Selective recovery of platinum from spent autocatalyst solution by thiourea modified magnetic biocarbons

The precious platinum group metals distributed in urban industrial products should be recycled because of their rapid decline in the contents through excessive mining. In this work, thiourea modified magnetic biocarbons are prepared via an energy-efficient microwave-assisted activation and assessed as potential adsorbents to recover platinum ions (i.e., Pt(IV)) from dilute waste solution. The physicochemical properties of prepared biocarbons are characterized by a series of spectroscopic and analytic instruments. The adsorption performance of biocarbons is carried out by using batch tests. Consequently, the maximum adsorption capacity of Pt(IV) observed for adsorbents is ca. 42.8 mg g−1 at pH = 2 and 328 K. Both adsorption kinetics and isotherm data of Pt(IV) on the adsorbents are fitted better with non-linear pseudo second-order model and Freundlich isotherm, respectively. Moreover, the thermodynamic parameters suggest that the Pt(IV) adsorption is endothermic and spontaneous. Most importantly, the adsorbents exhibit high selectivity toward Pt(IV) adsorption and preserve ca. 96.9% of adsorption capacity after six cyclic runs. After adsorption, the regeneration of the prepared adsorbents can be effectively attained by using 1 M thiourea/2% HCl mixed solution as an eluent. Combined the data from Fourier transform infrared and X-ray photoelectron spectroscopies, the mechanisms for Pt(IV) adsorption are governed by Pt–S bond between Pt(IV) and thiourea as well as the electrostatic attraction between anionic PtCl62− and cationic functional groups of adsorbents. The superior Pt(IV) recovery and sustainable features allow the thiourea modified magnetic biocarbon as a potential adsorbent to recycle noble metals from spent autocatalyst solution.

www.nature.com/scientificreports/ where C 0 and C e represent the initial and equilibrium metal concentrations (mg L −1 ), V denotes the solution volume (L) and M signifies the initial mass (g) of adsorbents. The spent biocarbons were regenerated by using different mixtures of ethylenediaminetetraacetic acid (EDTA)/HCl and thiourea/HCl solution to desorb metal ions. The selectivity adsorption of Pt (IV) onto prepared biocarbons was investigated in the presence of the simulated metal ions (e.g., Pt(IV), Ni(II), Cu(II), Zn(II) and Pb(II)) of spent catalysts from catalytic converter). The reusability of the prepared biocarbons for adsorption and desorption of metal ions was carried out. The regenerated adsorbents were washed with DI water for several times and then dried at 70 °C overnight before they were used for next batch adsorption.

Results and discussion
Physicochemical properties of prepared biocarbons. As 47,48 . Upon activating SCG-C with microwave-assisted KOH process (i.e., SCG-C-A), the structure of Fe 3 O 4 is almost intact. However, the peak intensity of Fe 3 O 4 observed for N-SCG-C-A is noticeably decreased after microwave-assisted N-doping process. At the same time, the peaks at 2θ = 44.8° and 41.8° are ascribed to the formation of α-Fe and Fe 3 C, respectively. Moreover, the crystalline phases of Fe 3 O 4 and α-Fe still can be found for Tu-N-SCG-C-A. The magnetic crystalline of iron in the Tu-N-SCG-C-A biocarbons can be further verified via TEM images as can be seen in Fig. 2. The wellidentified Fe 3 O 4 (311) and α-Fe (110) lattice fringes with d-spacings of 0.253 and 0.202 nm 49,50 , respectively can be observed. Furthermore, selected area electron diffraction (SAED) pattern confirms the above results, which are in good accordance with aforementioned XRD.
The FTIR spectra were used to understand the functional groups on the biocarbons. As displayed in Fig. 1b, the FTIR spectra of SCG-C and SCG-C-A show the similar patterns, which have broad bands at ca. 3343 cm −1 attributing to the stretching vibration of hydroxyl functional groups (i.e., -OH) 51 and two peaks at ca. 1600 and 1400 cm −1 respectively relating to the vibrations of aromatic rings and the C-C stretching of aromatic groups 52 .   53 . The N-SCG-C-A biocarbons have broad features in the range of 3200-3700 cm −1 , indicating the presence of -OH and -NH groups. In addition, the peak at ca. 1038 cm −1 is attributed to vibration of C-N groups, implying the successful N-doping on the biocarbons 54 . The peaks observed for Tu-N-SCG-C-A biocarbons at ca. 1650, 1530 and 1378 cm −1 are assigned to the C=N, N-C-N and C=S stretching vibrations, respectively. The above FTIR features suggest that the thiourea has been successfully modified on the magnetic biocarbons [55][56][57] . The amounts of N and S elements on the prepared biocarbons are summarized in Table 1. The N contents are decreased upon the carbonization and activation process. After microwave-assisted N-doping, the N amounts are increased. It is worthy to note that the S content in the Tu-N-SCG-C-A is increased to 0.49 wt% which is at least three times that of N-SCG-C-A, again proofing the effective modification thiourea onto N-SCG-C-A biocarbons. The porous structures of prepared biocarbons were studied by N 2 adsorption/desorption isotherms. In Table 1, the BET surface areas and pore volumes of SCG-C and SCG-C-A are remarkably increased as compared to that of SCG. However, the surface area of N-SCG-C-A is reduced, indicating that the N-doping via microwave-assisted treatments may alter the structure. Because the pores of N-SCG-C-A could be occupied via the  www.nature.com/scientificreports/ glutaraldehyde cross-linking reaction between thiourea and SCGs, the surface area (13.8 m 2 g −1 ) and pore volume (0.05 cm 3 g −1 ) of Tu-N-SCG-C-A are decreased compared to those of N-SCG-C-A. The high-resolution XPS N 1 s fitted spectrum of Tu-N-SCG-C-A is shown in Fig. 1c. The peaks at 399.0, 399.7 and 400.3 eV are assigned to nitrogen in the binding structure of =N-, -NH 2 and N-H, respectively which indicates that the possible formation of imine 57 due to the cross-linking reaction via glutaraldehyde between SCG and thiourea (also discuss later). The high-resolution XPS S 2p spectrum of Tu-N-SCG-C-A is shown in Fig. 1d. The S 2p peak is composed of S 2p 3/2 and S 2p 1/2 at 163.6 and 167.9 eV, respectively, suggesting the probable existence of C-S species 40 .

Adsorption performance of Tu-N-SCG-C-A
Effect of dosages and pH. The effect of adsorbent dosages (0.5-3.0 g L −1 ) on the adsorption of Pt(IV) (50 mg L −1 ) was evaluated by batch adsorption tests. Figure 3a, b illustrate that the Pt(IV) ions can be adsorbed onto the biocarbons rapidly in the initial stage and reach the equilbrium with time. Upon loading the dosage to 1.5 g L −1 , the adsorption rate can be increased to 98.3%. However, it takes ca. 90 min to reach the equilbrium for the dosage of 1.5 g L −1 . For the dosage of 2.5 g L −1 , the equilibrium can be achieved within 45 min. The increased dosages can provide more availability of surface active sites on the adsorbents 58 . The above results show that the dosage of 2.5 g L −1 is effective in terms of the maximum adsorption efficiency and rapid rate to equilibrium.
The influence of pH on the adsorption efficiency of Tu-N-SCG-C-A biocarbons between pH 2.0-8.0 is presented in Fig. 3c. It can be found that the higher values of pH, the lower of adsorption capacity of Pt(IV) ions. The highest removal efficiency of 99.4% Pt(IV) can be observed for Tu-N-SCG-C-A at pH = 2.0. However, the adsorption of Pt(IV) decreases to 88.0% at pH value = 8.0. The effect of pH on the adsorption of Pt(IV) could be explained by the point zero charge (pH PZC ), as shown in Fig. 3d. The pH PZC of Tu-N-SCG-C-A biocarbons is ca. 5.04. Whenever pH value is less than 5.04, the surface of adsorbents is charged positively, while charged negatively at pH > 5.04. It was reported that the speciation of Pt(IV) varied with the concentration of Cland pH 59   Kinetics of Pt adsorption. In order to study the kinetics and the possible mechanisms of adsorption process, the experimental data were elucidated by fitting with three kinetic models, namely, pseudo-first-order, pseudo-second-order and intraparticle diffusion models. Tables 2 and 3 list the fitting parameters of pseudofirst-order and pseudo-second-order kinetic models. The correlation coefficient (R 2 ) and normalized standard deviation (SD: ) are used to evaluate the fittness of these models. Compared to the pseudo-first-order model, the higher values of R 2 , the lower values of SD and better fitted between experimental and calculated q e values obtained from both linear and non-linear pseudo-second-order kinetic models indicate that the adsorption process of Pt(IV) onto Tu-N-SCG-C-A is reasonably interpreted on the basis of pseudosecond-order kinetic model. The above result is in good accordance with plots in Fig. 4a, b. The pseudo-second kinetic order model is based on the assumption that the rate-limiting step may be a chemisorption involving valance forces through sharing or exchanging electrons between adsorbent and adsorbate 62 . The theoretical adsorption capacity of adsorbents is calculated to be 20.  Fig. 4b. It also can be confirmed that the Pt(IV) adsorption onto Tu-N-SCG-C-A have a better prediction by using the non-linear pseudo-second kinetic order expression. Based on Weber and Morris's theory 63 , intraparticle diffusion also can be used to investigate the adsorption kinectics. It can be seen from Fig. 4c that the plots of q t versus t 1/2 consist of more than one step of adsorption process with three distinct reigons. The adsorption process of Pt(IV) on adsorbents could be described as three consecutive steps: (I) the bulk and film diffusion of Pt(IV) in solution; (II) internal diffusion on the surface of adsorbent; and (III) chemical interaction at the surface of adsorbents 64,65 . The slopes of three adsorption steps are decreased upon the increase of adsorption time, indicating that the adsorption rates are decreased due to the decrease of adsorption sites. At the same time, it is observed that the intercepts of three steps are increased by following the order: region III > region II > region I. The aforementioned results suggest that the chemical interaction at the surface of adsorbents is the rate-limiting step.
Adsorption isotherms and thermodynamics. In order to investigate the reaction behavior of adsorbents and adsorbates, the adsorption isotherms were employed. In this study, three adsorption isotherms at 298, 308, 318 and 328 K including Langmuir, Freundlich, and Temkin adsorption isotherms models were used to evaluate the adsorption of Tu-N-SCG-C-A. As shown in Fig. 5, the experimental values of q e increase gradually as the initial concentrations of Pt(IV) increase. Furthermore, the equilibrium concentrations gradually increase as the temperatures increase. The fitting results of adsorption isotherm models and their derived parameters are listed in Table 4. With the similar R 2 observed for Langmuir and Freundlich models, the SD values for Pt(IV) adsorption of Freundlich model are smaller than those of Langmuir model. As reported previously 52 , the evaluation of error analysis by using SD values should be better than R 2 values for adopting the best-fitted isotherm model. Hence, the obtained results are indicative that both of the adsorption of Pt(IV) onto Tu-N-SCG-C-A should follow Freundlich adsorption model, suggesting that the multilayer adsorption may be occurred on the Tu-N-SCG-C-A. Meanwhile, the values of n which can be expressed as adsorption intensity between adsorbents and the adsorbates are found to be greater than one, indicating the favorable adsorption of Pt(IV) onto Tu-N-SCG-C-A 66 . Table S1 shows the Pt(IV) adsorption capacities of Tu-N-SCG-C-A and previously reported adsorbents. The adsorption performance is mostly associated with the adsorbent preparation method  Table 3. Parameters of Pseudo-second-order-kinetic model for Pt(IV) adsorption on Tu-N-SCG-C-A.     www.nature.com/scientificreports/ and experimental conditions (e.g., adsorbent dosages, temperature, time, pH and agitation). The performance of fabricated Tu-N-SCG-C-A is analogous to those of earlier reported adsorbents. It is noteworthy that the microwave-assisted activation in the adsorbent preparation may provide time and energy-efficient route for possible applications. The thermodynamic parameters (ΔH°, ΔS° and ΔG°) observed for adsorption of Pt(IV) onto Tu-N-SCG-C-A at four different temperatures (298, 308, 318 and 328 K) were calculated and summarized in Table 5. The values of ΔG° are negative, showing that the adsorption processes of Tu-N-SCG-C-A are spontaneous. In particular, ΔG° becomes more negative when the temperatures increase from 298 to 328 K, indicating that the adsorption process is more favorable at higher temperatures. As confrimed in Fig. 6, the adsorption capacities of Pt(IV) are increased as the temperatures are elevated from 298 to 328 K. As a result, the values of ΔH° are calculated to be 20.27 kJ mol −1 for Pt(IV), revealing that the interaction between Pt(IV) and the Tu-N-SCG-C-A is an endothermal adsorption process. As such, a large amount of heat is consumed to transfer metal ions from aqueous to the solid phase. Also, it is in line with the increase of Pt(IV) adsorption capacity when the temperature is increased. Moreover, the positive value of ΔS° (94.39 J mol −1 K −1 ) is indicative of favorable Pt(IV) adsorption due to the increased randomness at the solid/solution interface. The adsorption of Pt(IV) onto Tu-N-SCG-C-A is more likely attributed to physical-chemical process because the standard enthalpy change is less than 80 kJ mol −167 .
Selectivity, desorption and reusability. In order to evaluate the practical applications, the selective adsorption of noble metals by using Tu-N-SCG-C-A was carried out. As shown in Fig. 7a, the adsorption capacity and recovery efficiency of Pt(IV) observed for the Tu-N-SCG-C-A are the highest among all the metals (i.e., Ni(II), Cu(II), Zn(II) and Pb(II)). With high concentration of chloride, Pt(IV) ions should mostly exist in the forms of [PtCl 6 ] 2− and [PtCl 5 (H 2 O)] − at the pH = 2, which results in the higher affinity toward the protonated functional groups on the Tu-N-SCG-C-A as compared to other metal ions (positively charged). The desorption efficiency of adsorbed Pt(IV) ions was studied by using various concentrations of EDTA/HCl and thiourea/HCl mixture solution. As shown in Fig. 7b, it can be found that the thiourea can effectively desorb Pt(IV) ions from the Tu-N-SCG-C-A compared to EDTA solution. Moreover, the recovery efficiency is decreased in more acidic solution. This result may be elucidated by weakening the electrostatic interactions between the Pt(IV) ions and the Tu-N-SCG-C-A adsorbents 68 . Among these desorption agents, 1 M thiourea/2% HCl mixture is the superior eluent for Pt(IV) recovery. To reveal the reusability of Tu-N-SCG-C-A, six adsorption-desorption tests were carried out via batch experiments by using 1 M thiourea/2% HCl mixture as an eluent. After each batch, the samples were collected by a strong magnet. As shown in Fig. 7c, the recovery efficiency of Tu-N-SCG-C-A is almost intact after the six successive sorption-desorption cycles. The FTIR spectra (see Fig. S1a    www.nature.com/scientificreports/ shifted to 1648 cm −1 , indicating that the N-C-N binding may have higher vibration energy after Pt(IV) adsorption. This finding may be due to the transformation of thioureas to the resonance structures. In addition, XPS measurements were used to reveal the adsorption mechanism. In the full survey spectrum of XPS (see Fig. 9b), the Pt 4f. peak can be observed for the Pt-Tu-N-SCG-C-A, indicating the successful accumulation of Pt onto the adsorbents. Further high-resolution Pt 4f. XPS spectrum of Pt-Tu-N-SCG-C-A is shown in Fig. 9c. It can be seen that the Pt 4f 7/2 spectrum could be deconvoluted into two peaks at 72.5 and 74.3 eV which are ascribed to Pt(IV) and Pt(II) ions, respectively 35,69 . As observed, some of Pt(IV) ions are reduced to Pt(II) after adsorption, which could be attributed to the reduction via C=N and -OH groups on the Pt-Tu-N-SCG-C-A 64 . The S 2p XPS spectrum of Pt-Tu-N-SCG-C-A is deconvoluted and shown in Fig. 9d. Compared to the fitted peaks observed for Tu-N-SCG-C-A (see Fig. 1d), the S 2p 3/2 and S 2p 1/2 peaks (at ca. 164.2 and 169.0 eV, respectively) are shifted to higher binding energies, which may be attributed to the coordination of S and Pt atoms 35 . The aforementioned results confirm the formation pf Pt-S bonding after adsorption. However, the adsorption capacity of Pt(IV) observed for Tu-N-SCG-C-A is ca. 0.22 mmol g −1 , while the sulfur amount of Tu-N-SCG-C-A is only 0.15 mmol g −1 , implying that other adsorption mechanism (instead of Pt-S formation) may occur between Tu-N-SCG-C-A and Pt(IV). In view of anionic (i.e., [PtCl 6 ] 2− ) and cationic (i.e., Tu-N-SCG-C-A) properties in acidic condition, the electrostatic interaction may also be involved in the adsorption process.

Conclusions
In summary, the magnetic biocarbons were prepared by carbonizing SCGs in the atmosphere of CO 2 , activating with microwave-assisted KOH process, N-doping treatments by microwave ammoxidation and modifying thiourea by cross-linking via glutaraldehyde in this work. The adsorption kinetics and isotherms observed for adsorption of Pt(IV) were well described by non-linear pseudo second-order model and Freundlich isotherm, respectively, indicating the occurance of multilayer chemical adsorption on the surface of Tu-N-SCG-C-A. The obtained thermodynamic data reveal that the adsorption of Pt(IV) onto the Tu-N-SCG-C-A can happen spontaneously and endothermically. The Tu-N-SCG-C-A adsorbent is able to selectively uptake Pt(IV) ions from simulated wastewater with mixed metal ions. In addition, Pt(IV) ions can be adsorbed and desorbed