Observation of magnetic domains in graphene magnetized by controlling temperature, strain and magnetic field

Since the production of ferromagnetic graphene as an extremely important matter in spintronics has made a revolution in future technology, a great deal of efforts has recently been done to reach a simple and cost-effective method. Up to now, controlling the magnetic properties at extremely low temperature have been investigated only by adding and removing atoms in graphene lattice. In this regard, the effect of strain on the magnetic and electronic properties of graphene has been probed. Here, the ferromagnetic properties are what have been created by strain, magnetic field, and temperature along with observation of the parallel magnetic domains in ferromagnetic graphene for the first time as a great achievement. In this way, we have represented the following: First, introducing three novel methods based on temperature, magnetic field, and strain for producing ferromagnetic graphene; Second, obtaining ferromagnetic graphene at room temperature by significant magnetization saturation in mass-scale; Third, probing the electronic systems and vibrational modes by Raman and IR spectroscopy; Fourth, introducing stacking and aggregation as two types of gathering process for graphene sheets; Fifth, comparing the results with leidenfrost effect-based method which the temperature, magnetic fields, and strain are simultaneously applied to graphene flakes (our previous work).

Graphene properties are extremely influenced by the preparation methods 46,47 . After first isolation of graphene by Scotch tape 1 , many methods are proposed to produce graphene, from exfoliation [48][49][50] of graphite to growing graphene on the substrate by nucleation of carbon atoms 51 . On the other hand, most mechanical exfoliation methods 49 such as the Liquid-phase exfoliation 52,53 produce high-quality graphene flakes in mass-scale. Therefore, in this study, the graphene flakes are obtained by the Liquid-phase exfoliation method by using of ethanol solution 54 .

Method
Graphene preparation. Graphene suspensions were prepared by the Liquid-Phase exfoliation method 52,54 .
To follow the instruction, pristine graphite was added to the ethanol solution in which the graphite concentration was 5 mg/ml and the ethanol/water ratio was 20:80. Then, sonication was performed for 30 min at high power and was subsequently centrifuged at 3000 and 4000 rpm (just for LFE) for 10 min, to eliminate larger flakes (Fig. 1a). At these speeds, according to the Beer-Lambert equation with absorbance coefficient ~ 3182 l/g m (660 nm) 54 , the concentration of graphene was calculated to 2.9 and 2.1 mM, respectively. To consideration the role of temperature, pressure, and magnetic field in the magnetization process, three different methods were recommended, depicted in Fig. 1. Fig. 1b, the graphene suspension was poured in autoclave vessel and heated 10 h at 200 °C. In 10 h, the graphene flakes were aggregated and deposited on the bottom of vessels. Then, aggregated graphene was collected and dried. In this method, the graphene flakes were exposed to high pressure inside of autoclave and high temperature. Fig. 1c, the resultant suspension was placed on the Heater-stirrer at 200 °C until the solution was evaporated and graphene flakes remained. In this case, the magnetic field of magnet in heater-stirrer is considered as an external magnetic field source. The magnetic field, measured by a magnetic field sensor, was ~ 1 μT. In this process, both magnetic field and temperature were applied to graphene flakes. Fig. 1d, suspension of graphene was placed in the oven at 200 °C until the solution was evaporated and graphene flakes totally dried. The deposition was collected and used for further characterization. Here, the graphene flakes were exposed to just temperature. dropped on the aluminum hot surface (300 °C) and after drying the ferromagnetic graphene particles were collected. Here, the droplets were rotated and evaporated base on the Leidenfrost effect. In this method, the source of the magnetic field and temperature were Heater-stirrer and the source of pressure was the pressure inside the droplets. In fact, all parameters were simultaneously applied to graphene flakes 45 .

Results
As mentioned above, the strain is presented as an important factor to control and manipulating the electronic system and the magnetic properties of graphene [42][43][44]55 . Apart from strain, temperature and the magnetic field are significant parameters to manipulating the electronic and magnetic properties of graphene which is studied here. In all the methods, ferromagnetic graphene powders (FGPs) was created ( Fig. 2) due to the change of electronic system and lattice structure of graphene. Also, the parallel magnetic domains are observed (Figs. 3 and 4) which can be related to self-arrangement behavior of electrons. In addition, the gathering process of graphene sheets can be directly affected the electronic systems and vibrational modes. Gathering of graphene flakes can be classified into two forms, stacking and aggregating. In the stacking process, the sheets can be gathered in a normal pattern (AB and ABC) but in the aggregation process, the sheets can be twisting and do not follow the normal pattern. It is expected that the gathering process is strongly related to the time of the process. In Oven-and Heater-methods, the gathering is performed slowly and sheets can be stacked in the normal pattern. Whereas, in LFE-and Autoclave-methods, the gathering process is performed rapidly and the sheets do not have time to stack in normal patterns, therefore the aggregation process happens. In Fig. 5, the difference between both aggregation processes are shown in the scanning electron microscopy (SEM). The electronic manipulation is detected by Raman spectroscopy (Fig. 6).

Ferromagnetic graphene. Ferromagnetic behavior of graphene powders is investigated by vibrating sam-
ple magnetometer (VSM) and the results are shown in Fig. 2. Due to the natural defects (I D /I G ~ 0.25) and negligible impurities such as Ni, Co, Fe, and Mn, the weak ferromagnetic behavior is observed in pristine graphite (Supplementary Fig. S1 and Table S1). As shown in Fig. 2, the magnetization saturation (Ms) of FGPs is different in the methods. In FGPs, ~ 0.08 emu/g as Ms is generated by temperature in Oven-method, ~ 0.24 emu/g is generated by temperature and pressure in Autoclave-method, and ~ 0.16 emu/g is generated by temperature and the magnetic field in Heater-method. Clearly, the contribution of temperature, pressure, and magnetic field in the magnetization of graphene is ~ 0.08, 0.16, and 0.08 emu/g, respectively. Because of the better comparison in magnetic domains, the 4000 rpm sample of LFE is used in this study (Ms ~ 0.16 is near to other FGPs). Therefore, for comparison of Ms, the Ms of 3000 rpm LFE-sample is reported ~ 0.4 emu/g, indicated the simultaneous effect of all parameters 45 . Similar to LFE-method, the slope created in the VSM diagram of Heater-method in high magnetic fields, related to the non-uniformity of particles. In Heater-method, the non-uniform conditions are felt by flakes because of gradual removal of the solution, non-uniform of the magnetic field, and non-uniform heat flow. In LFE-method, the slope created is related to the different behavior of droplets 45    www.nature.com/scientificreports/ widely considered as a key diagnostic tool for symmetric vibrational modes, whereas IR is used for recognizing the asymmetric modes. Although most vibrational modes are active, some of the modes are silent. Generally, in Raman spectroscopy, the laser light is used to excitation and Raman scattering is detected. In addition, the electronic band structures can involve in the resonance process such as double resonance (DR) and triple resonance  www.nature.com/scientificreports/ (TR) Raman process. For example, the phonon modes far from the center of BZ are activated by the DR Raman process. Whereas, at Γ point the phonon is usually activated by first-order Raman process [57][58][59][60] . Lattice structure, the lattice vibrations, the electronic systems, and their changes can be investigated by Raman spectroscopy. Therefore, each parameter that affects the electronic systems and phonon modes of graphene can be detected and probed by Raman spectroscopy 57 . Raman spectra of all FGPs and graphite are shown in Fig. 6. Also, all effective parameters in these Raman spectra such as the number of layers, stacking order, twisting, and also the external perturbation such as strain, magnetic field, and temperature are discussed below. The wavelength of the laser used in this study is 532 nm and its spot size is 0.72 μm. Indeed, the area of the sample is covered by laser light is 0.72 μm, in each measurement. Many flakes can exist in this area which is different at the number of layers, stacking order, twisting, and electronic systems.
Interpretation of the vibrational mode is based on the structural symmetry and group theory. The highest symmetry is D 6h , which is belonged to one-layer graphene and graphite, at the Γ point. The irreducible representation of them are Γ Graphene = A 2u + B 2g + E 1u + E 2g and Γ Graphite = 2(A 2u + B 2g + E 1u + E 2g ). Three optical modes in graphene are E 2g (Raman active) and B 2g (silent). In graphite, nine optical modes exist containing three IR active, five Raman active, and one silent (B 2g ) modes. IR active modes are a doubly degenerate E 1u appear in ~ 1588 cm −1 , generated from asymmetric in-plane vibrational mode and one asymmetric out-of-plane mode, A 2u , located in ~ 868 cm −1 57,61 . In Supplementary Fig. S3, the IR spectrum for all samples is presented and E 1u and A 2u modes are observed in most samples. Also, these two peaks are too weak in graphite and LFE-sample. For graphite and LFE-sample, E 1u peak is located at ~ 1525 cm −1 and in the other samples broadening and redshift of E 1u peak is observed.
By increasing the number of layers, the symmetry is reduced and subsequently, the modes are changed. For example, N-layer graphene (NLG) symmetry with the even and odd layers is D 3d and D 3h at Γ point, respectively. Also, the symmetry information along the Γ-K direction is different and their phonons are activated by DR Raman process. For example, the symmetry of monolayer and NLG (even) at M point is D 2h and C 2h , www.nature.com/scientificreports/ respectively 58,62 . Generally, changes in the electronic system and the lattice structure can create new modes, and even the combination and splitting of some modes can appear the complex modes in the Raman spectrum. Basically, three characteristic peaks exist in graphene Raman spectra. One of the main peaks in Raman spectra is generated from symmetric in-plane vibrational mode, located in ~ 1580 cm −1 . These doubly degenerate, E 2g , is called G mode and its position is sensitive to external perturbations, such as defects, doping, strain, and temperature. The two other peaks appear in ~ 2670 and 1200 cm −1 , called 2D and D, respectively 57 .
Ultralow-frequency modes. Ultralow-frequency modes are directly related to interlayer vibrations. In graphite, the out-of-plane and in-plane interlayer vibrations, layer-breathing (LB) and shear (C) modes, are related with B 2g (~ 128 cm −1 ) and E 2g (~ 43.5 cm −1 ) modes, respectively. As another example in AB-2LG, C and LB modes are located at ~ 31 cm −1 (E g ) and ~ 90 cm −1 (A 1g ), respectively. In addition, C mode in AB-3LG and ABC-3LG are located at ~ 33 cm −1 and ~ 19 cm −1 , respectively. Generally, interlayer vibrations depend on twisting, stacking order, and the number of layer 57,61,63-68 . In Fig. 4a, the Raman spectra of graphite, Heater-, and Oven-sample are shown and no special modes are observed between 0 and 1000 cm −1 , however, many modes are observed in Fig. 4b for LFE-and Autoclave-sample. These differences are related to the aggregation and stacking process. D mode. D mode requires defects for activation. This mode is used to characterize defects in graphene. Generally, the I D /I G is used to estimate the amount of defects in graphene flakes 57 . Only edge defects can be created in the samples, related to the sonication process in the dispersion of graphene. Because of using the same dispersion method in all samples, it is expected that the edge defects are the same and I D /I G are equal (Supplementary Table S3) but in aggregation process the number of edges, exposed to the laser light are more than the stacking. Moreover, the complexity of D mode in Supplementary Fig. S4b related to combination and creation new modes. G mode. G mode is originated from a first-order Raman scattering process and related to E 2g mode at Γ point. In graphene and graphite, G mode is located in ~ 1582 cm −1 and is sensitive to external magnetic fields, strain, and temperature. According to published studies, discrete Landau levels are created in graphene by perpendicular magnetic fields. In addition, the energy and filling factor of Landau levels depend on the strength of the external magnetic field. Electron transition between these levels and resonance process with optical phonons are known as magneto-excitons and magneto phonon resonances, respectively. These are led to new optically modes, detected by Raman spectra. For example, when one-layer graphene is exposed to the perpendicular magnetic field, the symmetry representation of allowed transitions are A 1 , A 2 , and E 2 which the E 2 and E 2g (origin of G peak) can interact. Therefore, the position and the shape of the G peak are changed and led to broadening and splitting 57,[69][70][71][72] . Broadening of G mode in Heater-method can be related to the new modes such as E 2 (Supplementary Fig. S4c).
Changing the interatomic distance in crystal lattice under stain and stress lead to manipulating electronic systems. Recently, the strain effect on electronic systems is considered and remarkable results such as shifting the Dirac cones, opening band gap, shifting and splitting Raman modes, and inducing strong pseudo-magnetic field is reported. As an example, in graphene, shifting the vibrational frequency and splitting the G mode (G + and G − ) is observed under stress. In addition, the G mode are sensitive to temperature 42,57,73-80 . 2D mode. 2D (~ 2700 cm −1 ) is activated by the resonance process. The component of the 2D peak dependent on electronic bands and laser wavelengths. Here, the laser wavelength is the same in all measurements. Due to several electronic bands are in multilayer graphene, the many 2D components can be expected. Because of these features, the 2D peak is used to determine the number of layers. In addition, the stacking order also can be strongly effected on electronic bands and change the spectral profile of the 2D peak. Therefore, the complexity of 2D LFE-sample and Autoclave-sample compared to other samples can be related to more manipulation of electronic systems (Fig. 6b). In addition, 2D peak is sensitive to temperature and strain. Moreover, in the presence of strain, deformation of Dirac cones is led to splitting 2D modes into 2D + , and 2D −57,66,76,81-84 .

Conclusion
Here, three different methods are reported for the production of FGPs. Characterize of FGPs are performed by VSM, MFM, Raman, and IR spectroscopy. Magnetism is determined and magnetic domains are observed. To discussion about the results, the electronic systems and phonons are investigated. Magnetization process is based on manipulating the electronic system under temperature, external magnetic field, strain, and is also related to the gathering process. In conclusion, all three parameters can manipulate the electronic systems and control the magnetic properties of graphene powders. The best parallel magnetic domains are observed in Autoclave-and LFE-samples.