Activated Carbon from Spent Co ﬀ ee Grounds: A Good Competitor of Commercial Carbons for Water Decontamination

Featured Application: Activated carbon from a renewable source (spent co ﬀ ee grounds) as a competitive substitute for commercial ones in water decontamination from phenols. Abstract: In the framework of the circular economy, spent co ﬀ ee grounds were converted into powdered activated carbon by means of pyrolysis, using potassium hydroxide as the activating agent. Its adsorption capacity on a panel of phenolic compounds was compared with those of two commercial powdered activated carbons, after preliminary studies on organic dyes with di ﬀ erent ionic properties, to assess the a ﬃ nity between adsorbates and adsorbents. Pseudo-ﬁrst-order and pseudo-second-order kinetic models were carried out, together with Freundlich and Langmuir isotherms. They were useful to calculate the breakthrough at 5%, 10%, and 50% of adsorption and the partition coe ﬃ cients for the comparison of performance between di ﬀ erent sorbent systems in a less biased manner (e.g., reducing bias associated with operational settings like sorbate concentration and sorbents dosage). The results showed that the removal e ﬃ ciency for SCGs-AC was comparable with that of the commercial activated carbons with the highest partition coe ﬃ cients for methylene blue (12,455 mg / g / µ M, adsorption capacity = 179 mg / g) and 3-chlorophenol (81.53 mg / g / µ M, adsorption capacity = 3765 mg / g). The lower e ﬃ ciency in bromothymol blue and bisphenol-A adsorption was due to its di ﬀ erent morphology and surface properties. with of two commercial to tests with preliminary studies were performed to assess the adsorption of three characterized by ﬀ erent blue (MB)–cationic, erythrosine-B (EB)–anionic, and bromothymol blue (BB)–non-ionic. These experiments compare the results with the data and to evaluate the a ﬃ nity between adsorbent and adsorbate.


Introduction
Industrialization, urbanization, and increased consumer demand are responsible for a continuous spread in the production of chemical substances. According to Gavrilescu et al. [1], the global production of chemicals grew from 1 million (in 1930) to nearly 350 million tons per year (in 2011). In 2012, approximately 143,000 new compounds were registered in the European market, 50% of them being harmful to the environment [2]. Chemical compounds can be released in the environment during production, storage,

Powdered Activated Carbons
Three different PACs (one from SCGs and two commercial PACs) were used to study the adsorption of organic dyes and phenolic compounds. SCGs (100% Arabica blend) were collected from an Italian local cafeteria after espresso coffee brewing. They were sieved to remove impurities, stirred to obtain a homogenous sample, and oven-dried at 105 • C for 24 h. Hereafter, 60 g were mixed with potassium hydroxide (KOH) powder in a 1:1 mass ratio under N 2 atmosphere. KOH is commonly used as activating agent to develop porosity during thermal treatments [42]. The obtained mixture was pyrolyzed in a tubular alumina reactor (Carbolite) at 800 • C for 4 h, under N 2 atmosphere (150 cc/min). The operating conditions were chosen according to Wang et al. [43]. After pyrolysis, the sample was washed with deionized water on a Gooch filter 4, until neutral pH was reached, and oven-dried at 105 • C for 24 h. The obtained PAC sample is called SCGs-AC in this work. The commercial PACs were purchased from Merck (Product ID 1.02514.1000, Darmstadt, Germany) and Ceca Italiana S.R.L., Arkema Group (Acticarbone NCL 1240, Rho (MI), Italy), here labelled C1-AC and C2-AC, respectively, and used as received.

Organic Dyes
Methylene blue (MB, MM 319.86 g/mol), erythrosine B (EB, MM 879.84 g/mol), and bromothymol blue (BB, MM 624.40 g/mol) were purchased from Sigma-Aldrich and used as received. Tests were carried out in 100-mL flasks, sealed with a plastic tarp, by adding fixed amounts of PACs to aqueous solutions containing a known initial concentration of organic dyes ( Table 1). The samples were continuously stirred at 800 rpm at ambient temperature (20 ± 2 • C), keeping the pH between 5 and 6. The experimental conditions were chosen according to Ahmad et al. (2007) [44]. Samples were collected at different time intervals, filtered with a PTFE 0.45-µm syringe filter, and analyzed to calculate the adsorption capacity according to Equation (1), and the adsorption kinetics using a pseudo-first-order (Equation (2)) or pseudo-second-order kinetic models (Equation (3)) [45]. The tests, conducted in triplicate, were stopped when the measured concentration was constant after two consecutive measurements, or it was below the detection limits.
Adsorption capacity (%) = where C i (in mg/L) is the initial concentration and C t (in mg/L) is the concentration at time t.
where q e and q t represent the amount of the adsorbed compounds (mg/g) at equilibrium and at time t (h), respectively; k 1 (h −1 ) is the first-order kinetic constant, and k 2 is the one for pseudo-second-order kinetic (g/mg·h).  [50]. They were carried out in 100-mL flasks, sealed with a plastic tarp, by adding fixed amounts of PACs to aqueous solutions containing a known initial concentration of phenolic compounds ( Table 1). The samples were continuously stirred at 800 rpm, at ambient temperature (20 ± 2 • C), keeping the pH between 5 and 6. The experimental conditions were chosen according to Laksaci et al. [37,38] and Zbair et al. [51]. Samples were collected after 5, 20, 60, and 180 min from the beginning of the tests, filtered with a PTFE 0.45 µm syringe filter, and analyzed to calculate the adsorption capacity, according to Equation (1), and the adsorption kinetics (Equations (2) and (3)). The tests were conducted in triplicate.

Equilibrium Isotherms
Determination of adsorption isotherms was performed according to ASTM D3860-98 [50] by adding different amounts of PACs to a series of 100 mL flasks, containing the same initial concentration used in the batch adsorption tests for both organic dyes and phenolic compounds ( Table 1). The flasks were kept under stirring in a six-place magnetic plate at 800 rpm and room temperature (20 ± 2 • C) until equilibrium, keeping the pH between 5 and 6. Freundlich and Langmuir models were considered in this study, as they are the most frequently used to describe the adsorption of organic dyes and phenolic compounds [52][53][54][55], even if other models have been considered in the literature [56]. The parameters in the Freundlich (Equation (5)) and Langmuir (Equation (6)) equations were determined through a linearization procedure.
ln q e = ln k F + 1 n ln C e (linearized Freundlich) where q e is the adsorption capacity (mg/g), C e is the residual concentration, k F and 1/n are the Freundlich constants, k L is the Langmuir constant, and Q 0 is the maximum saturation capacity at the isotherm temperature (mg/g).

Analytical Procedures
The specific surface area (SSA), total pore volume (TPV), and average pore diameter (APD) of the PACs were calculated with the Brunauer-Emmett-Teller (BET) method by recording N 2 adsorption-desorption isotherms at −196 • C on an automatic adsorption instrument (Micromeritics apparatus, ASAP 2010 V4.01 G, Norcross, Georgia, USA). The morphology and the elemental composition of the samples were determined with environmental scanning electron microscopy energy dispersive x-ray analysis (ESEM-EDX) on a FEI-QUANTA200 instrument (FEI Italia S.r.l., Milan, Italy). The chosen blow-up was 2000X for elemental composition and 400X and 1600X for morphology. The PACs skeletal density was measured by a helium pycnometer (Multivolume Pycnometer 1305 by Micromeritics, Norcross, Georgia, GA, USA). Finally, the point of zero charge (pH-PZC) was calculated using the solid addition method reported in Kumar et al. (2008) [57]. 50 mL of 0.1 M KNO 3 solution were prepared adding dropwise a 0.1 M HCl or 0.1 M NaOH solution to get different pH values (2, 4, 6, 8, 10, and 12). Then, 1.0 g of PACs was added in each solution and the samples were kept under stirring (800 rpm) in a six places magnetic plate for 48 h at room temperature (20 ± 2 • C). The final pH of the supernatant was recorded. The difference between the initial and the final pH (∆pH = pH i -pH f ) was plotted against the initial pH value (pH i ). The pH-PZC is the point in which the resulting curve intersects the x axis ∆pH = 0.
Organic dyes and BPA concentrations were measured using a UV/VIS spectrometer (PerkinElmer instruments, Lambda 25 UV/VIS Spectrometer, Milan, Italy), keeping a constant wavelength at maximum absorbance (664 nm for MB, 526 nm for EB, 433 nm for BB, and 224 nm for BPA). Ph and CP were determined using a UV-VIS spectrophotometer (Shimadzu Italia S.r.l. UV-1601, Milan, Italy) at a constant wavelength of 510 nm, according to the Italian analytical standard IRSA-CNR 29/2003, vol. 2, n. 5070-A2 [58]. The pH of each sample was adjusted to 4 with concentrated phosphoric acid, and 5 mL of copper sulphate were added. The samples were distilled for 10 min and the condensed solutions were collected. Then, 5 mL of ammonium chloride were added to each solution, then ammonium hydroxide until pH = 10. Finally, by mixing the solutions with 2 mL of 4-aminoantipyrene and 2 mL of potassium ferrocyanide for 10 min, the color was developed. The volumes were brought to 200-250 mL with deionized water before UV-VIS analyses. All concentrations have been determined by extrapolation from calibration curves obtained from standard solution of the target compounds, according to the Beer-Lambert law.

Physico-Chemical Properties of PACs
The porous nature of the samples was investigated by N 2 physisorption at −196 • C. All of them displayed very high surface area values, of 823 and 932 m 2 /g for the commercial C1-AC and C2-AC, respectively, up to 1199 m 2 /g for SCGs-AC, prepared by spent coffee grounds treated with KOH and then pyrolyzed at high temperature. In particular, the latter sample showed a 30% higher BET surface area with respect to the commercial active carbons used for comparison. The cumulative pore volume of the three sample was comparable and very high, in the range 0.491-0.523 cm 3 /g. According to the IUPAC classification [59], the mean pore diameter was borderline between micro-and mesopore regions for SCGs-AC (1.7 nm) and typical of mesoporous materials for C1-AC and C2-AC (2.5 nm and 2.1 nm, respectively).
It is well established that the adsorption characteristics of activated carbons derive from the nature of the precursor, production method, and experimental conditions during the preparation. Thus, they are recognized as versatile adsorbents due to their high specific surface area, porosity, high adsorption capacities, and surface properties [42]. In particular, the pH-PZC values of the activated carbons here considered, calculated using the solid addition method reported by Kumar et al. [57], ranged in the interval 8.8-10.5 (Table 2). Activated carbons are amphoteric materials in which surface charges depend on the pH of the solution. As an example, SCGs-AC sample displayed a pH-PZC of 9.5, which gave rise to a negative charge on the surface of the activated carbon for pH values higher than the pH-PZC, with an electrostatic interaction between the negatively charged surface of the AC and a cationic molecule. As the pH of the system decreases, the number of positive charges on the surface increases, and the surface sites on the activated carbon do not tend to absorb the cations of dye due to electrostatic repulsion. An inverted mechanism occurs in the presence of an anionic probe molecule, with stronger electrostatic interactions at acidic pH values and repulsive interactions in strong alkaline conditions when the pH is higher than the pH-PZC of the material. All the adsorption determinations here reported have been carried out at pH about 6, thus reducing the negative charge amount on the AC surface in the order C2-AC > SCGs-AC > C1-AC.
ESEM-EDX was applied to observe the morphology and pore size of PACs and to determine their composition. Figure 1a,b illustrate 1600X and 400X magnified ESEM micrographs of SCGs-AC sample. The surface was smooth, full of cavities and characterized by partially flattened cylindrical pores. In the pores, traces of foreign elements (i.e., Mg, Al, Si or S, ≤ 1% w/w) could be observed. The morphology is quite similar to that reported by Laksaci [38], where the pores observed (µm range) act as channels for the microporous network. The textural features of the commercial samples, observed from micrographs in Figure 1c-f, were significantly different from those obtained for the SCGs-AC sample. These images showed the extent of inhomogeneity of the surfaces of C1-AC and C2-AC, which displayed a more compact and less porous structure compared to the adsorbent generated from spent coffee grounds. sample. The surface was smooth, full of cavities and characterized by partially flattened cylindrical pores. In the pores, traces of foreign elements (i.e., Mg, Al, Si or S, ≤ 1 %w/w) could be observed. The morphology is quite similar to that reported by Laksaci [38], where the pores observed (μm range) act as channels for the microporous network. The textural features of the commercial samples, observed from micrographs in Figure 1c-f, were significantly different from those obtained for the SCGs-AC sample. These images showed the extent of inhomogeneity of the surfaces of C1-AC and C2-AC, which displayed a more compact and less porous structure compared to the adsorbent generated from spent coffee grounds. By comparing the ESEM-EDX elemental composition of the three samples (Table 2), it emerged that SCGs-AC contained a much higher amount of oxygen with respect to the commercial materials (C/O ratio 6.23 for SCGs-AC; 10.38 for C1-AC and 12.72 for C2-AC), which could be reasonably due to the peculiar preparation process carried out through KOH activation. Since adsorption of a dye onto adsorbents is also a physical phenomenon in which dye molecules attach onto the adsorbent surface under the influence of van der Waals forces and hydrogen bonding, the presence of a higher amount of oxygen species in the adsorbent matrix could be advantageous for the adsorption of By comparing the ESEM-EDX elemental composition of the three samples (Table 2), it emerged that SCGs-AC contained a much higher amount of oxygen with respect to the commercial materials (C/O ratio 6.23 for SCGs-AC; 10.38 for C1-AC and 12.72 for C2-AC), which could be reasonably due to the peculiar preparation process carried out through KOH activation. Since adsorption of a dye onto adsorbents is also a physical phenomenon in which dye molecules attach onto the adsorbent surface under the influence of van der Waals forces and hydrogen bonding, the presence of a higher amount of oxygen species in the adsorbent matrix could be advantageous for the adsorption of cationic dyes, such as methylene blue, and detrimental in the presence of anionic organic molecules, such as erythrosine-B [35,60].

Organic Dyes Removal
Adsorption tests on organic dyes were performed as preliminary studies with the aim of evaluating the affinity between PACs and adsorbates with different ionic properties. Adsorption occurred in a two-stage process, with a steep initial phase until 5 min for MB (Figure 2a) and 20 min for EB and BB (Figure 2b,c), as a result of the high concentration gradient. The only exception was for BB with C2-AC, whose trend continuously increased over time, reaching a plateau after approximately 2 h of contact time. It is usually reported that, at the beginning of the adsorption process, the filling of PACs pores is predominant, resulting in a marked removal rate ( Figure 2). Once the pores volume has been filled, chemisorption is the driving force, whose intensity is proportional to the number of active sites in PACs surfaces [42]. The adsorption trends were confirmed by the results of the kinetic modelling ( Table 3). The processes followed the pseudo-second-order kinetic in which the adsorption was proportional to the gradient squared (Equation (3)). Similar results were observed by Laksaci et al. [37], who performed adsorption tests on MB, and by Anastopoulos et al. [35], who reviewed the use of coffee as adsorbent for dyes and heavy metals in wastewaters. All PACs showed higher adsorption capacities for ionic adsorbates (Figure 2) as MB and EB. In particular, C2-AC presented always an intermediate removal rate with respect to SCGs-AC and C1-AC. SCGs-AC and C2-AC was the best PAC for MB adsorption, achieving 100% removal in 5 min of contact time (Table 4). With regard to EB, the removal rate of SCGs-AC was lower than those of C1-AC and C2-AC. However, the results were comparable if 180 min of adsorption time were considered (Table 4). Since the kinetics of adsorption could be affected by the appropriate surface functional groups of the matrix to attract the adsorbate molecule, it was suggested that the higher amount of oxygen species present in the SCGs-AC sample (as determined by ESEM-EDX, Table 2) could create a wider series of van der Waals interactions and hydrogen capable of attracting more rapidly MB.
Appl. Sci. 2020, 10, x 11 of 16 Figure 2. Removal rate (mmol/g of AC) during adsorption (min) for organic dyes and phenolic compounds. The reported data are the average of triplicate measurements; the standard deviations were always less than 5%. The adsorption time data for MB are reported until 25 min to see clearly the adsorption trend, because it is faster than the others.

Phenolic Compounds Removal
At pH about 6 all the phenols considered are in the protonated form (pKa = 9.89, 9.12, and 9.6-10.2 for Ph, CP, and BPA, respectively), thus the adsorption performance of the AC are reasonably due to a complex series of van der Waals and steric interactions.
Similarly, to the behavior of organic dyes, adsorption for Ph and CP occurred in a two-stage process with all the tested PACs (Figure 2d and e). After 20 min of contact time, the equilibrium was almost established, reaching a reduction of 71.2% for Ph and 91.9% for CP with SCGs-AC, of 58.4% for Ph and 89.7% for CP with C1-AC, and of 72.2% for Ph and 95.6% for CP with C2-AC (Table 4). As regards BPA, the adsorption curves (Figure 2f) continuously increased without reaching the equilibrium in 3 h of contact time. According with Çeçen and Aktaş [63], adsorption processes longer than 3 h are not technically and economically convenient in real applications, so the tests were stopped after such duration In agreement with what was observed for organic dyes and according to the literature [64], the pseudo-second-order kinetic was the best fitting model also for phenols' adsorption ( Table 3). The kinetic constants were higher in the adsorption of Ph and CP for all PACs, with respect BPA confirming the trends reported in Figure 2. Freundlich isotherms could be used for describing the The results of the adsorption isotherms (Table 5) showed that the equilibrium could be satisfactorily described by the Langmuir model, indicating a monolayer coverage onto char particles for commercial PACs and for SCGs-AC only with MB, as already reported by Spagnoli et al. [61]. Freundlich was the best fitting model for EB and BB with SCGs-AC. There are numerous reports on organic molecules adsorption in pores that are barely large enough to admit the adsorbate molecule and it has been pointed out that dyes and humic acids have a size (1.5-3.0 nm) that favors their adsorption in mesopores (i.e., pore diameter > 2 nm) [62]. Hence, the difference in pore size distribution can significantly affect the adsorption capacity for molecules of different sizes and shapes. For this reason, it is important to know the molecular dimensions of the pollutant adsorbate and ensure that the activated carbon used has sufficient pore diameter to achieve the removal of the adsorbate. Regarding the MB, EB, and BB molecules, they all have smaller molecular sizes (lower than 1.5 nm) compared with the average pore diameter of the active carbons used (Table 2).   To meaningfully evaluate the performance of the sorbents, PC values have been calculated and reported in Table 6 together with breakthrough levels at 5%, 10%, and 50% of adsorption based on the adsorption isotherms.
It is clear from the data reported in Table 6 that high adsorption capacities do not directly correlate to high PCs since the latter take into account also the residual concentration of pollutant. In general, the PCs of all the PACs considered are high for the dyes compared with the phenolic compounds, while the adsorption capacities have an opposite trend. An improvement in PC values implies enhancement in adsorption affinity: thus, the affinity of SCGs-AC for MB was significantly higher with respect C1-AC and C2-AC, in agreement with the surface properties of the ACs. Coherently, SCGs-AC exhibited a significantly lower affinity for EB and BB. The extremely high value of PC for C1-AC for BB could be considered in agreement with the extremely rapid adsorption rate observed in Figure 2. As for 3-chlorophenol (CP), the SCGs-AC shows both higher adsorption capacity (3765 mg/g) and partition coefficient (81.53 mg/g/µM) than the commercial PACs, representing and interesting candidate as compared with recent systems proposed in the literature (e.g., Batool et al. [56] report for 2,4-dichlorophenol a maximum adsorption capacity of 45.95 mg/g and PC = 41.77 mg/g/µM).  1 The q e (BT 100%) corresponds to saturation conditions.

Phenolic Compounds Removal
At pH about 6 all the phenols considered are in the protonated form (pKa = 9.89, 9.12, and 9.6-10.2 for Ph, CP, and BPA, respectively), thus the adsorption performance of the AC are reasonably due to a complex series of van der Waals and steric interactions.
Similarly, to the behavior of organic dyes, adsorption for Ph and CP occurred in a two-stage process with all the tested PACs (Figure 2d,e). After 20 min of contact time, the equilibrium was almost established, reaching a reduction of 71.2% for Ph and 91.9% for CP with SCGs-AC, of 58.4% for Ph and 89.7% for CP with C1-AC, and of 72.2% for Ph and 95.6% for CP with C2-AC (Table 4). As regards BPA, the adsorption curves ( Figure 2f) continuously increased without reaching the equilibrium in 3 h of contact time. According with Çeçen and Aktaş [63], adsorption processes longer than 3 h are not technically and economically convenient in real applications, so the tests were stopped after such duration.
In agreement with what was observed for organic dyes and according to the literature [64], the pseudo-second-order kinetic was the best fitting model also for phenols' adsorption ( Table 3). The kinetic constants were higher in the adsorption of Ph and CP for all PACs, with respect BPA confirming the trends reported in Figure 2. Freundlich isotherms could be used for describing the adsorption of Ph and CP for all the PACs as reviewed by Hussain et al. [29], while the Langmuir model was more suitable for BPA as confirmed by Zbair et al. [47].
It is to observe that considering the adsorption capacities and the PC values, the same trends were obtained only for Ph (SCGs > C1-AC > C2-AC). As for CP, the higher performance was demonstrated by SCGs-AC either considering the adsorption capacity or the PC values, while the behavior of C1-AC and C2-Ac resulted to be inverted considering the two different parameters. As for BPA, SCGs have lower performance with respect C2-AC in both cases, the latter being the best adsorbent after 3 h of contact time (Table 4).
In general, the adsorption capacity of PACs was higher for CP without any noticeable difference between the three adsorbents ( Figure 2e). The removal rates of both Ph and CP for SCGs-AC and C2-AC were comparable (about 140 mg/g), while those for C1-AC were lower (approximately 120 mg/g). Similar results were obtained for phenolic compounds by Laksaci et al. [37,38], Castro et al. [32], and even better than those described by Lamine et al. [64]. The same considerations could not be extended to SCGs-AC with BPA.
According to Pendleton et al. [65], the trend of BPA adsorption capacity is inversely related to the oxygen content of the adsorbents (Table 2). Fonseca Alves et al. [64] gained 98% of BPA removal with AC from SCGs activated with ZnCl 2 , under the same pH and temperature conditions of the reported study, even if he adsorption time was 24 h (3 h in this experiment). Better results could be found in the literature with other adsorbents, as reviewed by Bhatnagar and Anastopoulos [40]. In particular, ultra-high-surface-area porous AC from low cost asphalt seemed to be effective for the removal of both BPA and other endocrine disrupting chemicals from wastewater [66]. Techniques other than adsorption are available for BPA removal, including ozonation [67], membrane filtration [68], plasma [69], and advance oxidation processes [70]. Despite the high efficiency, all these techniques are characterized by high investment and operational costs, as opposed to adsorption processes [71]. The lower removal efficiency of SCGs-AC for BPA could be due to the morphology and to a poor electrostatic interaction established between BPA, which was reported to be in its neutral form at pH below 8 [72] and the partially negatively charged SCGs-AC surface.

Conclusions
Emerging pollutants pose a severe risk for the ecosystems and the human health, since they are continuously released into the environment. Among them, phenolic compounds play an important role being widely used in industrial applications. In this study, three different PACs were used to remove phenolic compounds after preliminary analyses on organic dyes. All the samples displayed very high surface area values, ranging from about 820 m 2 g −1 to 1200 m 2 g −1 . In particular, the sample SCGs-AC, prepared by spent coffee grounds treated with KOH and then pyrolyzed at high temperature, showed a 30% higher surface area value with respect the commercial active carbons used for comparison purposes. Moreover, the mean pore diameter was between micro-and mesopore regions for this sample and typical of mesoporous materials for the commercial ones. Adsorption occurred via multiple mechanisms, including pore filling and adsorbent-adsorbate π-π and H-bonding interactions. SCGs-AC proved to have similar adsorption capacities than commercial PACs for the removal of organic dyes and phenolic compounds. In particular, PACs showed higher affinity with ionic substances, as MB and EB, as confirmed by the high partition coefficients and in agreement with their surface properties. Regarding Ph and CP, the removal rate was around 70% and 95% for all PACs. BPA was harder to remove from solution in comparison to the other phenolic compounds, with the lowest value of PC for SCGs-AC. Future studies will be aimed at improving the chemical properties of the produced SCGs-AC, via surface functionalization, to obtain a higher removal efficiency for phenolic compounds and other endocrine-disrupting agents. Moreover, LCA and LCC studies will be carried out to evaluate whether this application could be feasible from both the environmental and economical point of view at larger scale.

Conflicts of Interest:
The authors declare no conflict of interest.