultrasonic method will hypothetically increase the rate of crystallization of ions to form crystals. Sonication can be used to accelerate reaction kinetics by acoustic cavitation. The process creates a number of effects, including micro-mixing, local input of energy, and high shear forces. It encourages nucleation, homogenizes the dispersed phase of the precursors, and counteracts diffusion restrictions. This process does not alter the thermodynamic equilibrium of a system but can be used to overcome the nucleation barrier and promote homogeneous growth of particles. It is superior to simple aging and mechanical mixing. The ultrasonication-assisted precipitation method synthesises ZnO, Fe₂O_3, and MnO₂ metal oxides NF were synthesised to perform adsorption and photodegradation of MB dye in wastewater treatment. For the MB dye degradation part of the study, both adsorption and photocatalysis reactions were carried out. As adsorption will always reach a point of saturation, the adsorbent should be removed and a new adsorbent introduced. This could increase the cost of treatment. However, the photocatalytic activity enables dye oxidation to the surroundings, and sufficient space is spared for the new dye to be removed (adsorbate). This study aims to utilise photocatalysis to address the drawback of the adsorption part (the need for a new adsorbent) and enhance the complete degradation of dye from a more sustainable perspective. Thus, the objectives are to analyze the adsorption and photocatalytic activities of selective metal oxide nanoflowers on methylene blue dye removal and to evaluate the dye degradation efficiency. Additionally, to optimize the results of the dye removal process and investigate the reusability of nanoflowers in dye degradation. The novelty of this work lies in the ultrasonication-assisted precipitation synthesis of various types of metal oxide NFs, which enables the adsorption and photodegradation of dyes in wastewater treatment.

Methodology

Materials

Zinc chloride (ZnCl₂), iron (III) chloride (FeCl₃), sodium hydroxide (NaOH), methylene blue powder, hydrochloric acid (HCl), and ethanol, purity ≤ 100% were purchased from Sigma Aldrich. While zinc oxide (ZnO), Manganese (II) sulphate (MnSO₄), HMTA (Hexamethylenetetramine), iron (III) oxide nanopowder (Fe₂O₃), and manganese (IV) oxide (MnO₂), purity ≤ 100% were procured from Merck.

Synthesis of metal oxide Nanoflower (NF)

ZnCl₂ (3g) was dissolved in 100 mL of water and stirred for 30 min on a hotplate stirrer to enhance the homogeneity or uniform distribution of precursor agents to avoid aggregations of ions¹⁹. Following that, 20 mL of the stirred solution was subjected to 30 min of ultrasonic treatment using Cole-Parmer Ultrasonicator 230 V at a frequency of 35 kHz^19,20. Zak et al. ²⁰ revealed that in the first 5 min, the nanoseed remained partially reacted due to the limited time of reaction; at 15 min, the nanorod was formed instead, which acted as the sheets or petals of the NF. At 30 min, a densely distributed, flower-like nanostructured morphology. However, beyond the 30-minute reaction time, at 50 min, aggregation of NF and nanosheets will result in a greater NF, which reduces the total surface area exposed and allows fewer active sites to react with the dye. With that, the ultrasonication time was set to 30 min. Then, 20 mL of 1 M NaOH and 0.3 g of HMTA (acts as a surfactant) were added, along with ultrasonic treatment²¹. This leads to the production of white precipitates¹⁹. The purpose of adding surfactant is to prevent agglomeration or aggregation of the nanocrystals and maintain the growth rate of the NF to ensure it is within the nanoscale range. Aside from that, surfactant aids in precisely controlling and monitoring the morphology of the NF to enhance the flower shape of the nanoparticle by slowly releasing NH₃/OH⁻, buffering supersaturation, and enabling facet-dependent crystal growth along preferred directions²².

Initially, the ultrasonicator was set to ambient temperature, but due to ultrasonic agitation, its temperature was increased to 50 ℃. The residue was rinsed with water, filtered, and dried overnight at a temperature of 60 ℃¹⁹. To filter the samples, centrifuge the residues from the ultrasonic cleaner in a centrifuge at 4000 rpm for 5 min. Then, a similar procedure was repeated by replacing the Zn²⁺ metal ions with Fe³⁺ and Mn⁴⁺. The precipitates of iron (III) chloride and manganese (IV) chloride were reddish brown and dark brown, respectively. The chemical equations for the production of ZnO, Fe₂O₃, and MnO2 NFs are shown in Eqs. (1–5), respectively.

1

2

3

4

5

.

Characterization techniques

Scanning electron microscopy (SEM, Hitachi S-4800, Tokyo, Japan) was used to examine the morphology of the ZnO, Fe₂O₃, and MnO₂ NFs produced to compare their differences. The elemental and oxide compositions of the NF could be studied by X-ray fluorescence (XRF) using a Brucker S2 Puma Energy Dispersion X-Ray Fluorescence spectrometer with CuK, proportional to radiations of λ = 1.5416 Å, which range from 10° to 90° at room temperature. A microanalysis method was carried out to identify the elemental composition (i.e., Zn, Mn, O, and Fe) for the samples. The particle size analyser was used to evaluate the size of the NF within the nanoscale using the Horiba Laser Scattering Particle Size Distribution Analyser LA-960. Perkin Elmer LS 55 Fluorescence Spectrometer was implemented to conduct the photoluminescence test for all 3 NFs to detect their excitation energy and intensity for each sample towards visible light. The ultraviolet-visible spectroscopy (UV-Vis) diffuse reflectance spectra of all three NFs were then determined by a spectrophotometer named Hitachi U-3900. To study the thermal degradation on the weight loss of the samples, the thermogravimetric analysis is performed using the Mettler Toledo Thermogravimetric Analyser Model TGADSCI 1600. The heating rate was set to 10 C/min for all three NF samples. Since the NF is derived from different oxidation numbers of feedstock, which result in + 2, +3, and + 4 charged NFs, a study of the interaction of surface charges between the positively-charged NFs and the methylene blue dye to predict the stability of long-term colloidal dispersion was conducted using a zeta potential test conducted using Malvern Zetasizer Nano ZS ZEN3600. To perform Fourier transform infrared spectroscopy (FTIR), Bruker Tensor II was used to examine the functional groups of the structure of ZnO, Fe₂O_3, and MnO₂ NFs. This is to study the differences between the functional groups of the commercial nanopowder and the NF produced.

Adsorption analysis

The adsorption test of metal oxide NF on the MB solution was conducted at room temperature in the dark²³. Initially, ZnO (20 mg/L) was mixed with a 25 mg/L dye solution of MB. Then, the suspensions were stirred continuously for 90 min without any light presence²³. This aimed to optimize the adsorption-desorption equilibrium²³. To create a conducive experimental environment with no light presence at all, a double-layered bottom-trimmed corrugated box was prepared to be placed on top of the shaker. A 5 mL syringe with a nanofiber 0.22 μm were used to extract the solution and filter the NF out, as the presence of the NF might impact the absorbance results via UV rays later. Then, the dye concentration was determined by the UV-Vis adsorption spectra on the Shimadzu UV-1800 spectrophotometer²³. The amount of dye adsorbed on the metal oxide NF was determined by Eq. (6):

6

Where, (mg/g) refers to the adsorption capacity at a specific time, t; (mg/L) refers to the initial concentration of MB dye; (mg/L) refers to the concentration of MB dye at a certain time, t;  V (L) refers to the initial volume of MB dye solution; W (mg) refers to the amount of nanocomposite.

Photocatalytic activity

To perform photocatalysis, photocatalysts should be activated by light radiation (i.e., visible light or ultraviolet) or photons under suitable wavelengths for the electron-transfer process to proceed²⁴. Photocatalysts should undergo oxidation and reduction reactions simultaneously to operate sustainably at the highest rate of reaction²⁴. As conducted for the adsorption test, ZnO (20 mg/L) was mixed with the 25 mg/L dye solution of MB and agitated at room temperature for 1.5 h. The suspension was then placed under natural sunlight for up to 3 h. To avoid the evaporation of suspension under the sun, the aluminium foil was used to cover the openings of the conical flask. The reason for choosing natural sunlight instead of artificial light sources is that natural sunlight is safe for the operators to conduct the experiment, unlike xenon lamps or ultraviolet rays. Besides, the condition was set up according to the industrial outdoors sedimentation or wastewater tank, which utilizes the natural sunlight to proceed with the wastewater treatment, as the introduction of safe LED light might induce an extra charge to the treatment process. Hence, directly using the cost-effective natural source – sunlight rays was the best option. The solar photocatalytic experiments were conducted outdoors at Curtin University Malaysia, Miri, Sarawak, Malaysia, approximately 4°24′10″N, 113°59′36″E. The experiments were performed during 12pm. A 5 mL syringe with a nanofilter 0.22 μ m was used to extract the solution and filter the NF out, as the presence of the NF might impact the absorbance results via UV rays later. The dye removal efficiency (%) was calculated using Eq. (7). The experiment was repeated by manipulating the concentration of ZnO to 40 mg/L, 60 mg/L, 80 mg/L, and 100 mg/L. Last but not least, a similar procedure was performed by swapping the ZnO NF with Fe₂O₃ and MnO₂ NFs.

7

Where, (mg/L) refers to the initial concentration of MB; whereas (mg/L) refers to the concentration of residual MB after photocatalysis occurs in a specific period (t).

Optimization studies

The parameters were manipulated, for instance, the concentration of dye ranged from 5 to 25 mg/L, the pH ranged from 6 to 12, and the concentration of metal ions ranged from 20 to 100 mg/L. 1 cycle of 20 trial runs was conducted to analyze the interaction of the parameters in the study for a single NF. To enhance the high consistency and accuracy of the results, each run ought to be performed 3 times, and the average values are taken into account. Thus, to study the interaction of parameters with all NFs, a total of 180 runs were performed. Table 1 displays the factors to be manipulated to study their relative outcomes on the dye degradation efficiency and the scale of values to be input. Despite the total 20 runs generated by the design of experiments approach there were 15 runs with different sets of parameter values, but the last five runs had similar sets of values. This can be explained by the factorial, axial, and centre points overlapping and showing a similar output due to the iterations of different sets of parameters. The relationship between increasing pH levels or decreasing concentration of metal ions, as well as the concentration of dye, with the degradation of dye was investigated through the three-dimensional interface response plots and contour diagrams.

Table 1.

Factors to be manipulated to study their relative outcomes on the dye removal efficiency and their range of values.

Factors	Units	Type of criteria	Lower bound	Upper bound	-ve alpha	+ve alpha
Concentration of metal ions	mg/L	Factor	40	80	20	100
pH	unitless	Factor	7.5	10.5	6	12
Concentration of dye	mg/L	Factor	10	20	5	25
Dark removal	%	Response
Light removal	%	Response

Reusability

To examine the reusability of metal oxide NF for practical application, the samples have been reused in the dye removal process for 4 cycles (1st original treatment and 4 reused cycles). After the end of the process, the ZnO, Fe₂O₃, and MnO₂ NFs were recovered with centrifugation, where the solution of photocatalytic dye was spun at 4000 rpm for 5 min to precipitate the solid photocatalyst, which is the metal oxide NF²⁵. The subsequent step was the washing of the NF with ethanol once and distilled water up to three times, filtered, and then dried in a vacuum oven at 60 °C for 24 h²⁵. Next, the samples were stored separately in petri dishes at ambient temperature to be used in the subsequent degradation cycles²⁵.

Results and discussion

Characterization results

Scanning electron microscopy (SEM)

Fig. 1 demonstrates that a huge number of homogeneous as well as evenly sized blossom-shaped NFs coated the surfactant. It has been discovered that a 3D hierarchical ZnO NF with a clearly defined shape exists. Simultaneously, the high-resolution SEM image of the ZnO NF exhibited in Fig. 1c claimed that the NF structure comprises numerous extremely thin leaves, which can be seen from its appearance. The diameters of the extremely thin leaves vary from the base (roots) to the apex (tips)²⁶. It can be observed that the ZnO NF has thin leaves and large roots which connect all mutually in one common core in such a way as to output a well-defined morphology of a 3D hierarchical NF²⁷. Meanwhile, the diameter of monodispersed independent ZnO NFs ranges from 500 nm to 900 nm, which is very tiny and within the nanoscale²⁷. Additionally, the ultrathin leaves or petals of ZnO NF generated a sizeable surface-to-volume ratio for the mass transport property of catalytic applications²⁷. The industrial ZnO powder was showing a tiny form as well, but in an irregular shape, which is very different from the synthesized ZnO NF. ZnO NF was displaying the rose-type NF.

Fig. 1.

SEM images of ZnO (a) NF at low magnification 60 k (b) control powder at low magnification 60 k (c) NF at high magnification 160 k.

For Fe₂O₃ NF (Fig. 2), the output is more identical to the shape of the sheets. Hence, it should be claimed as nanosheets. Nanosheets are also a type of NFs. From the looks of the pure Fe₂O₃ powder, they exist in tiny spherical form. This could be due to the higher concentration used during the synthesis phase²⁸. While Tu et al. ²⁹ have reported spherical NFs for their FE@NTC, which consists of different chemicals²⁹. Multiple layers or sheets of Fe₂O₃ NF increase the total surface area in contact with the dye, thereby enhancing the performance of Fe₂O₃ NF in the degradation of the MB dye. The sheets appear slightly irregular in shape, which may indicate the presence of several active sites on the surface of the nanostructure itself. However, Fe₂O₃ NF showed a lower total surface area exposed compared to ZnO NF and MnO₂ NF.

Fig. 2.

SEM images of Fe₂O₃ (a) NF at low magnification 60 k (b) powder at low magnification 60 k (c) NF at high magnification 160 k.

For MnO₂ NF (Fig. 3), the NF is spherical instead of spiky. However, the high porosity of the spherical flower enhances the active sites for the particles to react with the dye. MnO₂ powder possesses an obvious rod shape, which shows an intense difference to the final morphology of the ball-shaped NF. However, when using a higher magnification to observe the NF, it can be noticed that the inner side of the spherical MnO₂ NF is the spiky form, in which the long rods of nanoparticles were dissociated to form smaller spiky or needle-like stems or petals for the NFs. This increases the total surface area exposed of the NF to the dye and doubtlessly optimizes the degradation of the dye³⁰. This spiky structure later aggregates and assembles during ultrasonication to build up the spherical structure of the flower. Although the composite particle size is not in the nanoscale, it was referred to as a nanoflower. This is due to the fact that at least one dimension or feature of them falls into the nanometer (nm) range and their synthetic regimes are typically based on bottom-up mechanisms, including molecular/atomic assembly. Thus, the overall size of the structure may be larger than in the 1–100 nm range. The size of the petals in each nanoflower increases as crystal growth progresses over time, and also the nanorods became wider and larger³¹.

Fig. 3.

SEM images of MnO₂ (a) NF at low magnification 60 k (b) powder at low magnification 60 k (c) NF at high magnification 160 k.

X-Ray Fluorescence (XRF)

Table 2 tabulates the oxide composition of the ZnO, Fe2O3, and MnO₂ NF samples. The oxide composition through the microanalysis method for each sample was ZnO – 97.2%, Fe₂O₃ – 95.44% and MnO₂ – 98.26%. Negligible impurities, such as chlorine of less than 5%, were accounted for in all samples, which showed the majority of components are the desired ZnO, FeO, and MnO. This revealed that ZnO, Fe₂O₃, and MnO₂ flower-like nanoparticles barely consist of zinc oxides, iron oxides, and manganese oxides, respectively, since there were negligible traces of significant impurities in the fluorescence spectrum. This claimed that the NF generated consists of the important oxides to treat dye wastewater. Besides, the output also confirmed that the ZnO has successfully inherited from the synthesis of ZnCl₂ with HMTA and NaOH, and the same goes for FeO and MnO. The impurities might be attributed to the washing process of the glassware using dilute HCl to remove the stubborn stain. Nevertheless, the high percentage of desired oxides displayed a positive outcome of the XRF analysis. The high purity of the NF, synthesized with its high surface area exposed to the wastewater, suggests that all three metal oxide NFs are feasible as adsorbents to perform dye removal. The elemental composition for ZnO, Fe₂O_3, and MnO₂ NF Zn – 83.12%, Fe – 86.01%, and Mn – 86.56% were, respectively. The high percentage further enhances the high purity of ZnO, Fe₂O₃, and MnO₂ NFs, which aim for greater performance in dye removal.

Table 2.

Oxide composition of (a) ZnO, (b) Fe₂O₃, and (c) MnO₂ NF.

Particle size analyzer (PSA)

Fig. 4 clearly exhibits the particle size distribution for ZnO, Fe₂O₃, and MnO₂ NFs, respectively. For ZnO NF, the median particle size was about 2.6217 μ m whereas the diameter size was around 934.2 nm. For Fe₂O₃ NF, the particle size was about 2.5413 μ m whereas the diameter size was around 558.6 nm. For MnO₂ NF, the particle size was about 4.4537 μ m whereas the diameter size was around 1.5557 μ m. The greater size of MnO₂ NF is due to the fact that manganese is a heavy metal with a greater atomic weight. Besides, the Mn is located at period 4 in the periodic table compared to Fe and Zn. Furthermore, the oxidation number or nuclear charge might affect the size of the particle due to electron-electron repulsion, as MnO₂ shows the greatest charge of + 4.

Fig. 4.

Particle size distribution for (a) ZnO, Fe₂O₃, and MnO₂ NF.

It is observed where the frequency distribution reaches a maximum or prominent peak. Since the frequency distribution consists of merely one maximum, it is known as a monomodal type of distribution. If two maxima were noticed, it is then called a bimodal distribution. In contrast, a particle size distribution graph with more than two maxima is referred to as a multimodal distribution³². Based on the graphs shown in Fig. 4, a monomodal particle size distribution was observed, indicating a uniform size of all NFs. This concludes that the average size of NFs is highly constant, hence all of them are capable of effectively treating dyes with a great amount of active sites. Uneven sizes may result in poor dye removal performance due to the reduced number of active sites at larger NF sizes, resulting in a smaller total surface area exposed. Besides, a monomodal distribution showing only the major component considered within the sample, as two or more maxima, resembles different types of particles found in the sample³².

The diameter of the particle is within the nanoscale of 500 nm, which justifies the ‘nano’ term of the NF samples. Despite the particle size being in < 4 μ m, the ‘nano’ term can still be used to define the flower or sheet-like particles due to the particles being grown from a nanoscale to a final shape. The process of synthesizing the ‘nanoflowers’ from the nanocrystals themselves is claimed as nanotechnology¹⁷. Therefore, despite the final particle size exceeding the nanoscale, it is still claimed as a ‘nano’ term. Besides, the small value of μ m indicating the structure of the ‘NF’ is still extremely tiny. Technically, above 10 μ m particles should then be named as microflowers. In some cases, the addition of surfactant, which is HMTA, does not fully guarantee the prevention of aggregation; instead, at times it might cause slight aggregation and interfacial absorption of the NF due to the hydrophobic tails and hydrophilic heads³³. Even though the surfactant prevents the aggregation of particles, the ultrasonication process promotes the growth of nano seeds to NFs. Once the growth of the NF exceeds a certain threshold, this process offsets the surfactant’s ability and explains the aggregation of the NF to a larger size. All in all, the particle size as illustrated in the SEM is identical to the PSA results for all 3 NFs (all diameters are within the range of 500 nm), indicating the high accuracy of the output to support all judgments claimed. The aggregates of MnO₂ measured by PSA are in the micrometer range. On the other hand, the spherical flower-like structures of MnO₂ in the SEM images are built up from rod-/needle-like nanostructures. Therefore, the particle of MnO₂ consists of a hierarchical flower-like microstructure, hereinafter denoted by the term “MnO₂ hierarchical flower-like microstructure”. This term not only denotes structures on the nanometer scale but also structures that are built up from nanostructures as the building blocks of the flower-like structure.

Photoluminescence (PL)

Through the photoluminescence approach, the efficiency of the charge carrier transmission, immigration, and trapping in a semiconductor can be studied. This technique involves the emission of a light spectrum after the absorption of photons, which covers electromagnetic radiation studies³⁴. Fig. 5 exhibits the excitation and emission bands of ZnO, Fe₂O₃, and MnO₂ NFs under the optical or visible light spectrum, respectively.

Fig. 5.

Photoluminescence analysis for (a) ZnO excitation and emission, (b) Fe₂O₃ excitation and emission, and (c) MnO₂ excitation and emission.

For ZnO NF emission, the first sharp dominant peak resembles the luminescence band at about 438 nm. This is attributed to the near band emission of ZnO (~ 3.37 eV¹⁵) and the recombination of free electron-hole pairs, which indicates the excitation³⁴. Two additional broad, weak intensity emission peaks were observed at around 506 nm and 600 nm. For the ZnO excitation, a dominant peak at 338 nm was noticed.

For Fe₂O₃ NF emission, the first sharp dominant peak resembles the luminescence band at about 385 nm. This is attributed to the near band emission of Fe₂O₃ (~ 1.9 eV based on¹⁵) and the recombination of free electron-hole pairs, which indicates the excitation. Another broad weak intensity emissions peak at around 420 nm was observed. For the Fe₂O₃ excitation, a dominant peak at around 240–255 nm was noticed.

For MnO₂ NF emission, the first sharp dominant peak resembles the luminescence band at about 438 nm. This is attributed to the near-band emission of MnO₂ (~ 2.3 eV, based on¹⁵) and the recombination of free electron-hole pairs, indicating excitation. No obvious broad weak intensity emissions peak was observed. For the MnO₂ excitation, a dominant peak at around 235–250 nm was noticed.

Since the extension of broadband peaks for all 3 NFs from 400 to 800 nm recline in the visible light region, showing the ZnO, Fe₂O_3, and MnO₂ NFs are capable of showing positive results during the photocatalysis reaction under the natural sunlight. Among them, MnO₂ NF illustrates a greater intensity for the broad green emission peak at 415 nm than other NF samples because of greater oxygen vacancies or extrinsic defects on the surface of the structure. The photoluminescence and photocatalytic behaviour for all 3 NFs are strongly impacted by the defect of the crystal associated with oxygen vacancies on the particles’ surface³⁴.

Optical studies / DRS UV-Vis (Diffuse reflectance spectroscopy ultraviolet-visible)

Fig. 6 displays the absorption spectra of ZnO, Fe₂O₃, and MnO₂ NF, respectively, as a function of visible light wavelength (200 nm to 800 nm). For ZnO NF, the graph illustrates a fundamental absorption edge at about 405 nm. The absorption peak is about 4.5 a.u at around 390 nm, which was attributed to ZnO’s inherent band gap absorption, following Katiyar et al.’s value of 380 nm³⁵. This is due to the electrons directly transitioning between the conduction and valence bands³⁵. For Fe₂O₃ NF, the graph illustrates a fundamental absorption edge at about 782 nm, but there is a sign or possibility (oscillation noticed) for this particular NF to have slight absorption despite exceeding this value. The absorption peak is about 3.75 a.u at around 480 nm. For MnO₂ NF, the hump is located at about 634 nm, but there is a sign or possibility (oscillation noticed) for this particular NF to have slight absorption despite exceeding this value. Although the first absorption peak is about 4.3 a.u at around 300 nm, the subsequent absorption peak is 4.0 a.u. located at around 460 nm, which is within the acceptable range of natural sunlight. Since the photocatalysis reaction is performed under natural sunlight, which is considered visible light, it possesses a light spectrum in a wide range of 380 nm to 700 nm. Since all three metal oxide NFs exhibit a plausible absorption ability within this range, they are capable of proceeding with the photocatalysis process to react with the dye under natural sunlight.

Fig. 6.

DRS UV-vis analysis for (a) ZnO, (b) Fe₂O₃, and (c) MnO₂ NF.

Thermogravimetric analysis (TGA)

Fig. 7 illustrates the results of the thermogravimetric analysis (TGA) and differential thermal analysis (DTA) curves for the ZnO, Fe₂O₃, and MnO₂ NFs, respectively. Through the TGA test, the thermal properties of all three NFs are studied. While observing the significant peaks from the graph, only the trough would be accounted for instead of the crest. Besides, the reading of the peak should be conducted beyond the room temperature, as any peak before the room temperature could result in the baseline correction regarding the start-up of the equipment³⁶.

Fig. 7.

TGA and DTA curve for (a) ZnO, (b) Fe₂O₃, and (c) MnO₂ NF.

For ZnO NF, there are two significant peaks (steeper slope) which are at the temperatures of 177 °C and 238 °C. For Fe₂O₃ NF, there are two significant peaks, which are at temperatures of 77 °C and 282 °C. For MnO₂ NF, there are two significant peaks, which are at temperatures of 240 °C and 520 °C. Every significant peak shows a significant drop in the weight of the NF samples. Besides, it also illustrates that at that particular temperature, the NF experienced irreversible thermal degradation, which reduced the quality of the NF. This might affect the performance of the NF in the degradation of the dye at that certain temperature. However, the majority of thermal degradation temperatures for the NF are at an extremely high temperature of about 177 °C –520 °C, which is not common for every wastewater treatment. Despite the thermal degradation temperature for Fe₂O₃ NF being much lower, at about 77 °C, the dye wastewater treatment is usually performed at a temperature of 20 °C– 65 °C³⁷. Therefore, this shows that all three NFs are capable of degrading dye effectively at the desired or optimized temperature.

For ZnO NF, the sample weight is about 92.7% at 177 °C, indicating a weight loss of 7.3%; at 238 °C, the residue is about 90.18%, with approximately 7.215 mg of samples remaining at 700 °C. As the temperature increases from 0 °C to 700 °C, the drop from 25 °C to 177 °C is due to the water dehydration process to remove excess liquid or vapour within the NF samples, slightly beyond the room temperature³⁸. Another significant drop at 238 °C is due to the further elimination of surface-adsorbed moisture and surfactant species, which HMTA consists of ZnO NF^36,38. However, there is still an insignificant drop in NF samples (less weight loss) as the temperature increases to 800 °C. This is because the decomposition of the precursor during the heating process initiates the release of water-trapped crystalline structure within the ZnO NF³⁶.

For Fe₂O₃ NF, the sample weight is about 83% at 77 °C, indicating a weight loss of 17%; at 282 °C the residue is about 73.58%, approximately 5.8871 mg of samples remaining at 700 °C. For MnO₂ NF, the sample weight is about 88.02% at 240 °C, indicating a weight loss of 7.3%; at 520 °Cthe residue is about 87.37%, approximately 6.9892 mg of samples remaining at 700 °C. The explanation that fits the trend of the graph for Fe₂O₃ and MnO₂ is similar to the ZnO NF justification as mentioned above. Fe₂O₃ NF tends to experience the first significant dehydration earlier than ZnO NF, but for the subsequent round of elimination of surface-adsorbed moisture and surfactant species, it has a similar temperature to ZnO NF. From that, it could be deduced that Fe₂O₃ NF is more heat sensitive than the rest since it experiences thermal degradation earliest and has the least remaining sample weight throughout the TGA test. However, MnO₂ NF only experiences the first thermal degradation due to the elimination of surface-adsorbed moisture and surfactant species at a higher temperature of 240 °C, but it experiences greater heat impact for the decomposition of precursor during the heating process, which initiates the release of water-trapped crystalline structure within the MnO₂ NF. Therefore, it has a lesser remaining sample weight compared to ZnO NF.

Zeta potential / electrophoretic mobility

Zeta potential aims to determine the stability of colloidal dispersion. Table 3 exhibits the zeta potential values for ZnO, Fe₂O₃, and MnO₂ NFs, which are − 19.4 mV, 24.1 mV, and − 10.1 mV, respectively. All values mentioned were within the literature range for ZnO ( -24.1 mV), Fe₂O₃ (32.5 mV – -19.4 mV), and MnO₂ (-10 mV – -37 mV) particles^39–41. This strengthens the reliability and accuracy of the results obtained and supports the judgment made.

Table 3.

Zeta potential and conductivity for metal oxide NFs.

Samples	Zeta Potential (mV)	Conductivity (mS/cm)
ZnO NF	-19.2	0.773
Fe2O3 NF	24.1	0.392
MnO2 NF	-10.1	0.0752

Since the methylene blue dye is a cationic or positively-charged dye, the negatively-charged NF is capable of reacting with the MB dye to neutralize it (return to pH 7). In other words, the application of metal oxide NF towards adsorption and photocatalytic dye wastewater treatment should show a positive outcome, as the positive and negative charges will attract both particles to react. A high magnitude of zeta potential (regardless of the charge) reveals a greater ability of dispersion with opposite charge particle surfaces. On the other hand, a low zeta potential value indicates a higher tendency towards particle aggregation or agglomeration³⁹. Since the overall value is greater than 10 mV, which is considered a high zeta potential value, the agglomeration of NFs is avoided, and they are capable of being dispersed with other particles (such as MB dye) to react. As agglomeration is neglected, this states that the NF has a higher active site due to a higher total surface area exposed or in contact with the MB dye. As such, a higher percentage of dye removal would then be expected.

Despite the positive zeta potential value of Fe₂O₃ NF, it might show a negative outcome towards the methylene blue dye degradation. Some studies support that this type of NF is still possible to achieve a positive outcome in treating cationic dye⁴². According to Balu et al. (2018), – Fe₂O₃ nanocomposites are capable of degrading 80% of MB dye⁴². Besides, the medium tested is at pH 7; at different pH levels, the zeta potential values might vary, and the charge might be changed. According to Meng et al.⁴⁰, the zeta potential range for Fe₂O₃ is between 32.5 mV – -19.4 mV within different pH ranges. Nevertheless, the results show the positive outcome of Fe₂O₃ NF to degrade MB dye. Notably, ZnO should exhibit a greater tendency towards dye removal, followed by MnO₂ and Fe₂O₃ at pH 7. However, the performance of dye degradation might alter according to various pH values.

Fourier Transform Infrared Spectroscopy (FTIR)

Fig. 8 illustrates the FTIR spectra of ZnO, Fe₂O₃, and MnO₂ NFs through the ultrasonication-assisted co-precipitation method compared with the commercial ZnO, Fe₂O₃, and MnO₂ powders, respectively. The chemical composition, functional group, and purity could be studied⁴³.

Fig. 8.

FTIR pattern for (a) ZnO, (b) Fe2O3, and (c) MnO2 NF.

Multiple dominant absorption peaks of ZnO NF were noticed at 3447, 1632, 1100, 903, and 725 cm^− 1. The reading of a high peak at 725 cm^− 1 further proved the existence of a Zn-O bond, consistent with reported literature values in the range of 524–610 cm^− 143. Another significant peak at 3447 cm^− 1 was attributed to the absorption of water⁴³. Due to the fact that ZnO particles consist of vibrational peaks attributed to the interatomic vibrations in the fingerprint region, where the wavelength value is less than 1000 cm^− 1. The peak at 3447 cm^− 1 was ascribed to the stretching vibration of the surface of -OH of the ZnO⁴⁴. The adsorption of water from the air on the ZnO NF causes the formation of hydroxyl bonds. The formation of the hydroxyl group is due to the adsorption of water molecules on the surface of metal oxide materials during the synthesis of NFs at ambient temperature, which eventually dissociates to generate the hydroxyl group. The existence of this hydroxyl bond on the ZnO NF surface tends to act as the active site, particularly in aqueous solutions such as methylene blue dye solution, which is beneficial as it enhances reaction activity⁴⁴. As such, the rate and efficiency of dye degradation were greatly improved. Since the peak intensity of ZnO NF is high, this suggests that the ultrasonication-assisted co-precipitation method aids in strengthening the Zn-O bond.

For Fe₂O₃ NF, a large broad band observed at 3228 cm^− 1, aligning with previously reported at 3398 cm^− 1 was due to the O-H stretching vibration in the hydroxyl groups⁴⁵. Whereas the absorption peaks at about 1417 cm-1 and 1641 cm-1 were ascribed to the symmetric and asymmetric bending vibration of C = O. Another significant peak at about 1375 cm^− 1 resembled the PVP group, in agreement with the literature-reported at 1293 cm^− 145. A strong band beyond 600 cm^− 1 resembled the Fe-O stretching mode corresponding to the literature value of Fe-O stretching mode at 576 cm^− 145.

For MnO₂ NF, two significant peaks were observed at around 1100 and 1666 cm^− 1, which resemble C-C and C-N bonds of pyridine, respectively, in agreement with the literature-reported at 1389 and 1590 cm^− 1)⁴⁶. Meanwhile, the slight peaks at 1013 and 829 cm^− 1 proved the presence of C-N and Mn-N groups in the MnO₂ NF ⁴⁶. 2900 cm^− 1 showing the signs of C-H bond. Notably, the Mn-O bond was observed at the band of 600 cm^− 1, which is close to the literature value of 568 cm^− 1, since the peak was not fully shown in the graph due to the limited range, hence it was expected that the peak would fall in around 568 cm^− 146. A weak peak at 3455 cm^− 1 was attributed to the C-OH stretching as well as the stretching of vibrations of C = O bond at 1675 cm^− 146. These two peaks, allocated at 3455 and 600 cm^− 1 claim the existence of O-H bond, which reveals the presence of physisorbed water molecules that then attach to the Mn-O bond and MnO₂ NF⁴⁶.

Analysis of Variance (ANOVA) for Zinc Oxide NF

Table 4 shows the full experimental results obtained using the suggested parameter sets for 20 runs. The highest percentage for dark removal (adsorption) was 69.73%, whereas the lowest percentage was 13.65%. For light removal (photocatalysis), the highest percentage for light removal was 89.13%, whereas the lowest percentage was 38.65%. The high adsorption percentage proves that ZnO was a remarkable adsorbent, whereas the higher photocatalytic percentage indicates that Fe₂O₃ was a good photocatalyst. In short, the zinc oxide NF was a good absorbent and photocatalyst that was capable of performing the degradation of dye without the need to introduce absorbents or doping-enhanced processes. The differences between the predicted and actual results were within the acceptable range, which could be seen by the high R-squared value in the following section. The actual equations generated to predict the removal efficiency (%) under dark and light conditions are shown in Eqs. (8) and (9) below.

8

9

Table 4.

Experimental versus predicted results for each suggested run for ZnO NF.

Std	Run	Conc of dye (mg/L)	Conc of metal ions (mg/L)	pH	Experimental response 1: dark removal (%)	Experimental response 2: light removal (%)	Predict dark removal (%)	PredictLight removal (%)
9	1	5	60	9	63.13	89.13	57.94	92.69
4	2	20	80	7.5	20.93	40.93	28.20	45.29
2	3	20	40	7.5	13.65	38.65	14.97	37.17
5	4	10	40	10.5	53.92	83.92	51.81	78.02
11	5	15	20	9	22.07	52.07	23.97	55.96
12	6	15	100	9	48.41	65.41	41.30	63.06
7	7	10	80	10.5	52.07	77.07	55.91	77.01
17	8	15	60	9	48.42	58.42	46.71	58.68
14	9	15	60	12	69.73	85.27	73.15	87.67
20	10	15	60	9	43.42	58.42	46.71	58.68
10	11	25	60	9	20.93	50.93	20.91	48.91
8	12	20	80	10.5	54.55	69.55	54.33	69.56
19	13	15	60	9	48.42	58.42	46.71	58.68
1	14	10	40	7.5	45.04	75.04	50.42	73.49
18	15	15	60	9	48.42	58.42	46.71	58.68
16	16	15	60	9	48.42	58.42	46.71	58.68
6	17	20	40	10.5	46.86	66.86	43.63	66.47
13	18	15	60	6	54.27	59.73	45.64	58.87
3	19	10	80	7.5	48.66	78.66	57.05	77.51
15	20	15	60	9	48.42	58.42	46.71	58.68

Where A (mg/L) refers to the concentration of dye, B (mg/L) refers to the concentration of metal ions, whereas C refers to the pH values, which are unitless.

Adsorption (dark removal)

The R-squared value was 0.9060, whereas the adjusted R-squared value was 0.8213. The R-squared value was greater than 0.9, revealing the high reliability of the output of the experiment. Table 5 tabulated the analysis of variance (ANOVA) for the quadratic model. P-values less than 0.0500 indicate model terms were significant. In this case, A, B, C, AC, B², and C² were significant model terms. This judgement made sense, as all parameters, especially pH, played a crucial role in wastewater treatment, depending on the dye concentration. This validated the significance of model terms in the system. Values greater than 0.1000 indicated the model terms were not significant.

Table 5.

ANOVA for the quadratic model for dark removal of ZnO NF.

Coefficient terms	Sum of squares	Mean square	F - value	P - value
Model	3627.26	403.03	10.70	0.0005
A – concentration of dye	1370.85	1370.85	36.41	0.0001
B – concentration of metal ions	301.20	301.20	8.00	0.0179
C – pH	756.80	756.80	20.10	0.0012
AB	21.78	21.78	0.5785	0.4645
AC	371.83	371.83	9.88	0.0105
BC	3.20	3.20	0.0850	0.7766
A2	83.41	83.41	2.22	0.1675
B2	311.33	311.33	8.27	0.0165
C2	252.84	252.84	6.72	0.0269
R-squared value	0.9060
Adjusted R-squared value	0.8213

Photocatalysis (light removal)

The R-squared value was 0.9720, whereas the adjusted R-squared value was 0.9469. The R-squared value is greater than 0.9, revealing the high reliability of the output of the experiment. Since the predicted R-squared value is 0.7725, which differs by < 0.2 from the adjusted R-squared value, this was reasonably accepted and supported the high accuracy of the predictions. Table 6 tabulated the analysis of variance (ANOVA) for the quadratic model. P-values less than 0.0500 indicate model terms were significant. In this case, A, C, A², AC, and C² were significant model terms. This judgement made sense, as pH plays a crucial role in wastewater treatment, depending on the dye concentration. This validated the significance of model terms in the system. Meanwhile, the lack of fit was insignificant, indicating the model was the best fit for the predictions deduced.

Table 6.

ANOVA for the quadratic model for light removal of ZnO NF.

Coefficient terms	Sum of squares	Mean square	F - value	P - value
Model	3617.59	401.95	38.62	< 0.0001
A – concentration of dye	1916.25	1916.25	184.10	< 0.0001
B – concentration of metal ions	50.48	50.48	4.85	0.0522
C – pH	829.44	829.44	79.69	< 0.0001
AB	8.40	8.40	0.8075	0.3900
AC	306.78	306.78	29.47	0.0003
BC	12.65	12.65	1.22	0.2961
A2	230.97	230.97	22.19	0.0008
B2	1.09	1.09	0.1049	0.7527
C2	334.67	334.67	32.15	0.0002
R-squared value	0.9720
Adjusted R-squared value	0.9469

Analysis of variance (ANOVA) for iron (III) oxide NF

Table 7 shows the full experimental results obtained using the suggested parameter sets over 20 runs. The highest percentage for dark removal (adsorption) was 77.84%, whereas the lowest percentage was 1.66%. For light removal (photocatalysis), the highest percentage for light removal was 89.20%, whereas the lowest percentage was 10.12%. The moderate percentage in the adsorption process proves that the Fe₂O₃ might be an ordinary absorbent, but the greater percentage in the photocatalysis process illustrates that the Fe₂O₃ was a good photocatalyst. In short, the Fe₂O₃ NF was a good absorbent and photocatalyst that was capable of performing the degradation of dye without the need to introduce absorbents or doping-enhanced processes. The differences between the predicted and actual results were within the acceptable range, which could be seen by the high R-squared value in the following section. The actual equations generated to predict the removal efficiency (%) under dark and light conditions are shown in Eqs. (10) and (11) below.

10

11

Table 7.

Experimental versus predicted results for each suggested run for Fe₂O₃ NF.

Std	Run	Conc of dye (mg/L)	Conc of metal ions (mg/L)	pH	Experimental response 1: dark removal (%)	Experimental response 2: light removal (%)	Predict dark removal (%)	PredictLight removal (%)
15	1	15	60	9	1.66	10.12	4.40	12.96
16	2	15	60	9	1.66	10.12	4.40	12.96
6	3	20	40	10.5	66.80	71.80	69.12	68.01
3	4	10	80	7.5	62.25	64.89	43.48	51.61
17	5	15	60	9	1.66	10.12	4.40	12.96
2	6	20	40	7.5	34.50	38.34	29.69	32.54
1	7	10	40	7.5	37.64	69.83	38.96	62.72
12	8	15	100	9	26.13	39.92	44.17	50.15
8	9	20	80	10.5	67.50	70.18	49.73	60.21
4	10	20	80	7.5	31.27	32.27	26.90	29.23
5	11	10	40	10.5	56.80	80.96	44.71	66.92
20	12	15	60	9	1.66	10.12	4.40	12.96
18	13	15	60	9	1.66	10.12	4.40	12.96
14	14	15	60	12	77.84	89.20	89.20	100.19
7	15	10	80	10.5	44.29	62.59	32.64	51.32
9	16	5	60	9	20.42	52.44	32.78	66.75
10	17	25	60	9	36.53	42.70	40.61	45.46
13	18	15	60	6	55.53	58.94	60.61	65.01
11	19	15	20	9	60.64	62.23	59.04	69.06
19	20	15	60	9	1.66	10.12	4.40	12.96

Where A (mg/L) refers to the concentration of dye, B (mg/L) refers to the concentration of metal ions, whereas C refers to the pH values, which are unitless.

Adsorption (dark removal)

The R-squared value was 0.8720, whereas the adjusted R-squared value was 0.7569. The R-squared value was approximately or close to 0.9, revealing the robust and reliable output of the experiment. The reason for the slightly lower R-squared value compared to the rest might be due to the noise. Table 8 tabulates the analysis of variance (ANOVA) for the quadratic model. P-values less than 0.0500 indicated model terms were significant. In this case, A², B², and C² were significant model terms. This judgement made sense as pH plays a crucial role depending on the concentration of dye in wastewater treatment. This validated the significance of model terms in the system. Values greater than 0.1000 indicated the model terms were not significant. If there were many insignificant model terms (not considered those required to support hierarchy), model reduction might improve the model. Meanwhile, the lack of fit was insignificant, which showed the model was the best fit to support any predictions deduced.

Table 8.

ANOVA for the quadratic model for dark removal of Fe₂O₃ NF.

Coefficient terms	Sum of squares	Mean square	F - value	P - value
Model	11565.74	1285.08	7.57	0.0020
A – concentration of dye	61.27	61.27	0.3611	0.5613
B – concentration of metal ions	220.89	220.89	1.30	0.2805
C – pH	817.25	817.25	4.82	0.0529
AB	26.75	26.75	0.1577	0.6997
AC	566.67	566.67	3.34	0.0976
BC	137.70	137.70	0.8114	0.3889
A2	1639.32	1639.32	9.66	0.0111
B2	3502.17	3502.17	20.64	0.0011
C2	7812.31	7812.31	46.04	< 0.0001
R-squared value	0.8720
Adjusted R-squared value	0.7569

Photocatalysis (light removal)

The R-squared value was 0.9088, whereas the adjusted R-squared value was 0.8266. An R-squared value greater than 0.9 reveals the high reliability of the output of the experiment. Table 9 tabulated the analysis of variance (ANOVA) for the quadratic model. P-values less than 0.0500 indicated model terms were significant. In this case, C, A², B², and C² were significant model terms. This judgement made sense as pH played a crucial role depending on the concentration of dye in wastewater treatment. This validated the significance of model terms in the system. Values greater than 0.1000 indicated the model terms were not significant. If there were many insignificant model terms (not considered those required to support hierarchy), model reduction might improve the model. Meanwhile, the lack of fit was insignificant, which showed the model was the best fit to support any predictions deduced.

Table 9.

ANOVA for quadratic model for light removal of Fe₂O₃ NF.

Coefficient terms	Sum of squares	Mean square	F - value	P - value
Model	12726.13	1414.01	11.07	0.0004
A – concentration of dye	453.26	453.26	3.55	0.0890
B – concentration of metal ions	357.40	357.40	2.80	0.1254
C – pH	1237.63	1237.63	9.69	0.0110
AB	30.50	30.50	0.2387	0.6357
AC	488.91	488.91	3.83	0.0789
BC	10.08	10.08	0.0789	0.7845
A2	2924.58	2924.58	22.89	0.0007
B2	3419.11	3419.11	26.76	0.0004
C2	7621.10	7621.10	59.64	< 0.0001
R-squared value	0.9088
Adjusted R-squared value	0.8266

Analysis of variance (ANOVA) of manganese (IV) oxide NF

Table 10 shows the full experimental results conducted based on the suggested sets of parameters for 20 runs. The highest percentage for dark removal (adsorption) was 81.46%, whereas the lowest percentage was 22.84%. For light removal (photocatalysis), the highest percentage for light removal was 93.48%, whereas the lowest percentage was 45.70%. The high percentage in the adsorption process proved that the MnO₂ was a great absorbent, whereas the greater percentage in the photocatalysis process illustrated that the MnO₂ was a good photocatalyst. In short, the manganese (IV) oxide NF was a good absorbent and photocatalyst that was capable of performing the degradation of dye without the need to introduce absorbents or doping-enhanced processes. The differences between the predicted and actual results were within the acceptable range, which could be seen by the high R-squared value in the following section.

Table 10.

Experimental versus predicted results for each suggested run for MnO₂ NF.

Std	Run	Conc of dye (mg/L)	Conc of metal ions (mg/L)	pH	Experimental response 1: dark removal (%)	Experimental response 2: light removal (%)	Predict dark removal (%)	PredictLight removal (%)
19	1	15	60	9	57.97	63.42	56.96	64.29
7	2	10	80	10.5	73.59	82.00	69.22	81.45
20	3	15	60	9	50.76	65.47	56.96	64.29
15	4	15	60	9	57.97	63.42	56.96	64.29
2	5	20	40	7.5	22.84	45.70	26.04	43.09
13	6	15	60	6	48.94	64.79	44.34	65.45
8	7	20	80	10.5	59.50	69.07	57.03	68.52
1	8	10	40	7.5	56.60	78.86	57.91	76.25
9	9	5	60	9	80.42	93.48	83.12	97.81
5	10	10	40	10.5	81.46	93.07	77.37	86.79
3	11	10	80	7.5	68.81	81.67	69.40	79.30
18	12	15	60	9	57.97	63.42	56.96	64.29
6	13	20	40	10.5	63.25	71.16	61.50	70.37
4	14	20	80	7.5	38.29	46.51	41.21	49.63
17	15	15	60	9	57.97	63.42	56.96	64.29
10	16	25	60	9	40.60	52.88	39.06	51.72
16	17	15	60	9	57.97	63.42	56.96	64.29
12	18	15	100	9	52.24	60.47	53.32	59.08
11	19	15	20	9	46.22	53.32	46.30	57.88
14	20	15	60	12	73.87	92.37	79.62	94.88

The equation in terms of actual factors could be applied to perform complete predictions regarding the response for specified levels of each factor. Here, the levels should be designated in the original units for every single factor, respectively. This equation was prohibited from use during the ascertaining of the relative impact of each factor because the coefficients were scaled to accommodate the units of every factor, and the intercept was not located at the centre of the design space. The actual equations generated to predict the removal efficiency (%) under dark and light conditions are shown in Eqs. (12) and (13) below.

12

13

Where A (mg/L) refers to the concentration of dye, B (mg/L) refers to the concentration of metal ions, whereas C refers to the pH values, which are unitless.

Adsorption (dark removal)

The R-squared value was 0.9555, whereas the adjusted R-squared value was 0.9155. An R-squared value greater than 0.9 reveals the high reliability of the output of the experiment. Since the predicted R-squared value was 0.7047, which has a difference of < 0.2 from the adjusted R-squared value, the result was reasonably accepted and proved the high accuracy of the prediction results. Table 11 tabulated the analysis of variance (ANOVA) for the quadratic model. P-values less than 0.0500 indicate model terms were significant. In this case, A, C, A², AC, B², and C² were significant model terms. This judgement made sense as pH played a crucial role depending on the concentration of dye in wastewater treatment. Meanwhile, the lack of fit was insignificant, which showed the model was the best fit to support any predictions deduced.

Table 11.

ANOVA for the quadratic model for dark removal of MnO₂ NF.

Coefficient terms	Sum of squares	Mean square	F - value	P - value
Model	3747.93	416.44	23.88	< 0.0001
A – concentration of dye	1940.84	1940.84	111.28	< 0.0001
B – concentration of metal ions	49.28	49.28	2.83	0.1237
C – pH	1244.68	1244.68	71.36	< 0.0001
AB	6.77	6.77	0.3882	0.5472
AC	127.84	127.84	7.33	0.0220
BC	192.86	192.86	11.06	0.0077
A2	26.79	26.79	1.54	0.2435
B2	80.37	80.37	4.61	0.0574
C2	39.66	39.66	2.27	0.1625
R-squared value	0.9555
Adjusted R-squared value	0.9155

Photocatalysis (light removal)

The R-squared value was 0.9686 whereas the adjusted R-squared value was 0.9404. An R-squared value greater than 0.9 reveals the high reliability of the output of the experiment. Since the predicted R-squared value was 0.7533, which has a difference of < 0.2 from the adjusted R-squared value, the result was reasonably accepted and proved the high accuracy of the prediction results. Table 12 tabulated the analysis of variance (ANOVA) for the quadratic model. P-values less than 0.0500 indicated model terms were significant. In this case, A, C, A², AC, B², and C² were significant model terms. This judgement made sense as pH played a crucial role depending on the concentration of dye in wastewater treatment.

Table 12.

ANOVA for the quadratic model for light removal of MnO2 NF.

Coefficient terms	Sum of squares	Mean square	F - value	P - value
Model	3837.57	426.40	34.28	< 0.0001
A – concentration of dye	2124.29	2124.29	170.80	< 0.0001
B – concentration of metal ions	1.42	1.42	0.1139	0.7428
C – pH	866.12	866.12	69.64	< 0.0001
AB	6.09	6.09	0.4896	0.5000
AC	140.11	140.11	11.27	0.0073
BC	35.20	35.20	2.83	0.1234
A2	172.37	172.37	13.86	0.0040
B2	53.08	53.08	4.27	0.0657
C2	395.93	395.93	31.83	0.0002
R-squared value	0.9686
Adjusted R-squared value	0.9404

Interaction between parameters for metal oxide NFs

Figs 9, 10, 11, 12, 13 and 14 depict the 3D plot of the interaction of parameters, which are the concentration of metal ions (ZnO, Fe₂O₃, and MnO2 NF), the concentration of dye, and pH, towards the dark and light removal of methylene blue dye. Meanwhile, the actual versus predicted results were illustrated in a single graph as shown in Figs. 10, 11, 12, 13, 14 and 15. Merely negligible differences were observed between the predicted and actual output, showing a high consistency in the prediction of results. With increasing pH levels and increasing metal content, dye elimination rises. At high concentrations of metal ions associated with greater pH but lower dye concentration values, the greatest dye degradation efficiency could be accomplished. The existence of hydroxyl ions on the surface of ZnO, Fe₂O_3, and MnO₂ NFs may be what increased the binding effectiveness by creating an electrostatic attraction between the dye. The degradation of dye is reportedly high at greater pH (i.e., 9, 10.5, and 12), which correlates to the combined impacts of dye in adsorption and photocatalytic processes.

Fig. 9.

A 3D plot between the interaction of (a) AB (b) AC (c) BC (d) graph of the predicted versus actual outcomes for ZnO NF for dark removal of methylene blue dye.

Fig. 10.

A 3D plot between the interaction of (a) AB (b) AC (c) BC (d) graph of the predicted versus actual outcomes for ZnO NF for light removal of methylene blue dye.

Fig. 11.

A 3D plot between the interaction of (a) AB (b) AC (c) BC (d) graph of the predicted versus actual outcomes for Fe₂O₃ NF for dark removal of methylene blue dye.

Fig. 12.

A 3D plot between the interaction of (a) AB (b) AC (c) BC (d) graph of the predicted versus actual outcomes for Fe₂O₃ NF for light removal of methylene blue dye.

Fig. 13.

A 3D plot between the interaction of (a) AB (b) AC (c) BC (d) graph of the predicted versus actual outcomes for MnO₂ NF for dark removal of methylene blue dye.

Fig. 14.

A 3D plot between the interaction of (a) AB, (b) AC, (c) BC, (d) graph of the predicted versus actual outcomes for MnO₂ NF for light removal of methylene blue dye.

Fig. 15.

Reusability of ZnO, Fe₂O_3, and MnO₂ NFs in dye wastewater treatment for 5 cycles (1 cycle for original test and subsequent 4 cycles for reusability test).

The effect of the concentration of metal ions used to generate the NF on the rate of MB dye degradation has also been studied and discussed. Based on Fig. 14 (a), with the concentration of dye constant in a maximum value of 80 mg/L, the removal of dye merely shows almost similar outcome as the minimum value of 40 mg/L which is between 47 and 53% (dark removal), 71% (light removal) for ZnO NF; approximately 48% (dark removal), 52% (light removal) for MnO₂ NF. Increasing the concentration of metal ions of the photocatalytic material, ZnO, will decrease dye removal efficiency. This is because the ZnO NF will tend to aggregate to form more significant clusters, increasing the particle size as longer particles are generated⁴⁷. This will reduce the exposed surface area, reducing the amount of active sites for photocatalysis⁴⁷. Moreover, the aggregation might be the source of electron-hole recombination centre formation⁴⁷. Consequently, the photocatalysis performance drops, so the dye removal efficiency decreases. However, there are a few studies claiming that the increment of photocatalytic material leads to an increase in radical production as the surface of active sites is greater⁴⁸. Therefore, it can be concluded that at a certain greatest concentration of metal oxide NFs, maximum efficiency of dye removal can be achieved, resembling the saturation point at which the dye can be degraded. This explains the significance of the parameter pH and the concentration of dye instead of the concentration of metal ions.

However, for Fe₂O₃ NF at 20 mg/L of dye and 40 mg/L of metal ions, maximized dark removal efficiency of 65% could be achieved; whereas at 10 mg/L of dye and 40 mg/L of metal ions, maximized dark removal efficiency of > 65% could be achieved. The different trends of iron (III) oxide NF could be explained by the different surface charges (refer to section Zeta potential explanation). Therefore, different kinds of interactions occurred (observed through the significance values) between the parameters.

Overall, the pH parameter plays a crucial role since it manipulates the charge of the dye and NF, which affects the overall reaction. Oksida-Polivinilpirolidon¹⁸ has stated that the point of zero change (PZC) dramatically impacts the performance of the photodegradation of dye. PZC stands for the situation where the surface charge of NF is discovered to be neutral or zero¹⁸. Due to the absence of the electrostatic force, the interaction between the pollutants and NF at PZC is accounted for as minimal¹⁸. At pH 10.5, the optimum dye removal efficiency could be obtained, which illustrates that the value of PZC for ZnO, Fe₂O₃, and MnO₂ NFs have been discovered at pH 9 and pH 12.

As pH and concentration of metal ions rose, dye removal increased⁴⁹. In contrast to metal ion concentration, pH has a more outstanding impact on the removal of dyes. Several factors might influence the process, such as adsorption, precipitation, increased surface area, minuscule particle dimension, and a great amount of metal ion surface active sites. It may be assumed that dye was actively being eliminated⁴⁹. By using a 3D graph and contour diagrams as illustrated in Figs. 10, 11, 12, 13, 14 and 15, the cumulative impact of pH, concentration of dye, and concentration of metal ions, which refers to the adsorbent dosage in that particular NF, on the removal of methylene blue dye has been articulated. The dye degradation is maximised by the greater concentration of metal ions found in the metal oxide NF, and at certain values of pH, but usually, greater pH is associated with plausible outcomes⁴⁹.

Ultrasonication-assisted co-precipitation method in generating metal oxide NFs which probes sonication, was widely used to identify the adsorbent-adsorbate interaction^27,43. With the ultrasonication technique’s association, the solution’s agitation improves effectively due to the high ultrasonic frequencies. This further strengthens the reactivity of adsorbent adsorption sites exposed to the contaminants (adsorbate). Likewise, the adsorbent-adsorbate interaction increases, leading to the ultimate goal of the study, which is to increase the rate of adsorption. Hence, the dye removal efficiency increases.

Optimization of metal oxide NF towards the degradation of dye

Table 13 tabulates the optimization results generated by Design Expert V13 (trial version) for desirability analysis selected for the ZnO, Fe₂O_3, and MnO₂ NF. A total of 80 suggested optimization results have been generated for ZnO, Fe₂O₃, and MnO2 NF, respectively; however, only the first choice, which shows the greatest desirability, was chosen to proceed with the optimization process.

Table 13.

Optimization results generated by Design Expert V13 for ZnO, Fe₂O₃ and MnO₂ NF.

Material	Conc of dye (mg/L)	Conc of metal ions (mg/L)	pH	Dark removal (%)	Light removal (%)	Validation of dark removal (%)	Validation of light removal (%)
ZnO NF	10	64.454	10.5	57.67	77.21	59.11	79.19
Fe2O3 NF	20	40.000	10.5	69.12	68.01	70.71	71.88
MnO2 NF	10	40.000	10.5	77.37	86.79	74.98	87.68

Since the validation of both dark and light removal is within the difference of %, hence the recommended models are reliable and accurate. This also shows that the optimization of metal oxide NF towards dye wastewater treatment is successfully conducted with positive outcomes obtained. The highest removal (response) of methylene blue dye, with desirability close to 0.9, acknowledges the ideal value of parameters, according to the model optimization profile. Notably, MnO₂ NF exhibited the greatest performance in degrading dye, followed by Fe₂O₃ NF and ZnO NF.

Reusability of metal oxide NF

Fig. 15 illustrates the reusability of ZnO, Fe₂O₃, and MnO₂ NF, respectively, up to 4 cycles. The set of parameters was based on the previously obtained optimization results. According to Bashir et al. 19, ZnO NF exhibited promising stability of about four cycles without significant differences in photocatalytic activities. Since reusability is essential for sustainability, metal oxide NF should be capable of regeneration to prevent waste after only one cycle and to save costs. Fig. 15 displays that there is no remarkable loss in the dye removal efficiency observed during the 1st and 2nd recycling, which indicates that metal oxide NF possesses exemplary photocatalytic activity but is capable of being reused and recycled. However, the steeper slope indicating a greater drop of NF performance in the degrading dye is observed in the 3rd and 4th subsequent rounds. ZnO NF showed a drop of 31.38% for the adsorption process and 22.92% for the photocatalysis process after 4 cycles; Fe₂O₃ NF showed a drop of 58.44% for the adsorption process and 34.44% for the photocatalysis process after 4 cycles; MnO₂ NF showed a drop of 46.72% for the adsorption process and 34.62% for the photocatalysis process after 4 cycles. ZnO NF has the most outstanding dye removal efficiency despite 4 rounds of recycling, followed by MnO₂ and Fe₂O₃ NF. This could be a possible cause by the blockage of the active pores during the treatment process, which cannot be regenerated using mild washing and heat treatments.

Comparison of results with similar techniques and literature studies

From the results illustrated above in Table 14, it can be deduced that the results from the hands-on experiment followed a similar trend as illustrated in other literature studies (with similar techniques and procedures). However, slight variations in the data obtained were expected. If compared with the raw NF or nanoparticles (without doping with extra photocatalysts and adsorbents), the ZnO NF achieved a greater dye removal efficiency than the literature studies. Nevertheless, the laboratory work has been fully conducted to enhance the final conclusion on the manipulating variable (concentration of dye, pH, and concentration of metal ions) to the responding parameters, which is the dye removal efficiency (%). Meanwhile, the characterization of NFs from the hands-on experiment was also similar to the literature results, with slight deviation.

Table 14.

Comparison of experimental results with literature studies.

Synthesis techniques	Photocatalyst	Dye removal efficiency (%)	Reaction condition			References
			Conc of metal ions	Conc of dye	Light source
Solvothermal	ZnO / grapheme	100	20 mg/L	30 mg	Visible light	50
Sol – gel	ZnO NPs / rGO	99.5	10 mg/L	10 mg	Hg lamp	51
Solvothermal	ZnO @ GO	98.5	0.5 mM	10 mg	UV lamp	52
Sol – gel	ZnO / Ag	82.6	20 μ m	0.01 g/L	Xenon lamp	53
Ultrasonication assisted hydrothermal	ZnO / CNTs	99	20 mg/L	0.5 g/L	Hg lamp	54
Green synthesis	Sb – ZnO	98.6	50 μ m	0.05 mg/mL	UV lamp	55
Solvothermal	rGO / TiO2 / ZnO	92	0.3 mg/L	0.1 g/L	Xenon lamp	56
Solvothermal	rGO / TiO2	68	0.3 mg/L	0.1 g/L	Xenon lamp	56
Sol – gel	ZnO NPs	63	10 mg/L	10 mg	Hg lamp	51
Sol – gel	ZnO	65.6	20 μ m	0.01 g/L	Xenon lamp	53
Ultrasonication-assisted co-precipitation	ZnO NF	89.13	5 mg/L	60 mg/L	Natural sunlight	Current study
Ultrasonication-assisted co-precipitation	Fe2O3 NF	89.2	15 mg/L	60 mg/L	Natural sunlight	Current study
Ultrasonication-assisted co-precipitation	MnO2 NF	93.48	5 mg/L	60 mg/L	Natural sunlight	Current study

Conclusion and recommendations

This work concludes the synthesis of metal oxide NFs for different oxidation numbers or cation charges (+ 2, + 3, and + 4), which are ZnO, Fe₂O₃, and MnO₂ NF, respectively. For ZnO NF, the median particle size was about 2.6217 μ m whereas the diameter size was around 934.2 nm. For Fe₂O₃ NF, the particle size was about 2.5413 μ m whereas the diameter size was around 558.6 nm. For MnO₂ NF, the particle size was about 4.4537 μ m whereas the diameter size was around 1.5557 μ m. SEM images of the synthesized NFs reveal hierarchical flower-like architectures. XRF (X-Ray Fluorescence) analysis indicates that these NFs are mostly composed of the corresponding metal oxides, i.e., ZnO, Fe₂O₃, and MnO₂ NFs. The MB removal performance of the above three NFs for adsorption and for photocatalytic removal was evaluated under the optimized conditions.

ZnO NF is capable of achieving 60.29% and 89.13% dye removal efficiency for adsorption and photocatalysis processes, respectively. Fe₂O₃ NF is capable of achieving 77.84% and 89.2% dye removal efficiency for adsorption and photocatalysis processes, respectively. MnO₂ NF is capable of achieving 90.42% and 93.48% dye removal efficiency for adsorption and photocatalysis processes, respectively. Notably, MnO₂ NF exhibited the greatest performance in degrading dye, followed by Fe₂O₃ NF and ZnO NF.

While the reusability test for all NFs has been conducted, no remarkable loss in the dye removal efficiency has been observed. ZnO NF showed a drop of 31.38% for the adsorption process and 22.92% for the photocatalysis process after four cycles; Fe₂O₃ NF showed a drop of 58.44% for the adsorption process and 34.44% for the photocatalysis process after four cycles; MnO₂ NF showed a drop of 46.72% for the adsorption process and 34.62% for the photocatalysis process after four cycles. ZnO NF has the most outstanding dye removal efficiency despite 4 rounds of recycling, followed by MnO₂ and Fe₂O₃ NF. Although the material activity decreased to some extent after several reuse cycles, this decrease is assumed to be caused by the blockage of the active sites of the NFs or by the incomplete regeneration of the catalysts. The present study has several limitations. The performance of the photocatalytic materials was evaluated by using only MB as a model dye, and no experiments were carried out using real wastewater. In addition, the present study does not include any detailed experiments, such as XRD, BET, or TOC, and no evaluation of the economic feasibility was performed. These issues should be solved in future studies in order to evaluate the practical applicability of the synthesized NFs.

In conclusion, inorganic metal oxide NF is capable of effectively removing the dye in wastewater due to its high surface area exposed and high selectivity characteristics. As methylene blue dye has reportedly appeared in most of the wastewater globally in high proportions, it has been selectively chosen as the dye to be removed in this study. This further claims that this piece of research work will contribute to resolving the current issues in the community. Meanwhile, the metal oxide NF synthesised from the ultrasonicated-assisted co-precipitation method is capable of generating a large scale of NF in a shorter period with high surface morphology. Precipitation itself might generate low yield, poor regulation of surface morphology, weak magnetic properties, and a small size of the NF; however, with the assistance of the ultrasonication approach, the purity of the NF produced could be greatly improved. Besides, maghemite NF formed with lower structural distortion possesses a greater crystallite size. The new approach is thus able to produce a uniform size of metal oxide NF within a short period. The ultrasonication approach has been further implemented to cover the drawbacks of the traditional co-precipitation method as well as enhance the final results. This makes it worth studying to offset the recent research gap. Since both adsorption and photocatalytic reactions occur in the dye degradation process, this further shows that full degradation of the dye could be achieved, and enhances the dye removal efficiency.

Abbreviations

3D

Three-dimensional

A

Concentration of dye (model factor)

B

Concentration of metal ions (model factor)

C

pH (model factor)

AB

Interaction between concentration of dye and concentration of metal ions

AC

Interaction between concentration of dye and pH

BC

Interaction between concentration of metal ions and pH

ANOVA

Analysis of variance

DRS

Diffuse reflectance spectroscopy

DTA

Differential thermal analysis

FTIR

Fourier transform infrared spectroscopy

HMTA

Hexamethylenetetramine

LED

Light-emitting diode

MB

Methylene blue

NF

Nanoflower

NFs

Nanoflowers

NPs

Nanoparticles

PL

Photoluminescence

PSA

Particle size analyzer

PZC

Point of zero charge

SEM

Scanning electron microscopy

TGA

Thermogravimetric analysis

UV

Ultraviolet

UV-Vis

Ultraviolet-visible spectroscopy

XRF

X-ray fluorescence

Fe₂O₃

Iron(III) oxide

FeCl₃

Iron(III) chloride

HCl

Hydrochloric acid

MnO₂

Manganese(IV) oxide

MnSO₄

Manganese(II) sulphate

NaOH

Sodium hydroxide

TiO₂

Titanium dioxide

ZnCl₂

Zinc chloride

ZnO

Zinc oxide

Author contributions

Sing Ying Lee – Investigation, Formal analysis, Writing - Original Draft. Yie Hua Tan – Conceptualization, Supervision, Writing - Review & Editing. Sie Yon Lau – Supervision, Writing - Review & Editing. Bridgid Lai Fui Chin, Nabisab Mujawar Mubarak , Pui Vun Chai, Rama Rao Karri, and Shengli Niu– Writing - Review & Editing and Validation.

Funding

The authors received no financial support for this work.

Data availability

All data supporting the findings of this study are available within the paper.

Competing interests

The authors declare no competing interests.