Removal of selected pharmaceutical pollutants from groundwater using colloidal activated carbon

Abstract

Groundwater (GW) is a vital source of freshwater worldwide. However, the increasing occurrence of pharmaceutical contaminants originating from various anthropogenic activities, posing significant health risks. This study investigates the efficiency of using colloidal activated carbon (CAC) in- removing carbamazepine (CBZ), lamotrigine (LTG), and caffeine (CAF) under a simulated in-situ remediation of groundwater. The study conducted through a series of fixed-bed column experiments. Several characterization techniques, including SEM/EDX, BET surface area, particle size distribution, and zeta potential, were used to identify the physiochemical properties of the CAC. Water samples spiked with a mixture of the three pharmaceutical compounds were allowed to pass through a column filled with a mixture of a bed and CAC materials. Inflow and outflow concentrations of the pharmaceuticals were measured by High-Performance Liquid Chromatography (HPLC) system. The effect of several parameters on the removal efficiency were investigated, including water flow rate, inlet concentration, CAC dosage, and bed type. SEM/EDX results revealed that CAC has a porous surface with 95.68% carbon and 4.32% potassium. The CAC has a surface area of 1112.24 m²/g, particle size ranges between 6.68 and 39.44 μm, and a negative surface charge. The optimal Dose-Response model breakthrough curve was developed at a flow rate of 1 mL/min, 2 g CAC, an initial concentration of 20 mg/L, and 50:50 sand-carbonate mixture showed a removal efficiency of around 40% of the pharmaceuticals. The study demonstrates that the use of colloidal activated carbon adsorption is a promising technology for the in-situ remediation of groundwater contaminated with pharmaceuticals.

Introduction

Emerging contaminants (ECs) are compounds whose potential effects on human health and the environment have only recently drawn the attention of researchers¹. These chemicals are present in a wide range of products, such as pesticides, plasticizers, pharmaceuticals, and personal hygiene products². They can release hazardous wastes and non-biodegradable compounds into the environment, resulting in groundwater (GW) pollution³. Pharmaceuticals are compounds designed to have a therapeutic effect on the body⁴. Their presence in water has caused significant concerns over the potential for negative consequences on both humans and animals⁵. Sewage treatment plants (STP) release their effluent into the environment after incomplete breakdown, which often results in a certain amount of the parent compound (or conjugated derivative) ending up in sewage treatment facilities through biliary (feces) and renal (urine) excretions⁶. Since they are meant to be resistant to chemicals and challenging to get rid of with conventional treatment techniques, they have been found in drinking water, surface water, and groundwater^6,7.

Among pharmaceuticals, caffeine (CAF), which activates the central nervous system, is a member of the methylxanthine chemical group. It is among the most widely used pharmacologically active substances worldwide. Because of its widespread use, CAF is frequently found in water sources⁸. Carbamazepine (CBZ) is a mood stabilizer and antiepileptic medication. It is frequently detected in groundwater and found in aquatic habitats⁹. Another antiepileptic drug used to treat epileptic seizures is lamotrigine (LTG), which can be used alone or in combination with other drugs like CBZ¹⁰.

Physical, chemical, and biological methods have been employed in the remediation of such pharmaceuticals. Chemical and biological methods include advanced oxidation processes (AOPs) and anaerobic or aerobic mechanisms have been reported; however, these techniques can be costly, generate harmful byproducts, or take a long time. Adsorption is a preferred physical method due to its low cost and high efficiency; it can be applied by using different materials such as activated carbon⁵. Activated carbon (AC) is an efficient adsorbent for eliminating ECs from water. Its high porosity and surface area allow it to extract more than 90% of ECs from a variety of materials. It originates from carbon-rich sources such as wood, coal, and coconut shells^5,11. There are three types of AC commonly used for adsorption: granular activated carbon (GAC), powdered activated carbon (PAC), and colloidal activated carbon (CAC)^12,13.

Both ex-situ (pump & treat) and in-situ AC adsorption processes are used for groundwater remediation The in-situ treatment using CAC is injected directly into the groundwater. The deposition of CAC on the sediment grains surrounding the injection hole then creates a sorption-active zone in the aquifer, which can prevent pollutants from spreading with groundwater flow¹⁴.

Several studies reported the presence of various types of pharmaceuticals in groundwater, including psychiatric medications, anti-hypertensives, antibiotics, and antifungals. In Germany, Reh et al.¹⁵ detected CBZ and LTG concentrations of 9.8 ng/L and 38.4 ng/L, respectively. While, López-Serna¹⁶ and his team detected CBZ of 10.9 ng/L in Spain. Imma Ferrer¹⁷ detected Lamotrigine in 93% of sedimentary groundwater samples across the United States. Yen-Ching Lin et al.¹⁸ investigated emerging pollutants in Taiwanese groundwater, identifying potential sources of local contaminants and their influence on groundwater quality. In Saudi Arabia, Aasim M. Ali et al.¹⁹ found that 13 pharmaceuticals and personal care products (PPCPs) were present in seawater samples from high-concentration areas in the Red Sea coastal waters. The most common contaminants were metformin, diclofenac, acetaminophen, and caffeine. In addition, Yolanda Picó et al.²⁰ found that water samples were contaminated with caffeine, Paracetamol, and Ibuprofen. Additionally, Yolanda Picó et al.²¹ found pharmaceuticals in surface water, sediments, and vegetation from various sites in Saudi Arabia. Other studies have explored the adsorption of pharmaceuticals using AC in various applications. For instance, Sotelo et al.²² studied caffeine adsorption using GAC in a fixed-bed column setup, revealing that the breakthrough time increased with column length, inlet concentration, and flow rate. The study also found that higher bed depth, initial concentration, and flow rate affected the adsorption process. Facundo et al.²³ studied the removal of specific pharmaceutical active compounds, including caffeine, using commercial ACs and reused AC from a drinking water treatment facility. Higher bed height and initial concentration increased breakthrough time and removal, while PAC showed higher removal efficiency than GAC. Biological treatment, sequencing batch reactor (SBR) was applied, showing effective removal of caffeine, with a maximum removal efficiency of 95%. Delgado et al.²⁴ compared PAC and GAC and found that PAC had greater adsorption capacity and rate. Miera et al.²⁵ examined the use of GAC to extract caffeine from tainted tap and deionized water (DI water). Delgado et al.²⁶ created a technology that effectively removed CBZ using a fixed-bed adsorption column with PAC evenly dispersed throughout the sand. Removal efficiencies during the first six days of operation exceeded 80%. Almuntashiri et al.²⁷ demonstrated the importance of using GAC column adsorption for micropollutant removal, including CBZ. Increased flow rate resulted in faster saturation and depletion of the adsorbent due to fast breakthrough. According to the literature, most studies have focused on treating such pollutants in water by using GAC and PAC. On the other hand, not much research has been published regarding the use of CAC to remove pharmaceuticals. To the best of our knowledge, only one study investigated employing CAC to remove such pollutants. In 2024, a commercial PAC and a lab-prepared CAC were examined in batch studies to study the removal of CBZ, focusing on a single parameter (pH). The results of CAC revealed that it had a greater adsorbing efficiency than PAC¹³. The aforementioned study used a different kind of CAC in batch studies, which differs from our work. This study’s objectives are to use a lab-scale fixed-bed column to assess the effectiveness of CAC in removing specific pharmaceuticals (CBZ, LTG, and CAF), determining the optimal treatment conditions for the target pollutants by CAC, determining the isotherms and kinetic models, as well as studying the impact of specific treatment parameters (such as flow rate, contaminant concentration, water salinity, and adsorbent dosage) on the removal efficiency of the CAC.

Materials and methods

Materials

Colloidal Activated Carbon (CAC) (CP1-PAC-F/AquaSorb™ HYDROSOL™) was purchased from Jacobi Carbons and diluted 10 times. Sustainable raw materials are used in the production of CAC. Table 1 shows the characteristics of the CAC provided by the manufacturer. Acrylic columns (10 cm in length and 2 cm in diameter) were purchased from the local market. Ottawa sand (ASTM graded sand-C778) was purchased from U.S. Silica Company. Multichannel peristaltic pump BT100-1 L was purchased from Longer Pump. Caffeine (purity > 98%), Carbamazepine (purity > 97%), and Lamotrigine (purity > 98%) were purchased from TCI chemicals. For each experiment, a fresh CAF, LTG, and CBZ solution was prepared.

Table 1.

CAC properties provided by the manufacturer.

Visual test	homogeneous spreading
The amount of solution in AC	For 1 L of water (100–200 g)
The density of the solution	Between 1.05 and  1.25 g\cc
The median diameter of the (dry) adsorbing of materials	15–35 μm
Iodine index (after drying and dilution at 1/20)	> 900 mg/g
Viscosity (room temp)	= 0.5 Pa.s.

Analytical method

High-Performance Liquid Chromatography (Ultimate 3000 HPLC model with DAD detector) from Thermo Fisher Scientific, USA, was used for all analytical quantifications. The stationary phase was a C18 column (150 × 4.6 mm) and the mobile phase was a mixture of 50:50 (v/v) water/methanol pumped at a rate of 1.0 mL/min. The levels of CAF, CBZ, and LTG were measured before and after the removal. Standard solutions of each pharmaceutical were injected into the HPLC system to record peak areas, which were used for the determination of the calibration curve. With “Y” representing the peak area, “X” the concentration, “a” the slope, and “b” the intercept, the resulting linear regression equation was Y = aX + b. The correlation coefficient (R²), limit of detection (LOD), and limit of quantification (LOQ) for each pharmaceutical compound were calculated and shown in Table 2. The linearity was determined to be between 100 ppb and 50 ppm for CAF and between 200 ppb and 50 ppm for LTG and CBZ.

Table 2.

The analytical parameters.

	LOD (ppb)	LOQ (ppb)	Correlation coefficient (R²)
CAF	30	100	0.9994
LTG	60	200	0.9997
CBZ	60	200	0.9995

Characterization of colloidal activated carbon (CAC)

It is crucial to characterize CAC to determine its physiochemical characteristics and comprehend the adsorption mechanism. After being dried for 24 h at 110 °C, CAC was crushed and used for various characterizations. Zeta potential (Litesizer 500) by Anton Paar was used to determine the external electric surface charge at various pH levels (2–12). The stability of CAC is determined by the charge carried by tiny particles on CAC that disperse in water. A stable suspension results from a higher zeta potential (positive or negative), whereas clumping or settling particles are caused by a lower zeta potential (near zero). In a 200 mL flask of DI water, 0.2 g of potassium nitrate (KNO₃) (10 mM) was used as a background electrolyte to create the zeta potential solution. After that, it was put in 50 mL conical flasks, and the pH was adjusted using diluted sodium hydroxide (NaOH) and Nitric Acid (HNO₃) solutions. To allow the CAC to interact with the solution, 0.1 g of CAC was injected into each flask, mixed, and left overnight. Zeta potential was then determined after centrifuging 3 mL of each flask. JEOL, JCM-7000 Benchtop scanning electron microscopy coupled with energy dispersive X-ray spectroscopy (SEM/EDXS) was used to examine the morphology, surface structure, and composition of CAC. A dynamic laser diffraction particle size analyzer (PSD) (HELOS/BR) by QUIXEL was used to determine the particle size distribution for both the original and diluted CAC samples. To determine the availability of a surface for adsorption for porous materials, Anton Paar, Quantachrome Autosorb iQ Station 1 was used to evaluate the BET (Brunauer, Emmett, and Teller) surface area and CAC pore volume. N₂ adsorption-desorption isotherms at 77 K were used to produce it, and the density functional theory (DFT) approach was used to measure the pore volume.

Fixed-bed column experiments

Acrylic columns filled with Ottawa sand were used for the experimental work on fixed-bed columns. The columns utilized in this study were 10 cm length, 2 cm diameter. Figure 1 depicts the experimental schematic diagram. To optimize flow distribution and stop CAC washout, glass wool and 1.5 g of gravel layers were placed at both the upper and lower ends of the column. At the upper end, a 4 g of seawater sand layer was positioned below the gravel, while Ottawa sand was packed between the upper and lower support layers to serve as the main packing medium. Different doses of CAC were introduced into the columns through the bottom inlet to form the packed bed. The resulting fixed-bed porosity was determined to be 0.314. A solution of lamotrigine, carbamazepine, and caffeine in varying concentrations was pumped upward at a predetermined flow rate using a peristaltic pump at 25 ◦C. To ensure its mobility and dispersion throughout the columns, the CAC was diluted 10 times. The effects of the adsorbent (CAC) dosage, flow rate, initial contaminant concentration, and bed type, as indicated in Table 3, were assessed using fixed-bed experiments. The carbonate material was sieved to a particle size > 250 μm prior to packing to prevent clogging. A control experiment was carried out under different conditions without the addition of CAC in order to measure the adsorption of CAF, CBZ, and LTG by the bed material as well as the system losses. The breakthrough curves were assessed under different conditions. To determine the adsorption mechanism and measure the adsorption capacity, kinetic and isotherm models were employed. Samples were taken at various times and analyzed by HPLC.

Fig. 1.

Schematic diagram of the fixed-bed column experimental setup.

Table 3.

Experimental layout for fixed-bed column tests.

Influencing factor	Flow rate (mL/min)	Initial concentration (CAF-LTG-CBZ) (mg/L)	CAC dosage (g)	Bed type
CAC dosage	1	20	1	Ottawa
CAC dosage	1	20	2	Ottawa
CAC dosage	1	20	3	Ottawa
Flow rate	0.5	20	2	Ottawa
Flow rate	1	20	2	Ottawa
Flow rate	2	20	2	Ottawa
Initial concentration	1	10	2	Ottawa
Initial concentration	1	20	2	Ottawa
Initial concentration	1	30	2	Ottawa
Bed type	1	20	2	Ottawa
Bed type	1	20	2	Carbonate
Bed type	1	20	2	Ottawa + Carbonate

The breakthrough curve can be used to determine the adsorption capacity, which is a crucial indicator of column performance. Table 4 presents the formulas used to determine the breakthrough curve parameters for pharmaceutical adsorption on CAC. Where Co is the initial concentration, Ct is concentration at time t, q is the equilibrium uptake of pharmaceuticals (mg/g), m is the amount of adsorbent (g), R% is the pharmaceuticals removal efficiency in (%), m_total is the total quantity of pharmaceuticals passed over the column (mg), and m_adsorb (mg) is the adsorbed pharmaceuticals, and Q is the volumetric flow rate (mL/min).

Table 4.

Equations used to calculate the breakthrough curve parameters.

	(1)
	(2)
	(3)
	(4)

Breakthrough curve model

The performance of an adsorption column is mainly influenced by key operational factors, including the breakthrough and saturation times, the shape of the breakthrough curve, and the adsorption capacity of the column. These parameters were identified using a non-linear regression technique for the graphs of the concentration at time t divided by the initial concentration (C/C₀) vs. time. Several mathematical models are commonly used to fit the experimental data from column experiments and to forecast the concentration/time profiles, and breakthrough curves. The Dose-Response model is one of the empirical models applied in this study. It has been widely used to describe adsorption and removal behavior of various pollutants as well as to characterize the kinetics of fixed-bed columns. Equations (5) and (6) were used to fit the experimental breakthrough curves.

5

6

where a is a dimensionless fitting constant, b (or V50%) represents the treated volume (L), at which the effluent concentration reaches 50% of the influent concentration, q_o​ is the maximum adsorption capacity (mg/g), Co is the influent concentration, and C is the effluent concentration at time t²⁸.

Results and discussion

Characterization of colloidal activated carbon (CAC)

SEM micrographs of the CAC at different magnifications are depicted in Fig. 2. The micrographs confirm that the CAC surface is highly amorphous. The morphology at lower magnification (Fig. 2a), show presence of an irregular, fractured morphology with abundant cracks and cavities. This morphology can facilitate pharmaceutical transport toward internal adsorption sites. At higher magnification (Fig. 2b), a more detailed porous network is depicted, which include numerous pore openings and interconnected voids with non-uniform shapes. This morphology indicates a heterogeneous surface with multiple adsorption sites. Such surface roughness and pore development are desirable for adsorption because they increase the accessible surface area and support diffusion and pore-filling mechanisms. The corresponding EDX spectrum (Fig. 2c) shows that CAC is predominantly composed of carbon (95.68 wt%) with a minor contribution of potassium (4.32 wt%), which could be attributed to activating agents or stabilizing agents in the CAC. The obtained results are consistent with the morphology and elemental composition reported for porous activated carbon materials and support the suitability of CAC as an adsorbent in this study^29,30.

Fig. 2.

SEM micrographs of CAC at (a) low magnification and (b) higher magnification; (c) EDX spectrum.

According to the zeta potential measurements displayed in Fig. 3a, CAC exhibits a negative surface charge, as evidenced by the consistently negative values across all pH levels. Many researchers have observed similar findings^8,30. As the pH increased, the zeta potential charge became more negative. Increasing the solution pH from 2 to 12 resulted in decreasing the zeta potential from − 3.9 to -45.8 mV. The increase in surface negativity enhances electrostatic repulsion between CAC particles, which prevent agglomeration and improve the suspension stability. The well dispersed CAC also contribute to higher surface area, thereby greater adsorption capacity. The fact that CAC has a negative surface charge indicates stronger affinity for positively or neutral charged pollutants.

Fig. 3.

(a) Zeta potential measurements and (b) N₂-adsorption-desorption isotherms of the CAC.

The BET surface area was determined to be 1112.24 m²/g, indicating that it has a large surface area for adsorption applications and a highly porous structure. This value falls within the commonly reported range for commercial activated carbons (typically ~ 500–1400 m²/g)³¹. It is comparable to GAC materials used in water treatment. For example, Filtrasorb 400 has been reported with a BET surface area of 1105 m²/g³². According to the International Union of Pure and Applied Chemistry (IUPAC), the rapid uptake of nitrogen at low P/P_o indicated that the material had a microporous structure with pore diameters (< 2 nm), Type 1, as determined by the N₂ adsorption-desorption isotherm shown in Fig. 3b. The total pore volume was found to be 0.511 cm³/g. This value is comparable to previous studies, where total pore volume of 0.630 cm³/g was reported for the lab-manufactured CAC used¹³. The total pore volume determined by Density Functional Theory DFT method was 0.470 cm³/g. The mean pore diameter of 1.051 nm indicates that the majority of pores are microporous. The high surface area and microporous nature enhance the adsorption capacity of CAC for effective removal of Pharmaceuticals. Table 5 displays the particle size distribution results of both the original and diluted CAC samples. It reveals that the median particle size, or X₅₀, is 17.87 μm for the diluted CAC sample and 17.44 μm for the original CAC particles. Furthermore, X₉₀ values for the original and diluted CAC samples were 39.44 μm and 38.51 μm, indicating 90% of the CAC particle size or less. X₁₀ for the original and diluted samples was determined to be 6.68 μm and 6.94 μm, respectively. This suggests that 10% of CAC particle sizes have these values or below.

Table 5.

Particle size distribution of CAC.

Value	Particle size (µm)
	Original CAC	Diluted CAC
X10	6.68	6.94
X50	17.44	17.87
X84	32.35	32.28
X90	39.44	38.51

Breakthrough curves modelling

The Dose-Response model was found to provide the best fit to the fixed-bed breakthrough curves. The nonlinear model fitting was performed using OriginPro by plotting C/C₀ against time (min). The effectiveness of fitting the breakthrough curve was compared using the correlation coefficient (R²) values, where high R² values indicate better correlation. Table 6 shows the Dose-Response model parameters for the CAF, LTG, and CBZ, under different experimental conditions. The obtained results revealed that the calculated q₀ (µg/g) increased as flowrate decreased and the adsorbent dosage increased. This is attributed to the longer contact time and greater availability of active sites in the column. Fluctuated change in q₀ was observed at different concentrations and bed type.

Table 6.

Dose-Response parameters under different experimental conditions.

Variables	Test No.	CAF				LTG				CBZ
		a	b (mL)	q0 (µg/g)	R2	a	b (mL)	q0 (µg/g)	R2	a	b (mL)	q0 (µg/g)	R2
Flow rate (mL/min)	0.5	2.5	520	5199	0.98	2.0	501	5012	0.92	1.9	553	5533	0.98
	1	2.1	96	955	0.98	1.8	100	1000	0.97	2.1	101	1011	0.98
	2	4.1	81	813	0.98	2.9	93	925	0.97	3.1	98	979	0.97
CAC Dosage (g)	1	1.5	50	996	0.97	1.5	44	882	0.97	1.4	47	942	0.95
	2	2.3	96	959	0.98	1.8	101	1009	0.97	2.1	101	1011	0.99
	3	2.3	323	2156	0.99	2.4	307	2046	0.99	2.3	354	2362	0.98
Initial Conc. (mg/L)	10	3.5	451	2253	0.99	3.2	437	2184	0.98	3.4	474	2369	0.99
	20	2.3	96	959	0.99	1.8	101	1009	0.97	2.1	101	1012	0.99
	30	3.8	148	2218	0.99	3.8	140	2101	0.99	3.3	130	1950	0.97
Bed Type	Carbonate	1	118	1178	0.97	0.9	76	763	0.93	1.1	65	651	0.94
	sand	2.3	96	959	0.99	1.8	101	1009	0.97	2.1	101	1012	0.99
	sand, carbonate (50:50)	3	203	2030	0.95	2.0	211	2111	0.98	2.9	183	1827	0.95
Best conditions	All parameters	2.7	265	2649	0.99	2.8	243	2426	0.99	2.7	246	2463	0.99

Effect of operational conditions

The effect of flow rate

Figure 4 shows the breakthrough curves obtained at various flow rates of 0.5, 1, and 2 mL/min, using an initial concentration of 20 mg/L and 2 g of CAC. The results showed that longer breakthrough and saturation was exhibited at lower flowrates. As exhibited in Table 7 an increase in flow rate from 0.5 to 2 mL/min, the breakthrough time (C/C₀ ≈ 0.1) declined from 206 min to 28 min, while saturation time (C/C₀ ≈ 0.85) reduced from 446 min to 308 min. This behavior can be attributed to the longer contact time between the CAC and the spiked solution at low flow rate, which enhance solute diffusion into the CAC pores and improves removal efficiency. Pharmaceuticals move more quickly at high flowrates, which could lead to insufficient time for adsorption process, similar behavior was reported many studies^22,33,34. However, because of the larger mass transfer, this investigation demonstrated higher removal efficiencies for a flow rate of 2 mL/min³⁵. The adsorption capacity was found to be higher for higher flow rates; this is because the 0.5 mL/min flow rate did not reach saturation; however, if the experiment was extended, the adsorption capacity of the 0.5 mL/min flow rate would exhibit higher results^36,37.

Fig. 4.

The breakthrough curves at different flowrates: (a) CAF, (b) LTG, and (c) CBZ.

Table 7.

Effect of flow rate on the breakthrough parameters for the adsorption of CAF, LTG, and CBZ onto CAC.

Flow rate	0.5 mL/min			1 mL/min			2 mL/min
	CAF	LTG	CBZ	CAF	LTG	CBZ	CAF	LTG	CBZ
tb (min)	206	206	206	45	45	45	28	28	28
ts (min)	446	446	446	360	360	360	308	308	308
madsorb	1.09	1	0.76	2.13	1.71	1.72	3.87	3.04	3.05
mtotal (mg)	4.24	4.01	3.57	6.84	6.48	5.76	11.7	11	9.86
q (mg/g)	0.55	0.5	0.39	1.06	0.86	0.86	1.93	1.52	1.53
R%	25.83	24.94	21.72	31.19	26.38	29.90	33.05	27.47	30.98

The effect of the CAC dosage

The breakthrough curves for the dosage effect are shown in Fig. 5. Three adsorbent dosages of CAC (1 g, 2 g, and 3 g) were tested at a flow rate of 1 mL/min and an initial concentration of 20 mg/L. A control column without CAC was operated to quantify the removal by the bed material. The control column showed insignificant removal by bed material, confirming that the removal occurred through adsorption onto CAC. At lower CAC dosages, short breakthrough time was observed, indicating faster bed saturation. As summarized in Table 8, the breakthrough time (C/C₀ ≈ 0.1) decreased from 180 min for 3 g dosage to 45 min and 15 min for the 2 g and 1 g dosages, respectively. Moreover, saturation time was found to increase at higher dosage due to the high availability of adsorption sites. Even though it did not achieve saturation it had the greatest adsorption capacity. The saturation time increased from around 240 min to 420 min when dosages increased from 1 g to 3 g. Similar trends have been reported by other researchers using different adsorbents^38,39.

Fig. 5.

The breakthrough curves at different CAC dosage: (a) CAF, (b) LTG, and (c) CBZ.

Table 8.

Effect of dosage on the breakthrough parameters for the adsorption of CAF, LTG, and CBZ onto CAC.

CAC Dosage	1 g			2 g			3 g
	CAF	LTG	CBZ	CAF	LTG	CBZ	CAF	LTG	CBZ
tb (min)	15	15	15	45	45	45	180	180	180
ts (min)	240	240	300	360	360	360	420	420	420
madsorb	0.98	0.78	1.05	2.21	1.76	1.83	2.88	2.85	2.42
mtotal (mg)	4.72	4.44	5.1	7.09	6.66	6.12	8.27	7.77	7.14
q (mg/g)	0.98	0.78	1.05	1.10	0.87	0.91	0.96	0.95	0.80
R%	20.67	17.67	20.60	31.19	26.39	29.91	34.82	36.73	33.99

The effect of initial concentration

Figure 6 present the breakthrough curves at initial pharmaceutical concentrations of 10 mg/L, 20 mg/L, and 30 mg/L. The experiment was conducted using 2 g of CAC and a constant flow rate of 1 mL/min. The obtained results showed that the breakthrough time decreased as the initial concentration increased, which is attributed to the fast saturation of the available binding sites at high initial concentrations³⁴. At lower concentrations, however, the breakthrough occurs later, and the surface of the adsorbent takes a longer time to become saturated. Table 9 shows that when the inlet concentration increased from 10 mg/L to 30 mg/L, the breakthrough time (C/C₀ ≈ 0.1) decreased from 255 min to 45 min and saturation time (C/C₀ ≈ 0.85) from 435 min to 318 min. The breakthrough curves at lower concentrations were relatively flatter, indicating the formation of a broader mass transfer zone (MTZ) within the column. However, at higher concentrations the curves were steeper and more S-shaped, which suggests a narrower MTZ and enhanced intraparticle diffusion process^34,40. Similar observations have been reported in many studies^34,41,42. The observed increase in adsorption capacity with higher concentrations may be attributed to higher mass transfer due to higher concentration gradient at higher concentrations. These findings are consistent with previous studies which reported similar behavior^33,36,43.

Fig. 6.

The breakthrough curves at different initial concentration (a) CAF, (b) LTG, and (c) CBZ.

Table 9.

Effect of initial concentration on the breakthrough parameters for the adsorption of CAF, LTG, and CBZ onto CAC.

	10 mg/L			20 mg/L			30 mg/L
	CAF	LTG	CBZ	CAF	LTG	CBZ	CAF	LTG	CBZ
tb (min)	255	255	255	45	45	45	108	78	108
ts (min)	435	435	435	360	360	360	318	318	318
madsorb	1.20	1.34	1.02	0.95	0.91	0.88	3.61	3.43	3.40
mtotal	3.69	4.17	3.56	3.06	3.45	2.95	8.39	7.91	7.47
q (mg/g)	0.60	0.67	0.51	0.47	0.45	0.44	1.81	1.72	1.70
R%	32.69	32.10	28.79	31.19	26.39	29.91	43.02	43.36	45.51

The effect of bed material

Figure 7 shows the breakthrough curves for different bed materials, including carbonate, Ottawa sand, and carbonate -Ottawa sand mixture (50:50 w/w) at a flow rate of 1 mL/min and a CAC dosage of 2 g. The results showed that the mixed bed (carbonate-sand) exhibited the longest breakthrough time indicating more efficient retention of the pharmaceuticals. As presented in Table 11, the breakthrough time for the mixed bed were observed at 107 min, 62 min, and 77 min for CAF, LTG, and CBZ, respectively. In comparison, breakthrough time in carbonate bed observed at 32 min for CAF and CBZ and 47 min for LTG, while the sand bed exhibited a breakthrough time of 45 min.

Fig. 7.

The breakthrough curves at different bed materials: (a) CAF, (b) LTG, and (c) CBZ.

Table 11.

GW characterization.

Test	Result
pH	7.91
TDS (mg/L)	3681.28
EC (µS/cm)	5752

Although the mixed bed exhibited long breakthrough time, its saturation time was faster than carbonate and sand. The saturation time of carbonate, sand and mixed bed were 437 min, 360 min, and 377 min, respectively.

The enhanced performance of the mixed bed can be attributed to the synergistic interaction between the two bed materials, carbonate enhances surface interactions with the pharmaceuticals, while Ottawa sand contributes to hydrodynamic stability, ensuring uniform flow distribution and reducing the risk of clogging. Similar results of mixed granular media on improving adsorption kinetics and flow uniformity have been reported in previous studies^44,45.

The sandstone carbonate combination had a longer breakthrough time (C/C₀ ≈ 0.1), as listed in Table 10, which was observed at 107, 62, and 77 min for CAF, LTG, and CBZ. However, carbonate alone was observed at 32 min for CAF and CBZ and 47 min for LTG. While the breakthrough time for the column filled with Ottawa sand alone was determined at 45 min. Despite having the longest breakthrough time, the sand and carbonate mixture had a faster saturation time than carbonate by itself. For carbonate, Ottawa sand, and the two together, the saturation periods that corresponded to (C/C₀ ≈ 0.85) were 437, 360, and 377 min. Because carbonate interacts chemically with pollutants to enhance adsorption, the adsorption capacity of the two types combined in one column was higher than that of the two types alone^44,45. While carbonate promotes chemical adsorption, Ottawa sand facilitates uniform water flow and avoids clogging.

Table 10.

The effect of bed type on the breakthrough parameters for the adsorption of CAF, LTG, and CBZ onto CAC.

	Carbonate			Ottawa sand			Carbonate + Ottawa sand
	CAF	LTG	CBZ	CAF	LTG	CBZ	CAF	LTG	CBZ
tb (min)	32	47	32	45	45	45	107	62	77
ts (min)	437	437	437	360	360	360	377	377	377
madsorb	1.77	1.64	1.05	2.21	1.76	1.83	2.73	2.23	2.04
mtotal	7.25	7.56	4.41	7.09	6.66	6.12	6.25	6.52	5.24
q (mg/g)	0.89	0.82	0.52	1.10	0.87	0.92	1.37	1.12	1.02
R%	24.43	21.69	23.71	31.19	26.9	29.91	43.67	34.27	39.05

Treatment of real ground water

The breakthrough curves of the real groundwater spiked with the targeted pharmaceuticals are depicted in Fig. 8. Groundwater characterization results for pH, TDS, and Electrical Conductivity are shown in Table 11. The treatment experiment was conducted under the optimized operating conditions established based on the findings from the previous experiments. A control column (without CAC) exhibited insignificant removal, confirming that adsorption was attributed to the CAC. The optimum operating conditions were determined as follow; bed type (50:50 w/w) mixture of sand and carbonate, flowrate 1 mL/min, CAC dosage 2 g and initial pharmaceutical concentration 20 mg/L. Under these conditions the CAC demonstrated excellent adsorption performance. As shown in Table 12, the breakthrough time was observed at 132 min, while saturation time reached at 432 min and pharmaceutical adsorption capacity of 6 mg/g.

Fig. 8.

The breakthrough curves of the real groundwater treatment.

Table 12.

Breakthrough parameters for the treatment of real groundwater.

	Real groundwater
	CAF	LTG	CBZ
tb (min)	132	132	132
ts (min)	(C/C0 ≈ 0.85)
	432	432	432
madsorb	3.20	3.06	2.90
mtotal	7.51	6.82	6.69
q (mg/g)	1.60	1.52	1.45
R%	42.63	44.80	43.35

Conclusions

This study examined the utilization of a commercial CAC for the removal of certain pharmaceuticals (CAF-LTG-CBZ) through a series of fixed-bed studies. CAC was characterized using different instruments, including SEM/EDX, BET surface area, Zeta potential, and particle size distribution (PSD). SEM demonstrated that CAC has a porous surface structure, and its chemical composition (mainly carbon) was determined by EDX. The BET surface area was determined to be 1112.24 m²/g, and the zeta potential showed that CAC has a negative surface charge. The size distribution of CAC particles ranges between 6.68 and 39.44 μm for original CAC and between 6.94 and 38.51 μm for the diluted CAC. Effects of flow rate (mL/min), CAC dosage (g), initial concentration (mg/L), and bed type (50%:50%Ottawa sand to carbonate) were examined in fixed-bed column experiments. The fixed-bed column breakthrough curves of the best conditions were fitted using the Dose-Response model. With a correlation coefficient (R²) close to 1. The best conditions were determined to be at a flow rate of 1 mL/min, 2 g of CAC dose, an initial concentration of 20 mg/L, and mixture of 50:50 Ottawa sand: carbonate of bed material, with breakthrough and saturation times of (C/C₀ = 0.1) and (C/C₀ ≈ 0.85), respectively, and a removal efficiency of about 40%. The large surface area and microporous structure of CAC contribute to its strong adsorption performance, highlighting its potential for efficient in situ groundwater remediation. The CAC can be in-situ injected in subsurface to form permeable reactive barrier that prevents the pollutants from dispersing with groundwater movement (plume control).

Author contributions

Shahd Alghamdi performed the investigation and experimental work, conducted data analysis including kinetics and isotherms, and wrote the first draft of the manuscript. Bassam Tawabini contributed to conceptualization, supervision, data validation and curation and participated in editing and reviewing the manuscript. Abdullah Basaleh contributed to investigations, kinetics and break through curves modeling, conceptualization, and reviewing and editing and the manuscript. Mohammed Benaafi provided resources, contributed to data validation and assisted in reviewing and editing the manuscript. Shehzada Muhammad Sajid Jillani contributed to data analysis, analytical investigations and participated in reviewing and editing and the manuscript.

Data availability

The datasets generated during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.