Evaluation and Optimization of Azithromycin Removal by Raw and Alkali‐Modified Peanut Shells Using Taguchi‐Based Experimental Design Approach

TL;DR

This study shows that modified peanut shells can efficiently remove azithromycin from water, offering a sustainable and cost-effective solution.

Contribution

The novel use of alkali-modified peanut shells for azithromycin removal, optimized via the Taguchi method, is presented.

Findings

Modified peanut shells achieved 85% azithromycin removal under optimized conditions.

pH and initial concentration were the most influential factors in the adsorption process.

Surface analysis confirmed enhanced adsorption due to increased functional groups and electrostatic interactions.

Abstract

Advanced treatment methods for removing antibiotics are cost‐intensive. Subsequently, the goal of environmental and economic sustainability has switched attention towards bio‐adsorbents. This study evaluated the effectiveness of raw and alkali‐modified peanut shell powder as a cost‐effective, novel adsorbent for removing azithromycin, one of the most widely used drugs worldwide. Prepared adsorbents were characterized by FTIR and SEM equipped with EDX. Experiments designed using a Taguchi‐based approach were performed with a synthetic azithromycin solution to optimize initial concentrations, adsorbent dose, pH, and time. The results showed 63% removal with raw adsorbent at pH 11, an initial concentration of 20 mg/L, a time of 45 min, and an adsorbent dose of 0.4 g/L. With the modified adsorbent, an attractive 85% (maximum) removal was achieved at pH 11, an initial concentration of 30 mg/L, a time of 60 min, and an adsorbent dose of 0.4 g/L. Based on analysis of variance (ANOVA), pH and initial concentration are identified as the most influential factors for azithromycin removal. The improved adsorption performance of modified peanut shells (qmax = 192.1 mg/g compared to 159.2 mg/g for raw PS) was due to increased surface heterogeneity, enhanced electrostatic interactions, and greater accessibility of oxygen‐containing functional groups, as confirmed by kinetic, isotherm, and surface analysis.

The Taguchi approach optimizes adsorption parameters for enhanced removal.Modified peanut shell adsorbent removes up to 85% of azithromycin.π–π interaction and hydrogen bonding drive the adsorption mechanism.Chemical adsorption mechanisms are revealed through SEM and FTIR analysis.Bioadsorbents offer a sustainable solution to antibiotic removal.

The Taguchi approach optimizes adsorption parameters for enhanced removal.

Modified peanut shell adsorbent removes up to 85% of azithromycin.

π–π interaction and hydrogen bonding drive the adsorption mechanism.

Chemical adsorption mechanisms are revealed through SEM and FTIR analysis.

Bioadsorbents offer a sustainable solution to antibiotic removal.

This study demonstrates the optimization of azithromycin removal using raw and alkali‐modified peanut shell adsorbents via the Taguchi method, achieving up to 85% efficiency. The adsorption is governed by π–π interactions and hydrogen bonding, highlighting bio‐adsorbents as sustainable solutions for antibiotic removal from wastewater.

Figures

Chemical structure of azithromycin.

Calibration curve of azithromycin at λmax 208 nm.

SEM images of synthesized peanut shell (PS) adsorbents. (a) Raw PS. (b) Modified PS.

EDX spectrum with elemental composition. (a) Raw PS. (b) Modified PS.

FTIR spectrum. (a) Raw PS. (b) Modified PS.

Taguchi plots showing the effect of adsorption factors for azithromycin removal. (a) Raw PS. (b) Modified PS.

Percentage contribution of factors for the removal of azithromycin.

Interaction plots for RE (%) using data means.

Contour plots for RE (%) versus initial conc. versus adsorbent dose.

Plots for different kinetic models.

Plot of isotherm data for raw PS obtained at 20 mg/L initial concentration, a pH of 1, 45 min contact time, and 0.4 g/L adsorbent dose.

Plot of isotherm data for modified PS at C0 = 30 mg/L initial concentration, a pH of 11, 60 min contact time, and 0.4 g/L adsorbent dose.

Tables

TABLE 1. Azithromycin removal efficiency obtained by Taguchi‐based experimental runs.

Response
Experimental run	Initial concentration (mg/L)	pH	Time (min)	Adsorbent dose (g/L)	% RE (average ± SD)
					Raw PS	Modified PS
1	20	5	15	0.2	42.4 ± 1.5	55.8 ± 0.8
2	20	7	30	0.4	53.5 ± 0.7	68.9 ± 1.2
3	20	9	45	0.6	62.9 ± 0.9	77.6 ± 0.9
4	20	11	60	0.8	63.8 ± 1.2	78.4 ± 0.7
5	30	5	30	0.6	46.9 ± 2.0	56.9 ± 0.85
6	30	7	15	0.8	50.7 ± 0.4	66.3 ± 1.8
7	30	9	60	0.2	58.0 ± 2.1	81.7 ± 1.6
8	30	11	45	0.4	63.5 ± 2.3	82.5 ± 0.6
9	40	5	45	0.8	42.6 ± 0.9	55.0 ± 0.8
10	40	7	60	0.6	46.3 ± 1.7	64.2 ± 0.2
11	40	9	15	0.4	56.3 ± 0.7	72.2 ± 0.7
12	40	11	30	0.2	57.2 ± 0.9	75.1 ± 0.7
13	50	5	60	0.4	40.8 ± 1.2	57.2 ± 0.2
14	50	7	45	0.2	42.9 ± 1.9	64.7 ± 1.8
15	50	9	30	0.8	54.1 ± 2.1	71.1 ± 2.9
16	50	11	15	0.6	55.1 ± 0.8	73.6 ± 0.8

TABLE 2. Analysis of variance (ANOVA) for experimental data.

Source	p (raw PS)	p (modified PS)
Initial conc. (mg/L)	0.005	0.038
pH	0.000	0.001
Time (min)	0.198	0.134
Dose (g/L)	0.056	0.270

TABLE 3. Summary of parameters calculated for different kinetic models.

Kinetic models	Parameters	Raw PS	Modified PS
Pseudo first order	qe (mg/g)	32.40	62.01
	K1 (h−1)	5.28	5.56
	R 2	0.98	0.98
	RMSE	0.625	1.72
Pseudo second order	qe (mg/g)	36.32	69.60
	K2 (g/mg‐h)	0.22	0.114
	R 2	0.99	0.99
	RMSE	0.523	1.24

TABLE 4. Summary of parameters calculated for different isotherm models.

Isotherm model	Parameters	Raw PS	Modified PS
Langmuir model	KL (g/mg min−1)	0.037	0.074
	qmax (mg/g)	159.2	192.1
	RMSE	1.00	4.44
	R 2	0.99	0.94
Freundlich model	KF (mg/g)	9.92	18.37
	N	1.49	1.55
	RMSE	1.28	4.58
	R 2	0.989	0.94
Temkin	KT (L g−1)	36.8	67.3
	bT (kJ mol−1)	72.94	57.22
	RMSE	1.04	4.14
	R 2	0.99	0.95
D‐R	qmax (mg/g)	74.8	74.8
	β (mol2/J2)	8.3 × 10−6	1.43 × 10−6
	E (kJ/mol)	0.24	0.59
	RMSE	3.17	12.9
	R 2	0.95	0.57

TABLE 5. Comparison between the adsorption capacities of various adsorbents with studied adsorbents.

Antibiotics	Adsorbent	qmax (mg/g)	Exp/model	Conditions	Ref.
Norfloxacin	Rice husk	20.12	Langmuir	pH = 6.2, ad = 3 g/L, T = 298 K	Paredes‐Laverde et al. (2018)
	Coffee husk	33.56	Langmuir	—
Sulfamethoxazole	GOS	122	R‐P model	K = 298, pH = 6, ad = 0.01 mg	Rostamian and Behnejad (2016)
Tetracycline	PNS‐SO3H	303.0	Langmuir	pH 3.75, IC = 400 ppm	Islam et al. (2019)
Penicillin	Chestnut shell	100	Langmuir	pH = 3.0	Rostamian and Behnejad (2016)
Azithromycin	Azolla filiculoides –based activated porous carbon	374	Exp	T = 333 K, ad = 0.1 g/L	Balarak et al. (2021)
Azithromycin	Raw nano diatomite	68	Model	ad = 2.5 g/L, IC = 100 mg/L, pH = 9, T = 298 K	Davoodi et al. (2019)
	Saponin‐modified diatomite	91.7	Model	ad = 1 g/L, IC = 100 mg/L, pH = 9, T = 298 K
Azithromycin	Clinoptilolite	28.01	Model	ad = 5 g/L, pH = 11	Saadi et al. (2021)
Azithromycin	Peanut shell	159.2	Langmuir	pH = 11, ad = 0.4 g/L, IC = 30 mg/L, T = 298	This study
	NaOH PS	192.1	Langmuir

Full text

Introduction

1

Antibiotics have been considered an effective treatment for infectious diseases in humans and animals for several decades. These are used to treat diseases by inhibiting the growth of microorganisms (Sui et al. 2015; Wang et al. 2019). These therapeutic drugs enter the water cycle through diverse ways, including domestic and hospital sewage, sewage treatment facilities, and pharmaceutical industry effluents (Balarak et al. 2021). Recently, antibiotic discharge into the environment has become a global concern due to its potential risks (Yu et al. 2016). Among antibiotics, macrolides are commonly used to cure infectious diseases. Azithromycin (AZT) is a second‐generation macrolide derived from the first‐generation macrolide (erythromycin), with better pharmacokinetic properties (Landová and Vávrová 2017). During COVID‐19, azithromycin was also explored as a treatment for early‐stage symptoms of this infection (Schwartz and Suskind 2020). Macrolides are highly persistent, and their high percentage release into the environment occurs primarily through excretion in bile and feces (Sui et al. 2015; Singer et al. 2019). Six to twelve percent of azithromycin remains the parent drug in urinary excretions (Verdier et al. 2001). This results in significant loading of macrolides in hospital effluents and municipal wastewater treatment plants (WWTPs) (Kumar et al. 2019).

Conventional WWTPs are not specifically designed for antibiotic treatment, and some compounds, especially macrolides such as azithromycin metabolites, return to their parent compounds during primary and secondary treatments in conventional WWTPs (Kraemer et al. 2019; Pazda et al. 2020). Because of improper disposal and treatment of macrolides, they are found in surface water, sewage treatment plant effluents, drinking water, and soil ranging from ng/L to μg/L (Al‐Riyami et al. 2018; Anh et al. 2021; Mahdi et al. 2021). According to a study conducted in WWTPs, observed ranges of macrolides are 0.3 μg/L in the United States, 0.01–14.5 μg/L collectively with fluoroquinolones and sulfonamides in Australia, and 250–4000 ng/L in Hong Kong (Al‐Riyami et al. 2018). Different reports also show their presence in drinking water (up to 2 ng/L) and in soil manure (Dasgupta and Sengupta 2015). Macrolide traces in the environment pose serious risks to aquatic life and soil microorganisms by affecting their activity and, ultimately, human health (Ekwanzala et al. 2020; Meena et al. 2015). In soil and aquatic environments, macrolides inhibit the growth of many microorganisms, such as algae and cyanobacteria (Kumar et al. 2019; Cycoń et al. 2019; Rodriguez‐Mozaz et al. 2020). Due to their regular release into the environment, macrolide drugs are becoming ineffective against target bacteria because of the presence of antibacterial resistance genes (Fouz et al. 2020; Grenni et al. 2018).

Various treatment methods, including biological and advanced oxidation processes, coagulation and flocculation, electrochemical reduction, disinfection, membrane separation, and adsorption, can remove antibiotics from water. Despite their benefits, these technologies' shortcomings make their implementation difficult (Shearer et al. 2022). Adsorption is one treatment method for which there is no clear evidence of its drawbacks (de Ilurdoz et al. 2022; Eniola et al. 2019). Adsorption removal offers several benefits, including high removal rates, low cost, ease of application, and regeneration properties (Ahmed et al. 2015; Georgin et al. 2022; Şenol et al. 2024; Georgin et al. 2023; El Messaoudi et al. 2024). Antibiotics have often been removed from aquatic environments using adsorbents such as activated carbon, carbon nanotubes, and adsorbent clays like bentonite, and more recently, materials based on layered double hydroxides, graphene oxide, and chitosan have also been developed for this purpose (Genç et al. 2013; Rahman et al. 2024; Rahman and Raheem 2022a; Rahman and Raheem 2022b). Recently, biomass has replaced high‐cost adsorbents and is in demand as a cost‐effective adsorptive material for removing many pollutants. Bioadsorbents are generated from agricultural waste and can be employed directly or with little modification to treat water and wastewater (Eniola et al. 2019). They have advantages due to their large size, wide availability, and cost‐effective production. To enhance adsorption capacity and reduce the solubility of the adsorbent, biomass can be modified or treated with chemicals or other substances (Herbert et al. 2021).

The current research has led to a significant advance in elimination of pharma‐actively compounds (PACs), for example, antibiotics by the application of highly sophisticated technologies, for example, photocatalysis, surface‐engineered composites, and hybrid oxidation adsorption systems (Ahmad et al. 2023; Zhang et al. 2025). Indicatively, recent reports have indicated efficient degradation/removal of antibiotics by modified photocatalysts, heterojunction materials, and functionalized surfaces with increased reactivity and selectivity (Asiri et al. 2025; Shelash Al‐Ha‐Wary et al. 2025). Although these methods have shown to be highly efficient in terms of removal, the methods may be either based on complicated synthesis pathways, costly reagents, or energy‐demanding processes or require complex operational control, which restricts their practical use in decentralized or low‐resource wastewater treatment systems.

Furthermore, although the environmental issue that may arise with the use of macrolide antibiotics, especially azithromycin, has been progressively increasing, the majority of recent adsorption‐based works have centered on fluoroquinolones and tetracyclines. The research on the removal of azithromycin is relatively small, and the current literature often uses engineered nanomaterials, activated carbons, or surface‐modified inorganic adsorbents (Zisti, Kaur, et al. 2025). Significantly less attention has been paid to using biomass based on agricultural waste, which is readily accessible, to remove azithromycin, especially when the comparisons were made in a systematic way and the biomass in the form of raw and chemically modified forms are compared mechanistically (Sillanpää et al. 2025; Zisti, Abdullaev, et al. 2025).

The next weakness of the literature at hand is the fact that most of the studies have only focused on the traditional one factor at a time experimental designs, which fail to adequately demonstrate the interaction effects among the operational parameters as pH, adsorbent dose, contact time, and initial concentration. This makes optimization of the process ineffective and minimizes the statistical strength of the findings reported. Moreover, a relatively mechanistic understanding of the effects of simple chemical modification on adsorption performance, surface heterogeneity, and interaction pathways has not been adequately studied.

This study aimed to evaluate the use of abundantly produced agricultural waste, that is, peanut shell (PS), as an adsorbent for removing macrolides. In 2014, worldwide, the estimated production of peanuts with shells was 43.9 million tons (Islam et al. 2019; Li et al. 2018). However, a very small proportion of PS is utilized for beneficial purposes, whereas the rest is disposed of as waste. Though the adsorption behavior of raw and modified PS has been studied to remove many pollutants, such as metals, dyes, and a few antibiotics (Annida et al. 2018), removing azithromycin by raw and modified PS has not been explored. Therefore, the removal of macrolide compounds using raw PS and base‐modified PS as adsorbents was investigated, and the adsorption parameters, including time, initial concentration, pH, and adsorbent dosage, to enhance the efficiency of the removal process, were optimized using a Taguchi‐based design of experiments (DOE). The adsorption and kinetic studies were also performed to gain insights into the adsorption process and mechanism.

Materials and Methods

2

Materials

2.1

Azithromycin dihydrate with 99% purity was purchased from the Pharmaceutical Laboratory (Lahore, Pakistan). The chemical structure of azithromycin is shown in Figure 1. Raw PS were obtained from the local market. Laboratory‐grade concentrated HCl (95%–98%) and NaOH (> 97%) were purchased from Sigma‐Aldrich. Distilled water was obtained from the distillation apparatus in the Institute of Environmental Engineering and Research (IEER) lab.

Chemical structure of azithromycin.

Sample Preparation

2.2

A 100 ppm stock solution of azithromycin in a 250 mL volume was prepared. For this solution, 0.025 g of azithromycin was dissolved in 100 mL of ethanol in a 250‐mL flask and stirred with a magnetic stirrer for 20 min. Distilled water was added slowly to this solution up to the mark. The solution kept stirring until the drug was dissolved completely. After that, solutions of 20, 30, 40, and 50 ppm in 100 mL were prepared from the stock solution of azithromycin for batch adsorption experiments. Fresh solutions were prepared from the stock for every set of experiments.

Preparation of Adsorbents

2.3

PS were washed repeatedly with distilled water and dried in the oven at 50°C for 6 h. Dry shells were ground to a powder and sieved through a #10 mesh (0.2 mm), and only the fraction passing through was used to ensure a uniform particle size of < 0.2 mm. Dried powder was stored for experimentation and termed raw peanut shell. For modification, 10 g of raw PS powder and 5 g of NaOH were placed in a 100‐mL flask containing 100 mL of distilled water, and the mixture was refluxed. After 3 h of refluxing, the flask was removed and cooled down. The mixture in the flask was filtered and rinsed with distilled water. Then, the mixture was placed in an oven at 50°C for drying (Khan et al. 2015). The dried mixture was stored in a sealed container for experimental studies and referred to as modified PS adsorbent.

Characterization of Adsorbents

2.4

The synthesized adsorbents were characterized by Fourier transform infrared spectroscopy (FTIR) using a Bruker Vertex 70, scanning electron microscopy (SEM), and energy‐dispersive x‐ray (EDX) using a Hitachi S‐4800 scanning electron microscope to investigate the functional groups, morphology, and chemical composition of the adsorbents, respectively.

Taguchi DOE and Adsorption Studies

2.5

Minitab software version 21 was used to perform the DOE using the Taguchi method. Taguchi is a well‐known statistical approach for assessing the influence of factors on removal efficiency. This approach has recently been applied to adsorption studies on antibiotic removal (Zisti, Kaur, et al. 2025). A significantly greater number of experiments would be required to optimize four parameters at three levels using a traditional one‐factor‐at‐a‐time approach. Conversely, this study's Taguchi L16 (4^4^) orthogonal array achieved optimization with fewer runs while retaining statistical robustness.

The Taguchi L16 (4^4^) scheme was employed, with four adsorption factors at four levels each, allowing a systematic investigation of factor effects with minimal experiments. The selected factors were pH (5, 7, 9, and 11), initial concentration (20, 30, 40, and 50 mg/L), adsorbent dose (0.2, 0.4, 0.6, and 0.8 g/L), and time (15, 30, 45, and 60 min). In the Taguchi approach, experimental results are converted to a signal‐to‐noise (S/N) ratio, which is then used to determine the optimal variables with the highest S/N ratios. The S/N ratio was calculated using Equation (1).

$$[eqn]$$

where y _ i _ is the measured adsorption efficiency for the i ^th^ replicate and n is the number of replicates. Response (removal efficiency) was evaluated in Minitab, and adsorption factors were optimized for maximum removal efficiency. Experimentation was carried out in a flask with a mixture of test solutions of known concentration, adsorbent doses, pH, and contact time as specified by Taguchi‐based DOE. This mixture was shaken for a specific contact time by placing the flask on a magnetic stirrer at 100 rpm. The solution mixture was filtered using filter paper (42 μm), and the concentration of azithromycin was determined in the filtrate. All samples were run in triplicate, and their averages were reported in the results.

Method for Azithromycin Analysis

2.6

A sample of azithromycin, before and after treatment, was analyzed at an estimated λmax of 208 nm using a UV–visible spectrophotometer (PerkinElmer). A calibration plot of five concentrations (1, 10, 20, 30, 40, and 50 mg/L) versus absorbance was developed. Duplicate absorbance measurements were obtained for all concentrations at λ_max_ of 208 nm. In the tested concentration range, azithromycin followed Beer–Lambert's law. As shown in Figure 2, the R ^2^ of the plot was 0.99, showing good linearity of the method. Intra‐day and inter‐day measurements were used for precision analysis, whereas accuracy was assured using the recovery method. Measurements after 3‐h intervals within a day and every third day in a week showed a coefficient of variation of < 2%, which showed acceptable precision. The recovery, analyzed at three standard solutions of 70%, 100%, and 130%, ranged from 100.5% to 101.6%, which is also within an acceptable range.

Calibration curve of azithromycin at λmax 208 nm.

Data Analysis

2.7

Experimental data were analyzed using analysis of variance (ANOVA), and the significance level of the adsorption factor was determined by a p‐value (0.05). Data were also fitted to different isotherms and kinetic models to explain the adsorption mechanism.

Kinetic Models

2.7.1

The following models were used to evaluate adsorption kinetics using the experimental data.

Pseudo‐first order model

$$[eqn]$$

Pseudo‐second‐order model

$$[eqn]$$

where q_t_ is the adsorbent capacity (mg/g) at any time t. K_1_ and K_2_ are the adsorption rate constants of pseudo‐first order and pseudo‐second order kinetic models, respectively.

Isotherm Models

2.7.2

The batch adsorption experiment results were investigated using the following isotherm models.

• iLangmuir isotherm model

$$[eqn]$$

• iiFreundlich isotherm model

$$[eqn]$$

• iiiTemkin isotherm

$$[eqn]$$

• ivDubinin–Radushkevich

$$[eqn]$$

where q_max_ and n are the adsorbent's maximum adsorption capacity and sorption intensity, respectively. K_d_ is the adsorption distribution coefficient, K_L_ is the Langmuir constant related to the free energy of adsorption, K_f_ is the Freundlich constant related to sorption capacity, K_T_ is the Temkin binding constant, and K_ad_ is the Dubinin–Radushkevich constant. β = RT/b, where R = universal gas constant, T = temperature, and b = heat of adsorption.

Results and Discussion

3

Characteristics of Adsorbents

3.1

SEM images for raw and modified PS adsorbents are shown in Figure 3a,b. The porous structure in the SEM image of the raw PS adsorbent (Figure 3a) is compact and rigid. These pores are blocked by impurities, which may slow the adsorption process. These pores are cleaned after contact with NaOH (modified PS), as shown in Figure 3b. The adsorption rate could increase due to the modified adsorbent's rougher, more porous surface. This improved pore accessibility might have resulted from the removal of impurities and some lignin and hemicellulose.

SEM images of synthesized peanut shell (PS) adsorbents. (a) Raw PS. (b) Modified PS.

EDX analysis was used to estimate the chemical composition of adsorbents. EDX spectrum for raw and modified PS adsorbent with % weight of elements is illustrated in Figure 4a,b, respectively. From EDX analysis, it is inferred that both the raw and modified adsorbents are mainly composed of oxygen and carbon, reflecting the organic nature of the biomass. Raw PS adsorbent has a high % weight of C and O, whereas these percentages are reduced upon modification due to the elimination of some lignin and hemicellulose (Ikladious et al. 2019). Moreover, NaOH removes Si from the modified PS and increases Na content.

EDX spectrum with elemental composition. (a) Raw PS. (b) Modified PS.

FTIR spectroscopy was used to characterize the functional groups and investigate changes in the chemical structure of the adsorbent before and after modification. Figure 5a,b shows the raw and modified PS FTIR spectra. As shown in Figure 5a, a sharp peak at 3323 cm^−1^ indicates the presence of the hydroxyl (O–H) group (Herbert et al. 2021). The lignocellulosic material in the methylene (−CH_2_) and methyl (−CH_3_) groups was exposed due to C–H stretching at 2920 and 2869 cm^−1^. The peaks around 1728, 1640, 1420, and 1261 cm^−1^ were attributed to the carbonyl (C = O) group, aromatic compounds, and sp2‐hybridized (C = C) structures of lignin (Islam et al. 2019). A sharp peak at 1022 cm^−1^ is characteristic of the cellulosic ether group (C–O stretching) (Pączkowski et al. 2021).

FTIR spectrum. (a) Raw PS. (b) Modified PS.

Although the spectra of both the adsorbents have similar key features, a few differences have been identified that enhance the adsorption rate of the modified adsorbent. As a result of the modification, the peak at 3300 cm^−1^ broadened and its intensity was slightly reduced, as shown in Figure 5b. After treatment with NaOH, a minor shift of 31 cm^−1^ was observed at 3354 cm^−1^, attributed to the free hydroxyl (−OH) group (Swelam et al. 2018). The modified PS spectrum lacks the carbonyl group (1728 cm^−1^) found in the raw adsorbent's hemicellulose and lignin. The absence of a silica peak at 746 cm^−1^ after NaOH modification indicates the reaction of NaOH base and silica. Additionally, the reduction in the intensity of all peaks between 1600 and 1200 cm^−1^ in the modified PS spectrum was due to the removal of impurities and lignin, thereby exposing more porous sites for adsorption (Prabhakar et al. 2015).

Removal of Azithromycin by Raw and Modified PS

3.2

Table 1 shows the removal efficiency obtained at the end of each experimental run. Experimental results show that the removal efficiency with raw PS ranged from 40% to 63%, with an average of 52%. In the modified PS with NaOH base, the overall removal efficiency was 55%–82%, with an average of 68%. Experimental runs 4 and 8 in Table 1 show the maximum azithromycin removal by raw and modified PS, respectively. Maximum removal by both adsorbents occurred in the pH range of 9–11, initial concentration of 20–30 mg/L, with adsorbent dose of < 0.4 g/L and time interval of 45–60 min.

Effect of Treatment Variables on Removal Efficiency

3.3

The effects of adsorption factors (initial concentration, pH, time, and adsorbent dose) on the removal efficiency for each adsorbent were analyzed using the mean function in the Taguchi method. Figure 6a,b shows the main effects plots for the mean of azithromycin removal by raw PS and modified PS, respectively. The percentage contributions of each factor, calculated using ANOVA, are presented in Figure 7. pH is an important variable in the removal of azithromycin from both the raw and modified adsorbents. The point of zero charge (pH_PZC_) for raw and modified PS was found to be 7.2 and 7.6, respectively. At pH of 7.6–8.5, better removals were achieved due to electrostatic attraction, as the surfaces of raw PS and modified PS are negatively charged at pH > pH_PZC_, whereas azithromycin is positively charged (pK_a_ = 8.74). Hence, there is no repulsion between the adsorbent and pollutant, which favors good removal. Conversely, at pH of 5, adsorption is slow due to repulsion between the target drug and the adsorbent, resulting in lower removal rates (Čizmić et al. 2019). Relatively higher removal rates at pH of 9 and 11 were observed due to hydrogen bonding between the adsorbent's hydroxyl groups and the adsorbate (Čizmić et al. 2019). As the initial concentration increased, the adsorption rate decreased.

Taguchi plots showing the effect of adsorption factors for azithromycin removal. (a) Raw PS. (b) Modified PS.

Percentage contribution of factors for the removal of azithromycin.

At high initial concentrations, excess azithromycin molecules accumulate on the adsorbent, reducing removal efficiency. The effect of contact time was substantial at the start of the process when a large volume of active sites was exposed for azithromycin removal on the surface of both adsorbents. At 15 min, there was a significant increase in azithromycin removal by both adsorbents. The adsorption dose has little effect on azithromycin removal. At low adsorbent doses, the removal rate was sufficiently high. With increasing adsorbent dose, adsorption sites overlap due to the accumulation of adsorbent particles, thereby reducing the adsorbent surface area (Gaur et al. 2018). As a result, the adsorbent capacity to uptake azithromycin and the removal efficiency decrease.

The interaction between variables was also examined using the interaction plots shown in Figure 8. The plot (Figure 8) indicates that the effect of adsorbent dose on removal efficiency varies significantly with the initial azithromycin concentration, as evidenced by the crossing lines, which suggest a notable interaction. Moderate interactions are observed between adsorbent dose and pH, whereas interactions with contact time appear weaker.

Interaction plots for RE (%) using data means.

Given the strong interaction between adsorbent dose and initial concentration, contour plots were generated to examine their combined effect on removal efficiency (presented in Figure 9). As shown in Figure 9, greater removal of azithromycin can be achieved with a combination of lower adsorbent doses (< 0.5 g/L) and low to moderate initial concentrations (20–50 mg/L). The removal efficiency is lowest at higher adsorbent doses and higher initial concentrations.

Contour plots for RE (%) versus initial conc. versus adsorbent dose.

ANOVA

3.4

ANOVA was performed on experimental results. The influence of each adsorption factor on removal efficiency is justified by the p‐value (probability of occurrence). Table 2 displays the analysis of variance for both adsorbents. A factor's probability of occurrence (p‐value) must be less than 0.05 to be significant in the model (Aschale et al. 2021). pH and initial concentration are more significant, with p‐values < 0.05 for both adsorbents, whereas contact time and adsorbent dose are insignificant, with *p‐*values > 0.05.

Optimization and Confirmation Experiments

3.5

The response optimizer function in Minitab was used to optimize adsorption conditions for removing azithromycin using raw and modified PS adsorbents. Optimization was performed using the general linear model from ANOVA to identify the concentration, pH, dose, and time that facilitate maximum azithromycin removal. The optimum conditions for removing azithromycin from raw PS (65%) are an initial concentration of 20 mg/L, a pH of 11, a contact time of 45 min, and an adsorbent dose of 0.4 g/L. The maximum removal (83%) can be achieved with the modified adsorbent at an initial concentration of 30 mg/L, a pH of 11, a contact time of 60 min, and an adsorbent dose of 0.4 g/L. The response optimizer's predicted results for the general linear model were confirmed through validation experiments with triplicate runs. Validation results were within ±2% for the raw adsorbent and within +2.2% for the modified adsorbent.

Adsorption Kinetics

3.6

Time‐dependent kinetic studies were conducted under optimized conditions for both adsorbents. Data was plotted against pseudo‐first‐order (PFO) and pseudo‐second‐order (PSO) models, as shown in Figure 10a,b for raw PS and modified PS, respectively. Table 3 shows kinetics parameters for azithromycin adsorption by raw and modified PS adsorbents. Table 3 illustrates that the coefficient of determination for the PSO for raw and modified PS (R ^2^ = 0.99) is higher than that for the PFO (0.98), and the RMSE is lower, indicating that the pseudo‐second‐order model provides a better fit for the AMS model. This indicates that adsorption kinetics are predominantly governed by surface adsorption processes involving active sites on the adsorbent. This likely indicates that the adsorption of azithromycin by both adsorbents is regulated by reaction rather than mass transfer (Li et al. 2018). The pseudo‐second‐order model also specified that the adsorption of azithromycin on raw and modified PS requires two binding centers (Raheem et al. 2024). The adsorption capacities of azithromycin at equilibrium (q_e_) by raw PS and modified PS are in close agreement with experimental values. In particular, the higher values observed for modified PS (69.6 mg/g) are consistent with the improved PSO fitting, indicating enhanced surface adsorption due to increased availability of active sites following alkali modification.

Plots for different kinetic models.

Adsorption Isotherms

3.7

Experiments were performed at optimized conditions by varying initial concentrations for Langmuir, Freundlich, Temkin, and D‐R isotherm models. The adsorption isotherms of azithromycin onto raw and modified PS are shown in Figures 11 and 12, whereas the calculated adsorption parameters for these isotherms are shown in Table 4.

Plot of isotherm data for raw PS obtained at 20 mg/L initial concentration, a pH of 1, 45 min contact time, and 0.4 g/L adsorbent dose.

Plot of isotherm data for modified PS at C0 = 30 mg/L initial concentration, a pH of 11, 60 min contact time, and 0.4 g/L adsorbent dose.

The Langmuir model still adequately described adsorption on raw PS (R ^2^ = 0.99, RMSE = 1.0), whereas a slightly lower Langmuir fit for modified PS (R ^2^ = 0.94, RMSE = 4.4) indicates partial violation of the identical‐site assumption due to surface heterogeneity introduced by chemical modification. This suggests primarily monolayer adsorption on relatively homogeneous surface sites. The maximum adsorption capacity (q_max_) in the Langmuir model for raw PS is 159.2 mg/g, and for modified PS is 192.1 mg/g. For both adsorbents, favorable adsorption on heterogeneous surfaces was confirmed by the Freundlich constants (n = 1.49–1.55), with higher K_F_ values for the modified adsorbent indicating increased site availability and adsorption affinity. Significant adsorbate–adsorbent interactions and heterogeneous surface energies were further supported by the Temkin model, particularly for modified PS, where stronger interactions align with higher surface functionalization.

The computed mean adsorption energy increased from 0.24 kJ/mol for raw PS to 12.9 kJ/mol for modified PS, though the D‐R model indicated relatively lower fitting accuracy. This suggests a shift from weak physical adsorption to stronger electrostatic or ion‐exchange interactions. Together, these findings show that alkali modification increases surface heterogeneity, interaction strength, and electrostatic affinity, thereby improving adsorption efficiency. On the other hand, surface‐controlled interactions, including hydrogen bonding and electrostatic attraction, continue to govern the basic adsorption mechanism.

Proposed Adsorption Mechanism

3.8

The adsorption of azithromycin onto both raw and alkali‐modified PS powder is mainly driven by surface interactions rather than bulk diffusion, as demonstrated by kinetic modeling, equilibrium isotherms, surface functional group analysis, and solution chemistry factors. For both types of adsorbents, the pseudo‐second‐order kinetic model provides the best fit, indicating that the adsorption rate depends on the availability of surface‐active sites. NaOH modification alters the lignocellulosic structure of the PS by hydrolyzing ester linkages and partially removing hemicellulose and lignin, thereby exposing more hydroxyl (−OH) groups and increasing surface charge density. The disappearance of the carbonyl (C = O) peak at approximately 1730 cm^−1^ after NaOH treatment (Figure 5b) suggests hydrolysis of ester bonds connected to hemicellulose and lignin components. These structural changes significantly improve the adsorbent's capacity by enhancing the accessibility and reactivity of adsorption sites. Therefore, it is postulated that azithromycin adsorption mainly occurs through electrostatic attraction, especially on alkali‐treated PS. The SEM and EDX results show increased surface heterogeneity and chemical activation following alkali treatment, further supporting the proposed adsorption mechanism.

The PS surface becomes negatively charged at solution pH levels above the pH_PZC_ (7.2–7.6), whereas azithromycin (pK_a_ = 8.74) mainly exists in its protonated, positively charged form. This charge difference, caused by strong electrostatic attraction, greatly enhances adsorption, especially with the alkali‐modified adsorbent. The significantly higher q_max_ (192.1 mg/g) of the alkali‐treated PS compared to the raw material (159.2 mg/g) results from improved accessibility of adsorption sites after NaOH treatment, increased availability of surface functional groups, and stronger electrostatic interactions. Collectively, these factors enhance adsorption capacity and interactions between azithromycin and the adsorbent.

Isotherm analysis supports this mechanism: The Freundlich and Temkin models indicate increased surface heterogeneity and variable interaction energies after alkali modification, whereas the Langmuir model shows mainly monolayer adsorption, especially for raw PS. Additionally, the modified material's D‐R mean adsorption energy increased from 0.24 kJ/mol for raw PS to 12.9 kJ/mol, suggesting a shift away from covalent chemisorption towards stronger electrostatic or ion‐exchange interactions. Without altering the fundamental adsorption pathway, alkali modification enhances overall adsorption efficiency by increasing site accessibility and strengthening existing hydrogen bonding and electrostatic interactions.

Comparison of Adsorption Capacities With Other Adsorbents

3.9

Table 5 shows the maximum adsorption capacities of different biomass adsorbents used in previous studies. Adsorbents synthesized in this study exhibit significantly higher adsorption capacities than those of others. For the removal of numerous antibiotics from the environment, raw and modified PS adsorbents have the potential to serve as high‐capacity, inexpensive adsorbents.

Conclusions

4

This study shows, for the first time, the effective use of alkali‐modified PS powder as a low‐cost, lignocellulosic adsorbent for removing azithromycin from water environments. Treating the raw PS with NaOH created a more porous surface of the PS adsorbent, boosting its adsorption efficiency. pH appeared to be the most significant factor, with alkaline conditions (9–11) favoring higher removal rates. Raw PS achieved a maximum removal of 65% at pH of 11, initial concentration of 20 mg/L, an adsorbent dose of 0.4 g/L, and a contact time of 45 min. The modified PS adsorbent achieved an optimal removal of 83% at pH of 11, an initial concentration of 30 mg/L, a contact time of 60 min, and an adsorbent dose of 0.4 g/L. Kinetic, isotherm, and surface analysis results showed that alkali modification greatly enhanced adsorption capacity by increasing surface heterogeneity, improving electrostatic interactions, and exposing more oxygen‐containing functional groups. These findings suggest that modified agricultural wastes could be effective for removing antibiotics in water and wastewater treatment applications. Using readily available agricultural waste and simple chemical modifications highlights the potential of this adsorbent for decentralized or low‐cost wastewater treatment systems to address antibiotic contamination. Despite these encouraging results, the study has some limitations, including the absence of thermodynamic analysis, regeneration studies, and post‐adsorption surface characterization, which would provide deeper insights into adsorption reversibility and long‐term performance. Future research should therefore focus on adsorbent regeneration and reuse and on validation in real wastewater matrices to evaluate scalability and economic viability.

Author Contributions

Rohab Asad: investigation, writing – original draft, software. Ghulam Hussain: investigation, methodology. Muhammad Usman: supervision, writing – review and editing, resources, visualization. Sahar Aurangzeb: writing – review and editing, formal analysis. Sana Afzal: formal analysis. Yasser Fouad: writing – review and editing, data curation. Muhammad Imran Masood: writing – review and editing, data curation. Naseem Abbas: writing – review and editing, supervision, validation. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by King Saud University (ORF‐2026‐698).

Ethics Statement

The authors state that these experiments were conducted according to established ethical guidelines, and informed consent was obtained from the participants. Also, it is stated that the study complies with all regulations and confirms that informed consent was obtained.

Conflicts of Interest

The authors declare no conflicts of interest.