Sugarcane bagasse ash for fipronil removal

The scientific consensus is that many pesticides, even in trace concentrations (μg L⁻¹ or ng L⁻¹), represent a potential risk to aquatic biota and human health. Their adverse effects on human health can be acute or chronic. Acute effects, at high concentrations, include favoring alterations in biological functions or even death. Chronic effects, at lower concentrations and with long exposure times, manifest as uncontrolled cell growth, leading to malignant tumors, mutagenic effects – with heritable changes in cell genetic material – and/or teratogenic effects, causing non-hereditary congenital deformations.

There is evidence that conventional water treatment involving chemical coagulation, flocculation, sedimentation or flotation, and rapid filtration is not very effective in removing pesticides. This is probably because the coagulants usually employed are ineffective in destabilizing and precipitating these compounds, so that separation in subsequent treatment steps is not possible. Selective adsorption may occur, onto clay particles that confer turbidity to natural waters, with their subsequent removal in sedimentation, flotation or filtration stages. In this scenario, adsorption is a complementary treatment step to remove pesticides from water for human consumption. Powdered activated carbon (PAC) of plant, mineral or animal origin is the most widely used adsorbent for this purpose in developing countries. The application of granular activated carbon, as the last step before disinfection, is more commonly seen in developed countries, to remove the precursors of chlorination byproducts.

However, the high cost of activated carbon often makes its use in water treatment infeasible, especially for small systems. For coconut shell-based activated carbon produced in Brazil, for instance, a dose of around 10 mg L⁻¹ is estimated to cost approximately US$3.2/100 m³-treated, approximately 60–70% of the total cost of the water treatment chemicals.

The number of studies involving natural materials or industrial waste as adsorbents has been increasing recently. Several natural adsorbents – e.g., Jatropha curcas, cassava bark, Brazil nuts, cashew nuts and pinus bark, cattail (Typha angustifolia), natural bushing (Luffa cylindrica fruit), coconut fiber, silk cotton rice husk ash and Moringa oleifera seeds (Gomes et al. 2022) – have demonstrated pesticide adsorption efficiency from water.

Of the pesticides that can cause harm to health, 40 are included in the Brazilian drinking water standards (Regulation 888 2021), 97 in the US standards (USEPA 2019) and 149 in the Australian standards (Araújo 2018). In Brazil, 15 pesticides (including fipronil) are mentioned in the drinking water standards. Fipronil (5-amino-1-[2,6-dichloro-4-(trifluoromethyl) phenyl-4-(trifluoromethylsulfinyl)-1H-pyrazole-3-carbonitrile) is an insecticide in the phenyl pyrazole chemical group. Classified in toxicological class II, it is considered highly toxic and is easily found for sale in various formulations (emulsions, granules and powders).

Fipronil acts specifically on insect central nervous systems, blocking chloride channels regulated by gamma aminobutyric acid receptors, causing paralysis and death (Bae & Kwon 2020). The insecticide is widely used in agriculture, in foliar treatment on cotton and soybean crops, and in soil treatment in potato, sugarcane and corn crops. It is also used in veterinary medicine to control a wide variety of insects, including cockroaches, mosquitoes and fleas (USEPA 2019; Bae & Kwon 2020).

Fipronil is highly toxic to non-target organisms including lizards (LC₅₀ for Acanthodactus dumerili 30 μg g⁻¹ body weight) and poultry (LC₅₀ = 11.3 mg kg⁻¹ quail). However, it has low toxicity to waterfowl (LC₅₀ > 2,150 mg kg⁻¹ mallard) and moderate toxicity to mammals by oral exposure (LC₅₀ = 97 mg kg⁻¹ for rats; LC₅₀ = 91 mg kg⁻¹ for mice) (Tingle et al. 2003). There are few reports of poisoning in humans. In studies of acute exposure in mammals, fipronil showed high toxicity through the gastrointestinal tract. Chronic exposure in rats led to convulsions (sometimes death), decreased body weight, and hematological and biochemical changes – e.g., in cholesterol, calcium, proteins and hormones. Fipronil was carcinogenic in male and female rats, producing benign and malignant thyroid tumors at higher doses (Tingle et al. 2003).

Studies have shown high acute fipronil toxicity in 96 h to several fish species, and induces physiological alterations, affecting larval growth, with a ‘no observed effect concentration’ (NOEC) of 0.0066 mg L⁻¹ and an ‘observed effect concentration’ (OEC) of 0.015. It has also been shown to be toxic in studies with freshwater invertebrates including Daphnia magna, oysters, shrimps and aquatic plants (USEPA 2019; Dallarés et al. 2020).

The concentrations of several pesticides were evaluated in the Brazilian rivers Vacacaí and Vacacaí-Mirim, in Rio Grande do Sul State, in the harvests from 2005/06 to 2007/08. In the Vacacai in 2005/06, at least one pesticide was found in 94% of samples, fipronil, carbofuran and imazethapyr having the highest proportional detection rates. The Vacacaí-Mirim had the highest fipronil concentrations in its samples. Fipronil stood out in the 2006/07 season as the only pesticide detected in all samples. Fipronil concentrations varied from 0.05 to 26.20 μg L⁻¹ in the 2005/06 and 2007/08 harvests (Marchesan et al. 2010).

The occurrence of fungicides and pesticides was evaluated in surface- and groundwaters, and sediment, in Idaho, Maine and Wisconsin in the USA, and fipronil was found in 8% of the groundwater samples at concentrations of about 2.2 ng L⁻¹ (Orlando et al. 2009). Fipronil sulfide was present in 5% of the sediment samples, at a concentration of 1.4 μg kg⁻¹. Previously, fipronil and three degradation products – fipronil desulfinil, fipronil sulfide and fipronil sulfone – in the Mermentau River basin in Louisiana, with surface water concentrations of between about 0.83 and 5.26 μg L⁻¹. The predominant degradation product in the sediments was fipronil sulfide, with concentrations of 0.64–24.80 μg kg⁻¹.

Fipronil's toxicological characteristics and frequency of appearance in water bodies led to its inclusion in the recent Brazilian drinking water standards (Regulation 888 2021) with a maximum permissible concentration of 1.2 μg L⁻¹. Internationally, only the Australian standard exceeds this with a limit of 0.7 μg L⁻¹, although, until 2017, no studies indicated the presence of fipronil in Australian water bodies (Araújo 2018).

Microcontaminant molecules adsorbed onto an empty surface are held by forces from the adsorbent surface. If the attraction is physical, the process is classified as physical adsorption or physisorption. Generally, in this, the attraction arises from van der Waals forces, which are weak so the adsorption is reversible. If chemisorption – i.e., chemical bonding – is involved, however, by electron rearrangement or sharing, a monolayer forms on the adsorbent surface, which is practically irreversible.

Natural materials – Moringa oleifera seeds, jatropha, cassava, rice husks, different types of nuts, vegetable loofah, etc. – have been used to adsorb organochlorine and organophosphorus pesticides and trace metals (Cd, Pb and Cr), sometimes without any prior chemical treatment (Rodrigues et al. 2014). In the wider context, some wastes, industrial byproducts or low-cost synthetic materials – e.g., slag, boiler ash, sugarcane bagasse, red mud, etc. – have been used as adsorbents. Almeida et al. (2020) showed that enrichment of polyurethane foams with petroleum industry catalyst waste (20–80% by weight) increased their adsorption capacity for high concentration trifluralin pesticide (30 mg L⁻¹) in synthetic waters. The highest removal results (95.3%) were achieved using only the catalyst residue as an adsorbent.

Increased global demand for ethanol from renewable sources, associated with large cultivable areas and favorable climatic conditions, have made Brazil the world's largest sugarcane producer. Some 654,839 tonnes of sugarcane were processed for sucrose and ethanol production in 2019/2020 (Conab 2017), making SBFA one of the most abundant industrial wastes in the country. On average, 1 tonne of sugarcane produces 280 kg of bagasse, with about 50% moisture content. Bagasse is usually burned to generate energy for industry, making bioelectricity from sugarcane the fourth source in the Brazilian energy matrix (Canilha et al. 2012).

SBFA, as a by-product of the sugar and alcohol industry, has great potential as an adsorbent in water and wastewater treatment (Sarker et al. 2017; Yadav & Singh 2020). This arises because its main components – alumina, silica, ferric oxide, calcium oxide, magnesium oxide and carbon – and physical properties, such as porosity, particle size distribution, specific surface and alkaline nature (Ahmaruzzaman 2010). Quite extensive research has been done on sugarcane bagasse ash as an adsorbent for removal of metal ions – Zn⁺², Cu⁺² and Cr⁺³, Ni⁺², Cd⁺² and Pb⁺² – drugs, acetylsalicylic acid and diclofenac (Shah et al. 2011; Ferreira et al. 2015; Salomão et al. 2019; Rodríguez-Díaz et al. 2021).

The SBFA was provided by the Minas Gerais Sugar-Energy Industries Association and generated in a plant in the state. The ash was characterized according to its carbon, metal and oxide content, and specific mass. Total organic carbon (TOC) was quantified using an SSM-5000ª module coupled to a Shimadzu TOC-L CSH/CSN instrument. The determination was made by the indirect method, with total carbon measured from the catalytic oxidation at 900 °C to CO₂ of all carbon present. Inorganic carbon was measured after acid-purging the CO₂, and the TOC value obtained as the difference between total and inorganic carbon (Adarme 2018).

Soluble metals and metalloids – i.e., Al, As, Ba, Be, Ca, Cd, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Mo, Na, Ni, Pb, Sn, Sr, Ti, V and Zn – the EPA 3050B methodology was used (USEPA 1996). Initially, a 1.5 g ash sample (moisture content 8.3%) was subjected to partial acid digestion before subsequent analysis by inductively coupled plasma optical emission spectrometry (ICP-OES) (Adarme 2018).

Oxides, including SiO₂, Al₂O₃, Fe₂O₃, TiO₂, MnO, MgO, CaO, K₂O and P₂O₅, were determined in an X-ray fluorescence spectrometer (Philips Panalytical MagiX with a PW2540 autosampler and a 2.4 kW rhodium tube). The spectrometry pellets were constructed from 2.5 g ash samples (undigested solids) with 1.0 g of binder. The calcination loss (%) was calculated gravimetrically after heating the sample to 1,050 °C for 60 min.

Specific mass was determined using the ABNT NBR 6458:2017 procedures (ABNT 2017). The sample density was determined gravimetrically in triplicate, by weighing 125 g of it, placing it immediately in a calibrated, 500 mL pycnometer, and applying a vacuum (88 kPa) for 60 min (Adarme 2018).

Ash surface morphological properties were determined by the scanning electron microscopy (SEM), using a TESCAN VEJA 3 electron microscope coupled to an Oxford microanalysis system. This provided the adsorbent's particle shape and texture.

Particle size analysis was performed with a vibrating sieve and Tyler/Mesh series sieves. The material was weighed with a Shimadzu AUX220 digital balance – resolution 0.0001 g maximum capacity 220 g. Sieving time was 5 min for all tests in a set of five sieves (0.038–0.300 mm, 400–48 mesh), with no openings in the bottom unit. The material retained on each sieve was weighed after sieving, and the tests were performed in triplicate with initial masses of 15.0040, 15.0131 and 15.0072 g. The aim was to identify the material's most representative diameter, so the sieve retaining most material indicated the predominant granulometry and was adopted as the standard granulometry in the adsorption tests.

The point of zero charge (PZC) was determined using three distilled water matrices prepared for pH values 3, 6 and 11, each divided into seven, 20 mL aliquots. pH was adjusted with nitric acid and 0.1 M sodium hydroxide solutions. Different, known amounts of ash were added, yielding seven Erlenmeyer flasks for each pH. The 21 flasks were stirred constantly at 118 rpm and 35 °C to reach equilibrium pH, and the final pH in each flask was measured. Plots of pH versus ash content were used to determine PZC graphically.

The 1.0 g-fipronil·L⁻¹ solution was prepared in methanol and stored at −20 °C to minimize degradation. Fipronil, diluted in methanol, was stirred into the study water in a 2 L flask for 10 min, before 100 mL aliquots (20 μg-fipronil·L⁻¹ in water) were transferred to seven conical flasks, which received fixed masses – 1.0–7.0 mg, in 1 mg steps – of the adsorbent. 100 mL of the study water and 2.0 mg of adsorbent were placed in another flask as a control. The experiments were performed in triplicate using groundwater at pH 7.2.

The Erlenmeyer flasks were shaken at 145 rpm and 25 °C for 60 min, after which, 10 mL samples were taken from all flasks. The samples were filtered under positive pressure using a Millipore 0.22 μm membrane. Finally, 150 μL aliquots were transferred to vials and stored in a freezer prior to HPLC-MS/MS analysis. After the assays, the Langmuir, Tempkin and Freundlich isotherm models were used to fit the fipronil adsorption data.

Determination of the equilibrium condition enables the maximum adsorption capacity under controlled conditions to be inferred. The same stock standard fipronil solution was mixed with the study water, with 10 min contact time. 100 mL aliquots of the 20 μg-fipronil·L⁻¹ solution were transferred to Erlenmeyer flasks and equal (60 mg L⁻¹) adsorbent doses added. 60 mg-adsorbent·L⁻¹ was determined as the optimum for fipronil removal in the adsorption tests, and fixed contact times of between 20 and 120 min were used. A blank test – with no adsorbate – was also run. All other test conditions were the same – i.e., shaking at 145 rpm at 25 °C, filtration under positive pressure on a 0.22 μm membrane, etc.

Adsorption tests were also done using commercial PAC of vegetable origin, under the same conditions defined for the ashes, to compare the adsorbents. A 20 mg-PAC·L⁻¹ dose and the same fipronil standard stock solution were used.

There were three phases in the jar test assays. Phase I corresponded to the application of 60 mg-SBFA·L⁻¹ and 20 mg-PAC·L⁻¹, in three flasks for each adsorbent. The hydraulic parameters in this phase reproduced those in the treatment plant noted above – i.e., rapid mix velocity gradient 800 s⁻¹ (∼400 rpm), agitation time 5 s, flocculation velocity gradient 35 s⁻¹ (∼40 rpm), flocculation time 10 min and settling velocity 6.0 cm min⁻¹ (sedimentation time ∼70 s).

The remaining turbidity was determined in the six samples after settlement before the samples were filtered under positive pressure on a 0.45 μm membrane. Subsequently, 150 μL aliquots were stored in a freezer prior to fipronil analysis by HPLC-MS/MS.

In Phase II, fipronil removal by ash (3 vials) and without adsorbent was compared, using the same dose of aluminum sulfate (100 mg L⁻¹) in all flasks. The methodology was the same as in Phase I, except that flocculation time was 40 min, which was adopted following indication from the tests of maximum fipronil removal at this contact time.

Finally, in Phase III, using the same procedures as in Phase II, the adsorbent was changed to PAC (at 20 mg L⁻¹) and flocculation time to 20 min.

The chemical composition of SBFA is shown in Table 1.

As shown in Table 1, the loss on calcination (LC), which represents the carbon not lost as CO or CO₂ during sugarcane combustion, was approximately 43.7%. This remaining carbon could provide adsorptive capacity, and its amount depends on both the proportion of water in the bagasse and the efficiency of its combustion.

In addition to the high residual carbon content, the predominance of silicon oxides, accompanied by aluminum and iron oxides, is noted. The presence and chemical characteristics of such oxides favor the accumulation of surface hydroxyls, which can contribute to specific interactions with the adsorbate. Thus, hydrogen bond-type intermolecular interactions probably contribute to organic compound removal. For fipronil, hydrogen bonding interaction could arise between the surface hydroxyls and the amino group (NH₂) present in the pesticide.

The fibrous particles are linear and skeletal, with holes and threads, as reported by others (Rodríguez-Díaz et al. 2015; Sarker et al. 2017; Mor et al. 2019). The irregular particles were caused by the melting and solidification of the oxides of silicon and metallic elements during combustion. The large amounts of silicon and oxygen corroborate the high SiO₂ content found in the XRF analysis. In this context and given the ability of surfaces with high silica content to enable the ion exchange of metallic species, it is advantageous for its use as an adsorbent (Shah et al. 2011). Similar morphologies were observed in other sugarcane bagasse ashes (Rodríguez-Díaz et al. 2015; Sarker et al. 2017; Mor et al. 2019; Toledo-Jaldin et al. 2019; Yadav & Singh 2020; Kumari et al. 2021).

In application of SBFA in water treatment plants, where coagulation pH values are generally around 6.2–6.8, the ash surface charge is expected to be predominantly positive, which may favor electrostatic interactions between negatively charged species and the adsorbent. This seems to be related to changes in the SBFA's surface charges, and the ionization of atoms or groups of molecules, facilitating the adsorption mechanisms. Therefore, at low pH, ionization of nitrogen atoms, chlorine and/or NH groups of the fipronil molecule is likely to occur (Figure 5(b)).

The parameters obtained by adjusting the Langmuir and Freundlich models are presented in Table 2.

The isotherm R² values show that the Freundlich model (R² = 0.9290) describes the adsorption data better than the Langmuir model (R² = 0.8870). The isotherm result of this study is in accordance with the literature (Yadav & Singh 2020).

Although the value of ‘n’ is <1 (0.8), SBFA performs better than commercial PAC, especially considering, among other things, the economic and environmental factors.

The Freundlich model indicates that the ash may present a heterogeneous surface with adsorption occurring in multiple layers. A plausible explanation for this is the hydrogen interactions generated between the surface hydroxyls present on the ash and the NH₂ group present in the fipronil molecule.

As can be seen, fipronil adsorption capacity increased rapidly for the first 40 min, achieving 73% efficiency. Subsequently, the rate slowed until saturation was reached. The highest efficiency observed was just below 78% at 80 min contact time. This is slightly higher than that obtained by Yadav and Singh – about 76% in 24 h – in their study (2020). It should be noted that very long contact times – e.g., 24 h – are infeasible in municipal-scale water treatment.

The contact time for fipronil removal acquires relevance in relation to the use of SBFA in conventional water treatment plants. Such plants – equipped with rapid mix, flocculation, sedimentation and filtration units – supply approximately 75% of Brazil's population, a proportion that increases to 94% when considering only surface water sources (Libânio 2016). As adsorbent application, usually PAC, usually occurs in the rapid mix unit, the theoretical contact time, in the plant, is about 2–3 h.

PAC adsorptive capacity was evaluated by comparing the fipronil concentration in the aqueous medium before (control) and after addition. PAC's fipronil removal efficiency was 79% for 20 min contact time, rising to 97% with 120 min contact time. In other words, it exceeded SBFA's fipronil removal efficiency in all situations studied. Nevertheless, while SBFA's performance (60 mg·L⁻¹ dose) has been 17% lower on average than PAC's (20 mg·L⁻¹ dose), it is noted that PAC's cost is much higher than that of SBFA (Varsha et al. 2022). The costs of the chemical products usually applied in water treatment are approximately US$1.6–2.0/100 m³. Due to Brazil's continental dimensions, which significantly increase transport costs, this estimate was based on water treatment plants in the country's Southeast Region, the richest part and home to approximately 44% of the Brazilian population. The rare use of PAC in water treatment plants in Brazil and developing countries is largely due to the high cost. Assuming an average applied dose of 10 mg L⁻¹ and the use of coconut shell activated carbon (US$3.0/kg), the cost would be around US$3.2–3.5/100 m³. Thus, PAC application will increase the cost of treatment chemicals by approximately 200–275%, in addition to personnel and electricity costs (Libânio 2016).

A number of conclusions can be drawn from the results of this study:

• (i)

  SBFA were successfully shown to adsorb fipronil, an insecticide and potential pollutant. SBFA use is quite favorable compared with commercial PAC, especially in the context of economic and environmental feasibility.

• (ii)

  Although SBFA's removal efficiency is lower than that of PAC, it is a viable solution, because, without prior treatment, its removal efficiency was almost 80% with an hour contact time.

• (iii)

  SBFA's low fipronil removal contact time is especially relevant in relation to the material's use in full-scale conventional water treatment plants, where the theoretical contact time is about 2–3 h.

• (iv)

  It is likely that fipronil adsorption is more efficient in low pH systems, since the fipronil molecule presents the amino group, which may favor surface interactions on the adsorbent by giving rise to polar electrostatic forces.

• (v)

  The isotherm R² values showed that the Freundlich model (R² = 0.9290) describes the adsorption data better than the Langmuir model (R² = 0.8870).

• (vi)

  Finally, the method developed is low-cost and efficient, and comprises the reduction, reuse and recycling of sugar and ethanol industry waste, promoting the circular economy and therefore sustainability.

The authors would like to thank Coordenação de Aperfeiçoamento de Pessoal de Pesquisa (Capes) and Universidade Federal de Ouro Preto (UFOP) for their financial support.

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

The authors declare there is no conflict.