Structure evolution, amorphization and nucleation studies of carbon-lean to -rich SiBCN powder blends prepared by mechanical alloying

Abstract

A number of carbon-lean and -rich SiBCN powder blends were subjected to mechanical alloying by high-energy ball milling generating a composite microstructure with varying proportions of amorphous and crystalline phases. Microstructural characterization at different milling stages and the evolution of free carbon and nanocrystalline SiC are discussed by using X-ray diffraction, Raman spectroscopy, transmission electron microscopy and nucleation magnetic resonance. Furthermore, the chemical bonding states of various SiBCN powder blends at different stages of milling were studied by FT-IR. The amorphization of carbon for carbon-lean powder blends was somewhat quicker than for carbon-rich analogues, which retained t-carbon and multiple graphene structures after being subjected to 40 h of milling. The chemical bonding state changes are similar for all investigated powder blends, while detailed microstructure changes are evident and consist of a considerable amorphous nature and a small amount of nanocrystallites after 40 h of milling. The forming nanograins are assigned to Si and SiC for carbon-lean and -rich powder blends, respectively. Ball milling leads to alloying, complete or partial solid state amorphization, accompanied by strain-induced heterogeneous or homogeneous nucleation of nanocrystalline phases from an amorphous matrix.

---

1. Introduction

Amorphous, thermodynamically stable materials, such as SiBCN ceramics, with relatively low density, outstanding high temperature performance, excellent mechanical properties and special structures, have generated growing research interest over the last decades.^1–3 These ceramics and their composites are especially promising candidates for reentry aircraft nose cones, wing leading edges and turbine blades.⁴ Polymer-derived SiBCN ceramics (PDC-SiBCNs) prepared using controlled thermolysis of preceramic polymers including silicon, boron, carbon and nitrogen elements with desirable compositions and microstructures can be produced.⁵ PDC-SiBCNs with distinct compositions exhibit different properties, revealing a close relationship between the initial chemistry, processing route and subsequent final nano-and interdomain regions.^6–8 The main advantages of PDC-SiBCNs are the stability of their disordered long-range structures and superior oxidation resistance compared to analogues at high temperatures.⁹

To prepare a promising SiBCN candidate, it is essential to obtain a homogenous distribution of the constituent elements; however, the low self-diffusion rates of silicon and boron in matrices limit their long range migration.¹⁰ Even though an expected distribution of the elements in SiBCN ceramics can be achieved by careful synthesis, this route involves complex preparation processes and multiple synthetic stages that often suffer from poor yields.^11,12 Moreover, the thermolysis stage requires extensive densification which in monoliths leads to substantial pores and flaws in the final matrix. Finally most processing steps must be conducted in an inert atmosphere.¹³ These unfavorable results constrain the utility of PDC-SiBCNs primarily to laboratory scales. Thus, dense amorphous SiBCN ceramics processed suing PDC routes require a significant investment in time and equipment. These problems mandate the development of new routes to SiBCN ceramics.

Mechanical alloying (MA) is a rapidly developing technique capable for producing a wide range dispersion strengthened alloys, with partially or completely amorphous/glassy microstructures from elemental powder blends.¹⁴ The materials processing is very simple and seems to be readily scalable: powders of new materials are produced by high energy ball milling of the starting ingredients.¹⁵ Solid-state amorphization by mechanical alloying involves deformation at high strain rate, cold welding, fragmentation and dynamic recrystallization.¹⁶ The development of amorphous microstructures also in ceramic powders with positive enthalpy changes implies that increases in interfacial and strain energy contributions could contribute significantly in solid-state amorphization.¹⁷ Besides strains, the chemical reactions may also contribute to amorphization in solid state over a wide composition range.¹⁸ Since the microstructures produced by MA, the strain and lattice disorder can be remain in a highly metastable state,¹⁹ therefore, the composition range for solid-state amorphization is usually larger in mechanically driven processes than that by rapid solidification methods.²⁰ The above results provide the possibility for the preparation of completely amorphous or partially nanocrystalline SiBCN powders.

Early work^21–23 on mechanical alloyed SiBCN powder blends mainly concerned the influence of ball milling parameters on structure, physical and surface characteristics of specific ingredients Si₂B₁C₃N₁ powders, and also focused on their thermal stability in inert atmospheres. However, research on the preparation of SiBCN powders by MA is still in its infancy, and much work remains to be done. It is known that the powder characteristics, such as particle size, morphology and elemental distribution, agglomeration, composition, impurity and surface chemistry, likely have a strong impact on compressibility, sinterability and the resulting ceramic microstructures.²⁴ In this regard, the powder constituents may play an important role in determining the morphology and structural evolution, amorphization and nucleation of the powder mixtures during MA. A further issue in the design and processing of optimal MA SiBCN powders, is their relatively low thermal stability, inhomogeneity and increased sensitivity to oxidation compared with PDC-SiBCNs.²³ The elemental composition design by mechanical alloying silicon, boron nitride and graphite in arbitrary molar ratio allows the possibility to tailor powder structures and phase compositions, which in turn determines and optimizes the properties of the subsequent sintered ceramics.

In the PDC-SiBCN systems, boron plays a key role in the formation of amorphous structures, retards decomposition of silicon-nitride bonds and reduces diffusion of different species involved in the carbothermal reaction at high annealing temperatures.^5,25–27 In the case of amorphous PDC-SiBCNs ceramics, an increased resistance to decomposition is beneficial for applications at high temperature compared with PDC-SiCN systems.^28–30 These effective methods provide materials for applications at (ultra)high temperature under extreme environments. To date, the role of boron in the structures and properties evolution of the PDC-SiBCNs is well studied; however, other elements such as carbon and nitrogen still require further study.

To clarify the effect of N content over a wide compositional range on materials' properties and structural evolution, Petrman prepared amorphous SiBCN materials using reactive magnetron sputtering, focusing on their microstructure, electronic structure, electrical and optical properties by using a combined approach of experiment and ab initio calculations.³¹ This work determined that decreasing the N content lowered the electrical resistivity and optical gap. Unfortunately, this approach allowed only one to tailor SiBCN compositions which can combine different functional properties, such as high thermal stability and electrical conductivity. In a similar work,³² Liao employed molecular dynamics simulations to study the effects of N content on the structural and mechanical properties of SiBCN ceramics and the simulation results gave a key insight into the nanostructure and mechanical properties of SiBCN ceramics with different compositions. To date, the art of carbon in SiBCN ceramics has been little studied with a few reports on carbon-rich PDC-SiCNs and -SiOC systems suggesting that carbon-rich PDC-SiCNs is strongly dependent on the carbothermal reactions between the amorphous SiCN matrix and excess carbon, forming micro- and mesopores in these materials.^33–37 The phase composition of PDCs-SiCN after pyrolysis at 800 °C consists primarily of amorphous nanodomains of sp² carbon and silicon nitride with an interfacial region characterized by mixed bonding between N, C, and Si atoms, likely stabilized by the presence of hydrogen.³³ The atomistic simulations used to generate amorphous SiCN with different carbon contents, reveal a tendency to include a “free carbon” phase as carbon content increases.^38–41

Moreover, electrochemical experiments on different carbon-rich PDC-SiCNs-based ceramics indicate that they offer high capacity/high stability anode materials for lithium-ion batteries applications.^42–44 Based on these investigations, it is assumed that carbon is essential for nanodomains changes, and may inhibit crystallization promote phase transitions in sintered SiBCN ceramics by MA; although there is no literature evidence either way. Thus, in order to obtain an optimized stability of amorphous state for technological applications and to produce tailor-made microstructures, it is highly desirable to investigate and understand the formation of nanosized crystallites, solid-state amorphization during mechanical alloying and its structure changes of carbon-lean to carbon-rich SiBCN powder blends. This then is the goal of the current work.

To this end, we have synthesized a number of carbon-lean and -rich amorphous or partially nanocrystalline SiBCN powder blends by mechanical alloying. Besides characterizing the morphology and microstructure changes, we find a strong influence of initial composition and milling time on the process of solid-state amorphization. Yet the precise mechanisms of amorphous phase formation are not easily delineated, in contrast, concerning on the formation of nanocrystalline SiC during MA process. Moreover, the structural evolution of the free carbon phases in amorphous matrix is also discussed. Finally, NMR studies are undertaken for chosen SiBCN powder blends to illustrate the SiC formation process, and we also attempt to offer a mathematical model to discuss about this.

2. Experimental section

2.1 Materials

The starting constituent powder blends, including cubic silicon (99.2% in purity, average particle size 45.0 μm), graphite (99.5% in purity, average particle size 8.7 μm) and hexagonal boron nitride (98.0% in purity, average particle size 0.6 μm) were procured from Beijing Mountain Technical Development Center of China.

2.2 Mechanical alloying

The high energy ball milling used in current work was a Fritch P4 series planetary mill. The powder blends were poured into Si₃N₄ milling vials with Si₃N₄ balls inside a glove box operated under a high purity argon atmosphere. The nominal rotational speed of the main disk was set as 350 rpm, and the vials were 600 rpm in reverse. Milling was done for 0.5, 1, 3, 5, 10, 20 and 40 h using a ball-to-powder mass ratio (charge ratio, C_R) set as 20 : 1 and the measured diameters of the ball used were 10 mm. The machine was paused for 20 min every 40 min to dissipate heat. Mechanical alloying was carried out with ten different components of powder blends with the same Si and h-BN contents but varying proportions of graphite. It may be noted that the carbon content is varied over a wide range as the XRD and TEM related to free carbon were studied in this work to monitor structure evolution. The molar ratio of the starting materials and milling parameters are presented in Table 1. The C7 powder blends are regarded as nominally stoichiometric powders for this powder composition consisting of only SiC and BN(C) phases without free carbon phase after hot pressing, as a result, the C1–C5 are denoted as carbon-lean powder blends while C9 and C10 are marked as carbon-rich ones.

Table 1 Elemental molar ratio and milling parameters of SiBCN powder blends prepared within the present study

Sample codes	Powder blends	Molar ratio of the starting materials			Milling parameters
		Si	C	BN	Milling time (h)	Ball-to-powder (wt)	Rotational speed (rpm)
C1	Si2B1C0.1N1	2	0.1	1	0.5–40	20 : 1	600
C2	Si2B1C0.5N1	2	0.5	1	0.5–40	20 : 1	600
C3	Si2B1C1N1	2	1	1	0.5–40	20 : 1	600
C4	Si2B1C2N1	2	2	1	0.5–40	20 : 1	600
C5	Si2B1C2.5N1	2	2.5	1	0.5–40	20 : 1	600
C6	Si2B1C2.9N1	2	2.9	1	0.5–40	20 : 1	600
C7	Si2B1C3N1	2	3	1	0.5–40	20 : 1	600
C8	Si2B1C3.1N1	2	3.1	1	0.5–40	20 : 1	600
C9	Si2B1C3.5N1	2	3.5	1	0.5–40	20 : 1	600
C10	Si2B1C4N1	2	4	1	0.5–40	20 : 1	600

---

2.3 Microstructural characterization

The identity and sequence of phase evolution in different milling stages were detected by X-ray diffraction (XRD) analysis using Rigaku Corp. diffractometer with Cu-K_α radiation. The particle size distribution was measured by a laser scattering particle-size analyzer (LA-920, Horiba Corp., Japan). In addition to monitoring the evolution of microstructure and chemistry, XRD was also used to determine the crystallite size and internal strains. The results of the XRD analysis concerning grain size and amorphization were verified by transmission electron microscopy (TEM) using a FEI Corp. Tecnai G² F30 instrument operated at 300 kV using high-resolution, bright filed images and selected area diffraction (SEAD) analysis. Energy dispersive spectrometer (EDS, Oxford instruments INCAx-act, Oxforshire, UK) attached to the TEM was also adopted to study the elemental arrangement and phase distribution. The morphology and size distribution of the as-milled powder blends were analyzed using a FEI Corp. Quanta 200 FEG scanning electron microscopy.

Fourier transform infrared spectroscopy (FT-IR) was conducted on an FT-IR spectrophotometer (FT-IR, 510P, Nicolet Corp., USA). Raman microprobe spectra were taken on an (Invia, Renishaw, USA) with 514.5 nm Ar⁺ laser excitation. To investigate the oxidation behavior on heating of the as-milled powder blends, TG-DTA mass spectroscopy (TG-DTA-MS, STA449F3, NETZSCH GmbH, Germany) was employed. The analysis was conducted on a thermoanalyzer with a crucible of ZrO₂ (Netzsch Group, Germany) at temperatures ranging from 40–1200 °C with a heating rate of 10 K min⁻¹ in high pure synthetic air (gas flow, 80 mL min⁻¹). The solid-state ²⁹Si nuclear magnetic resonance (solid-state NMR) spectra were performed by a Varian 400WB NMR spectrometer operating at a static magnetic field of 9.39 tesla, employing an 4 mm magic angle spinning (MAS) probe at spinning rates of 3 kHz at room temperature. The single-pulse ²⁹Si NMR spectra were recorded using a 90° pulse of 9 μs and recycle delays of 120 s.

3. Results and discussion

3.1 Effect of the milling time and carbon concentration on morphology and sizes reduction

The Fig. 1 SEMs present the variations in morphologies and sizes of the C7 powder blends for different milling times (0.5–40 h). The particles are mainly non-spherical of variable sizes below milling times of 3 h. Fig. 1(c) shows the surface of some mixed powder particles uniformly covered by nanoparticles of h-BN and carbon based on EDS results (not shown). Both h-BN and graphite have similar crystal structures, and the interlayers of both can be destroyed easily by mechanical alloying. Therefore, in the early stages of milling, the soft h-BN and graphite show a strong tendency to weld together to form nanocrystallites or amorphous structures.

Fig. 1 Surface morphologies of as-milled C7 powder blends: (a–h) morphologies of the C7 powder blend milled for (a) 0.5 h, (b) 1 h, (c) 3 h, (d) 5 h, (e) 10 h, (f) 20 h, (g) 40 h and (h) 40 h at higher magnification.

As the powder continues to blend these components flatten and cold welded together, brittle particles of lamellar silicon are covered by the nanosized particles, which construct this morphology in Fig. 1(c). The different morphologies of 5 h as-milled mixtures exhibit mostly larger agglomerated particles with spherical shapes, as displayed in Fig. 1(d). Agglomeration occurs at an early stage in ball milling (3–5 h), superseded by fracture and fragmentation. For milling times greater than 5 h, the structure of the powder mixtures is steadily refined due to deformation at much high strain rates, cold welding, fragmentation and dynamic recrystallization, and the particle sizes continue to decrease in the decline of strong agglomerating forces. After 10 h of milling, the particle size distribution is more uniform and the particles become more spherical and flat. Ball milling for 20 h and causes no further changes compared with Fig. 1(e), but decreases the particle size as a whole. However, on further milling to 40 h, the particle sizes increase because of cold welding and higher specific surface areas.^45,46 The average particle size of agglomerated powder is 5.2 μm, however, the size of individual powder particles is merely 100 nm or less, as indicated in Fig. 1(h). Fortunately, the smaller particles are beneficial to hot consolidation as the larger surface to volume ratio generally enhances densification.

The morphologies of as-milled SiBCN powder blends with various carbon concentrations are presented in Fig. 2. The surface morphologies of the resulting powders are similar, however, the particle sizes differ. For carbon-lean powders, the particles are almost regular with rough surfaces and uniform distribution inside the nanoparticles. Further increases in carbon content lead to a spherical morphology and diminishes the particle size effectively. That is to say increasing carbon content leads to a reduction in particle sizes, as also confirmed by the XRD results. It is suggested that there are two possible amorphization mechanisms during MA process; one is pressure induced and the other is crystallite-refinement induced. However, another amorphization mechanism is possible when milling gas/solid systems, that is, composition-induced amorphization or amorphization driven by chemical effects.

Fig. 2 SEM images of SiBCN powder blends with various carbon concentrations milled for 20 h.

3.2 Studies on microstructural evolution by XRD and TEM

Fig. 3 summarizes the typical XRDs of MA SiBCN powder blends after milling from 0.5 to 40 h. The characteristic peaks of the constituent elements gradually decrease in intensity, increase in width (full width at half maximum, FWHM) disappearing completely beyond 40 h depending on the carbon concentrations. The diffraction intensities of raw materials used such as h-BN and graphite drop dramatically after 0.5 h of milling and increase in width to finally disappear beyond 1 h of milling.

Fig. 3 Typical XRD patterns of the various SiBCN powder blends ball milled at different times.

Continued milling up to 5 h leads to a small diffuse halo coexisting with unprocessed starting silicon, indicating formation of amorphous phases. However, no distinct features were detected other than a vibration of broadening and intensity for diffraction lines of remaining silicon for all investigated powder blends with prolonging milling times. It should be mentioned that for carbon-lean blends, such as C1 and C2, the silicon remains till 40 h of milling while others present a contrary result, indicating the contribution of carbon to amorphization. Moreover, the XRD spectra of the prepared powder blends reflect weak diffraction peaks of silicon carbide, implying the formation of SiC crystallites. Since the increase of the milling times beyond 40 h of milling could not change the mechanical energy input into the blends, the microstructures of SiBCN powders with certain constituent will show nothing different, and the crystallite content and crystalline size may have little variation. The deduction is evidenced by the particle size distribution and grain size of the remaining silicon after MA process.

Grain sizes were obtained from the analysis of peak broadening in the XRD. Such broadening coincides with refinement of crystals, introduction of internal strains and instrumental effects. After correction for Kα₂ intensity and instrument broadening, the broadening owing to small crystal sizes was evaluated by the Scherrer formula.⁴⁷

B_p(2θ) = 0.9λ/(d cos θ) (1)

where d is the average dimensions of crystalline particles, B_p (2θ) is the broadening of the diffraction line collected at half maximum intensity, λ is the wave length of the X-ray and θ is the Bragg angle. The strain broadening can be measured using:

B_s(2θ) = (η sin θ)/cos θ (2)

where B_s (2θ) is the broadening depending on the internal strains, η is the effective internal strains. Finally, by assuming a Cauchy size and broadening profile and Gaussian strain broadening profile, it follows as:

B_T(2θ) = 0.9λ/(d cos θ) + (η sin θ)/cos θ (3)

Therefore, by construction of a least squares fit to plot B_T(2θ) cos θ against sin θ, the size and strain can be obtained from the slope and ordinate intercept. The evaluated grain sizes of various SiBCN powders as a function of milling times are given in Fig. 4(c). It is noted that the silicon crystallite sizes decrease with prolonging milling, suggesting effective introduction of internal strain. Moreover, the carbon-rich SiBCN powder blends show smaller grain sizes compared with the carbon-lean counterparts, indicating the effective contribution of carbon on size decreases. Also, the average particle sizes (APSs) of powders milled at 20 h are 6.9–3.4 μm and the median particle size is 6.6–3.2 μm, see Fig. 4(b) collected from laser scattering analysis. However, further milling decreases the crystallite sizes but results in an increase in the particle size, as shown for the 20 h and 40 h as-milled powders in Fig. 4(a).

Fig. 4 The particle size distribution for (a) the C7 powder blend prepared using different ball milling times and (b) SiBCN powders with different carbon concentrations milled for 20 h; (c) crystallite sizes of the remaining crystalline Si as a function the milling times for SiBCN powders with various carbon contents.

Experimentally determined lattice parameters for the remaining silicon of C7 powder blends as a function of milling time are given in Table 2. For the reliable calculation grain sizes (Scherrer formula) range 1–100 nm, thus the data collected in the initial milling stages may contain some errors. As indicated, the lattice parameters increase steadily with increasing milling time accompanied by increases in lattice strain. From a knowledge of the respective phase diagrams of Si–C and Si–N, the light atoms such as boron, carbon and nitrogen are not incorporated in silicon lattice during mechanical alloying, so the peak broadening of Fig. 3 can likely be attributed to a progressive reduction in crystalline size and enhanced lattice strain arising from crystal imperfections and plastic deformation despite it being very limited. The XRD profiles and previous work^21–23 have confirmed that a completely/almost amorphous state could be obtained after milling beyond 20 h, therefore, microstructure characterization by TEM was done to investigate the structure evolution of carbon-lean to carbon-rich SiBCN powder blends after 40 h of milling.

Table 2 Summary of the microstructural characteristics of C7 powder blends as a function of milling times

Milling time (h)	Lattice parameter (nm)	Mean particle size (μm)	Crystallite size (nm)	Lattice strain (%)
0.5	0.3136 ± 0.0001	9.33	952 ± 35	0.032 ± 0.002
1	0.3138 ± 0.0002	7.52	542 ± 27	0.096 ± 0.004
3	0.3139 ± 0.0001	6.02	132 ± 16	0.13 ± 0.01
5	0.3141 ± 0.0002	5.90	102 ± 15	0.19 ± 0.01
10	0.3142 ± 0.0003	5.18	85 ± 12	0.22 ± 0.02
20	0.3143 ± 0.0002	5.02	61 ± 11	0.26 ± 0.01

---

Fig. 5(a) and (b) show bright-field TEM images taken from a carbon-lean C1 powder blend after 40 h of MA. The microstructure of this powder consists mostly of nanocrystalline grains and a completely amorphous zone, confirmed by the SEAD patterns inserts. The HRTEM images in Fig. 5(c) and (d) show the remaining silicon and an amorphous matrix, in accord with the XRDs. However, the volume fraction of amorphous region dominates after 40 h of milling. Moreover, no other structures, such as nanocrystalline SiC and turbostratic (t-)BN(C) phases are observed.

Fig. 5 TEM images of carbon-lean C1 (a–d) and C3 (e–h) powder blends milled for 40 h; HRTEM images of detailed structures and SEAD patterns collected from larger region.

Fig. 5(e) and (f) provide the microstructural information of carbon-lean C3 powder. The SEAD patterns inserted in Fig. 5(e) exhibit typical halos of amorphous materials in agreement with the Fig. 3 XRD. However, the diffraction spots reveal some crystalline silicon after 40 h. The HRTEM in Fig. 5(f) shows turbostratic BN(C) 10 nm.²³ Fig. 5(g) and (h) also show completely amorphous local regions and crystalline zones. The crystalline region exhibits a d-spacing that matchs d(111) = 0.317 nm of silicon. The amorphous regions are more obvious and dominant compared with carbon-lean C1 powder. MA, drive formation of BN(C) phase with BN rather than SiC.⁴ However, in carbon-lean C2 powder blend, the forming SiC is prior than BN(C). Thus, increased in carbon may contribute to the formation of t-BN(C) which is more prone to amorphization.

The elemental distribution of carbon-lean C1 powder is given in Fig. 6. The elemental distribution of Si and C clearly show segregation, moreover, C and N contents decline sharply across the Si-rich region to the Si-lean zone suggesting C tends to react with B and N rather than Si. The bright spots in the C and N mapping images reveal formation of turbostratic BNC phases. Fig. 7 provides the elemental distribution of interface between BNC and amorphous zone. The B, C and N elements completely overlap supporting the above comment that BNC phase formation is favored with increases in carbon content.

Fig. 6 TEM and elemental distribution images of carbon-lean C1 powder blend milled for 40 h.

Fig. 7 TEM and elemental distribution images of carbon-lean C4 (a) and C5 (b) powder blends milled for 40 h showing interface of amorphous matrix and BNC phase.

Fig. 8(a) shows the bright-field image of nominally stoichiometric C7 powder blends milled for 40 h. The corresponding SEAD pattern inserted in Fig. 8(a) records both typical halo-pattern and spots due to the present of both amorphous matrix and nanocrystallites of BN(C) and SiC phases. Analysis the Debye rings reveals that the nanograins are distributed in a crystallographically random orientation. The variation in intensity of the rings is because of localized and inhomogeneous deformation and fracture by shear deformation.⁴⁸ The HRTEM image shown in Fig. 8(b) confirms the coexistence of amorphous regions as well as the nanocrystalline areas. Careful observation in other regions reveals similar nanocrystalline + amorphous microstructures with varying degree of relative volume fraction. In fact, the nanosized regions in C7 powders are larger the former. Both the SEAD patterns collected from larger regions in Fig. 8(c) and (d) record the amorphous or disorder structure. Based on the above analysis, it can be argued that the increase of carbon concentration in SiBCN powder blends may result in the formation of nanocrystalline SiC uniformly distributed in amorphous matrix. Moreover, it should be noted that the t-BN(C) phases are evident and increase in volume fraction at this carbon concentration.

Fig. 8 TEM images of nominally stoichiometric C7 powder blend milled for 40 h; (a) TEM image and corresponding SEAD pattern, (b–d) HRTEM images of detailed structures and SEAD patterns collected from larger region.

To investigate the role of carbon on microstructural evolution of the present quaternary systems, two powder blends with higher carbon concentrations were subjected to mechanical alloying under identical conditions. The corresponding SEAD pattern in Fig. 9(a) indicate a diffuse intensity halo with additional low intensity spots substantiating the XRD evidence of amorphous + nanocrystalline microstructure after 40 h of milling. The t-BN(C), nanocrystalline SiC and amorphous regions appear without obvious differences compared with the Fig. 9(b) and (c) materials. The HTREM of Fig. 9(d), illustrates some local ordered regions with widths of 5 nm and lengths of over 50 nm. An EDS line inserted in Fig. 9(d) reveals the ordered structure mainly comprises carbon, a multiple graphene forms during MA. An ordered microstructure exists even after 10 h of milling.

Fig. 9 TEM images of carbon-rich C9 (a–d) and C10 powder blends (e–h) milled for 40 h; HRTEM images of detailed structures and SEAD patterns collected from larger region.

Possibly, milling at higher strain rates (related to high rotational speed and charge ratio, C_R) or with different compositions is necessary for complete amorphization. Even so, the nanocrystallites dispersed in an amorphous matrix rather than completely amorphous or nanosized microstructures are more likely to yield greater strength and toughness, especially with uniformly distributed multiple graphene structure in SiBCN matrix after hot consolidation process. Comparison of microstructures at 20 and 30 h milling shows that the volume fraction of the nanocrystalline phase marginally decreases (not given here). Further microstructural disorder/defect introduced by continued milling over 40 h would transform the microstructure completely amorphous.

A similar structure is also found in carbon-rich C10 powder blends subjected to 40 h of milling. The only difference in structure change between C9 and C10 powder blends exists in the volume fraction of multiple graphene, and the contrast is obvious. From the present results, it appears that carbon concentration play a key role on the formation of crystalline structure and amorphous matrix, especially multiple graphene for the carbon-rich SiBCN powder blends. The increasing carbon concentrations lead to the formation of nanocrystalline SiC and t-BN(C), but also contribute to the amorphization of SiBCN matrix. For the carbon-lean powders, such as C1–C4, the amorphous regions are in a dominant class with some remaining silicon after 40 h of milling. For the near-stoichiometric or stoichiometric of C6, C7 and C8 powders, the volume fraction of nanosized SiC and t-BN(C) increase rapidly but still the amorphous regions are obvious and dominant, however, the multiple graphene has not been constructed compared with the carbon-rich analogues. Combined with the XRD results, we find that the ordered structure of graphite is easily being destroyed after 3 h of milling for all the target powder blends and reaches a well amorphous state with further milling. Thus, it seems that the multiple graphene structure stems from the amorphous carbon structure or disorder carbon. However, we could not confirm the above inference because XRD is insufficient to detect nanodomains of free carbon. To further assess the formation of multiple graphene structure, Raman and HRTEM are adopted.

3.3 The structure evolution of free carbon in SiBCN matrix

Raman spectroscopy provides information on the evolution of carbon species.³³ The most prominent features in the Raman spectra of as-milled SiBCN powders are the so-called disorder induced D band at approx. 1355 cm⁻¹, the G band at approx. 1580 cm⁻¹ and the G′ band due to the overtone of the D band always detected in defect-free carbon materials at 2700 cm⁻¹.⁴⁹ Another Raman peak at about 2950 cm⁻¹, is associated with a D + G combination model induced by disorder and the 2nd order Raman peak 2D appears at 2600 cm⁻¹ (here we only show the two main peaks of D and G bands).^50,51 The intensity ratio of the D and G bands, I(D)/I(G), enables the evaluation of the carbon-cluster size by using the formula reported by Ferrari and Robertson:⁵²where, L_a² is the lateral size of carbon domains along the sixfold ring plane and C′ is a coefficient that depends on the excitation wavelength of the laser. The value of the coefficient for the wavelength of 514.5 nm of the Ar-ion laser employed here is 0.0055 Å⁻².

The Raman spectra of the carbon-lean and -rich SiBCN powder blends milled for 20 h are presented in Fig. 10(a). For the carbon-lean powders of C1–C3, no Raman peaks assigned to free carbon are found. However, carbon-lean C4 and C5 powders exhibit broad and strong overlapped signals as a consequence of the pronounced structural disorder of the carbon phase. For the nominally stoichiometric C7 powder, relatively distinct and narrow peaks of D and G bands are detected due to the formation of disordered carbon (t-carbon) and ordered graphene-like structure. With increases in carbon concentration, the peaks corresponding to D and G models are more evident, seen as C9 and C10 spectra in Fig. 10(a). This means that multiple graphene are more frequent observed in these SiBCN powder blends. Combined with the TEM images, emergence of D band can attributed to the turbostratic carbon or lateral size reduction of nano graphite sheets. The in-plane crystallite sizes (for t-carbon) L_a for free carbon are 1.12–1.23 nm. The I(D)/I(G) ratio remains almost the same with values higher than 1 for C7–C9 powders, while decreases value less than 1 for C10 powder. The decrease in ratio of I(D)/I(G) and consequently in the cluster size of disordered carbon stems from a higher order degree of graphitization of C sp² layers. From the above, it can be inferred that carbon-rich powders consist mainly of t-carbon and multiple graphene, especially the latter. In contrast the carbon-lean powders are free of both materials due to their low carbon concentrations or strong tendency to amorphize. The above results are highly consistent with the TEM results.

Fig. 10 Raman spectra of (a) SiBCN powders with various carbon contents milled for 40 h and (b) C7 powder blend milled at different times.

The Raman spectra of nominally stoichiometric C7 powder milled at different times are given in Fig. 10(b). The D and G bands remain sharp after 1 h of milling, but decline after 3 h of milling, suggesting an amorphous carbon state arises in accord with the XRD patterns. In contrast, D and G bands rise obviously after 5 h of milling as broadened and overlapped signals; with higher intensity at 40 h. The results shown above reveal that the graphite structure is mostly destroyed after 3 h forming disordered carbon or multiple graphene at and after 10 h of milling. In other words, MA amorphizes carbon at low milling times but also promotes formation of t-carbon and multiple graphene structures upon 10 h of milling.

The HRTEM images of carbon-lean C4 and carbon-rich C9 powders subjected to different milling times are displayed in Fig. 11. Results clearly show that nanocrystalline structures form after 5 h of milling changing to a nanocrystalline plus amorphous structure after 20 h of milling for carbon-lean C4 powder, Fig. 11(a)–(c). Prolonging the milling time up to 40 h leads to a completely amorphous structure. The carbon-rich C9 powder reveals a distinct structure changes subjected to various milling times, as indicated in Fig. 11(e) and (f). After 5 h of milling, the disordered carbons combined with multiple graphene structure are observed. Continuing milling leads to no further structural changes but alters the volume fraction of disordered carbon and multiple graphene.

Fig. 11 Microstructures of the carbon-lean C4 (a–d) and -rich C9 (e–f) powder blends under the milling times of (a), (e) 5 h; (b), (f) 10 h, (c), (g) 20 h and (d) 40 h, observed by HRTEM.

The lattice structure of graphite has been described in detail.^53–57 During the high energy ball milling, graphite sheets are thinned by delamination or cleavage.⁵⁴ Defects are introduced and the basal plane stacking order is probably reduced by a shearing.⁵⁵ Actually, the short milling time (below 1 h of milling) may lead to nanocrystalline graphite embedded into the amorphous matrix. The shearing force on lattice planes generates a great amount of disordered carbon and slight rotations of sp² sheets in intermediate stage (up to 3 h and more). Coincidentally, numerous half Frank loops, interstitial loops, bending and buckling of the basal planes occur. The accumulation of these defects finally causes a collapse of hexagonal backbone to a very fine scale until a complete amorphous-like state is obtained.⁵³

At this intermediate stage of ball milling, thinner graphite sheets form uniform nanoscale t-carbon. However, for the carbon-lean powders, the disordered-like structure of carbon is not found due to rapid amorphization or reactions with silicon and BN. The carbon-rich powders generate multilayer graphene with lateral sizes of 5–10 nm and t-carbon. Prolonging milling leads to graphite or multiple graphene that is progressively disordered, curled and corrugated along the edges.^56,57

Only at long milling times are the multiple graphene-structures converted to t-carbon or amorphous structures beyond at 40 h, as indicated in Fig. 11. Although, the carbon in such a carbon-rich SiBCN powder seems completely amorphous after 3 h by XRD, Raman spectra and HRTEM confirm the presence of considerable multilayer graphene and t-carbon in amorphous matrix even after 40 h of milling. However, free carbon is essential for inhibiting crystallization and lowering carbothermal reactivity in Si–C–O and Si–C–N systems.^33,35 In toto, multiple structural changes induced by the ball milling lead to fluctuations in sp³/sp² ratios and the carbon concentration acts as a key factor in structure changes of SiBCN powder blends milled at various times.

3.4 Study on the changes of chemical bonds

FTIR spectroscopy was used to investigate bonding in the amorphous phases and the effects of milling time on bonding, FT-IR spectra of the various SiBCN blends are shown in Fig. 12(a). FT-IR spectra for all powders milled for 40 h are nearly the same, showing only variations in peak intensities. As previously reported,^7,9,33 MA or precursor-derived SiBCN ceramics present ν-OH stretching vibrations at 3496 cm⁻¹, ν C=N 1652 cm⁻¹, ν C=C 1582 cm⁻¹, ν B–N 1350 cm⁻¹, ν Si–O 1183 cm⁻¹, ν C–N 1050 cm⁻¹, ν Si–N 956 cm⁻¹ and ν Si–C 843 cm⁻¹. As for ν-OH, the intensity increases with increasing carbon concentration, indicating moisture absorption.²¹ For all the powder mixtures, increases in intensity of ν C=C at 1582 cm⁻¹ is detected. The broad peak corresponding to ν C=N can be traced to a nitrogen containing graphene-like structure, confirmed in polymer-derived SiCN powders.³³ Moreover, the ν Si–N band at 956 cm⁻¹ appears and increases in intensity with the carbon concentration.¹⁰ A band at 1350 cm⁻¹ emerges assigned to ν B–N and the ν Si–C appears at 843 cm⁻¹. Unexpectedly, the ν Si–O at 1183 cm⁻¹ results from oxygen contamination.

Fig. 12 FT-IR spectra of (a) SiBCN powder blend with various carbon contents milled for 40 h and (b) the nominally stoichiometric C7 powder blend milled for different times.

Fig. 12(b) shows the FT-IR spectra of nominally stoichiometric C7 powder milled for different times suggesting similar feathers compared with Fig. 12(a). However, the moisture absorption behavior weakens and the intensities of ν C=N, ν C=C, ν B–N, ν Si–N and ν Si–C bonds increase with prolonging of the milling time, showing an effective contribution of high energy ball milling to reactions among the four elements. All the spectra lines reveal the appearance of all investigated peaks though only 1 h of milling, however, the most important bands related to the presence of ν Si–C and ν Si–N increase with prolonging milling time. It should be noted that the presence of ν Si–N bond is not evidenced by the XRD and TEM results after 20 h of milling, hence it should come from the amorphous matrix. In a whole, the experimental findings collected from the FT-IR patterns are consistent with the aforementioned analysis.

3.5 Formation of silicon carbide during mechanical alloying by NMR investigation and its mathematical model

Reaction milling, is a modification of mechanical alloying which is mainly used to prepare supersaturated crystalline solid solutions, amorphous phases, nanocrystalline solids and solid solutions containing two or more immiscible component. With the influence of the high energy ball milling, not only various new chemical bonds and nanocrystalline SiC formed among the four elements Si, B, C and N, but also these elements were mixed at the level of atom and formed amorphous structure. In this part, we will discuss the formation of SiC during ball milling by NMR and attempt to provide a mathematical model.

Fig. 13 shows the ²⁹Si NMR spectra of carbon-lean C2 and carbon rich C9 powder blends at various interval of milling. As seen in Fig. 13(a), the peak corresponded to pure Si (at −80 ppm) is distinct at the initial stage of milling (1 h), and maintains in intensity with increased broadening of linewidth until 5 h of milling. Prolonging milling up to 10 h, the resonance part from pure Si falls dramatically in intensity and becomes much broader in linewidth. However, the trace of pure Si is detected in NMR spectra until 20 h of milling. The XRD results of C2 powder blends at similar milling stage substantiate that Si undergoes mutual deformation to from amorphous and crystalline phases beyond 20 h of milling. Expected the pure Si resonance line determined above, a broad structure in a range of −40 to 20 ppm is captured after 10 h of milling. Typically, a single resonance line is observed in the spectrum of β-SiC while up to three closely spaced resonances characterize the α-SiC materials in a range of −40 to 0 ppm.^58–62 Hence, at least two peaks can be distinguished in the broad area, suggesting the occurrence of two different phases, both incorporating Si as a component, at this milling stage. However, it should be argued that the large broadening may also result from the strain-induced distortion around the Si nucleus due to the MA effect. Moreover, the XRD results of C2 powder blends in this milling stage (20–40 h) showed the constituent elements, though with obviously amorphous structure, but could not reveal the presence of any alloying products.

Fig. 13 Solid-state ²⁹Si NMR spectra of carbon-lean C2 (a) and carbon-rich C9 (b) powder blends subjected to different hours of milling.

The spectra obtained in carbon-rich C9 powder blend are presented in Fig. 13(b), exhibiting a similar variation trend compared with the former. But it should be noted that the broad structure is found only for 5 h of milling, and its intensity increases rapidly with the extension of the milling time up to 20 h. At this milling stage, the resonance component from pure Si is greatly reduced in intensity, and the broad structure is dominant and tends to become symmetric at 20 h of milling, positioned close to the reference chemical shift.¹³ Thus, it can be inferred that the carbon concentration contribute to the broad structure formation during MA process. The broad structure formed at the initial milling stage and its subsequent narrowing for carbon-rich C9 powder blends can be interpreted as follows: amorphous extend solid solution phases are formed at the early stage milling (<5 h); continued milling leads to strain-induced nucleation of the stable coordination or environment of the constituent elements which are remained in the form of nanocrystallites or amorphous intermediate phases, as products until the final stage of milling (20 h and more).

In order to extract the various components of the broad structure, the spectra have been simulated, by using Gaussian–Lorentzian multi-peaks fitting procedure, and results are reported in Fig. 14 and Table 3. Four components (P1–P4) have been introduced. The spectra obtained in C7 powder blends after subjecting to 40 h of milling can be fitted as a sum of four components of resonance lines. The larger or dominant component has a linewidth site of about −15.2 ppm attributed to β-SiC (57.2%), while the smaller one has a site of 2.7 ppm corresponded to amorphous SiC (6.3%). For carbon-rich C9 blends, the spectra exhibit one peak at −35.1 ppm, which is assigned to SiCN₃ sites (site fraction = 24.3%). Besides, three signals at −26.1, −15.4 and 2.3 ppm are attributed to α-SiC (23.4%), β-SiC (40.6%) and amorphous SiC (11.7%) sites, respectively. It should be noted that the signal intensity, in case of carbon-lean C3 powder blends, is much smaller and weak enough than in case of carbon-rich ones, and in former case, the data collected time was three times longer than the latter. Hence, the resonance lines are not considered to be fitted in order to avoid errors. For the differences in enthalpy and free energy of the SiC polytypes are negligible, especially for the β-SiC and α-SiC,²³ thus, the reasonable factor leaded to the coexistence of the two phase, is the milling intensity and constituent composition which is also found in other MA materials.⁶³

Fig. 14 Solid-state ²⁹Si spectra in SiBCN system after 40 h of milling for carbon-lean C3, nominally stoichiometric C7 and carbon-rich C9 powder blends.

Table 3 Chemical shifts and site fraction of amorphous SiC, β-SiC, α-SiC and SiCN₃ sites derived from Gaussian–Lorentzian multi-peaks fitting from the ²⁹Si NMR spectra in Fig. 14

Sample codes	Amorphous SiC		β-SiC		α-SiC		SiCN3		β-SiC/α-SiC
	δ, ppm	Site fraction, %	δ, ppm	Site fraction, %	δ, ppm	Site fraction, %	δ, ppm	Site fraction, %
C9	2.3	11.7	−15.4	40.6	−26.1	23.4	−35.1	24.3	1.67
C7	2.7	6.3	−15.2	57.2	−25.4	18.9	−33.2	17.6	3.25

---

In summary, the NMR studies indentify the characteristic new environment of atomic coordination around Si atoms for both amorphous and nanocrystalline phases. The spectra appeared at higher chemical shift compared to that of pure Si at −80 ppm indicating the density of electrons at Fermi level at Si site is increased as a result of formation of these phases, e.g., amorphous SiC, α-SiC, β-SiC and SiCN_3. Moreover, the MA time act as a key role in the grain size reduction and amorphization, which leads to a large change in the atomic environment around Si atoms, resulting in a distribution in local field around the Si nuclei and giving rise to the enlarged linewidth. MA, such process is intimately connected with quantification phase transformation kinetics on both microscopic and macroscopic scales, thus it seems more easily for carbon-rich powder blends having chances to contact with Si nuclei.

Substantial defects could be produced at particle surface during milling, which can be regarded as nucleation sites due to its high energy offering driving force of nucleation. From the above results, we find that SiC have been formed in the early milling (5 h), seen in Fig. 11 and 13. During this milling time, part of silicon are amorphous while the graphite are a well amorphous state except for the free carbon (t-carbon and multiple graphene) of carbon-rich powder blends. So the formation of SiC may likely induced by heterogeneous nucleation as well as homogeneous nucleation.

The nucleation of SiC in powder blends can be treated as a kind of solid state reaction between silicon and carbon, referring to the changes of Gibbs free energy, interfacial energy and elastic energy. For heterogeneous nucleation, defects such as vacancy and dislocation in nanocrystalline silicon surface and grains boundaries can offer nucleation sites. The energy changes for heterogeneous or homogeneous nucleation can be described as:^4,64

ΔG_N = ε + ΔG_s + ΔG_V (5)

where ΔG_N is the total nucleation energy, ε is the elastic energy caused by the new phase mismatched with the matrix, ΔG_s is the surface energy changes between the new born phase and the disappeared phase, ΔG_V is the Gibbs energy change of the new formed phase and the disappeared one. For homogeneous nucleation, to simplify the analysis, the crystal nucleus is regarded as a small sphere. Thus, the ε and ΔG_s are expressed as follows:

ΔG_s = 4πr²σ (7)

where r is the radius of the SiC unit cell, ΔG_V is the difference in the Gibbs free energy between the amorphous matrix and crystal SiC, per unit volume, σ is the interfacial energy per unit area of the SiC and amorphous matrix. At certain milling conditions, the ΔG_V and σ are the constant.

According to classical nucleation and growth theory, the critical free energy required to form a sphere nucleus for the crystal, ΔG_N, is given by

For heterogeneous nucleation, the ΔG_s is much smaller than that of homogeneous nucleation. Hence, the heterogeneous nucleation can low the nucleation energy by decreasing the area of interface. For nucleation process of the SiBCN powders under plastic deformation and elastic strains, ΔG_N is positive at the certain milling conditions. Therefore, the critical free energy needed to form a nucleus for SiC phase is supported by the input mechanical energy. However, not all collisions are valid, so prolonging the milling time leads to the formation of SiC as well as amorphization, which is confirmed by the XRD and TEM observations. In fact, the formation and growth of a crystal in an amorphous matrix is almost depending on the atomic diffusion. Under normal pressure and low temperature, the formation and growth of SiC in amorphous matrix, which are strongly dependent on the long range diffusion of Si and C atoms, are very difficult due to their low diffusion rates; only the diffusion of atoms in a local area is possible. So at the early stage of milling (<5 h), not any SiC are formed due to the low volume ratio of the amorphous phase and crystalline starting materials, however, depending on the carbon concentration. Therefore, the repeated violent collision with prolonging time among the powder blends makes the volume of amorphous structure increase, which can cause annihilation of the crystalline volume and increase the structure defeats, such as vacancy and dislocation, thus reconstructing the atomic configuration. This will contribute a short-range diffusion of the Si and C atoms. Moreover, during the diffusion of the atoms some clusters with a similar atomic structure to a crystal may form, which could act as nucleation sites during the continuous milling process. Further increasing the milling time contributes to the growth of the based unites of nanocrystalline, such as α-SiC and β-SiC, growth along the close-packed planes which possess the lowest surface energy. Prolonging the milling time can favor the nucleation form and growth due to it increases the effective collisions (total collision frequency) among the atoms. Moreover, the annihilation of crystalline volume can change the local atomic arrangement, which may also promote the formation of some crystals. Therefore, the milling time at certain milling intensity (related to the rotation speed and ball-to-mass ratio) is very important for amorphization, nucleation and growth, which are strongly dependent on the atomic diffusion of quaternary MA SiBCN powders. Finally, the carbon contents play a key role in the phase transition and amorphization of SiBCN powder mixtures during MA process. For carbon-lean powder blends, i.e., C1 and C2, crystalline SiC are not found even after 40 h of milling, but the carbon-rich powder blends form crystalline SiC after subjecting to 20 h of milling. It thus suggests that, a proper ratio among these atoms is also required and the nucleation and growth of SiC during MA process is summarized in Fig. 15.

Fig. 15 Schematic showing the local atomic arrangement of SiC in an amorphous SiBCN matrix: (a) the atomic arrangement is random at early stage of MA (<5 h); (b) the local atomic arrangement changes to form some cluster with similar atomic structures of crystals (α-SiC and β-SiC units) at medium term of milling (5–20 h); (c) and (d) continuing milling (over 20 h) leads to the growth of nanocrystalline β-SiC and α-SiC in amorphous matrix, respectively (the growth close packed planes corresponded to β-SiC and α-SiC are (111) and (0002), respectively).

4. Conclusion

Within the present study, it appears that the preparation of varying proportions of completely amorphous or partially nanocrystalline microstructure dispersed matrix composite is viable by mechanical alloying of appropriate SiBCN powder blends. The grain sizes of the powder mixtures become to steadily refine due to the deformation at much high strain rate, cold welding, fragmentation and dynamic recrystallization during MA process; however, the carbon-rich powder blends reach to well solid state amorphization while the carbon-lean ones present remaining silicon in amorphous matrix upon 40 h of milling. The carbon concentration plays a key role on grain size reduction, solid state amorphization, formation of nanocrystalline SiC and free carbon structures. All the powder blends consist of dominantly amorphous matrix and a small amount of nanocrystallites after 40 h of milling, while the nanocrystallites are assigned to silicon and silicon carbide for carbon-lean and -rich powder blends, respectively. Continuing milling for carbon-rich powders leads to nothing different in structure changes but alter the volume fraction ratio of disordered carbon and multiple graphene in amorphous matrix. The species of chemical bonds changes for all investigated powder blends are similar, indicating the same chemical reaction process during mechanical alloying, but varies at the bonds content. The NMR analysis suggest that solid state amorphization in the present powder blends proceeds by developing a new atomic environment or coordination (there may develop more than one Si component as in carbon-rich C9) with the constituent elements during mechanical alloying that can be identified by XRD and HRTEM. Further research are in process to consolidate the as-milled powder blends and assess structure evolution and mechanical properties of the composites.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (NSFC, Grant number 51072041, 50902031 and 51021002). We want to show our appreciation to Prof. Jiancun Rao for his kind help on TEM analysis. The authors also want to show our appreciation to Richard M. Laine (university of Michigan, USA) for improving the language use.

References

1. R. Riedel, A. Kienzle, W. Dressler, L. Ruwisch, J. Bill and F. Aldinger, Nature, 1996, 382, 796.
2. N. V. R. Kumar, S. Prinz, Y. Cai, A. Zimmermann, F. Aldinger, F. Berger and K. Muller, Acta Mater., 2005, 53, 4567.
3. D. X. Li, Z. H. Yang, Z. B. Mao, D. C. Jia, D. L. Cai, B. Liang, X. M. Duan, P. G. He and J. C. Rao, RSC Adv., 2015, 5, 52194.
4. B. Liang, Z. H. Yang, Q. Q. Chen, S. J. Wang, D. X. Ming, D. C. Jia, Y. Zhou, K. Luo, D. L. Yu and Y. J. Tian, J. Am. Ceram. Soc., 2015, 98, 3788.
5. J. Kong, M. J. Wang, J. H. Zou and L. N. An, ACS Appl. Mater. Interfaces, 2015, 7, 6733.
6. Y. Gao, S. J. Widgeon, T. B. Tran, A. H. Tavakoli, G. Mera, S. Sen, R. Riedel and A. Navrotsky, Scr. Mater., 2013, 69, 347.
7. Y. Gao, G. Mera, H. Nguyen, K. Morita, H. J. Kleebe and R. Riedel, J. Eur. Ceram. Soc., 2012, 32, 1857.
8. R. Hauser, S. N. Borchard, R. Riedel, Y. H. Ikuhara and Y. Iwamoto, J. Ceram. Soc. Jpn., 2006, 114, 524.
9. J. Yuan, S. Hapis, H. Breitzake, Y. P. Xu, C. Fasel, K. H. J. Kleebe, G. Buntkowsky and R. Riedel, Inorg. Chem., 2014, 53, 10443.
10. C. Zhou, H. Min, L. Yang, M. Chen, Q. B. Wen and Z. J. Yu, J. Eur. Ceram. Soc., 2014, 34, 3579.
11. L. Gottardo, S. Bernard, C. Gervais, M. Weinmann and P. Miele, J. Mater. Chem., 2012, 22, 17923.
12. T. Ouyang, L. Gottardo, S. Bernard, R. Chiriac, C. Balan and P. Miele, J. Appl. Polym. Sci., 2013, 128, 248.
13. Z. H. Yang, D. C. Jia, X. M. Duan and Y. Zhou, J. Non-Cryst. Solids, 2010, 356, 326.
14. I. Manna, P. Nandi, B. Bandyopadhyay, K. Ghoshray and A. G. Ghoshray, Acta Mater., 2004, 52, 4133.
15. S. K. Pabi, D. Das, T. K. Mahapatra and I. Manna, Acta Mater., 1998, 46, 3501.
16. F. Delogu, Scr. Mater., 2012, 67, 340.
17. H. Y. Kim, M. Mizutani and S. Miyazaki, Acta Mater., 2009, 57, 1920.
18. P. A. Carvalho, I. Fonseca, M. T. Marques, J. B. Correia, A. Almeida and R. Viliar, Acta Mater., 2005, 53, 967.
19. C. Suryanarayana and A. A. Nasser, Prog. Mater. Sci., 2001, 58, 383.
20. F. Delogu and L. Takacs, Acta Mater., 2014, 80, 435.
21. P. F. Zhang, D. C. Jia, Z. H. Yang, X. M. Duan and Y. Zhou, Ceram. Int., 2012, 38, 6399.
22. P. F. Zhang, D. C. Jia, Z. H. Yang, X. M. Duan and Y. Zhou, Ceram. Int., 2013, 39, 1963.
23. Z. H. Yang, Y. Zhou, D. C. Ha and Q. C. Meng, Mater. Sci. Eng., A, 2008, 489, 187.
24. S. Zghal, M. Y. Hytch, J. P. Chevakier, R. Twesten, F. Wu and P. Bellon, Acta Mater., 2002, 50, 4695.
25. A. R. Puerta, E. E. Remsen, M. G. Bradley, W. Sherwood and L. G. Sneddon, Chem. Mater., 2003, 15, 478.
26. R. Ngoumeni-Yappi, C. Fasel, R. Riedel, V. Ischenko, E. Pippel, J. Woltersdorf and J. Clade, Chem. Mater., 2008, 20, 3601.
27. K. Su, E. E. Remsen, G. A. Zank and L. G. Sneddon, Chem. Mater., 1993, 5, 547.
28. X. Long, C. W. Shao, H. Wang and J. Wang, Dalton Trans., 2015, 35, 15463.
29. B. Jaschke, U. Klingebiel, R. Riedel, N. Doslik and R. Gadow, Appl. Organomet. Chem., 2000, 14, 671.
30. P. Colombo, G. Mera, R. Riedel and G. D. Soraru, J. Am. Ceram. Soc., 2010, 93, 1805.
31. V. Petrman, J. Houska, S. Kos, P. Calta and J. Vlcek, Acta Mater., 2011, 59, 2341.
32. N. B. Liao, W. Xue, H. M. Zhou and M. Zhang, RSC Adv., 2014, 3, 14458.
33. G. Mera, R. Riedel, F. Poli and K. Muller, J. Eur. Ceram. Soc., 2009, 29, 2883.
34. G. D. Soraru, S. Modena, P. Bettotti, G. Das, G. Mariotto and L. Pavesi, Appl. Phys. Lett., 2003, 83, 749.
35. T. Rouxel, G. Massouras and G. D. Soraru, J. Sol-Gel Sci. Technol., 1999, 14, 87.
36. A. Saha and R. A. Rishi, J. Am. Ceram. Soc., 2006, 89, 2188.
37. R. M. Prasad, G. Mera, K. Morita, M. Muller, H. J. Kleebe, A. Gurlo, C. Fasel and R. Riedel, J. Eur. Ceram. Soc., 2012, 32, 477.
38. A. Saha and R. Raj, J. Am. Ceram. Soc., 2005, 88, 232.
39. V. Tomar, M. Gan and H. S. Kim, J. Eur. Ceram. Soc., 2010, 30, 2223.
40. N. Resta, C. Kohler and H. R. Trebin, J. Am. Ceram. Soc., 2003, 86, 1409.
41. N. B. Liao, W. Xue and M. Zhang, J. Eur. Ceram. Soc., 2012, 32, 1275.
42. L. M. Reinold, M. Graczyk-Zajac, Y. Gao, G. Mera and R. Riedel, J. Power Sources, 2013, 236, 224.
43. J. Kaspar, G. Mera, A. P. Nowak, M. Graczyk-Zajac and R. Riedel, Electrochim. Acta, 2010, 56, 174.
44. D. Su, Y. L. Li, Y. Feng and J. Jin, J. Am. Ceram. Soc., 2009, 92, 2962.
45. S. Pasebani, I. Charit, Y. L. Wu, D. P. Butt and J. I. Cole, Acta Mater., 2013, 61, 5605.
46. K. W. Liu and F. Mucklich, Acta Mater., 2001, 49, 395.
47. P. Scherrer, Göttinger Nachrichten Math. Phys., 1918, 2, 98.
48. M. S. EI-Eskandarany, J. Saida and A. Inoue, Acta Mater., 2002, 50, 2725.
49. M. A. Pimenta, G. Dresselhaus, M. S. Dresselhaus, L. G. Cancado, A. Jorio and R. Saito, Phys. Chem. Chem. Phys., 2007, 9, 1276.
50. A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K. S. Novoselov, S. Roth and A. K. Geim, J. Phys. Chem. Lett., 2006, 97, 13831.
51. A. C. Ferrari, Solid State Commun., 2007, 146, 47.
52. A. C. Ferrari and J. Robertson, Phys. Rev. B: Condens. Matter Mater. Phys., 2000, 61, 14095.
53. O. Mondal, S. Mitra, M. Pal, A. Datta, S. Dhara and D. Chakravorty, Mater. Chem. Phys., 2015, 161, 123.
54. N. Pierard, A. Fonseca, J. F. Colomer, C. Bossuot, J. M. Benoit, G. Van Tendeloo, J. P. Pirard and J. B. Nagy, Carbon, 2004, 42, 1691.
55. A. G. Bannov, N. F. Uvarov, A. V. Ukhina, I. S. Chukanov, K. D. Dyukova and G. G. Kuvshinov, Carbon, 2012, 50, 1090.
56. J. Y. Huang, Acta Mater., 1999, 47, 1801.
57. A. Kukovecz, T. Kanyo, Z. Konya and I. Kiricsi, Carbon, 2005, 43, 994.
58. D. C. Apperley, R. K. Harris, G. L. Marshall and D. P. Thompson, J. Am. Ceram. Soc., 1991, 74, 777.
59. M. Narisawa, T. Iseki, Y. Katase, K. Okamura, K. Oka and T. Dohmaru, J. Am. Ceram. Soc., 2003, 86, 227.
60. K. R. Carduner, S. S. Shinozaki, M. J. Rokosz, C. R. Peters and T. J. Whalen, J. Am. Ceram. Soc., 1990, 73, 2281.
61. A. Bouclé Kassiba, J. Emery, I. V. Kityk, M. Makowska-Janusik, J. Sanetra, N. Herlin-Boime and M. Mayne, Phys. Lett. A, 2002, 302, 196.
62. S. Harrison, X. Xie, K. J. Jakubenas and H. L. Marcus, J. Am. Ceram. Soc., 1999, 82, 3221.
63. J. Y. Huang, Y. K. Wu and H. Q. Ye, Acta Mater., 1996, 44, 1201.
64. Y. J. Zhang, X. W. Yin, F. Ye and L. Kong, J. Eur. Ceram. Soc., 2014, 34, 1053.

---