CO₂ Reactivation of Activated Carbon to Improve Its Adsorption Capacity and Rate Toward Chlorpheniramine

Abstract

The adsorption of chlorpheniramine (CPA) on a series of activated carbon (AC) samples was investigated. Commercial AC Megapol M (MM) samples were reactivated with CO₂ at 800 °C for different times. The ACs were designated as MM4, MM8, and MM8A, corresponding to 4, 8, and 8 h cumulative (4 h and 4 h), respectively. The textural properties of MM8A were the highest due to an additional CO₂ reactivation process. Quantification of the carboxylic sites showed a decrease in the order MM > MM8A > MM4 > MM8. The Raman spectra of MM, MM4, MM8, and MM8A indicated that for longer CO₂ reactivation times, the D to G band intensity ratio (I_D/I_G) of all ACs increased due to surface defects. The CPA-adsorption capacity of the ACs revealed that MM8A had the highest adsorption capacity, attributed to its low density of carboxylic sites and disordered structure. Increasing the pH solution enhanced the CPA adsorption on MM8A, while temperature had a minor effect. The isosteric heat of adsorption indicated the CPA adsorption on MM8A occurred via physical interactions, with π–π stacking and hydrophobic interactions governing the process at pH = 11. The rate of CPA adsorption on MM8A was studied using diffusional models. The external mass transport model satisfactorily estimated the experimental data. Finally, it was found that CPA adsorbed more rapidly on MM8A than on MM.

1. Introduction

The existence of emerging contaminants (ECs) in water bodies, like pharmaceutical compounds, personal care products, endocrine disruptors, and pesticides, has become a serious issue because they are associated with cancer, reproductive, and endocrine disorders [1,2]. In addition, their accumulation in the food chain is a substantial risk to both the environment and human health [3]. The pervasiveness of ECs is related to their excessive use, ineffective removal from wastewater, and structural stability [4]. With the increasing demand for pharmaceutical compounds to meet requirements of humans and animals, ECs have been frequently detected in water bodies, representing a latent risk to human health and environment [5]. Chlorpheniramine (CPA), also known as chlorphenamine, is a nonprescription antihistamine with several clinical uses. For instance, it is used to treat the common cold, asthma, urticaria, depression, and, more recently, for the respiratory disease provoked by SARS-CoV-2 virus [6]. Although low concentrations of CPA (in nanograms per liter) have been frequently detected in effluents, investigations focused on its ecotoxicity and studies on the chronic responses of human body are limited [7]. CPA adversely affects human health, for example, via serotonergic effects due to overdose [8]. In addition, CPA is a potential precursor of N-nitrosodimethylamine (NDMA), a carcinogenic molecule [9]. Thus, CPA poses a threat to the environment and human health. Therefore, it is important to remove CPA from water to safeguard humans and the environment.

Advanced oxidation and ozonation are some common treatments used for CPA removal. However, both treatments promote the formation of NDMA as a byproduct of CPA degradation, thus complicating their application [9,10]. Against this backdrop, adsorption has been used as an alternative treatment method considering its ease of operation, efficiency, and lack of byproducts [11].

Carbon-based materials are the most commonly used adsorbents for pharmaceutical compounds. In particular, activated carbon (AC) has been extensively studied from the viewpoint of adsorption of pharmaceutical compounds owing to its exceptional characteristics such as large surface area, well-developed porosity, adequate pore size distribution, affinity for organic compounds, and surface chemistry favorable for adsorption [12]. Adsorption of a wide variety of pharmaceutical compounds on ACs has been studied [12,13,14]. CPA adsorption on ACs has also been investigated. For example, a high-surface-area AC prepared from date palm leaflets exhibited an improved adsorption capacity as the solution temperature increased (117.6 mg/g at 45 °C) [15]. In another work, an AC obtained from agricultural wastes was used in CPA adsorption. The behavior of CPA adsorption was influenced by solution pH, and the highest adsorption capacity was 72.41 mg/g at pH = 8 [16]. In addition, a granular AC with acidic character presented a CPA-adsorption capacity of ~195.4 mg/g. The adsorption capacity was improved with rising pH, and the adsorption mechanism was ascribed to π–π stacking interactions [17]. To a similar extent, an AC made of date palm leaflets was superficially modified using ethylamine. The modification produced a hydrophobic surface and enhanced the affinity of the AC toward CPA (q = 455 mg/g) due to π–π stacking interactions [18]. These studies emphasized that CPA adsorption is greatly affected by the solution pH and that the main driving force for CPA adsorption is π–π stacking interactions.

Despite the good adsorption capacity of ACs toward pharmaceutical compounds, diverse approaches, such as chemical and physical modifications, have been used to modify ACs. Chemical modifications were adopted to modify the surface chemistry of an AC (F400) using HNO₃ (F400-HNO₃), and its effect on the adsorption capacity for metronidazole was examined. F400-HNO₃ had more acidic sites than its pristine counterpart (F400). Nevertheless, the adsorption of metronidazole decreased as the concentration of carboxylic sites on the AC surface increased [13]. Regarding physical approaches, Moral-Rodríguez and collaborators modified a commercial AC via CO₂ activation with various activation times to improve the textural properties and capacity of the AC to adsorb diclofenac. As expected, the specific surface area, total volume, and micropore volume of the AC increased with activation time, resulting in a higher adsorption capacity for diclofenac. An increase in activation time remarkably enhanced the surface area; consequently, the adsorption capacity increased because more basal planes of the AC were available for interacting with diclofenac via π–π interactions [19]. These studies highlighted that CO₂ activation is a suitable method for improving the adsorption capacity of ACs toward pharmaceutical compounds.

Conversely, the rational design of adsorption systems requires information on the adsorption velocity and capacity of adsorbents [20]. The adsorption velocity of pharmaceutical compounds on ACs was modeled using diffusion and kinetic models [21]. In diffusion models, the overall adsorption velocity is governed by molecular diffusion in the pore volume, surface, or both [22]. Nonetheless, to the best of our knowledge, the modeling of CPA-adsorption velocity on CO₂-reactivated ACs using diffusion models has not been explored as of yet.

This work aims to study the adsorption rate and capacity for adsorbing CPA on a commercial AC without modification and modified by CO₂ reactivation. The AC was modified to enhance the adsorption capacity and rate. First, the pristine and modified ACs were characterized by different analytical techniques. Subsequently, the dependence of the CPA-adsorption capacity on the textural and physicochemical properties of the ACs was examined comprehensively. The effect of the solution pH and temperature upon the adsorption capacity of the AC exhibiting the highest adsorption capacity was investigated in detail. Finally, the adsorption rate of CPA on ACs was analyzed using diffusional models.

2. Diffusional Model

The CPA-adsorption velocity was interpreted by diffusion models. These models are derived based on mass balance and under the assumption that the adsorption process undergoes three sequential stages [22]: transport of the substance to the external surface of the particle, diffusion within the particle, and adsorption on the particle’s active sites.

2.1. External Mass Transfer Model

The external mass transport model (EMTM) [23] was used in this study. This model considers the solute mass transport from the bulk solution to the outer surface of the adsorbent as the stage controlling the overall adsorption rate. In addition, the following assumptions were made: there are no concentration gradients inside the adsorbent particles. This assumption is true when the solute undergoes fast diffusion within the particles and instantaneous adsorption on an active site. To describe this situation, a mass balance in the solution was performed. The differential equation and its corresponding initial condition are as follows [23]:

$$V{\displaystyle \frac{{dC}_A}{dt}}=-mS{k}_L(C_A-C_{Ar}{|}_{r=R})$$

(1)

$$IC: t=0; C_A=C_{A0}$$

(2)

where C_A is the CPA concentration in the water solution (mg/L); C_A0 is the initial concentration of CPA in the water solution (mg/L); C_{Ar|r = R} is the CPA concentration in the solution at the AC external surface r = R (mg/L); k_L is the external mass transfer coefficient in the liquid phase (cm/s); m is the mass of AC (g); S is the external area of AC per unit mass (cm²/g); t is time (s); and V is the volume of the CPA solution (L).

The first term of Equation (1) indicates the accumulation rate of CPA in the solution, while the second term indicates the concentration decay as CPA is transported from the bulk solution to the adsorbent external surface (r = R).

Likewise, it is necessary to conduct a mass balance of the contaminant in the adsorbent using the following equation and initial conditions:

$${\displaystyle \frac{m{\epsilon }_p}{{\rho }_p}}{\displaystyle \frac{d{C}_{Ar}}{dt}}+m{\displaystyle \frac{dq}{dt}}=mS{k}_L(C_A-C_{Ar}{|}_{r=R})$$

(3)

$$IC: t=0; C_{Ar}=q=0$$

(4)

where q stands for the mass of CPA adsorbed per unit mass of AC (mg/g); ε_p is the porosity or void fraction of AC; and ${\rho }_p$ is the particle density of AC (g/cm³).

This equation describes the accumulation of the contaminant in the pore volume and surface of the adsorbent, and it is equal to the mass transport of contaminant from the solution bulk to the external surface of the adsorbent.

2.2. Pore Volume and Surface Diffusion Model

The pore volume and surface diffusion model (PVSDM) is the general model, considering that intraparticle diffusion is due to both pore volume diffusion and surface diffusion. The PVSDM is derived by performing a mass balance of CPA in an aqueous solution, represented by Equations (1) and (2). Subsequently, the mass balance of CPA is carried out in a differential element of the adsorbent particles, which are assumed to be spherical. This equation is shown below:

$${\epsilon }_p{\displaystyle \frac{\mathsf{\partial }{C}_{Ar}}{\mathsf{\partial }t}}+{\rho }_p{\displaystyle \frac{\mathsf{\partial }q}{\mathsf{\partial }t}}={\displaystyle \frac{1}{r^2}}{\displaystyle \frac{\mathsf{\partial }}{\mathsf{\partial }r}}\left[r^2\left(D_{ep}{\displaystyle \frac{\mathsf{\partial }{C}_{Ar}}{\mathsf{\partial }t}}+D_{es}{\rho }_p{\displaystyle \frac{\mathsf{\partial }q}{\mathsf{\partial }r}}\right)\right]$$

(5)

where C_Ar is the CPA concentration in the solution inside the AC pores at a distance r (mg/L); D_ep is the effective pore volume diffusion coefficient (cm²/s); D_es is the surface diffusion coefficient (cm²/s); and r is the radial distance in the AC particle (cm).

The first term in the left side of Equation (5) represents the CPA accumulation within the volume of the pores, while the second term indicates CPA accumulation on the surface of the pores. The right side of the equation expresses the CPA intraparticle diffusion occurring in the volume and on the surface of the pores.

The partial differential Equation (5) is solved using the following initial and boundary conditions.

$$IC: t=0; C_{Ar}=0; 0\le r\le R$$

(6)

$$BC: {\left.{\displaystyle \frac{\mathsf{\partial }{C}_{Ar}}{\mathsf{\partial }r}}\right|}_{r=0}=0$$

(7)

$${\left.{BC: D}_{ep}{\displaystyle \frac{\mathsf{\partial }{C}_{Ar}}{\mathsf{\partial }r}}\right|}_{r=R}+{\left.D_{es}{\rho }_p{\displaystyle \frac{\mathsf{\partial }q}{\mathsf{\partial }r}}\right|}_{r=R}=mS{k}_L(C_A-C_{Ar}{|}_{r=R})$$

(8)

where R is the radius of AC particles (cm).

The PVSDM assumes that CPA adsorption on an active site proceeds rapidly. Thereby, CPA concentration in the solution inside the AC pores is in equilibrium with the mass of CPA adsorbed on the surface of pores. This equilibrium is described accurately by the isotherm model (Equation (9)), which expresses the relation between q and $C_{A_r}$.

$$q=f\left(C_{Ar}\right)$$

(9)

The PVSDM can be simplified to the pore volume diffusion model (PVDM) by assuming that the intraparticle diffusion is governed by the pore diffusion (D_es = 0 and D_ep ≠ 0) or into the surface diffusion model (SDM) by assuming that the intraparticle diffusion is controlled by surface diffusion (D_ep = 0 and D_es ≠ 0). The D_ep can be evaluated by the following mathematical relationship:

$$D_{ep}={\displaystyle \frac{D_{AB} {\epsilon }_P}{\tau }}$$

(10)

where D_AB is the coefficient of molecular diffusion of CPA in water (cm²/s) and τ is a factor that corrects the path of the diffusion of CPA molecules inside the pore volume of an AC. Assuming that the diffusion trajectories vary and do not follow straight lines, τ must take values greater than 1. Various experimental studies suggest that τ = 3.5 is suitable for ACs [22].

3. Materials and Methods

3.1. CO₂-Reactivation of Activated Carbons

The commercial powdered wood-based AC Megapol M (MM) was supplied by Carbotecnia, Zapopan, Jalisco, Mexico, and the particle diameter was 0.0075 cm. The MM AC was modified by CO₂ activation at 800 °C (ramp of 5 °C/min) over various activation times in a furnace (Carbolite Gero 30–3000 °C, Hope Valley, UK) under a constant CO₂ flow (100 cm³/min) through the furnace from the beginning; no other gas was supplied. Three ACs were obtained after activation times of 4, 8, and 8 h cumulative (4 h and 4 h), and designated as MM4, MM8, and MM8A, respectively. MM4 and MM8 were obtained as follows: (1) Some mass of MM was introduced into the furnace reactor, and the CO₂ stream was passed through the reactor immediately. (2) The furnace temperature increased evenly up to 800 °C; once this temperature was attained, the timer was switched on. (3) The furnace was turned off when the CO₂ activation time elapsed, and it was cooled down to room temperature (~25 °C) while the flow of CO₂ was maintained. For obtaining MM8A, some mass of MM4 was placed in contact with air for a few hours before subjecting it to a new CO₂ reactivation run; subsequently, this sample was reintroduced into the furnace reactor and subjected to steps 1, 2, and 3 to convert it from MM4 to MM8A.

3.2. Chemical Reagents

CPA maleate salt, acquired from Sigma–Aldrich, St. Louis, MO, USA, was used to conduct adsorption experiments.

Table 1. Physical–chemical properties of CPA.

Molecular Structure	Dimensions (nm)	pKa	Solubility (g/L)	DAB × 106 (cm2/s)
	x = 1.25 y = 0.54 z = 1.44	pKa1 = 4.0 pKa2 = 9.2	5.5	4.54

summarizes the physicochemical properties of CPA. Figure 1a,b show the chemical structure and species distribution diagram of CPA.

3.3. Characterization of Modified Activated Carbons

The textural properties of MM, MM4, MM8, and MM8A were characterized by interpreting their N₂ adsorption–desorption isotherms at −196 °C. These isotherms were obtained using a N₂-physisorption analyzer instrument (ASAP 2020 model, Micromeritics, Norcross, GA, USA). Prior to the analysis, all adsorbents underwent a vacuum treatment to remove moisture. The Brunauer–Emmett–Teller and Dubin–Radushkevich models were used to compute surface area (S_BET) and micropore volume (V_mic), respectively [24,25]. The pore volume (V_p) and average diameter (D_p) were computed following the methods suggested by Rouquerol et al. [26]. The mesopore volume of the ACs were determined using the Barrett–Joyner–Halenda (BJH) method [27]. Finally, the pore size distribution of ACs was computed using density functional theory (DFT).

The solid density (ρ_S) of MM and MM8A was evaluated using a helium pycnometer (Accupic 1330, Micromeritics, Norcross, GA, USA). The ε_p and ρ_p were estimated using the following equations:

$${\epsilon }_p={\displaystyle \frac{V_p}{V_p+\frac{1}{{\rho }_s}}}$$

(11)

$${\rho }_p={\displaystyle \frac{{\rho }_s}{1+V_p{\rho }_s}}$$

(12)

where V_p is the total pore volume of the adsorbent (cm³/g) and ${\rho }_s$ is solid density of AC (g/cm³).

To quantify the active site concentrations in the surfaces of MM, MM4, MM8, and MM8A, the Boehm titration method was employed using a previously developed procedure [28]. In summary, 0.4 g of adsorbent was placed in contact with neutralizing solutions in centrifuge tubes, which were agitated for 15 min daily, and the temperature was kept constant for 7 days.

A titration procedure using 0.01 N NaOH or HCl solutions was conducted to quantify the total acidity and basicity of the adsorbents; NaHCO₃ and Na₂CO₃ were used as neutralizers to determine the carboxylic and lactonic sites, respectively. Phenolic sites were determined from the difference between the total acidic sites and the summation of the carboxylic and lactonic sites.

The charge distribution on the surface and point of zero charge (pH_PZC) of MM, MM4, MM8, and MM8A were obtained from potentiometric curves. Briefly, 11 neutralizing solutions were prepared by adding different aliquots of 0.01 N NaOH or HCl in a volumetric flask (50 mL); 0.01 N NaCl was added to fill up to the mark of the flask. Subsequently, 0.05 g of MM, MM4, MM8, or MM8A was incorporated to 50 mL centrifuge tubes, and 40 mL of the neutralizing solutions were added to the centrifuge tubes. The rest of the neutralizing solutions were used as control. The centrifuge tubes were incubated in a constant-temperature bath for 7 days and agitated daily. The final pH values of the control and the AC-containing solutions were plotted versus the respective volume of the neutralizing solutions. Surface charge and pH_PZC were computed using an equation disclosed in a previous work [29].

The zeta potential measurements of bare MM8A and MM8A saturated with CPA (MM8A-CPA) were performed in a Zetasizer instrument (Malvern-Panalytical Ltd., Malvern, UK). In short, various dispersions of MM8A and constant ionic strength solutions (0.01 N) with a ratio of 0.1 g per 20 mL were prepared in centrifuge tubes. The pH values of the solutions were adjusted to be in the range of 2–11 and were maintained for 7 days by adding an appropriate number of drops of 0.01 N HCl or NaOH. The centrifuge tubes were maintained at 25 °C and stirred for 7 days. Similarly, the zeta potential measurements of MM8A-CPA with initial CPA concentrations of 500 and 750 mg/L were performed at pH values of 5, 7, 9, and 11. After 7 days, 1 mL aliquots of MM8A and MM8A–CPA dispersions were transferred into a sample cuvette to measure their zeta potential. The mass of CPA adsorbed on MM8A was evaluated.

The surface morphologies of MM and MM8A particles were examined using a scanning electron microscope (SEM), model JEOL (JSM-6610LV, Akishima, Japan). The Raman spectra of MM, MM4, MM8, and MM8A were obtained using a Raman spectrometer (XploRA Plus, Kyoto, Japan) using the 532 nm Ar line as an excitation source. Raman spectra were collected in the range of 800–2000 cm⁻¹. The D and G intensity ratio (I_D/I_G) was computed using the maximum intensity of the D and G bands at 1345 and 1600 cm⁻¹, correspondingly.

3.4. Determination of CPA in Aqueous Solutions

To quantify the CPA concentration in aqueous solutions, ultraviolet–visible light (UV–Vis) spectrophotometry was employed. The absorbance of the samples at the UV wavelength of 261 nm was estimated using (Shimadzu 2600, Tokyo, Japan) UV–Vis spectrophotometer. Several calibration curves were constructed using CPA stock solutions (1–50 mg/L) in the pH range 5–11 and used to compute the CPA concentration.

3.5. Adsorption Equilibrium Experiments

CPA-adsorption experiments were conducted in batch adsorbers (50 mL centrifuge tubes). To prepare the CPA solutions (50–1500 mg/L), an aliquot of CPA stock solution (1500 mg/L) was added into a volumetric flask (50 mL), diluted with a constant ionic strength solution, and adjusted to a specific pH by adding reasonable volumes of 0.01 N HCl and NaOH solutions. Subsequently, 0.05 g of MM, MM4, MM8, or MM8A was added to the batch adsorbers (50 mL centrifuges tubes); thereafter, the CPA solutions (50–1500 mg/L) were added, and the pH was kept constant using drops of 0.01 N HCl or NaOH solutions. The adsorbers were placed in a temperature-controlled bath at a stable temperature for 7 days and stirred daily for 15 min to reach equilibrium.

To quantify the CPA concentration at equilibrium, all the samples were centrifuged and filtered, obtaining 10 mL aliquots for analysis. The mass adsorbed by CPA was computed using the following equation:

$$q={\displaystyle \frac{V}{m}} \left(C_0-C_e\right)$$

(13)

where C_e stands for the equilibrium concentration of chlorpheniramine (mg/L), and C₀ is the initial concentration of CPA (mg/L).

The experimental error associated with determining the concentration of CPA in water solution was less than 2.0%. This ensured that the overall experimental error in estimating the mass adsorbed of CPA was under 4%.

3.6. Adsorption Rate Experiments

The experimental concentration decay curves data of CPA adsorption on ACs were procured in the stirred tank batch adsorber shown in Figure 2. The system consisted of a 250 mL acrylic funnel (1), in which 200 mL of CPA solution of known concentration and pH and 0.1 g of MM or MM8A were added. The CPA solution and the adsorbent were mixed using a propellant stirrer (2) powered by a motor with variable speed (3) set up at 300 rpm. The temperature of the solution in the adsorber was maintained constant by recirculating water through a glass coil submerged in the adsorber solution. The pH of the CPA solution was measured using a potentiometer (4), and the initial pH was kept constant by adding drops of 0.1 N HCl or NaOH as needed. To keep the mass of the adsorbent constant, a 4.5 cm diameter cellulose acetate membrane with a 0.2 μm pore opening (5) was placed at the bottom of the funnel. The solution in the adsorber was sampled by filtering a specific volume of the CPA solution through the membrane using a vacuum pump (6). The filtered sample was collected in a centrifuge tube (7) inside a vacuum flask (8). A peristaltic pump (9) recirculated the remaining solution in the sampling line (10). The experiments were initiated when the peristaltic pump was turned on to circulate the CPA solution into the sampling lines, and the timer was switched on immediately. During the experiments, 0.5 mL of solution samples were collected at specific times, and the CPA concentration of the samples was computed from the dilution ratios and using the method described in Section 3.4. The experiments lasted 30 min for both MM8A and MM, as the CPA concentration in the solution remained constant beyond this time, indicating that equilibrium was reached.

4. Results and Discussion

4.1. Textural Properties of Modified Activated Carbons

Table 2. Textural properties of the activated carbons.

AC	${\mathbf{S}}_{\mathbf{BET}}$ (m2 g−1)	${\mathbf{S}}_{\mathbf{mic}}$ a (m2 g−1)	Smeso (m2 g−1)	${\mathit{V}}_{\mathit{p}}$ b (cm3 g−1)	${\mathbf{V}}_{\mathbf{mic}}$ c (cm3 g−1)	Vmeso d (cm3 g−1)	${\mathbf{d}}_{\mathbf{p}}$ e (nm)	${\mathit{L}}_0$ f (nm)	${\mathit{\rho }}_{\mathit{p}}$ (g/cm3)	${\mathit{\epsilon }}_{\mathit{p}}$
MM	1107	984	123	1.15	0.43	0.71	4.15	0.87	0.57	0.66
MM4	1048	898	150	1	0.43	0.59	3.82	0.96	-	-
MM8	474	152	320	0.78	0.25	0.49	6.58	3.24	-	-
MM8A	1393	1343	50	1.32	0.58	0.75	3.79	0.86	0.57	0.75

Notes: ^a Micropore surface area calculated using Dubinin–Radushkevich method [25]. ^b Total pore volume determined at P/P₀ ≈ 0.99. ^c Micropore volume calculated by Dubinin–Radushkevich method [24]. ^d Mesopore volume calculated using Barrett–Joyner–Halenda model [27]. ^e Mean pore diameter [24]. ^f Mean micropore width estimated with Stoeckli equation [25].

lists the textural properties of MM, MM4, MM8, and MM8A. The N₂ adsorption/desorption isotherms of MM, MM4, MM8, and MM8A are shown in Figure 3. All the ACs showed a pronounced step in the micropore zone, and a capillary condensation step was observed in all the isotherms. These characteristics are typical of the type IIb isotherm according to IUPAC classification. In addition, this type of isotherm shows a hysteresis loop of type H4 associated with the filling of slit-shaped pores in ACs constituted predominantly of micropores and a few mesopores [26,30].

The values of S_BET, V_p, micropore surface area (S_mic), and micropore volume (V_mic) of MM, MM4, MM8, and MM8A varied with activation time. Pristine MM had an S_BET of 1107 m²/g, which reduced to 1048 and 474 m²/g after 4 h (MM4) and 8 h (MM8) of activation, respectively. The V_p values of MM, MM4, and MM8 were 1.15, 1.0, and 0.78 cm³/g, respectively. S_mic and V_mic decreased in the order MM > MM4 > MM8, possibly due to the reverse Boudouard reaction (${\mathrm{C}}_{\mathrm{s}}+{\mathrm{CO}}_{2(\mathrm{g})}\to 2\mathrm{C}{\mathrm{O}}_{\left(\mathrm{g}\right)}$) [31], in which the reactivation time and gasification temperature fostered the release of volatiles, leading to the formation of some wider mesopores and new micropores to a certain extent, because after a given period of activation time, the continuous devolatilization eventually induces the collapse of the pore walls of ACs and causes a reduction in S_BET, V_p, S_mic, and V_mic [31,32]. The increase in the mesoporous surface area (S_meso) in the order MM < MM4 < MM8 corroborates this tendency, as reported in various works on the two-step CO₂ reactivation of ACs [31,33,34]. Compared to MM, MM4, and MM8, MM8A has remarkably higher S_BET, V_p, S_mic, and V_mic values since it was subjected to a different CO₂ reactivation process (MM → MM4 → MM8A). In particular, S_mic increased drastically. The micropore development mechanism was as follows: (1) The microporous structure of the AC (MM → MM4) decreased in the first run of CO₂ reactivation, and some mesoporous structures were formed by the enlargement of the micropores. (2) Subsequently, the surface of MM4 was oxidized in an air atmosphere prior to the second run of CO₂ reactivation, and new oxygen-containing functional groups were created. (3) When MM4 was reintroduced to the furnace reactor and heated in the CO₂ stream, some surface functional groups were removed, and the CO₂ gasification reaction eventually generated new micropores. The previous mechanism of micropore formation was proposed by Lawtae and coworkers [35]. In addition, the removal of oxygen groups from the edges of basal planes and the formation of micropores can lead to a higher degree of disorder in the ACs, which is discussed later.

The pore size distributions of MM, MM4, MM8, and MM8A are also shown in Figure S1, and the pore size distributions are bimodal, indicating that the ACs were mainly constituted of micropores (d_p > 2 nm) and mesopores (2 nm > d_p > 50 nm)). This figure corroborates the formation of micropores in MM8A and MM4 and the decrease of micropores in MM8. Noticeably, the increase in the micropores was more pronounced for MM8A carbon.

4.2. Quantification of the Active Sites on MM, MM4, MM8, and MM8A

Table 3. Physical–chemical characteristics of MM, MM4, MM8, and MM8A.

AC	Total Acidic Sites (meq/g)	Total Basic Sites (meq/g)	Carboxylic Sites (meq/g)	Lactonic Sites (meq/g)	Phenolic Sites (meq/g)	pHPZC	pHPIE
MM	0.626	0.002	0.456	0.086	0.083	2.9	-
MM4	0.396	0.142	0.141	0.075	0.179	3.6	-
MM8	0.349	ND	0.136	0.072	0.141	4.4	-
MM8A	0.351	ND	0.169	0.065	0.117	3.2	3.1

Note: ND = Not determined.

lists the results of quantification of active sites by Boehm’s titration method. The total acidic sites decreased with CO₂ reactivation time in the following order: MM > MM4 > MM8 ≈ MM8A. The carboxylic sites were the prevalent functional group on the surface of MM, and after CO₂ reactivation, the number of carboxylic sites decreased by ≈30% for MM4 and MM8. However, MM8A (0.169 meq/g) had slightly more carboxylic sites than MM4 (0.141 meq g⁻¹) and MM8 (0.136 meq/g) because of the reoxidation of the surface prior to the second run of the CO₂ reactivation. The number of lactonic sites of MM, MM4, MM8, and MM8A was maintained in the range of 0.086–0.065 meq/g. Additionally, MM had an insignificant number of total basic sites; however, after 4 h of CO₂ reactivation (MM4), some basic sites were formed (0.142 meq/g). The formation of basic sites is attributed to the acidic sites that react with CO₂ gas and form hydroxide ions that subsequently react with carbon to generate basic sites [36]. Similarly, a few phenolic sites were formed during CO₂ reactivation.

4.3. Surface Charge Distribution and Point of Zero Charge of Activated Carbons

Figure 4 shows the charge distribution on the surfaces of MM, MM4, MM8, and MM8A as a function of solution pH. The pH_PZC of MM was 2.9; in other words, at pH < 2.9, the MM surface was positively charged, and at pH > 2.9, the MM surface was negatively charged owing to the deprotonation of its acidic sites, indicating the acidic nature of MM. For MM4, pH_PZC = 3.6 shifted to basic values because of the removal of the acidic sites and the formation of basic sites during CO₂ reactivation. For MM8, pH_PZC = 4.4 shifted further to basic values because of a greater loss of acidic sites. Meanwhile, for MM8A, pH_PZC = 3.2 shifted slightly to the right because of a smaller loss of carboxylic sites by the reoxidation of the surface. In general, the pH_PZC values of MM4, MM8, and MM8A shifted to the right, revealing that CO₂ reactivation modified the acidic nature of the ACs by reducing the number of acidic sites.

4.4. Surface Morphology and Raman Spectra of ACs

Figure S2a,b depict the SEM images of the surface of MM and MM8A, and the images of ACs were obtained at the same magnification ratio to facilitate the comparison of the surface morphology of both ACs. It can be observed that the irregular morphology of the MM particles is maintained in MM8A despite the 8 h of accumulated thermal treatment with CO₂. It can be noted that both ACs are comprised of small granules and the existence of particles with right-angled laminar shapes.

Detailed investigations revealed that Raman spectroscopy is suitable for obtaining structural information of CO₂-reactivated ACs [37,38]. In general, the Raman spectra of ACs show two typical bands located at around 1345 (D) and 1600 cm⁻¹ (G). The former band is associated with the breathing vibrations of sp2-bonded C atoms at the edges of basal planes (i.e., disordered graphitic lattices), while the latter band is related to the in-plane elongation of sp2-bonded C atoms in basal planes (optimal graphitic lattices). The ratio I_D/I_G reveals the degree of disorder in the AC structure [38]. The higher the I_D/I_G ratio, the more disordered is the structure of the AC, and vice versa. The Raman spectra of MM, MM4, MM8, and MM8A are shown in Figure 5a. The Raman spectrum of MM has an intense G band at 1590 cm⁻¹ and a flat D band at 1345 cm⁻¹. The I_D/I_G of MM was 0.83, indicating that the optimal graphitic lattices are more abundant than the defects on MM. The Raman spectrum of MM4 showed a shift in the D and G bands to lower values because graphitic lattices are less developed. The I_D/I_G (1.02) of MM4 indicates that MM4 has a high density of defects. Note that the Raman spectrum of MM8 indicates a shift in the G band to higher values, which may indicate a reduction in the number of disorders; this result is consistent with its I_D/I_G (0.98), which is slightly less than that of MM4. The MM8A Raman spectrum showed that the D band underwent the greatest shift to lower values. This result may be attributed to the higher number of edges in the basal planes, which is attributed to the formation of micropores and the removal of oxygen groups [38]. However, the I_D/I_G of MM8A did not change.

4.5. Adsorption Equilibrium of CPA on ACs

4.5.1. Adsorption Isotherms of CPA on MM, MM4, MM8 and MM8A

The CPA-adsorption equilibrium on ACs was interpreted using the Freundlich, Langmuir, and Radke–Prausnitz (R-P) isotherms, as shown in the following equations:

$$q={ k}_F{C}_{\mathrm{e}}^{1/n}$$

(14)

$$q={\displaystyle \frac{q_m{K}_L{C}_{\mathrm{e}}}{1+K_L{C}_{\mathrm{e}}}}$$

(15)

$$q={\displaystyle \frac{a{C}_{\mathrm{e}}}{1+b{C}_{\mathrm{e}}^{1-\beta }}}$$

(16)

where k_F represents the Freundlich isotherm constant (mg^1−1/n L^1/n/g); n represents the adsorption intensity in the Freundlich isotherm; K_L is a parameter of the Langmuir isotherm (L/mg); q_m stands for the maximum adsorption capacity (mg/g); and a (L/mg), b (L^1−β/mg^1−β), and β (L/mg) are constants of the R-P model.

The parameters of the isotherms were optimized using a nonlinear least-square procedure and the Levenberg–Marquardt algorithm implemented in Origin software (version 2021). Equation (17) was used to compute the average percentage deviation.

$$\%D={\displaystyle \frac{1}{N}}\sum _{i=1}^N\left|{\displaystyle \frac{q_{i,pr\mathrm{e}d}-q_{i,\mathrm{e}xp}}{q_{i,pr\mathrm{e}d}}}\right|\times 100\%$$

(17)

where N denominates the number of experimental data; q_i,pred and q_i,exp are the estimated and experimental uptake adsorbed of CPA (mg/g).

Table 4. Optimized parameters of the adsorption isotherms and their corresponding average percentage deviation values. Ionic strength = 0.01 N.

AC	pH	T (°C)	Freundlich			Langmuir			Radke–Prausnitz
			kF (mg1−1/nL1/n/g)	$\mathit{n}$	$\%\mathit{D}$	qm (mg/g)	KL (L/mg)	$\%\mathit{D}$	a (L/g)	b (L1−β/mg1−β)	β	$\%\mathit{D}$
MM	7	25	102.6	4.96	15.7	315.0	0.34	7.82	241.4	1.42	0.11	6.5
	11	25	34.36	2.10	12.8	706.4	0.01	3.75	6.85	0.97 × 10−2	0.0	3.8
MM4	7	25	90.06	3.45	22.0	515.5	0.047	4.85	24	0.047	0.0	4.9
	11	25	104.78	3.11	14.18	811.24	0.021	6.59	18.81	0.024	0.0	6.7
MM8	7	25	97.10	3.78	12.3	485.1	0.046	4.55	22.11	0.046	0.0	4.6
	11	25	55.10	2.46	17.66	642.44	0.020	8.36	12.75	0.020	0.0	8.4
MM8A	5	25	123.25	5.15	7.54	417.83	0.14	7.98	120.84	0.58	0.11	5.5
	7	15	90.96	3.44	16.60	607	0.027	7.98	16.00	0.026	0.0	7.9
		25	69.25	2.72	13.25	651.855	0.021	3.13	13.72	0.021	0.0	3.1
		35	80.06	3.13	7.79	641.87	0.021	5.63	13.99	0.021	0.0	5.6
	9	25	91.97	2.77	16.0	859.36	0.021	7.05	18.20	0.021	0.0	7.1
	11	25	111.86	2.99	14.75	881.54	0.024	5.84	21.83	0.027	0.01	6.0
Continuous Stirring
MM8A	7	25	138.89	4.30	5.58	650	0.029	13.67	64.75	0.31	0.17	1.5
MM	7	25	51.82	2.85	2.11	568.58	8.42 × 10−3	72.49	645.13	12.17	0.34	1.8

summarizes the optimized parameters for the Freundlich, Langmuir, and R-P isotherms, as well as their corresponding %D. It is worth pointing out that the R-P isotherm is a three-parameter isotherm, while the Freundlich and Langmuir isotherms are two-parameter models. The isotherm model with the lowest %D was chosen to interpret the experimental data. For the first twelve experimental conditions in Table 4, the R-P and Langmuir isotherms interpreted the experimental data reasonably well since the %D values for both isotherms were the lowest and were the same or slightly different. In addition, the β values for the R-P isotherm were β = 0 or approximately β ≈ 0. It is well known that the R-P isotherm simplifies to the Langmuir isotherm model when β = 0. Hence, CPA adsorption equilibrium was interpreted satisfactorily by the Langmuir model. The R-P isotherm was chosen for the last two experimental conditions because the %D values for the R-P isotherm were lower than those for the Langmuir isotherm.

4.5.2. Adsorption Capacity of MM, MM4, MM8, and MM8A Toward CPA

Figure 6a shows a comparative graph of the CPA-adsorption capacity of MM, MM4, MM8, and MM8A at pH = 7 and 25 °C. At this pH, CPA¹⁺ is the prevalent species, and all the ACs are negatively charged (see Figure 4), favoring the electrostatic attraction between CPA¹⁺ and the surface of ACs. At an equilibrium concentration of 450 mg/L, the amount of CPA adsorbed (Q₄₅₀) onto MM8A, MM4, MM8, and MM was 589.60, 492.00, 462.55, and 336.50 mg/g, respectively. Among the ACs, MM8A had the highest capacity for adsorbing CPA, which was 1.75, 1.27, and 1.20 times that of MM, MM8, and MM4, respectively.

Although the S_BET of ACs is closely associated with the adsorption capacity, there exists no linear relationship between the amount of CPA adsorbed and S_BET since S_BET increased in the following order: MM8 < MM4 < MM < MM8A. Meanwhile, the capacity for adsorbing CPA increased in the following order: MM < MM8 < MM4 < MM8A. Note that the density of the total acidic sites on the surface of ACs is crucial for determining their adsorption capacity. Figure 7a reveals the relationship between the amount of CPA adsorbed at an equilibrium concentration of 450 mg/L (Q₄₅₀) on the density of acidic sites of MM, MM4, MM8, and MM8A. As the concentration of acidic sites decreases, the adsorption capacity increases. However, there is a discrepancy with regard to the adsorption capacity of MM and MM8. Meanwhile, Figure 7b shows that a lower density of carboxylic sites on the surface of ACs improves CPA adsorption, and this tendency agrees well with the isotherms at pH 7 (Figure 6a). This behavior is attributed to the presence of carboxylic sites on ACs basal planes since they can disturb the electron cloud of aromatic rings, leading to poor π–π stacking interactions [39].

In previous works, some authors related the I_D/I_G to the AC adsorption capacity. For instance, Nguyen et al. [40] suggested that the higher I_D/I_G of an AC improved the adsorption of acetaminophen by establishing more π–π stacking interactions. In this context, Figure 5b shows the variation in the amount of CPA adsorbed on MM, MM4, MM8, and MM8A as a function of their I_D/I_G. The graph indicates that the CPA-adsorption capacity of the ACs augmented with the higher I_D/I_G, namely, possibly a higher disordered structure of Acs, favored their capacity for adsorbing CPA.

The CPA-adsorption isotherms corresponding to MM, MM4, MM8, and MM8A at pH 11 are shown in Figure 6b. At this pH, CPA neutral species prevail (see Figure 1b), while the surfaces of all ACs are negatively charged; accordingly, there is no electrostatic attraction. At an equilibrium concentration of 450 mg/L, the uptakes of adsorbed CPA (Q₄₅₀) on MM, MM4, MM8, and MM8A were 574.60, 708.16, 577.80, and 809.70 mg/g, respectively, representing a drastic improvement in their adsorption capacity with respect to their isotherms at pH = 7. These findings implied that the adsorption capacity of all ACs toward CPA is better when there is no electrostatic attraction.

4.5.3. Solution pH and Temperature Effects on the Adsorption Capacity of MM8A Toward CPA

The influence of temperature and solution pH on the adsorption capacity of MM for CPA was studied in detail since MM8A exhibited the best performance. The adsorption isotherms of CPA at pH 5, 7, 9, and 11 are shown in Figure 8a. The adsorption capacity increased with pH. For an equilibrium concentration of 450 mg/g, the Q₄₅₀ values for pH 5, 7, 9, and 11 were 410.4, 589.6, 777.73, and 809.71 mg/g, respectively. The change in pH from 5 to 7 increased the adsorption capacity 1.44 times. According to the species distribution of CPA, at pH = 5 and 7, CPA¹⁺ is the prevalent species. In addition, the magnitude of the negative charge of the MM8A surface increased as pH increased from 5 to 7, favoring electrostatic attraction toward CPA¹⁺. At pH = 9, the distribution of neutral CPA accounted for 40%; nevertheless, the adsorption capacity was 1.32 times that at pH = 7, suggesting that electrostatic interactions are not the only mechanism of CPA adsorption. Furthermore, the greatest amount of CPA was adsorbed at pH = 11 when the predominant species was neutral CPA (100%) and the surface of MM8A is negative, so that no electrostatic attractions were involved in CPA adsorption.

Figure 8b illustrates how temperature influences the CPA-adsorption capacity of MM8A. The rise in temperature from 15 °C to 25 °C and then to 35 °C resulted in similar adsorption capacities for MM8A. Moreover, the isosteric heat of adsorption was computed using Equation (18) and was used to verify whether an adsorption process occurs through physical (<40 kJ/mol) or chemical interactions (>40 kJ/mol).

$${\left({∆H}_{ads}\right)}_q={\displaystyle \frac{RLn\left(\frac{C_{\mathrm{e}1}}{C_{\mathrm{e}2}}\right)}{\frac{1}{T_1}-\frac{1}{ T_2}}}$$

(18)

where R denotes the ideal gas constant (8.314 J/mol K); C_e1 and C_e2 (mg/L) indicate the equilibrium concentration of CPA for T₁ and T₂ (K), correspondingly; and ${\left({∆H}_{ads}\right)}_q$ stands for the isosteric heat of adsorption (J/mol) at a specific mass of CPA adsorbed (mg/L).

The isosteric heat of adsorption ${\left({\Delta H}_{\mathit{ads}}\right)}_q$ was computed using the following data: a constant mass of CPA adsorbed (q) = 550 mg/g, and CPA equilibrium concentrations C_e1 = 345 mg/L and C_e2 = 274.64 mg/L for temperatures T₁ = 288.15 K and T₂ = 308.15 K, respectively. Under these conditions, ${\left({\Delta H}_{\mathit{ads}}\right)}_q$ was 8.42 kJ/mol, demonstrating that the CPA adsorption on MM8A occurred through physical interactions.

4.5.4. Mechanisms of CPA Adsorption on MM8A

The mechanisms of CPA adsorption on MM8A at pH = 5, 7, 9, and 11 were elucidated using zeta potential measurements. Figure 9a shows the isoelectric point (pH_IEP) of the MM8A is 3.1; the zeta potential of MM8A decreased as the solution pH increased. The zeta potentials of MM8A–CPA at various initial CPA concentrations (500 and 750 mg/L) were determined at pH = 5, 7, 9, and 11. Note that higher initial CPA concentrations resulted in less negative zeta potential values because the negative charge of the MM8A surface was balanced by the CPA-adsorbed molecules. The variation in the zeta potential (ΔZP) of MM8A–CPA (mV) concerning the mass of CPA adsorbed was estimated using Equation (19).

$$∆ZP={ \left(ZP\right)}_{MM8A-CPA}-{\left(ZP\right)}_{MM8A}$$

(19)

where ΔZP is the change in zeta potential (mV) due to the adsorption of CPA; (ZP)_MM8A-CPA and (ZP)_MM8A (mV) designate the zeta potential of CPA-saturated MM8A and MM8A without CPA, respectively.

The ΔZP of MM8A-CPA was the highest at pH = 5; the ΔZP was slightly reduced when the solution pH was 7 and further decreased as the pH increased to 9. This behavior demonstrates that electrostatic attractions played a fundamental role in the CPA-adsorption mechanism at pH = 5 and 7 but did not affect it at pH = 9 when π–π interactions occurred. At pH = 11, the zeta potentials of MM8A-CPA approach those of MM8A, proving that electrostatic attractions are not involved in the CPA-adsorption mechanism and that the π–π interactions between the basal planes of MM8A and the CPA molecules were enhanced by the neutralization of acidic sites and the loss of carboxylic sites during CO₂ reactivation; the loss of carboxylic sites in turn strengthened the interaction between the π electrons of the basal planes of MM8A and the π electrons of CPA [39,40].

Notably, ΔZP becomes negative or positive when the ZP decreases or increases depending on the amount of CPA adsorbed. Figure 9b shows the variation in ΔZP as a function of the CPA adsorbed. At pH = 5 and 7, ΔZP increases almost linearly with the mass of CPA adsorbed since CPA¹⁺ species adsorbed balances the negatively charged surface of MM8A. At pH = 9, the ΔZP has a slight slope, revealing that electrostatic attraction is still present. At pH = 11, ΔZP decreases slightly, reaching negative values. This behavior agrees with the result that electrostatic interactions do not play any role in CPA adsorption; therefore, π–π stacking interactions are a fundamental adsorption mechanism of CPA on MM8A at pH = 11.

Furthermore, note that π–π stacking interactions are not the only possible mechanism of CPA adsorption; in fact, the hydrophobic nature of the CPA molecule can foster its interactions with MM8A. Figure 10 shows the coefficient distribution (logD) diagram of CPA as a function of pH. The data of logD were predicted using the ChemAxon’s JChem software (version 14.12.15.0) [41]. The logD is the ratio of concentrations of species of a pharmaceutical compound in octanol to the concentrations of all its species in water. The logD diagram indicates how a compound with ionizable groups exists in the solution and its proportion at a given pH [42]. According to Figure 10, it can be noted that CPA exhibits hydrophilic affinity at pH values below 7, while it becomes hydrophobic at pH values above 7, and the hydrophobicity increases by raising the pH from 7 to 11 and remains constant at pH above 11. On the other hand, the hydrophobic nature of the MM8A surface is caused by the loss of oxygen functional groups (see Figure 7a,b). Hence, hydrophobic interactions between the CPA molecules and the MM8A surface are likely to occur. A comparison of Figure 8a and Figure 10 reveals that when the hydrophobicity of CPA was raised, the adsorption capacity of MM8A also increased, confirming that hydrophobic interactions were present in the CPA adsorption.

4.6. Rate of CPA Adsorption on MM8A

4.6.1. CPA-Adsorption Isotherm on MM8A and MM for Continuous Stirring

The adsorption equilibrium data were determined in a batch adsorber where stirring was discontinuous for 15 min daily (Section 3.5). The adsorption rate data were obtained in a stirred tank batch adsorber with continuous stirring. The stirring may decrease the particle size and affect the adsorption equilibrium. Hence, the adsorption equilibrium has to be determined in the stirred tank adsorber and was designated as adsorption isotherm for continuous stirring (CSt).

The adsorption equilibrium data for CSt were procured by performing experiments at different initial concentrations of CPA in the stirred tank batch adsorber and determining the final concentration of CPA at equilibrium, C_Ae, and the mass of CPA adsorbed at equilibrium, q_e. The adsorption equilibrium data for CSt of CPA on MM8A and MM at pH = 7 and 25 °C are listed in

Table 5. Experimental conditions for the CPA adsorption rate from aqueous solution on MM8A and MM.

AC	Experiment No.	m (g)	CA0 (mg/L)	CAe (mg/L)	qe (mg/g)	kL,opt × 102 (cm/s)	%D	Dep,opt × 104 (cm2/s)	%D
	RMM8A_1	0.1	100.3	7.90	184.73	16.20	13.41	11.4	12.0
	RMM8A_2	0.1	245.9	55.19	381.39	9.80	8.54	12.9	8.6
MM8A	RMM8A_3	0.1	496.7	238.16	517.10	8.94	2.51	11.4	2.6
	RMM8A_4	0.1	739.4	451.15	576.54	5.96	2.22	9.39	1.4
	RMM8A_5	0.1	978.2	669.08	618.15	7.18	1.62	8.70	1.3
	RMM8A_6	0.1	1467.2	1116.55	701.36	5.30	0.81	13.4	0.8
	RMM_1	0.1	50.87	5.30	91.13	5.88	12.19	5.11	4.1
MM	RMM_2	0.1	245.88	108.33	275.60	5.62	4.83	3.91	3.7
	RMM_3	0.1	506.62	317.20	378.83	7.19	1.68	5.98	3.9
	RMM_4	0.1	903.86	651.25	505.21	6.93	1.07	30.1	0.9

and are plotted in Figure 11. The equilibrium data for CSt were interpreted by the Langmuir, Freundlich, and R-P isotherms (Equations (14)–(16), respectively), and their %D was calculated using Equation (17). The R-P isotherm best fitted the adsorption equilibrium data for CSt, and the parameters and D% for the R-P isotherm are given in Table 4.

The adsorption isotherms for CSt and for discontinuous stirring (DSt) are plotted together in Figure 11 for comparison. It can be observed that the adsorption capacity for CSt of MM8A is approximately the same as that of DSt of MM8A. Furthermore, the adsorption capacity for CSt of MM is slightly above that for DSt of MM at concentrations of CPA at equilibrium below 200 mg/L. However, for CPA concentrations at equilibrium above 200 mg/L, the adsorption capacity for CSt of MM is higher than that for DSt of MM. This behavior is because the particle size of MM may be slightly reduced in CSt. It has been reported that the decrease in particle size of powdered ACs slightly increases its adsorption capacity [43].

4.6.2. Parameters for the Mass Transfer Evaluation

The molecular diffusivity of CPA (D_AB = 4.54 × 10⁻⁶ cm²/s) in water solution was estimated using the Wilke–Chang correlation [44]. For calculating the external mass transport coefficient (k_L), a method devised by Furusawa and Smith was employed. In this method, the following conditions are assumed: at t = 0, C_A→C_A0, and ${\left.{\mathrm{C}}_{\mathrm{AR}}\right|}_{\mathrm{r}=\mathrm{R}}\to 0$; substituting these parameters in Equation (1), we obtain the following equation:

$${\left[{\displaystyle \frac{d\left(\raisebox{1ex}{$C_A$}\!\left/ \!\raisebox{-1ex}{$C_{A0}$}\right.\right)}{dt}}\right]}_{t=0}={\displaystyle \frac{-mS{k}_L}{V}}$$

(20)

The slope of Equation (20) is approximated using the first three experimental data from the decay curves. The values of k_L of MM8A and MM are listed in Table 5.

4.6.3. EMTM- and PVDM-Based Interpretation of the Rate of CPA Adsorption on MM8A and MM

First, the adsorption rate data of CPA on MM8A and MM were interpreted employing the EMTM and Equations (1)–(4), which were solved numerically using COMSOL Multiphysics software (version 5.6). The experimental data of CPA adsorption rate on both MM8A and MM were interpreted using a numeric solution of the EMTM, and k_L was optimized (k_L,opt) so that the EMTM matched the experimental data. The lowest value was obtained by minimizing the following objective function:

$$minimum=\sum _{n=1}^N{\left({\varphi }_{A,exp}-{\varphi }_{A,pred}\right)}^2$$

(21)

where ${\varphi }_{A,\mathit{exp}}$ and ${\varphi }_{A,\mathit{pred}}$ correspond to the experimental and predicted dimensionless concentration of CPA.

The %D values were calculated (Equation (21)) as a criterion for the fittings. Figure 12a,b show the experimental data and the EMTM predictions for the rate of CPA adsorption on MM8A and MM corresponding to experiments RMM8A_3 and RMM_3 (Table 5), respectively. The EMTM interpretation of CPA adsorption rate on MM8A shows that the experimental k_L = 7.06 × 10⁻² cm/s predicts the experimental data well with a %D of 2.6%, while the k_L,opt = 8.94 × 10⁻² cm/s had a value %D of 2.5%. As for MM, the k_L = 8.40 × 10⁻² cm/s and k_L,opt = 9.71 × 10⁻² cm/s had %D values of 1.9 and 1.7%, correspondingly. These results indicate that the EMTM predicted the experimental data of CPA adsorption rate on MM8A and MM satisfactorily. The values of k_L,opt and %D are listed in Table 5.

The adsorption rate of CPA on MM8A and MM was interpreted by the PVDM. In this model, CPA diffusion occurs exclusively by pore volume diffusion (D_ep ≠ 0; D_es = 0). The values τ = 3.5, ε_p = 0.75, and D_AB of CPA were substituted in Equation (10) to obtain a D_ep = 9.74 × 10⁻⁷ cm²/s for MM8A. For MM, the same data were used to calculate the D_ep = 8.57 × 10⁻⁶ cm²/s with the exception of ε_p = 0.66. Figure 13a,b show that the PVDM prediction underestimates the rate of CPA adsorption on MM8A and MM as it indicates that equilibrium is attained in 1200 and 650 min approximately, while in reality, the experimental equilibrium is clearly reached in 5 min in both cases. Figure 13a,b also show the PVDM prediction for D_ep,opt = 1.14 × 10⁻³ cm²/s and 4.71 × 10⁻⁴ cm²/s for MM8A and MM corresponding to the concentration decay curves in RMM8A_3 and RMM_3 (Table 5), respectively. Although the PVDM predicts the experimental data fairly well, the D_ep,opt are much higher than the D_AB, indicating that CPA diffuses more rapidly inside MM8A and MM than in an aqueous solution, and this has no physical meaning. The previous results demonstrated that the external mass transport controls the adsorption rate of CPA on both MM8A and MM, and that it is slower than the intraparticle diffusion of CPA inside the pores of MM8A and MM [23]. This result can be explained by considering that the ACs are powdered, so the diffusion path inside the ACs is very short.

Figure 14a,b show the variation of k_L,opt with the CPA mass adsorbed for both MM8A and MM with different initial concentrations of CPA. Except for experiment RMM8A_1, the k_L,opt changed slightly but with no tendency, which is logical since k_L does not depend on the amount of mass adsorbed but on the solution hydrodynamics [23]. It can be recommended using the arithmetic averages of k_L,opt, which are k_L,opt = 6.4 × 10⁻² cm/s and k_L,opt = 5.5 × 10⁻² cm/s for MM8A and MM. Finally, comparing the decay curves of MM8A and MM, it is observed that for both adsorbents, equilibrium is reached at similar times; however, the concentration decays for MM8A are more pronounced because it adsorbs a higher amount of CPA at a similar rate. Therefore, the adsorption rate of CPA on MM8A is higher than that on MM.

5. Conclusions

The textural properties of MM, MM4, MM8, and MM8A varied with CO₂ reactivation time. S_BET, V_p, S_mic, and V_mic decreased in the case of MM4 and MM8 because of devolatilization and CO₂ gasification. The latter also induced an increase in the S_meso in MM4 and MM8. MM8A showed a large increase in S_BET, V_p, S_mic, and V_mic because of the different CO₂-reactivation processes that it was subjected to; these processes aided further removal of oxygen functional groups and the generation of new micropores. The quantification of active sites revealed that total acidic sites decreased with CO₂ reactivation time in the following order: MM < MM4 < MM8 ≈ MM8A. Meanwhile, the carboxylic sites decreased in the following order: MM < MM8A < MM4 < MM8. In MM8A, there was a minor loss of carboxylic sites because of reoxidization of its surface. The PH_PCZ values of MM4, MM8, and MM8A were shifted to the right, indicating that CO₂ reactivation reduced the number of acidic sites. The Raman spectra of MM, MM4, MM8, and MM8A showed that with higher CO₂ reactivation time, the I_D/I_G increased, revealing that the density of defects was more abundant.

The adsorption capacities of the ACs were assessed at pH = 7. MM8A had the highest CPA adsorption capacity, which was attributed to the following reasons: (1) The reduced density of the carboxylic sites fortified the π–π stacking interactions with CPA. (2) MM8A had a disordered structure because of the formation of micropores and the removal of carboxylic sites on the edges of basal planes. The effect of pH and temperature on CPA adsorption on MM8A was analyzed comprehensively as this material had the best adsorption capacity. The pH had a major effect on MM8A adsorption capacity. An increase in pH resulted in the adsorption of a higher mass of CPA. Temperature did not affect CPA adsorption much. The isosteric heat of adsorption implied that the adsorption of CPA on MMA occurred by physical interactions. The adsorption mechanisms of CPA on MM8A were elucidated by zeta potential measurements. At pH = 5 and 7, electrostatic attractions governed the process. At pH = 9, π–π stacking interactions predominated, and at pH = 11, the main driving force was π–π stacking interactions. However, hydrophobic interactions also participated in the CPA adsorption on MM8A.

The CPA adsorption rate was studied by using diffusion models. The PVDM could not be physically interpreted. The EMTM-based interpretation revealed that external mass transport dominated the adsorption process and that the intraparticle diffusion is much faster. Finally, the adsorption of CPA on MM8A occurred more rapidly on MM8A than on MM. Hence, this work demonstrated that the CO₂ reactivation of commercial ACs is a suitable process for increasing their capacity of adsorbing contaminants from water.