Visible light treatment of azo dye-contaminated water by Ni- and Co-doped-ZnO nanoparticles supported on carbon-covered alumina

Eliminating organic pollutants from wastewater is a critical step in protecting the environment. A large percentage, 15%, of all dyes produced worldwide is lost and discharged as textile waste due to the dyeing process (Pierce 1994). The removal of hazardous dyes and pigments has attracted the attention of environmental remediation efforts. Wastewaters from the dyestuff and colouring industries contain significant non-fixed chromophores, particularly azo types. Azo dyes, which contain aromatic moieties linked together by the azo (–N = N–) chromophore, are the most common dye type used in textile processing and allied industries. It has also been established that some azo dyes and their degradation products, such as aromatic amines, are highly carcinogenic (Ganesh et al. 1994; Zollinger 2003). Congo red (CR) is an organic compound that is the sodium salt of 3,3′-([1,1′-biphenyl]-4,4′-dial)bis(4-amino naphthalene-1-sulphonic acid), repeatedly used for a variety of industrial processes that include printing, textile colouring, pharmaceutical, and cosmetics. CR is stable, with a complex aromatic structure. It is notoriously difficult to remove mainly because it is engineered to resist fading (due to its stability to UV attack) and is also resistant to biodegradation. It is an anionic diazo dye with two azo bonds in its molecular structure, and this structural stability makes it highly toxic and resistant to degradation (Liu et al. 2015). Methyl orange (MO) is also an azo dye that has mutagenic properties and is hazardous on skin contact (irritant), eye contact (irritant), ingestion, and on inhalation. MO and its derivatives are carcinogenic and disrupt biological ecosystems. It can induce skin rashes, headaches, nausea, diarrhoea, muscle, and joint pain, irregular heartbeat, seizures, and toxicity to aquatic life (Yu et al. 2013). The discharge of these coloured wastewaters causes significant harm to aquatic life, while severe over-exposure may prove fatal if not treated timeously. The challenges of treating dyeing wastewater have resulted in sustained interest in developing advanced water purification methods. As international environmental standards have become more inflexible, more advanced technological systems for removing organic pollutants, such as dyes, have recently been developed. Many studies have been conducted to develop water dye removal techniques to meet the standards. The methods of removal are classified as:

• Physical methods: membrane filtration (Dickhout et al. 2017), adsorption (Değermenci et al. 2019), flocculation (Tang et al. 2016), and precipitation.

• Biodegradation: biological oxidation (Fu & Viraraghavan 2001; Senthilkumar et al. 2012).

• Chemical methods: photochemical decolourisation (Taka et al. 2017), chlorination (Miranda et al. 2016) and ozonation (Muñoza et al. 2008).

The key disadvantage of the earlier procedures is mainly that they only create a more concentrated pollutant-containing phase. Since they are not destructive but only change the contamination from one phase to another, they cause a different type of pollution and require additional treatments. Recent innovations in wastewater chemical treatment have improved the oxidative degradation of organic compounds dispersed in aqueous media. Among the new oxidation method (advanced oxidation processes-AOP), heterogeneous photocatalysis has appeared as an emerging technology leading to the most organic pollutants' total mineralisations (Fenoll et al. 2011; Ajmal et al. 2014; Anwer et al. 2019). Blake (1994) have provided a nearly-exhaustive list of various families of organic pollutants that photocatalysis can treat. In most cases, dissolved compounds in water are degraded using UV-illuminated titania (Blake 1994). Some recent studies (Nair et al. 2011; Aruna devi et al. 2018; Manjunatha et al. 2020; Reddy et al. 2020) have reported photosensitised degradation of organic dyes induced by visible light, which from an economic perspective, the use of solar/visible light irradiation is more encouraging than artificial UV irradiation sources.

Zinc oxide (ZnO) has been described as a viable substitute for titanium dioxide in water decontamination due to its numerous structural defects, mainly from oxygen vacancies and its ability for higher production of hydroxyl (^.OH) radicals, and a photoactivity that is better by a factor of 2–3 in UV and sunlight irradiation (Udom et al. 2013). ZnO is a metal oxide with a large bandgap energy of 3.37 eV, a vast excitation binding energy (60 meV), n-type conductivity, abundance in nature and environmentally friendly (Dodd et al. 2006). Photocatalysts based on ZnO mainly absorb UV and, to some extent, visible light. It is important to note that an ideal catalyst needs to be stimulated in the visible region for practical application, especially since visible light accounts for 45% of solar radiation energy while UV accounts for only a small portion of the spectrum (Aruna devi et al. 2018; Sarmah et al. 2018). As a result, various modifications to enhance the efficiency of metal oxides are of great interest. For ZnO to be stimulated by visible light, its bandgap has to be narrowed or split into many sub-gaps, which can be accomplished by embedding transition metal ions. Doping of ZnO is an effective method to improve its catalytic activity (Manjunatha et al. 2020). Yarahmade & Sharifnia (2014) employed dye-sensitised ZnO nanoparticles along with phthalocyanines (Pc) co-doped with cobalt (Co) and nickel (Ni) for the photocatalytic conversion of carbon dioxide (CO₂) and methane (CH₄). Notably, the experiments were conducted under visible light conditions (Yarahmadi & Sharifnia 2014).

Similarly, Chai et al. (2021) employed density functional theory to examine the structural, electronic, and optical properties of an optimised ZnO monolayer doped with Co and Ni. The study demonstrates the effectiveness of strain engineering in altering the electronic characteristics of ultrathin nanofilms, thereby improving the viability of these heterojunctions for optoelectronic uses (Chai et al. 2021). In another study, diluted magnetic semiconductors were synthesised with 10% of Fe-, Co-, and Ni-doped-ZnO, and their semiconducting, magnetic, chemical, and optical characteristics were investigated. The results indicate that Co-doped-ZnO has the lowest band gap and average optoelectronic properties (Ghosh et al. 2019). Ni-doped-ZnO has been reported for the steam reforming of ethanol. There was a strong correlation between Ni loading and the selectivity of the reaction (Yang et al. 2006).

Besides the aforementioned, the co-doping of ZnO with Ni/Co nanomaterials to produce ZnO hybrid compounds has been reported to be effective in the photodegradation of dyes and recalcitrant organic compounds (Aruna devi et al. 2018; Reddy et al. 2020; Goktas et al. 2022). Cobalt and nickel have abundant electronic states and are highly compatible with the ZnO matrix for tuning its physical and chemical characteristics (Pascariu et al. 2018; Poornaprakash et al. 2020). However, adequate metal doping (optimal limit) is required to ensure that the metal particles act solely as electron traps, not recombinants, thereby aiding electron–hole separation. These metal-ion-doped-ZnO nanoparticles are visible light active and have a higher photocatalytic activity rate than bare ZnO (Aruna devi et al. 2018; Manjunatha et al. 2020; Reddy et al. 2020). The other method deployed to improve the photocatalytic activity of the ZnO is by combining it with other materials like activated carbon (Loo et al. 2018), graphite (Lonkar et al. 2018), carbon nanotubes (Sapkota et al. 2019) and alumina (Zheng et al. 2008). Although alumina is the most common support material, interest in carbon and carbon-covered alumina (CCA) as catalyst supports has increased in recent years. The benefits of both alumina and carbon are combined in the CCA support. Before catalyst implantation, the alumina is uniformly coated with a thin carbon layer, resulting in a support material with both the beneficial surface properties of carbon and the alumina's textural and mechanical properties (Błachnio et al. 2007; Zheng et al. 2008). The CCA has been used as a support material in hydrotreating catalysis (Maity et al. 2009) and ammonia preparation (Masthan et al. 1991). CCA supports have lately been used as packing material for HPLC (Paek et al. 2010) and have also been synthesised in various forms. Also, Ag nanocatalysts have been immobilised on CCA to screen microorganisms in drinking water (Shashikala et al. 2007). CCA-supported titania (CCA/TiO₂) (Mahlambi et al. 2014) nanoparticles have been used in the degradation of Rhodamine B dye under the irradiation of visible light, and metal-ion-doped-titania on CCA for the photocatalytic degradation of the same dye (Mahlambi et al. 2014). The incorporation of the CCA will aid in reducing electron–hole recombinations, increasing available surface area for catalysis, improved catalyst recovery and reusability, and shifting the optical response of the ZnO semiconductor to the visible light region, thus, increasing the rate of photocatalytic degradation activity (Mahlambi et al. 2014). The photodegradation of CR and MO dyes as major pollutant models under visible-light irradiation is used as a benchmark to test the photodegradation ability of the prepared catalysts. To our knowledge, reported herein are the first examples of effective photocatalysts based on CCA-supported ZnO-doped with earth-abundant late transition metals Co/Ni. The key aims of this report are to (i) prepare hybrid metal (Co, Ni)–doped-ZnO materials, (ii) produce the metal-doped-ZnO–CCA heterostructured composite in which the metal-doped-ZnO is embedded in the CCA, (iii) improve nanocatalyst recovery and reusability in the photodegradation of CR and MO dyes, and (iv) further expand the optical absorbance of the metal-doped-ZnO/CCA nanocatalyst into the visible light region, thereby improving photocatalytic performance.

All reagents were of analytical grade. Zinc acetate, cobalt acetate, nickel acetate, calcium hydride, potassium chloride (99%), and oxalic acid were purchased from Industrial Analytical (South Africa). Capital Research Distributor Co., Ltd (South Africa) supplied absolute ethanol (EtOH), CR, and MO dyes. Sulphuric acid (98%), γ-alumina (calcined at 500 °C for 3 h to remove any organic impurities), disodium ethylene tetraacetic acid (EDTA), n-xylene (dried on calcium hydride overnight before usage), p-benzoquinone (BQ), and toluene 2,4-diisocyanate (TDI, used without purification) were purchased from Merck (Pty) Ltd. Analytical grade n-propanol and isopropyl alcohol (IPA) were purchased from Industrial Analytical (Pty) Ltd. IPA was distilled before usage.

ZnO was synthesised by the sol–gel method. A solution (tagged A) was prepared from 0.01 mol of zinc acetate dissolved in 60 mL of EtOH and stirred at 60 °C for 30 min. Then another solution (tagged B) was made by dissolving 0.02 mol of oxalic acid dihydrate in 80 mL of EtOH and stirring at 50 °C for 30 min. Solution B was added to the warm solution A dropwise and stirred for 1 h. A white sol was obtained, aged, and dried at 80 °C for 24 h. Finally, ZnO was obtained by calcination at 450 °C (Chen et al. 2017).

The metal-ion-doped-ZnO was prepared by modifying the method reported by Chen et al. (2017): A solution (A) was made from 0.01 mol of zinc acetate dissolved in 60 mL of EtOH. Then, 0.02 mol of cobalt acetate or nickel acetate (at various mole percentages) was dissolved in EtOH and added in drops to solution A, stirred at 60 °C for 30 min. A second solution (B) was obtained by dissolving 0.02 mol of oxalic acid dihydrate in 80 mL of EtOH, stirred at 50 °C for 30 min, then dropwise added to warm A and stirred for 1 h. A coloured sol was obtained, aged and dried at 80 °C for 24 h. The metal-doped-ZnO was obtained by calcination at 450 °C (Chen et al. 2017).

CCA supports were prepared by modifying the adsorption–pyrolysis method. A blend of γ-alumina (5 g) and 1% TDI in n-xylene (115 mL) were stirred at ambient temperature for 24 h. The resulting mixture was filtered and washed severally with n-xylene (100 mL). A feathery white precipitate was obtained and dried overnight at 80 °C. The precipitate obtained was then pulverised and placed into a quartz cell. The temperature was gradually increased to 700 °C under a nitrogen flow (30 mL/min⁻¹) and held at the same temperature for 3 h to complete the pyrolysis of TDI (Sharanda et al. 2006).

A colloidal suspension of the metal-ion-doped-ZnO (ZnO-M) in deionised water was prepared and sonicated for 30 min at ambient temperature, followed by the addition of the CCA, sonicated for 1 h and left at ambient temperature for 24 h to dry. The resulting product was dried at 80 °C overnight to yield the ZnO-M/CCA, which was then pulverised and treated at 450 °C for 3 h to afford the metal-doped-ZnO-impregnated CCA support (Sharanda et al. 2006).

Homogeneous powders of various samples were loaded into the sample holder on a glass slide and levelled to the correct height of a multi-purpose X-ray diffractometer D8-Advance (manufactured by Bruker AXS-, Germany) operated in a continuous ϑ-ϑ scan in locked coupled mode with Cu-K_α radiation. The measurements run within a range in 2ϑ defined by the user with a typical step size of 0.034° in 2ϑ (λKα₁ = 1.5406 Å). A position-sensitive detector, Lyn-Eye, is used to record diffraction data at a regular speed of 0.5 s/step, equivalent to an adequate 92 s/step time for a scintillation counter. A Spectrum 100 spectrophotometer (Perkin Elmer, USA) acquired the FTIR spectra at room temperature using an attenuated total reflectance (ATR) sampling accessory. All spectra were acquired in transmission mode in the 400–4,000 cm⁻¹ range, with 32 scans and a resolution of 2.0 cm. To determine the surface area (S_BET), pore volume, and pore size distributions of the as-prepared samples, a Micromeritics TriStar II (USA) surface area and porosity analyser was used. The sample was degassed in N₂ under vacuum for 24 h at 200 °C before the analysis. Typically, the sample was heated to 90 °C at a heating rate of 5 °C min⁻¹ and held at the same temperature for 180 min. The sample was then evacuated at a pressure of 50 mmHg for 30 min. The temperature was then ramped to 180 °C at a heating rate of 10 min⁻¹ and degassed for 24 h under N₂. The microstructure and morphology of the nanocatalysts were studied utilising Carl Zeiss Ultra Plus FEG (Germany) fitted with an energy-dispersive Oxford X-max detector (United Kingdom).

Similarly, for scanning electron microscopy (SEM), with an operating voltage of 30 kV, and transmission electron microscopy (TEM) operating at both standard and high-resolution (HR-TEM) modes with an operating voltage of 200 kV, a JEOL JEM 2100 (JEOL Ltd, Japan) was used. Before the TEM analysis, the powdered samples were dispersed in EtOH and sonicated for 15 min, after which the sample holder (grid) was dipped into the dispersion. The sample holder was removed and dried with a UV lamp before reinsertion into the TEM sample port for imaging. The SEM and TEM equipment were outfitted with an EDX detector, which was used to identify all the prepared samples. The photoluminescence (PL) spectra were collected using a Perkin Elmer LS 55 (USA) spectrofluorimeter with an excitation wavelength of 325 nm and emission slit widths of 10 nm. The PL spectra were acquired in the wavelength range 390–800 nm, and a few mg of the sample powder was placed in a sample holder for the PL studies. UV/Vis reflectance spectra were collected using a Perkin Elmer Lambda 55 (USA) UV–Vis spectrophotometer equipped with an integrating sphere (Labsphere-USA). Bandgap energies were calculated by plotting (F(R)-hʋ)² against hʋ from the corresponding Kubelka–Munk functions (F(R), which are directly proportional to radiation absorption. The electrochemical measurements were carried out with a VersaSTAT 3F electrochemistry workstation (AMETEK Scientific Instruments, USA). A three-electrode system consisting of a reference, counter and working electrodes was employed. The working electrodes were made by modifying glassy carbon electrodes (GCE) with the respective photocatalysts (ZnO and ZnO-M/CCA). The modification typically involved adding ethanol (600 μL) to a mixture containing a 10 mg sample and 3 mg carbon paste (binder) and sonicated for 30 min. The resulting mixture was drop-casted with a micro-pipette onto a clean, glassy carbon electrode and allowed to dry overnight. A platinum wire was used as the counter electrode, and Ag/AgCl was the reference electrode. A 0.5 M of Na₂SO₄ solution was used as the electrolyte. Any dissolved oxygen in the solution that would otherwise interfere with the redox activity of the working electrode during measurements was purged out by pumping nitrogen gas through the system for 5 min. The electrochemical impedance spectroscopy (EIS) measurements were performed using the same three-electrode system (Yang et al. 2021). The thermal stability of the samples was determined with a thermogravimetric analyser with a small furnace (SF) (Mettler Toledo-Germany). The TGA was measured at a heating rate of 10 °C/min from 25 to 800 °C in air.

The diffractogram of ZnO displays characteristic sharp and well-defined peaks at 2θ = 31.77, 34.38, 36.29, 47.59, 56.63, 62.88, 66.37, 67.99, and 69.15, marked by their miller indices [(100), (002), (101), (102), (110), (103), (200), (112) and (201)] referred to the existence of ZnO nanoparticles with hexagonal wurtzite structure (JCPDS no. 00-036-1451) with (a = b = 3.24982 Å, and c = 5.20661 Å) with a preferential orientation of (101) plane with respect to other planes. No other impurity peaks were detected, indicating that the obtained ZnO was phase pure. While the data for both the ZnO–Co/CCA and ZnO–Ni/CCA composites show peaks associated with metal-doped-ZnO as observed (ZnO traced with the red line, blue line for CCA and aqua line for Ni ions). Additionally, the Ni⁺² (200) phase peaks around 44° and Co⁺² at 43° were detected on the diffraction patterns of ZnO–Co/CCA and ZnO–Ni/CCA composites, respectively (designated ‘*’).

The most intense peaks of both ZnO and metal-doped-ZnO corresponding to (100), (002) and (101) phases were broadened and found to reduce in intensity due to the transition metal dopants in ZnO–Co/CCA and ZnO–Ni/CCA. A progressive shift to lower diffraction angles of the peak indexed to the (002) phase is also observed (as the inset in Figure 1(a)) (Estévez-Hernández et al. 2017). The shifting of peak position is ascribed to induced internal strain in the lattice due to the substitution by smaller Co/Ni ions in the ZnO framework [ionic radius difference between Zn⁺² (0.74 Å) and Co⁺² (0.65 Å), Ni⁺² (0.55 Å) ions] (Kalita & Kalita 2017).

The surface area, pore volume, and pore size distribution were chosen to study the effect of carbon loading on the alumina surface, metal-doped-ZnO, and the effect of embedding the metal-doped-ZnO nanomaterials on the CCA (Table 1). The adsorption–desorption isotherms of alumina, CCA, ZnO, ZnO–Co (0.5:0.1)/CCA (0.1:1), and ZnO–Ni (0.5:0.25)/CCA (0.75:1) are depicted in Figure 2(a)–2(e). The alumina, CCA, ZnO–Co/CCA, and ZnO–Ni/CCA composite all have heterogeneous pore structures and adsorption–desorption isotherms classified as type IV according to the Brunauer–Emmett–Teller (BET) classification, indicating that they are mesoporous materials. Adsorbent–adsorptive interactions and interactions between molecules in the condensed state determine the adsorptive nature of mesopores. Mesoporous materials have properties that include pore condensation, a process in which a gas condenses into a liquid-like phase in a pore at a pressure P less than the saturation pressure P₀ of the bulk liquid (Monson 2012; Landers et al. 2013). ZnO revealed bimodal pore size distribution and a type H3 hysteresis loop linked with capillary condensation and multilayer adsorption on the nanocatalyst surface.

The mesoporous structure of ZnO allows pollutants to disperse through its channels and be destroyed on the catalyst surface. All prepared samples feature a large hysteresis loop, with alumina having P/P₀ = 0.4–0.9, indicating that it is mesoporous and has a high adsorption capacity (219 cc/g at P/P₀ = 1). In contrast, the CCA exhibited (170 cc/g at P/P₀ = 1) with similar hysteresis loop. Following impregnation, the formed ZnO–Co/CCA (240 cc/g at P/P₀ = 1) and ZnO–Ni/CCA (270 cc/g at P/P₀ = 1) also have large type H2 hysteresis loops of P/P₀ = 0.5–0.9 with the highest adsorption capacity. This type of isotherm describes the process of nitrogen adsorption on the material's surface. In addition, this type is linked with a sharp desorption branch at relative pressures near the lower end of hysteresis (adsorption–desorption). The highly steep desorption observed in H2 loops can be ascribed to pore-blocking/percolation in a narrow range of pore necks or cavitation-induced evaporation. Furthermore, when compared to pure ZnO, the ZnO–Co/CCA and Zn–Ni/CCA catalysts had higher specific surface area and pore volumes. In comparison, CCA has a surface area of 144.62 m²g⁻¹, similarly observed previously (Souza Macedo et al. 2019; Mendes et al. 2020; Kazakova et al. 2021). ZnO had an active area value of 13.78 m²/g⁻¹. In contrast, ZnO–Co/CCA and ZnO–Ni/CCA have surface areas of 213.93 and 152.37 m²/g⁻¹, respectively, which are wider and better for more active reaction sites and the separation of photogenerated excitons. The observed increase in the active area for ZnO-M/CCA compared to bare ZnO is due to the carbon loading, consistent with previous research (Mahlambi et al. 2014; Lin et al. 2019). As a result, the available active area for dye adsorption and removal increases. According to the BET grouping (Sing et al. 1985), it has channel-like pores with non-uniform pore size distribution within the mesoporous regions (Nishikiori et al. 2012; Thommes et al. 2015).

The SEM image of alumina (Figure 3(a)) shows that it is made up of large particles on a granular surface with a size distribution between 21 and 23 nm. In contrast, the CCA surface seen in Figure 3(b) is smooth and porous due to the homogeneous covering of carbon. The ZnO's SEM image showed spherical and clustered particles due to calcination (Behnajady et al. 2011; Bekele et al. 2021). Figures 3(c) and 3(d) represent the hybrid combination of the Ni and Co ion-doped-ZnO catalyst on the CCA surface. The images of the Co and Ni-doped-ZnO catalysts supported by CCA are similar and show uniform dispersion of the metals in the matrices of the ZnO–Ni/CCA and ZnO–Co/CCA heterostructures.

TEM was used to further investigate the differences in crystal arrangement. The micrograph of the ZnO revealed the typical crystalline hexagonal wurtzite structure (at a higher magnification of 50 nm (Figure 4(a)). The ‘hexagonal’ structure of ZnO is clearly seen (indicated by white hexagonal shape). The ZnO–Co (0.5:0.1)/CCA (0.1:1) and ZnO–Ni (0.5:0.25)/CCA (0.75:1) were selected for the TEM and HR-TEM study. Figures 4(c) and 4(d) show well-separated grains with hexagonal and spherical structures on the surface. Also, Figure 4(b) displays the TEM image of the CCA support, which shows that it is microporous and fluffy due to the carbon covering. The triangular sublattice of complete graphene is well understood to comprise equivalent carbon atoms. However, the sublattice composition of defective graphene may differ, resulting in corresponding changes in the carbon content of a monolayer coating of disordered carbon material, as seen in the TEM image (Figure 4(b)), confirming the XRD analysis (see Supplementary material, Figure S4(b)) (Kazakova et al. 2021).

Since metal ions are incorporated into the ZnO lattice, and the CCA supports are made of an amorphous carbon layer, it is critical to maintaining the crystallinity of the nanoparticles because it is directly responsible for their catalytic activity. The metal-doped-ZnO were uniformly distributed on the CCA supports, a common unique feature of the CCA supports (Lin et al. 2005; Shashikala et al. 2007). This observation may be due to the porosity of CCA, which enabled the even distribution of the embedded catalysts. In addition, the CCA-supported catalyst did not lose crystallinity, which agrees with the PXRD data. The average interlayer spacing was determined as 0.214 nm for ZnO–Co/CCA and 0.199 nm ZnO–Ni/CCA, which agrees well with the 0.255 and 0.180 nm obtained from the XRD data, respectively. In addition, the particle size of ZnO–Co/CCA and ZnO–Ni/CCA range from 10–12 and 10–15 nm, respectively, in close agreement with the PXRD data. The metal-ion-doped-ZnO catalyst appears to be set in the pores of the CCA, signifying that they can be used in catalysis without unravelling from the support.

Additionally, the HR-TEM images of ZnO–Co/CCA and ZnO–Ni/CCA in Figures 5(a) and 5(b) reveal concentric rings indicating polycrystallinity (a blend of amorphous and crystallinity) in the nanocomposite. It comprises many crystallites, and discrete spots that make up the rings. Furthermore, the observed selected area electron diffraction (SAED) pattern differs from pure hexagonal ZnO (Supplementary material Figure S4), indicating the formation of metal-doped-ZnO nanocomposites. Similarly, Figures 5(c) and 5(d) show the determined d-spacing.

Supplementary material (Figure S5) shows the EDX spectra of the two-metal-(Co and Ni)-doped-ZnO supported on the CCA, confirming that the as-prepared sample contains only Zn, O, Ni, Al, C, and Co ions in the samples. In addition, the elemental mapping images in Supplementary material, Figures S5(d)–5(h) show that Al, O, Zn, and C elements are spatially and homogeneously distributed in the maps. These results agree with the PXRD data; no other impurity phases were found.

The ZnO–Co (0.5:0.1)/CCA (0.1:1) and ZnO–Ni (0.5:0.25)/CCA (0.75:1) display peaks at 405 and 412 nm, respectively. This emission peak is reduced in the Co-doped-ZnO/CCA band and intensified in the Ni-doped-ZnO/CCA. Low fluorescence intensity indicates less charge recombination and high separation rates. Perhaps, the increased recombination in the photogenerated excitons in the Ni-doped-ZnO/CCA may be due to overwhelmed separation capability of CCA. In addition, a minimal crystal defect that would have otherwise served as a trap (Salem et al. 2013; Gopchandran 2016) for the electrons to enable capture by the CCA but served as a point source of recombination for charges (Rochkind et al. 2014), hence the observed increased fluorescence intensity in the ZnO–Ni/CCA.

Furthermore, deep-level defect emission signifies the existence of intrinsic defects in the nanostructures. However, two emission peaks were observed in the blue emission region. As observed from Figure 6, the emission spectra peak in the visible region for all the ZnO-modified nanocatalysts is weak and broad. The peak at 422 nm has been linked to transitions from the bottom of the CB to the O_i levels (Gandhi et al. 2014). In contrast, the peak at 486 nm is associated with electronic transitions from the shallow donor (Zn_i) to the shallow acceptor (V_zn) levels. The broad, intense green-to-red band emission for undoped ZnO centred at 528 nm is caused by transitions from deep donor levels to the VB caused by oxygen vacancies (V_o). The emission spectra of Co-doped-ZnO/CCA samples are quenched when compared to undoped ZnO under the same excitation conditions, as observed in Figure 6 (Ashokkumar & Muthukumaran 2015).

The results obtained from the UV–Vis diffuse reflectance were used to estimate the bandgap directly (Figure 7 as insets) (Ravidhas et al. 2015). The bandgap of bare ZnO was determined to be 3.21 eV, which is close to the known literature value of 3.37 eV for pure ZnO nanocatalysts. The bandgap energies were significantly reduced when the metal-ion-doped-ZnO catalysts were supported on the CCA. The bandgap of the free ZnO nanoparticles was reduced to 1.83 eV after supporting the materials onto the CCA. The decrease in the bandgap of the CCA-supported catalysts is much more significant than the 0.05 eV decrease reported for carbon-doped-ZnO (Srinivasan et al. 2019). The presence of carbon, alumina, and metal ions on the ZnO matrix may have contributed to the catalysts' bandgap narrowing, resulting in an enhanced shift of the band edge towards the visible region of the spectrum (Singh et al. 2017).

Investigation of the IR peaks below 1,000 cm⁻¹ is significant because they indicate the presence or absence of Zn–O bonds and their functional groups. As observed in Figure 8, the shift in vibration frequencies of the ZnO–Co and ZnO–Ni nanoparticles between the 400 and 1,000 cm⁻¹ range is caused by the incorporation of nickel/cobalt ions into the ZnO hexagonal wurtzite lattice. A strong absorption band seen in the 400–600 cm⁻¹ range is for the Zn–Ni–O or Zn–Co–O stretching frequencies. Although broad absorption is observed in this range for all the samples, the positioning of the peaks varied depending on the Ni/Co-doping. The type of doping species affects the spectra, and it has been observed that the broadening at the shoulder of the ZnO band at 495 cm⁻¹ is due to the stretching of M–O peaks (M = Co, Ni, or Zn), which is shifted to 542 cm⁻¹ for the ZnO–Ni and 581 cm⁻¹ for the ZnO–Co materials. However, the peak at 495 cm⁻¹ on ZnO and ZnO–Co/CCA spectra was very weak on ZnO–Ni/CCA (Supplementary material, Figure S7). Perhaps, this observation is due to the low quantity of Ni in the composites (Figure 8). The existence of a shift (blue shift) in frequencies with dopant species (Ni/Co ions) may be due to changes in bond strength with the replacement of Zn²⁺ with Co²⁺/Ni²⁺, confirming the incorporation of Co²⁺/Ni²⁺ into the ZnO lattice (Shinde et al. 2014). In addition, a very weak peak at 2,251 cm⁻¹ (‘*’) was observed on the ZnO–Ni/CCA, corresponding to the N = C = O group. Additionally, the absorption band at 2,000–2,400 cm⁻¹ on the CCA and ZnO-M/CCA represents vibrations of oxygen-carrying reactive species (He et al. 2012b). It should be noted that the data obtained by XRD, and optical studies support the obtained results.

The typical alumina Al–O–Al band at 670 cm⁻¹ (Jun-Cheng et al. 2006) was shifted to 821 cm⁻¹ on CCA and then to a lower wavenumber of 803 cm⁻¹ on the ZnO–Ni/CCA and ZnO–Co/CCA spectra. Furthermore, the band at 1,100 cm⁻¹, which is also typical of alumina due to Al–O vibration mode (Xu et al. 2017), was shifted to 1,090 cm⁻¹ on the CCA spectrum and 1,065 cm⁻¹ on both ZnO–Ni/CCA and ZnO–Co/CCA spectra due to C–O–C vibrational. The absence of the characteristic band (Jun-Cheng et al. 2006) associated with alumina between 1,000 and 435 cm⁻¹ on the CCA spectrum confirmed the formation of CCA, as observed in Figure 8. Compared to pure alumina, these peaks shifted to higher wavenumbers (redshift). The shift revealed a unique interfacial interaction between alumina and carbon, stabilising the composite. These findings confirmed the alumina-to-CCA transformation.

Thermogravimetric analysis (TGA) was used to determine the thermal stability and degree of surface modification. The ZnO, ZnO–Co (0.5:0.1)/CCA (0.1:1) and ZnO–Ni (0.5:0.25)/CCA (0.75:1) nanocomposites underwent four, six, and two stages of weight loss, resulting in 96, 91, and 90% loss of weight, respectively. The removal of physically adsorbed water and some organic components from the ZnO, ZnO–Ni/CCA, and ZnO–Co/CCA nanocomposites resulted in weight losses of 1.59% (100–542 °C), 3.79% (100–150 °C), and 6.17% (100–189 °C) as the first step respectively. The second step involved the removal of physisorbed and chemisorbed carbon materials resulting in weight losses of 1.39% (542–714°C), 1.56% (142–244 °C), and 6.17% for ZnO, ZnO–Ni/CCA, and ZnO–Co/CCA, respectively. For the ZnO sample, the third step was more of a continuation of the second step of 0.21% (714–784 °C) weight loss due to physisorbed and chemisorbed carbon materials, and then the final step, which was a rapid 0.35% (from 784 °C) due to complete removal of organic materials. While for the ZnO–Co/CCA, the second step was the final weight loss of about 4% (from 190 °C).

The third step, 0.84% (244–440 °C) to the fourth step, 0.78% (440–581 °C) weight losses for ZnO–Ni/CCA are due to physisorbed and chemisorbed carbon materials. The fifth and final weight losses of 1.73 and 0.35% (581 °C and above) are due to the complete removal of organic materials (Supplementary material, Figure S4).

Charge transport across a photocatalytic semiconductor and charge transfer across the interface of its surface to the adsorbed species play essential roles in semiconductor-photocatalysis. Solid-state EIS is a vital tool for studying the electrical properties of semiconductor materials. All the prepared samples displayed a semi-circle component at a high frequency with a diameter corresponding to the transfer resistance (R_ct–) and a linear component at a high frequency (Begum et al. 2017; Gul et al. 2019; Merlo et al. 2021).

The ZnO–Co/CCA composite displayed the smallest diameter (incomplete circle), indicating that it had the best conductivity (smallest transfer resistance) compared to ZnO or ZnO–Ni/CCA. The most credible reason for this may be that CCA acts as an electron sensitiser due to its surface's variety of reactive groups (Mahlambi et al. 2014; Tonda et al. 2014; Liu et al. 2019).

In the experiment, the suspension of nanocatalyst and dye was shaken in a dark enclosure for 30 min before irradiation by visible light. It is important to note that the ZnO–Co (0.5:0.1)/CCA (0.1:1) and ZnO–Ni (0.5:0.25)/CCA (0.75:1) samples showed higher adsorption of the dyes than pure ZnO.

The data presented in Figures 10(a) and 10(b) illustrate the superior catalytic performance of the modified ZnO catalyst due to larger specific active areas and hence more reactive sites. All the (ZnO-M/CCA) were effective in the removal of the dyes. Also, the CCA must be present in a critical quantity to ensure the optimum light-harvesting capacity of the support. Overly large amounts of the catalyst block the CCA's pores and reduce its performance (Huang et al. 2014). This is due to the metal-doped-ZnO catalyst being locked into the CCA pores (Lin et al. 2019; Solodovnichenko et al. 2019).

The adsorption of the dye molecules to the reactive surface of the catalyst is critical for effective photocatalysis. The results showed photolysis of 5% for the CR and 22% for the MO azo dyes. Adsorption (in the dark) removed 17, 36, and 30% of CR over ZnO, ZnO–Co/CCA and ZnO–Ni/CCA, respectively. Similarly, adsorption without photolysis removed 42, 38, and 48% of MO over ZnO, ZnO–Co/CCA, and ZnO–Ni/CCA, respectively. The observed absorptive capacity is in close agreement with the BET data discussed in Section 4.1.2. The amount of dye adsorbed is due to the high specific surface areas of the prepared catalyst (Table 2). In addition, these findings suggest that visible light and a photocatalyst are required for complete pollutant dye destruction to be effective. In addition, the data presented in Figure 10(a) and 10(b) shows that the ZnO–Co/CCA photocatalyst degraded 100% of the CR in 150 min and 97% of the MO dye in 180 min. Similarly, the ZnO–Ni/CCA degraded 85% of CR and 98% of MO. The results confirmed that complete dye removal occurs via a photocatalytic pathway involving the prepared ZnO-M/CCA catalyst.

The excellent photocatalytic performance of the ZnO-M/CCA material is attributable to many factors, including the CCA, which caused structural defects that changed the property of ZnO (reduced bandgap energy), thereby increasing light absorption (Srinivasan et al. 2019). Furthermore, as shown in Figure 6, photogenerated charge carrier recombination is almost completely suppressed in ZnO–Co/CCA, resulting in its high photoactivity. Similarly, despite increased recombination of the photogenerated excitons, the photocatalytic performance of the ZnO–Ni/CCA is far better than pure ZnO. The increased photocatalytic performance could also be attributed to the prepared material's increased specific area of about 213.93 and 152.37 m²g⁻¹ for ZnO–Co/CCA and ZnO–Ni/CCA, respectively, which increased the number of active sites, combined with enhanced visible light absorption. However, it should be noted that while increased surface area is important for photoactivity, other factors, such as the light-gathering ability of the composite (due to reactive groups Sapkota et al. 2019 on CCA) and suppression of the photo-induced charge carrier due to metal doping are also crucial for photocatalytic performance (Nenavathu et al. 2018; Praveen et al. 2018). Table 2 compares the photocatalytic activities of the as-prepared nanocomposites to related works with a comparable support material reported in the literature.

The gradient of each plot and the rate constants k for the removal of CR by ZnO, ZnO–Ni/CCA, and ZnO–Co –Co/CCA catalysts are 0.0014, 0.0088, and 0.027 min⁻¹, respectively. While the rate constants for removing MO by ZnO, ZnO–Ni/CCA and ZnO–Co/CCA catalysts are 0.0029, 0.0169, and 0.0022 min⁻¹, respectively. These results are consistent with data reported for the ZnO catalyst (Wu et al. 2019).

The rate constants for the photolysis of CR and MO are very low, with values of 0.0002 and 0.0019 min⁻¹, respectively. ZnO–Co/CCA, photodegradation of CR and MO, is 20 and seven times faster than with bare ZnO. Also, the photolysis of CR and MO by degradation using ZnO–Co/CCA are 141 and 12 times faster. These remarkable results are attributed to the lowest bandgap exhibited by ZnO–Co/CCA, as shown in Figure 7(b). Perhaps, the efficient suppression of the recombination of the photo-induced charge carrier due to the metal doping, coupled with the larger active area, are responsible for the observed catalytic properties. Accordingly, embedding the metal-doped-ZnO onto CCA causes a fundamental alteration of the material's electronic band structure, resulting in improved performance. More importantly, the carbon coating due to the CCA allowed for cross-plane movement (diffusion channels), which improved charge and mass transfer and increased photocatalytic efficiency (Li et al. 2020; Lu et al. 2020). The kinetics data of the photocatalytic degradation of the model dyes are displayed in Table 3.

The main factors responsible for the catalysts' improved removal performance are enhanced light absorption, the pollutant adsorption capacity of the nanocomposite, and suppression of recombination exciton through efficient separation. This prepared catalyst met these requirements. In general, UV is the only way to stimulate ZnO. However, some factors provide visible light absorbance (Baruah et al. 2010; Bouarroudj et al. 2021; Senasu et al. 2021).

The exciton electron–hole (e⁻/h⁺) pair produced during photocatalysis is prone to irradiative recombination, which must be hindered to make available the electron for further dye degradation reactions. The photo-induced charged pair recombination is successfully hindered if they are separated efficiently for long periods. In the illustrations shown in Figure 13, when the catalyst system was exposed to visible light irradiation, the electron in the VB of the main semiconductor ZnO is agitated and stimulated to the CB, where it is quickly captured by CCA, while the h⁺ fragment remains in the VB. For the Co/Ni-doped-ZnO/CCA, the Co/Ni dopants trap the photoexcited electrons from the ZnO CB before capturing by CCA, while h⁺ remained in the VB, limiting electron–hole recombination in these catalysts.

Therefore, the superior photocatalytic performance of ZnO-M/CCA may be attributed to the dual ease of generation of photo-charges (due to the low bandgap energy) and reduced exciton recombination.

Many variables were monitored during the wastewater cleaning process, including pH, colour, COD, and total dissolved solids (TDS). The effect of pollution on these parameters reveals the level of contamination. Physicochemical analyses are necessary during the photocatalytic removal of pollutant dyes. Inspection of the untreated CR and MO dye solutions revealed characteristic red/pinkish and yellow colours. When the MO and CR dyes were treated with the ZnO–Co (0.5:0.1)/CCA (0.1:1) and ZnO–Ni (0.5:0.25)/CCA (0.75:1) catalyst, their colour changed to light pink and clear solution, respectively. The extent of COD reductions and other data obtained in this study are presented in Table 4.

After 180 min of visible light exposure with bare ZnO and ZnO–Co/CCA, the COD for MO was reduced by 37% (entry 3) and 80% (entry 4), respectively. For CR, the COD reduction for bare ZnO and ZnO–Co/CCA was 26% (to 39 mg) and 65% (to 19 mg/L), respectively. This could be because intermediate products are more difficult to degrade and oxidise more slowly than the dyes (Ince et al. 1997; Bilinska et al. 2015; Nagarajan & Venkatanarasimhan 2019). The results confirmed the ZnO–Co/CCA system's higher mineralisation potential for dye removal from the wastewater. As evidenced by an increase in the TDS value 229.7(363) [MO (CR)-entry 3], the presence of ZnO nanoparticles may be responsible for the whitish appearance of the cleaned water. TDS analysis revealed an increase in the number of solid particles in aqueous solutions, which measures how much nanocatalyst was retained. When ZnO–Co/CCA nanocatalysts were used, TDS increased by 21% for MO and 11% for CR, compared to bare ZnO, which increased by 169 and 151%, respectively. This demonstrated the catalyst support's effectiveness in reducing catalyst leaching into the treated water. Based on the TDS and COD study, it is clear that the fabricated catalysts were very effective in decolourising and decontaminating the wastewater. The COD results closely match those of the photodegradation analysis. Thus, indicating that the pollutant dyes were effectively mineralised into harmless water.

After each run, the catalyst material was retrieved by centrifugation and filtered, rinsed severally with deionised water, dried overnight at 80 °C, and ready for the next cycle. The test showed little or no loss in catalyst performance by the fifth cycle. There was only a one-point decrease in dye removal efficiency from 100% in the first cycle to 99% in the fifth cycle towards CR removal. Similarly, the photocatalytic performance of MO decreased from 98 to 96% by the fifth cycle [Figure 14(a) and 14(b)]. Thus, attesting to the catalyst's high reusability and stability.

Metal-ion-doped-ZnO nanocatalysts with high visible light photocatalytic activity were successfully implanted on CCA supports. These were applied to the photodegradation of CR and MO dyes, where all the materials displayed high photocatalytic activities. The CCA-supported version maintained its crystallinity and catalytic effectiveness, directly correlated to its better performance than the unsupported bare metal-ion-doped-ZnO and pure ZnO variants. This effectiveness may be connected to enhanced surface area and pore volume due to the support. The presence of CCA helps to reduce the recombination of excitons and increase the active site, thus improving the overall photocatalytic performance of the composites. Based on the obtained results, it is concluded that, for the first time, highly active metal-ion-doped-ZnO catalysts have been synthesised and incorporated onto CCA supports for the visible light photodegradation of azo dyes. This is a significant step forward in using visible light for wastewater remediation through ZnO-driven materials.

We thank ESKOM (TESP programme) and the National Research Foundation of South Africa for financial support.

The work reported in this paper does not require any ethical approval.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.