Phosphorus removal in domestic wastewater treatment plant by calcined eggshell

The circular economy aims to keep the value of products, materials, and resources in the production process, minimizing the generation of waste. This approach seeks to separate economic development from the consumption of finite resources (Ellen MacArthur Foundation 2017). Various anthropogenic wastes and materials are integrated in this concept of circular economy.

Phosphorus (P) is an essential nutrient and is present in most fertilizers used in the global food production chain (Cordell et al. 2011; Wilfert et al. 2015). The main source of P is phosphate rocks, a non-renewable natural resource. The size of the world's phosphate reserves is a matter of divergence, with estimates of duration varying from decades to hundreds of years (Van Vuuren et al. 2010). However, there is general agreement that the quality of reserves is deteriorating in terms of content and quality (Desmidt et al. 2014), with extraction increasingly complex and costly. The production of phosphate rocks in 2018 was 270 million tons (US Geological Survey 2019).

Phosphorus is present in wastewater and is a pollutant when released into water bodies. It accelerates eutrophication, which can cause serious damage to aquatic ecosystems and water supply (Peng et al. 2018). Phosphorus concentration in domestic wastewater is in the range of 4–12 mg L⁻¹ (Metcalf & Eddy 2014). Within wastewater treatment plants (WWTP), a major source of P is the supernatant of the anaerobic digestion of sludges formed in primary sedimentation tanks and biological treatment. This supernatant contains high concentrations of solids, biochemical oxygen demand (BOD) and nutrients, and it is usually recycled to the head of the plant. P concentration in the supernatant of anaerobic digester varies in the range 110–470 mg/L (Metcalf & Eddy 2014).

Yuan et al. (2012) suggested that municipal wastewater is a potential source for P recycling and estimated that its recovery could satisfy 15% to 20% of world demand. One alternative for P recovery in wastewater and sludge treatment plants is its adsorption to a residue (Maroneze et al. 2014).

Eggs are one of the most important foods in the world, as they are sources of nutrients essential to human diet. According to the Brazilian Animal Protein Association (ABPA), in 2019 the national consumption of eggs was 230 units per person annually (ABPA 2020). The world production of eggs reached 7.2 × 10⁷ t year⁻¹ in 2017, representing an increase of 150% with respect to 1990 (FAO 2016). Eggshells are produced as waste when eggs are used directly for food or as raw material in the food industry. The shell represents approximately 11% of the egg's weight (Quina et al. 2017). Approximately 94% of the shell is composed of calcium carbonate (Mittal et al. 2016). Quina et al. (2017) classified the potential reuse of this material, including its use as food additive, soil correction, and as pollutant adsorbent (Figure 1).

Köse & Kivanç (2011) reported the use of calcined eggshell (CES) for adsorption of phosphate from aqueous solution. The authors found that desorption was not completely reversible. However, instead of recovering P to its natural form, an alternative is to use the resulting material (eggshell + P) as fertilizer.

Several studies using raw or modified calcined eggs have been reported. Morales-Figueroa et al. (2021) synthesized a material composed of eggshells, magnesium oxide and acetic acid. They found that total P could be eliminated from an industrial wastewater using 13.2 mg L⁻¹ calcium magnesium acetate, pH 12 and 12 min contact time. Ribeiro et al. (2020) prepared an adsorbent using eggshell, hydrochloric acid and sodium hydroxide. The authors found that 329 mg P g⁻¹ could be recovered using samples prepared from a P-stock solution. Using eggshell calcined at 800 °C, Torit & Phihusut (2018) measured 80% P removal from domestic wastewater containing 1.7 mg L⁻¹ P. Almeida et al. (2020) produced adsorbents composed of raw or CES modified by solutions of ferric chloride hexahydrate and magnesium chloride hexahydrate. The best adsorbent removed 99% P in samples prepared from stock solutions and 51.8% from an effluent containing 10 mg L⁻¹ P. They concluded that the complex matrix of the effluent had a negative effect on adsorption. Panagiotou et al. (2018) used eggshells calcined at 900 °C for 30 min to adsorb P that was leached by sulfuric acid (H₂SO₄) from anaerobic digester sludge. The authors observed the formation of Brushite (CaHPO₄·2H₂O), a material suitable to use as fertilizer.

This study aimed to evaluate the use of calcined and ground eggshells as adsorbent of P present in synthetic samples, in effluent from preliminary wastewater treatment, and in the supernatant of anaerobic digester used for sludge treatment. Eggshells were not modified other than calcination, and one focus of this work was the application of the adsorbent to the supernatant of anaerobic digester, a highly concentrated and complex liquid that is formed in wastewater treatment. It was hypothesized that a complex formed by Ca and P could be produced that has potential use as a soil conditioner in agriculture. However, this beneficial use was not tested in the present study.

Eggshells were collected from household consumption of one of the authors. To remove impurities, the shells were rinsed with deionized water and subsequently dried at 100 °C for 24 h in an oven. After drying, the eggshells were separated into two parts. One was calcined in a furnace at 600 °C for 4 h, and the other at 800 °C for 2 h. The resulting products were named, respectively, CES600 and CES800. The CES were subsequently ground with a mortar and pestel. The resulting powder was screened through a sieve with size less than 0.595 mm (30 mesh). According to Panagiotou et al. (2018), CES with particle diameters less than 1 mm enhanced P adsorption capacity, since there was an increase in the specific surface area (SSA) of the material, as quantified through gas sorption data.

Preliminary tests were carried out with stock solutions composed of deionized water and monopotassium phosphate (KH₂PO₄), with P concentrations of 15 mg L⁻¹ P. In addition, domestic wastewater samples were collected for adsorption and kinetics tests from an activated sludge municipal WWTP in Porto Alegre, Brazil. Samples from preliminary treatment effluent and supernatant from anaerobic digester were used for the tests. These sampling points were chosen due to the high concentrations of P present. All samples were collected on three different days. Table 1 shows preliminary treatment effluent and anaerobic digester supernantant composition of samples collected at the wastewater treatment plant. Parameters conductivity, total Kjeldahl nitrogen (TKN), total P, and alkalinity were more concentrated in supernatant than in preliminary effluent. In particular, the concentration of P was almost 19 times higher in supernatant, which makes this liquid attractive for P recovery techniques.

Thermogravimetry/derived thermogravimetry tests (TG/DTG), scanning electron microscopy (SEM), elemental composition, and SSA were used to characterize the adsorbents produced by eggshells (CES600 and CES800).

For the eggshell TG/DTG test, a term scale (SDT Q600, TA Instruments-Waters) was used. The test was carried out in nitrogen, using a heating ramp of 10 °C min⁻¹, at a temperature between 25 and 900 °C. SEM was performed with the Phenom ProX Desktop SEM equipment, operating in the range of 10 and 15 kV and nominal resolution of 8 nm. The microscope was equipped with an energy dispersive X-ray spectroscope, which allowed the chemical composition of CES600 and CES800 samples before and after the P adsorption tests to be determined. The SSA of the samples was examined using the nitrogen sorption method (Brunauer-Emmett-Teller (BET) method) using the equipment Micrometrics Tristar^® II 3020.

Kinetic models are useful for understanding the dynamics of contaminant adsorption in the adsorbent, as well as the adsorption mechanism. Kinetic tests were performed using eggshell concentrations that showed the highest P removals. In these tests, the flasks were placed into a shaking table operating at 150 rpm and temperature of 20 °C. Aliquots were taken after contact times of 1, 5, 10, 15, 30, 45, 60, 90, and 120 min of agitation. The aliquots were filtered through 0.45 μm filters for P analysis. The adequacy of the results to the pseudo-first and second-order models was verified.

Total P concentrations in samples were determined using the Stannous Chloride Method (4500-PD) (Standard Methods 2005). Absorbance was measured at 690 nm wavelength using spectrophotometer UV-1600, Pro-Analise. The samples collected at the WWTP were digested using the Persulfate Method (4500-PB.5) (Standard Methods 2005) to convert organic and acid-hydrolysable P to ortho-phosphate. Samples were not filtered before digestion. Other analysis made in wastewater and supernatant used standard methods from APHA (Standard Methods 2005): pH (method 4500–H⁺ A), conductivity (method 2510 B), turbidity (nephelometry, method 2130 B), COD (closed reflux, titrimetric, Method 5220 C), 5-day BOD (respirometric, method 5210 D), TKN (Method 4500-N_org B), alkalinity (titration, method 2320 B), and TSS (gravimetric, method 2540 D).

The Wilcoxon–Mann–Whitney test was used to compare adsorption at temperatures 600 °C and 800 °C for each of the quantities of calcined eggs added. Analysis of variance (ANOVA) was used to compare removal means among the same sample. Two-way ANOVA was used to compare removal means and CES concentrations in raw, CES600, and CES800. The Tukey test was used for multiple comparisons for all possible pairs of CES800 eggshell concentrations. All tests considered a significance level of 0.05. The correlation between pH and P removal was determined with the Spearman coefficent. R software from RSudio plataform was used for statistical analysis. Nonlinear regression models were calculated using Excel software.

The TG/DTG curve of the eggshell thermal decomposition reaction is shown in Figure 2. The data showed that the thermal decomposition of the eggshell could be divided into three events of mass losses. The first (Δm₁ = 2.7%), which occurred between 25 °C and 100 °C, can be attributed to the loss of moisture in the eggshell. The second event occurred between 100 °C and 600 °C, with mass loss Δm₂ equal to 15.1%. Decomposition of organic matter took place in this range of temperature. The greatest reduction in eggshell mass happened between 600 °C and 765 °C, with mass loss Δm₃ of 35.8%. Higher temperatures promote the release of carbon dioxide (CO₂) and calcium oxide (CaO) formation. Upon reaching 900 °C, there was a total loss of mass of 54%. Rodrigues & Ávila (2017) found reductions of 1.0%, 7.7% and 39.5% for Δm₁, Δm₂ and Δm₃.

The SSA of the eggshell particles were measured by the nitrogen sorption method (BET method). The gas sorption measurements indicated that SSA increased as the calcination temperature rose. The raw eggshells (ES), CES600 and CES800 presented SSA of 0.30 m² g⁻¹, 0.65 m² g⁻¹ and 2.8 m² g⁻¹, respectively. CES800 surface area increased 4.3 times in relation to CES600 and 9.4 times in relation to ES. Panagiotou et al. (2018) reported SSA for ES, CES600, CES800, and CES900 of 0.31 m² g⁻¹, 0.41 m² g⁻¹, 1.3 m² g⁻¹ and 1.6 m² g⁻¹, respectively.

Figure 3 shows an illustration of raw ground eggshells and those that have been calcined at 600 °C for 4 h and 800 °C for 2 h. The increase in calcination temperature resulted in residues with different visual characteristics, such as size and color of particles.

The eggshell has a porous structure, responsible for the gas exchange between the external and internal media. Figure 4 shows SEM images of CES600 and CES800 in different magnifications (×500, ×1,000, and ×2,000). The images indicated that the increase in temperature resulted in greater porosity of the residues, corroborating the results of the BET analyses.

After carrying out the adsorption tests, the CES800 adsorbents were again visualized with SEM (Figure 5) and had chemical composition measured by energy dispersive X-ray spectroscopy. Table 3 shows the elemental compositions of the CES800 adsorbent after the adsorption test with the supernatant of the anaerobic digester. The CES800 composition remained similar as before adsorption. The evidence that P was not found as a major element in the composition of CES800 after the adsorption test may be related to the fact that its increase was not sufficient to represent a significant fraction of the adsorbent material.

Figure 6 presents the results of the P removal tests with different concentrations of eggshell applied to samples from synthetic and preliminary treatment effluent with CES600 (a), with CES800 (b), and sample from the anaerobic digester supernatant, using CES800 (c).

The results show that for CES600 the efficiency of P removal varied according to the sample. For the synthetic solution, which had an initial P concentration of 15 mg L⁻¹ P, 10 g L⁻¹ CES600 removed more than 90% of P. In preliminary treated effluent, this removal could not be reached even using a concentration five times as high. For the preliminary treated effluent, the removal efficiency grew with increasing adsorbent concentration, reaching up to 85% removal. Almeida et al. (2020) observed P removal decreased from 99% in stock solutions to 51.5% in wastewater effluent. They attributed the reduction to the complex nature of the effluent in comparison to the stock solutions.

The eggshell concentrations of the CES800 adsorbent were 10–25 times lower than the concentrations that were used with CES600 (Figure 6(a) and 6(b)). A CES800 concentration of just 0.1 g L⁻¹ was sufficient to remove more than 90% of P in the synthetic sample. For the preliminary effluent, 1.0 g L⁻¹ CES800 removed 98% P. At a significance level of 0.05, application of Wilcoxon–Mann–Whitney test rejected the null hypothesis of equality in the efficiency of P removal by eggshells calcined at temperatures 600 °C and 800 °C (p < 0.001).

Application of the Tukey test for comparisons of all possible pairs of CES800 eggshell concentrations showed there were no significant differences in P removal for CES800 concentrations higher than 0.1 g L⁻¹ in synthetic solution. For P adsorption from preliminary effluent, CES800 concentrations above 0.3 g L⁻¹ were not significant. Therefore, the concentration of 0.3 g⁻¹ could be chosen to remove 85% of the initial concentration of P with 30 min contact time.

The initial P concentration in the supernatant from anaerobic digester was 137 ± 6 mg L⁻¹, which was higher than those in the other samples. For this reason, concentrations of CES800 had to be much higher than those used for synthetic and preliminary effluent samples. For instance, the application 1.0 g L⁻¹ of CES800 removed more than 95% of P in preliminary effluent (Figure 6(b)), but only 39% in supernatant (Figure 6(c)). To achieve 95% P removal from supernatant, a CES800 concentration of 30 g L⁻¹ was needed.

ANOVA was used to verify whether there were significant differences in the percentages of P removal from the supernatant of the digester with respect to the increase in the concentration of CES800 eggshell. For significance level of 0.05, the hypothesis that there were no differences between the concentrations of CES800 used for P removal was rejected (p < 0.05). The results of application of the Tukey test for comparisons of all possible pairs of CES800 eggshell concentrations are shown in Figure 7. All comparisons of CES800 eggshells concentrations that were significantly different are shown in blue. The pairs of concentrations that were not significantly different cross the zero value in the estimated range, and are shown in red. The only pairs of concentrations that were not significantly different were those involving concentrations higher than 20 g L⁻¹. At this concentration, CES800 removed 90% of the initial concentration of P with a contact time of 30 min.

Figure 8 shows the pH values measured in anaerobic digester supernatant after 30 min contact time for different CES800 concentrations. The initial pH of the supernatant was approximately 7.5 and gradually increased as the adsorbent concentration increased. With CES800, a concentration 50 times greater than used in preliminary effluent was required to reach pH 12. This behavior was due to the buffer capacity of the supernatant that resulted from the production of bicarbonate during decomposition of proteins in anaerobic digestion.

Correlation analysis was used to assess the relationship between pH and P removal. The non-parametric Spearmans coefficient was 0.92, indicating a strong correlation. Figure 9 shows the fraction of P removed as function of pH. Above pH 11, P removal was over 90%. A possible explanation for the correlation between pH and P removal is the reaction between Ca and P, with formation of hydroxyapatite (Metcalf & Eddy 2014). CaO is more soluble than CaCO₃, and, in solution, can lead to precipitation of calcium phosphate (Ca₃(PO₄)₂) (Molle et al. 2005; Paradelo et al. 2016). However, these results cannot be understood only because of the pH increase because higher pH values were reached using higher eggshell concentrations and therefore had larger surface areas available for P adsorption.

Köse & Kivanç (2011) measured positive values of zeta potential on the surface of CES in the pH range 2–10. Because of the positive charge in eggshell surface and the negative charge of phosphate ions, these authors considered electrostatic attraction an important mechanism of P removal. They found that more than 99% P removal could be reached in the pH range 2–10. This was different from what was measured in the present study, as shown in Figure 9, with maximum removal at pH above 10.

Based on values of EC presented in Table 1, μ of preliminary effluent and digester supernatant were 0.011 and 0.059, respectively. Because of its much higher ionic strength, the adjusted equilibrium solubility product constant in digester supernatant was approximately 4.5 times lower than in preliminary effluent. Both samples had lower solubility constants than in solutions with ideal behavior, in which molar concentrations were used. Estimations of the adjusted solubility product constants for preliminary effluent and digester supernatant is presented in the Supplementary Material.

The nonlinear equations of Langmuir and Freundlich models were applied to adsorption of P present in synthetic water, preliminary treated effluent, and supernatant from anaerobic digester. For the latter, only CES800 was used considering previous results showing that this adsorbent needed much lower concentrations than CES600. The Langmuir isotherm model assumes a homogeneous surface with a fixed number of accessible sites for adsorption, all with the same activation energy and reversible chemical equilibrium. The Freundlich equilibrium model is suitable for heterogeneous surfaces (Nazaroff & Alvarez-Cohen 2001).

The Langmuir (q_max, b) and Freundlich (k_f, 1/n) constants are presented in Tables 4 and 5, together with the coefficients of determinations R² and the normalized mean square error (NMSE). Lower NMSE values describe a better fit (Foo & Hameed 2010). The experimental data and model for CES600 and CES800 are shown in Figure 10.

Both models fitted relatively well experimental data. Phosphorus removal is normally described as adsorption on a heterogeneous surface, as the Freundlich model (Köse & Kivanç 2011). On the other hand, Panagiotou et al. (2018) found that the Langmuir model described better P adsorption to CES.

The adsorption capacities q_max in Langmuir equation were 263, 36, and 1.6 mg P g⁻¹ CES800, respectively, for synthetic sample, preliminary treatment effluent, and digester supernatant. The adsorption capacity depended on the sample, and was higher for stock solution and lower in digester supernatant. Complex samples have more chemicals constituents competing for the adsorption sites (Benjamin & Lawler 2013). The lower q_max indicated there were less sites available for P adsorption compared with synthetic and preliminary effluent samples.

Table 6 shows maximum adsorption capacities for CES and other adsorbents reported in other studies. Except for this study, all isotherm tests were made using P diluted in distilled water, which may not represent adsorption in complex liquids. Adsorption capacities varied 69 times from lower to higher values of q_max. The adsorption capacity of CES800 for P in anaerobic digester supernatant was much lower than those for samples prepared from distilled water.

Kinetic tests were carried out to analyze the behavior of P adsorption by the adsorbent over time. A 20 g L⁻¹ CES600 concentration was used for tests with synthetic and preliminary effluent samples. For the CES800 adsorbent, the concentrations used were 0.1 g L⁻¹ for the synthetic sample, 0.2 g L⁻¹ for the preliminary effluent, and 20 g L⁻¹ for supernatant from the anaerobic digester. These concentrations were chosen for P removal efficiencies above 70%. The tests lasted 120 min. Samples were collected during the testing period. Figure 11 shows the results from the tests.

Phosphorus removal was fast in the synthetic sample, reaching almost 100% in less than 5 min. In preliminary treated effluent, removal reached 70% with CES600 and 80% with CES800. Most of the adsorption occurred within 30 min. Köse & Kivanç (2011) and Panagiotou et al. (2018) tested adsorption to CES over 10 h and 24 h, and found that most of the adsorption took place in the first 30 min.

It is interesting to note that removal of P from the digester supernatant was higher than in effluent from preliminary treatment, reaching almost 100% after 90 min. P concentration in supernatant was about 19 times higher than in the effluent (137 versus 7.3 mg L⁻¹P) and this might have improved the adsorption to CES800 adsorbent. Eggshell concentration for digester supernatant was 100 times the concentration used in preliminary effluent. Calcium released into solution from CES could have reacted with phosphate ions to form precipitates as discussed earlier.

The parameters calculated for the pseudo-first and second-order kinetics models are shown in Tables 7 and 8, respectively, together with the values of R² and NMSE. For synthetic sample and preliminary effluent, the results showed that the pseudo-second-order model represented the experimental data better compared to the pseudo-first-order model. For supernatant of anaerobic digester, both models fitted the data well. These results support findings from previous studies on the adsorption of P (Köse & Kivanç 2011; Oliveira et al. 2015; Panagiotou et al. 2018).

Adsorbent produced by eggshell calcined at 800 °C for 2 h was able to remove P in samples from wastewater. At concentrations of 0.3 g L⁻¹ and 20 g L⁻¹, CES800 removed more than 70% of the P present in preliminary effluent and supernatant from anaerobic digester after 30 min contact time.

The thermal decomposition of the eggshell was divided into three mass loss events, totaling 54% loss at 800 °C. Also, at this temperature of calcination, the SSA of the eggshell increased by 4.3 times the area formed at a temperature of 600 °C and 9.3 times the area of the raw eggshell.

In addition to morphological changes that were induced by the increase in the calcination temperature, there were also chemical changes. With higher calcination temperatures, atomic calcium concentration increased, and carbon decreased. The elementary compositions of the adsorbents before and after the contact times with samples were similar.

Langmuir and Freundlich isotherm models fitted the experimental data well. There was a clear effect of pH on the removal process. Ca₃(PO₄)₂ precipitation was a potential secondary mechanism for P removal, in addition to the adsorption process. The kinetics of P adsorption was fast, occurring especially during the first 30 min. The pseudo-second-order model fitted the experimental data better, although the first-order model was also a good fit for P adsorption from the supernatant of digester.

It is possible to expand the tests with CES800 for continuous and mixed flow reactors considering the excellent removal that was achieved in 30 min of contact time. Further studies can advance knowledge about the potential use of the eggshell–P aggregate in agriculture as a soil conditioner, reusing a non-renewable resource (P) and a reusable waste (eggshells), which is usually sent to landfills. An economic feasibility evaluation for producing the adsorbent is also required.

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