Adsorption of ibuprofen from aqueous solution on chemically surface-modified activated carbon cloths

2.1 Materials

2-[4-(2-Methylpropyl) phenyl]propanoic acid, also named ibuprofen (IBP), was purchased from Sigma–Aldrich (>98% purity). The atomic positions of IBP were determined by molecular modelization using chemsketch 3D viewer. The dimensions of the molecule included in a parallelepiped (1.36 (length) × 0.74 (width) × 0.52 (thickness) nm³) were obtained by adding the Van der Waals radius value (0.1 nm) to the Hydrogen atoms at extreme positions. These dimensions are in agreement with those found by Azaïs et al. (2006).

The raw activated carbon cloth (900-20) referred to C0 was provided by Kuraray Chemical Co. Ltd. (Japan). Prior to modification, it was stirred in 0.1 mol L⁻¹ hydrochloric acid solution for 24 h to remove metal salt impurities. After filtration, it was rinsed with distilled water (∼200 mL per g) until reaching a constant pH of the filtrate and dried at 383 K for 24 h. The dried sample was further modified by (i) bleaching or (ii) thermal treatment, and (iii) further ibuprofen loaded as following:

1. 1 g of C0 was stirred in 200 mL of a 0.13 mol L⁻¹ NaOCl solution at room temperature for 24 h. After filtration, washing and drying, the recovered bleached sample was named CNaOCl.

2. About 2.5 g of C0 was heated under a nitrogen flow in a tubular furnace at 973 K (ramp of 10 K min⁻¹) and maintained at this temperature for 1 h. Sample was kept under inert atmosphere until cooling to room temperature. Recovered cooled sample was named C700N2.

3. To investigate the porosity accessible to ibuprofen, 200 mg of each sample (C0, CNaOCl and C700N2) was saturated by ibuprofen (1 L, 100 mg L⁻¹) at pH 7, at maximum uptake for 5 days and compared to the raw fabric activated carbon.

2.2 Textural and chemical characterization of activated carbons

The porosity of the activated carbon cloths was characterized by N₂ adsorption–desorption at 77 K and CO₂ adsorption at 273 K using an automatic sorptometer (ASAP 2020, Micromeritics). Prior to measurements, samples have been degassed for 12 h at 523 K under vacuum.

The N₂-isotherms were used to calculate the specific surface area using the Brunauer–Emmett–Teller (BET) equation, assuming the area of the nitrogen molecule to be 0.162 nm². As negative unrealistic C factors were obtained by applying the BET model in the relative pressure range from 0.05 to 0.3, the BET specific surface areas were preferentially computed in the relative pressure range of 0.01–0.05, as for microporous materials (Kaneko and Ishii, 1992).

The total pore volume was estimated as the liquid volume of N₂ adsorbed at a relative pressure of 0.995. Pore size distributions (PSD) of the activated carbon samples were determined by using NLDFT (non local density functional theory) models applied on the adsorption isotherms of N₂ at 77 K. Additionally, the distribution of pores smaller than 0.7 nm (narrow micropores or ultramicropores) was evaluated from CO₂ adsorption isotherms at 273 K. For that, infinite slit pore model was assumed for CO₂ adsorption (pores diameter lower than 1.1 nm), while finite slit pore model was used for N₂ adsorption simulations (Jagiello and Olivier, 2009). N₂ adsorption data at P/P₀ < 0.01 were obtained using incremental fixed doses of ∼10 cm³ g⁻¹ (STP), setting the equilibration interval at 300 s. CO₂ adsorption data were obtained at P/P₀ ranging from 4 × 10⁻⁴ to 3.5 × 10⁻², using 45 s equilibration interval (Jagiello and Thommes, 2004).

The DFT pore size distributions of the IBP loaded activated carbon cloths (at pH 7) were studied in the same condition as that of the raw cloth but after degassing at 323 K for 4 days under secondary vacuum to avoid the decomposition of ibuprofen of which melting point is equal to 348–350 K (Lerdkanchanaporn and Dollimore, 2000).

The pH of each activated carbon cloth (0.5 g) was measured in a distilled water suspension (12.5 mL) after heating at 90 °C and then cooling to room temperature (Reffas et al., 2010).

The pH_PZC, i.e. the pH of the solution when the net surface charge equals zero, was determined by the so-called pH drift method (Franz et al., 2000). Mixtures of 0.15 g of activated carbon cloths and 50 mL of deoxygenized 0.01 mol L⁻¹ NaCl solutions of initial pH values varying from 2 to 12 were stirred for 48 h under N₂ in order to avoid the formation of dissolved CO₂. The final pH was measured and plotted against the initial pH. The pH_PZC was equal to the value for which pH (final) = pH (initial).

Boehm titrations quantify the basic and oxygenated acid surface groups on activated carbons (Boehm, 2002). Surface functional groups such as carboxyl (R—COOH), lactone (R—OCO), phenol (Ar—OH), carbonyl or quinone (RR′C⚌O) and basic groups were determined using different reactants, assuming that NaOC₂H₅ reacted with all groups; NaOH did not react with RR′C⚌O groups; Na₂CO₃ did not react with RR′C⚌O nor R—OH groups and that NaHCO₃ only reacted with R—COOH groups. About 0.15 g of each carbon sample was mixed in a closed Erlenmeyer with 50 mL of a 0.1 mol L⁻¹ aqueous reactant solution (NaOH, Na₂CO₃ or NaHCO₃). In the case of NaOC₂H₅, only 0.1 g of carbon sample was added to 50 mL of 0.01 mol L⁻¹ solutions which were prepared in absolute ethanol. The mixtures were stirred for 24 h at constant speed (650 rpm) and room temperature and then filtered off. Back-titrations of the filtrate (10 mL) were then achieved with standard HCl (0.01 mol L⁻¹) to determine the oxygenated group contents. Basic group contents were also determined by back titration of the filtrate with NaOH (0.01 mol L⁻¹) after stirring the activated carbon (0.15 g) in HCl (50 mL, 0.01 mol L⁻¹) for 24 h.

The surface of activated carbons was characterized by ATR-FTIR using a Thermo Scientific Nicolet iS10 spectrometer equipped with a germanium crystal. A DTGS KBr detector was used for detection and the incident angle of the beam was 45°. All spectra were collected in the infrared region [500–4000 cm⁻¹] with a spectral resolution of 4 cm⁻¹. 64 scans were accumulated for each analysis.

X-ray photoelectron spectroscopy (XPS) measurements were performed using an ESCALAB 250 spectrometer (Thermo Fisher Scientific) at monochromatic Al-Kα anode X-ray radiation, on a 150 × 800 μm² analysis region, under 2 × 10⁻⁹ mbar vacuum. The high-resolution scans (0.1 eV) were obtained over the 280.1–299.9 eV (C1s) and 523.1–539.9 eV (O1s) energy ranges with a pass energy of 20 eV. After baseline subtraction, the curve fitting was performed assuming a mixed Gaussian–Lorentzian peak shape (the ratio of Gaussian to Lorentzian form was equal to 0.3). The carbon 1s electron binding energy corresponding to graphitic carbon was referenced at 284.6 eV for the calibration (Xie and Sherwood, 1990).

2.3 Adsorption experiments: Isotherms and kinetics

All the IBP solutions were prepared from UHQ water (Ultra High Quality, 18.2 MΩ purity) containing 10 vol.% of methanol (99.9%) in order to increase the IBP solubility.

The kinetics were studied at 298 K at pH 3 and 7. The initial pHs of IBP solutions were adjusted at pH 3 and 7 by adding either 0.1 mol L⁻¹ HCl or NaOH. Suspensions of 10 mg of the activated carbon cloth in 100 mL of 100 ppm IBP solution were stirred at 250 rpm and then filtered at different times between 5 min and 7 days.

Adsorption isotherms of IBP on the different activated carbon cloths were performed at pH 3 at constant temperatures of 298, 313 and 328 K. 10 mg of activated carbon cloths was introduced in IBP solutions (100 mL) of varying initial concentrations (5–100 mg L⁻¹). Suspensions have been stirred for 5 days in a thermostatically controlled orbital shaker (New Brunswick Scientific, Innova 40, and stirring speed of 250 rpm) to reach equilibrium and then filtered. The initial and residual IBP concentrations were measured by high performance liquid chromatography (HPLC) using a Waters chromatograph equipped with a high pressure pump (Waters 515), a photodiode array detector (Waters 996) and a Sunfire C18 column (5 μm, 4.6 × 250 mm). A methanol/ultrapure water solution (80/20, v/v), containing 0.1 vol.% of concentrated phosphoric acid (95 wt.%) in isocratic mode at a flow rate of 1 mL min⁻¹ was used as mobile phase. Detection was operated at 220 nm.

The equilibrium IBP uptake Q_ads (mg g⁻¹) was calculated from equation: $$Q_{\text{ads}}=\frac{(C_{\text{e}}-C_{\text{i}})·V}{m}$$ where V is the solution volume (L), C_i is the initial IBP concentration (mg L⁻¹), C_e is the equilibrium IBP concentration (mg L⁻¹) and m is the mass of the dry activated carbon cloth (g).

3.1 Characterization of activated carbons

3.1.1

3.1.1 Surface chemistry

The results of Boehm titrations (Table 1) indicated that C0 was mainly acidic: the amount of oxygenated groups (∼0.84 meq g⁻¹) was about twice the one of basic groups (∼0.43 meq g⁻¹). This was confirmed by the slightly acidic pH measured value (6.05). The higher value of pH_PZC (8.40) might be assigned to the presence of high contents of carbonyl groups (pK_a–carbonyl ∼16–20).

Table 1 Surface groups (in meq g⁻¹) obtained from “Boehm” titrations, pH and pH_PZC values of raw and modified activated carbon cloths.

Activated carbon sample	C0	CNaOCl	C700N2
Carboxylic groups	0.04	0.56	0
Lactonic groups	0.12	0.42	0
Phenolic groups	0.02	0.19	0.03
Carbonyl groups	0.66	0.51	1.36
Total oxygenated groups	0.84	1.68	1.39
Total basic groups	0.43	0.28	0.62
pH	6.05	5.59	7.52
pHPZC	8.40	6.54	7.80

Thermal treatment under N₂ removed virtually the totality of the carboxylic and lactonic groups and significantly increased the number of carbonyl group compared to C0 (Table 1). The substrate has become slightly basic (pH = 7.52 and pH_PZC = 7.80) due to the high reactivity of the newly treated surface with regards to dioxygen upon re-exposure in air (Menéndez et al., 1996). As a consequence, new oxygenated groups (mainly carbonyl groups) were generated after aging of C700N2.

The treatment of the activated carbon cloth with NaOCl has increased the amount of total oxygenated groups (from 0.84 to 1.68 meq g⁻¹) and led to a decrease in the number of basic sites (Table 1). As expected, this oxidative treatment increased the carboxylic group content, created lactones and phenols but slightly decreased the number of carbonyl groups. As a consequence, the measured pH_PZC and pH values for CNaOCl were the lowest.

The evolution of the surface chemistry induced by both treatments was also characterized by XPS. The C1s spectra (not shown) were deconvoluted into four peaks (Table 2) attributed to carbon present in: graphitic layers (peak I), phenolic or alcohol groups (peak II), carbonyl or quinone groups (peak III), and carboxylic or ester or lactonic groups (peak IV). The three peaks issued from the deconvolution of the O1s spectra (Table 2) were attributed to C⚌O groups (peak A), C—O groups (peak B) and to chemisorbed oxygen (peak C).

Table 2 Atomic percentages of carbon (C1s) and oxygen (O1s) obtained by XPS analysis of raw and modified activated carbon cloths.

Peak	Binding energy (eV)	Signal attribution E(eV)	C0	CNaOCl	C700N2
I	284.6	C—C	74.89	40.82	57.34
II	285.09–285.24	C—O	13.57	17.98	12.60
III	286.21–286.35	C⚌O	6.83	7.29	8.99
IV	287.93–288.36	O—C⚌O	0	6.50	8.10
A	531.21–531.90	O⚌C	1.68	4.84	5.16
B	532.96–533.33	O—C	1.66	19.85	6.36
C	535.01–535.75	Chemisorbed O	1.37	2.71	1.46
Total atomic O			4.71	27.40	12.98

For all samples, the peak of highest intensity was attributed to graphitic carbon (∼284.6 eV). The relative intensity of this main signal decreased for both treated samples as oxygenated functional groups were formed. Indeed, the total percentage of atomic oxygen measured increased from 4.71% for C0 to 12.98 at.% and 27.40 at.% for samples C700N2 and CNaOCl, respectively. An increase in the intensities of the signals attributed to carbon atoms linked to oxygen ones was also observed for both treated samples compared to sample C0 (peaks labeled II, III and IV).

For sample CNaOCl, the increase in oxygen groups observed by XPS (Table 2) is mainly due to the formation of additional phenolic and carboxylic groups, in agreement with the Boehm titrations (Table 1). Similarly, Perrard et al. (2012) reported after oxidation of an ex-cellulose activated carbon cloth by NaOCl (0.53 mol L^-1, 90 min), an increase in the carboxylic and phenolic group content was observed by XPS, but later phenolic groups were oxidized up to lactonic groups. Cagnon et al. (2005) reported an increase in carboxylic and carbonyl groups observed by XPS, after bleaching of an activated carbon. Indeed, in one of our previous studies (Guedidi et al., 2013), we also found that a commercial granular activated carbon oxidized by NaOCl presented a slight increase in carbonyl and carboxylic contents compared to the raw activated carbon.

Moreover, the XPS O—C contribution has increased for both modified cloths and especially for CNaOCl (from 1.66 to 19.85 at.%), which is in agreement with Boehm titrations. Furthermore, the chemisorbed oxygen increases after oxidation by NaOCl from 1.37 to 2.71 at.% (Table 2). It can be explained by the presence of water attributed to the increase in the surface hydrophilicity for CNaOCl.

For C700N2, the oxygen increase observed by XPS (Table 2) originates from the formation of C⚌O and O—C⚌O bonds. The increase in carbonyl (C⚌O) is in agreement with Boehm titration. The carboxylic groups were found absent by Boehm titrations, thus the O—C⚌O bonds were possibly ester or lactonic groups. Virtually, most of the studies (Rong et al., 2003; Swiatkowski et al., 2004) on the thermal treatment of activated carbon under inert atmosphere reported that all oxygen containing functional groups were removed during this treatment. But, the work of Menéndez et al., (1996) showed the temporary efficiency of such treatment for removing the oxygen containing surface groups, as reactive surface sites are known to be generated. As the resultant surface is very active, it can react with oxygen present in air, giving new surface oxide such us carbonyls. These groups have a basic character, conferring basic properties to the activated carbon.

The infrared ATR spectra of C0 and C700N2 are very similar (Fig. 1 a and b) and show a broad band between 1000 and 1250 cm⁻¹ assigned to C—O stretching in acids, alcohols, phenols, ether and esters (Park et al., 1999). These observations suggest that these groups remained present on the activated carbon cloth surface even after treatment at 973 K under N₂, in agreement with the results obtained by Boehm titrations and XPS (similar content of phenolic and lactonic groups before and after thermal treatment under N₂).

Close

Figure 1 FTIR spectra of (a) C700N2; (b) C0 and (c) CNaOCl.

Export to PPT

In CNaOCl ATR spectrum (Fig. 1c), the intensity of this band was increased exhibiting two maxima at 1120 and 1200 cm⁻¹ assigned to C—O stretching in ester or ether groups due to bleaching (Tables 1 and 2).

Moreover, after the bleaching treatment, a new broad band appeared in the 3200–3600 cm⁻¹ range due to the O—H stretching vibration of phenolic groups and chemisorbed water (Biniak et al., 1997). The peak centered at 1720 cm⁻¹, only observed on the spectrum of CNaOCl, is assigned to the C⚌O stretching vibration of ketone, ester and/or aromatic carboxylic groups (Perrard et al., 2012). All spectra present a peak at 1580–1600 cm⁻¹ assigned to C⚌C stretching modes of aromatic rings or to C⚌O groups stretching vibrations conjugated with aromatic rings (Sabio et al., 2004). This peak (at 1600 cm⁻¹) was also noticeably increased after NaOCl treatment (Fig. 1c), possibly after the formation of carboxylic, lactonic and quinone groups (Table 1). As conjugation moves absorptions to lower wavenumbers, this peak is shifted to 1580 cm⁻¹ for C700N2 and C0 (Domingo-Garcia et al., 2000).

3.1.2

3.1.2 Porosity characterization

Fig. 2a and b shows the nitrogen adsorption–desorption isotherms of raw and modified fabrics. All the isotherms belong to the type I, typical of microporous materials, according to the IUPAC classification. They present a sharp knee in the low relative pressure region and a very low slope in the multilayer range (at higher pressure), indicating the absence of significant mesoporosity. Indeed, insignificant mesopore volumes were measured for the three samples (Table 3). The isotherm profiles are characteristic of narrow micropore samples. After NaOCl treatment, the N₂ uptake at high relative pressure was significantly reduced, explaining the decrease of the BET specific surface area to 1690 m² g⁻¹ in comparison with the C0 one (1910 m² g⁻¹). The decrease in the BET specific surface area is related to the decrease in the micropore volume which might be filled by the surface functional groups formed through NaOCl oxidation. This treatment yields also to the supermicropore slight decrease from 0.32 to 0.26 cm³ g⁻¹, possibly explained by the collapse of some pore walls during oxidation. Table 3 shows for C700N2 a very slight increases both in BET specific surface area (1946 m² g⁻¹) and in micropore volumes, which might result from the thermal decomposition of the oxygenated surface groups that were either inside the micropores or blocking micropore entrances (Pereira et al., 2003). However this increase is not significant as the measured ultramicropore and supermicropore volumes remain constant after thermal treatment.

Close

Figure 2 (a) N₂ adsorption–desorption isotherms at 77 K of the pristine C0(♢: adsorption, ♦: desorption), CNaOCl (□: adsorption, ■: desorption) and C700N2 (△: adsorption, ▴: desorption); (b) Zoom of N₂ adsorption–desorption isotherms of C0, CNaOCl and C700N2.

Export to PPT

Table 3 Textural properties obtained by N₂ adsorption/desorption and CO₂ adsorption studies.

Sample	C0	CNaOCl	C700N2	C0(IBP)	CNaOCl(IBP)	C700N2(IBP)
SBET (m2 g−1)	1910	1690	1946	1244	1161	1240
Total pore volume(cm3 g−1)	0.79	0.69	0.80	0.51	0.48	0.51
Micropore volumea(cm3 g−1)	0.75	0.65	0.76	0.47	0.44	0.48
Mesopore volumea (cm3 g−1)	0.04	0.04	0.04	0.04	0.04	0.03
Ultramicropore volumeb (cm3 g−1)	0.31	0.31	0.30	0.12	0.14	0.14
Supermicropore volumea (cm3 g−1)	0.32	0.26	0.32	0.25	0.22	0.24

a From N₂ DFT.

b From CO₂ DFT.

The study of the PSD of the IBP-loaded samples and the initial adsorbents brought out that this molecule might mainly adsorb in the ultramicropores and slightly in the supermicropore (Fig. 3) as previously reported by Guedidi et al. (2013). The ultramicropore volumes clearly decreased after IBP adsorption while the supermicropore volume slightly decreased (Table 3). From PSDs obtained from CO₂ isotherms, it can be concluded that whatever the fabric sample: C0, C700N2, or CNaOCl, the adsorption of IBP might occur mainly in the ultramicropores of diameter lower than 0.7 nm (Fig. 3).

Close

Figure 3 NLDFT pore size distributions either for narrow micropores (<0.7 nm) from CO₂ adsorption isotherms at 273 K, or for supermicropores (1 nm < ∅ < 2 nm) from N₂ adsorption isotherms at 77 K: of C0 (♢) and 134.7 mg g⁻¹ IBP loaded C0 (♦), of CNaOCl (□) and 119 mg g⁻¹ IBP loaded CNaOCl (■) and of C700N2 (Δ) and 127.05 mg g⁻¹ IBP loaded C700N2 (▴).

Export to PPT

However, a rough calculation shows that IBP can only access to slit-pores of inter-wall spacing higher than 0.7 nm by taking into account the thickness of the molecule (0.5 nm) and the atomic radius of the adsorbent on both sides of the pore walls (almost 0.1 nm). This means that theoretically IBP should not be accommodated in the ultramicropores. Possibly, the IBP molecule could access the larger ultramicropores after a modification of its geometry.

Table 4 shows also the IBP occupied volumes in ultramicropores and supermicropores calculated from the difference between porous volume of the initial samples and the IPB loaded samples (ultramicropore and supermicropore volumes were obtained from PSD DFT models from CO₂ and N₂ isotherms, respectively). These volumes were compared to the volume uptake obtained from the IBP measured adsorption uptake taking into account its calculated volume (0.52 nm³). The whole IBP occupied micropore volumes (ultramicropore + supermicropore) calculated for all carbons (Table 4) were higher than the IBP volumic uptake suggesting the IBP pore space blocking leading to the formation of sealed inaccessible porosities particularly in ultramicropores. This also suggests the IBP accommodation both in larger ultramicropores and supermicropores.

Table 4 Comparison of the occupied ultramicropore volumes (from DFT PSD) and the IBP occupied volume calculated from the uptake.

	Vultramicropore occupied by IBP (cm3 g−1)a	Vsupermicropore occupied by IBP (cm3 g−1)a	IBP volume uptakeb (cm3 g−1)
C0	0.19	0.07	0.20
CNaOCl	0.17	0.04	0.18
C700N2	0.16	0.08	0.19

a Calculated from the PSD volume differences.

b Calculated from the IBP measured uptake and dimension size.

3.2 Ibuprofen adsorption kinetics and pH effect

The pK_a of ibuprofen is equal to 4.91 (Lindqvist et al., 2005). The effect of the initial pH on the IBP adsorption kinetics onto the C0 cloth was studied at pH ∼ pK_a − 2 and pH ∼ pK_a + 2 (i.e. pH = 3 and 7) at 298 K, where IBP was either molecular, or negatively charged. The plot of C/C₀ vs time (Fig. 4) shows that the adsorption kinetic is faster at pH 3 than at pH 7. Indeed, the equilibrium adsorption times, corresponding to the appearance of the plateaus on both curves, are of about 600 min and 1440 min at pH 3 and 7, respectively.

Close

Figure 4 Ibuprofen adsorption kinetics onto C0 cloth at 298 K at pH 3 and 7 ([IBP] = 100 mg L⁻¹, m_adsorbent = 10 mg, V_IBP = 100 mL, stirring speed = 250 rpm).

Export to PPT

Moreover, Fig. 4 shows that the IBP uptake increases when the pH decreases. As the C0 fabric surface is positively charged and the molecule is neutral at pH 3, (pH_PZC equal to 8.4), the IBP adsorption is promoted by dominant dispersive interactions as previously reported (Guedidi et al., 2013). The higher uptake of IBP at acidic pH with regard to neutrality is also probably related to the lower IBP solubility (of the molecular form) while the pH decreases, as its lower solubility promotes its adsorption on the carbon cloth (Shaw et al., 2005). The experimental IBP adsorption kinetics on C0 fabric were fitted to four models namely, the Lagergren pseudo-first-order model (1) (Lagergren, 1898), the pseudo-second-order model (2) (Ho and McKay, 1999), the Elovich model (3) (Chein and Clayton, 1980) and the intraparticle diffusion model (4) (Weber and Morris, 1963).

A pseudo first-order equation can be expressed as:

The second-order equation is:

The Elovich equation is:

The intraparticle diffusion model is formulated by:

In order to estimate the best fit of the kinetic models to the experimental kinetic data, the optimization procedure requires different statistical parameters to be defined. In our study, the best fit model was evaluated by adjusted R², the Sum of the Square of the Errors (SSE), the Residual Root Mean Square Error (RMSE) and the chi-square test χ² expressed as follows:

Table 5 shows the adsorption kinetic parameters calculated from the kinetic models. The profiles of each fitted curve at pH 3 and 7 are displayed in Fig. 6a and b respectively. The highest adjusted correlation coefficient R² = 0.99 value obtained at pH 3 and lowest values of SSE (2810), RMSE (14.7) and χ²(8.90) for the Elovich model made it the best model for ibuprofen adsorption. It confirmed that the adsorption data were well represented by Elovich model (Table 5). The α (Elovich parameter) adsorption initial rate of IBP was equal to 3.929 × 10⁵ mg g⁻¹ min⁻¹. 1/β, indicative of the number of sites available for adsorption, was equal to 23. Besides, 1/β × ln(α β) is the adsorption amount when ln(t) is equal to zero. This value is helpful in understanding the adsorption behavior of the first step (Tseng, 2006); i.e., the IBP adsorption uptake by C0 when t = 1 min is equal to 231.22 mg g⁻¹ which represents more than 50% of the total ibuprofen uptake.

Among all the tested kinetic models, the pseudo second order model was the best one for fitting IBP adsorption at pH 7 (it gives the best adjusted-R² = 0.99, the lowest SSE = 442.5, RMSE = 5.26 and χ² = 15.72) (table 5). The intraparticle diffusion model proposed by Weber and Morris (1963) was not consistent with the experimental kinetics at pH 3 (Fig. 5a). However, at pH 7, the intraparticle diffusion model describes well the adsorption of IBP up to 600 min (intraparticle diffusion rate constant equal to 5.764 mg g⁻¹ min^−1/2). Fig. 6 shows the intraparticle diffusion plot of the adsorption of ibuprofen on the C0 cloth at 298 K, plot indicating three stages. The first stage represents the instantaneous adsorption or external surface adsorption up to 600 min, for which, the linear portion passes through the origin indicating that the rate of adsorption is controlled by the intraparticle diffusion. In the second stage, the regression is nearly linear but does not pass through the origin, suggesting that the intraparticle diffusion is not the only rate limiting mechanism in the adsorption process. The third region is the final equilibrium stage where the intraparticle diffusion starts to slow down due to the extremely low IBP concentrations remaining in the solutions. The Kp_i values which represent the rate parameter of i stage were found to be decreased (Kp₁ = 5.79, Kp₂ = 1.98, Kp₃ = 0.24) indicating that the adsorption rate is initially faster and then slows down when the time is increased (Tan et al., 2009).

Close

Figure 5 Fitted profiles of ibuprofen adsorption kinetics of C0 sample at pH 3 (♦) (a) and at pH 7 (■) (b) by pseudo-first-order, pseudo-second order, Elovich and intraparticle diffusion models.

Export to PPT

Close

Figure 6 Plot of intraparticle diffusion model for adsorption of IBP on C0 fabric at pH 7.

Export to PPT

3.3 Adsorption isotherms

3.3.1

3.3.1 Effect of temperature

Adsorption experiments were carried out at 298, 313 and 328 K. To investigate the temperature dependency of the adsorption capacity, four equilibrium models were tested: Langmuir (Zhu et al., 2009), Freundlich, Langmuir–Freundlich (Yao, 2000) and Redlich-Peterson (Kumar and Porkodi, 2007). All the isotherms were simulated using an iterative procedure based on a non-linear least-squares algorithm.

The experimental data of IBP adsorption on C0 and modified activated carbon cloths at pH 3 were well fitted by the Langmuir–Freundlich model. In terms of adjusted-R², the Langmuir–Freundlich model gives the best fitting (adjusted R² > 0.96) in agreement with our previous work about the IBP adsorption on a granular activated carbon (Guedidi et al., 2013). Fig. 7 displays the Langmuir–Freundlich fits of isotherms at pH 3. With the rise of temperature the IBP maximum uptake increases for each cloth whatever the treatment which means that adsorption phenomenon is endothermic.

Close

Figure 7 Experimental IBP adsorption isotherms on C0, CNaOCl and C700N2 at pH 3 and at 298 K (♦), 313 K (▴) and 328 K (■) fitted by the Langmuir–Freundlich model (lines).

Export to PPT

The adsorption equilibrium capacity of the activated carbon cloth C0 at 298 K was compared with the ones published recently of some other commercial and prepared activated carbons (Table 6). We can notice that the adsorption capacity of the activated carbon cloth C0 toward ibuprofen is the highest and reaches about 492 mg g⁻¹ at acidic pH. This value can be correlated its highest surface area (S_BET ∼ 1910 m² g⁻¹) compared to the ones of the other sorbents (S_BET < 1160 m² g⁻¹). However, the equilibrium time needed to reach this value is higher than for the granulated or powdered samples, which has to be correlated to a slower diffusion of the pollutant within the cloth fibers’ porosity.

Table 6 Comparison of the performance toward the adsorption of ibuprofen on the C0 activated carbon cloth and other activated carbons reported in the literature.

Activated carbon (AC) origin	SBET (m2g−1)	Time for equilibrium	Adsorption capacity (mg g−1)	References
AC cloth C0	1910	600 min (10 h)	491.9	This study
Commercial granular AC	800	4000 min (67 h)	138.1	Guedidi et al. (2013)
AC from artemisia vulgaris	358.2	>5 h	16.73	Dubey et al. (2010)
AC obtained by chemical activation of cork powder	891	2 h	112.4	Mestre et al. (2009)
AC obtained by physical activation of cork powder	1060	2 h	106.4	Mestre et al. (2009)
AC obtained by physical activation of PET	1426	1 h	138.9	Mestre et al. (2009)
AC obtained by physical activation of coal	1156	0.5 h	131.6	Mestre et al. (2009)
AC obtained by physical activation of wood	899	0.2 h	131.6	Mestre et al. (2009)
AC obtained by physical activation of wood + boiling in 20% HNO3	879	0.2 h	89.30	Mestre et al. (2009)
AC from olive waste cake	793	5 h	9.09	Baccar et al. (2012)
Commercial activated charcoal	–	∼2 h	64.5	Khalaf et al. (2013)