Mechanistic Investigation of the Formation of Nickel Nanocrystallites Embedded in Amorphous Silicon Nitride Nanocomposites

Herein, we report the mechanistic investigation of the formation of nickel (Ni) nanocrystallites during the formation of amorphous silicon nitride at a temperature as low as 400 °C, using perhydropolysilazane (PHPS) as a preformed precursor and further coordinated by nickel chloride (NiCl2); thus, forming the non-noble transition metal (TM) as a potential catalyst and the support in an one-step process. It was demonstrated that NiCl2 catalyzed dehydrocoupling reactions between Si-H and N-H bonds in PHPS to afford ternary silylamino groups, which resulted in the formation of a nanocomposite precursor via complex formation: Ni(II) cation of NiCl2 coordinated the ternary silylamino ligands formed in situ. By monitoring intrinsic chemical reactions during the precursor pyrolysis under inert gas atmosphere, it was revealed that the Ni-N bond formed by a nucleophilic attack of the N atom on the Ni(II) cation center, followed by Ni nucleation below 300 °C, which was promoted by the decomposition of Ni nitride species. The latter was facilitated under the hydrogen-containing atmosphere generated by the NiCl2-catalyzed dehydrocoupling reaction. The increase of the temperature to 400 °C led to the formation of a covalently-bonded amorphous Si3N4 matrix surrounding Ni nanocrystallites.


Introduction
Transition metal (TM)-based compounds have been widely developed as heterogeneous catalysts in reactions involving energy-relevant transformations [1][2][3]. Growing interest in "sustainable growth", in industry and science as well as politics, still demands innovation to store, transmit, and/or convert any form of energy; thus, materials discovery appears to be a key element in the innovation cycle of energy conversion, transmission, and storage technologies [4].
Chemical routes toward ceramics are well-suited approaches for designing and synthesizing such highly active catalysts with the required ceramic robustness, especially in harsh environments. A very convenient precursor route is the polymer derived ceramics (PDCs) route, which allows fine control over the chemical composition of the final ceramic materials as well as their phase distribution and nanostructure [5][6][7]. In recent years, TM/Si-based (oxy-)carbide (SiC, SiOC) and carbonitride (SiCN) matrix nanocomposites, such as Ni/SiC, Ni/SiOC, Ni/SiCN(O), and M/SiCN (M = Pd, Ru, Pd 2 Ru, Cu, Ir, Ni, Pt, Co, and Fe) derived from metal-modified polycarbosilanes, polysiloxanes, and polysilazanes

Synthesis of Single Source Precursor for SiNiN System
Commercially available perhydropolysilazane (PHPS, NN120-20, 20 wt% in dibutyl ether solution, Sanwa Kagaku, Corp., Shizuoka, Japan) and nickel chloride (NiCl 2 , 98% purity, Sigma-Aldrich Japan, Tokyo, Japan) were used without further purification. All the reactions and handling of the chemicals for precursor syntheses were carried out under an inert atmosphere of pure argon (Ar), using standard Schlenk line and grove box techniques. The dibutyl ether solvent of as-received PHPS was substituted by superanhydrous toluene (99.5% purity, Wako Pure Chemical Industries, Ltd., Osaka, Japan) to be 20 wt% for the PHPS contents. Then, the synthesis of Ni-modified PHPS samples was performed at the atomic ratios of Ni in NiCl 2 to Si in PHPS (Ni/Si) = 0.05 and 0.2. The synthesized precursors were labeled as 0.05NiPHPS and 0.2NiPHPS, respectively. In a typical experiment, a 300 mL two-neck, round-bottom flask equipped with a magnetic stirrer was charged with the PHPS (20 wt% toluene solution, 21 mL, 93.1 mmol), then NiCl 2 (0.564 g, 4.44 mmol, Ni/Si = 0.05) and anhydrous toluene (20 mL) were added to the solution at room temperature. The resulting orange-colored solution was refluxed for 15 h under flowing Ar with stirring. During refluxing, the color of the solution changed from orange to brown-gray. After the reaction mixture was cooled down to room temperature, the solvent was removed under vacuum at 50 • C to give Ni-modified PHPS (0.05NiPHPS).

Conversion to Ni/Amorphous SiN Composite
The 0.05NiPHPS was subsequently pyrolyzed in a quartz tube furnace (Model FUW220PA, Advantec Toyo Kaisha, Ltd., Chiba Japan) under flowing N 2 (200 mL/min) up to specific temperatures of 200, 300, and 400 • C with the ramping rate of 5 • C min −1 . Then, each sample was quenched without holding time to afford Ni nanoparticle-dispersed amorphous silicon nitride composite sample labeled as Ni/SiNX (X represents the pyrolyzed temperature). As a reference, Ni-free PHPS was also pyrolyzed under the same manner to give amorphous silicon nitride sample denoted as SiNX. To study the Ni-nanoparticle formation within the PHPS-derived amorphous Si-N network, pyrolysis of as-received NiCl 2 alone was performed under flowing NH 3 or 10% H 2 /Ar (200 mL/min) with the ramping rate of 5 • C min −1 . When the furnace temperature reached 300 • C, the flowing gas was switched from the reductive gas to N 2 before quenching as mentioned above.

Characterization
The synthesized single source precursors were characterized by an attenuated total reflection flourier transform infra-red (ATR-FTIR) spectroscopy using FTIR spectrometer (FT/IR-4200IF, JASCO Corporation, Tokyo, Japan) with a diamond prism under an incidence angle of 45 • loaded in an ATR attachment (ATR PRO 550S-S/570S-H, JASCO Corporation, Tokyo, Japan). The ATR-FTIR spectra were collected at a resolution of 4 cm −1 with a cumulative number of 128 and the spectra were normalized by the strongest peak in the measured region.
To investigate the chemical bonding states, Raman spectra were recorded on pyrolyzed samples and as-synthesized samples (Model inVia, Renishaw, UK) using a single mode solid laser with wavelength 633 nm and power 25 mW for Raman excitation. The Raman spectra were also normalized by the strongest peak.
The thermal behavior up to 800 • C of as-received PHPS and 0.05NiPHPS was studied by simultaneous thermogravimetric (TG) and mass spectroscopic (MS) analyses (Model STA7200, Hitachi High Technologies Ltd., Tokyo, Japan) coupled with a quadrupole massspectrometry (Model JMS-Q1050GC, JEOL Ltd., Tokyo, Japan)]. The measurements were performed under flowing He (100 mL/min) with a heating rate of 10 • C min −1 .
The chemical composition of the pyrolyzed samples was calculated according to following equation: where the carbon (C) content was measured using a carbon analyzer (non-dispersive infrared method (Model CS844, LECO Corporation, MI, USA), and the oxygen (O) and nitrogen (N) contents were measured using an oxygen nitrogen hydrogen analyzer (inert gas fusion method, Model EMGA-930, HORIBA, Ltd., Kyoto, Japan), while the Si, Ni, and Cl contents were analyzed by the energy dispersive X-ray spectroscopy (EDS) mounted on a scanning electron microscope (SEM, Model JSM-6010LA, JEOL Ltd., Tokyo, Japan).
The powder X-ray diffraction (XRD) pattern analysis of the pyrolyzed samples was performed on a flat sample stage using Ni-filtered CuKα radiation (Model X'pert, Philips, Amsterdam, The Netherlands). XRD patterns were collected twice with two different scan rates: 4.7 and 0.8 • min −1 in the regions of 10 to 90 • and 35 to 55 • , respectively. The average crystallite size (L) of metal nickel formed in situ, was obtained from the Scherrer equation as follows: where λ is the X-ray wavelength in nanometer, K is a constant related to crystallite shape, generally applied as 0.9, and β is the peak width of the diffraction peak at half maximum height in radian. The β in this study was collected from the diffraction peak at 44.5 • measured at the scan rate of 0.8 • min −1 .
Transmission electron microscope (TEM) observations and high-angle annular darkfield scanning transmission electron microscope (HAADF-STEM) observations were performed on the pyrolyzed samples in a JEOL JEM-ARM200F operated at an accelerating voltage of 200 kV. The size of the electron probe was approximately 0.1 nm. The convergent angle and the detector collection angle were 22 mrad and 68-280 mrad, respectively. The Ni average size and size distribution histogram were obtained by measuring 50 particles in their longer direction using ImageJ program (NIH free software ver.1.52a).

Chemical Structure of Ni-Modified PHPS
To investigate the chemical bonding states of Ni-modified PHPS, ATR-FTIR spectroscopic analysis was performed on two samples, namely 0.05NiPHPS and 0.2NiPHPS, which display the Ni:Si atomic ratio of 0.05 and 0.2, respectively. As-received PHPSwho served as a reference-(black line in Figure 1a (Table S1). In addition, Ni-modified PHPS samples show shifts of νSi-N-Si band towards the lower wavenumber from 831 to 820-809 cm −1 , and the shifts increase consistently with the increase of the Ni/Si atomic ratio; thus, with the increase of the Ni content ( Figure 1b).
It has been reported that the catalytic formation of bonds between Group 14 and Group 15 elements, such as Si-N bonds resulting from the dehydrocoupling reaction between Si-H and N-H, could be conducted in the presence of various transition metal compounds at the temperature ranges of 20-90 • C [35]. Considering our synthesis conditions (i.e., precursor synthesis conducted in toluene at reflux (~110 • C)), the NiCl 2 -catalyzed dehydrocoupling reactions of Si-H/N-H are expected to proceed in our system to form further Si-N bonds.
Generally, the frequency of vibration (harmonic-oscillator) is inversely proportional to the square root of the mass of vibrating moiety. For instance, Lucovsky et al. reported that the Si-N stretching band attributed to (≡Si) 3 Si-N= was identified at a lower wavenumber in comparison with (≡Si) 2 HSi-N=, in which one of the silicon moieties (≡Si) was replaced with a H atom [36]. The shift of νSi-N-Si band in Ni-coordinated PHPS toward lower wavenumber is-in this analogy-due to the occurrence of dehydrocoupling reactions catalyzed by NiCl 2 leading to the formation of ternary silylamino group ((≡Si) 3 N:).
Based on these observations, the Ni-modified PHPS synthesized in this study can be characterized as follows: the ternary (≡Si) 3 N: group formed in situ via the NiCl 2 -catalyzed dehydrocoupling reaction between the (≡Si) 2 HN: group and the H-Si≡ bond is suggested as a possible ligand that coordinates on the Ni(II) cation of NiCl 2 to afford a complex structure such as a 4-coordinate Ni(II) complex (Scheme 1). It has been reported that the catalytic formation of bonds between Group 14 and Group 15 elements, such as Si-N bonds resulting from the dehydrocoupling reaction between Si-H and N-H, could be conducted in the presence of various transition metal compounds at the temperature ranges of 20-90 °C [35]. Considering our synthesis conditions (i.e., precursor synthesis conducted in toluene at reflux (~110 °C)), the NiCl2-catalyzed dehydrocoupling reactions of Si-H/N-H are expected to proceed in our system to form further Si-N bonds.
Generally, the frequency of vibration (harmonic-oscillator) is inversely proportional to the square root of the mass of vibrating moiety. For instance, Lucovsky et al. reported that the Si-N stretching band attributed to (≡Si)3Si-N= was identified at a lower wavenumber in comparison with (≡Si)2HSi-N=, in which one of the silicon moieties (≡Si) was replaced with a H atom [36]. The shift of νSi-N-Si band in Ni-coordinated PHPS toward lower wavenumber is-in this analogy-due to the occurrence of dehydrocoupling reactions catalyzed by NiCl2 leading to the formation of ternary silylamino group ((≡Si)3N:).
Based on these observations, the Ni-modified PHPS synthesized in this study can be characterized as follows: the ternary (≡Si)3N: group formed in situ via the NiCl2-catalyzed

Chemical Reaction during in Situ Nano Structuring Process
The occurrence of these mechanisms affects the ceramic conversion of PHPS. Therefore, the thermal behavior of the 0.05NiPHPS sample has been monitored using TG-MS analysis under flowing inert gas (He). The TG-curve and total ion current chromatogram (TICC) (Figure 2a) indicate that the weight loss contentiously occurs at 150 to 800 °C associated with the evolution of gas species in three temperature ranges as revealed by the simultaneous MS analyses (Figure 2b-e). They show that the weight loss up to 200 °C is

Chemical Reaction during in Situ Nano Structuring Process
The occurrence of these mechanisms affects the ceramic conversion of PHPS. Therefore, the thermal behavior of the 0.05NiPHPS sample has been monitored using TG-MS analysis under flowing inert gas (He). The TG-curve and total ion current chromatogram (TICC) (Figure 2a) indicate that the weight loss contentiously occurs at 150 to 800 • C associated with the evolution of gas species in three temperature ranges as revealed by the simultaneous MS analyses (Figure 2b-e). They show that the weight loss up to 200 • C is due to the evaporation of residual reaction solvent as the evolved species are attributed to toluene (C 6

Chemical Composition of Nanocomposites
The XRD analysis reveals that Ni(II) keeps the initial state as dichloride at 200 °C (Ni/SiN200 sample) since the diffraction pattern composed of the three peaks (2θ = 15.1, 32.8, 52.6°) completely matches the three main peaks detected for the NiCl2 used in this study ( Figure S2) The release of NH 3 is due to the transamination reaction which is one of the typical cross-linking reactions occurring in polysilazanes [37,38]. Interestingly, the transamination reaction in 0.05NiPHPS sample, i.e., within the cross-linked PHPS polymer network by the NiCl 2 modification (Figure 2), was found to proceed approximately 50 • C lower than that in the as-received PHPS sample ( Figure S1).
The identification of monochlorosilanes and HCl is most probably due to the reaction between the Si center and/or NH units of PHPS with NiCl 2 that could form a nitride phase.
To investigate the transformation of Ni-modified PHPS samples into ceramics in more depth, we selected the 0.05NiPHPS sample to be pyrolyzed at intermediate temperatures (200, 300, and 400 • C) under an inert gas (N 2 ) atmosphere. Then, the pyrolyzed intermediates labeled Ni/SiNX -with X the temperature at which the sample has been exposedhave been characterized by elemental analyses, X-ray diffraction, and Raman spectroscopy. Table 1 lists the chemical compositions of the pyrolyzed samples (Ni/SiNX) along with those of PHPS-derived amorphous silicon nitride samples (SiNX) isolated at the same temperatures. At first, the Ni:Si ratio fixed in the early stage of the process, at the polymer level, increased during the pyrolysis indicating the release of Si-containing species, such as monochlorosilane, as identified by TG-MS (Figure 2). To support this observation, the N/Si and Cl/Si atomic ratios in Ni/SiNX samples decrease with the pyrolysis temperature increase, which is consistent with the evolution of NH 3 and monochlorosilane, as shown during TG-MS analysis (Figure 2).

Chemical Composition of Nanocomposites
The XRD analysis reveals that Ni(II) keeps the initial state as dichloride at 200 • C (Ni/SiN200 sample) since the diffraction pattern composed of the three peaks (2θ = 15.1, 32.8, 52.6 • ) completely matches the three main peaks detected for the NiCl 2 used in this study ( Figure S2 (Figure 3a,b). It should be noted that Ni nanocrystallite cannot be formed by pyrolysis of pure NiCl 2 powder under reductive H 2 or NH 3 conditions, as well as in inert N 2 conditions (Figures S2 and S5). Therefore, it is strongly suggested that the Ni nanocrystallites are formed in situ through chemical reactions of PHPS with NiCl 2 leading to the formation of a covalently-bonded amorphous Si 3 N 4 matrix surrounding Ni nanocrystallites.   The average crystallite size was calculated for the Ni (111) plane by using the Scherrer equation, and the initial average crystallite size at 300 °C of 18.4 nm was evaluated for the Ni/SiN300 sample, then the crystallite size reached 18.5 nm at 400 °C (Ni/SiN400 sample).

Mechanistic Investigation of the in Situ Formation of Nanocomposites
Interestingly, FTIR spectroscopy and TG-MS analysis suggested that NiCl2 may produce a Ni nitride as a consequence of potential interactions with ((≡Si)3N:) or N-H bonds; a Ni nitride compound was not detected as an intermediate by the XRD analysis ( Figure  3). Within this context, we conducted further investigation focused on the 0.05NiPHPS sample, the intermediate states (Ni/SiN200 and Ni/SiN300 samples) and the Ni/a-SiN nanocomposite (Ni/SiN400 sample) and obtained a clear understanding of the in situ formation pathway by the Raman spectroscopic analysis (Figure 4). The average crystallite size was calculated for the Ni (111) plane by using the Scherrer equation, and the initial average crystallite size at 300 • C of 18.4 nm was evaluated for the Ni/SiN300 sample, then the crystallite size reached 18.5 nm at 400 • C (Ni/SiN400 sample).

Mechanistic Investigation of the In Situ Formation of Nanocomposites
Interestingly, FTIR spectroscopy and TG-MS analysis suggested that NiCl 2 may produce a Ni nitride as a consequence of potential interactions with ((≡Si) 3 N:) or N-H bonds; a Ni nitride compound was not detected as an intermediate by the XRD analysis ( Figure 3). Within this context, we conducted further investigation focused on the 0.05NiPHPS sample, the intermediate states (Ni/SiN200 and Ni/SiN300 samples) and the Ni/a-SiN nanocomposite (Ni/SiN400 sample) and obtained a clear understanding of the in situ formation pathway by the Raman spectroscopic analysis (Figure 4).
The spectra have been recorded for the samples at polymer state as well as those listed in Table 1. At polymer state (0.05NiPHPS sample), the spectrum presents two sharp peaks at 169.4 cm −1 and 265.1 cm −1 , which are slightly red-shifted against those of asreceived NiCl 2 at 171.0 cm −1 and 266.6 cm −1 shown in Figure S2 and one additional distinct peak centered at 503.3 cm −1 assigned to a Si-Si bond [39,40]. After pyrolysis at 200 • C (Ni/SiN200 sample), the two peaks due to NiCl 2 further red-shifted to 168.8 cm −1 and 263.9 cm −1 , respectively.
Feshin et al. reported that the coordination of N(CH 3 ) 3 to GeCl 4 causes the electron density redistribution to elongate the Ge-Cl bond length and increase the polarity of the Ge-Cl bond [41]. It is generally accepted that the red-shifts of Raman peaks can be interpreted as a decrease in the force constant of chemical bonds, since the frequency is proportional to the square root of the force constant [42]. Therefore, it could be analogically assumed that the N:→Ni dative bond formation makes the Ni-Cl bond of NiCl 2 weaker and more polar and, hence, more reactive. The spectra have been recorded for the samples at polymer state as well as those listed in Table 1. At polymer state (0.05NiPHPS sample), the spectrum presents two sharp peaks at 169.4 cm −1 and 265.1 cm −1 , which are slightly red-shifted against those of as-received NiCl2 at 171.0 cm −1 and 266.6 cm −1 shown in Figure S2 and one additional distinct peak centered at 503.3 cm −1 assigned to a Si-Si bond [39,40]. After pyrolysis at 200 °C (Ni/SiN200 sample), the two peaks due to NiCl2 further red-shifted to 168.8 cm −1 and 263.9 cm −1 , respectively.
Feshin et al. reported that the coordination of N(CH3)3 to GeCl4 causes the electron density redistribution to elongate the Ge-Cl bond length and increase the polarity of the Ge-Cl bond [41]. It is generally accepted that the red-shifts of Raman peaks can be interpreted as a decrease in the force constant of chemical bonds, since the frequency is proportional to the square root of the force constant [42]. Therefore, it could be analogically assumed that the N:→Ni dative bond formation makes the Ni-Cl bond of NiCl2 weaker and more polar and, hence, more reactive.
Meanwhile, the peak at 503.3 cm −1 became broader in the Ni/SiN200 sample. The deconvolution of the broad signal yields the Si-Si bond peak centered at 498.7 cm −1 (peak 1) and an additional one at 514.8 cm −1 (peak 2) assigned to Ni-N bond in Ni3N [43]. At 300 °C (Ni/SiN300 sample), the two peaks due to NiCl2 disappear and the peak area ratio of Meanwhile, the peak at 503.3 cm −1 became broader in the Ni/SiN200 sample. The deconvolution of the broad signal yields the Si-Si bond peak centered at 498.7 cm −1 (peak 1) and an additional one at 514.8 cm −1 (peak 2) assigned to Ni-N bond in Ni 3 N [43]. At 300 • C (Ni/SiN300 sample), the two peaks due to NiCl 2 disappear and the peak area ratio of peak 2/peak 1 decreases from 1.58 (Ni/SiN200 sample) to 1.02 (Ni/SiN300 sample), then at 400 • C, the peak 2, attributed to the Ni-N bond, disappears and the Ni/SiN400 sample presents the peak 1, due to the Si-Si bond, as a single peak (Figure 4a). It should be noted that Raman spectrum of the Ni/SiN400 sample in the wider Raman shift range is free from typical peaks attributed to amorphous silicon oxides such as Si-O-Si (820 cm −1 ) and Si-(OH) x (920 and 1079 cm −1 ) [44], while other peaks are found to correspond to the Raman peaks detected for SiN400 and thus attributed to PHPS-derived Si-N matrix networks ( Figure S4). These results are consistent with the result of elemental analysis, showing that oxygen impurity is very low in both samples.
Under the present pyrolysis condition up to 400 • C, it is important to note that NiCl 2 keeps its initial state without thermal decomposition ( Figure S5). Thus, the results obtained by the TG-MS, XRD, and Raman spectroscopic analyses reveal that the reaction of the activated NiCl 2 with PHPS begins to start below 200 • C and the subsequent metal Ni nucleation and crystallization occurs between 200 to 300 • C. Moreover, it is strongly suggested that the in situ formation of metal Ni proceeds via the formation of Ni nitride species as intermediates. The Raman spectroscopic analysis reveals that Ni 3 N is one possible intermediate phase (Figure 4a), which is well consistent with the reported thermodynamic data: Ni 3 N has the lowest formation energy of NiN x (x = 0 to 1/3) [45] and is a thermodynamically favorable phase at lower temperatures in the phase diagram for Ni-N system [46].
Regarding the Ni-N bond formation, since the temperature range detected by the Raman spectroscopic analysis was as low as 200 to 300 • C, the direct nitridation of Ni by atmospheric N 2 is excluded. Therefore, the Ni-N bond is intrinsically formed by the chemical reaction of NiCl 2 with PHPS as the nitrogen source via ternary silylamino groups.
It should be noted that the formation of Si-Si bond is reasonable since the N/Si atomic ratio of all the Ni/SiNX samples is lower than 1 (Table 1); indeed, all the SiNX samples with the N/Si atomic ratio <1 (Table 1) exhibited the typical Raman peak attributed to Si-Si bond (Figure 4b).
Based on these results, reactions during the polymer to metal Ni/a-SiN conversion process can be suggested as follows (Scheme 2): After the chemical modification of PHPS with NiCl 2 to afford Ni complexes (1), Ni-N bond formation would proceed via the nucleophilic attack of N atom of the silylamino group on the center Ni(II) cation to replace Cl. Since the N:→Ni dative bond has been already formed, the N atom would be easily accessible to Ni(II), and thus a S N 2 reaction via the transition state (2) is proposed. Subsequently, the nucleophilic attack of the released chloride anion (Cl -) on other electrophiles, such as Si center of PHPS moieties (3), would proceed, leading to the evolution of monochlorosilane, as detected by the TG-MS analysis. Then, the formation of Ni nitride species (5) could proceed through the thermal decomposition of in situ preformed species, having Ni-N bond (4). . The resulting species with a Ni-N bond (4) decompose to give nickel nitride species (a possible one is Ni3N) (5). Subsequent decomposition of (5) is facilitated by an in situ-formed H2 leads to the formation of Ni (0) (6), followed by the nucleation and crystallization of Ni (7) at 200 to 300 °C.
In addition, the transamination reaction leads to the formation of the byproducts, NH4NiCl3 and NH4Cl, as shown in Figure 3, resulting from the nucleation reactions of the gaseous NH3 formed in situ with Ni complexes and the NH3 with HCl formed in situ, respectively. Then, above 300 °C, the byproducts thermally decompose to yield gaseous NH3 and HCl, while the regenerated NiCl2 species would participate in the Ni-N bond formation reaction as shown in Scheme 2.
To assess the phase nanostructure of the metal Ni/amorphous silicon nitride composites in more detail, TEM observation was performed on the Ni/SiN400 sample. As shown in Figure 5a, many dark dots correspond to metallic Ni crystallites embedded within amorphous silicon nitride matrix. In fact, the fringe spacing calculated from Figure 5b   The resulting species with a Ni-N bond (4) decompose to give nickel nitride species (a possible one is Ni 3 N) (5). Subsequent decomposition of (5) is facilitated by an in situ-formed H 2 leads to the formation of Ni (0) (6), followed by the nucleation and crystallization of Ni (7) at 200 to 300 • C.
In general, under N 2 or NH 3 atmosphere, Ni 3 N, which is considered as the possible intermediate in this work, starts decomposing thermally at about 400 • C, accompanied by the formation of metallic Ni and gaseous nitrogen, while the Ni 3 N decomposition temperature has been found to be as low as 157 • C due to its chemical fragility under an H 2 atmosphere [47,48]. Thus, it could be speculated that the local hydrogen partial pressure around the Ni nitride species embedded within the polymer-derived amorphous Si-N network reached a sufficiently high level to promote the immediate decomposition of the Ni nitride species (5), followed by the nucleation and subsequent crystallization of metal Ni at 200 to 300 • C, shown as (6) and (7) in Scheme 2. As previously mentioned, NiCl 2 exhibited high catalytic activity to facilitate Si-H/N-H dehydrocoupling reactions in PHPS moieties. It should be noted that this NiCl 2 -catalyzed dehydrocoupling reaction proceeded contentiously during pyrolysis, which facilitated the polymer to the ceramic conversion of PHPS up to 400 • C.
In addition, the transamination reaction leads to the formation of the byproducts, NH 4 NiCl 3 and NH 4 Cl, as shown in Figure 3, resulting from the nucleation reactions of the gaseous NH 3 formed in situ with Ni complexes and the NH 3 with HCl formed in situ, respectively. Then, above 300 • C, the byproducts thermally decompose to yield gaseous NH 3 and HCl, while the regenerated NiCl 2 species would participate in the Ni-N bond formation reaction as shown in Scheme 2.
To assess the phase nanostructure of the metal Ni/amorphous silicon nitride composites in more detail, TEM observation was performed on the Ni/SiN400 sample. As shown in Figure 5a, many dark dots correspond to metallic Ni crystallites embedded within amorphous silicon nitride matrix. In fact, the fringe spacing calculated from Figure 5b

Conclusions
In this study, we reported on the in situ formation of metallic nickel nanocrystallites, with an average Ni size of 18.0 nm, finely dispersed within a polymer-derived amorphous silicon nitride matrix at pyrolysis temperatures as low as 400 °C. The results can be summarized as follows: 1. ATR-FTIR spectroscopic analysis revealed that NiCl2 had high catalytic activity for dehydrocoupling reactions between Si-H and N-H of PHPS under the present A brief size distribution analysis by measuring the size of 50 Ni nanocrystallites revealed a relatively wide range of size distribution from 7.32 to 43.4 nm; however, most of the Ni crystallites were below 25 nm in size, and the mean size was found as 18.0 nm. This is confirmed by the high-resolution image (Figure 5c) which reveals the presence of a nanocrystal with a lattice spacing of 0.202 nm, corresponding to the d-spacing of the lattice plane of the Ni structure, i.e., the (111) direction of the fcc cubic Ni structure which agrees with the phase identified, as well in the X-ray diffractogram (Figure 3), as seen in the SAED pattern (Figure 5b) of the same sample.

Conclusions
In this study, we reported on the in situ formation of metallic nickel nanocrystallites, with an average Ni size of 18.0 nm, finely dispersed within a polymer-derived amorphous silicon nitride matrix at pyrolysis temperatures as low as 400 • C. The results can be summarized as follows: 1.
ATR-FTIR spectroscopic analysis revealed that NiCl 2 had high catalytic activity for dehydrocoupling reactions between Si-H and N-H of PHPS under the present precursor synthesis condition conducted in toluene at reflux (~110 • C), leading to the formation of ternary silylamino groups ((≡Si) 3 N:). Consequently, the ternary silylamino group coordinated the Ni(II) cation of NiCl 2 to afford a complex such as the 4-coordinated Ni(II) complex.

2.
TG-MS and Raman spectroscopic analyses revealed that the Ni-N bond in Ni nitride species is intrinsically formed via the S N 2 reaction of the preformed 4-coordinated Ni(II) complex at 200 • C: the nucleophilic attack of the N atom of the silylamino group on the center Ni(II) cation and simultaneous elimination of Clas a leaving group. Subsequently, the nucleophilic attack of the released Clon other electrophiles, such as Si center of PHPS moieties, would proceed accompanied by the evolution of monochlorosilane, while the N-bonded Ni species subsequently decomposed to give the Ni nitride species. 3.
XRD, ATR-FTIR, and Raman spectroscopic analyses revealed that Ni nanocrystallites started to form at temperatures as low as 200 to 300 • C through the decomposition reaction of the in situ formed Ni nitride species facilitated by H 2 , which was generated through the NiCl 2 -catalyzed dehydrocoupling reaction of PHPS. In addition, this NiCl 2 -catalyzed dehydrocoupling reaction of PHPS accelerated the polymer to the ceramic conversion of PHPS up to 400 • C.
Therefore, it can be concluded that the low temperature in situ formation of Ni/amorphous silicon nitride composites was governed by the synergistic effect of the following two reactions attributed to the high catalytic performance of NiCl 2 for dehydrocoupling reaction of PHPS: (i) the Ni(II) complex formation allows in situ formation of the Ni nitride species as an intermediate below 200 • C and (ii) the immediate decomposition of the Ni nitride species promotes Ni nucleation at 200 to 300 • C.
Based on our results obtained in this study, it is highly expected that the ability to control material properties such as the size of Ni nanocrystallites, chemical composition, and pore size distribution by designing polymeric precursors can be achieved.
Thus, further investigation on the molecular structure of single source precursors for the ideal metal/nitride nanocomposites is in progress and the nanocomposites are expected to be applied as a new, highly active catalyst for important reactions, such as the hydrogenation of CO 2 , which is a promising way to solve the environmental and energy affairs caused by CO 2 emissions. Such a study is under investigation and will be published separately.