Article pubs.acs.org/Langmuir
Migration of Silver Nanoparticles from Silver Decorated Graphene Oxide to Other Carbon Nanostructures K. A. Shiral Fernando,*,† Venroy G. Watson,† Xifan Wang,† Nicholas D. McNamara,† Mary C. JoChum,‡ Dylan W. Bair,‡ Barbara A. Miller,† and Christopher E. Bunker‡ †
Energy Technology and Materials Division, University of Dayton Research Institute, Dayton, Ohio 45469, United States Propulsion Directorate, Air Force Research Laboratory, Wright-Patterson Air Force Base, Ohio 45433-7103, United States
‡
S Supporting Information *
ABSTRACT: Decoration of graphene oxide (GO) sheets with Ag nanoparticles has been demonstrated using a simple sonication technique. By changing the ratio between Ag-decorated-GO and GO, a series of Ag-decorated-GO samples with different Ag loadings were synthesized. These Ag-decorated-GO samples were characterized using transmission electron microscopy (TEM), X-ray diffraction (XRD) spectroscopy, thermal gravimetric analysis (TGA), and differential scanning calorimetric (DSC) techniques. TEM analysis showed that Ag nanoparticles were evenly distributed on GO sheets, and the size analysis of the particles using multiple TEM images indicated that Ag nanoparticles have an average size of 6−7 nm. TEM analysis also showed that Ag nanoparticles migrated from Ag-decorated-GO to lateradded GO sheets. In XRD, all the Ag-decorated GO samples showed the characteristic peaks related to GO and face-centered-cubic (fcc) Ag. Thermal analysis showed peaks related to the combustion of graphitic carbon shifted to lower temperatures after GO sheets were decorated with Ag nanoparticles. In addition, further experiments performed using Agdecorated-GO and multiwalled carbon nanotubes (MWNTs) confirmed that Ag nanoparticles migrated from Ag-decorated-GO to later-added carbon nanotubes without a noticeable coalescence of Ag nanoparticles.
■
INTRODUCTION Much recent effort has gone into research of metal-decorated carbon nanostructures because these nanostructures possess very attractive featues.1−6 As a result, a wide range of metaldecorated carbon nanostructures have been prepared that exhibit promising applications in nanotechnology and nanobiotechnology.2,3,7−11 For example, Sun and co-workers demonstrated that metal-decorated carbon nanoparticles might be used to convert carbon dioxide to fuel and split water in order to produce hydrogen in the presence of light.2,8 Silver is widely known for its antimicrobial activity, and therefore, Ag-decorated carbon supports are used in biomedical applications, such as in wound dressing and medical catheters.12,13 Ruiz et al. have reported that GO was highly biocompatible with bacterial and mammalian cells, with no cytotoxicity, and that in fact GO promotes bacterial and mammalian cell growth by enhancing cell attachment and proliferation, while Ag-decorated-GO inhibits the growth of bacterial cells.7 In this paper, we present the preparation and characterization of Ag-decorated-GO samples with different Ag concentrations/loadings using simple sonochemical methods that mix different ratios of Ag-decorated-GO and GO. GO was prepared using a modified Hummers method.14−16 This method involves the chemical oxidation of graphite to produce highly exfoliated graphene sheets that contain a large amount of oxygen functional groups (epoxides, alcohols, ketones, carbonyls, and carboxylic groups).17−23 The oxygen functional groups © XXXX American Chemical Society
on the GO surface have a major role in the nucleation and growth of Ag nanoparticles.25−42 For example, Pasricha et al. reported that oxygen functional groups act as nucleation sites for formation and anchoring of metal nanoparticles.42 In addition, metal (Ag and Au) nanoparticles can interact with GO sheets through physisorption, electrostatic interactions, and charge-transfer interactions.42 Furthermore, this study shows that Ag nanoparticles migrate from Ag-decorated-GO to lateradded GO and produce a new Ag-decorated-GO sample with less Ag content. The microscopy characterization shows that Ag nanoparticles are evenly distributed without any noticeable ripening effect or coalescence. To the best of our knowledge, this is the first report showing migration of metal nanoparticles from one carbon nanostructure to another (inter). However, Lu et al. have reported that, at elevated temperatures (>100 °C), small Ag nanoparticles moved on GO surface to form larger Ag nanoparticles.51 Lu et al. observed that the coalescence of Ag nanoparticles happened in same GO sheet (intra), and they did not report any Ag nanoparticles migration between GO sheets (inter). They speculated that two possible mechanisms, Oswald ripening and particles coalescence, are likely to have been responsible for the ripening process.51 Received: June 20, 2014 Revised: August 21, 2014
A
dx.doi.org/10.1021/la502401n | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 1. TEM images for Ag-GO samples with different Ag loadings.
■
RESULTS AND DISCUSSION GO was prepared by oxidation of natural graphite powder using a modified Hummers method.14−16 Graphite consists of stacked graphene layers with a plane distance of 0.335 nm.17,43 The severe oxidation process in a modified Hummers method separates these stacked graphene layers and introduces many oxygen-containing functional groups, as noted before.17−23 GO powder readily exfoliates into individual or few-layers-thick nanosheets in polar solvents (water, DMF, ethanol, etc.) upon a simple sonication forming a suspension that is stable for a long period of time. In addition, it has been reported that metal ions could be adsorbed on GO nanosheets because these oxygen functional groups provide reactive sites for nucleation and growth of metal nanoparticles.24−42 Both GO and silver acetate are readily soluble in DMF, and so DMF was used as the solvent in our Ag decoration of GO reactions. In addition, Pastoriza-Santos and co-workers have reported that DMF is a mild reducing agent that reduces silver acetate to silver nanoparticles easily at elevated temperatures.44 Furthermore, DMF has a desired vapor pressure and viscosity,
which are important properties to use in a sonochemical reaction. Controlling the Ag Content of GO. A 33 wt % Ag loaded sample was prepared by mixing 50 mg of GO with 25 mg of silver acetate in DMF and sonicated for 20 min. The same experiment was repeated five more times to obtain a sufficient amount of 33 wt % Ag-decorated-GO stock sample. The other Ag-decorated-GO samples with different Ag loadings (20, 15, 10, and 5 wt %) were synthesized by changing the ratio between 33 wt % Ag-decorated-GO and GO. In synthesis the precise ratio between 33 wt % Ag-decorated-GO and GO was sonicated in DMF for an hour in a bath sonicater. The reproducibility of this experiment was performed by replicating above experiments multiple times in the same conditions. These Ag-decorated-GO samples (33−5 wt %) were characterized using TEM, XRD, and simultaneous TGA/DSC analysis. TEM analysis was performed to determine morphology, extent of nanoparticle decoration, and size distribution of Ag nanoparticles. TEM samples were prepared by dropping the dispersions of GO and Ag-decorated-GO samples in B
dx.doi.org/10.1021/la502401n | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
isoproponol onto carbon-coated or plain copper TEM grids and then drying at room temperature. As seen in Figure 1a, GO has a typical flaky structure that is a few layers thick and has a wrinkled surface. Ag-decorated-GO with different Ag loadings (33−5%) are shown in Figure 1b−f. Ag nanoparticles appear as dark dots and are evenly decorated on GO sheets in each sample. The population of Ag nanoparticles on GO surface decreases from 33 wt % sample to 5 wt % sample, as expected. For Ag-decorated-GO samples with different Ag loadings, we did not observe in TEM the areas where Ag nanoparticles are agglomerated or larger undecorated areas of GO. The size analysis of Ag nanoparticles on GO surface was performed using multiple TEM images of each Ag-decorated-GO. The size analysis of Ag nanoparticles using TEM showed an average size of 6−7 nm, with a size distribution of 1−22 nm, as shown in Figure 2a−e.
Figure 3. XRD analysis of the GO and Ag-decorated-GO samples.
addition to GO peak and fcc Ag peaks, a broad peak appeared at a 2θ value of 24°, which is related to reduced graphene oxide (r-GO). Qui et al. reported a similar observation of a new broad peak appearing at 2θ value of ∼24° after GO was decorated with Pt nanoparticles, and later they identified it as a r-GO peak.28 The observation of the peak at 2θ value of ∼24° indicates GO is partially reduced during the sonochemical reaction. The intensity ratio between XRD peak of GO and Ag peaks increases when Ag percentage decreases, as expected. Therefore, 33 wt % Ag-decorated-GO sample has the lowest intensity and 5 wt % Ag-decorated-GO sample has highest intensity of GO peak. Simultaneously, 33 wt % Ag-decoratedGO sample has the highest intensity and 5 wt % Ag-decoratedGO sample has the lowest intensity of fcc Ag peak. In addition, TGA and DSC analyses were performed using a simultaneous TGA/DSC instrument at a temperature range from 25 to 800 °C to obtain the Ag loading and thermal behavior of Ag-decorated-GO samples, as shown in Figure 4. It has been reported that Ag nanoparticles are stable in air up to ∼1000 °C, and therefore, in thermal analysis, Ag nanoparticles are not oxidized.48,49 In DSC analysis, GO and Ag-decoratedGO samples showed two exothermic peaks, with the first peak centered at ∼222 °C and the second peak centered at 578 °C. The first peak is related to the removal of oxygen functionality from GO surface, and the second peak is related to the combustion of graphene sheets.17,50 The intensity of the first peak at ∼222 °C depends on the Ag content of sample. A 33 wt % Ag-decorated-GO sample has the lowest intensity for the first peak because it has the highest percentage of Ag loading, while a 5 wt % sample has the highest intensity for first peak because it has the lowest percentage of Ag loading. As noted before, the oxygen functional groups on the surface of GO act as nucleation sites for formation and anchoring of the metal nanoparticles. Therefore, the highest Ag loaded sample has the lowest amount of free oxygen functional groups and the lowest Ag loaded sample has the highest amount free oxygen functional groups. As a result, the 33 wt % Ag-decorated-GO
Figure 2. Size distributions of Ag nanoparticles on GO for 33 (a), 20 (b), 15 (c), 10 (d), and 5 wt % (e).
XRD analysis was performed to determine the crystallinity of GO and Ag-decorated-GO samples, as shown in Figure 3. In XRD analysis, GO exhibits a carbon (002) diffraction peak at 2θ value in the range of 10°−12°, which is considerably lower than the diffraction peak of graphite at 2θ value of ∼26°.45−47 GO shows a lower 2θ value in XRD because the vigorous oxidation process during GO synthesis introduces oxygen functional groups and water molecules between graphene layers; therefore, GO has a higher interlayer distance. The interlayer distance of graphene sheets of GO is ∼0.709−0.804 nm, and the interlayer distance of graphene sheets of graphite is ∼0.335 nm.17,43,45 In XRD, all the Ag-decorated-GO samples show the characteristic peaks related to GO and face-centeredcubic (fcc) Ag, which are well matched to ICDD card data. In C
dx.doi.org/10.1021/la502401n | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 4. DSC (a) and TGA (b) traces for GO and Ag-decorated-GO samples with different Ag loadings.
sample showed the lowest intensity for the first peak at ∼222 °C and the 5 wt % Ag-decorated-GO showed the highest intensity for the same peak. This DSC results provide direct evidence that nucleation and the growth of Ag nanoparticles occurred in oxygen functional groups of GO surface. The second peak at 578 °C of GO, which is related to the combustion of carbon of graphene sheets, is shifted to lower temperature for all the Ag-decorated-GO samples. A 33 wt % Ag-decorated-GO has the highest shift, while 5 wt % Agdecorated-GO has the lowest shift of the carbon combustion peak, as shown in Table 1. The exact reason behind this carbon
and increase the interlayer distance further. Subsequently, Agdecorated-GO samples have a larger surface area than regular GO to absorb heat from heating source. Therefore, the combustion of Ag-decorated-GO samples occurs at a lower temperature than regular GO. In TGA, weight loss between 25 and 100 °C is attributed to the removal of water between graphene sheets of GO. Then, there are two regions of noticeable weight loss: from 150 to 250 °C and 550 to 600 °C. The former is attributed to the removal of oxygen functional groups from GO surface while the latter is related to the combustion of graphitic carbon sheets of GO. The two weight loss regions in TGA correlate well with two exothermic peaks observed in DSC. As noted before, Ag nanoparticles are stable in air up to ∼1000 °C without making their oxide. As GO is completely burned out before ∼600 °C, TGA analysis could be utilized to determine the Ag content of Ag-decorated-GO samples (Table 1). Ag loading of Ag-decorated-GO samples that were obtained from TGA showed a good agreement with calculated values. Migration of Ag Nanoparticles. In the synthesis of Agdecorated-GO samples with different Ag loadings, a 33 wt % Ag-decorated-GO sample was used as a starting material, and it was mixed with GO to obtain desired Ag loading. In order to synthesize new Ag-decorated-GO samples with different Ag loading, Ag nanoparticles should leave the 33 wt % Agdecorated-GO sample and migrate to later-added GO. TGA, DSC, and XRD analyses show the presence of Ag in newly synthesized Ag-decorated-GO samples. Nevertheless, these characterization techniques do not provide any information about the migration of Ag nanoparticles from 33 wt % Agdecorated-GO to later-added GO. Therefore, these techniques cannot be utilized to characterize the migration and distribution of Ag nanoparticles. Microscopic techniques, especially TEM, are useful to characterize the distribution of Ag nanoparticles on GO sheets. As shown in Figure 1, all the newly prepared Agdecorated-GO samples show that Ag nanoparticles are evenly distributed on GO surface. The content of Ag nanoparticles on
Table 1. Actual Ag Content, Average Nanoparticle Size from TEM and the Position of Carbon Combustion Peak from DSC for Ag-Decorated-GO Samples Synthesized by Changing the Ratio between 33 wt % Ag-Decorated-GO and GO sample
actual Ag loading from TGA (wt %)
av size from TEM (nm)
position of carbon combustion peak
GO 33 wt % Ag-GO 20 wt % Ag-GO 15 wt % Ag-GO 10 wt % Ag-GO 5 wt % Ag-GO
32 15 12 8 5
6 7 7 6 7
578 418 457 461 478 499
combustion peak shift to a lower temperature is not yet understood. However, we speculate this could likely be due to three effects. First, Ag behaves as a catalyst and catalyzes the combustion of graphitic carbon. Second, silver is an excellent conductor, absorbing heat and passing to graphene sheets efficiently, and thus the combustion of GO occurs at lower temperature in the presence of Ag nanoparticles on GO sheets. Third, it has already been published that metal nanoparticles intercalate between graphene sheets during the metal decoration of GO,28 so metal nanoparticles behave as spacers D
dx.doi.org/10.1021/la502401n | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
GO surface of each sample depends on ratio between 33 wt % Ag-decorated-GO and GO used during synthesis. A 5 wt % Agdecorated-GO sample has the lowest content of Ag nanoparticles, while 33 wt % Ag-decorated-GO has the highest content of Ag nanoparticles. Even though TEM analysis confirms that Ag nanoparticles are evenly distributed in newly synthesized Ag-decorated-GO samples with less Ag loading, we need further investigation to confirm the migration of Ag nanoparticles from 33 wt % Agdecorated-GO to later-added GO because later-added GO has the similar morphology to GO with Ag nanoparticles. Therefore, a couple of new experiments were designed to confirm the migration of Ag nanoparticles from Ag-decoratedGO to later-added GO. The first experiment was designed to determine whether the migration of Ag nanoparticles from Agdecorated-GO to GO is continuous. In this experiment, 10 wt % Ag-decorated-GO sample was synthesized from 20 wt % Agdecorated-GO sample instead from 33 wt % Ag-decorated-GO sample, as shown in Figure 5. Analysis of this sample confirmed that successive migration of Ag nanoparticles occurred and that 10 wt % Ag-decorated-GO can be synthesized from 20 wt % Ag-decorated-GO sample. As noted before, size analysis was performed using multiple TEM images of 10 wt % Agdecorated-GO sample, and it indicated the average size of Ag nanoparticles is ∼10 nm with a size distribution of 1−20 nm. In addition to TEM analysis, this sample was analyzed by XRD and simultaneous TGA/DSC. These characterizations showed that there were no meaningful differences between the 10 wt % Ag-decorated-GO samples synthesized from 20 or 33 wt % Agdecorated-GO. Therefore, this experiment has proved the successive migration of Ag nanoparticles from Ag-decoratedGO to later-added GO. The second experiment was designed by adding MWNTs to 33 wt % Ag-decorated-GO sample instead of GO because in TEM the tubular structure of MWNTs can be easily differentiated from GO sheets. A 33 wt % Ag-decorated-GO sample was mixed with MWNTs in DMF and sonicated similar to previous Ag-decorated-GO synthesis experiments. TEM analysis of product (Ag-decorated-GO-MWNTs) indicated both GO and MWNTs are evenly decorated with Ag nanoparticles, as shown in Figure 6. Size distribution analysis was performed to evaluate the size distribution of Ag nanoparticles on MWNTs and GO. Analysis of particles revealed the average size of Ag nanoparticles on both GO and MWNTs surfaces are ∼6−7 nm, with a size distribution of 1− 24 nm. These numbers are well matched with an average size and a size distribution of Ag nanoparticles obtained for our previous Ag-decorated-GO samples. Therefore, this experiment confirms that Ag nanoparticles are migrating from Agdecorated-GO to later-added MWNTs without any noticeable ripening or damage of particles. Similar to this, Ag nanoparticles should migrate from 33 wt % Ag-decorated-GO to later-added GO in the synthesis of 5−20 wt % Ag-decorated-GO samples. To the best of our knowledge, no papers have been published related to the migration of metal nanoparticles from one carbon substrates to another (inter). However, as mentioned before, Lu et al. reported that, in elevated temperatures, Ag nanoparticles, especially smaller ones, moved on same GO sheet (intra) and agglomerated to form larger nanoparticles.51 In addition, there are other reports of the migration, coalescence, and agglomeration of small metal nanoparticles to form larger metal nanoparticles in a same GO sheet.51−55 The exact mechanism of the migration of Ag nanoparticles
Figure 5. TEM analysis to prove the migration of Ag nanoparticles from Ag-decorated- GO to GO is not limited to one time. The successive migration of Ag nanoparticles occurs between Agdecorated-GO to GO.
from Ag-decorated-GO to later-added GO or MWNTs is not clear at this point. We speculate that sonochemistry activates some of the silver nanoparticles that are attached to GO surface, so silver nanoparticles release to DMF solution. Lateradded GO and MWNTs also have reactive sites, including defect sites and oxygen functional groups. Therefore, Ag nanoparticles in DMF solution can attach to the surface of later-added GO or MWNTs. We should note here that the simple mixing of 33 wt % Ag-decorated-GO with GO or MWNTs in the absence of aforementioned experimental conditions (in DMF and sonication) did not show any silver presence on later-added GO or MWNTs. Therefore, sonication and a solvent (DMF) are essential for migration of Ag nanoparticles from one carbon substrate to another. In the synthesis of Ag-decorated-GO samples (20−5 wt %), the 33 wt % Ag-decorated-GO sample was mixed with GO and sonicated E
dx.doi.org/10.1021/la502401n | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 6. TEM analysis to show the migration of Ag nanoparticles from Ag-decorated-GO to MWNTs: acid-treated MWNTs (a), 33 wt % Agdecorated-GO (b), and 20 wt % Ag equivalent Ag-decorated-GO-MWNTs (c).
in a round-bottom flask. In this reaction both Ag-decorated-GO and later-added GO are in close proximity in DMF solvent. Sonication provides the required kinetic energy to mix and collide Ag-decorated-GO with later-added GO. Therefore, Ag nanoparticles might also migrate from Ag-decorated-GO to later-added GO because of these rapid collisions. It has been reported in the literature that synthesis of Ag and other metal nanoparticles using passivation agents to prevent agglomerations and then attaching these passivated-metal nanoparticles to GO and other carbon nanostructures via covalent bonds or weak intermolecular interactions such as π−π stacking, hydrophobic, and electrostatic interactions.3,56 The nature of chemical interaction between GO and Ag nanoparticles is completely different in these samples compared to our Ag-decorated-GO. Our Ag nanoparticles do not have a passivation agent, and therefore, Ag nanoparticles are directly attached to the surface of GO or any other carbon nanostructures. This migration of metal nanoparticles between nanostructures is significant because this method might be utilized to introduce metal nanoparticles to the substrates that are inherently difficult to functionalize with metal nanoparticles. GO, carbon nanoparticles, and purified carbon nanotubes are easy to functionalize with metal nanoparticles because they have oxygen functional groups on the surface. As noted before, these oxygen functional groups provide platforms for nucleation and growth of metal nanoparticles. Metal nanoparticles do not agglomerate because the nucleation and growth of metal nanoparticles happens in distinct places on these
nanostructures. However, the materials and substrates that do not have defects and functional groups cannot be decorated with metal nanoparticles easily because of the absence of nucleation centers. Therefore, migration of metal nanoparticles might be useful in decorating the substrates and materials with metal nanoparticles that do not possess nucleation centers.
■
CONCLUSIONS GO was successfully decorated with Ag nanoparticles in DMF using sonochemistry. The loading of Ag nanoparticles on GO surface was controlled by mixing 33% Ag-decorated-GO with known amounts of GO. TEM analyses revealed that GO surfaces are evenly decorated with Ag nanoparticles in all the Ag-decorated-GO samples prepared in this study. Size analysis using TEM indicated an average size of 6−7 nm, with a size distribution of 1−22 nm. Simultaneous TGA and DSC analysis of GO and Ag-decorated-GO samples with different Ag loadings indicated that the carbon combustion peak of GO at ∼578 °C shifted to lower temperature after Ag nanoparticles were added to GO surfaces. Further experiments performed using MWNTs confirmed that Ag nanoparticles migrated from Ag-decorated-GO to later-added MWNTs.
■
EXPERIMENTAL SECTION
Materials. Silver acetate (99.99%), ammonium persulfate, phosphorus pentoxide, potassium permanganate, multiwall carbon nanotubes (MWNTs), sulfuric acid, hydrochloric acid, 13.3 M nitric acid (>69%), and N,N-dimethylformamide (DMF) (99.6%) were
F
dx.doi.org/10.1021/la502401n | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Synthesis of Other Ag-Decorated-GO with 15, 10, and 5 wt % Ag Loading. These Ag-decorated-GO samples were synthesized in the same way as described in 20 wt % Ag-decorated-GO synthesis procedure. In these reactions, the weight ratio between GO and Agdecorated-GO were changed to obtain the new Ag-decorated-GO samples with desired Ag loadings. Successive Dilution of 33 wt % Ag-Decorated-GO To Obtain 20 and 10 wt % Ag-Decorated-GO. First, the 20 wt % Agdecorated-GO sample was synthesized in the same way as described earlier. In the preparation of 10 wt % Ag-decorated-GO sample, 20 wt % Ag-decorated-GO (25 mg) was mixed with GO in DMF (15 mL) and sonicated for 1 h. The recovery and cleaning procedures were performed in the same way as stated in previous synthesis procedures. Synthesis of 20 wt % Ag-GO-MWNTs. In a round-bottom flask, 9.13 mg of MWNT and 15.38 mg of 33 wt % Ag-decorated-GO were mixed with 15 mL of DMF and sonicated in a bath sonicator for 1 h. Cleaning and recovery of the sample were performed in the same way as described in the previous synthesis procedures. Measurements. Transmission electron microscopy (TEM) images were obtained using a Hitachi H-7600 operated at 100−120 kV. Samples for TEM were prepared in butanol and drip spotted on copper grids coated with carbon. X-ray diffraction (XRD) studies were performed using a Bruker D8-Advance equipped with a Cu Kα source and a Sol-X detector. Observed XRD patterns were identified with the ICDDS crystallographic database. The mass and energetic behavior of the samples as a function of temperature were determined using thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) on a TA Q600. The solid sample was heated in an opened alumina pan from room temperature to 800 °C at 10 °C/min with air flow rate kept at 100 mL/min. A PerkinElmer 400F equipped with a continuous 785 nm laser was used to record the Raman spectrum over a wavenumber ranging from 100 to 3000 cm−1 by placing the solid sample on a glass slide in the sample holder for analysis.
purchased from Sigma-Aldrich. Graphite powder was obtained from Asbury Carbon (grade 3805). Dialysis tubes (MWCO ∼ 6000−8000) and reagent ethanol (90%) were also purchased from Fisher Scientific. All the solvents were used as received without any further treatment or purification. Synthesis of Graphene Oxide. Graphene oxide was synthesized by using modified Hummers method. H2SO4 (10 mL) was heated at 80 °C in a 500 mL round-bottom flask, and then (NH4)2S2O8 (0.9 g) and P2O5 (0.9 g) were added. The mixture was stirred well until all the reagents were dissolved well and make a homogeneous solution. Graphite powder was introduced to the flask, and the mixture was allowed to heat for 4.5 h at 80 °C and then cooled down to room temperature. The reaction mixture was diluted with 250 mL of deionized water and kept for 12 h. The mixture was filtered, and a solid residue in filter paper was washed repeatedly with water and dried in an oven for overnight. The solid residue was added to H2SO4 (40 mL) in a 500 mL round-bottom flask and cooled in an ice bath. KMnO4 (5 g over 40 min) was added to the mixture while the temperature was kept
