Efficient removal of pharmaceuticals from water using graphene nanoplatelets as adsorbent — TEI-tagged view

Fatin Ahza RoslipersName, Haslina AhmadpersName, Khairulazhar JumbripersName, Abdul Halim AbdullahpersName, Sazlinda KamaruzamanpersName, Nor Ain Fathihah AbdullahpersName (2021date). “Efficient removal of pharmaceuticals from water using graphene nanoplatelets as adsorbent.” Royal Society Open Sciencetitle·j.

Object of study — adsorptive removal of sulfamethoxazoleterm and acetaminophenterm·key from water by graphene nanoplateletsterm·dom

Research question — how efficiently GNPsterm·dom remove SMXterm and ACM from water by adsorptionterm·mth

Contribution — GNPterm·dom C300term identified and optimised as an adsorbent reaching 99% removal, with regenerationterm·key and real-water tests

Full text — tagged

The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.

Efficient removal of pharmaceuticals from water using graphene nanoplateletsterm·dom as adsorbent Abstract Recently, pharmaceutical pollutantsterm·key in water have emerged as a global concern as they give threat to human health and the environment. In this study, graphene nanoplateletsterm·dom (GNPsterm·dom) were used to efficiently remove antibiotics sulfamethoxazoleterm (SMXterm) and analgesic acetaminophenterm·key (ACM) as pharmaceutical pollutantsterm·key from water by an adsorptionterm·mth process. GNPsterm·dom; C750term, C300term, M15 and M5 were characterized by high-resolution transmission electron microscopy, Raman spectroscopy, X-ray diffraction and Brunauer–Emmett–Teller. The effects of several parameters viz. solution pHterm·dom, adsorbent amount, initial concentrationterm·dom and contact time were studied. The parameters were optimized by a batch adsorptionterm·mth process and the maximum removal efficiencyterm·mth for both pharmaceuticals was 99%. The adsorptionterm·mth kinetics and isotherms models were employed, and the experimental data were best analysed with pseudo-second kinetic and Langmuir isothermterm·key with maximum adsorption capacityterm·mth (Q m ) of 210.08 mg g −1 for SMXterm and 56.21 mg g −1 for ACM. A regenerationterm·key study was applied using different eluents; 5% ethanol-deionized water 0.005 M NaOH and HCl. GNPterm·dom C300term was able to remove most of both pollutants from environmental water samples. Molecular dockingterm·key was used to simulate the adsorptionterm·mth mechanism of GNPterm·dom C300term towards SMXterm and ACM with a free binding energy of −7.54 kcal mol −1 and −5.29 kcal mol −1 , respectively, which revealed adsorptionterm·mth occurred spontaneously.

Introduction Water pollutants found in the environment can be divided into several categories viz. heavy metals, dyes, pesticides, pharmaceuticals and personal care products, phenolics, and others (hydrocarbons, inorganic anions, etc.) [ 1 ]. The source of these noxious pollutants come from untreated discharges of irresponsible textile, pharmaceutical, agricultural and mining factories. Pharmaceutical pollutantsterm·key have attracted interest among researchers to solve this problem which can cause serious health effects towards humans and may harm the environment. Research on the fate of pharmaceuticals in water prove that there are still traces found downstream [ 2 ]. Long-term exposure to low concentrations of pharmaceuticals in water and the synergistic effects of mixtures may pose risks to human health, especially to sensitive subpopulations [ 3 – 5 ]. It gains more concern as it is reported most pharmaceuticals are non-biodegradable [ 6 – 8 ]. Sulfamethoxazoleterm (SMXterm) is regularly prescribed as an antibiotic either as individual medication or combined with other drugs such as trimethoprim, a combination commercially known as Bactrim. Bactrim is used to treat several diseases, for instance, many types of bacterial infection namely urinary tract infection including pneumonia. Known SMXterm contraindications cause allergy, hypersensitivity to either component, hyperkalaemia, severe hepatic or renal dysfunction [ 9 ]. Acetaminophenterm·key (ACM) are used as an analgesic for mild to moderate pain and an anti-pyretic [ 10 ]. The use of ACM is prevalent among surveyed consumers and has the highest consumption levels in Malaysia [ 11 ]. To put it more simply, because the water also contains other unwanted drugs, the incident of unconscious consummation of other drugs causes interaction with other drugs, thus leading to the worst condition in a prolonged period. It is known that not all drugs are completely digested in the human body which are then secreted as urine into wastewater and it is found that several amounts of concentration of SMXterm are found in water bodies; approximately 15% of the unmetabolized fraction goes into sewage after being ingested and subjected to human metabolism [ 12 ]. Table 1 shows the maximum concentration of SMXterm and ACM in surface water in Asia-Pacific and globally. Table 1. Maximum concentration of sulfamethoxazoleterm and acetaminophenterm·key found in water [ 13 ]. pharmaceutical substance therapeutic group Asia-Pacific (μg l −1 ) global (μg l −1 ) a sulfamethoxazoleterm antibiotics 14.3 29.0 acetaminophenterm·key analgesics 9.17 230.0 a Includes Asia-Pacific, Africa, Eastern Europe, Latin America and Caribbean, western Europe and others in United Nations regions.

There are various removal methods that are used to remove pollutants from water such as bioreactors [ 14 ], degradation and photodegradation [ 15 – 18 ], filtration [ 19 ], photocatalysis [ 20 , 21 ], photo-Fenton and Fenton-like [ 22 , 23 ], coagulation [ 24 ] and adsorptionterm·mth [ 25 , 26 ]. Among these methods, adsorptionterm·mth is a widely used and most efficient method for wastewater treatment. This technique is superior and considered as powerful alternatives to conventional methods [ 27 ] because of its low cost, ease of operation, great efficiency, high capacity, simplicity, reliability and less energy consumption [ 28 , 29 ]. Graphene has gained interest as an adsorbent owing to its large surface area, accessible surface, high hydrophobicity and high mechanical properties [ 30 , 31 ]. As such, many researchers have explored graphene or graphite-based materials as an adsorbent for different pollutants [ 32 – 34 ] owing to its enormous, delocalized π-π electron system that can form strong bonding with various pollutants [ 35 ]. In previous research, graphene nanoplateletsterm·dom (GNPterm·dom) C750term have been studied as an adsorbent with good adsorption capacityterm·mth, yet are quite expensive and other types of GNPsterm·dom with different surface areas are unexplained [ 35 , 36 ]. In this study, the removal efficiencyterm·mth of SMXterm and ACM by the adsorptionterm·mth process from water using GNPsterm·dom was evaluated. Different types of GNPsterm·dom: C750term, C300term, M15 and M5, were screened to choose the best adsorbent and further optimized by several parameters: solution pHterm·dom, GNPterm·dom amount, initial concentrationterm·dom and contact time. Langmuir and Freundlich isotherms were applied to analyse the adsorptionterm·mth equilibrium. Pseudo-first order and pseudo-second order were also discussed for the kinetic study of removal capacity and to identify the behaviour and mechanism of adsorptionterm·mth. Molecular dockingterm·key studies were also implemented to give interaction insight of GNPterm·dom with SMXterm and ACM.

Materials and methods Chemicals and materials GNPsterm·dom C750term, C300term, M15 and M5 were purchased from XG Sciences, USA. SMXterm, 98% was obtained from Alfa Aesar, USA and ACM (4-acetamidophenol), 98% purity was purchased from Acros Organic ( table 2 ). Deionized water was used as a solvent for a stock solution and further dilution. The solution pHterm·dom was measured using a HI-2002 Edge pH Meter, Hanna Instrument, USA. Table 2. Physicochemical properties of pharmaceuticals. pharmaceutical chemical formula CAS No solubility in water (mg l −1 ) dissociation constant, pKa molecular weight (g mol −1 ) sulfamethoxazoleterm C 10 H 11 N 3 0 3 S 723-46-6 610 1.6 and 5.7 253.28 acetaminophenterm·key C 8 H 9 NO 2 103-90-2 14 000 9.5 151.16 Characterization of adsorbents The compositional and morphological properties of GNPsterm·dom were determined using a high-resolution transmission electron microscope (HRTEM) JEM-2100F operated at 200 kV (JEOL, Japan). Specimens were prepared by dispersing the GNPsterm·dom in 100% acetone and adding a drop of dispersed GNPterm·dom on the copper grid. Raman spectroscopy was performed using WITec Alpha 300R (WITec GmbH, Germany). Brunner–Emmett–Teller (BET) and Barrett–Joyner–Halenda (BJH) for specific surface area, pore volume, pore size and pore width distribution of GNPsterm·dom were analysed using MicroActive for TriStar II Plus 2.03 (Micromeritics, USA), the samples were degassed with nitrogen gas at about 150°C with saturation pressure of approximately 756 mmHg overnight before analysis. X-ray diffraction analysis used LabX XRD-6000 (Shimadzu, Japan) by scanning the GNPsterm·dom at 2 θ of 20°–80° at a speed angle of 2° min −1 with Cu K α radiation. Zeta potentialterm·mth and average diameter were measured using Zetasizer Nano (Malvern Panalytical Ltd, UK).

Preparation of solution The stock solution of SMXterm was freshly prepared by dissolving 10 mg in a 100 ml volumetric flask to produce 100 mg l −1 of the pharmaceutical solution then further diluted to lower concentration using deionized water. The calibration curve of SMXterm was plotted by absorbance at 264 nm and ACM at 243 nm. The regression equation from E xcel was used to calculate the concentration after the adsorptionterm·mth process, C e .

Preliminary adsorptionterm·mth study Different types of GNPsterm·dom with different surface areas were used towards the removal of SMXterm and ACM to select the best adsorbent among four types of GNPsterm·dom: C750term, C300term, M15 and M5. The study was carried out by preparing a 10 ml of 20 mg l −1 SMXterm solution and using 20 mg of GNPsterm·dom as adsorbent. The solution was then sonicated for 20 min which was later centrifuged, and the supernatant was filtered by using a 0.22 µm cellulose acetate (CA) syringe filter. The concentration of the supernatant was obtained by detecting absorbance using a UV-Vis spectrophotometer (Shimadzu UV-1650PC).

Batch adsorptionterm·mth experimentsterm·mth Batch adsorptionterm·mth experimentsterm·mth were performed by preparing 10 ml of the pharmaceutical solution to investigate the effect of adsorptionterm·mth parameters: solution pHterm·dom, GNPterm·dom amount, initial concentrationterm·dom and contact time. The stock solution was prepared then further diluted to set the initial concentrationterm·dom. The solution pHterm·dom was adjusted using 0.1 M NaOH and 0.1 M HCl and measured using a HI-2002 Edge pH Meter (Hanna Instrument, USA). GNPsterm·dom were added into the solution and were sonicated continuously for a certain time. Then, the solution was centrifuged to collect the supernatant and was filtered through a 0.22 µm CA syringe filter, to ensure there is no trace of GNPterm·dom in the solution before running the sample through the UV-Vis spectrophotometer to avoid ambiguous results. The removal efficiencyterm·mth of the solution was acquired from the calculation by using the following equation: 2.1 removal efficiencyterm·mth ( % ) = ( C i − C e ) C i × 100 , where C i is the initial concentrationterm·dom of the pharmaceutical solution (mg l −1 ) and C e is the equilibrium concentration of the solution retrieved after the removal process (mg l −1 ). The adsorption capacityterm·mth was calculated according to the equation: 2.2 q e = ( C i − C e ) V m , where q e is the amount of pharmaceutical adsorbed by the GNPsterm·dom (mg g −1 ) or known as adsorption capacityterm·mth at equilibrium, V is the volume of pharmaceutical solution (l) and m is the mass of GNPsterm·dom used (g). The reported values were the means of triplicates. Isotherms and kinetics studies were further studied to understand the adsorptionterm·mth mechanisms of SMXterm and ACM onto GNPterm·dom C300term. Table 3 exhibits the expressions of both isotherms and kinetics. For the adsorptionterm·mth isotherm, Langmuir type and Freundlich models were chosen while pseudo-first-orderterm·mth, pseudo-second-orderterm·key and intraparticle diffusion model were applied to understand the kinetics. Table 3. Linearized expressions for each adsorptionterm·mth isotherm and kinetic. isotherms and kinetics expressions linear expression plot Langmuir isothermterm·key q e = q m [ K L C e 1 + K L C e ] 1 q e = [ 1 K L I I q m ] [ 1 C e ] + 1 q m 1 q e V s 1 C e Freundlich isotherm q e = K F C e 1 / n ln q e = ln K F + 1 n ln C e ln q e Vs ln C e pseudo-first-orderterm·mth kinetic q t = q e ( 1 − e − k 1 t ) ln( q e − q t ) = − K 1 t + lnq e ln( q e − q t ) Vs t pseudo-second-orderterm·key kinetic q t = k 2 t q e k 2 t + 1 t q t = [ 1 q e ] [ t ] + 1 K 2 q e 2 t q t V s t intraparticle diffusion q t = K id t (0.5) q t = K id t (0.5) + C q t Vs t (0.5) Regenerationterm·key A regenerationterm·key study was carried out using the optimized parameters for adsorptionterm·mth. The used adsorbent was dispersed in 10 ml of different eluents: 0.005 M NaOH, 0.005 M HCl and 5% ethanol-deionized water then sonicated for 30 min for desorption process. The washed adsorbent was then centrifuged, and the supernatant was discarded. The adsorbent was dried in the oven around 55°C. The dried adsorbent was reused for the adsorptionterm·mth process and the steps were repeated for several cycles.

Real water application Three different real water samples: Langat river water, Kuyoh river water and laboratory tap water were collected for environmental application. The samples were filtered using filter paper and kept at 4°C in the dark. The properties of the three real water samples were analysed. Real water samples were spiked with a concentrated solution of ACM and SMXterm (100 mg l −1 ) were added into the real water samples to obtain a sample with final concentrations of 20 mg l −1 . Twenty milligrams of GNPsterm·dom was added to the spiked samples and sonicated for 20 min for the removal process. The sample was centrifuged, and the supernatant was filtered with a 0.22 µm CA syringe filter into a vial ready to be analysed. Chemical oxygen demand (COD, mg l −1 ) and biochemical oxygen demand (BOD, mg l −1 ) were measured following the APHA 5220 and APHA 5210 methods, respectively. Total soluble solids (TSS, mg l −1 ) was analysed, a Hach DR 900 (Hach, USA) and a WTW Inolab meter (WTW GmbH, Germany) was used to measure total dissolved solids (TDS, mg l −1 ). The concentration of sodium, magnesium and calcium were analysed using inductively coupled plasma-optical emission spectrometry by referring to the APHA 31110b method ( table 4 ). Table 4. Analysis of the Langat river water, Kuyoh river water and tap water. Langat river water Kuyoh river water tap water chemical oxygen demand, COD (mg l −1 ) 43.63 77.57 14.60 biochemical oxygen demand, BOD (mg l −1 ) 2.15 7.20 1.40 total soluble solids, TSS (mg l −1 ) 35.00 18.00 0 total dissolved solids, TDS (mg l −1 ) 171.00 282.00 150.00 sodium, Na (mg l −1 ) 5.50 10.00 7.58 magnesium, Mg (mg l −1 ) 1.25 1.84 1.27 calcium, Ca (mg l −1 ) 13.32 24.86 14.64 Results and discussion Characterization of graphene nanoplateletsterm·dom By using an HRTEM at 10 000 times magnification, physical features of GNPsterm·dom were analysed. Figure 1 displayed the structure of GNPsterm·dom which possessed wrinkle flakes with multiple layers. C750term ( figure 1 a ) has an irregular shape which resembles more like platelets, more wrinkle features and smaller size with greater aggregation compared to C300term, M15 and M5. As shown in figure 1 c,d , M5 and M15 exhibit bigger flakes with a less overlapping layer. Figure 1. HRTEM images of ( a ) C750term, ( b ) C300term, ( c ) M15 and ( d ) M5 at 10 000 times magnification.

The thickness of layer GNPsterm·dom was evaluated at atomic resolution via Raman spectroscopy which shows vibrational bands of D, G and 2D ( figure 2 ). The D bandterm·dom, known as the disorder band appears around 1350 cm −1 because of out of plane vibrations attributed to the presence of structural defects. The presence of this band proved the GNPsterm·dom have ragged edges [ 37 , 38 ]. This result is also confirmed in the HRTEM images. The G bandterm·dom which is a sharp peakterm·dom that exists around 1587 cm −1 resulted from the in-plane vibration of sp 2 hybridized carbon atoms. The 2D band is the second order of the D bandterm·dom, sometimes denoted as the G’ band that appears around 2700 cm −1 , resulted from a two-phonon lattice vibrational process without defect. The shape of 2D bands was wide and low and the full width at half maximum of the 2D bands for all GNPsterm·dom greater than 90 cm −1 perfectly categorized as multilayer graphene [ 39 ]. Intensity ratio of 2D and G peaks were used to determine the number of graphene layers. The ratios I 2D / I G of four different GNPsterm·dom are less than 1, thus confirming the multilayer properties of these adsorbents. Figure 2. Raman spectra of GNPsterm·dom C750term, C300term, M15 and M5.

Figure 3 shows the X-ray diffraction patterns of GNPterm·dom C750term, C300term, M15 and M5. Peaks shown in the figure correspond with the characteristic peaks of pristine graphite which could be assigned to the (002), (100), (004) and (110) hexagonal planes by referring to the JCPDS Card No (75-1621). C300term ( figure 3 b ) shows more distinct peaks with a higher intensity that resemble graphite compared with the X-ray diffraction pattern of C750term ( figure 3 a ) that only exhibits peaks of (002) and (100) with a broader peak. Figure 3. X-ray diffraction spectra for GNPsterm·dom ( a ) C750term, ( b ) C300term, ( c ) M15 and ( d ) M5.

Through BET and BJH analysis, the surface area, micropore area, pore size and pore volume of adsorbent were analysed and summarized in table 5 . The surface area C750term (738.44 m 2 g −1 ) was found to be the highest compared to others followed by C300term (390.30 m 2 g −1 ), M15 (61.51 m 2 g −1 ) and M5 (63.55 m 2 g −1 ). The pore diameter of C750term, C300term, M15 and M5 can be categorized as mesopores as they fall in the range of (2–50 nm) according to the International Union of Pure and Applied Chemistry classification. Table 5. Summary of BET and BJH analysis for GNPterm·dom C750term, C300term, M15 and M5. GNPterm·dom BET surface area (m 2 g −1 ) micropore area (m 2 g −1 ) total pore volume (cm 3 g −1 ) external surface area (m 2 g −1 ) BJH pore width (nm) C750term 738.44 218.71 0.78 519.73 4.96 C300term 390.30 83.65 0.41 306.65 4.66 M15 61.51 10.00 0.10 51.51 5.38 M5 63.55 17.96 0.08 45.59 4.94 Figure 4 exhibits adsorptionterm·mth/desorption isotherm type IV which is similar to reported by Al-Khateeb et al . [ 36 ] with an H3 type hysteresis loop that confirms there were aggregates (loose assemblages) of platelet like particles forming slit-shaped mesopores [ 40 ]. The intermediate flat region in the isotherm corresponds to monolayer formation at the lower pressure region of the graph then followed by multilayer [ 41 ]. Figure 4. N 2 adsorptionterm·mth/desorption isotherms for GNPterm·dom ( a ) C750term, ( b ) C300term, ( c ) M15 and ( d ) M5.

Zeta potentialterm·mth measured the electrostatic repulsion between similarly charged particles and hence indicates a degree of dispersion efficiency thus giving information tendency of agglomeration, and a larger zeta potentialterm·mth may mean there is a better dispersion [ 42 , 43 ]. Table 6 exhibits the measured zeta potentialterm·mth of GNPterm·dom C300term (−50.80 mV) is the most stable compared to other three GNPsterm·dom. GNPterm·dom M5 has the least absolute value with −20.87 mV, which considered its tendancy to aggregate since it is lower than −30 mV [ 44 ]. Both GNPterm·dom C750term and C300term possessed an average diameter of less than 2 µm, while M5 and M15 with 4.21 µm and 6.79 µm, respectively. Table 6. Zeta potentialterm·mth and average diameter of different types of graphene nanoplateletsterm·dom using dynamic light scattering. GNPterm·dom average diameter (nm) zeta potentialterm·mth (mV) C750term 933.67 −44.63 C300term 1010.00 −50.80 M15 6794.67 −33.90 M5 4205.33 −20.87 Preliminary adsorptionterm·mth study Figure 5 presents the preliminary adsorptionterm·mth study of four different types of GNPsterm·dom C750term, C300term, M15 and M5 towards SMXterm. Based on the characterization, C750term possessed the advantage of a higher surface area of the four materials tested. However, it is found that C300term has the highest removal efficiencyterm·mth compared to other types of GNPsterm·dom despite the fact C750term has a higher surface area. Furthermore, C300term has the highest stability based on the zeta potentialterm·mth. This finding was contrary to expectations that a higher surface area will have better adsorptionterm·mth efficiency [ 45 ], though the result of removal efficiencyterm·mth is similar to what has been reported by Al-Khateeb [ 35 ]. Thus, other factors must be taken into consideration to achieve a higher percentage of adsorptionterm·mth. This might be owing to clumping or aggregation of C750term because of its high surface area which causes lower yet good adsorptionterm·mth of SMXterm. After centrifugation and filtration, some GNPsterm·dom may have not filtered thoroughly owing to the small size of C750term able to go through 0.22 µm CA syringe filter. Because the objective of this preliminary study was to choose the best adsorbent, C300term was chosen. On top of that, C300term has a lower manufacturing cost compared to C750term which reduces the total cost for the overall process [ 46 , 47 ]. Thus, C300term was further used for the batch adsorptionterm·mth study with different parameters: solution pHterm·dom, GNPterm·dom amount, initial concentrationterm·dom of SMXterm and contact time. Figure 5. Preliminary test of different type of graphene nanoplateletsterm·dom (GNPsterm·dom).

Evaluation of adsorptionterm·mth behaviours of graphene nanoplateletsterm·dom C300term towards sulfamethoxazoleterm and acetaminophenterm·key Effect of solution pHterm·dom To assess the adsorptionterm·mth behaviours of GNPterm·dom C300term, a batch adsorptionterm·mth study was performed to remove SMXterm and ACM by using GNPterm·dom C300term as adsorbent. Solution pHterm·dom is an important factor in the adsorptionterm·mth of pharmaceuticals. The effect of solution pHterm·dom on the adsorptionterm·mth of pharmaceuticals was studied from pH 2 to 10 in a 10 ml solution volume of 20 mg l −1 concentration using 20 mg GNPterm·dom C300term for 20 min. As displayed in figure 6 , the adsorptionterm·mth ability of GNPterm·dom C300term towards both SMXterm and ACM was affected by the pH value of the solution. As the initial pH of the solution increased from 2 to 4, the adsorption capacityterm·mth of SMXterm onto GNPterm·dom C300term gradually increased ( figure 6 a ). On the other hand, the adsorption capacityterm·mth and removal efficiencyterm·mth for ACM increased gradually from acid to neutral then decreased as it further increased ( figure 6 b ). Maximum adsorptionterm·mth was detected at pH 4 (99.27%) for SMXterm and ACM at pH 8 (98.38%). The lowest removal for SMXterm was detected at pH 10 with 85.65% and for ACM at 76.8% at pH 2. The increasing solution pHterm·dom was unfavourable for the removal of SMXterm molecules and resulted in a low adsorption capacityterm·mth at higher pH. Because SMXterm is known as an acidic organic compound with dissociation constant (pKa) of 1.6 and 5.7, the removal of the targeted compound was better under acidic conditions (pH < pKa) than at basic pH (pH > pKa). In other words, when solution pHterm·dom is less than their pKa values, they are present in aqueous solution as a protonated form and these observations are in agreement with the results in the previous works [ 33 , 48 ]. Figure 6. The effect of solution pHterm·dom on ( a ) sulfamethoxazoleterm and ( b ) acetaminophenterm·key using adsorptionterm·mth by GNPterm·dom C300term.

Effect of graphene nanoplateletsterm·dom amounts Different amounts of GNPterm·dom C300term were used for adsorptionterm·mth towards a 10 ml of 20 mg l −1 adjusted to pH 4 in 20 min for SMXterm and pH 8 for ACM based on previous optimum conditions. From the results presented in figure 7 , with the increase in GNPterm·dom C300term amount from 5 mg to 60 mg, the removal efficiencyterm·mth increased; however, the adsorption capacityterm·mth for both pharmaceuticals decreased. At low adsorbent dosage, adsorptionterm·mth sites were fully available, resulting in higher adsorption capacityterm·mth. At a higher GNPterm·dom C300term amount, the active sites of the adsorbent which prepared to bind with the pharmaceutical molecules were not fully used in comparison to lower adsorbent dosages. The increase in the amount of GNPterm·dom C300term may cause aggregation of adsorbent and possibility of collisions between solid particles [ 49 , 50 ], and consequently, the available adsorptionterm·mth sites may decrease as well owing to the adsorption capacityterm·mth of GNPterm·dom C300term for SMXterm and ACM. Figure 7. The effect of GNPterm·dom amount on ( a ) sulfamethoxazoleterm and ( b ) acetaminophenterm·key using adsorptionterm·mth by GNPterm·dom C300term.

Effect of initial concentrationterm·dom It is crucial to study the effect of initial concentrationterm·dom upon revealing any adsorptionterm·mth process. Thus, a set of experimentsterm·mth was carried out at different initial concentrationsterm·dom of 10 to 100 mg l −1 using previously optimized parameters. Figure 8 shows that the adsorption capacityterm·mth of pharmaceutical increased as the initial SMXterm and ACM concentration increased from 10 to 100 mg l −1 with an almost linear trend. The higher the initial concentrationterm·dom, the more adsorbate molecules available for adsorptionterm·mth, resulting in a higher adsorption capacityterm·mth at higher pharmaceutical concentrations. However, as the initial concentrationterm·dom increased more than 40 mg l −1 , the removal efficiencyterm·mth slightly decreased. Nevertheless, the range of removal efficiencyterm·mth is still more than 98%. Figure 8. The effect of initial concentrationterm·dom on ( a ) sulfamethoxazoleterm and ( b ) acetaminophenterm·key using adsorptionterm·mth by GNPterm·dom C300term.

Effect of contact time The contact time between both pharmaceuticals and C300term was studied with several sets of batch adsorptionterm·mth experimentsterm·mth using the previously optimized condition at different times of 2 to 60 min, and the results are presented in figure 9 . As shown in the figure, SMXterm and ACM adsorptionterm·mth by GNPterm·dom C300term occurred very fast initially and then the rate slowed down as the contact time increased. Both adsorption capacityterm·mth and removal efficiencyterm·mth have the same trend. During the initial period, the adsorption capacityterm·mth of the pharmaceuticals rapidly increased with time as many active sites were available on the surface of GNPterm·dom C300term for binding of pharmaceutical molecules [ 51 ]. These available active sites were gradually occupied as time increased thus resulting in reducing adsorptionterm·mth rate [ 50 ]. Figure 9. The effect of contact time on ( a ) sulfamethoxazoleterm and ( b ) acetaminophenterm·key using adsorptionterm·mth by GNPterm·dom C300term.

Adsorptionterm·mth isotherm and kinetic Langmuir and Freundlich isotherm models were applied to investigate the adsorptionterm·mth equilibrium between the pharmaceuticals and GNPterm·dom C300term. As shown in table 7 , the adsorptionterm·mth data fitted well with both Langmuir and Freundlich isotherms and this finding was similarly reported in [ 52 ]. The R 2 value obtained from the Langmuir isothermterm·key model were 0.9952 for SMXterm and 0.9946 for ACM, revealing that the Langmuir isothermterm·key model slightly fit better and is more suitable for describing the adsorptionterm·mth process of the pharmaceuticals onto GNPterm·dom C300term. This illustrated that adsorptionterm·mth of SMXterm and ACM on GNPterm·dom C300term was monolayer adsorptionterm·mth occurring on a homogenous surface without interaction between the adsorbate. Table 7. Comparison of adsorptionterm·mth isotherm and kinetic for sulfamethoxazoleterm and acetaminophenterm·key. isotherm Langmuir Freundlich n K L 2 R 2 n K F R 2 SMXterm 210.084 0.374 0.9952 1.193 52.034 0.9848 ACM 56.211 0.721 0.9946 1.435 21.230 0.9842 kinetic pseudo-first order pseudo-second order K 1 Q e R 2 K 2 Q e R 2 SMXterm 0.108 0.129 0.6743 7.059 6.653 0.9999 ACM 0.115 0.435 0.8218 1.815 6.667 0.9999 Adsorptionterm·mth kinetic with a sample concentration of 20 mg l −1 was further studied, and the kinetic parameters as well as the correlation coefficient, R 2 were acquired by linear regression. According to the data presented in the table, the pseudo-second-order modelterm·mth fitted better than the pseudo-first-orderterm·mth kinetic models. Based on the pseudo-second-order modelterm·mth, the correlation coefficients ( R 2 ) value was 0.9999 for both SMXterm and ACM which was greater than pseudo-first order. Thus, the pseudo-second-order modelterm·mth was more suitable for describing the process of SMXterm and ACM adsorptionterm·mth on GNPterm·dom C300term in agreement with chemisorption being the rate-controlling step [ 53 – 55 ]. The intraparticle diffusion model by Weber and Morris derived from Fick's second law of diffusion was applied to explain further on the diffusion mechanism which includes three steps. The first two steps are transport steps, and the last step is a reaction step [ 1 ]: (i) film diffusionterm·dom (external diffusion), which is the transport of adsorbate from the bulk phase to the external surface of the adsorbent; (ii) pore diffusionterm·dom, which is the transport of adsorbate from the external surface into the pores; and (iii) surface reaction, which the adsorbate will attach to the internal surface of the adsorbent. The data show multilinearity (electronic supplementary material, table S1) which indicates that the intraparticle diffusion controls the adsorptionterm·mth process within the early stage, but was not the rate-determining step for the whole adsorptionterm·mth process similarly found in previous research [ 35 ].

The maximum adsorptionterm·mth capacities for both ACM and SMXterm from this paper were compared with other adsorptionterm·mth capacities reported in the literature. Table 8 summarizes the previous research using graphene and activated carbon as adsorbent using the adsorptionterm·mth method. Table 8. Previous research on sulfamethoxazoleterm and acetaminophenterm·key. adsorbents pollutant adsorption capacityterm·mth (mg g −1 ) references graphene nanoplateletterm·dom C300term SMXterm 210.08 this work powder activated carbon SMXterm 86 [ 56 ] mesoporous silica-magnetic graphene oxide nanocomposite SMXterm 15.46 [ 34 ] multi-walled carbon nanotubes SMXterm 29 [ 57 ] graphene nanoplateletterm·dom C300term ACM 56.21 this work graphene nanoplateletsterm·dom C750term ACM 18.07 [ 35 ] graphene ACM 18.28 [ 58 ] activated carbon from Quercus Brantii (oak) ACM 45.45 [ 59 ] After adsorptionterm·mth analysis Table 9 compares the BET results before and after the adsorptionterm·mth of SMXterm and ACM. After the adsorptionterm·mth process, one can see the surface area of adsorbent decreased to that before the adsorptionterm·mth process, C300term (390.30 m 2 g −1 ) while after the adsorptionterm·mth process, C300-SMX (325.62 m 2 g −1 ) and C300-ACM (291.70 m 2 g −1 ). This shows adsorptionterm·mth had occurred as the SMXterm molecules adhered to the surface of GNPterm·dom C300term thus resulting in a notably lower surface area, micropore area and external surface area. Table 9. Comparison before and after adsorptionterm·mth of SMXterm and ACM towards GNPterm·dom C300term using BET analysis. BET surface area (m 2 g −1 ) micropore area (m 2 g −1 ) total pore volume (cm 3 g −1 ) external surface area (m 2 g −1 ) C300term 390.30 83.65 0.41 306.65 C300-SMX 325.62 38.33 0.37 287.29 C300-ACM 291.70 11.86 0.36 279.84 Field emission scanning electron microscopy (FESEM)/energy dispersive X-ray (EDX) was used to observe the adsorbent surface and elemental analysis ( figure 10 ). After the adsorptionterm·mth process, other elements were present on the surface of GNPterm·dom C300term of other elements besides carbon. The elements of oxygen and nitrogen were detected on the surface of GNPterm·dom C300term after adsorptionterm·mth of SMXterm and ACM. There was also the presence of sulfur found on the C300-SMX sample. Figure 10. FESEM images of GNPterm·dom C300term after adsorptionterm·mth of ( a ) SMXterm and ( b ) ACM. EDX analysis of GNPterm·dom C300term after adsorptionterm·mth of ( c ) SMXterm and (d) ACM.

To get a better view of the adsorptionterm·mth mechanisms, molecular dockingterm·key was applied, and the data are detailed in table 10 . Binding energy measures the affinity of the GNP-SMX complex and the sum of energies of each molecule separately. The lower the energy value, the more stable it is. The binding energies for both molecules were negative values; −7.54 kcal mol −1 and −5.29 kcal mol −1 for SMXterm and ACM, respectively, which means the SMXterm and ACM were bound spontaneously to the surface of GNPterm·dom C300term without consuming energy. The ΔGbind obtained for GNP-SMX is higher 2.25 kcal mol −1 than GNP-ACM which is in good agreement with experimental. In the experimental results, GNPterm·dom shows high maximum adsorption capacityterm·mth towards SMXterm (210.08 mg g −1 ) compared to ACM (56.21 mg g −1 ). This different adsorption capacityterm·mth of GNPterm·dom towards pollutants can be explained by the strength of the binding energy obtained from the docking results. Both pollutants preferred to bind at the surface layer of GNPsterm·dom. Van der Waals, hydrogen bonds and desolvation energy described energy loss with the interaction between pharmaceuticals and GNPterm·dom and solvent upon binding which revealed the chemisorption occurred exothermically (electronic supplementary material, figure S2). Table 10. Molecular dockingterm·key of GNP-SMX and GNP-ACM. (VDW, Van der Waals.) binding energy, Δ Gbind (kcal mol −1 ) VDW + H-bond + desolvation energy (kcal mol −1 ) electrostatic energy (kcal mol −1 ) GNP-SMX −7.54 −8.74 0.00 GNP-ACM −5.29 −5.89 0.00 The adsorptionterm·mth mechanism of GNPterm·dom towards the targeted pollutants can take place in the form of electrostatic interactions (cation or anion attraction), π–π interaction, hydrophobic interaction and hydrogen bonding [ 60 , 61 ]. Both SMXterm and ACM molecules contain an aromatic group or benzene ring (electronic supplementary material, figure S3). The adsorptionterm·mth mechanism is mainly owing to interaction of the electron-rich GNPterm·dom surface and the protonated aniline ring of pharmaceutical molecules. When the pH is near or equal to the pKa of pharmaceutical, the pharmaceutical molecules exist in neutral form thus it favours hydrogen bonding and π–π interaction with GNPsterm·dom. When the solution pHterm·dom drifted to more acid or basic conditions, electrostatic interactions will occur between the GNPterm·dom and pharmaceutical molecules.

Regenerationterm·key As depicted in figure 11 , the removal efficiencyterm·mth for both SMXterm and ACM using GNPterm·dom C300term through four regenerationterm·key cycles by different eluents: 0.005 M NaOH, 0.005 M HCl and 5% ethanol-deionized water is shown. By using chemical regenerationterm·key, interaction forces formed between the adsorbed pharmaceutical pollutantsterm·key and the eluent molecules causing the SMXterm and ACM molecules to leave the adsorptionterm·mth sites of the adsorbent and dissolve in the eluent [ 62 ]. Among the used eluents for the adsorptionterm·mth of SMXterm, 0.005 M HCl and 5% ethanol-deionized water were able to regenerate GNPterm·dom C300term where high removal efficiencies could be achieved even after the fourth cycle with 98.79%–95.18% and 98.33%–95.95%, respectively. When washed with 0.005 M NaOH, graphene efficiency was reduced from 99.28% to 78.8% in the second cycle, having the lowest percentage after the fourth cycle (32.84%). GNPterm·dom C300term was able to remove 95.58%–71.45% of ACM when washed with 5% ethanol-deionized water which is the highest followed by 0.005 M NaOH and 0.005 M HCl, with removal efficiencies of 96.01%–64.35% and 97.23%–26.08%, respectively. The surface of GNPterm·dom was held by weak intermolecular interactions viz. Van der Waals forces and dipole–dipole interaction [ 38 ]. After several cycles, the removal efficiencyterm·mth of GNPterm·dom is reduced because of the loss of GNPterm·dom during the washing process as it was not fully separated by centrifugation. Conclusively, regenerationterm·key of the GNPterm·dom C300term by 5% ethanol-deionized water was considered the suitable eluent for both pollutants, SMXterm and ACM with high removal efficiencies after several adsorptionterm·mth–desorption cycles. Figure 11. Removal efficiencyterm·mth of GNPterm·dom C300term ( a ) SMXterm ( b ) ACM versus regenerationterm·key cycle using different eluents.

Real water application Based on figure 12 , it can be concluded that GNPterm·dom C300term managed to remove most of the pharmaceutical pollutantsterm·key from the spiked real water samples. More than 98% of SMXterm were removed from all three real water samples: Langat river water, Kuyoh river water and tap water, while for ACM the removal efficiencyterm·mth was somewhat lower than SMXterm yet more than 90%. As shown in the figure, GNPterm·dom C300term able to remove, respectively, 93.83%, 95.76% and 96.45% of ACM, as for SMXterm 98.10%, 98.79% and 99.88% from Langat river water, Kuyoh river water and tap water. The results proved that GNPterm·dom C300term is also suitable for the real water application. Figure 12. Removal efficiencyterm·mth of GNPterm·dom C300term on real water samples; Langat river water, Kuyoh river water and tap water.

Conclusion In conclusion, all four types of GNPsterm·dom were able to remove SMXterm and ACM from water. GNPterm·dom C300term shows the highest adsorption capacityterm·mth compared to GNPterm·dom C750term, M15 and M5. The unexplored potential of GNPterm·dom C300term and the rest has revealed the other types GNPterm·dom that work better compared to C750term, which is a bit costly and faces an agglomeration problem. The removal of pharmaceuticals with different pKa values was strongly affected by pH. It is proved that the adsorptionterm·mth of SMXterm and ACM using GNPterm·dom C300term was owing to chemisorption as the data were best fitted with pseudo-second order. The adsorptionterm·mth data fitted well with both Langmuir and Freundlich isotherms. Owing to slightly higher R 2 , the Langmuir model was more suitable for describing the adsorptionterm·mth process of GNPterm·dom C300term towards both pharmaceuticals. In addition, GNPterm·dom C300term can remove up to 99% of SMXterm and ACM. On top of that, a negative value of binding energy from docking showed that the adsorptionterm·mth was spontaneous and exothermic. In addition, 5% ethanol-deionized water was found to be the best eluent for the regenerationterm·key of GNPterm·dom C300term as adsorbent for the removal of SMXterm and ACM. Real water application results showed that GNPterm·dom C300term managed to remove most of the pharmaceutical pollutantsterm·key from the environmental samples. GNPterm·dom C300term exhibits great potential as an excellent adsorbent with high adsorptionterm·mth ability for the removal of SMXterm and ACM, and thus may also be used towards other pharmaceutical pollutantsterm·key.

Supplementary Material