Purification of Carbon-Based Magnetic Nanoparticles (CMNs) Produced by Submerged Arc Discharge in Liquid Ethanol/Ethylenediamine

Surface modification of covalently amine-attached carbon-based magnetic nanoparticles (CMNs) can be produced quite simply in submerged arc discharge using the amine-containing liquid medium ethylenediamine (ED), resulting in CMNs-ED. However, after the arc discharge processing, the resulting nanoparticles possibly contain physically absorbed amine-containing molecules from a liquid medium on the particle surface. To remove the non-covalently bound molecules, a purification process is required. In this study, the purification was conducted using polar and non-polar solvents following the synthesis process. The surface property was initially characterized by a dispersion test in water, showing that CMNs-ED purified by water have better dispersion than CMNs produced in ethanol alone, CMNs-ED before purification, and CMNs-ED after purification by immersion in toluene. Before and after purification, the diffraction pattern showed definitive peaks corresponded to the crystal planes of C(002), Fe3C(220), and Fe3O4(311) at 26.51°, 44.65°, and 35.42°, respectively. The amine functional group on the nanoparticles before purification thought to come from decomposed ethylenediamine assigned by the vibration peaks appeared at a wavenumbers ~3400 cm -1 and 1020-1220 cm -1 , which corresponds to N-H and C-N, respectively. After purification, the vibration peaks of amine groups were still observable, indicating that the amine groups were still covalently attached to the nanoparticles. Magnetic analysis showed that CMNs before and after purification have superparamagnetic properties, with the magnetic saturation value around 10-17 emu/g. The electron microscope images show that the CMNs-ED before purification have a spherical form with a diameter larger than CMNs-ED after purification.


INTRODUCTION
Magnetic nanoparticles are the most interesting and exciting material to be developed this decade. Magnetic nanoparticles have unique chemical and physical properties, such as small size, large surfaceto-volume ratio, easy synthesis and functionalization, and good biocompatibility (Huang, Lu, & Yang, 2019). These properties offer unique opportunities for many applications, such as adsorption of heavy metal ions from industrial wastewater (Almomani, Bhosale, Khraisheh, Kumar, & Almomani, 2020;Liu et al., 2020), serving as a catalyst (Javadi, Zareyee, Monfared, Didehban, & Mirshokraee, 2020;Maleki, Taheri-Ledari, Ghalavand, & Firouzi-Haji, 2020), or in Li-ion batteries (Moradi, Wang, & Botte, 2020), biomedicine (Kiplagat, Martin, Onani, & Meyer, 2020;Suleman & Riaz, 2020), and drug delivery (Azizi, Nosrati, & Danafar, 2020;Zhang et al., 2020). One of the most frequently used magnetic nanoparticles is iron. However, because iron can be easily oxidized, thus further reducing its magnetism, it is necessary to case the iron with another material that can inhibit further oxidation, such as graphite. Graphite has inert characteristics, low toxicity, stable mechanical and thermal properties, and biocompatibility (Du et al., 2014). The other compound containing iron with magnetic properties is the iron oxide of Fe3O4, which has been proven to be biocompatible in bioapplications, especially when combined with carbon material (Fang, Cheng, Zhang, Zhang, & Keidar, 2018;Wang et al., 2018). Some methods that have been used to synthesize iron oxide/carbon (Fe3O4/C)-based materials include the solvothermal method (Ren, Lin, Meng, & Zhang, 2019;Vinodhkumar et al., 2020), the hydrothermal method , and the arc discharge method (Peggiani et al., 2020;Saraswati, Retnosari, Hayati, Amalia, & Hastuti, 2018;Zhao et al., 2019). In general, the arc discharge method can be performed in both a low-pressure chamber and in a liquid medium. The arc discharge method conducted in a liquid medium has several advantages: the relatively short processing time, the simple tools required, and easier purification that yields a high product purity (Sano et al., 2002). Moreover, the submerged arc discharge can easily provide the functionalized product with specified characteristics for the desired application.
One way to achieve functionalization is by adding the functional groups on the particle surface. For its application in biomedicine, functionalization can be carried out by adding amino groups (Gojny, Nastalczyk, Roslaniec, & Schulte, 2003;Saraswati, Ogino, & Nagatsu, 2012). The magnetic nanoparticles produced are generally hydrophobic; however, the hydrophobic characteristic can be changed to hydrophilic by amino groups' attachment on surface particles. Furthermore, the functionalization using amino groups improves dispersion and increases interfacial interaction between nanoparticles and the functional group (Gojny et al., 2003).
In the submerged arc discharge, surface nanoparticles with functional groups can be modified by replacing the liquid medium with protic and hydrophilic functional groups. This technique has been already developed by Saraswati et al. using the addition of ammonia for surface functionalization . Furthermore, they also used additional ethylenediamine for surface modification of the nanoparticle (Rismana, Astuti, Suselo, Anwar, & Saraswati, 2018).
However, the purity of the product produced is also considered important. The by-product commonly formed in submerged arc discharge is amorphous carbon; thus, the amorphous carbon and other impurities need to be removed. Demonstrations of the various purification methods of carbon nanomaterial have frequently been published for CNTs (Hou, Liu, & Cheng, 2008). Several purification methods for carbon nanotubes (CNTs) have been reported (Hou et al., 2008), including by filtration (Bandow et al., 1997) and centrifugation (Yu et al., 2006). The centrifugation method is an efficient and low-cost method for the purification of carbon materials (Deng et al., 2009). Another technique for purification uses an annealing method (Li et al., 2012). The annealing method was also reported to remove impurities with high effectiveness using vacuum and thermal methods (Meyer-Plath et al., 2012).
In this work, we synthesized CMNs nanoparticles with the arc discharge method in a liquid medium of ethylenediamine mixed with ethanol to modify the particle surface with amine groups followed by the purification process using a combined purification technique by dissolving impurities in organic solvents, annealing, and centrifugation. To the best of our knowledge, the analysis of physical and chemical properties of CMNs before and after purification, including the detail of the changing characteristics influenced by purification in each step, have not been reported elsewhere. These modified techniques considerably improved the purity of the product. The effect of purification on the nanoparticle structure was investigated using X-ray diffraction (XRD), Fourier transform infrared (FTIR), scanning electron microscope-energy dispersive X-Ray (SEM-EDX), vibrating sample magnetometer (VSM), and transmission electron microscopy (TEM). The other characteristics of nanoparticles, such as magnetism and hydrophilicity after purification, were also investigated by simple tests with a magnetic bar and a dispersion test in distilled water, respectively.

Synthesis of Ethylenediamine-Modified Carbon Magnetic Nanoparticles (CMNs-ED)
As an anode, the modified graphite electrode was filled with a mixture of Fe3O4, carbon powder, and fructose binder in a 1:3:2 ratio (w/w/w). The modified electrode was heated at 250 °C under vacuum conditions for six hours. CMNs were prepared with the arc discharge method using 300 mL ethanol 50%. Meanwhile, for the CMNs-ED preparation, ethanol with ethylenediamine addition (50%:50%) in 300 mL total volume was used as a liquid medium of the arc discharge. The process was carried out using 20 V and 10 A of current. Each arc process was continued for about 2 minutes. After arc discharge terminated, the CMNs-ED were collected from the liquid medium by centrifugation and decantation. The heating treatment at 150 °C was then applied to get dried CMNs-ED as a product. Purification of Ethylenediamine-Modified Carbon Magnetic Nanoparticles (CMNs-ED) We compared several purification techniques as follows. The first purification technique was carried out by washing the collected nanoparticles using distilled water several times. The water and the nanoparticles were separated by centrifugation. After the supernatant was discharged, the sediment, or product, was dried at 150 °C. CMNs dried in this stage were named CMNs-ED purified by water. The next purification technique was the continuation of the previous stage. CMNs-ED purified by water were annealed by utilizing a thermal vacuum annealed in horizontal tubular furnaces at a temperature of 250 °C and a heating rate of 5 °C/min for six hours. CMNs dried in this stage were named CMNs-ED purified by annealing. A further purification step was conducted by immersing CMNs-ED annealed in the previous stage in toluene for 24 h, followed by annealing treatment at 250 °C. CMNs dried in this stage were named CMNs-ED purified by immersion in toluene, followed by annealing treatment.

Dispersity and Magnetic Tests
The magnetization and dispersion tests were used to determine the magnetic response and the surface characteristics of CMNs-ED nanoparticles, respectively. The dispersity test was done by mixing 0.01 gram of nanoparticles in distilled water and was then continued by sonication for 15 minutes. Dispersion testing was carried out by observing the mixture after sonication 60s and 1h. Magnetic tests were carried out by using a magnetic bar. The test was analyzed by observing the attractivity response of the magnetized nanoparticles.

Characterization of the Materials
The products of each phase were investigated using X-ray diffraction (XRD) (Philips X'pert Pro; Cu-Κα source λ = 1.54060 Å, step = 0.017°; range = 5-80°). The structural properties of CMNs-ED were studied using transmission electron microscopy (TEM) (FEI Tecnai G2 20 S-Twin; 200kV voltage acceleration class; sub angstrom resolution of 0.24 nm (point) and 0.188 nm (line); magnification range of ×30.0k). The CMNs-ED particle size distribution before and after purification was estimated based on the diameter data of nanoparticles measured by the ImageJ application. The infrared spectra measured by Fourier transform infrared (FTIR) (Shimadzu IR Prestige-21; range = 500 and 4000 cm -1 ; KBr pellet) were used to analyze attached functional groups. Further characterizations were also performed using a scanning electron microscope (SEM) (Hitachi SU-3500; widescreen GUI; Ultra Variable-Pressure Detector) and vibrating sample magnetometer (VSM) (VSM250; T= 298-773 K; external magnetic field [H = 100 Oe-21 kOe]) to analyze the morphological structure and the magnetic property of the nanoparticles, respectively.

RESULTS AND DISCUSSION
The differences of the surface characteristics of the CMNs before and after purification, including with and without ED addition, were easily recognized by dispersing CMNs in distilled water, as shown in Figure  1. A through E in Figure 1 respectively correspond to CMNs, CMNs-ED before purification, CMNs-ED purified by water, CMNs-ED purified by annealing, and CMNs-ED purified by immersion in toluene followed by annealing treatment. The results of the dispersion test were carried out under four conditions, namely before sonication (A1-E1), just after sonication (A2-E2), after being allowed to settle for 60s (A3-E3), and 1h (A4-E4). Before sonication, the unmodified nanoparticles CMNs had a hydrophobic character (A1). The effect of functionalization using the amine group caused CMNs, which were initially hydrophobic, to become hydrophilic. The purification also affected the color of the dispersion liquid produced. Before purification, the dispersion liquid was yellowish, which is thought to be due to excess ethylenediamine in the arc discharge product (B1). On the other hand, the purification using distilled water removed the remained ethylenediamine, producing a clear dispersion liquid (C1-E1).
Figures 1 (A2-E2) show the CMNs dispersion just after sonication, which indicates that sonication treatment can be used to improve the dispersion of the nanoparticles. The nanoparticle dispersion test's stability was confirmed 60 s after sonification whether the nanoparticles were settling or not, as shown in Figures 1 (A3-E3). In Figure 1 (A3), even though the sonication treatment had been performed, the CMNs nanoparticles remained in a non-dispersible form in water, shown as floating particles above the water due to the absence of the hydrophilic functional groups. On the other hand, as shown in Figures 1 (B3-E3), the nanoparticles have good dispersion stability after 60s.
Figure 1(A4) shows that after one hour of settling, CMNs return to float on the water as in the initial dispersion state. In the Figures 1 (B4 and D4), the nanoparticles also began to fall and settle. Interestingly, the nanoparticles shown in Figure 1 (C4) have the best dispersion stability due to the influence of the successful surface modification with amine groups. Figure 1(E4) shows that the dispersion of CMNs-ED in water is worse after purification treatment with toluene. The toluene may physically be absorbed on the CMNs-ED nanoparticles, resulting in the hydrophobic characteristics on the nanoparticle surface triggering the worst dispersity, which accords with the work reported by Tasca et al. (Tasca, Ghajeri, & Fletcher, 2017).
The functional groups present in CMNs-ED before purification, CMNs-ED purified by water, CMNs-ED purified by annealing, and CMNs-ED purified by immersion in toluene followed by annealing treatment were further analyzed by FTIR spectroscopy, as shown in Figure 2. CMNs produced without ED added were not further analyzed using FTIR because the surface characteristics could already be significantly seen from the dispersion properties in Figure 1. For CMNs produced with ED added, the characterization using FTIR was analyzed to study the effect of the purifications and the amine groups, which possibly attached to the nanoparticles' surface. The FTIR results featured in Figure 2 Table 1. FTIR analysis was used to compare the effect of purifications on the functional groups of nanoparticles. Free O−H (A) was observed at wavenumber 3600-3700 cm -1 (Saraswati, Retnosari, Hayati, Amalia, & Hastuti, 2018). The FTIR spectra after purification are shown in Figure 2 (b-d).
The absorption of O-H stretching decreases, which indicates the decreasing H2O vavor content in nanoparticles due to the purification effect, especially from the annealing method.   Table 1. The absorption of the N−H bond was observed at a wavenumber ~3400 cm -1 (B) (Ajmal et al., 2015;Avedian et al., 2018). Before purification, the CMNs-ED show a peak at 3437 cm -1 (B) with strong intensity, as shown in Figure 2(a), which is thought to correspond to the excess of ethylenediamine. The purification might decrease the intensity of the N-H stretch bend. This phenomenon suggests that the purification decreases the physical bonding between the N-H in ethylenediamine and the C atom in the carbon nanomaterials' surface. The vibrations of the N−H bond were observed at wavenumber ~1645 cm -1 , ~1500 cm -1 , and 775-850 cm -1 (Ramanathan et al., 2005;Stuart, 2004). The peak observed at wavenumber ~1645 cm -1 (F) is the primary amine of NH2 groups (Ramanathan et al., 2005), which is thought to indicate the absorption of primary amine NH2 attached to the carbon. The purification effect triggered the primary amine decrease on the surface of the nanoparticles. The absorption at wavenumbers around 1020-1220 cm -1 (I) was detected as C-N bond stretching vibrations (Ramanathan et al., 2005;Stuart, 2004). A band at 1025 cm -1 (Figure 2a) (I) indicates the presence of C−N, which suggests that the functionalization of nanoparticles with ethylenediamine resulting in the covalent attachment of amine groups to the C atom was achieved. After purification (see Figures 2(b-d)), the C−N bond intensity was stronger than before purification. An increase in the peak intensity generally means an increase in the functional group's number. The absorption band of C−N after purification created a spectral shift that occurred in 1086 cm -1 , 1076 cm -1 , and 1075 cm -1 , as shown in Figures 2(b-d), respectively. The peak at 540-590 cm -1 (M) is attributed to the stretching vibrations of the Fe-O (Fe 3+ bond) of Fe3O4 in octahedral sites (Baye, Abebe, Appiah-Ntiamoah, & Kim, 2019;Li, Wu, Li, Tang, & Chen, 2018). Before purification, the absorption of Fe−O had a strong intensity; however, after purification, the intensity of Fe−O was decreased. The decrease of Fe-O intensity is further analyzed in the discussion of diffraction data.
To study the purification effect on the crystalline phase, CMNs-ED nanoparticles before and after purification were analyzed by using XRD equipment. The diffraction patterns for CMNs-ED before purification, CMNs-ED purified by water, and CMNs-ED purified by immersion in toluene followed by annealing method are shown in Figure 3. Due to the similar surface characteristics of CMNs-ED purified in the second stage with CMNs-ED purified by annealing, as shown in Figures 1 and 2, this sample was not further characterized by XRD.
Before purifications, the peaks of Fe3O4 had a strong intensity; however, after purification with annealing, as shown in Figures 3(b) and (c), the intensity of Fe3O4 decreased. The annealing method carried out under vacuum conditions minimized the presence of oxygen. Under limited oxygen conditions, oxygen present in Fe3O4 will decrease, which is in line with EDX data. Therefore, the reaction with other elements than oxygen, e.g., carbon, occurred more, resulting in the greater dominance of iron carbide as indicated by the increasing intensity of the peak at 2θ 54.67° as Fe3C (040). This phenomenon shows that the purification has omitted the Fe3O4. The effect of purification also decreased the intensity of Fe3O4 and Fe2O3, suggesting that it influences the magnetization of the purified products. Moreover, the purification effect of CMNs-ED impacts the reduced intensity of the amorphous carbon diffraction peak at 2θ 21.36° (Shi et al., 2015).  The magnetic property of CMNs-ED before purification, CMNs-ED purified by water, and CMNs-ED purified by immersion in toluene followed by annealing treatment was investigated with a vibrating sample magnetometer (VSM). The magnetization curve of the samples was measured at room temperature, as shown in Figure 4. The magnetization curve shows that the magnetic saturation (Ms) value of CMNs-ED before purification, CMNs-ED purified by water, and CMNs-ED purified by immersion in toluene followed by annealing treatment are 11.38 emu/g, 16.87 emu/g, and 9.95 emu/g, respectively. Saturation magnetization is a primary factor for successful magnetic separation. According to the previous study, the magnetic saturation value >10 emu/g was enough for magnetic separation from the aqueous solution using an external magnetic bar (Dhoble, Lunge, Bhole, & Rayalu, 2011;Ma, Guan, & Liu, 2005). The hysteresis profiles have completely reversible S-shapes. These data confirmed that the superparamagnetic characteristics were maintained in the CMNs-ED before and after the purification process.
The magnetization test was also carried out by magnet attraction. The visible magnetic testing results are shown in Figure 5, which presents the CMNs-ED before purification, CMNs-ED purified by water, and CMNs-ED purified by immersion in toluene followed by annealing treatment. The products in powdery form and dispersed in distilled water are shown to be magnetic and attractable to the magnet. Therefore, the purification presented in this study give no effect on the magnetism attractions visually shown in Figure 5. Figures 6(a-d) show the SEM image of CMNs-ED before purification, CMNs-ED purified by water, CMNs-ED purified by annealing, and CMNs-ED purified by immersion in toluene followed by annealing treatment. As shown in Figures 6(a-b), the product's surface consists of grains and layer, suggesting that the Fe3O4 particles are attached to the carbon layers. As shown in Figures 6(c-d), the nanoparticles' surface is covered by granular particles densely distributed on the surface probably as Fe3O4. Figure 5. The magnetism of (a,d) CMNs-ED before purification, (b,e) CMNs-ED purified by water, and (c,f) CMNs-ED purified by immersion in toluene followed by annealing treatment. Interactions NPs with a magnet (upper) and interactions NPs in distilled water by using an external magnetic bar (lower). Figure 6. SEM images of (a) CMNs-ED before purification, (b) CMNs-ED purified by water, (c) CMNs-ED purified by annealing, and (d) CMNs-ED purified by immersion in toluene followed by annealing treatment. Table 2. EDX results of CMNs-ED before purification, CMNs-ED purified by water, CMNs-ED purified by annealing, and CMNs-ED purified by immersion in toluene followed by annealing treatment.

Sample
Weight The elemental analysis of the EDX data of CMNs-ED before purification, CMNs-ED purified by water, CMNs-ED purified by annealing, and CMNs-ED purified by immersion in toluene followed by annealing treatment is shown in Table 2. The SEMinterpreted data was correlated with the elemental analysis of Fe. As shown in Table 2, the weight percentages of Fe atoms in CMNs-ED purified by annealing and CMNs-ED purified by immersing in toluene followed by annealing treatment are 27.48 and 11.06, respectively. This weight percentage of Fe contributed to Fe3O4 and Fe3C. On the other hand, the purifications possibly caused the decreased weight percentage of elements O and N.
The TEM images of CMNs-ED before purification, CMNs-ED purified by water, CMNs-ED purified by annealing, and CMNs-ED purified by immersion in toluene followed by annealing treatment are respectively shown in Figures 7(a-d). The results show that purification affects the nanoparticles' structure and morphology. Figure 7(a) shows that iron compounds which are possibly dominant as Fe3O4 particles (according to XRD analysis) -due to the content of metal elements -represented as a darker area are relatively larger than those of particles in the other TEM images [see Figures 7(b-d)] that are located outside of the carbon layers. Figure 7(b) shows the structure and morphology of CMNs-ED purified by water. As can be seen in Figure  7(b), the nanoparticles CMNs-ED were not agglomerated. After the purification, the large particles are reduced. The smaller iron compounds are left behind trapped in the carbon layer, rendering the nanoparticles hard to lose. Figure 7(c) shows the structure and morphology of CMNs-ED purified by annealing. The results indicated that Figure 7(c) is almost the same as Figure 7(b) in a specific area. Additionally, the carbon layers were observed with less amorphous carbon in some areas. The TEM images of nanoparticles CMNs-ED purified by immersion in toluene followed by annealing treatment (Figure 7d) indicated that the Fe3O4 was wrapped in the carbon layers, even though in some areas they only show the carbon layers. In general, purification with any method performed in this study confirms the characteristics of particles produced as iron compounds -according to XRD data, these compounds are likely to be Fe3O4 and dominant Fe3C -wrapped by carbon layers. The CMNs-ED before purification has a slightly larger size than CMNs-ED after purification. Figure 7. TEM images and histogram of particles size of (a) CMNs-ED before purification, (b) CMNs-ED purified by water, (c) CMNs-ED purified by annealing, and (d) CMNs-ED purified by immersion in toluene followed by annealing treatment.

CONCLUSIONS
Amine-modified CMNs were successfully synthesized by submerged arc discharge in ethylenediamine. Hence the dispersion test of CMNs produced in submerged arc discharge with and without ethylenediamine addition shows different characteristics. Several purification steps performed in this study variously affect dispersity, in which purification using water with sonication is revealed to be the better purification method compared to the others. The XRD diffractogram of CMNs-ED reveals the main characteristics of carbon, Fe3C, and Fe3O4. Additionally, the purification conducted in this study reduces the intensity of Fe3O4 and increases the crystalline peaks of carbide Fe3C. FTIR analysis indicates that the amine group was successfully attached, indicated by C-N peaks observable at wavenumber 1020-1220 cm -1 . The CMNs-ED before purification have a dominant particle size of 10-15 nm. The structure of CMNs-ED after purification was observed as an iron compound wrapped by carbon layers with the dominant particle size less than 10 nm. After purification, the magnetism of nanoparticles remained, as confirmed by powder's attraction to the magnet and by the hysteresis loop analyzed using VSM showing the CMNs magnetization at ⁓10 emu/g. Synthesis, characterization and in vitro evaluation of methotrexate conjugated fluorescent carbon nanoparticles as drug delivery system for human lung cancer targeting. Journal of Photochemistry and Photobiology B: Biology,153,[111][112][113][114][115][116][117][118][119][120]