GO/PAMAM as a High Capacity Adsorbent for Removal of Alizarin Red S: Selective Separation of Dyes

Adsorption of Alizarin Red S (ARS) on graphene oxide/poly(amidoamine) (GO/PAMAM) was studied at different ARS initial concentrations, temperatures, pHs, shaking rates and contact times. Adsorption sites of GO/PAMAM were phenolic –OH (Ph) group of GO and amine groups (–NH2, –NH3 and –NHR2) of PAMAM dendrimer moieties of GO/ PAMAM. At pH = 2 and 318 K, maximum adsorption capacity (qe,max) of the adsorbent was 1275.2 mg g–1 which is one of the highest capacity in the literature. Thus, GO/PAMAM in this work acted as a superadsorbent for ARS. At the incipient of adsorption, ARS– molecules were adsorbed on Ph sites that was reaction-controlled step, (Ea = 114.5 kJ mol–1). Adsorption of ARS–on the remaining sites was diffusion–controlled. In alkaline media, two other types of ARS molecules were identified during that were adsorbed on Ph and –NH3 sites. Further increasing the pH of the solution, decreased the number these two sites and yielded a reduced adsorption capacity (qe,max). Methylene blue (MB), thionine (Th), pyronin Y (PY), acridine orange (AO), methyl blue (MEB) and janus green (JG) dyes were selectively separated from their mixtures with ARS molecules using GO/PAMAM at pH of 2. The used adsorbent was recycled efficiently by using ethylenediamine very fast.


Introduction
Wastewater generated by different industries contains pollutant compounds that are hazardous to the health of human beings and animals. Dye compounds included in these pollutants of wastewater are produced by industries like food, paper, rubber, textile, printing, tanning, dyestuff and pigment industries. During the last decades extensive progress has been made for treatment of industrial wastewaters. A number of techniques used for this purpose are filtration, 1 chemical oxidation, 2 ion exchange, 3 biological degradation, 4 reverse osmosis, 5 coagulation, 6 and adsorption. 7 Adsorption is a facile, cost effective and widely-applied method to remove dye compounds from wastewater systems and in many cases can be recycled easily.
In this study, graphene oxide/poly(amidoamine) (GO/PAMAM) was synthesized by grafting poly(amidoamine) (PAMAM) dendrimer to graphene oxide. [24][25][26] The as-synthesized GO/PAMAM and ARS-adsorbed GO/PAMAM were characterized by different techniques like BET (Brunauer-Emmett-Teller), SEM (Scanning Electron Microscope), EDS (Energy Dispersive X-Ray Spectroscopy), XRD (X-Ray Diffraction) and FTIR Rafi  The adsorption process of ARS on GO/PAMAM surface was carried out under different experimental conditions including ARS concentration, ionic strength, pH, temperature, shaking rate and contact time. Due to high adsorption capacity of GO/PAMAM for ARS, GO/PAMAM was considered as a superadsorbent for ARS dye, Table 1.
Kinetics and thermodynamics of adsorption of ARS on GO/PAMAM surface were investigated using the ARI-AN and KASRA models respectively that elucidated the process mechanism. Also, high capacity adsorption of GO/ PAMAM in acidic pHs for ARS was used for selective separation of a number of dyes from their mixtures with ARS.

Synthesis of GO/PAMAM
Generation 2 PAMAM (G2 PAMAM) dendrimer (the first generation in this kind of nomenclature is called G-0.5), graphene oxide (GO) and GO/PAMAM were synthesized based on the published procedure. 26

3. Characterization of GO/PAMAM
The nitrogen-based BET specific surface area of GO/ PAMAM was determined by a Pore Size Micromet-rics-tristar 3020 instrument, Figure S1. The obtained BET isotherm for GO/PAMAM was type IV and its BET surface area, maximum pore volume (slit pore geometry), adsorption average pore diameter (by BET) and pore volume were 9.59 m 2 g -1 , 0.0044 cm 3 g -1 , 18.9 nm and 0.045 cm 3 g -1 , respectively. It is noticeable that BET surface and pore volume of GO/PAMAM in this work are several times higher than those of reported. 26 Also, its hysteresis loop was H3 which in this case aggregates of platelike GO/ PAMAM nanoparticles form slit-like pores. 27 Scanning electron micrographs (SEM) of GO/ PAMAM and ARS-adsorbed GO/PAMAM samples at pHs of 0, 2 and 13 were taken using a MIRA3 TESCAN equipment at 15 keV. The SEM photos of pristine GO/PAMAM and its samples obtained under different conditions exhibited similar surface morphology which were aggregations of PAMAM-covered particles with a porous morphology,  Figure S2.
EDS spectrum of the as-synthesized GO/PAMAM was acquired by a MIRA3 TESCAN equipment. According to the results of EDS, the atomic percentages of elements on its surface were similar to the published results 26 which confirmed the formation of GO/PAMAM, Figure  S3. The XRD pattern of as-synthesized GO/PAMAM was recorded by a Rigaku D-max C III, X-ray diffractometer using Ni-filtered Cu-Kα radiation (λ = 1.5406 Å). The smooth intensity with broad peaks at around 2θ of 26.7° showed that the main structure of GO/PAMAM was amorphous, Figure 2(a). The XRD spectrum of GO/ PAMAM was similar to that in the literature. 26 Also, the FTIR spectrum of GO/PAMAM was taken by a Nicolet IR 100 (Thermo Scientific) FTIR spectrophotometer using KBr pellet technique, Figure 3(a). The peak at 1185 cm -1 was assigned to the stretching vibration of C-OH (phenolic) groups of GO/PAMAM. 26,28,29 The spectrum was very similar to that of the previous report for GO/PAMAM. 26 The absence of the peak at 1680-1760 cm -1 confirmed the lack of carboxylic acid group in GO/ PAMAM. Diminishing the C=O band of GO at 1731cm -1 and the appearance bands at 1637 cm -1 (C=O amide I stretching vibration mode), 24,29,30 in the IR spectrum of as-synthesized GO/PAMAM confirmed that the GO/ PAMAM was synthesized.

1. Thermodynamics of Adsorption of ARS on GO/PAMAM
The ARS molecule has two pK a values, pK a1 = 5.5 and pK a2 = 11.5. 39,40 This compound, under experimental conditions, at pH ≤ 3 and pHs of 11 and 13 was in the forms of ARS -(yellow), ARS 2-(red) and ARS 3-(violet) molecules respectively, Figures 6(a) and 6(b). The chemical structure used for ARSwas confirmed before. 41 Thus, there were one or two ARS forms in solutions within the applied pH range in this work. These ARS forms are potential species in the solution, interacting with GO/PAMAM surface.
As published, 29,42 protonated primary amine (-NH + 3 ), phenolic -OH and protonated tertiary amine (-NHR + 2 ) groups of GO/PAMAM started to be deprotonated at pHs higher than 9.54 and 8.24 and 4, respectively. Thus, in acidic solutions these three functional groups and in alkaline media protonated and deprotonated primary amine groups of GO/PAMAM were its probable candidate groups for the adsorption of ARS molecules. Role of the -CO-NH-(amide) groups of the GO/PAMAM in the adsorption process was studied using FTIR spectra.
The peak of stretching vibration of phenolic C-OH group at 1185.5 cm -1 in the IR spectrum of GO/PAMAM, Figure 3 This red-shifting validated interaction of this group with sulfonate group of ARSmolecules. Also, the peak of amide group in GO/PAMAM at 1637 cm -1 , Figure 3(a), was observed at 1637, 1637 and 1638 cm -1 in spectra of ARS-adsorbed GO/PAMAM at pHs of 0, 2 and 5, Figures 3(c)-3(e), respectively and did not have wavelength shift. Therefore, from FTIR spectra, it was concluded that in acidic solutions -NH + 3 and phenolic -OH groups of GO/PAMAM interacted with sulfonate group of ARSmolecules. In this work, the functional groups of GO/PAMAM that interacted with ARS molecules are called its adsorption sites.
At first, the adsorption of ARS molecules in alkaline media was studied. From its pK a1 and pK a2 value, ARS de- . At pH = 10, ARS 2molecules and at pH = 13, ARS 3molecules interacted first with -NH + 3 (in the first curve) and subsequently with -NH 2 (in the second curve) sites of GO/PAMAM surface, respectively.
The observed decrease in q sssB value in isotherm at pH = 10 compared to isotherm at pH = 13 was resulted from a decrease in the number of -NH + 3 groups of adsorbent surface. An increase in the negative charge of ARS molecule, as solution Ph was increased from 10 to 13, resulted in an increase in the electrostatic repulsion between adsorption sites and ARS molecules which in turn resulted in a decrease in K values of regions I, IIA and IIB from pH = 10 to pH = 13, Table 2.
At pH = 14, ARS 3molecules interacted with -NH 2 sites. As seen from Table 2, adsorption binding constants of ARS 3molecules with -NH 2 site at pH = 14 (region IIA) and -NH 2 site at pH = 13 (region IIB) and the adsorption binding constant of ARS 2molecules with -NH 2 site at pH = 10 (region IIB) were comparable.
In alkaline solutions, due to the low adsorption of ARS molecules on the surface, no CRAC was observed in adsorption isotherms, Table 3.
Then, the adsorption of ARS molecules on the surface of GO/PAMAM was studied in a water solution. The pH of these series of ARS solutions was about 5. Both ARSand ARS 2species appeared in the aqueous solution, in which ARSwas the predominate species. Under the pH environment, Ph and -NH + 3 sites were present on the surface of adsorbent. Given that the polarity of O-H bond is stronger than that of N-H bond, in region I and IIA (the first curve), ARSmolecules interact with Ph sites first and then with -NH + 3 sites, respectively. Before adding adsorbent to the solutions, the color of solutions was brown (a mixture of ARSand ARS 2molecules). At the end of region IIB (the second curve) and CRAC C2-C3 (the CRAC between regions IIB and IIC), the color of the solutions in region IIC (the third curve) became red that confirmed ARSmolecules adsorbed on the surface of GO/PAMAM in regions I, IIA and IIB. With an increase in the initial concentration of solution the color became paler, Figures 8(a) and 8(b). By considering the changes in the color of ARS solutions, it was speculated that in regions I, IIA and IIB, ARSmolecules (in brown color) adsorbed on Ph and then on -NH + 3 sites respectively. This adsorption step is followed by ARS 2molecules (in red color) adsorbed on the remaining -NH + 3 sites in region IIC. As seen from experimental data in Table 2, the   Table 2. Adsorption equilibrium constants obtained from the Henry and Temkin equations based on the ARIAN model and experimental ssc A , q sssA , ssc B , q sssB , c max and q e,max values for adsorption of ARS on GO/PAMAM in water, alkaline and acidic solutions in regions I, IIA, IIB and IIC at 100 rpm and 308-328 K.  maximum adsorption capacity of regions I and IIA is q sscB = 291,1 mg g -1 , which was much bigger than those of reported for ARS 43 and orange IV 44 which interacted with GO. However, our q sscB was smaller than that of the second and third curves, q max,e -q sscB = 345,9 mg g -1 , Table 2.

Solvent
Then, it was concluded that in region IIC, ARS 2molecules interacted with some remaining -NH + 3 sites. As seen from Table 2, at pH = 5 and 318 K, ARSin region IIB and then ARS 2in region IIC interacted with -NH + 3 sites. The K value of the former interaction was higher than the latter one. Given the negative charge of ARS 2was higher than that of ARS -, the K values observed in region IIC were less than those in region II, Figures 8 and 9. The observation of K values was attributed from two aspects: (1) the steric hindrance caused by the adsorbed ARS molecules; (2) increasing repulsion interaction between ARS molecules and negatively charged adsorbent surface.
Alizarin (AZ) was used to determine the functional group responsible for interacting ARS with GO/PAMAM in alkaline media. AZ molecule shares the same structure as ARS except for the missing sulfonate group, Figure 10. The pK a of AZ is 6.77 and both hydroxyl groups of AZ are ionized under an alkaline environment. 45 Experiments showed that AZ adsorption capacity of GO/PAMAM at pH = 12 was 21.6 mg g -1 that was much less than the ARS adsorption capacity of 113.4 mg g -1 under similar conditions. This test verified that the sulfonate group of ARS 2and ARS 3molecules but not their -Ogroups interacted with GO/PAMAM in alkaline media.
Finally, the adsorption of ARS on GO/PAMAM was studied in acidic solutions, pH ≤ 3. Visible spectroscopy technique showed that only ARSmolecules appeared in this range of pHs. As reported before, 42 with the decrease in pH (pH < 4) all the tertiary amines were protonated. Due to the repulsion between primary and tertiary amines of PAMAM dendrimer, its structure was opened, yielding  an ideal interaction between the Ph groups of GO moiety of adsorbent and ARSmolecules. Thus, the adsorption of ARS on the pristine GO/PAMAM obeyed only from the Henry equation (region I) without forming region IIA. Under these conditions more primary and tertiary amine groups exposed to ARSmolecules. Due to binding three alkyl groups to N atom of tertiary amines, the polarity of N-H bond of protonated tertiary amines form is less than that for N-H bond of protonated primary amines -NH + 3 and O-H bond of Ph groups.
As shown in Tables 2, S1-S3 and Figure 7, the adsorption of ARSmolecules at pH≤ 3 occurred in three regions (I, IIB and IIC). However, the adsorption of ARSmolecules at pH = 0 happened in two regions (I and IIB) and their related adsorption isotherms were formed from two curves.
Thus, ARSmolecules adsorbed on Ph, -NH + 3 and -NHR + 2 sites in region I and sections IIB and IIC, respectively. As shown in Tables 2, 3, S1-S3 and Figures 7(a) and 7(b), the maximum experimental adsorption capacity (q e,max ) of the adsorption of ARSmolecules on GO/PAMAM at pHs of 1, 2 and 3 were comparable. This implied that an increase in ionic strength of solutions at pHs from 3 to 1 did not affect the q e,max of process. The adsorption binding constants (K) in regions I and IIB at 318 K increased from pHs 1 to 3. However, K values in region IIC (on the -NH + 3 sites) were approximately constant. Each of the adsorption isotherms at pHs of 1, 2 and 3 consisted of two curves and a CRAC appeared between them, Table 3. The presence of  CRACs was due to the high adsorption capacity of adsorbent for ARSmolecules in the concentration range of the first curve which resulted in steric hindrance of adsorbed ARSmolecules on the adsorbent surface and electrostatic repulsion between adsorbent surface and ARSmolecules.
At pH = 2, thermodynamic parameters of interaction of ARSmolecules with Ph, -NH + 3 and -NHR + 2 sites in regions I, IIB and IIC respectively, were calculated using the adsorption binding constants obtained from the Henry, Temkin and Temkin isotherms respectively, Table 2. ΔH and ΔS values were 63.2 kJ mol -1 and -57.1 J mol -1 K -1 in region I, 75.8 kJ mol -1 and -127.2 J mol -1 K -1 in region IIB and 48.9 kJ mol -1 and 237.8 J mol -1 K -1 in region IIC, respectively.
Afterwards, the Thermodynamic parameters of the adsorption were analyzed. Adsorption in the liquid phase, occurred through two kinds of interactions. One of them is the interaction of adsorbate molecules with surface of adsorbent and the other is due to replacing water molecules attached to the surface by adsorbate molecules.
The former interaction, by immobilization of adsorbate molecules, lowers the entropy of adsorption system. The latter one increases the disorder and thus the entropy of adsorption system which is result of the mobility of detached water molecules. At pH = 2, the negative ΔS values observed in adsorption of ARSmolecules on the surface of GO/PAMAM in region I and sections IIB, showed electrostatic (ion-ion or ion-dipole) interaction occurred between ARSmolecules with Ph and -NH + 3 adsorption sites of GO/PAMAM. The negative ΔH values of these interactions indicated that the adsorption process in these two regions were exothermic.
On the other hand, the large positive ΔS value in region IIC showed after initial electrostatic interaction between ARSmolecules and -NHR + 2 adsorption sites of GO/PAMAM surface, a hydrophobic interaction happened between ARSmolecules and hydrocarbon chains of -NHR + 2 adsorption site that detached a large number of water molecules solvating ARSmolecules and GO/ PAMAM surface. The positive sign of ΔH value of this interaction was due to: (1) required energy to separate more number of water molecules from the GO/PAMAM surface and ARSmolecules and (2) subsequent hydrophobic interactions between ARSmolecules and GO/PAMAM surface. 46,47 It was observed that adsorption capacities of these two curves of isotherms (specially the second curve) decreased from pH = 1 to pH = 0. Adsorption capacities of the second curve decreased about 50% from pH = 1 to pH = 0.3 and further disappeared at pH = 0 which showed gradual decrease in the availability of the second adsorption site with an increase of HCl concentration.
This observation was further investigated using the following tests. The comparison of EDS spectra of ARSadsorbed GO/PAMAM modified by 1 M HCl for 10 hours and pristine GO/PAMAM showed that atomic percentages of nitrogen were comparable, which were 24.75% and 23.13% in the former and latter samples, respectively (without considering atomic percentage of its chlorine), Figures S3(a) and S3(b).
According to the measurements of nitrogen physisorption BET technique for pristine GO/PAMAM, ARS-adsorbed GO/PAMAM at pH = 2 and ARS-adsorbed GO/PAMAM at pH = 0, BET surface area, maximum pore volume for slit pore geometry, adsorption average pore diameter (by BET) and pore volume were 9.59 m 2 g -1 , 0.0044 cm 3 g -1 , 18.9 nm and 0.045 cm 3 g -1 for the first sample, 1.92 m 2 g -1 , 0.001 cm 3 g -1 , 14.2 nm and 0.0069 cm 3 g -1 for the second sample and 1.35 m 2 g -1 , 0.0006 cm 3 g -1 , 10.7 nm and 0.0036 cm 3 g -1 for the third sample respectively, Figures S1(a)-S1(c). The t-plot micropore area for ARS-adsorbed GO/PAMAM at pH = 0 was 0.0075 m 2 g -1 and ARS-adsorbed GO/PAMAM at pH = 2 and pristine adsorbent were lack of micropores. These results showed that as GO/PAMAM at pH = 0 adsorbed less ARS molecules than GO/PAMAM at pH = 2, the BET surface area and slit pore volume of adsorbent at pH = 0 were less than those values at pH = 2 and part of adsorbent pores at pH = 0 changed to micropores.
The comparison of IR spectra of ARS-adsorbed GO/ PAMAM modified by 1 M HCl for 10 hours and pristine GO/PAMAM showed that those IR spectra were very similar, Figures 3(a) and 3(b). Also, the shift of broad peak around 2θ of 26.7° in the XRD spectrum of GO/PAMAM to 26.4° in the XRD spectrum of GO/PAMAM modified at pH = 0 showed a change in the structure of the latter one, Figures 2(a) and 2(b).
This evidence confirmed that the decrease in the adsorption capacities of the first and second curves of adsorption isotherms at pHs of 0.3 and 0 was due to masking the adsorption sites of GO/PAMAM. Indeed, a decrease in pH to less than 1 and an increase in ionic strength of solutions resulted in a shrinkage in PAMAM structure 42 and possible interactions were between -NH + 3 and polyamide groups of PAMAM dendrimer and between functional groups of PAMAM with a number of functional groups on the GO planes of adsorbent like its C = C bonds and compressed its structure. This resulted in a decrease in the number of free Ph and -NH + 3 groups and entrapment of -NHR + 2 groups in internal structure of PAMAM and thus, resulted in a decrease in the availability of Ph, -NH + 3 and -NHR + 2 adsorption sites of GO/PAMAM for ARSmolecules.
Finally, relative adsorption capacities of isotherms for regions I and IIC were given in Table S4. For example, because at pH = 5 the concentration of ARS 2molecules was less than ARSmolecules, the adsorbent's relative adsorption capacity for ARS 2-(region IIC) was 0.15 of q e,max value.

2. Kinetics of Adsorption of ARS on GO/ PAMAM
The adsorption mechanism of ARS on GO/PAMAM was studied by the KASRA model, and intraparticle diffusion and ISO equations using different ARS initial concentrations, temperatures, pHs and shaking rates. At 318 K, using 0.25 mM ARS at pH = 0 and using 0.07 mM ARS at pH = 2, only one adsorption curve was observed in their kinetic diagrams without the presence of TRAK, implying that the adsorption of ARSmolecules occurred on the Ph sites of adsorbent, Figures 11(a) and 11(b) and Table 4. gions (curves) were observed in all kinetic diagrams: ARSmolecules interacted with Ph sites in the first curve, followed by interaction with -NH + 3 and -NHR + 2 sites in the second one. It was observed that at pH = 2, ISO rate constants for the adsorption on the Ph sites (in the first region) increased with an increase of the shaking rate. Also, an increase in pH from 2 to 3 increased the ISO rate constants for the adsorption on Ph, -NH + 3 and -NHR + 2 sites. At pH = 2, adsorption acceleration, initial velocity and k dif for adsorption on Ph, -NH + 3 and -NHR + 2 sites increased with temperature, shaking rate and ARS initial concentration. The k I1 values at 0.7 mM and 100 rpm, that

a) b)
As seen from Tables 5 and 6 and S5-S7, adsorption acceleration, initial velocity, k dif , k I1 and k I2a decreased as pH decreased from 2 to 0. This was resulted from the compression of the adsorbent structure at pH = 0, that blocked interior adsorption sites. Using 0.7 mm ARS at pHs of 2 and 3 at different shaking rates and temperatures two re- implied the interaction of Ph sites with ARSmolecules increased with temperature. The activation energy of the adsorption in region I was 114.5 kJ mol -1 . Therefore the adsorption of ARS on Ph sites was reaction-controlled. The adsorption in region I in region I only spanned about 6-8% of the whole adsorption duration but was accounted Unit of k dif is in mg g -1 min -0.5 . Units of a 1 , a 2 and a 2 are in mg g -1 min -2 . Units of t 1 , t 2 and t 3 are in min and those of q 1 , q 2 and q 3 are in mg g -1 and t 1 = q 1 = 0. Boundary points coordinates of diffusion regions, (t n , q n ), are similar to those of the KASRA model, (t 0n , q 0n ) in Table S5. *At pH = 5, data from left to right belong to the first and second kinetic curves and the third kinetic curve (corresponding to the ARIAN section IIC) starts after second TRAK (90-120 min) and for that k dif and a 4 are 36.3 mg g -1 min -0.5 and -0.006 mg g -1 min -2 , respectively. Adsorption sites and ARS types involved at pHs = 5, 10 and 13 are discussed in text. for about 34-41% of the adsorption capacity of GO/ PAMAM. The adsorption on Ph sites was much faster than those on the -NH + 3 and -NHR + 2 sites. On the other hand, under similar conditions, other rate constants obtained from the ISO equation changed randomly with temperature and thus the adsorption on the -NH + 3 and -NHR + 2 sites was diffusion-controlled. At pH = 2 and 0.07 mM ARS in 0.1 M NaCl, adsorption acceleration, initial velocity and k I1 increased slightly compared to those values in a solution without NaCl. This was due to more availability of ARSmolecules to internal structure of PAMAM dendrimer with the ionic strength of solution. The slight increase in k I1 value with an increase in the ionic strength of solution confirmed that the adsorption interaction in region I occurs between ARSmolecules and uncharged Ph groups of adsorbent.
At pH = 5, 318 K and 1 mM ARS, adsorption acceleration, initial velocity, k dif and k I1 of the adsorption on Ph and -NH + 3 sites (first curve) were less than those at pH = 2, 318 K and 0.7 mM ARS. Because, at pH < 4, all the primary and tertiary amines were protonated and the repul-sion interaction between these amine groups of PAMAM dendrimer, its structure swelled 42 to expose more adsorbent inner surface compared to that at pH = 5, Tables 5  and 6.
At 318 K and 0.07 mM ARS, ARS 2molecules at pH = 10 and ARS 3molecules at pH = 13 were adsorbed on -NH + 3 sites (first curve) and -NH 2 sites (second curve) respectively. As the pH increased from 5 to 13, the electron charge density on the adsorbent surface became higher, that increased the electrostatic repulsion between ARS molecules and adsorbent surface. The repulsive interaction yielded the appearance of TRAK in the kinetic diagrams Table 4.

3 Recycling ARS-adsorbed GO/PAMAM
For recycling ARS-adsorbed GO/PAMAM, several solvents were tested, such as, diethyl ether, methanol, ethanol, benzene, DMF, THF, DMSO, diethylenetriamine and alkaline and acidic aqueous solutions, none of them could extract ARS from the used adsorbent. However, we found ethylenediamine extracted ARS from the ARS-adsorbed GO/PAMAM very fast and effectively. In a series of experiments, 0.006 g samples of pristine GO/PAMAM were added to 30 ml of 10 -3 M ARS solution at pH = 3 at ambient temperature. The ARS-adsorbed adsorbents were separated from supernatant after 10 h followed by washing and drying. Then, about 45 ml of ethylenediamine was added to the about 0.0025 g of the used adsorbent in three steps (15 ml/step). The adsorption capacity of pristine GO/ PAMAM was fully regenerated. The EDS spectrum of the recycled adsorbent showed no trace of sulfur atom. The SEM images of the recycled adsorbent showed that its morphology was similar to that of the pristine GO/ PAMAM, Figures 12(a)-12(d). Finally, ARS was separated from the ethylenediamine solution (in purple) by the distillation process.
At this pH, GO/PAMAM had its maximum adsorption capacity for ARS. Using the adopted initial ARS concentration, ARS was fully removed from the solution. Comparison of the UV-Vis spectra of pure dyes solutions with those of ARS-dye mixtures (in the absence of adsorbent) showed that the strength of dye-ARS interaction followed the trend: JG (by formation of deposit) > MB, AO, PY >Th > MEB, peaks B and D of Figures 14(b) -14(g). The final concentrations of Th, MB, AO and PY in their mixtures with ARS were 95%, 56%, 45% and 56% of their initial concentrations, respectively. As shown in peaks B and D of Figure 14   concentrations were estimated as 76%, 79%, 59%, 80%, 45% and 30% of their initial concentrations, respectively.
An increase in the absorbance intensity when the used dyes were adsorbed in the presence of ARS and GO/PAMAM simultaneously, compared to that in the presence of ARS alone, confirmed that due to the complete adsorption of ARS on adsorbent, most dyes molecules remained in the supernatant. Furthermore, some adopted dyes molecules interacted both with surface adsorption sites and adsorbed ARS molecules, peaks of B, C of Figures 14(b)-14(g). It was observed that in the used initial concentrations of ARS and JG, the ARS-JG deposit formed in their mixture was completely dissolved in the presence of GO/PAMAM and the solution turned blue from colorless which was due to complete adsorption of ARS, Figure S5 and peak D of Figure 14(g).

Conclusions
The adsorption of ARS on the surface of GO/ PAMAM was studied under different temperatures, pHs, ARS initial concentrations, contact times and shaking rates. Adsorption isotherms were analyzed by the ARIAN model and its kinetic curves were investigated by the KAS-RA model and ISO and intraparticle diffusion equations. Analysis of adsorption isotherms and kinetic curves showed that phenolic -OH groups of GO moiety (Ph) and -NH 2 , -NHR + 2 and -NH + 3 groups of PAMAM dendrimer moiety of GO/PAMAM were adsorption sites of the adsorbent. Depending on the solution pH, due to deprotonation, ARS formed three different species as ARS -, ARS 2and ARS 3-. At acidic pHs, ARSmolecules interacted with Ph, -NH + 3 and -NHR + 2 in sequence. The adsorption on the Ph and -NH + 3 sites were exothermic and the adsorption on the -NHR + 2 sites was endothermic, respectively. At pHs of 1, 2 and 3, q e,max of the adsorbent for ARSattained to maximum value and were similar together. At pH = 2 and 328 K, q e,max of GO/PAMAM (as a superadsorbent) for the adsorption of ARSwas 1275.6 mg g -1 . This was attributed to the open GO/PAMAM structure after the protonation of all primary and tertiary amine groups of PAMAM dendrimer. Due to the repulsion interaction between these groups, GO/PAMAM structure became very open and resulted in an increase in adsorption capacity of GO/PAMA for ARS. At pH = 2, binding constant values of ARSmolecules to Ph sites were higher than those of -NH + 3 and -NHR + 2 sites of adsorbent. The interaction of ARSmolecules with Ph sites was faster than that of -NH + 3 and -NHR + 2 sites. The adsorption of ARSon Ph sites was reaction-controlled. The activation energy obtained from its ISO rate constant was 114.5 kJ mol -1 . However, the adsorption of ARSon the -NH + 3 and -NHR + 2 sites was diffusion-controlled. The decrease in the pH of solutions from 1 to 0 resulted in a decrease in q e,max , which could be owe to fewer available Ph, -NH + 3 and -NHR + 2 groups of GO/PAMAM for ARS molecules that was caused by a shrinkage in GO/ PAMAM structure and masking some of functional groups of GO/PAMAM. Also, this resulted in a decrease in ISO rate constants with a decrease in pH from 2 to 0. At pH = 5, ARSinteracted first with Ph sites and then with -NH + 3 sites. In continuation, ARS 2interacted with -NH + 3 adsorption sites.
At pHs of 11 and 13, ARS 3molecules at first interacted with -NH + 3 and then -NH 2 adsorption sites. At pH of 14, ARS 3molecules interacted with -NH 2 adsorption sites only. Different types of ARS molecules in acidic and alkaline media interacted with GO/PAMAM adsorption sites through their sulfonate head.
By using GO/PAMAM, several dyes including MB, AO, Th, PY, MEB and JG were separated from their mixtures with ARS at pH of 2. Our extensive study illustrated that used adsorbent was regenerated completely and effi-ciently using ethylenediamine at ambient temperature.
Furthermore, GO/PAMAM can be used to selectively remove ARS from dye mixtures at pH of 2. The finding could be beneficial to fabricate photochemical and electrochemical sensors.