Improving the mechanical behavior of reduced graphene oxide/hydroxyapatite nanocomposites using gas injection into powders synthesis autoclave

In this study, we show the synthesis of reduced graphene oxide/hydroxyapatite (rGO/HA) composites using a hydrothermal autoclave with argon-15% hydrogen gas injection. This both increases the hydrothermal pressure and uses hydrogen as a reductive agent in the process. The synthesized powders were then consolidated with spark plasma sintering method. The analysis of the consolidated samples included Vickers Indentation technique and cell viability. The results showed that injected gases in the autoclave produced powders with a higher crystallinity compared to synthesis without the gases. Also, hydrogen gas led to increased reduction of GO. The microscopic analysis confirmed existing graphene sheets with folding and wrinkling in the powders and indicated that various preferential directions played a role in the growth of hydroxyapatite crystals. The results showed that in general, graphene sheets increased the mechanical properties of HA. In the samples synthesized with injected gases, this increase was more significant. Interface analysis results indicate that reduced graphene oxide (rGO)/HA interface is likely coherent. These nanocomposites were biocompatible and showed some hydrophobicity compared to pure HA.


experimental
The primary chemicals used in this study, along with their specifications, are presented in Table 1. The initial solution (S1) was first prepared (diethylene glycol + dimethyl formamide + deionized water with a volume ratio of 10: 10: 80) and the following steps were performed in order. The solution containing Ca +2 (4.7 grams of calcium nitrate tetrahydrate in 120 mL of S1) was added dropwise to a 20 mL stirred suspension of GO (HA/1.5% rGO) with stirring continued for 1 h. The solution containing phosphate ions (1.56 grams of diammonium hydrogenphosphate in 80 mL of S1) was dropwise added to the solution. The pH of the solutions was adjusted to >10 with ammonium solution. The resulting solution was poured into the Teflon (PTFE) vessel and transferred to the autoclave. The hydrothermal process was carried out for 8 h at 180 °C by injection of nitrogen gas at 0, 5, and 10 bar (The volume of the PTFE container was 340 mL). The powders were dried at oven for 12 h at 60 °C. The resulting powders were sintered after drying and ball milling (250 rpm, 12 h) 46,48,49 . Figure 1 shows the hydrothermal system used in this research, load-displacement, and indentation affected zone. Descriptions of this system (Fig. 1a) have already been published 54 . Evaluation of the consolidated samples. The powders were consolidated via SPS, as previously reported 54 , at a temperature of 950 °C. Exactly similar to the previous report, the biocompatibility assays were performed 54 .
To calculate the relative density of sintered samples the Archimedes method was used. In this method, wet weight, dry weight and immersed weight with a precision of 0.0001 were measured and the Archimedes ratio was used for relative density calculations (ASTM C373-88) 54 .

Vickers indentation. Instrumented microindentation experiments (Grindosonic tester with a Vickers tip)
were conducted on the polished surfaces of samples at a maximum load of 2 N and ramp dwell time of 10 s. Nine tests were performed at different locations of each sample. Elastic modulus and hardness were calculated from the load-displacement curves (Fig. 1b) using Oliver-Pharr method 55 . The modified Antis method was used to evaluate the fracture toughness (K1C) of the samples (Eq. 1) 56 : W t is the area below the load-displacement curve and W e the area below the unloading curve which corresponds to the elastic deformation. The energy W t is the total of elastic and plastic deformation (W e and W p respectively). λ is a dimensionless constant is close to 0.0498 for Vickers tip. C is the average crack length and P is the applied force. The use of experimental parameters is the major advantage of this method; it is easy to calculate when using instrumented indentation.
Characterization techniques. Different evaluation methods have been used in this research, some of which are described below. The other instruments used to characterize the samples include inductively coupled plasma (ICP) (DV7300, Optima Co.), X-ray photoelectron spectroscopy (XPS, Thermo ESCALAB 250XI), FESEM (SIGMA VP-500, ZEISS), and TEM (CM120, Philips). ImageJ 1.52d and Diamond 3.2 softwares were used in this study.
X-ray diffraction. X-ray diffraction (XRD, X′ Pert Pro, Panalytical Co.) was used to determine the phase constituents of the samples, contained a detector Cu Kα radiation (40 kV, 40 mA, λ = 1.5406 Å) and 2 theta from 10° up to 80° in steps of 0.02°. Equation 2 was used to estimate the crystallinity of HA (Xc) 57 .
Where υ (112/300) and I 300 are the intensity of the hollow between diffraction peaks of HA in the planes (300) and (112) and the intensity of the peak of HA in the plane (300), respectively. Equation 3 was used to calculate the crystallite size (Williamson-Hall method) 58 .
In this equation, d, θ, and λ are grain size, Bragg diffraction angle, and wave length of used X-ray (Cu), respectively. β and ε are full width at half height (FWHM), and crystalline lattice strain, respectively.
Field emission scanning electron microscopy. Field emission scanning electron microscope (FESEM, Hitachi S4700 equipped with energy dispersive X-ray spectroscopy) and a portable scanning electron microscope (SEM, TM-1000) were used to observe the morphology of samples (mounted in an adhesive carbon film and Au coated by sputtering for its observation). www.nature.com/scientificreports www.nature.com/scientificreports/ Fourier transform infrared spectroscopy. Fourier transform infrared spectroscopy (FTIR, VERTEX 70, Bruker Corp.) was used to identify the functional groups of the samples (resolution of 4 cm −1 , spectral region from 400 to 4000 cm −1 using 2 cm −1 steps, scan number of 8). The samples were prepared and mixed with potassium bromide (1 mg powdered samples and 300 mg KBr). The mixture was pressed into discs by applying 200 MPa pressures (1 mm thickness). The spectra were collected at room conditions (60% relative humidity, 25 °C) 48,49 .
Raman spectroscopy. Raman spectroscopy (Renishaw inVia spectrometer) was used in the range of 300-3500 cm −1 , recording 5 times for 10 seconds of each accumulation, with a wavelength of 532 nm, green laser line in a backscattering configuration using a microscope (100× objective, 100% power, an acquisition time of 10 s), which had been excited from an argon ion laser. The samples were subjected on Al foil in order to remove the fluorescence background 48,49 .
High-resolution transmission electron microscopy. High-resolution transmission electron microscopy (HRTEM, TALOS F200A with a twin lens system, X-FEG electron source, Ceta 16 M camera and a super-X EDS detector) was used to observe atomic structure of the samples and spatially resolved elemental analysis, with a spatial resolution higher than 2 nm (using TALOS microscope in STEM mode, exposure times of 5 minutes were used to create elemental distribution maps). High angle annular dark field detector (HAADF) was used to obtain STEM images (RG overlays of the STEM EDX elemental maps were made using the FIJI). To study the atomic structure, fast fourier transform (FFT) and inverse fast fourier transform (IFFT) analysis were used 48 . Figure 2 shows the XRD patterns and FESEM images of rGO-HA (P10) powders. According to the XRD pattern of the synthesized powders (Fig. 2a), full conformity is achieved between the peaks obtained and the reference standard of pure HA (JCPDS 09-0432) 45,54 . Accordingly, the HA has a high purity hexagonal structure. In other words, the XRD pattern of the rGO-HA powders is quite similar to pure HA. According to studies, graphene oxide has a peak in the range of 2theta = 10 48 . After reduction of graphene oxide, this peak is removed and a new one appears around 2theta = 26. This peak is much weaker and wider than the HA (002) peak due to the amorphous structure of rGO. Therefore, the rGO peak is covered by the HA (002) peak which is highly intensified due to its high crystallinity. Table 2 shows the specification of the HA scatter planes obtained. According to the XRD pattern (002), (211), and (300) planes are the main growth planes in HA crystals where, (002) and (300) planes are perpendicular. It is clear that injection of a gas mixture of and thus an increase in total autoclave pressure will increase the crystallinity from 75% (P0) to 86% (P5), 89% (p10) (Eq. 2) and average crystallite size from 32 nm (P0) to 43 nm (P5), 50 nm (P10) (Eq. 3). Increasing the pressure has increased the intensity of the peaks in some directions and decreased in some directions. In the direction of the (002) plane, the injection of gas has reduced the peak intensity (Fig. 2b), but in case (211) and (300) planes the peak intensity has increased (Fig. 2c). Probably due to the increase in pressure applied, the growth of crystals in directions <211> and <300> has continuously increased, and this increase in pressure in direction <002> has first reduced the growth rate and then increased www.nature.com/scientificreports www.nature.com/scientificreports/ it. For this reason, increasing the intensity of the peaks in the <211> and <300> directions is regular, but in the <002> direction it is irregular. Due to the increase in pressure with gas injection, the growth rate in directions <211> and <300> has increased more than direction <002>, and with the further increase of this pressure, the growth rate has increased in all directions. The peaks are somewhat sharper in the samples that have been synthesized in the presence of injected gas. Also, the peaks have been transmitted slightly to the right, which is probably due to increased pressure from the gas injection. Increasing the crystallinity in the use of injected gas caused the HA to be synthesized with a more precise stoichiometric ratio. As shown in the FESEM image (Fig. 2d,e), the graphene sheets (rGO) are wrinkled and folded while the HA particles are stuck on their surfaces, between them, and on the edges. The HA particles have agglomerated in some places. It is clear that morphology of these particles is nanorod shaped. These nanorods are less than 50 nanometers in diameter but have longitudinal variations. As with similar reports previously published, here is the direction of the growth of nanorods in the C-axis 33,46,49 . Figure 3 shows the TEM images, HAADF image, EDS analysis, elemental map of calcium, phosphorus, carbon, and oxygen for the synthesized powders with the presence of injected gas (P10). Figure 3a shows the high-density of HA particles at the edges of the folded graphene sheets. The size of the HA particles is very fine and their morphology is in the form of rod and prism (Fig. 3b), indicating that there are preferential growth directions in the case of gas injection mode. The TEM images also show that there is a particles size distribution. Smaller particles of HA have remained on the graphene surface due to stronger contact when preparing the samples. Also, these particles have pores due to the polycrystalline structure of HA and the presence of gases. ICP analysis of the residual solution was performed after the hydrothermal process (The findings showed that the amounts of calcium and phosphorus remaining in the solution were very low and negligible). Due to the ratio of calcium to phosphate in the input chemicals was 1.67, the ratio of calcium to phosphate in the synthesized powders is about 1.67. EDS analysis confirms the presence of trace elements (Fig. 3c-h). Elemental maps show that trace elements are homogenously distributed in the powders (P10) 49 . Figure 4 illustrates the HRTEM image with the FFT and IFFT analysis of composite powders, which is synthesized in the presence of injected gases (P10). Three areas were considered in the HRTEM image (Fig. 4a). As the analysis of region A shows (Fig. 4b,c), the HA particles grow along the (211) planes and show a d-spacing of 0.28 nm. But in area C, the (100) and (300) planes are marked with the d-spacing (Fig. 4e,g). The FFT analysis of region B shows the hexagonal structure graphene sheets (Fig. 4d). It is also the place where the crystal planes collide (Fig. 4f) 48   www.nature.com/scientificreports www.nature.com/scientificreports/ Schematic 1 shows the crystal structure of HA and a diagram of graphene atomic matching with HA crystalline planes. According to the schematic images (Schematic 1c), the atomic alignment of the crystalline planes with graphene sheets is less than the limit (0.25). Therefore, the interface between the two phases on the graphene surface is likely coherent. According to these findings, (300) planes are likely in contact with the surface of graphene sheets. In this research, during the synthesis of HA, its (300) planes are prior to the (100) planes 48 . Figure 5 shows FTIR analysis and Raman spectroscopy for synthesized powders and GO. Table 3 shows the results of FTIR analysis as shown in Fig. 5a-c. The FTIR analysis reveals that the synthesized powders contain graphene sheets and HA. By comparing the GO peaks with the final rGO-HA powders peaks, it is found that the bonds related to the oxygen-containing functional groups on the GO surface have reduced well and some of the peaks have completely disappeared. The peaks at 1395 cm −1 and 1730 cm −1 are shifted upwards, indicating reduction of GO, since these two peaks are characteristic of GO. As shown in the diagrams, the presence of hydrogen gas in the mixture of gases has led to more reduction. Also, the quality of the peaks obtained for the powders which is synthesized under the gas injection conditions is more accurate and uniform than the other, indicating a better bonds quality 45,49 .

Results and discussion
The Raman spectroscopy has been done to confirm the presence of rGO. Table 4 shows the results of Raman spectroscopy as shown in Fig. 5d-f. The rGO-related Raman signals in this spectrum are much clearer than the HA signal, although its weight percent in the powders is much lower. The D and G peaks in the rGO have not had any displacement at the Raman spectra, indicating that the rGO-HA powders have been successfully synthesized.    www.nature.com/scientificreports www.nature.com/scientificreports/ Regarding the Raman spectrum of rGO, the peaks intensity ratio (ID/IG) has increased compared to primary GO (ID/IG (GO) ≈ 0.73, ID/IG (P10) ≈ 0.91) which shows the chemical and thermal reduction that has caused structural disorder in the graphene network. It is also clear that the presence of injected gas has led to more reduction of GO 45,49 .
Density calculations from the Archimedes method showed that the sample synthesized by the classical method (P0) reached to relative density of 96.66% and the samples that were synthesized in the presence of injected gas (P5, P10) reached to relative density of 95.91% and 96.9%, respectively. The temperature of 950 °C has been chosen to reduce the probability of decomposition of HA crystals and degradation of graphene sheets.  www.nature.com/scientificreports www.nature.com/scientificreports/ Figure 6 shows the images of the P10 sample fracture surface after sintering and Raman spectroscopy for P10. According to the results of Raman analysis (Fig. 6e), no chemical reaction occurred during sintering and the graphene sheets were rescued from the consolidation process. However, as a result of the sintering process (high pressure and high temperature), the graphene sheets have been partially degraded, reducing the D/G ratio compared to the powder state. Also, some structural change has occurred in the sintered samples, causing the D peak intensity to change. Figure a-c, show the presence of graphene in three dimensions and in Fig. 6d,e, the black spots on the image are the locations of rGO-HA particles, which are distributed as acceptable in the sample volume. The HA nanorods are located on the graphene surfaces and between their layers (Graphene sheets are assembled to create a 3D structure). This porous three-dimensional structure reduces the relative density and mechanical properties of the sintered samples but according to previous studies, the presence of these porosities can increase osteoconductivity for these materials 48 . Figure 7 shows the force-displacement diagrams of the sintered samples along with the mechanical properties extracted from these graphs. To compare the effect of injected gas during the synthesis of powders on the final properties of the composites, all samples were subjected to a Vickers test. As the curves show the contact depth for P0 is greater than that for P5, P10 samples. In other words, more force is needed to achieve a constant contact depth in P5 and P10. This conclusion is also valid for P10 compared to P5. Considering the same conditions for the preparation of samples, it is likely that another mechanism including a higher degree of GO reduction or higher crystallinity, and a stoichiometric most likely is responsible for this phenomenon. Also, according to these diagrams, the elastic work in P0 is greater than the other samples. Also, the plastics work is slightly higher in P0, but with a smaller ratio, which is obtained from the surface below the curves. In these diagrams, the transition to the left means the improvement of mechanical properties. In Fig. 7d, the force-displacement curve shows that   the Vickers indenter has hit a hole in its path. The part shown with the arrow shows the contact depth where the cavity is located. These changes are more evident in samples with more porosity. In some curves, these changes appear several times in a curve. These cases involve some errors in the calculations. The indentation analysis results show that the hardness, the Young's modulus of P5 and P10 samples are higher than that of P0. Also, P10 showed better mechanical properties than the P5. The reason for this increase in mechanical properties should be examined from two perspectives. First, increasing the hydrothermal pressure increases the crystallinity of the primary powder and improves the properties of the HA, and secondly, the presence of hydrogen gas in the mixture of gases increases the reduction degree of GO and increases the mechanical properties of the graphene sheets. Table 5 shows the hardness and elastic modulus values for the samples 48,54 . According to Eq. 1, the fracture toughness (K1C) is directly related to the ratio of total work to elastic work (Wt/We), and is inversely proportional to the crack length. Therefore, comparison of these two factors is useful in analyzing the effect of graphene on fracture toughness. Figure 8 shows the three-dimensional and two dimensional diagrams of elastic-plastic work analysis for P0, P5, and P10 nanocomposites after consolidation along with the crack analysis of consolidated samples after Vickers indentation testing. According to the results, the Wt/ We ratio for the P5 and P10 has increased. The size of the cracks in the P5 and P10 nanocomposites are smaller compared to the P0 cracks. In all cases there is a deflection mechanism. This mechanism is due to the presence of nano-sized particles, which increases the strength. In a general conclusion, increasing the Wt/We ratio in the P5, P10 nanocomposites and decreasing the crack length have a dual effect on increasing the P5, P10 nanocomposites K1C compared to P0. Figure 9 shows the XPS analysis of the GO and P10 sample along with C 1 s and O 1 s high resolution fitted spectra for the P10 sample. XPS analysis is a useful way to evaluate the chemical composition of carbon-based nanomaterials. By comparing Fig. 9a peaks, Ca 2p and P 2p signals confirm the presence of HA phase after sintering process at 950 o C. As a result of these findings, HA has retained its composition after high temperature consolidation and has not been decomposed. According to Fig. 9b,c, the carboxyl groups of GO surface still remain the same as the original GO, due to the strong electrostatic interaction between calcium phosphate and the carboxyl groups present on the surface of GO 48 . Figure 10 shows fluorescent cell culture images on P10 sample after 72 hours and results of the MTT assay. The results are similar to those previously published 54 . The only difference is the morphology of the cells that are globular. The presence of rGO sheets, which has become hydrophobic by the loss of its surface agents, may have