Chitosan impregnation of coconut husk biochar pellets improves their nutrient removal from eutrophic surface water

The presence of excess nutrients in water resources can be harmful to human health and aquatic ecosystems. To develop an affordable water treatment method, the agricultural waste material coconut husk was converted into a low-cost adsorbent by thermal conversion to biochar, pelletized without (CH), and with chitosan (CHC), or eggshell powder (CHEG) modifications. The physical and chemical properties of all adsorbents were characterized using Brunauer-Emmett-Teller (BET) surface analysis, Fourier transform infrared spectroscopy, scanning electron microscopy, pHzpc, iodine number and elemental analysis. The adsorption of ammonium (NH4+), nitrate (NO3−), and phosphate (PO43−) in single and mixed solute solutions was investigated for initial concentrations of 10 mg L− 1. Langmuir, Freundlich, Sips, Dubinin-Radushkevich (D-R) and BET isotherm models were used to investigate the adsorption mechanisms. The maximum adsorption capacity of NH4+ on CH, CHC, and CHEG from mixed solute solution was 5.0, 4.7 and 5.9 mg g− 1, respectively, while the adsorption capacity of mixed:single solute solution was 0.95, 0.93, and 1.04, respectively. CH, CHC, and CHEG had greater ability to remove the cation NH4+ than anions NO3− and PO43− from aqueous solution. The highest maximum adsorption capacity for anions NO3− and PO43− was found on CHEG (1.7 mg g− 1) and CH (6.7 mg g− 1), respectively. NH4+ and NO3− were bound by chemisorption as indicated by D-R isotherm E values (> 8 kJ mol− 1), and enthalpy ∆H values (> 80 kJ mol− 1). In contrast, PO43− adsorption was mainly by physical interaction, including pore-filling, and electrostatic attraction. Pseudo first order and pseudo second order models provided good fits of the sorption kinetics data (R2 > 0.9). The initial concentrations of NH4+, NO3−, and PO43− in surface water sampled from a canal in Bangkok were 10.4, 1.2, and 3.9 mg L− 1, respectively, which indicated eutrophication. At a dose of 20 g L− 1, CHC achieved the best nutrient removal from this surface water, by 24% for NH4+, 25% for NO3−, and 66% for PO43− after 48 h contact, respectively.


Introduction
Nitrogen and phosphorus are the main elements causing surface water eutrophication. Eutrophication occurs when the total N and P concentrations in surface water exceed 1.5 mg-N L − 1 and 0.75 mg-P L − 1 , respectively [1]. Eutrophication stimulates aquatic plant and algae growth, leading to an imbalanced aquatic ecosystem, oxygen depletion, and production of cyanotoxins which pose a risk to human and ecosystem health. Inorganic nutrients exist mainly in the forms of ammonium (NH 4 ). For combinations of these ions, it was found that NH 4 + with PO 4 3− has greater effect on phytoplankton growth than NO 3 − with PO 4 3− [2]. The primary source of nutrient pollution in water resources is untreated domestic

Open Access
Sustainable Environment Research *Correspondence: anawat.pin@mail.kmutt.ac.th wastewater and runoff from agriculture, affecting fisheries and raw water used for aquaculture [3]. Also, water supply is often based on raw water abstraction from a river, pond, lake, reservoir, or groundwater, and high nitrate levels in drinking water can disturb oxygen transport in the bloodstream, consequently affecting infants at risk for methemoglobinemia, or blue baby syndrome [4]. The allowable NO 3 − concentration for drinking water set by the World Health Organization (WHO) is 50 mg L − 1 equivalent to 11.3 mg N L − 1 . In a nutrient-rich environment, cyanobacteria can quickly create algae blooms that release cyanotoxins into water. The occurrence of cyanotoxins is another threat to drinking water resources, and the WHO has set a preliminary guideline for microcystin in drinking water of 1.0 μg L − 1 [5].
Conventional water treatment uses a coagulation/ clarification process which is not effective in removing inorganic nutrients. Activated carbon (AC) filtration processes are also widely used for water purification, but AC is a high-cost adsorbent. Alternatively, biochar from agriculture residues has recently been promoted as a low-cost adsorbent and waste valorization opportunity in a circular economy [6]. Biochar can be produced by thermal conversion of biomass under oxygen-limited conditions. Biochar is an effective adsorbent to bind contaminants in soil and water [7] due to the porous and surface properties of biochar which facilitate pollutant adsorption. The application of biochar to adsorb pollutants from the water is attractive and the numbers of related research studies is rapidly increasing [7][8][9]. Biochar and modified biochar are effective materials to adsorb NH 4 + [10,11]. Typically, biochar has a negative charge on the surface which effectively adsorbs NH 4 + , but negligibly adsorbs NO 3 − and PO 4 3− because of charge repulsion. This deficiency can be addressed with biochar modifications. Biochar composites with eggshell powder [12] or Mg-Al [13] have the potential to adsorb NO 3 − . Also, chicken-eggshell contains calcium and magnesium cations and has capacity for NO 3 − adsorption at 2.8 mg g − 1 [12]. Modified biochars can also enhance PO 4 3− adsorption [13,14]. Chitosan is another environmentally friendly material that is available from agricultural waste like shrimp shells, and effectively adsorbs inorganic anions in water due to the amine functional groups (R-NH 3 + ). Zhao and Feng [15] reported that chitosan microspheres can remove NO 3 − and PO 4 3− with an adsorption capacity of 32.2 and 33.9 mg g − 1 , respectively. Chitosan coated biochar could thus potentially facilitate the NO 3 − and PO 4 3− adsorption from aqueous solution. Moreover, biochar with Mg and Ca contents has a great capacity for PO 4 3− adsorption, as reported by Almanassra et al. [16]. Biochar is typically used as an adsorbent in powdered form or small aggregates. However, powder is not suitable for adsorption filters as it clogs easily. Using biochar pellets as media in the filtration process for wastewater treatment would reduce such detrimental head loss. Pelletization of biochar can be done by mixing biochar with a binder such as starch, molasses, rubberwood, or chemical reagents. Hu et al. [17] found that starch could be used as a binder with good hydrophobicity but lower mechanical strength than NaOH. Aransiola et al. [18] found that using cassava starch, corn starch or gelatin at 10-30% (w/w) as binders produced biochar briquettes of storable stability. Therefore, using a starch solution with a pellet machine is a promising technology for biochar pellets production.
In most previous studies, adsorption studies were carried out as single solute batch adsorption experiments, which are difficult to extrapolate to field conditions [9]. Many pollutants and ions coexist in the water system, significantly influencing the equilibrium adsorption capacity [8]. Furthermore, most nutrient adsorption studies were conducted with high initial concentrations that are unrealistic in environmental remediation applications. For applications of biochar to remove nutrients from surface water, lower initial nutrient concentrations should be used for adsorption studies.
The aims of this study were: (i) to investigate the properties of biochar pellets produced from coconut husk biochar with and without surface modification; (ii) to establish the adsorption mechanisms of the three key nutrients (NH 4 + , NO 3 − , and PO 4 3− ) by these biochar pellets in single solute and mixed solute solutions with environmentally relevant initial concentrations; (iii) to investigate the effect of initial pH on nutrient adsorption; and (iv) to demonstrate the application of these biochar pellets to the treatment of nutrient polluted surface water.

Materials and methods
Coconut husk was obtained from central Thailand and carbonized to biochar by a community 200-L drum kiln method at a temperature of around 400 °C [19]. The carbonized husks were crushed to powder and used for making three types of biochar pellets as summarized in Fig. 1.

Biochar and pelletization
Commercial corn flour (McGarrett) was used as a binding agent for biochar pelletization. To make a gelatinized starch, 10 g of corn flour was thoroughly mixed with 100 mL deionized (DI) water and boiled until the solution was a viscous liquid. The liquid was mixed with 50 g of coconut husk biochar powder and agitated until a homogeneous mixture was observed. The mixture was processed by a pellet mill with diameter of 4 mm, the pellets were cut into 5-10 mm lengths and dried in the oven (AS ONE, OFW-600B) at 60 °C for 24 h. The coconut husk biochar pellets (CH) were kept in a plastic bottle in a dry place and used for all experiments.

Chitosan coating of biochar pellets
Shrimp chitosan powder provided from Biolife ELAND, Thailand, was used to coat CH pellets (CHC). Five grams of chitosan powder were placed in a 250 mL glass bottle containing 500 mL of 1% (v/v) acetic acid solution. Then the mixture was shaken in an incubator shaker (New Brunswick, Innova 42/42R) at 100 rpm and 30 °C for 12 h. After addition of 25 g of CH pellets the chitosan solution was continuously shaken at 100 rpm and 30 °C for 24 h. The CHC pellets were placed on filters and rinsed with DI water until excess chitosan had been removed, then dried in the oven at 60 °C for 24 h. CHC was kept in a plastic bottle in a dry place and used for all experiments.

Pellets of biochar mixed with eggshell powder
Chicken eggshell was cleaned and crushed into powder. The coconut husk biochar powder was mixed with eggshell powder in a ratio of 1:1 (w/w), then the mixture was pelletized using corn flour as a binding agent followed the steps described above (2.1). Pelletized biochar mixed with eggshell powder (CHEG) was kept in a plastic bottle at a dry place and used for all experiments.

Characterization of biochar pellets
The C, H, N, and S contents of biochar pellets were analyzed using a Perkin Elmer 2400-II CHNS elemental analyzer. Oxygen content was obtained by calculation as 100 -%C -%H -%N -%S. The specific surface area was determined using Brunauer-Emmett-Teller (BET) nitrogen gas adsorption-desorption isotherms at 77 K with a surface analyzer (Autosorb-1, Quantachrome, BEL model, USA). The pore size distribution was measured from the nitrogen gas desorption using a Barrett-Joyner-Halenda analyzer. The iodine number was measured as a relative indicator of porosity following the ASTM D4607  method. Surface morphology images were analyzed by scanning electron microscopy (SEM, BRUKER). Fourier Transform Infrared (FTIR) spectroscopic analysis of biochar pellets indicated the functional groups on the surfaces as observed by a Thermo Scientific Nicolet 6700 spectrometer. The ions released from the biochar pellets were determined after shaking 20 g L − 1 of biochar in DI water at 120 rpm and 30 °C for 48 h. Then, the mixture was filtered through 0.45 μm membrane filters (Sartorius) before measuring Na + , NH 4 4 3− and SO 4 2− ions by Ion Chromatography (761 Compact IC, Metrohm; column Metrosep A Supp5-150/4.0 at flow rate of 0.7 mL min − 1 for anion analysis and Metrosep C4-100/4.0 at flow rate of 0.9 mL min − 1 for cation analysis).
The pH at zero-point of charge (pH zpc ) was analyzed using the pH drift method with a pH meter (WTW, pH 3210). Biochars (0.15 g) were placed in 125 mL flasks containing 50 mL of 0.01 N NaCl solution adjusted to different pH values of 1.0, 3.0, 5.0, 7.0, 9.0 and 11.0 using NaOH and HCl. The mixture was shaken at 120 rpm and 30 °C for 48 h, before measuring the final pH. The pH zpc was where the initial pH equaled the final pH.

Batch experiments
Batch experiments were conducted to study the adsorption capacities of biochar pellets to remove nutrients in single solute and mixed solute systems. Stock solutions of nutrients at 1000 mg L − 1 concentration were prepared using analytical grade ammonium chloride; NH 4 Cl (Qrec), potassium nitrate; KNO 3 (Ajax Finechem), and potassium dihydrogen phosphate; KH 2 PO 4 (Ajax Finechem). Then, the initial concentration of each nutrient was prepared by dilution of the stock solutions to 10 mg L − 1 both in single and mixed solute solutions. All solutions were adjusted to pH 7 ± 0.5 using HCl and NaOH.
All biochar adsorbents (CH, CHC, and CHEG) were sterilized to prevent microbial activities in nutrient adsorption studies by using an autoclave (121 °C for 15 min) and then dried in an oven at 60 °C. The sterilized adsorbents were kept in sterilized and cleaned bottles in a dry place before being used for batch experiments. All experiments were done in triplicates.
To study the effect of the contact time of adsorption, 20 g L − 1 of sterilized CH, CHC, and CHEG pellets were added into 50 mL centrifugal tubes containing each single and mixed solute solution. Control experiments were conducted using no adsorbate and no adsorbent. The mixtures were shaken and the water samples were collected periodically until reaching sorption equilibrium. The suspensions were filtered using sterilized 0.45 μm membranes (Satorius) before measuring the final pH, and NH 4 + , NO 3 − , and PO 4 3− concentrations by ion chromatography. The average aqueous concentration values were determined and the amount of adsorbate on the biochar, q t (mg g − 1 ), was calculated using Eq. (1). , and PO 4 3− released by biochar into DI water for the same condition of the nutrient adsorption study. The adsorption kinetics of nutrients in each single and mixed solute solutions were determined. The result was fitted with pseudo first order and pseudo second order kinetic models.

Adsorption isotherm study
Each single solute of NH 4 + , NO 3 − , and PO 4 3− solution was prepared at concentrations ranging from 10 to 100 mg L − 1 . A mixed solution with NH 4 + , NO 3 − , and PO 4 3− each at concentrations ranging from 10 to 100 mg L − 1 was also prepared. All solutions were adjusted to pH 7 ± 0.5 using HCl and NaOH. Sterilized CH and CHC pellets at a dosage of 20 g L − 1 were used in this study. The solutions were shaken at 120 rpm and 30 °C for 48 h. Then, the final pH and concentration of NH 4 + , NO 3 − and PO 4 3− was measured for the filtrates. All experiments were done in triplicates. The average values were calculated and fit with typically monolayer and multilayer adsorption isotherm models, which were Langmuir, Freundlich, Sips, Dubinin-Radushkevich (D-R), and BET models according to Eqs. (2)-(6), respectively.
where, C e is the equilibrium concentrations (mg L − 1 ) and q e the amount of nutrient adsorbed (mg g − 1 ) at BET; q e = q BET K 1 C e (1 − K 2 C e )(1 − K 2 C e + K 1 C e ) equilibrium, with q max being the maximum capacity (mg g − 1 ). K L is the Langmuir constant (L mg − 1 ), K f (mg 1-1/nf L 1/nf g − 1 ) the Freundlich constant, and 1/n f (dimensionless) the Freundlich exponent. K s (L ns mg -ns ) and n s (dimensionless) are Sips constants. The ε value of the Dubinin-Radushkevich model can be calculated as and T is the temperature (K). C is a dimensionless concentration ratio C e,molar /C°, where C° is the standard state of the solute in aqueous solution equal to 1 M and C e,molar is the equilibrium concentration in units of M [20]. β is the Dubinin-Radushkevich constant (mol 2 J − 2 ). K 1 and K 2 value are BET constants (L mg − 1 ).

Effect of initial pH
To study the effect of initial pH, each of the sterilized adsorbents at a dosage of 20 g L − 1 was placed into 50 mL centrifuge tubes containing a single or mixed solute solution with different initial pH values of 3, 5, 7, 9, and 11 (initial concentration of 10 mg L − 1 ). The mixtures were shaken for 48 h at 120 rpm and 30 °C. Then, the filtrates were used to determine the final pH and concentrations of NH 4

Effect of temperature
To study the effect of temperature, the dosage of each biochar was prepared as above. The initial concentration of single solute solution was 10-100 mg L − 1 (initial pH of 7 . The Gibbs free energy of adsorption (∆G) is given by Eq. (7), The relationship of ∆G to the change of enthalpy (ΔH) and entropy (ΔS) of adsorption can be expressed as Eq. (8).
where the K a is the dimensionless thermodynamic equilibrium constant [21,22] and T is the temperature (K). Then, K a can be estimated from the Langmuir isotherm as K a = (K L,molar / γ e ) × C s , which K L,molar is the Langmuir constant in units of M − 1 , C s is the standard reference solute concentration in aqueous solution, which is equal to 1 M, and γ e is activity coefficient at adsorption equilibrium. γ e is a function of the ionic strength (I e ) of the aqueous solution which depends on the concentration and the charge carried by the dissolved ions (z) according to log γ e = − Az 2 I e 1/2 [21]. For neutral adsorbates or adsorbates with weak charges γ e → 1 [21]. In our experiments the γ e value was calculated to be in the range of 0.75-0.98, and therefore K a was approximated by K a = K L, molar × C s . The change of enthalpy (ΔH) and entropy (ΔS) can then be determined by the slope and intercept of the plot of lnK a versus 1/T.

Effect of surface water on nutrient adsorption
Surface water from the Khwang canal in a residential area of Bangkok, Thailand, was collected and sterilized in an autoclave (121 °C for 15 min). Each of the adsorbents at 20 g L − 1 was placed in 50 mL centrifugal tubes containing sterilized surface water. The control was using surface water without biochar addition. The mixtures were shaken at 120 ppm and 30 °C for equilibrium time 48 h (same condition as 2.

Characteristics of biochar and modified pellets
The properties of CH, CHC, and CHEG are compared in Table 1, which shows that the surfaces modification of biochar enhances the properties of the CH pellets. The iodine number of CHC was higher than that of CHEG and CH pellets, which corresponded to the trend for the BET surfaces. The low BET surface area of CH was due to the high mineral content of alkaline biochar that can block access to the micropores [23]. Modification of CH biochar gave more potential for adsorption. The mean pore diameter of CH, CHC, and CHEG indicated mesopores with diameters of 2-50 nm. The mean pore diameter was CHC (7.5 nm) > CH (6.3 nm) > CHEG (3.3 nm). Eggshell powder might have blocked the pores, decreasing total pore volume and diameter. The pH of CH, CHC, and CHEG in DI water (pH DI ) was 7.21, 6.10, and 8.10, respectively. CHEG was moderately alkaline from the CaO and CaCO 3 in eggshell powder, since raw eggshell consists of 98.5% CaO [24]. CHC was weakly acidic due to the acetic acid used for chitosan coating.
The pH values in aqueous solution of all adsorbents are greater than the pH zpc , thus the adsorbent surfaces exhibit a net negative charge. The atomic ratios of H/C and O/C indicate the aromaticity and polarity of biochar, respectively. In this study, the order of aromaticity was CH > CHEG > CHC, and the order of polarity was CHEG > CHC > CH. The higher polarity was caused by more oxygen-containing functional groups and surface hydrophilicity which improves the removal of inorganics [25]. The O/C ratio of CHEG had a 2-time increase compared with the original CH pellet ( Table 1), indicating that CHEG had more oxygen-containing groups which promote the adsorption potential.
Biochar has functional groups on the carbonaceous surface such as hydroxylic, carboxylic, or phenolic moieties [26]. FTIR spectra of CH, CHEG, CHC, chitosan powder, and eggshell powder are shown in Fig. 2. The approximate peak at 3400 cm − 1 was ascribed to -OH groups and -NH stretching vibrations [15]. The band at 2900 cm − 1 was attributed to O-H and aliphatic C-H stretching, which indicated hemicellulose and cellulose structures [12]. CH, CHC, and CHEG showed a peak at 1740 cm − 1 ascribed to C=O stretching. CHC had an acidic surface (pH = 6.1) as indicated by acidic functional group peaks of 1740 and 1370 cm − 1 , representing carboxylic and phenolic groups [19,25]. The peak at 1585 and 1215 cm − 1 in CHC and chitosan (Fig. 2a) showed that CHC had chitosan coated on the surface, indicating the amine group. The sharp peaks of 712, 872, and 1400 cm − 1 in eggshell and CHEG (Fig. 2b) were ascribed to Ca-O bonding, C-O stretching and C-H bonding, respectively, with the presence of CaO and CaCO 3 [7,12].
SEM images show the morphologies of CH, CHC, and CHEG biochar in Fig. 3. Chitosan coating might be covering some surface area of CHC (Fig. 3b) which by and large retained the original biochar surface features. The CH SEM image also resembled that of CHEG. CHEG contained eggshell powder, and some eggshell fragments are indicated in Fig. 3c.
Inorganic components in biochar are dependent on the feed stocks. Figure 4 shows the concentration of anions and cations released from biochar into DI water. CH released ion concentrations greater than CHEG and CHC. The average total ion concentrations from CH, CHC, and CHEG were 52.6, 7.2 and 20.1 mg g − 1 , respectively. High alkaline mineral contents of biochar may block biochar pores which was confirmed by low BET surface values. The order of ions released from CH was According to Khawkomol et al. [1 9] th e content of the metal components of coconut husk biochar was K > Cl > Na > Ca > Mg > P. The ions released from biochar could compete with nutrients for adsorption sites. Hence, these anions and cations may affect the nutrient adsorption.

Nutrient sorption and kinetic
The adsorption kinetics of NH 4 + , NO 3 − , and PO 4 3− in single and mixed solute solutions were investigated. The mixed solution system is more similar to actual water and wastewater systems, which consist of various ions including counter ions (Cl − and K + ). The various ions in water or wastewater play an important role as competitors with the nutrients NH 4 + , NO 3 − , and PO 4 3− for adsorption sites and consequently decrease the adsorption efficiency for the individual ions by biochar [9]. An initial concentration of 10 mg L − 1 with initial pH of 7 was used in this study due to the existence of nutrients at these typical concentration levels in surface water. The result showed that, contrary to the initial expectation, NH 4 + , NO 3 − , and PO 4 3− adsorption in single solute solution was less than for mixed solute solution (Fig. 5).
The adsorption kinetics of NH 4 + in single and mixed solute solutions of CH, CHC, and CHEG are shown in Fig. 5a [12]. Although the biochar typically already contains Ca 2+ , CHEG showed NO 3 − sorption better than CH (Fig. 5c-d). This result showed how Ca 2+ from eggshell powder can promote NO 3 − removal. The sorption The initial PO 4 3− concentration was 10 mg L − 1 , but the CH, CHC, and CHEG also released PO 4 3− at 1.9, 0.5, and 0.3 mg g − 1 , respectively into DI water. Thus, low adsorption capacities of all adsorbents were observed in single solute solution. Zhang et al. [9] also reported low PO 4 3− adsorption at low initial PO 4 3− concentration. At high concentration, steep gradients between biochar and the bulk solution result in better filling of reactive adsorption sites [27]. CH, CHC, and CHEG effectively adsorbed PO 4 3− from the mixed solute solution. The pseudo first order and pseudo second order kinetic models were used to fit the adsorption data. The calculated kinetic parameters are given in  4 3− adsorption on CH, CHC, and CHEG from single and mixed solute solutions the R 2 value was more than 0.9 except for the NO 3 − adsorption on CH in single solute solution with an R 2 value of 0.83. NO 3 − adsorption on CHC was well fit by the pseudo second order model in single and mixed solute solutions, while CH and CHEG data were best fit by the pseudo first order model. Based on R 2 values, the pseudo second order model fitted better than the pseudo first order model for PO 4 3− adsorption from mixed solute solution. Biochar pellets have heterogeneous surfaces with the proposed mechanism for PO 4 3− being that it is not adsorbed directly onto the biochar surface [28]. On the other hand, PO 4 3− could precipitate with metal ions (Ca 2+ and Mg 2+ ) out of solution and onto biochar surfaces [29]. However, the PO 4 3− adsorption kinetics of CH, CHC, and CHEG in single solute solution could not be determined in this study due to the strong compounding effect of PO 4 3− released from biochar.
The rate of PO 4 3− adsorption rates onto CH was highest, followed by NO 3 − and NH 4 + , respectively, with k  ).

Nutrient sorption isotherm
The NH 4 + , NO 3 − , and PO 4 3− adsorption capacities increased in a nonlinear way as the initial concentration increased both in single and mixed solute solutions (Fig. 6). Table 3 shows the adsorption parameters of each adsorbent using various adsorption isotherm models. The NH 4 + adsorption capacities for the investigated concentrations were in the ranges of 0.2-2.2, 0.07-1.3 and 0.3-2.2 mg g − 1 for CH, CHC, and CHEG, respectively (Fig. 6a). Nitrate adsorption capacities in single and mixed solute solutions of CH, CHC, and CHEG were in the ranges of 0.01-1.6, 0.2-1.4, and 0.2-3.5 mg g − 1 , respectively. It was found that PO 4 3− was released from adsorbents if the initial PO 4 3− concentration was lower than 10 mg L − 1 (Fig. 6c). The amount of PO 4 3− adsorbed increased as the initial PO 4 3− concentration increased, while the amount of PO 4 3− released from biochar would remain the same [9]. PO 4 3− adsorption on biochar increased with high initial concentration possibly due to higher concentration gradients, when PO 4 3− is better at filling active sites [27].
The equilibrium isotherms could be classified as S or L curves (Fig. 6). Ammonium adsorption on CH and CHEG in single and mixed solute solution was classified as L curves (type I) with a plateau that represented the maximum adsorption capacity as predicted by the Langmuir model. The curve of NH 4 + adsorption on CHC (Fig. 6a) was classified as S curve (type VI) which represented a shorter plateau. This indicated that the solutes interacted with each other on the biochar surface resulting in multilayer adsorption [30]. A similar trend was observed for NO 3 − adsorption as single solute on CH (Fig. 6b). NO 3 − adsorption on CHC reached the maximum capacity of 1.6 and 1.3 mg g − 1 for the single and mixed solute solution, respectively. NO 3 − in mixed solution also reached a maximum capacity for CHEG (0.9 mg g − 1 ). However, PO 4 3− adsorption on CH was described by a type III (concave upward curve) adsorption isotherm, meaning that the adsorbate-adsorbate interaction was more significant compared to adsorbate-sorbent interactions [31]. Furthermore, PO 4 3− adsorption on CHC and CHEG from single solute solution was described by a type V isotherm which is similar to type III, with adsorbate-adsorbate interactions being dominant, and filling of mesopores could have occurred [31].
The Langmuir model indicates monolayer adsorption on a homogenous surface, while the Freundlich  − adsorption on CH, CHC, and CHEG was lower than 1 except for CH in single solute solution (1/n f > 1). The CH, CHC, and CHEG was unfavorable to adsorb PO 4 3− in single solute solution (1/n f > 1) while CHC and CHEG were favorable to adsorb PO 4 3− in mixed solute solution (1/n f < 1). It was reported that PO 4 3− adsorption is not by direct interaction with the carbon surfaces but by precipitation with Ca 2+ [29]. Eggshell has high calcium content which could precipitate with PO 4 3− in CHEG and PO 4 3− could also be attracted by chitosan surfaces (CHC). However, based on other research, biochar  is an effective adsorbent for PO 4 3− even without modification [9,13].
Furthermore, the D-R adsorption isotherm is applicable for heterogeneous surfaces, describing the physisorption and chemisorption of adsorbates on biochar by calculation of the mean free energy (E = (1/2β) 1/2 ) [30,33]. An E value less than 8 kJ mol − 1 indicates that adsorption is predominantly by physical interactions, while E value higher than 8 kJ mol − 1 signifies chemisorption. It was found the mechanism of NH 4 + and NO 3 − adsorption on CH and CHEG in single and mixed solutes indicated chemical adsorption. The E values were in the range of 8-16 kJ mol − 1 . CHC showed physical interaction with NH 4 + and chemical interaction with NO 3 − . PO 4 3− adsorption in this study was thus mainly physical interactions (E < 8 kJ mol − 1 ), i.e., pore filling or electrostatic attraction. The prediction of q max values using the D-R isotherm was unreliable for PO 4 3− adsorption in this study (Table 3), due to the measured data not reaching the anticipated plateau in the curve (Fig. 6c). It should also be noted that the D-R isotherm might not always reliably distinguish between physical or chemical adsorption in complex solid/liquid adsorption systems and is mainly used for curve fitting and parameter value predictions [33].
The three-parameter Sips isotherm model is well suited to predict adsorption on heterogeneous surfaces and identify adsorption without adsorbate-adsorbate interactions as it combines the Langmuir and Freundlich isotherm models [34]. Adsorption then follows the Freundlich model at low initial concentration and the Langmuir model at high initial concentration. Therefore, the Sips model is used to describe monolayer adsorption. A n s value of the Sips isotherm between 0 and 1 indicates heterogeneous surfaces while an n s value close to 1 indicates more homogeneous surfaces [34]. Accordingly, CH, CHC, and CHEG had heterogeneous surfaces unfavorable for PO 4 3− adsorption both in single and mixed solute solution studies (n s > > 1) except CHEG in mixed solute solution (n s = 1.0).
The BET model can describe multilayer adsorption behavior [31,34]. It has the same assumptions as the Langmuir model, but with multiple adsorption layers having different adsorption energy [31]. NH 4 + and NO 3 − adsorption studies in single and mixed solute solution of CH, CHC, and CHEG indicated multiple layers of adsorption since the data were well fitted with the BET isotherm (R 2 > 0.93).

Effect of initial pH on adsorption
The initial pH of the solution affects the adsorption mechanism by determining the surface charge of an adsorbent. Figure 7 shows the NH 4 The NH 4 + adsorption capacity increased when the initial pH increased from 3 to 7. However, the NH 4 + sorption capacity decreased when the initial pH value rose from 9 to 11. This result was confirmed by other researchers [10,35,36]. Biochar with high negative zeta potential facilitates NH 4 + adsorption through electrostatic attraction, but when pH is increased to 10, the NH 4 + cation would be transformed into NH 3 in the aqueous phase [37]. This study confirmed that the adsorption capacity decreased when pH increased towards 11 in closed batch systems which indicates ammonia stripping was not the mechanism for NH 4 + removal, similar to the findings of Vu et al. [38]. The NH 4 + adsorption by CH, CHC, and CHEG from mixed solute solution was better than for single solute solution. The highest NH 4 + sorption capacities of CH, CHC, and CHEG were 0.41, 0.37, and 0.40 mg g − 1 , respectively, in a single solute solution and they were 0.63, 0.69, and 0.66 mg g − 1 in mixed solute solution at the initial pH in the range between 7 and 11. The high amount of H + in the lower pH range might compete with NH 4 + for adsorption sites. At low pH electrostatic repulsion occurs between cations and the positive charge of the protonated biochar surfaces [8].
The NO 3 − adsorption on CH, CHC, and CHEG from mixed solute solution was higher than from single solute solution for initial pH 3-11 except for CHC and CHEG at initial pH 11. The highest NO 3 − sorption capacity of CH was found at initial pH 3, being 0.31 and 0.61 mg g − 1 in single and mixed solute solution. At low pH, H + attachment to the biochar surface can attract NO 3 − by electrostatic charge and facilitate the removal of NO 3 − in solution. However, when the initial pH was acidic following HCl addition, the Cl − concentration would have increased in the solution. Cl − can compete with NO 3 − for the attractive sites on the adsorbent surface [12]. The highest NO 3 − sorption capacity of CHC was found at initial pH 5 and 7 of 0.52 and 0.62 mg g − 1 in single and mixed solute solution, respectively. The highest NO 3 − sorption capacity of CHEG was found at initial pH 11 and 3 of 0.60 and 0.65 mg g − 1 in single and mixed solute solution, respectively. At alkaline pH (pH = 11) the eggshell fragments in CHEG significantly adsorbed NO 3 − in solution.
Depending , where pK 1 , pK 2 , pK 3 values are 2.13, 7.20, and 12.33 [16]. The PO 4 3− adsorption of all adsorbents in mixed solute solution increased compared with single solute solution at initial pH of 3-11. An initial pH of 3 represented the highest PO 4 3− sorption capacity of CH, CHC, and CHEG in mixed solute solution. CH had the highest adsorption capacity of 1.9 mg g − 1 followed by CHC (1.1 mg g − 1 ) and CHEG (0.9 mg g − 1 ). The H + ions at acidic conditions can protonate the active sites (positive charge) of the biochar surface to adsorb PO 4 3− . The R-NH 2 amino functional group of chitosan on CHC is protonated (R-NH 3 + ) when pH is low which enhances the PO 4 3− adsorption by electrostatic attraction [15]. However, the highest PO 4 3− sorption capacity in single solute solution of CH, CHC, and CHEG was found at an initial pH of 11. It was 0.44, 0.37, and 0.18 mg g − 1 of CH,  are the predominant species at pH solution of 7-9, and these species may also form complexes with Ca 2+ and Mg 2+ but precipitation of phosphate and Ca 2+ is unlikely to occur at a pH of 6.5 [29].

Effect of temperature
Biochar strongly affected PO 4 3− the adsorption because it released PO 4 3− into the solution especially at low initial concentration, and the measured isotherm data did not reach a plateau at high concentration and was therefore poorly fitted by the Langmuir adsorption model. Thus, NH 4 + and NO 3 − were selected to investigate thermodynamics in this study. The thermodynamic parameters of nutrient adsorption in single solute solution are shown in Table 4. The negative ∆G value of NH 4 + and NO 3 − adsorption on CH, CHC, and CHEG indicated favorable and spontaneous processes. However, it is noted that the batch adsorption study was conducted in an incubator shaker, and the agitation of the shaker might have enhanced nutrient adsorption. Moreover, the trend of more negative ∆G value with increasing temperature for CH, CHC and CHEG means that nutrient adsorption improves at higher temperature. It suggests sufficient energy is available to transport ions into the inner biochar pores [10]. Depending on the biochar and solution, the results showed endothermic processes (positive ΔH values) of NH 4 + and NO 3 − adsorption. Only NH 4 + adsorption on CH showed exothermic processes. Furthermore, ΔH can indicate the interactions between the adsorbent and adsorbate. For physical adsorption, such as van der Waals interactions, ΔH is usually lower than 20 kJ mol − 1 and from 20 to 80 kJ mol − 1 it shows electrostatic interaction [30]. A ΔH value from 80 to 450 kJ mol − 1 indicates chemical bonding [29]. Apart from CH, ΔH values were higher than 80 kJ mol − 1 , indicating the bonding between adsorbate and biochar surface. For CH, NH 4 + had a negative ΔH value, while for NO 3 − physisorption was indicated, i.e., van der Waals interactions (ΔH = 7.8 kJ mol − 1 ). CHC showed the highest ΔH value on NO 3 − adsorption of 246 kJ mol − 1 . It means chemical bonding between chitosan (R-NH 2 + ) and NO 3 − occurred. Moreover, the eggshell in CHEG can also react with NO 3 − . The positive values of ΔS showed randomness at the solid-liquid interface during adsorption, which is due to the presence of non-pyrolyzed biochar from the biochar production.
Based on the results from the thermodynamic analysis, CHC and CHEG can strongly adsorb NH 4 + and NO 3 − in actual water and wastewater treatment applications as indicated by negative free energy and enthalpy values of the adsorption.

Nutrient adsorption mechanisms
The original and modified biochar surfaces contain different characteristics, which consequently affected the adsorption capacity. It was found that nutrient adsorption capacity at equilibrium in mixed solute solution was greater than in single solute solution (Fig. 7). In contrast, Yin et al. [13] found adsorption of NH 4 + , NO 3 − , and PO 4 3− on soybean straw biochar in tri-solute solutions was decreased compared to single solute solutions. Different types of biochars have different physical and chemical properties that affect nutrient adsorption. The nutrient adsorption capacity of different adsorbents is summarized in Table 5. A diagram of nutrient adsorption mechanisms of CH, CHC, and CHEG in mixed solute solution is shown in Fig. 8 ions in solution can readily migrate into the pores of biochar. Although PO 4 3− can access the pores, the phosphate leaching from biochar is also significant, especially for CH. For PO 4 3− adsorption in the single solute study, the high concentration gradient brought about a net release of phosphate ions instead of adsorption (Fig. 5e). However, PO 4 3− might also be removed from solution by other mechanisms.
According to the FTIR spectra of biochar in this study (Fig. 2), there are oxygen containing groups including hydroxyl and carboxyl on biochar surfaces. These functional groups on the biochar can be protonated or deprotonated depending on the solution pH, as shown in Eq. (10). Fan et al. [10] reported that -O − was typically the dominant oxygen species on biochar surfaces. CH, CHC, and CHEG had a net negative charge on their surfaces at neutral pH (pH solution > pH zpc ) which enabled strong interactions with NH 4 + in solution following Eqs.
The Ca 2+ and Mg 2+ content in biochar could have facilitated precipitation or surface deposition as mechanisms for PO 4 3− adsorption in the form of CaHPO 4 and MgHPO 4 .

Nutrient removal from surface water
The existence of various ions and other solutes in surface water could affect nutrient adsorption by competition for the surface adsorption sites. To avoid microbial activities during adsorption, surface water with high nutrient content from a canal in central Bangkok was sterilized. The average initial concentration of sterilized surface water was 10 PO 4 3− L − 1 with an initial pH = 7.9. It was found that the final pH after 48 h of CH, CHC, and CHEG adsorption was slightly decreased to 7.7, 7.2, and 7.7, respectively. CH, CHC, and CHEG at 20 g L − 1 dosage removed NH 4 + by 33, 24, and 37%, respectively (Fig. 9). NO 3 − could not be removed by CH and CHEG due to the strong repulsion effect of the negatively charged surfaces. However, the efficiency of NO 3 − removal by CHC from sterilized surface water was 25%. Coexisting ions in surface water will affect biochar adsorption. For example, NO 2 − and NO 3 − ions could compete on the biochar surface due to NO 2 − adsorption capacity being higher than NO 3 − in binary (NO 2 − + NO 3 − ) solution [39]. PO 4 3− removal could not be achieved with CH, while CHC and CHEG effectively removed PO 4 3− by 66 and 58%, respectively. The R-NH 3 + functional group on CHC was thus confirmed to facilitate NO 3 − and PO 4 3− removal by electrostatic attraction in solution. In addition, calcium is a main component of CHEG, and the precipitation with Ca 2+ out of solution is likely the main mechanism of PO 4 3− removal, which can explain greater PO 4 3− than NO 3 − removal. Overall, CHC was the best adsorbent for simultaneously treating NH 4 + , NO 3 − , and PO 4 3− pollution in surface water. This removal of nutrients in this study was by the adsorption process only without biodegradation and biosorption. Therefore, the influence of microbial attachment to biochar surfaces should be investigated in further studies to optimize nutrient removal.  ) adsorption, affecting charge of surfaces and functional groups of the biochar. The adsorption mechanism of nutrient adsorption could be well described with the Sips isotherm model, which at low initial concentration follows the Freundlich model and at high initial concentration follows the Langmuir model. The nutrient adsorption mechanisms showed both physisorption and chemisorption. PO 4 3− adsorption was predominantly physical interaction, including pore-filling and electrostatic attraction by charge and functional group on biochar surfaces. Besides, chemical bonding can happen in NH 4 + and NO 3 − adsorption on biochar surface. CH, CHC, and CHEG were able to adsorb NH 4 + in surface water which was present at of low initial concentration. The surface modification of biochar as CHC and CHEG enhanced the adsorption of NO 3 − and PO 4 3− . On the other hand, the potential of biochar to treat water is compromised by the release of PO 4 3− from the biochar. The results show that biochar is removing PO 4 3− better at high initial concentration, which increased PO 4 3− adsorption. Furthermore, CHC showed the best capability to simultaneously adsorb all nutrients in this study (NH 4 + , NO 3 − , and PO 4 3− ) from real eutrophic surface water. Although CHC is acidic on the surface, the final pH value was approximately neutral (≈ 7.1), which is acceptable for water treatment. CHC has abundant and versatile functions on its surface. Therefore, CHC can be used as an adsorbent for the simultaneous treatment of the main inorganic nutrients in surface water. Pelletization enables use of this biochar for water filtration, and microbial attachment to the biochar should be investigated in future studies to enhance the nutrient removal in biofiltration processes.