Nutrient Recovery from Dairy Processing Wastewater Using Biochar

Abstract

In this study, we examined the capacity of magnesium-amended biochar to recover nutrients from dairy processing wastewater (DPW). Our results suggest that biochar engineered with magnesium (Mg–BC) was successful at recovering over 3 mg of PO₄³⁻-P per gram of biochar (96%) in synthetic and raw DPW through a combination of adsorption and chemical precipitation. The addition of Mg–BC to increase the pH of the synthetic and raw DPW was likely important in promoting chemical precipitation and increased nutrient recovery. The Mg-dosed biochar also recovered 1.7 mg of NH₄⁺-N per gram of biochar (24%) from raw DPW. However, the raw biochar (R-BC) was only capable of recovering a maximum of 0.5 mg of PO₄³⁻-P and an insignificant amount (˂0.1 mg) of NH₄⁺-N per gram of biochar.

1. Introduction

An important treatment technique used in the dairy processing industry is anaerobic digestion. However, anaerobic digestion does not recover valuable dissolved nutrients like phosphorus, in the form of phosphate (PO₄³⁻-P), and/or nitrogen, as ammonium (NH₄⁺-N) [1]. As a result, dairy processing wastewater (DPW) has very high concentrations of PO₄³⁻-sP and NH₄⁺-N [2]. Phosphorus (P) is an essential nutrient for crop production and is traditionally mined from phosphate rock, a finite resource that exists in only a handful of countries worldwide [3]. Nitrogen is also an essential plant nutrient, and while bioavailable forms can be produced via the conversion of gaseous N into NH₄⁺, the release of N into waterways causes ecosystem degradation. Chemical precipitation of struvite and other PO₄³⁻-P minerals is known to be effective for recovering P from various waste streams including swine wastewater, landfill leachate, urine, and dairy manure [4]. NH₄MgPO₄·6H₂O (Struvite) is a crystalline compound consisting of equimolar concentrations of magnesium (Mg⁺²), NH₄⁺, and PO₄³⁻ ions. Struvite has the potential to be used as a commercial slow-release fertiliser due to its high concentration of beneficial nutrients and low solubility, providing a sustainable alternative to liquid mineral fertilisers [5]. Previous research on P recovery from DPW using chemical precipitation has shown that magnesium dosing was effective at recovering over 90% of the phosphate (PO₄³⁻-P) in the DPW [6]. In this case, precipitation was carried out on DPW that had undergone anaerobic digestion.

While chemical precipitation of DPW for the recovery of P has been proven to be successful, it can generate large amounts of sludge and precipitates containing heavy metals and pathogens, which require further treatment [7,8]. Furthermore, precipitation has also been shown to be ineffective at recovering NH₄⁺-N from DPW because it is either in excess concentration compared to the required stoichiometric ratio for struvite precipitation, or non-struvite phosphate-containing minerals form at the expense of struvite [6]. To address this challenge, this study explores the use of magnesium-modified biochar for nutrient recovery from DPW, aiming to overcome the limitations of chemical precipitation by enhancing both PO₄³⁻-P and NH₄⁺-N removal.

An alternative to chemical precipitation of nutrients from nutrient-rich wastewater is adsorption onto biochar. Biochar is a low-cost adsorbent produced by heating biomass in an anoxic environment, a process known as pyrolysis. Biochar can be produced from various types of organic waste streams, including agricultural waste and wastewater sludge, helping to reduce pollution and increase carbon sequestration [9]. Raw and modified biochar have both been shown to be effective at removing PO₄³⁻-P and NH₄⁺-N from wastewaters [10]. The recovery mechanisms vary greatly depending on the nature of the biochar and the catalysts used to enhance them, ranging from electrostatic interaction to surface precipitation [11]. Metals such as magnesium, aluminum, and calcium play an important role in removing PO₄³⁻-P and NH₄⁺-N from wastewater [12,13,14,15]. Modifying biochar with metal oxides, especially MgO, enhances phosphate removal through mechanisms such as surface precipitation and electrostatic attraction [16,17,18,19].

Magnesium-enriched biochar has been investigated as an effective sorbent for phosphate and ammonium ions in many studies. However, most of these studies have employed only synthetic phosphate and ammonium solutions to evaluate the adsorption performance of the biochar composites [20,21,22]. Only a few numbers of studies have utilised real wastewater sources, such as piggery digestate, acid extracts from incinerated sewage sludge ash, municipal wastewater, and contaminated irrigation water for the adsorption of phosphate and ammonium on biochar and modified biochar [23,24]. To the best of our knowledge, no study has evaluated nutrient recovery from dairy processing wastewater using both raw and magnesium-enriched biochar. This is especially relevant for anaerobic digester effluents with naturally high magnesium and calcium levels, which differ from typical wastewater streams. DPW contained up to ~34 mg/L PO₄³⁻-P and ~70 mg/L NH₄⁺-N. For comparison, piggery digestate generally contains PO₄³⁻-P in the range of 10–20 mg L⁻¹ and NH₄⁺-N between 20 and 40 mg L⁻¹, while municipal wastewater typically exhibits much lower nutrient concentrations, with PO₄³⁻-P < 10 mg L⁻¹ and NH₄⁺-N < 30 mg L⁻¹ [24,25]). These relatively high nutrient levels in DPW not only complicate downstream treatment but also present an underutilized opportunity for resource recovery through technologies such as biochar-based adsorption.

Here, we investigated the recovery of PO₄³⁻-P and NH₄⁺-N from post-AD dairy processing wastewater (DPW) onto biochar. Two types of biochar were tested, raw coconut shell biochar and engineered biochar (coconut shell with a magnesium catalyst). Mg was selected as a catalyst for the biochar because it can enhance surface precipitation and electrostatic interaction while also potentially triggering struvite formation, simultaneously recovering PO₄³⁻-P and NH₄⁺-N [26]. Mg treatment has also been shown to enhance pore size, thereby increasing physical adsorption capacity and the presence of oxygen-containing functional groups that facilitate electrostatic attraction and surface complexation [26,27]. Nutrient recovery onto the biochars was also tested on synthetic wastewater since the DPW can contain other compounds (metals, salts, and organics) that can potentially influence the adsorption chemistry.

2. Materials and Methods

2.1. Biochar Synthesis

The raw biochar (R-BC) used in this study was obtained from Carbon Activated Australia, Victoria, Australia. This material (#ACOC-L60) is a very high-activity Granular-Activated Carbon (GAC) manufactured by steam-activating coconut shell in a slow-rotating kiln heated to 900–1100 °C. The R-BC is then acid-washed and pH neutralised. The specifications of R-BC are summarised in

Table 1. The specifications of raw biochar (GAC).

Parameters	Values
Particle size	1.70–0.43 mm
Total Surface Area	1150–1200 m2/g
Apparent Density	0.50–0.52 g/cm3
pH	7–8
Ash content (maximum)	0.5%
Moisture content	2%
Iodine Number	1150–1200 mg/g
CTC No (Carbon Tetrachloride No) (minimum)	60

.

The engineered biochar (Mg–BC) was prepared from the R-BC using the method for Mg–O biochar nanocomposite synthesis reported by Zhang et al. [28]. Briefly, R-BC was first washed with Milli-Q deionised water (deionised water), followed by oven drying at 80 °C. Next, 40 g of magnesium chloride hexahydrate (MgCl₂.6H₂O) (purchased from Merck Life Science Pty Ltd., Victoria, Australia) was dissolved in 60 mL of deionised water to prepare a magnesium chloride (MgCl₂) solution (≈3.3 M). The washed R-BC was then immersed in the solution for 2 h before oven drying at 80 °C. After drying, the soaked R-BC was transferred to a crucible and placed inside the vessel of a muffle furnace. The vessel was then filled with N₂ to purge any oxygen from it and the crucible was heated to 600 °C (10 °C/min) under N₂ flow for 1 h. All chemicals used in these experiments, including acid, bases and analytical standards, were of reagent grade or higher.

2.2. Wastewater Sample Collection

The DPW was collected in 20 L sealed buckets from the outlet of the anaerobic digester at a local dairy processing facility in Casino, NSW, Australia. Synthetic wastewater (stock solution) was produced with ammonium chloride (NH₄Cl) and monopotassium phosphate (KH₂PO₄) to mimic the NH₄⁺-N and PO₄³⁻-P concentrations in the DPW, respectively. These chemicals were also purchased from Merck Life Science Pty Ltd., Victoria, Australia.

2.3. Adsorption and Kinetic Experiments

The adsorption and adsorption kinetics of PO₄³⁻-P and NH₄⁺-N on the R-BC and Mg–BC were investigated with the raw DPW. The experiments of adsorption and adsorption kinetics were performed at room temperature by adding R-BC or Mg–BC into test tubes containing full-strength raw DPW. Samples (0.05 g) of each of the R-BC and Mg–BC were placed separately into a series of polyethylene centrifuge tubes containing 50 mL of raw DPW. The tubes were then capped tightly and the mixtures were kept in suspension by shaking at 240 rpm at 23 ± 2 °C on a shaker for 24 h. Each tube was removed from the shaker at specified reaction time intervals of 1, 2, 3, 4, 6, 8, 16 and 24 h. The supernatant liquor from each tube was analysed for PO₄³⁻-P and NH₄⁺-N. All adsorption experiments were triplicated.

The experimental results were fitted to the following pseudo-first and pseudo-second adsorption kinetic models [29]:

Pseudo-first-order:

$$q_t=q_e\left(1-\mathit{exp}\left(-K_{1}t\right)\right)$$

(1)

Pseudo-second-order:

$$q_t=K_2{q_e}^{2}t/1+K_2{q}_{e}t$$

(2)

where K₁ and K₂ are the first-order and second-order apparent adsorption rate constants, respectively, (h⁻¹ and g mg⁻¹ h⁻¹) and q_t and q_e are the amount of sorbate adsorbed (mg g⁻¹) at time t and equilibrium, respectively. To fit the model, q_t was plotted over time and the rate constants (K₁ or K₂) and equilibrium adsorption values (q_e) were estimated from the fitting functions. The Excel Solver is used to calculate the models’ parameters.

2.4. Adsorption Isotherms Experiments

An adsorption isotherm describes how a solute interacts with a solid surface at a constant temperature. It shows the relationship between the amount of substance adsorbed (usually per unit mass of adsorbent) and its concentration in the surrounding phase (gas or liquid). Batch experiments were undertaken at room temperature with both the raw DPW and the synthetic wastewater to generate adsorption isotherms. Different concentrations of the raw DPW were prepared by diluting it with deionised water to achieve 0%, 25%, 50%, 75% and 100% strength. These corresponded to PO₄³⁻-P concentrations of 0.05, 8.1, 16.0, 24.6, and 34.0 mg L⁻¹, and NH₄⁺–N concentrations of 0.55, 17.2, 34.8, 54.1, and 72.4 mg L⁻¹, respectively. Similarly, synthetic wastewater was diluted to the same percentages (0%, 25%, 50%, 75% and 100%) using deionised water, with resulting PO₄³⁻-P concentrations of 0.10, 10.5, 18.6, 27.7, and 36.3 mg L⁻¹, and NH₄⁺–N concentrations of 0.3, 20.3, 35.5, 53.3, and 72.6 mg L⁻¹, respectively, to match the corresponding nutrient levels in raw DPW. These were adjusted with sodium hydroxide (NaOH) to match the pH of the raw DPW. Samples (0.05 g) of each of the R-BC and Mg–BC were placed separately into a series of polyethylene centrifuge tubes along with 50 mL of different concentrations of the raw DPW. The tubes were then capped tightly and the mixtures were kept in suspension by shaking at 240 rpm at 23 ± 2 °C on a shaker for 24 h, ensuring concentration equilibrium was reached before sampling. The equilibration time was determined from adsorption kinetics experiments. After sampling, the supernatant liquor was analysed for NH₄⁺-N and PO₄³⁻-P. Similar set of experiments for different concentrations of synthetic wastewater with each of the R-BC and Mg–BC was performed. All isotherm experiments were triplicated.

Langmuir and Freundlich isotherm models were used to simulate the adsorption isotherm experimental data. Two isotherm models were used to simulate the experimental data and the governing equations can be written as equations [17]:

Langmuir model

$$q_e=KQ{C}_e/\left(1+K{C}_e\right)$$

(3)

Freundlich model,

$$q_e=K_f{C}_e^{1/n}$$

(4)

where K represent the Langmuir bonding term related to interaction energies (L mg⁻¹) and K_f is the Freundlich constant related to adsorption capacity (L g⁻¹), Q denotes the Langmuir maximum adsorption capacity (mg g⁻¹) (also defined as the adsorbate surface concentration when all available adsorption sites are occupied), C_e is the residual nutrient concentration in solution at equilibrium, also known as the equilibrium concentration (mg L⁻¹), n is the Freundlich linearity constant and q_e is the amount of nutrient adsorbed per mass of adsorbent at equilibrium (mg g⁻¹) [30].

To fit the model, q_e was plotted against the equilibrium concentration (Ce) and the maximum adsorption capacity was estimated from the fitting functions. Measurements were taken before and after the 6 h isotherm experiments in order to investigate the ability of the R-BC and Mg–BC to affect the pH of the wastewater. The Excel Solver is used to calculate the Q, K and K_f using q_e and Ce data from the measurements.

Isotherm experiments are often conducted to study the adsorption of solutes (like heavy metals or nutrients) on solid phases under controlled temperature and concentration conditions. However, the solutes may not only adsorb but also precipitate as minerals if the solution becomes supersaturated. The chemical equilibrium model, Visual MINTEQ 4.0 (MINTEQ), was used to determine the potential species precipitating during the isotherm experiments by simulating saturation indices for various minerals [31]. Oversaturated and undersaturated mineral indices were determined with MINTEQ by using the chemical component and pH data from the isotherm experiments. This helps in distinguishing between adsorption and precipitation as removal mechanisms.

2.5. Sampling and Analysis

Analysis of dissolved nutrients was performed according to standard methods of APHA 2017 [32] with samples collected from the test tubes using a syringe and filtered (0.22 uM) into vials for nutrient analysis. Samples for dissolved metals analysis were filtered through 0.45 μm cellulose acetate and then acidified with nitric acid prior to analysis by Inductively Coupled Plasma–Mass Spectrometry (ICP–MS) or Flow Injection Analysis (FIA). The amount of PO₄³⁻-P or NH₄⁺-N removed from the DPW was calculated based on the difference between initial and final aqueous concentrations. All experiments were performed in triplicate.

X-ray diffraction (XRD) analysis was carried out to identify any crystallographic structure for the samples (Mg–BC, Mg–BC treated with raw DPW, and Mg–BC treated with synthetic DPW) using a computer-controlled X-Ray diffractometer (Philips Electronic Instruments, 310 Orchard Rd, Singapore 238864) equipped with a stepping motor and graphite crystal monochromator.

3. Results and Discussion

The nutrient concentrations in the raw DPW sourced from the AD system had an average concentration of 31.8 mg L⁻¹ of PO₄³⁻-P and 72.4 mg L⁻¹ of NH₄⁺-N. The concentration of PO₄³⁻-P was 85% of total P and NH₄⁺-N was 78% of Total N. The DPW also contained significant potassium, sodium, chloride, sulfur and calcium concentrations (

Table 2. Elemental composition of untreated DPW.

Elements	Concentration (mg/L)
Phosphorus	31.8
Ammonium	72.4
Iron	1.12
Silicon	11.8
Magnesium	11.1
Potassium	28.8
Sodium	390
Chloride	37.4
Sulfur	21.4
Calcium	35

).

Figure 1 shows the wide-angle XRD patterns of the various MgO-biochar samples. The XRD results confirmed that nanosized MgO particles were formed in the sample of Mg–BC. This indicates that the synthesis method can produce MgO-biochar nanocomposites.

3.1. Adsorption and Kinetics of Biochar

The effects of the Mg content on the phosphate (PO₄³⁻-P) and ammonium (NH₄⁺-N) adsorption are illustrated in Figure 2.

The results suggest that the removal of PO₄³⁻-P and NH₄⁺-N from DPW exhibited a strong dependence on Mg content in the biochar. There was a significant reduction in the PO₄³⁻-P concentration of the DPW when treated with the Mg–BC (biochar engineered with Mg), with 96% (33.1 g L⁻¹) of PO₄³⁻-P removed after 24 h (Figure 2). However, treatment with R-BC (raw biochar) only achieved a 7% recovery of PO₄³⁻-P (2.4 g L⁻¹) after 24 h. The results showed that the Mg–BC significantly enhanced the removal of PO₄³⁻-P with more than a 1250% increase compared to R-BC, which had a very low affinity of removal (PO₄³⁻-P < 3.5 g L⁻¹) (Figure 2). Also, the NH₄⁺-N recovery efficiency of Mg–BC was significantly greater than R-BC, achieving a 24% recovery of NH₄⁺-N (16.6 g L⁻¹) compared to almost 0% after 24 h, respectively. These results are consistent with previous studies demonstrating that Mg-containing biochar composites exhibited strong affinity for aqueous phosphate adsorption [16,19]. Similarly, MgO-modified diatomite [14] and lanthanum-assisted oak sawdust biochar [33] have shown effective phosphate removal capabilities. The previous studies reported that magnesium-modified biochar achieved phosphorus removal rates ranging from 85% to 92% when treating piggery digestate wastewater and reclaimed water streams [20,34]. Our findings further confirm that the Mg content in Mg–biochar composites significantly improve the removal of PO₄³⁻-P from aqueous solutions. The highest phosphate removal (96%) was achieved with the Mg–BC, indicating that biochar engineered with Mg plays a key role in optimising adsorption performance.

Adsorption kinetics is one of the most important factors in evaluating the efficiency of an adsorbent. The PO₄³⁻-P and NH₄⁺-N adsorption by the R-BC and the Mg–BC as a function of time are presented in Figure 3 and Figure 4. Both the R-BC and the Mg–BC showed recovery of PO₄³⁻-P from the raw DPW stabilising after approximately 200 min, indicating concentration equilibrium had been reached. Initially, PO₄³⁻-P adsorption occurred rapidly, followed by a slower phase, consistent with observations from previous studies [14,17]. PO₄³⁻-P recovery was greatly improved with the Mg–BC compared to the R-BC, with 3.2 mg g⁻¹ and 0.3 mg g⁻¹ removed, respectively (Figure 3). This might be due to the electrostatic attraction between the positively charged magnesium oxide surfaces and the negatively charged phosphate ions [17,35]. This was also observed in our study, where MgO particles were formed in the Mg–BC sample (Figure 1). The point of zero charge of MgO is approximately 12 [36], indicating that its surface remains positively charged under most natural aqueous conditions [37]. Consequently, MgO particles in the biochar matrix could effectively bind anions, such as phosphate ions, in aqueous solution through mono-, bi- and trinuclear complexion mechanisms [28]. In contrast, the surfaces of most unmodified carbon-based adsorbents (e.g., R-BC) are negatively charged; they have little or no ability to remove anions, particularly with respect to phosphate and nitrate [38,39].

The kinetic data of PO₄³⁻-P adsorption were fitted with the pseudo-first-order and the pseudo-second-order kinetic models. The estimated kinetic models’ parameters are presented in

Table 3. Estimated kinetic model parameters.

Adsorption Kinetics	Pseudo-First-Order		Pseudo-Second-Order
	K1 (h−1)	qe (mg g−1)	K2 (g mg−1 h−1)	qe (mg g−1)
PO43−-P Raw R-BC	0.0083	0.27	0.038	0.31
PO43−-P Raw Mg–BC	1	3.2	0.074	3.3
NH4+-N Raw R-BC	1	0.03	0.0217	0.07
NH4+-N Raw Mg–BC	0.6	1.5	0.0448	1.6

. Mg–BC shows significantly higher adsorption capacity and rate constants for both PO₄³⁻-P and NH₄⁺-N compared to R-BC (Table 3). The pseudo-second-order model generally shows higher q_e values, which may indicate a better fit for describing the adsorption behaviour. The kinetic model parameter showed that the phosphate adsorption was better described by the pseudo-second-order kinetic model with a relatively greater correlation coefficient (R²). The high coefficient of determination for Mg–BC and the pseudo-second-order model indicates that the PO₄³⁻-P recovery with the engineered biochar likely involves a process of chemical bonding, meaning electrons are shared or exchanged between the adsorbate (raw DPW) and the adsorbent (Mg–BC) [18]. The pseudo-second-order model assumes a monolayer adsorption system and the mechanism of adsorption may be determined by chemisorption [29].

NH₄⁺-N recovery was also found to be improved when utilising the Mg–BC compared with the R-BC, with 1.7 mg g⁻¹ and 0.2 mg g⁻¹ removed, respectively (Figure 4). NH₄⁺-N recovery persisted for approximately the first 1000 min of the experiment for the Mg–BC. Again, the pseudo-second-order model was a better fit for the experimental Mg–BC data, favouring a process of chemisorption rather than physisorption. NH₄⁺-N adsorption may be influenced by electrostatic interactions between the positively charged ammonium ions and the negatively charged inner surface of the biochar [18]. The relatively slow adsorption of NH₄⁺-N indicated that an intra-particle diffusion mechanism may be involved, as pore diffusion is typically slow and often serves as the rate-determining step. The previous studies [18,29] suggest that the slow adsorption of NH₄⁺-N likely involved the biochar’s porous network and surface, which facilitated the dispersion of MgO flakes. This structure may have enabled ammonium ions to diffuse through pores into the negatively charged carbon matrix via a combination of film and intra-particle diffusion.

3.2. Adsorption Isotherms

For comparative analysis, both raw DPW and synthetic DPW comprising only PO₄³⁻-P and NH₄⁺-N were used in the isotherm study. The use of synthetic DPW allowed for a controlled assessment of PO₄³⁻-P adsorption without the influence of complex matrix components present in raw DPW. In contrast, raw DPW contains a range of competing ions and potential chemical interferences (e.g., Ca²⁺, Mg²⁺, and other metals) (Table 2) that can influence adsorption behavior through precipitation or complexation reactions. The PO₄³⁻P adsorption isotherms for both synthetic and raw DPW highlight an important distinction between the q_e/C_e relationship for the R-BC and the Mg–BC (Figure 5).

Treatment with R-BC resulted in a slight increase in q_e with a greater Ce. However, the Mg–BC treatment presents an inverse relationship between the q_e and Ce concentrations. The experimental data for Mg–BC show a significantly higher adsorption capacity, especially at lower Ce, with values reaching up to 3.5 mg g⁻¹ and 3.2 mg g⁻¹ for synthetic and raw DPW, respectively. Comparatively, the results for the R-BC were significantly lower, with q_e values of 0.5 mg g⁻¹ and 0.3 mg g⁻¹ reached at higher Ce concentrations for synthetic and raw DPW, respectively.

The maximum adsorption capacity (Q) and the coefficients of determination (R²) for both Langmuir and Freundlich adsorption isotherms were calculated for PO₄³⁻-P and NH₄⁺-N adsorption in raw and synthetic wastewater (

Table 4. Maximum adsorption capacity (Q) and coefficient of determination (R²) values.

Adsorption Isotherm	Q (mg g−1)	R2 (Langmuir)	R2 (Freundlich)
PO43−-P Raw R-BC	0.3	0.63	0.63
PO43−-P Raw Mg–BC	1.8	0.34	0.1
PO43−-P Synthetic R-BC	0.5	0.99	0.99
PO43−-P Synthetic Mg–BC	1.9	0.36	0.07
NH4+-N Raw R-BC	1.0	0.72	0.72
NH4+-N Raw Mg–BC	1.0	0.42	0.98
NH4+-N Synthetic R-BC	0.1	0.50	0.50
NH4+-N Synthetic Mg–BC	0.6	0.03	0.02

). Langmuir and Freundlich isotherm curves were removed from Figure 4 due to their low R². For the adsorption of PO₄³⁻-P, Mg–BC showed the greatest Q of 1.9 mg g⁻¹ and 1.8 mg g⁻¹ in synthetic and raw DPW, respectively. R-BC presented a significantly decreased Q, when compared with Mg–BC, of 0.5 mg g⁻¹ and 0.3 mg g⁻¹ in synthetic and raw DPW, respectively. R² values for the PO₄³⁻-P adsorption isotherms for both the Langmuir and Freundlich models were between 0.63 and 0.99 for the R-BC, compared with R² values ranging from 0.07 to 0.36 for the Mg–BC. These adsorption capacities were similar to those reported in previous studies, such as 1.24 mg g⁻¹ for magnetic orange peel biochar [40] and 1.0 mg g⁻¹ for lodgepole pine wood biochar [41]. However, relatively high values of maximum adsorption capacity for PO₄³⁻-P on magnesium-enriched biochar were reported in the studies with synthetic phosphate solutions [21,22].

The poor fit displayed between the isotherm models and the Mg–BC experimental results suggests that multiple mechanisms of PO₄³⁻-P recovery occur during the mixing of the Mg–BC and the DPW rather than simply adsorption. Other processes, such as precipitation, may be combining with adsorption to increase overall nutrient recovery, increasing the q_e of the Mg–BC above its assumed Q. This theory is supported by the reasonable to excellent fit displayed between isotherm models and the R-BC experimental results, highlighting adsorption as the main method of nutrient recovery for the R-BC since, unlike the Mg–BC, it contains no ions with which PO₄³⁻-P precipitation could occur. A MINTEQ model used to simulate the PO₄³⁻-P adsorption isotherm was used to calculate the oversaturation of trimagnesium phosphate (Mg₃(PO₄³)₂) in solution during the PO₄³⁻-P experiment. Therefore, it is highly likely that chemical precipitation and subsequent recovery of PO₄³⁻-P from the DPW is co-occurring during Mg–BC addition. This may also explain the slightly higher recovery rates noted during the experiments undertaken using synthetic DPW as opposed to raw, since the raw DPW contains competing phosphate counterions (e.g., calcium), which can bind up the magnesium introduced by the Mg–BC, limiting PO₄³⁻-P recovery.

The raw wastewater contains a complex mixture of ions (Table 2) that can lead to multiple competing reactions occurring during treatment. Chemical equilibrium modelling performed with MINTEQ predicted the oversaturation of calcium phosphate compounds such as hydroxyapatite (HAP) and tricalcium phosphate during experiments involving raw wastewater and Mg–BC. HAP is an effective fertiliser for industrial and agricultural applications requiring the slow release of P and Ca nutrients [42]. However, HAP is also an effective adsorbent of heavy metals and any precipitate formed from the DPW should also be analysed for the presence of unwanted metals that could impact soil and plant health [43].

NH₄⁺-N adsorption in the raw DPW reached 1.6 mg g⁻¹ at a Ce of 57 mg, which was significantly better than the 0.7 mg g⁻¹ adsorbed by the R-BC at a Ce of 65 mg (Figure 6). The Freundlich model (R² = 0.98) fits the Mg–BC data more precisely than the Langmuir model (R² = 0.42), which suggests that multilayer adsorption or adsorption on heterogeneous surfaces may be occurring with the Mg–BC [44]. R-BC had minimal effect on synthetic DPW, adsorbing less than 0.2 mg g⁻¹ of NH₄⁺-N. Mg–BC in synthetic DPW produced unusual results of adsorption followed by desorption at increasing Ce. It is possible that the divalent Mg²⁺ cations from the biochar were competing with the monovalent NH₄⁺ cations in the synthetic wastewater. In this case, displaced NH₄⁺ ions would result in a higher concentration of NH₄⁺-N in solution.

For NH₄⁺-N adsorption, Q is 1.0 mg g⁻¹ for both R-BC and Mg–BC in the raw DPW but is only 0.1 mg g⁻¹ to 0.6 mg g⁻¹ in the synthetic wastewater for the R-BC and Mg–BC, respectively (Table 4). The lack of a significant trend for the Mg–BC in synthetic DPW isotherm data resulted in neither model representing the behaviour of this adsorbent due to irregular adsorption/desorption patterns.

3.3. Effect of Solution pH on Adsorption

pH is a critical factor that influences the adsorption ability of an adsorbent in solution. The pH of the raw and synthetic DPW was measured before and after the adsorption isotherm experiments (Figure 7). The initial pH of the raw and synthetic DPW was 7.4 and 7.2, respectively. The average final pH of the raw and synthetic DPW treated with Mg–BC was 9.4 and 9.5, respectively. However, the average final pH of the raw and synthetic DPW treated with R-BC was only 8.1 and 8.0, respectively. This indicates that the presence of the Mg–BC in both the raw and synthetic DPW is causing an increase in pH.

This elevation in pH by Mg–BC may be attributed to the inherent alkalinity of Mg compounds and their surface chemistry. Magnesium hydroxide and other Mg-based species on the biochar surface can hydrolyze and release OH⁻ ions into solution, thereby raising the pH. This chemical behavior is consistent with previous findings that Mg-enriched biochars typically exhibit a liming effect, increasing the pH of acidic and neutral waters or soils [45].

Solution pH plays a pivotal role in adsorption processes, particularly when recovering nutrients such as phosphate (PO₄³⁻-P) and ammonium (NH₄⁺-N) from wastewater using engineered biochars. It influences not only the ionization of adsorbate species but also the surface charge of the adsorbent, thereby affecting adsorption efficiency and mechanisms. Recent studies have shown that the recovery efficiency of PO₄³⁻-P is significantly improved at a pH of >9 [46]. Since the solubility of PO₄³⁻-P starts to decrease as pH increases above ~8, studies have found that when pH increases beyond this range, PO₄³⁻-P tends to precipitate as various insoluble compounds such as hydroxyapatite, brucite and struvite [47,48]. This substantiates the possibility of precipitation occurring with adsorption and helps to explain why PO₄³⁻-P recovery was greater when treated with Mg–BC as opposed to R-BC (Figure 3).

Generally, a higher pH can also enhance the adsorption of NH₄⁺-N, due to increased electrostatic attraction since there is a lesser concentration of H⁺ competing for binding space in the biochar with the NH₄⁺ [49]. Within the pH range of ~7 to 9.5, an increase in pH typically results in a reduction in H⁺ competition at the adsorbent surface, thus improving NH₄⁺-N uptake through electrostatic attraction [50]. Mg–BC’s enhancement of pH likely supports this mechanism, as confirmed by higher NH₄⁺-N recovery compared to R-BC. Additionally, modifications with Mg may introduce functional groups that increase cation exchange capacity (CEC), further supporting NH₄⁺ retention [51]. Another study reported that the adsorption capacity of NH₄⁺-N increased with rising solution pH, suggesting that NH₄⁺-N responded to the decreasing positive charge on the Mg-modified biochar surface [18]. This may be the reason for the increased NH₄⁺-N recovery occurring with the Mg–BC, since it is increasing the pH of the DPW.

4. Conclusions

This study underscores the potential of coconut shell biochar engineered with a magnesium catalyst to significantly enhance the recovery of phosphate from dairy processing wastewater. Magnesium-enhanced biochar was more successful than raw biochar at removing PO₄³⁻-P from the DPW through a combination of adsorption and chemical precipitation. This study used adsorption kinetics, adsorption isotherms and chemical equilibrium models to evaluate the nutrient recovery of the biochars. While not as significant as the PO₄³⁻-P recovery, the recovery of NH₄⁺-N was also improved by the altered physiochemical properties of Mg–BC. Mg–BC’s ability to elevate pH directly supports its superior performance in nutrient recovery from wastewater compared to R-BC. The magnesium-enhanced biochar also provided an alternative method for increasing wastewater pH while simultaneously producing a PO₄³-P-enriched biochar that may be of value to agriculture. Further research is required to improve NH₄⁺-N recovery and to investigate the chemical composition of the enriched biochar. Future work is also needed to evaluate the bioavailability of these captured PO₄³⁻-P and NH₄⁺-N to plants in soil application. Future studies should assess the bioavailability of the captured PO₄³⁻-P and NH₄⁺-N when applied to soil, as this will determine their effectiveness as a slow-release fertiliser. Evaluating plant uptake under realistic soil conditions and over extended periods will provide crucial insights into the long-term agronomic value of the recovered nutrients. Such investigations are essential to support the practical application of nutrient-laden biochar in sustainable agriculture and to promote resource-efficient nutrient recycling.