Silica extraction from rice hull ash through the sol–gel process under ultrasound

Rice is among the main foods produced in the world and is part of the daily diet of most families. The main waste from rice processing is rice husk (RH), which has been used as biomass for energy generation through combustion. In this process, rice husk ash (RHA) is generated as a residue, and its silica (SiO2) content varies from 85 to 98%. The present work describes the study of the extraction of silica from RHA by the ultrasound-assisted sol–gel method. An experimental design based on the response surface methodology (RSM) with the symmetrical, second-order rotational central composite design (RCCD) was applied to determine the best extraction conditions considering extraction time and molar ratio (n) as variables = nNaOH/nSilica). These optimal conditions were then applied to three ash samples, two obtained by the combustion process in a boiler furnace, with a mobile grate system (RHAC1 and RHAC2), and one obtained by the pyrolysis process (RHAP) carried out in a fixed bed reactor. Results showed that a molar ratio of 4.4, and an extraction time of 107 min were the best extraction conditions, leading to a yield of 73.3% for RHAP, 43.9% for RHAC1, and 31.1% for RHAC2. It was found that the extraction yield and textural properties of the silica obtained depend on the characteristics of the ash used. The silica extracted from RHAC1 presented a surface area of 465 m2.g−1, mesopores of 4.69 nm, purity greater than 95%, and an ultra-fine granulometric distribution, reaching nanoparticle dimensions, characteristics comparable to commercially available silicas.


Introduction
The decrease in the availability of natural resources, coupled with the need to reduce CO 2 emissions, has been stimulating the increase in energy production through renewable resources, such as lignocellulosic biomass, which includes sugarcane bagasse, peach kernel, elephant grass, and rice husk (RH) (Borges et al 2016;Huang and Lo 2018;Mahlia et al. 2019;Pedroso et al. 2018;Silva et al. 2021;Teixeira et al. 2020;Valadão et al. 2021).
Rice husk is an agricultural waste, considered a residue from rice processing in rice industries, found in abundance in all rice producing regions of the world (FAO-AMIS 2021). This waste production is a problem in the Rio Grande do Sul-Brazil, which is responsible for 70.4% of the national rice production, with 8,278 tons in the 2020/2021 harvest (CONAB 2021). The issue is worse in the Pelotas region, which has some of the largest rice processing plants in the state of Rio Grande do Sul (IRGA 2021) and RH constitutes one of the main residues of the region. Local industries seek to reduce the environmental impact through modernization combined with social and environmental responsibility.
Rice husk has a calorific value of approximately 16.25 MJ/kg, which enables its application as biomass for energy generation through thermochemical processes (Adams et al. 2018). When RH is used as biomass for energy generation, through combustion in boiler furnaces, it results in a solid residue called rice husk ash (RHA).
Another alternative to eliminate the excess RH generated in rice processing is the thermochemical conversion of Responsible Editor: George Z. Kyzas biomass by the pyrolysis process, which differs from combustion because it occurs in the absence of oxygen. The process of pyrolysis of RH has been explored through several studies and papers published in recent decades. Among its benefits, the ease of storage and transportation of the liquid product (bio-oil) stands out the most. The bio-oil can be used as fuel or as a source of raw materials for various chemical industries (Betemps et al. 2017;Huang and Lo 2018;Sanches Filho et al. 2018, 2020Téllez et al. 2021;Zhang et al. 2019). The products resulting from pyrolysis can be classified as condensable volatiles (bio-oil), noncondensable volatiles (gas), and solid residue, also known as biochar (Wang et al. 2020;Zhang et al. 2018).
Both biochar and the solid residue from the combustion of RH (RHA) correspond to approximately 20 to 40% of the rice husk mass, depending on the efficiency of the thermal process (Fernandes et al. 2016;Vieira 2018). The physical and chemical characteristics of RHA, especially its low density and high silica content, contribute to its characterization as a residue of difficult degradation, with very few nutrients for soil application, while also being difficult to transport and store. The difficulty in dealing with the large volume of ash generated from combustion leads to most industries paying for its transportation and disposal in licensed landfills (Kieling et al. 2020).
The inorganic composition of biochar and RHA are very similar: both being formed by oxides, most particularly silicon oxide (SiO 2 ), whose composition ranges from 85 to 97% by weight (Fernandes et al. 2016;Flores et al. 2021). Based on these values, the use of this waste for silica extraction is an interesting alternative for obtaining a product with higher added value.
Studies on silica extraction from thermochemical solid wastes show the feasibility of obtaining silica nanoparticles with high surface area and high purity. The applications of nano-silica have been growing over the years, and the research aimed at its extraction and use has been receiving increasing attention. Due to its exceptional physical and chemical properties, nano-silica can be used in various areas, such as cosmetics, electronics, refractories, medicine, and dentistry (Athinarayanan et al. 2015;Patel et al. 2017).
Silica can be present in thermochemical solid wastes in two morphological phases: amorphous and crystalline. The amount of amorphous silica in thermochemical waste depends particularly on three aspects: the minority constituents, the process temperature, and the firing conditions. Studies indicate that at temperatures above 800 °C, the formation of the crystalline phase starts . Amorphous silica is an allotropic form of silicon that has no crystallographic orientation (CETEM/MCTIC 2019), and is more reactive and suitable for extraction by the sol-gel method.
The main methods of silica production from agricultural waste, as in RH, RHA, and biochar, can be divided into two groups: thermal methods and chemical methods. The sol-gel process is a chemical method that stands out for its low energy cost (Patel et al. 2017). Sol-gel processing comprises a series of operations that include chemical reactions and physical processes (dissolution, neutralization, condensation, polymerization, phase transition, phase separation, evaporation, etc.) promoting the formation of porous solids from liquid solutions of molecular precursors (Landau 2006). The replacement of chemical compounds with natural materials in the sol-gel route has been studied to make the process more environmentally friendly (Razavi et al. 2020;Mahdiani et al. 2017). Several scientific investigations have been carried out over the years based on the application of the sol-gel method on RH, RHA, and various waste materials with great success (Ebisike et al. 2020;Irigon et al. 2020;Mourhly et al. 2019). Kalapathy et al. (2001) obtained a 91% yield in silica extraction with 93% purity, starting from RHA. Whereas Santana Costa and Paranhos (2018) started from RH and, in the laboratory with controlled conditions, produced the RHA and obtained the silica with a yield of 99.44%.
The recurrent method in these studies consists of dispersing the thermochemical solid waste (ash) in an alkaline solution of sodium hydroxide, where the dissolution of amorphous silica occurs through the conversion into soluble sodium silicate (Eq. 1). The second step consists of silicic acid formation (Eq. 2), through pH decrease with the help of an acid, which then condenses and polymerizes forming siloxane bonds (Eq. 3) (Mor et al. 2017).
The dissolution of silica (Eq. 1) was performed under stirring and at constant temperature (Fuzinatto et al. 2021;Irigon et al. 2020). Alternatively, the ultrasound-assisted extraction methodology has been present in the literature for different areas (Amigh and Taghian Dinani 2020;Arduim et al. 2019;Carrillo-lopez et al. 2019;Luz 1998;Souza et al. 2022). There are few studies using ultrasound-assisted methods for silica extraction from RHA, and its use is restricted to the steps before extraction, such as for acid washing to remove impurities (Franco et al. 2017;Lee and Oh 2019;Peres 2018). Ultrasound has also been used to functionalize the silica surface using organic functional groups (Kuvayskaya and Vasiliev 2019; Lim et al. 2021). These studies demonstrate the influence of the ultrasound step on the quality of the silica obtained, with an increase in pore volume, pore diameter, density, porosity, and purity of the silica. Carrillo-Lopez et al. (2019) describe that the ultrasound waves apply physical, mechanical, and chemical effects, which induce structural and physicochemical changes and accelerate chemical reactions (Carrillo-Lopez et al. 2019).
The transformations associated with ultrasound are due to the effect of cavitation, which is the formation, growth, and implosive collapse of bubbles in a liquid, caused by sound waves that create alternating regions of compression and expansion that can form bubbles up to 100 microns in diameter. When the micro bubble reaches a critical size, it implodes generating intense local heat and pressure, around 5000 °C and 1000 atm. The implosion of micro bubbles also sends shock waves through the liquid and, when this liquid has dispersed solid particles, drives small dust particles to collide with each other at speeds of over 500 km per hour (Suslick 1995). In heterogeneous solid-liquid surface reactions, unlike cavitation bubble collapse in liquid, the collapse is on, or near, a surface and is asymmetric, because the surface offers resistance to the flow of liquid from that side. The result is an inrush of liquid predominantly from the side of the bubble away from the surface, which results in the formation of a powerful, surface-directed, jet of liquid. It can increase the mass and heat transfer in the surface by disrupting the interfacial boundary layers. In this medium, acoustic cavitation can produce dramatic effects on particulate matter or agglomerates. In our case, the solid surface is not a continuous surface, but a suspended powder, and the cavitation can achieve a particle size reduction and efficient dispersion (Liu et al. 2022;Mason and Pétrier 2004).
Ultrasound has applications in the field of organic chemistry and is very effective in dissolving, increasing reaction rates, and increasing product yield. It has also been used to improve the hydrolysis and separation of lignocellulosic materials for the production of biofuels in biorefineries, be it chemical or physical mechanism enhancement (Ullah et al. 2019). Ultrasound has been employed in material synthesis, such as ceramic powders and catalysts, due to its effect of increasing surface area and decreasing synthesis time, in comparison with conventional methods. Furthermore, ultrasound can change dissolution, nucleation and crystals growth processes in catalysts synthesis (Khoshbin and Karimzadeh 2017;Wang et al. 2008). Khoshbin and Karimzadeh (2017) used the ultrasound in template free synthesis of nanostructured ZSM-5 zeolite from rice husk ash silica and found that the milder conditions offered by ultrasound improved the surface area and led to a decrease in the crystal size and crystallinity of ZSM-5.
Although there are few studies describing the effect of ultrasound on silica extraction from RHA using NaOH, a positive effect on the characteristics of the silica extracted by ultrasound can be expected. The ultrasound cavitation effect will improve the mass transfer between the fluid phase, which contains the NaOH, and the solid phase, which contains rice husk ash (silica). This cavitation will also help expose the silica, facilitating the contact between the leaching agent and the silica. Studies describing the effect of ultrasound in the alkali attack step of silica, through the sol-gel process from rice husk ash, correlating yield, characteristics of the silicas, extraction conditions, and the influence of the ash structure, are practically non-existent. This information constitutes a novelty of great scientific interest.
Considering the above, this work aims to develop a methodology for silica extraction from solid thermochemical waste-from combustion (RHAC) and pyrolysis (RHAP)by using the sol-gel method with ultrasound assistance and to characterize the obtained silica. This process will generate a product with higher added value and reduce the environmental impact of these residues.

Materials and methods
This study was carried out on IF Sul-Campus Pelotas, in the Environmental Contaminants Laboratory, the Fuel Analysis Laboratory 2, and the Chemical Engineering Process Laboratory.
For the development of this study, three samples of thermochemical solid waste were obtained by different processes; two samples were generated by combustion (RHAC 1 ; RHAC 2 ) and one by pyrolysis (RHAP). The samples RHAC 1 and RHAC 2 were obtained from two industries in the region of Pelotas (RS Brazil), both from the combustion process in a boiler furnace, with a mobile grate system. The RHAP was obtained using the organic materials volatilization system (SIVOMO-250), which is a semiindustrial scale apparatus installed at IFSUL-Pelotas campus, with a stainless steel reactor with a fixed bed and using inert atmosphere (nitrogen). In this process, the following conditions were used: 100 g of RH with 2 mm in diameter; a temperature of 600 °C with a residence time of 10 min; heating rate of 25 °C. min −1 ; and nitrogen flow with an average flow of 60 mL.min −1 .
RHAC 1 , RHAC 2 , and RHAP were dried at 105 °C and ground to a particle size smaller than 0.710 µm.
The ultrasound equipment (US) used in the study is from the Unique brand, model USC-4800 A, us frequency of 40 kHz and power of 220 W.
For silica extraction by the alkaline sol-gel extraction method, analytical grade reagents NaOH and HCl were used without further purification.
As a reference sample for quality control, high purity commercial silica produced by Macherey-Nagel was used, with granulometry 70-230 mesh ASTM.

Immediate analysis
The samples were prepared and analyzed according to ASTM D1762 (ASTM 2007), for analysis of moisture, volatile matter, and ash. The determination of the fixed carbon content was obtained mathematically, subtracting from 100 the moisture content, volatile materials, and ash of the samples, according to Fernandes et al. (2016).

X-ray diffraction analysis (XRD)
The mineralogical characterization of RHAC 1 , RHAC 2 , and RHAP was analyzed by X-ray Diffraction (XRD) in a Miniflex 300 X-ray diffractometer (Rigaku) with copper Kα line radiation ( CuKα → λ = 1.5418 Å), located at the CADEQ Analytical Center (UFSM-RS). The operating conditions were 30 kV, 10 mA, and 0.03° per second sweep speed in the 2θ range between 5 and 99.98°. The identification of the peaks, the structural parameters of the unit cell, and the width at half height (FWHM) values were calculated using the diffractometer software PDXL (integrated X-ray powder diffraction software).
The crystallite size was estimated through the Scherrer equation, using the most intense reflection in 2θ, (Eq. 4) (LU et al. 2015).
where k is Scherrer's constant, the value of 0.9 was used; λ is the wavelength in nm; β is the width at half height of the crystallite's highest intensity diffraction peak (FWHM), and θ is the Bragg angle.
The estimation of the crystalline and amorphous percentage of the samples was performed by applying a simple area separation method, which compares the integrated intensities of the crystalline and amorphous phases, from the diffraction scan, and is based on the conservation of intensity. The integration is performed over the entire range of 2θ from 0° to 180°, where θ is the Bragg angle. The degree of crystallinity is calculated according to Eq. 5, using the Origin software (Stern and Segerman 1968).
where A cris is the crystalline integral area and A am is the amorphous integral area. An amorphism index of 100 − I cris was used.

X-ray fluorescence analysis (XRF)
The semi-quantitative chemical composition was measured using XRF. The samples were ground, sieved, and analyzed in the Epsilon 1 apparatus (Panalytical), using the semi-quantitative Ominion calibration curve, under the following analysis conditions: Ag anode; kV: 50; µA: 100; Filter: Cu-500; Detector: SDD 5; atmosphere: air. Analyses were carried out at the ITT Fossil laboratory at UNISINOS-RS (University of Vale do Rio dos Sinos).

Sol-gel process
The silica was obtained using the sol-gel process, which consisted of applying alkaline extraction, with and without the ultrasound aid, followed by neutralization with HCl for precipitation, condensation, and gelation (Fig. 1).
The RHAC 1 sample was used to develop the ultrasoundassisted extraction method and to compare with the traditional method (without ultrasound).
The determination of the silica yield, for both methods, was calculated using the ash content of the samples, obtained in the immediate analysis, (Eq. 6).
where Si O2 = final silica weight, in grams; mRHA = initial sample mass, in grams; and TC = ash content, in mass fraction.

Traditional method
For comparison with the ultrasound aided sol-gel method, a reference method (traditional method) was adopted in this study, which is based on the sol-gel method performed by Kalapathy et al. (2001) and later adapted by Fuzinatto et al. (2021). Typically, 5 g of sample was mixed with 100 ml of 1 M NaOH for 1 h at 90 °C with constant magnetic stirring. Subsequently, filtration was carried out on quantitative filter paper. The carbonic residue was reserved for further analysis, while the filtrate was cooled at room temperature and neutralized to pH 7 with 2 M HCl. The sample was then aged for 18 h, washed with distilled water, and centrifuged for 15 min at 2500 rpm. Silver nitrate was added to indicate the removal of NaCl, by the absence of turbidity. Finally, the sample was dried in an oven at 80 °C for 24 h, obtaining silica.

Ultrasound method
To evaluate the intensity distribution inside the ultrasound bath or possible intense cavitation regions, the methodology of Santos and Capelo (2007) was adopted. Strips of aluminum foil (5 cm × 10 cm) were immersed in the bath for thirty seconds; they were placed at various points in the bath and kept along the surface. The intensity of the ultrasonic wave was evaluated by the wear caused on them.
The ultrasound method was developed according to the following steps. 5 g of RHAC 1 (in triplicate) was mixed with different molar ratios of NaOH to silica of rice husk ash (n = n NaOH /n Silica ), with a constant NaOH solution volume of 100 ml in a 200 ml Erlenmeyer flask with a plastic cap. The Erlenmeyer was placed in the ultrasound for a set amount of time, according to statistical planning. The molar ratio (6) % η = (SiO2∕(TC × mRHA)) × 100 (n = n NaOH /n Silica ) was determined based on the ash content of the average RHAC 1 sample, considering that 100% of the ash is SiO 2 , to ensure that NaOH was in excess, according to the preliminary study (Fuzinatto et al. 2021). The temperature used was 60 °C (maximum temperature of the equipment). After the extraction, the procedures of precipitation, filtration, gelation, aging, washing, and drying followed the methodology described in the traditional method.

Experimental design
The response surface methodology (RSM) with symmetrical and second-order rotational central composite design (RCCD) was used to evaluate the effects of process variables on the yield of ultrasound assisted silica extraction. The experiments used 4 repetitions in the central point and axial part, totaling 12 experiments (Ferreira et al. 2018).
The following conditions were kept fixed: 5 g sample with granulometry below 0.710 µm, 100 ml of NaOH, 60 °C ultrasound bath temperature, and 2 M HCl concentration. The coded and natural values of the selected levels of the independent variables are presented in Table 1.
The analysis of the data obtained from the planning and its effects, coefficients, response surface, and significance were obtained using the software ®Statistica 7.1 (STAT-SOFT, USA).
The validity of the results obtained is strongly dependent on the normality of the analyzed data (Rodrigues and Iemma 2005). To verify the normality of the responses, the Shapiro-Wilk normality test was applied.
The relevance of the suggested model and its significance to the experimental data fit were verified using analysis of variance (ANOVA) (Cecon and Silva 2011). For this, the value of the coefficient of determination (R 2 ) and the Fisher's F test were used. An R 2 close to 1 indicates an adequate model. For the F test, the ratio between the mean square of the regression terms and the mean squared error, the F calculated , was calculated and compared to the critical value of F for a given level of significance, F table . If F calculated is greater than F tabled , then the model is adequate and significant (Calado and Montgomery 2003).
In addition, the lack of fit was calculated, which relates the mean square of the lack of fit to the mean square of the pure error. Finally, the assumption of normality of errors was verified through the normal probability plot of residuals. If they were normally distributed, then approximately 95% of the standardized residuals will fall in the range of (− 2 to + 2).
The percentage yield of silica production (Y) was considered the process response. This response is assumed to be affected by the relationship between the independent variables and was analyzed to fit a second-order polynomial equation (Eq. 7). The goodness of fit for the silica production yield equation was expressed by the R 2 .
where Y = response variable in terms of silica yield (%); X 1 and X 2 = input variables; β o = represents the global average of the observations; β 1 and β 2 = linear coefficient; β 11 and β 22 = quadratic coefficient; and β 12 = interactive coefficient. The analysis of the effects of factors on silica yields was performed using the Pareto diagram. Through this, the effects of both variables and the interaction between them were evaluated, comparing them with the minimum value of standardized significance.
Once the best conditions were defined, the method was carried out in triplicate for the other RHAC 2 and RHAP samples.
To control the quality of the procedures and evaluate possible losses during the process, a sample of high purity commercial silica was evaluated through the same process.

Thermal analysis (TGA)
Thermal characterization was carried out in a Thermal Analyzer TGA-50-Shimadzu equipment, with a heating ramp of 10 °C /min up to 1000 °C, with a nitrogen flow of 50 ml/ min, in the CIA/FURG laboratory of the Federal University of Rio Grande (FURG).

X-ray diffraction analysis (XRD)
For the mineralogical characterization of silica, its degree of amorphism was determined by X-ray Diffraction (XRD). As described in the characterization of rice husk ash.

Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDX)
The morphology was investigated by Scanning Electron Microscopy (SEM) in an EVO LS15 (Zeiss) apparatus using high vacuum, with BSE (backscattered electron) detectors. Before the analysis, the sample was dried for 22 h at 70 °C, and then, the particulate material was deposited on double-sided carbon tape and coated with Au (gold) in a metallizer (Quorum, model Q150R) for 60 s. A qualitative analysis of the chemical composition of silica was also performed using the energy dispersive technique (X-Act model from Oxford), carried out at the ITT Fuse laboratory, at UNISINOS-RS.

X-ray fluorescence analysis (XRF)
In this stage of the work, analyses of the qualitative chemical composition were carried out through the technique of dispersive energy. Also using XRF, the semi-quantitative analysis was performed as described in the rice husk ash characterization section.

Textural properties
Nitrogen adsorption was used to determine the specific surface area of the samples by the BET method (Brunauer-Emmett-Teller). Pore volume and diameter were calculated by the Barrett-Joyner-Halenda (BJH) method. The analysis was performed using a Gemini VII 2390ª apparatus, at the CIA/FURG laboratory of the Federal University of Rio Grande (FURG). The degassing of the samples was carried out in a vacuum at 200 °C for 4 h.

Immediate analysis
From the immediate analysis of the ash presented in Table 2, it can be seen that the two ash samples from the combustion in a mobile grate system (RHAC 1 and RHAC 2 ) have a low proportion of volatile material and fixed carbon, in contrast to the pyrolysis ash (RHAP), which showed high content of volatile material and fixed carbon. These results reflect the thermochemical processes where these ashes were generated. It can be seen that RHAC 2 , when compared to RHAC 1 , underwent a more intense combustion process, reducing both fixed and volatile carbon levels and presenting a higher ash content. For the RHAP samples, the fixed, volatile, and ash carbon levels are explained by the low process temperature and residence time (Dhyani and Bhaskar 2019). The RHAP sample was obtained in the slow pyrolysis process, where it is observed the highest biochar formation. Thus, the chemical bonds of the lignocellulosic material are broken slowly without significantly affecting the polymeric structure, favoring rearrangement reactions on the solid surface, secondary reactions associated with a longer residence time in the reactor, which benefits the formation of a carbon rich solid residue (Valadão et al. 2021). The results described in Table 2 are in agreement with those determined by Fernandes et al. (2016),  (2015) and Vieira (2018) for RHAP. Figure 2 shows the comparison of diffractograms for the three samples used in this study. It can be seen that the RHAP diffractogram shows only the amorphous halo characteristic of silica with a centralized peak, located between the 2θ positions of 15° and 30°, confirming that the silica present in this sample is amorphous. On the other hand, the RHAC 1 and RHAC 2 ash diffractograms show a decrease in the silica amorphism halo and show the presence of crystalline phases of silica. In the RHAC 1 diffractogram, where the halo of amorphism decreases, it is possible to observe a peak of Cristobalite in the 2θ position of 22° with an intensity of up to 600 cps. In the RHAC 2 diffractogram, where the halo of amorphism is significantly smaller, the cristobalite peak reaches the intensity of 2000 cps. Table 3 lists the amorphism indices of RHAC 1 and RHAC 2 , RHAP, and their respective crystallite sizes. When comparing these data with the immediate analysis (Table 2), it can be seen that the ash amorphism index increases in the following order: RHAP < RHAC 1 < RHAC 2 , which is exactly the reverse order of volatile material and fixed carbon in the ash, which follows the following ascending order: RHAC 2 < RHAC 1 < RHAP. Considering the literature, an increase in temperature for the thermochemical process is one of the deciding factors for the silica transformation from amorphous to crystalline. In addition, when comparing the crystallite size increase for RHAC 1 and RHAC 2 ash samples, which present silica crystalline phases, the crystallite size is greater for the RHAC 2 sample, as it has a lower volatile material and fixed carbon content. Possibly, these characteristics are a result of the burning conditions of the rice husks, mainly due to the temperature used. Some authors point out that crystallization starts at temperatures above 700 °C (Cordeiro 2009;Fernandes et al. 2016), while others suggest temperatures above 800 °C (Chauhan & Kumar 2013; Ghorbani et al. 2013;Soltani et al. 2015). Cordeiro et al. (2014) proved the correlation between the amorphism index and the firing temperature, where it is possible to verify that the increase in process temperature causes a drop in the amorphism index. It should also be noted that, in addition to the firing temperature, there are other variables that directly affect the reactivity of these ashes, such as residence time, type of process, cooling, granulometry, and material composition. Finally, the XRD agrees with the immediate analysis of the samples, where RHAC 1 and RHAC 2 (combustion) showed lower levels of fixed and volatile carbon, higher amounts of ash, and also showed an increase in their crystallinity, as a reflection of the thermochemical conditions that were generated. These characteristics decrease the silica extraction efficiency. This situation does not apply to RHAP, generated by the pyrolysis process (600 °C), which presented 100% amorphous silica. Table 4 presents the composition of the oxides determined by the XRF for the RHAC 1 , RHAC 2 , and RHAP samples used in this study, and compares with other authors results. Similar composition is observed between RHAC 1 and RHAC 2 with silica contents of 88.2% and 88.9%. Ash similar to the ones in this study was used in a work by Flores et al. (2021), which were obtained from the combustion of the same type of RH. The results are also in agreement with those described by Fernandes (2016), for RHA obtained by combustion in mobile grate systems similar to the RHA of this study.

X-ray fluorescence analysis (XRF)
As observed in the immediate analysis, RHAC 2 has a lower loss of ignition (LOI) content and higher levels of K 2 O. The potassium contained in RHA accelerates both the

Assessment of cavitations zones
Based on the tests with the aluminum foil strips, the regions of greatest wear on the paper were identified (Fig. 3), which shows that the bath presents differences in ultrasonic intensity depending on the location in the device's bowl. To overcome these variations during the extractions the Erlenmeyers were attached to a support, which alternated positions every 10 min to ensure a uniform effect of the ultrasonic bath during the process.  1 and 7 showed the lowest yields and were also the experiments with the shortest extraction times, which demonstrates the strong influence of time, because even increasing the molar ratio did not lead to a significant difference in yields. This influence is also observed when comparing samples 1 and 2, where the increase in time caused a significant improvement in yield.

Model development and statistical analysis
Using the values of the dependent variables (income), the Shapiro-Wilk normality test was performed, obtaining a value of p-value = 0.08044 greater than 0.05; there is no evidence to reject the null hypothesis, so the data follows the normal distribution.
The statistical analysis of the effects, considering the yield as a response, confirms that the linear time is the most significant (1.6 times greater than the linear molar ratio effect) and positive variable. The linear molar ratio, interaction, and quadratic time effects were also significant, with the last two causing a decrease in silica yield. The quadratic molar ratio effect was not significant, with significance below p = 0.05, being taken out of the regression calculation. The analysis of standardized variables using the Pareto diagram (Fig. 4) demonstrates that the effects that are above the threshold value for the confidence interval are significant.
Multiple regression analyses were performed to correlate the silica production responses (Y%) with the selected independent variables, using the polynomial illustrated by Eq. 7. The quadratic regression equation that leads to optimal silica production yield for this study, after eliminating the statistically non-significant coefficients, is shown as Eq. 8.
The statistical model presented in Eq. 8 must be predictive and meaningful to adequately represent the experimental data. The prediction and significance of the statistical model were evaluated using analysis of variance and Fischer's F test. Fischer's F test showed that the F calculated (8)  Fig. 3 Photo of aluminum strips, after an ultrasound test (F calculated = 537.48) was more than 119 times the value of F tabulated (F tabulated = 4.5). This implies that the use of the quadratic regression described in Eq. 8 is highly significant, with a significance level of 99%. The calculation of the coefficient of determination (R 2 = 0.993) shows that 99.30% of the variability in the response is compatible with the data predicted by the selected model. The predicted yield values of amorphous silica production are similar to the experimentally observed ones, showing that the experimental data follows a predictive model.
Through the normal probability graph of the residuals (Fig. 5), the assumption of normality of the errors was verified. They were normally distributed, and the interval was between − 1.2 and + 1.0, demonstrating the normality of the errors.
From the validated model, the response surface was constructed (Fig. 6). It can be observed that the increase in the reaction time and the molar ratio (n = n NaOH /n Silica ) increases the silica yield, as discussed earlier. It is possible to verify that, at low molar ratios and high reaction times, yields close to those with high molar ratios are obtained, a  Fig. 4 Pareto diagram for the effects of silica extraction yield fact that represents an economy both in the obtaining process and in the generation of residues. It is worth noting that the minimum ratio used in the process must respect the reaction stoichiometry for silica with NaOH (Eq. 1), which is 1, to avoid NaOH becoming the limiting reagent.

Silica extraction yield: traditional and modified method assisted ultrasound
The yield of silica extraction using ultrasound was evaluated in comparison with extraction by the traditional method. The results, presented in Table 6, show that the ultrasonic extraction method proved to be reproducible, with coefficients of variation (CV) lower than 5% for the different ash samples analyzed. It is also possible to observe the dependence of the matrix to be extracted, where samples with higher crystallinity showed lower yields. Effect also observed using the traditional method. A more complex matrix, such as that of RHAC 1 and RHAC 2 , which present differences both in their compositions and in the variables of the thermochemical process applied to AC, such as operating temperature, residence time, and cooling, can result in different morphologies, structures, and reactivity (Fernandes 2020), which ends up influencing the extraction yield. The silica yields calculated considering the percentage of silica present in the ashes, determined by XRF, and the percentage of amorphism, determined by XRD, were higher than those initially calculated. They better represent the extraction efficiency by considering the real influence of the sample on the extraction, as it is known that only amorphous silica is extracted and not all ash is silica. It can be seen that the highest percentage of increase in yield occurred in RHAC 2 due to its lower degree of amorphism, proving the influence of the rice husk combustion process conditions on the obtained ash and, consequently, on the extraction yield.
When applied to a commercial amorphous silica sample, the yields reached the recovery value of 98.8%, indicating minimal losses during the silica extraction process. The influence of the morphology in the extractive process was confirmed when the dissolution of commercial silica was reached within 20 min. The situation is also confirmed when compared with the traditional method. Samples with a higher degree of amorphisms, such as RHAP and commercial silica, showed less difference in yields between the methods, but this behavior was not so evident among combustion ash samples (RHAC), which suggests the influence of other components of the matrix in addition to the morphology of the silica to be extracted.
Considering the rice production of the South region, 2020/2021 harvest of 1,521,934 t/year of which 20% are rice husk, which corresponds to 304,387 t/year. In the conversion via the thermochemical process of CA, the average generation of 20% of RHA occurs (Fernandes 2020), which would result in the generation of 60,877 t/year of environmental residue. Subjecting these ashes to the extraction process carried out in this work, considering yields of 73.3%, there would be the production of 44,622 t/year (3,719 t/month) of silica with purity above 95%.
The reaction temperature in this work remained at 60 °C due to limitations of the ultrasound equipment, results by the traditional method with 90 °C and bibliographic data suggest that better yields can be achieved with higher temperatures ( Mourhly et al. 2019;Zaky et al. 2008).

Characterization of the silica obtained
Thermal analysis (TGA) Figure 7 shows the comparison of TGA for the silicas obtained by the traditional method and under ultrasound. Two main sections of mass loss stand out in the results. The first, up to approximately 100 °C, is probably due to the elimination of water molecules adsorbed on the surface with a weight loss of 9.9% for SRHAC 1 ultrasound sample (a) and 6.5% for SRHAC 1 Traditional sample (b). This result in agreement with the greater surface area of the silica obtained under ultrasound, since the silanol groups are found on the surface of the material. The second section is in the range from 200 °C to 744 °C for the traditional sample and up to approximately 790 °C for the ultrasonic sample. In this range, there is the conversion of silanol bonds into siloxane bonds on the surface of the silica, releasing the water molecules generated from this step and forming the product SiO 2 . These results agree with those observed by Nassar et al. (2019), who extracted silica from rice husk without applying ultrasound. Nassar et al. (2019) performed differential thermal analysis (DTA) and found endothermic peaks corresponding to the weight loss. Figure 8 shows the comparison of the XRD for the silicas obtained by the traditional and ultrasound methods. It is observed the characteristic amorphism halo of the silica with a centralized peak, located between the 2θ positions of 15° and 30°, and the absence of a crystalline peak, suggesting that the silica is in its amorphous form. Differences between the tested silicas were not evidenced by XRD, regardless of the origin of the ash or the extraction method adopted. The amorphism of the extracted silica denotes  Fig. 7 Thermal analyses (TGA)-silica obtained from RHAC 1 . a SRHAC 1 by the ultrasound method and b SRHAC 1 by the traditional method that only amorphous silica was extracted from the ashes, which may explain, in part, the lower extraction yields for the ashes that showed crystalline phases of silica (RHAC 1 and RHAC 2 ).

X-ray diffraction analysis (XRD)
Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDX) Figure 9 shows the SEM micrographs for the silicas obtained from the SRHAC 1 sample by the two methods, traditional and ultrasound. Both samples show small particles of different sizes and large agglomerates. It is not possible to identify significant differences between the silicas obtained by the traditional process (Fig. 9b) and the silica obtained by ultrasound (Fig. 9a). The images prove the presence of particles smaller than 10 µm, which can reach nanometric dimensions. EDX analysis confirms the purity of the silica product, with silicon and oxygen representing 100% of the composition of the silicas obtained by both processes. The gold peak (2.10 keV) is due to the preparation of the samples, which were coated with gold for analysis, while the carbon peak (0.277 keV) is due to the carbon tape used to mount the sample. The elemental composition of the samples was confirmed by EDX, proving they are majorly composed of silicon and oxygen, as shown in Table 7 and Fig. 10. Table 8 presents the results of the chemical analysis by XRF. It can be seen that the silicas obtained by ultrasound and by the traditional method are similar to each other. The SRHAC 1 obtained by the traditional method still showed an accentuated amount of Na 2 O remaining from the process. According to Kalapathy et al. (2001), metal ions (Na) are trapped in the silica gel network during gel formation, making it difficult to remove them when washing this gel. Sodium impurities can be removed by repeated washing with deionized water. In the work presented by Eliana dos Santos (2017), where the results are also listed in Table 9, it was verified that the amount of Na 2 O and SO 3 was very high, and a subsequent washing was necessary after drying the silica. This washing with water avoids acid washing, leading to less consumption of chemical products.

X-ray fluorescence analysis (XRF)
The Fire Loss values of the ultrasonic and traditional SRHAC 1 samples can be explained by the higher concentration of silane groups, which are located on the silica surface. These groups transform into siloxanes upon heating above 600 °C, releasing H 2 O and transforming into a hydrophobic surface. To confirm this procedure, the SRHAC 2 samples were heated to 600 °C in Muffle to remove the OH groups,   as observed by the reduction of the LOI value and increase of oxides in XRF analysis.
In the work of Steven et al. (2021), a series of industrial applications were presented for silicas with purity above 95%, including civil construction (high-performance cement, concrete, pozzolan additive); chemical industry (cellulose and paper, detergent, petrochemicals, fertilizer, carrier); special materials (polymer, glass, ceramic, solar panel, graphene, zeolite, catalyst support); electronics (battery, capacitor, semiconductor); insulators (thermal insulators, refractory materials, flame retardants); adsorbent materials (moisture adsorbent in the form of silica gel, activated carbon); and pharmaceutical and health industry (as an active ingredient vehicle in formulations). This demonstrates that the silicas obtained in this work, except for the sodium contaminated traditional SRHAC 1 , are suitable for use. Considering only the oxides, all the silicas obtained in this work, except for the traditional SRHAC 1 , have SiO 2 values above 95%. Figure 11 shows the N 2 adsorption-desorption isotherms for the silica samples SRHAC 1 , SRHAC 2 , and SRHAP obtained  by the utrasound method, and the SRHAC 1 obtained by the traditional method. All curves exhibited type IV isotherms with hysteresis, as shown in Fig. 11 (according to IUPAC classification). This indicates the mesoporous nature of the obtained silica products, as well as solids that have this type of hysteresis are of non-uniform sizes or shapes, with agglomerates of particles forming slit shaped pores, confirming the amorphism of silica (Razavi et al 2020).

BET (Brunauer, Emmett, Teller) and BJH (Barrett-Joyner-Halenda) analyses
The BET and BJH results obtained for the SRHAC 1 , SRHAC 2 , and SRHAP samples are shown in Table 9. It is verified that the silica surface area for the SRHAC 1 sample is 1.65 times greater when using the ultrasound method, instead of the traditional method, and the pore size is approximately half. This confirms the effect of ultrasound on the improvement in the textural properties of silica, which provides greater adsorption power and greater reactivity, in agreement with the studies by Peres (2018) and Ortiz (2016). The surface areas of the silicas are in agreement with the values found in the literature using the sol-gel method, such as Fernandes et al. (2016) and Yuvakkumar et al. (2012Yuvakkumar et al. ( /2014. It can also be seen that there are differences in the values of surface area and pore size for the silicas of the three ashes Sample isotherms a SRHAC 1 ultrasound, b SRHAC 1 traditional, c SRHAC 2 ultrasound, d SRHAP ultrasound extracted by ultrasound. The smallest surface area observed was precisely for the silica obtained from RHAC 2 , indicating once again that the thermochemical method of ash generation and the ash matrix itself can influence not only the amount of silica extracted but also the textural properties of the silica obtained.

Conclusion
Through this study, it was possible to develop a method for high purity silica extraction, aided by ultrasound, that used thermochemical solid waste generated by combustion and pyrolysis (CCAC and CCAP). Through the application of the DCCR model, the influence of the variables, molar ratio, and reaction time was verified, the latter being the most significant. The study suggests that with longer reaction times, the yields obtained increase. The average silica yield reached values of 31.1% for RHAC 2 , 43.9% for RHAC 1 , and 73.3% for RHAP, with n NaOH /n Silica molar ratio of 4.4 and extraction time of 107 min, values that can also be obtained with lower molar ratios. The second-order adjusted polynomial function, with the variables time and molar ratio, adequately describes the extraction of silica from thermochemical residues, with an explained variation around the mean of 99.3%. The mathematical model was predictive and showed no lack of fit.
By comparing the extraction of rice husk ash obtained by combustion and by pyrolysis, it was noticed that the extraction yield and the textural properties of the silica obtained depend on the characteristics of the ash used, and the degree of amorphism of the silica present in the ash has a strong influence since only amorphous silica was extracted. In addition, when the traditional extraction method was compared with the ultrasound for RHAC 1 , an increase in surface area and a decrease in pore size were verified, thus demonstrating the effect of ultrasound.
Obtaining silica with a high degree of amorphism, with particles possibly in the nano scale, purity above 95%, high surface area, and mesopores is of great importance. There is interest from the industry and many possible applications, such as electronics, ceramics, rubber, medicine, and catalysts. The possibility of obtaining a product with such characteristics, from waste generated in large quantities locally and around the world, constitutes an alternative to add value and an environmentally friendly destination for this residue.
Data availability Data sharing does not apply to this article. All data generated or analyzed during this study are included in this published article and the referenced articles.

Declarations
Ethics approval Not applicable.

Consent to participate Not applicable.
Consent for publication All authors read and approved the final manuscript. Therefore, all consent to publish this work in the journal Environmental Science and Pollution Research.

Competing interests
The authors declare no competing interest.