《复合材料 Composites》课程教学资源（学习资料）第二章 增强体_Growth of One-Dimensional Nanostructures in Porous Polymer?Derived Ceramics by Catalyst-Assisted Pyrolysis. Part I：Iron Catalyst

C)2009 The American Ceramic Society urna Growth of One-Dimensional Nanostructures in Porous Polymer- Derived Ceramics by Catalyst-Assisted Pyrolysis. Part I: Iron Catalyst Cekdar Vakifahmetoglu, Eckhard Pippel. Jorg Woltersdorf, and Paolo Colombo* dIpartimento di Ingegneria Meccanica, Settore Materiali, University of Padova, 35131 Padova, Italy Max-Planck-Institut fur Mikrostrukturphysik, D-06120 Halle, Germany Department of Materials Science and Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802 The presence of Fe Clz catalyst enabled the growth of one- Si3 N4 nanowires/nanobelts, using FeCl2 as catalyst. A similar dimensional nanostructures directly during the pyrolysis of was also used to synthesize powders containing Sic duced from a polysiloxane pree nanorods, SiNg nanobelt ngle-crystalline Si3N4 nano- ramic polymer with the aid of a gas-generating porogen. Either wires with or without2aluminum doping, and finally silicon silicon nitride or silicon carbide nanowires were formed, with a doped boron nitride (Bn)nanotubes having a bamboo struc- length of several micrometers, depending on the processing at- ture. Furthermore, pyrolysis of pCs was shown to yield ceramic mosphere. Increasing the pyrolysis temperature caused an in- powders containing SiC whiskers, when nickel ferrite was incor- crease in the length and the amount of nanostructures produced. porated in the precursor The remaining matrix consisted of an incompletely crystallized owever, very few studies have so far investigated pyrolysis/or containing SiC crystals and either graphitic(N3 SiO-C phase mation of nanostructures during pyrolysis, via CAP. rphous carbon(Ar pyrolysis). X-ray diffrac- pores of preceramic polymer-derived monoliths tion data and high-resolution transmission electron microscopy al. pyrolyzed nickel acetate-containing poly(methyl- g: vestigations combined with electron energy loss and energy- phenyl)silsesquioxane( PMPS)and observed the formation of growth mechanisms for the nanowires, which depended on the the polymer-to-ceramic conversion of the polymeric matrix. The lySIS atmosphere (gas phase reaction for N2 pyrolysis; authors described these pores as ""catalytic microreactors, and vapor-Hiquid-solid for Ar pyrolysis). afterwards showed the in situ formation of CNTs. when silicon as added, and of Sic/siOz nanowires when nickel acetate and silicon were incorporated together into the PMPs precursor. L Introduction he formation of nickel silicide tips was observed for both types tubes, nanowires, nanobelts, etc. )has been the subject of a posed 2, x cess as a possible growth mechanism was,pro- steadily growing interest owing to their unique and often superior and enabled to produce nanostructures, according to the TEM properties compared with their bulk and/or microscale counter- and scanning electron microscope(SEM) images reported in parts, and a great number of manufacturing techniques have those studies, the nanofibers/ wires/tubes produced in the pores been developed for the production of these materials. In partic ular, the use of preceramic polymers is very promising. due to the SiC nanowires having spherical particles on their tips was also great tailorability of their structures on a molecular scale and ease observed in the channels of porous SiC ceramics fabricated from of processing. It has been shown that various types of nanostruc- a-SiC powder and PCS precursor as a binder.29, 30 Catalyst pa tures, such as whiskers, nanotubes, -and nanocables/wires-1 ticles for the growth of the nanowires were considered to fibers- of different compositions can be produced directly originate from unwanted iron impurity in the starting SiC pow- from preceramic polymers, without the use of any transition der, suggesting a similar VLS mechanism of growth. In this metal additives as catalyst. Recently, great progress has been case also, the amount of nanostructures produced was very lin made in the production of nanostructures from preceramic poly- ited. Instead, Yoon et al. very recently reported the formation mers(mostly polysilazanes or polycarbosilanes(PCSs) by apply of highly aligned macroporous SiC ceramics decorated with ho- ing catalyst-assisted pyrolysis(CAP), leading to improved yield mogeneously distributed SiC nanowires, produced by unidirec- and the formation of varied morphologies. For example, carbon tional freeze casting of SiC/camphene slurries with different amounts of the PCs precursor. Iron originated again from the nanoparticles,or from a borazine-based precursor including arting SiC powder, and was found in the tips of the nanowires. nickel as catalyst, while Sic/siOz core/shell nanocables were The presence of nanostructures led to a remarkable increase in the specific surface area(SSA), from 30 to 86 m/g, when the Likewise, a polysilazane was used to produce amorphous silicon initial PCS nt was varied 5-20 wt%. due to enhanced bonitride(SiCN) powder with in sitte-grov growth of the SiC nanowires The present paper further investigates the H-J. Klecbe--contributing editor tionally building nanostructures in the pores by one-pot in situ CAP of a commercially avai In a companion work(Part ID), a different type of catalyst 出品 (CoCl2) will be discussed, and the main characteristics of all po- rous components, including the SSa values, will be reported as a function of processing conditions and catalyst type. The general 959

Growth of One-Dimensional Nanostructures in Porous Polymer￾Derived Ceramics by Catalyst-Assisted Pyrolysis. Part I: Iron Catalyst Cekdar Vakifahmetoglu,w,z Eckhard Pippel,y Jo¨rg Woltersdorf,y and Paolo Colombo* ,z,z z Dipartimento di Ingegneria Meccanica, Settore Materiali, University of Padova, 35131 Padova, Italy y Max-Planck-Institut fu¨r Mikrostrukturphysik, D-06120 Halle, Germany z Department of Materials Science and Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802 The presence of FeCl2 catalyst enabled the growth of one￾dimensional nanostructures directly during the pyrolysis of highly porous monoliths, produced from a polysiloxane prece￾ramic polymer with the aid of a gas-generating porogen. Either silicon nitride or silicon carbide nanowires were formed, with a length of several micrometers, depending on the processing at￾mosphere. Increasing the pyrolysis temperature caused an in￾crease in the length and the amount of nanostructures produced. The remaining matrix consisted of an incompletely crystallized Si–O–C phase, containing SiC crystals and either graphitic (N2 pyrolysis) or amorphous carbon (Ar pyrolysis). X-ray diffrac￾tion data and high-resolution transmission electron microscopy investigations combined with electron energy loss and energy￾dispersive X-ray spectroscopy methods enabled to ascertain the growth mechanisms for the nanowires, which depended on the pyrolysis atmosphere (gas phase reaction for N2 pyrolysis; vapor–liquid–solid for Ar pyrolysis). I. Introduction THE synthesis of one-dimensional (1D) nanostructures (nano￾tubes, nanowires, nanobelts, etc.) has been the subject of a steadily growing interest owing to their unique and often superior properties compared with their bulk and/or microscale counter￾parts, and a great number of manufacturing techniques have been developed for the production of these materials.1 In partic￾ular, the use of preceramic polymers is very promising, due to the great tailorability of their structures on a molecular scale and ease of processing. It has been shown that various types of nanostruc￾tures, such as whiskers,2 nanotubes,3–5 and nanocables6 /wires7–9/ fibers10–12 of different compositions can be produced directly from preceramic polymers, without the use of any transition metal additives as catalyst. Recently, great progress has been made in the production of nanostructures from preceramic poly￾mers (mostly polysilazanes or polycarbosilanes (PCSs)) by apply￾ing catalyst-assisted pyrolysis (CAP), leading to improved yield and the formation of varied morphologies. For example, carbon nanotubes (CNTs) were synthesized from a PCS containing iron nanoparticles,13 or from a borazine-based precursor including nickel as catalyst,14 while SiC/SiO2 core/shell nanocables were produced using poly(dimethylsiloxane) coupled with ferrocene.15 Likewise, a polysilazane was used to produce amorphous silicon carbonitride (SiCN) powder with in situ-grown single-crystal Si3N4 nanowires/nanobelts, using FeCl2 as catalyst.16 A similar methodology was also used to synthesize powders containing SiC nanorods,17 Si3N4 nanobelts,18–20 single-crystalline Si3N4 nano￾wires with21 or without22 aluminum doping, and finally silicon￾doped boron nitride (BN) nanotubes having a bamboo struc￾ture.23 Furthermore, pyrolysis of PCS was shown to yield ceramic powders containing SiC whiskers, when nickel ferrite was incor￾porated in the precursor.24,25 However, very few studies have so far investigated the for￾mation of nanostructures during pyrolysis, via CAP, in the pores of preceramic polymer-derived monoliths. Scheffler et al. 26 pyrolyzed nickel acetate-containing poly(methyl– phenyl)silsesquioxane (PMPS) and observed the formation of multiwall CNTs, only within the pores that were formed during the polymer-to-ceramic conversion of the polymeric matrix. The authors described these pores as ‘‘catalytic microreactors,’’ and afterwards showed the in situ formation of CNTs, when silicon was added, and of SiC/SiO2 nanowires when nickel acetate and silicon were incorporated together into the PMPS precursor.27 The formation of nickel silicide tips was observed for both types of systems, and therefore, a well-known vapor–liquid–solid (VLS) process as a possible growth mechanism was pro￾posed.27,28 Although, the processing technique used was simple and enabled to produce nanostructures, according to the TEM and scanning electron microscope (SEM) images reported in those studies, the nanofibers/wires/tubes produced in the pores of the pyrolyzed monoliths were very few. In situ formation of b￾SiC nanowires having spherical particles on their tips was also observed in the channels of porous SiC ceramics fabricated from a-SiC powder and PCS precursor as a binder.29,30 Catalyst par￾ticles for the growth of the nanowires were considered to originate from unwanted iron impurity in the starting SiC pow￾der, suggesting a similar VLS mechanism of growth.29 In this case also, the amount of nanostructures produced was very lim￾ited. Instead, Yoon et al. 31 very recently reported the formation of highly aligned macroporous SiC ceramics decorated with ho￾mogeneously distributed SiC nanowires, produced by unidirec￾tional freeze casting of SiC/camphene slurries with different amounts of the PCS precursor. Iron originated again from the starting SiC powder, and was found in the tips of the nanowires. The presence of nanostructures led to a remarkable increase in the specific surface area (SSA), from 30 to 86 m2 /g, when the initial PCS content was varied 5–20 wt%, due to enhanced growth of the SiC nanowires. The present paper further investigates the possibility of inten￾tionally building nanostructures in the pores of cellular ceramics by one-pot in situ CAP of a commercially available polysiloxane. In a companion work (Part II), a different type of catalyst (CoCl2) will be discussed, and the main characteristics of all po￾rous components, including the SSA values, will be reported as a function of processing conditions and catalyst type. The general H.-J. Kleebe—contributing editor *Member, The American Ceramic Society. w Author to whom correspondence should be addressed. e-mail: cekdar@unipd.it Manuscript No. 26277. Received May 22, 2009; approved September 16, 2009. Journal J. Am. Ceram. Soc., 93 [4] 959–968 (2010) DOI: 10.1111/j.1551-2916.2009.03448.x r 2009 The American Ceramic Society 959

960 Journal of the American Ceramic Society-Vakifahmetoglu et al Vol 93. No 4 aim of this development is the production of macroporous ce- tached to a confocal microscope(x 50 objective)using the 633 ramic components possessing high SSA values, for gas adsorp- Im line of a He-Ne laser as the excitation wavelength. Samples tion, catalyst support, and pollutant removal applications were ground and the powders were used for analysis, using a low aser power(5%) II. Experimental Procedure IlL. Results and discussio Cellular ceramics were produced by using a commercially avail- able PMPs preceramic polymer(H44, Wacker Chemie AG, (1) Foaming, Crosslinking, and Thermal Analysi Burghausen, Germany), denoted PMPS in the remainder of the Many studies can be found in the literature concerning foaming text. Solid PMPS 96 wt% was ball milled together with I wt% of various polymeric systems using ADA. ADA has also azodicarbonamide(ADA 97%, Sigma-Aldrich, St Louis, Mo) been used to produce macrocellular Sioc and SicN3ceram acting as a physical blowing agent, and 3 wt% of a transition ics from preceramic polymers. It was shown that the decompo- mixed batch was then l or the groIN ure, Sigma-Aldrich), serv- sition process of this porogen can be controlled by varying its metal halide powder(FeCl2, >98% nsferred to an oven for foaming and presence and concentration of an activator.34.3Activators, in- thermal crosslinking(I h at 90C and then 5 h at 250.C: 2C cluding transition metal compounds in particular reduce the de- min heating rate). The porous thermoset monoliths thus ob- omposition temperature of ADa to values as low as 150C. tained were then individually pyrolyzed under N2 or Ar(both After heating at 250C. cellular monoliths with a high amount of 99.999% pure)in an alumina tube furnace (2 h at the required open porosity were obtained. The presence of porosity can be emperature, in the range 1250%-1400C: 2C/min heating and attributed both to the continued release of volatiles oligomers during the PmPs curing and to the good match between the Thermal analysis(TG-DTA)measurements were carried out temperature of decomposition of the blowing agent and the under Ar or N,(Netzsch STA 429, Selb, Germany: 2C/mil heating rate)on the already cured samples. The morphological polymer viscosity at this temperature (ADA decomposes features of the es were analyzed from fresh fracture su pyrolysis, with no evident signs of melting or formation faces using a SEM(SM-6300F SEM, JEOL, Tokyo, Japan). cracks, indicating that the amount of cross linking achieved SEM images were subsequently analyzed with the Image Tool during foaming was sufficient to prevent thermoplastic flow software(UTHSCSA, University of Texas, San Antonio, TX)to the polysiloxane, and that the open-pore structure of the mate- quantify the cell size and cell-size distribution. The raw data rial allowed the release of the decomposition gases. The PMPs obtained by image analysis were converted to 3D values to ob- precursor can be cured thermally in air at temperatures >100C tain the effective cell dimension by applying the stereological without the need of any peroxide radical initiator and curing equation: Dsphere=Deirde/0. 785. Specimens appropriate for agent. The ceramization of the polymer takes place pre- solution transmission electron microscopy(HRTEM) dominately between 400 and 600C, and the resulting SiO-C cross-sectioning techni material ast up to pyrolysis temper ally resolved characterization as well as electron energy ures of 1200 Increasing the pyrolysis temperature results in LS)and energy-dispersive X-ray spectroscopy(EDXS) crystallization of the amorphous matrix. TGA analysis mea ormed using an aberration-corrected(Cs probe correc- surements(see Figs. 1(a)and(b)were in agreement wit TITAN 80-300 analytical scanning transmission elec literature data, confirming that the ceramization of cured bodies roscope(STEM), allowing a spatial resolution of better occurred between 400 and 600oC,with a ceramic yield >75% in Stem mode and an energy resolution of the eels It 1200oC. We observed that the incorporation of iron ions measurements of about 0. 2 eV, which was of special importance affected the stability of the resulting ceramics at high tempera for the recording of the fine structure signals near the ionization tures(T>1000.C). in particular leading to a decrease in ceramic dges (ELNES). yielding information on chemical bonding. X ield at high temperatures. This enhanced decomposition(de ray diffraction data (XRD, Bruker D8-Advance, Karlsruhe rease in thermal stability) occurs with the release of gas (SiO Germany) were collected using CuKol radiation (40 kv and CO)and crystallization, similarly to what was reported for a 0 mA: step scan of 0.05, counting time of 5 s/step and Ni-containing polysiloxane. In particular, while pure A=1. 54060 A). Raman spectra were recorded with an Invia heated in N2 did not show any significant weight loss Renishaw Raman microspectrometer(Gloucesteshire, U.K. )at 1000.C, and no peaks were evident in the dta curve up to PMPS-ADA PMPS-ADA-, PMPS-ADA-FeCL -20 30 TGA DTA DTA 020040060080010001201400020040060080o100012001400 2004000008001000120014 Temperature(C) emperature(C) (a)Nitrogen (b)Argon Fig 1. TG/DTA data for PMPS samples treated under(a)N, (b)Ar

aim of this development is the production of macroporous ce￾ramic components possessing high SSA values, for gas adsorp￾tion, catalyst support, and pollutant removal applications. II. Experimental Procedure Cellular ceramics were produced by using a commercially avail￾able PMPS preceramic polymer (H44, Wacker Chemie AG, Burghausen, Germany), denoted PMPS in the remainder of the text. Solid PMPS 96 wt% was ball milled together with 1 wt% azodicarbonamide (ADA 97%, Sigma–Aldrich, St. Louis, MO) acting as a physical blowing agent, and 3 wt% of a transition metal halide powder (FeCl2, 498% pure, Sigma–Aldrich), serv￾ing as a catalyst source for the growth of nanostructures. The mixed batch was then transferred to an oven for foaming and thermal crosslinking (1 h at 901C and then 5 h at 2501C; 21C/ min heating rate). The porous thermoset monoliths thus ob￾tained were then individually pyrolyzed under N2 or Ar (both 99.999% pure) in an alumina tube furnace (2 h at the required temperature, in the range 12501–14001C; 21C/min heating and cooling rate). Thermal analysis (TG-DTA) measurements were carried out under Ar or N2 (Netzsch STA 429, Selb, Germany; 21C/min heating rate) on the already cured samples. The morphological features of the samples were analyzed from fresh fracture sur￾faces using a SEM (JSM-6300F SEM, JEOL, Tokyo, Japan). SEM images were subsequently analyzed with the ImageTool software (UTHSCSA, University of Texas, San Antonio, TX) to quantify the cell size and cell-size distribution. The raw data obtained by image analysis were converted to 3D values to ob￾tain the effective cell dimension by applying the stereological equation: Dsphere 5 Dcircle/0.785.32 Specimens appropriate for high-resolution transmission electron microscopy (HRTEM) were prepared using an adapted cross-sectioning technique. Atomically resolved characterization as well as electron energy loss (EELS) and energy-dispersive X-ray spectroscopy (EDXS) was performed using an aberration-corrected (Cs probe correc￾tor) FEI TITAN 80-300 analytical scanning transmission elec￾tron microscope (STEM), allowing a spatial resolution of better than 1 A˚ in STEM mode and an energy resolution of the EELS measurements of about 0.2 eV, which was of special importance for the recording of the fine structure signals near the ionization edges (ELNES), yielding information on chemical bonding. X￾ray diffraction data (XRD, Bruker D8-Advance, Karlsruhe, Germany) were collected using CuKa1 radiation (40 kV, 40 mA; step scan of 0.051, counting time of 5 s/step and l 5 1.54060 A˚ ). Raman spectra were recorded with an Invia Renishaw Raman microspectrometer (Gloucesteshire, U.K.) at￾tached to a confocal microscope ( 50 objective) using the 633 nm line of a He–Ne laser as the excitation wavelength. Samples were ground and the powders were used for analysis, using a low laser power (5%). III. Results and Discussion (1) Foaming, Crosslinking, and Thermal Analysis Many studies can be found in the literature concerning foaming of various polymeric systems using ADA.33–37 ADA has also been used to produce macrocellular SiOC33 and SiCN37 ceram￾ics from preceramic polymers. It was shown that the decompo￾sition process of this porogen can be controlled by varying its particle size, heating rate and processing temperature, and the presence and concentration of an activator.34,38 Activators, in￾cluding transition metal compounds in particular reduce the de￾composition temperature of ADA to values as low as 1501C.38 After heating at 2501C, cellular monoliths with a high amount of open porosity were obtained. The presence of porosity can be attributed both to the continued release of volatiles/oligomers during the PMPS curing39 and to the good match between the temperature of decomposition of the blowing agent and the polymer viscosity at this temperature (ADA decomposes B2101C).33,40 All the foams retained their morphology during pyrolysis, with no evident signs of melting or formation of cracks, indicating that the amount of cross linking achieved during foaming was sufficient to prevent thermoplastic flow of the polysiloxane, and that the open-pore structure of the mate￾rial allowed the release of the decomposition gases. The PMPS precursor can be cured thermally in air at temperatures 41001C without the need of any peroxide radical initiator and curing agent.39–41 The ceramization of the polymer takes place pre￾dominately between 4001 and 6001C, and the resulting Si–O–C material remains amorphous at least up to pyrolysis tempera￾tures of 12001C.42 Increasing the pyrolysis temperature results in crystallization of the amorphous matrix.43 TGA analysis mea￾surements (see Figs. 1(a) and (b)) were in good agreement with literature data, confirming that the ceramization of cured bodies occurred between 4001 and 6001C,39 with a ceramic yield 475% at 12001C. We observed that the incorporation of iron ions affected the stability of the resulting ceramics at high tempera￾tures (T410001C), in particular leading to a decrease in ceramic yield at high temperatures. This enhanced decomposition (de￾crease in thermal stability) occurs with the release of gas (SiO and CO) and crystallization, similarly to what was reported for a Ni-containing polysiloxane.44 In particular, while pure PMPS heated in N2 did not show any significant weight loss above 10001C, and no peaks were evident in the DTA curve up to Fig. 1. TG/DTA data for PMPS samples treated under (a) N2, (b) Ar. 960 Journal of the American Ceramic Society—Vakifahmetoglu et al. Vol. 93, No. 4

April 2010 Growth of ID Nanostructures in Porous Polymer-Derived Ceramics (b) Fig. 2 Scanning electron microscope micrographs taken from the fracture surfaces of sample MPS-FeCirADA pyRolyzed under N2: (a and b)at 1500C(Fig. I(a), the data for the sample containing FeCl, different length(see Figs. 2(a)and (b) for the general and for owed the presence of an endothermic peak around 1450%C, detailed view, respectively). No significant change was observed associated with a large weight loss. This was attributed to the in the general morphology of the samples heated in the 1250 occurrence of a carbot 1400C temperature range(Figs. 2(aHD), and the foams had nd C, to form SiC(see later) ewise, while pure PMPs spherical cells (-300+150 um in diameter) with connecting cell heated in Ar did not show any significant mass loss abo indows(135+87 um in diameter). At all pyrolysis tempera 1000C, the one including FeCl] showed a continuous mass loss, tures, a large amount of nanowires were homogeneously d nd at 1500oC, the difference between these two samples became tributed on the surface of the macro-porous components. No >5 wt%(Fig. I(b)) tips (or particles) were observed to be present at the end of the nanowires(see embedded high-magnification images in Figs 2(b),(d)and(D), and the length of the nanowires was as high as (2) Microstructural and Nanochemical characterization A) Nitrogen Pyrolysis. The pyrolysis of the PMPS- XRD and Raman spectroscopy were performed to under FeCly-ADA sample under N, atmosphere at 1250C yielded a stand the phase evolution in dependence on the pyrolysis tem- macrocellular ceramic decorated with bundles of nanowires of perature. Figure 3(left) shows the XRD patterns for PMPS- FeaS SimOn 巴 LV 901000 ig. 3. X-ray diffraction pattens(left) and Raman spectroscopy (right)of the samples pyrolyzed in N2(a)1250C,(b)1300C(c)1350C and d)1400° C treatment

15001C (Fig. 1(a)), the data for the sample containing FeCl2 showed the presence of an endothermic peak around 14501C, associated with a large weight loss. This was attributed to the occurrence of a carbothermal reduction reaction between Si3N4 and C, to form SiC (see later).17,45,46 Likewise, while pure PMPS heated in Ar did not show any significant mass loss above 10001C, the one including FeCl2 showed a continuous mass loss, and at 15001C, the difference between these two samples became 45 wt% (Fig. 1(b)). (2) Microstructural and Nanochemical Characterization (A) Nitrogen Pyrolysis: The pyrolysis of the PMPS– FeCl2–ADA sample under N2 atmosphere at 12501C yielded a macrocellular ceramic decorated with bundles of nanowires of different length (see Figs. 2(a) and (b) for the general and for the detailed view, respectively). No significant change was observed in the general morphology of the samples heated in the 12501– 14001C temperature range (Figs. 2(a)–(f)), and the foams had spherical cells (B3007150 mm in diameter) with connecting cell windows (135787 mm in diameter). At all pyrolysis tempera￾tures, a large amount of nanowires were homogeneously dis￾tributed on the surface of the macro-porous components. No tips (or particles) were observed to be present at the end of the nanowires (see embedded high-magnification images in Figs. 2(b), (d) and (f)), and the length of the nanowires was as high as 500 mm. XRD and Raman spectroscopy were performed to under￾stand the phase evolution in dependence on the pyrolysis tem￾perature. Figure 3 (left) shows the XRD patterns for PMPS– Fig. 2. Scanning electron microscope micrographs taken from the fracture surfaces of sample PMPS–FeCl2–ADA pyrolyzed under N2: (a and b) at 12501C (c and d) at 13501C, and (e and f) at 14001C. Insets show high-magnification images of the nanowires. Fig. 3. X-ray diffraction patterns (left) and Raman spectroscopy (right) of the samples pyrolyzed in N2 (a) 12501C, (b) 13001C, (c) 13501C and, (d) 14001C treatment. April 2010 Growth of 1D Nanostructures in Porous Polymer-Derived Ceramics 961

962 Journal of the American Ceramic Society-Vakifahmetoglu et al Vol 93. No 4 Si-I C-K NK 200 Si 150 SiK 100 NK 0.2 Energy (kev pa [110] pole graph 5 nm hanochemical measurements: EELS of matrix ph(p上E⊥Smnw( nidle, and edx of snowi s htm over iw it corre- TEM/EELS/EDXS analyses of nic regularity of nanowires, with diffraction pattern, (c) HRTEM image of matrix phase directly surrounding the Si3 N4 nanowires, showing FeCly-ADA samples pyrolyzed at different temperatures under ies showed that the thermolysis of a polysiloxane and silicon(si) N, atmosphere. Although the sample obtained at 1250C mixture under N, yielded Si,N2O, due to the reaction of the displayed a typical diffraction pattern of amorphous SiCxOy, released Sio gas from the precursor with the pyrolysis atmo- with a broad hump in the 20-30 range(20), it also contained sphere(N2), and a further increase of the temperature resulted in well-defined crystalline features, attributable to B-SiC (JCPDs a continuous increase in the absorption of nitrogen up to the #29-1129), a-Si3N4 (JCPDS #41-0360) and Fe3Si (JCPDS #4 1207), together with a small amount of Si2ON2 (JCPDs showed that the equilibrium stable phases formed on nitriding #47-1627) phase. The formation of Si2 N20(sinoite: silicon SiO2: C mixtures in the temperature range between 1300 and nitride) phase has been shown to occur after the pyrolysis of 1500C are either B-Si3N4+C or SiN20+C, depending on the polysiloxane precursors in n/ ia atmosphere at low-process-. ygen partial pressure. Though the stability of phases depends ing temperatures(<1400C) A small peak at 33.7(20) strongly on impurities such as iron, increase in O2 partial pres- (marked with in Fig 3(left, (a) is usually attributed to planar sure clearly promotes the Si2N20 formation, and with the de. talline structure. With an increase in the pyrolysis temperat a-Si3N4 is known to be the low-temperature polymorph of the crystalline peaks of B-SiC and a-Si3 N4 became more intense i3N4, yet if the Sio partial pressure in the reaction bed is high, it remains stable up to the temperature where carbother relative to SiN,o decreased in intensity, while the ones for mal reduction of Si3 N4 with C occurs providing SiC. In this B-Si3 N4 became noticeable(20=27 1and 33.7). Previous stud- study, as deduced from DTA(see DTA in Fig. I(a)), the reac-

FeCl2–ADA samples pyrolyzed at different temperatures under N2 atmosphere. Although the sample obtained at 12501C displayed a typical diffraction pattern of amorphous SiCxOy, with a broad hump in the 201–301 range (2y),43 it also contained well-defined crystalline features, attributable to b-SiC (JCPDS #29-1129), a-Si3N4 (JCPDS #41-0360), and Fe3Si (JCPDS #45- 1207), together with a small amount of Si2ON2 (JCPDS #47-1627) phase. The formation of Si2N2O (sinoite: silicon oxy￾nitride) phase has been shown to occur after the pyrolysis of polysiloxane precursors in N2/NH3 atmosphere at low-process￾ing temperatures (o14001C).11,47–49 A small peak at 33.71 (2y) (marked with in Fig. 3 (left, (a)) is usually attributed to planar defects (stacking faults and rotational twins) in the b-SiC crys￾talline structure.50 With an increase in the pyrolysis temperature, the crystalline peaks of b-SiC and a-Si3N4 became more intense. At 13501C, the broad hump completely disappeared, the peaks relative to Si2N2O decreased in intensity, while the ones for b-Si3N4 became noticeable (2y 5 27.11 and 33.71). Previous stud￾ies showed that the thermolysis of a polysiloxane and silicon (Si) mixture under N2 yielded Si2N2O, due to the reaction of the released SiO gas from the precursor with the pyrolysis atmo￾sphere (N2), and a further increase of the temperature resulted in a continuous increase in the absorption of nitrogen up to the melting point of Si (Tm(Si)B14141C).11,51 Siddiqi and Hendry 46 showed that the equilibrium stable phases formed on nitriding SiO2:C mixtures in the temperature range between 13001 and 15001C are either b-Si3N41C or Si2N2O1C, depending on the oxygen partial pressure. Though the stability of phases depends strongly on impurities such as iron, increase in O2 partial pres￾sure clearly promotes the Si2N2O formation, and with the de￾crease in O2 partial pressure, Si3N4 is stabilized over Si2N2O.46,52 a-Si3N4 is known to be the low-temperature polymorph of Si3N4, yet if the SiO partial pressure in the reaction bed is high, it remains stable up to the temperature where carbother￾mal reduction of Si3N4 with C occurs providing SiC.46 In this study, as deduced from DTA (see DTA in Fig. 1(a)), the reac￾Fig. 4. HRTEM/EELS/EDXS analyses of sample PMPS–FeCl2–ADA pyrolyzed at 14001C under N2 atmosphere; (a) TEM overview with corre￾sponding nanochemical measurements: EELS of matrix phase (top), EELS of nanowires (middle), and EDX of nanowires (bottom), (b) HRTEM image of the atomic regularity of nanowires, with diffraction pattern, (c) HRTEM image of matrix phase directly surrounding the Si3N4 nanowires, showing SiC with graphitic regions. 962 Journal of the American Ceramic Society—Vakifahmetoglu et al. Vol. 93, No. 4

April 2010 Growth of ID Nanostructures in Porous Polymer-Derived Ceramics 963 ure surfaces of sample PMPS-FeCly-ADA pyrolyzed under Argon:(a and b)at 1250C(c and d)at 1350.C, and (e and f)at 1400.C. Insets show high-resolution images of the nanow tion of Si3 Na nanowires with carbon occurred only at around atmosphere(see later). Similar investigations on samples heat 1450.C under N,: therefore. Sic did not form due to the car treated at lower temperature indicated that nitrogen was always bothermal reduction of Si3 N4 but through a different mecha- present only in the nanowires. The HRTEM image of the silicor nism(see later), and the Si2N20 phase transformed to Si3N tride nanowires, she 4(b), shows that they ha when the pyrolysis temperature increased. Indeed, the phase erfect single crystalline silicon nitride structure, without defects ransformation from Si2 N20 to Si3N4 has been shown to be fa (as confirmed also by the selected area electron diffraction pa ciliated with increased annealing temperature, especially aboy ternsee inset), which was identical over the entire nanowire. 1300C or with the extension of the heat treatment time in n Both the hrTEM image and the Saed pattern suggest that the atmosphere.. This is in good agreement with the XRD results nanowires grew along the [lll] direction. Figure 4(c)shows a of the present study. For the sample pyrolyzed at 1400.C lattice plane imaging of an agglomeration of nanoparticles in the composite ceramic containing B-SiC, a mixture of iron silicide of nanowires, revealing that these particles contain hases(predominantly Fe Si) and Si3N4(both a(JCPDS #41 ic regions (together with amorphous carbon distributed 0360) and p CPDs #33-1160) polymorphs) lout any in the sample, as supported by Raman investigations) and sil were obtained icon carbide in accordance with eels data Raman spectroscopy was used to acquire information about As mentioned before, XRD results indicated that the b-si3N e structural evolution, in particular, of the free carbon phase se became more visible with increasin dispersed in the resulted matrix obtained at different p ture, while the Si2N20 phase gradually disappeared. This result, temperatures. The ceramic foam obtained at 1250C pyrolysis in combination with TEM data, implies that B-Si, N4 formed via under N2 atmosphere exhibited features typical of amorphous a phase transformation of the o Weimer et all the, with a broad D band(1330 cm-), more intense than have in fact shown that the control of the intermediate theg band( 1580 cm ) and a small 2D band centered around being either carbon-rich (Si-O-C) or nitrogen-rich (Si-O-N) ncreased,besides a separation of the G-peak into two separate although the factors controlling the B-Si N4 formation are ver maxima at 1580 cm al G-band)and 1 620 cm(D-b complex. In order to understand the growth mechanism of the all the peaks narrowed. While the int of the g-band i nanowires. microstructural investigations were carried out both creased continuously, the intensity of the D-band decreased. This by SEM and HrTEM, but attempts tify transition metal- graphite,with increasing pyrolysis temperature 5.56 Raman were not successful. Consequently, the iron silicide he nanowires indicates an enhancement in the ordering of carbon toward containing compounds on the tips and roots of the nanowires data demonstrate the permanence of carbon in the structure at presence is evident from XRD data, is believed to be randomly high pyrolysis temperature, in accordance with eels distributed within the matrix phase. It was shown that SiO and In Fig 4, the results of the HRTEM, EDXS, and EELS an- CO are the main gaseous species that form during the pyrolysi alyses are reported. Figure 4(a) shows a TEM overview of the of a similar polysiloxane precursor at temperatures>1000oC ple pyrolyzed at 1400C in N2. The eels (top right)shows and that the partial pressure of both of the gases increases with that the matrix phase surrounding the nanowires consisted of increasing pyrolysis temperature up to 1400.C.ThErefore,it silicon carbide, containing some percent oxygen, and of graphi be assumed that Sio gas reacted with N2 together with the tic carbon(hence, it was a not completely crystallized SioC ma- free carbon, to nucleate Si3 N4 crystallites according to the pro- terial). The EEls(middle right) and the eDX profile(bottom posed reaction right)were taken from the nanowires, both indicating that the nanowires contained only silicon and nitrogen. The quantifica 3Sio(g)+3C(s)+2N2(g)- N4(s)+3CO(g) (1) tion of the spectra evidenced that the nanowires consisted of pure Si3N4. Obviously, the nitrogen originated from the flowing pyrolysis as conant The nucleation thus occurred via a vapor-s that no si3 Na phase was observed when Ar was used as pyrolysis wing reaction(1), when the concenti

tion of Si3N4 nanowires with carbon occurred only at around 14501C under N2; therefore, SiC did not form due to the car￾bothermal reduction of Si3N4 but through a different mecha￾nism (see later), and the Si2N2O phase transformed to Si3N4 when the pyrolysis temperature increased. Indeed, the phase transformation from Si2N2O to Si3N4 has been shown to be fa￾cilitated with increased annealing temperature, especially above 13001C 53 or with the extension of the heat treatment time in N2 atmosphere.51,54 This is in good agreement with the XRD results of the present study. For the sample pyrolyzed at 14001C, a composite ceramic containing b-SiC, a mixture of iron silicide phases (predominantly Fe3Si) and Si3N4 (both a (JCPDS #41- 0360) and b (JCPDS #33-1160) polymorphs), without any Si2ON2, were obtained. Raman spectroscopy was used to acquire information about the structural evolution, in particular, of the free carbon phase dispersed in the resulted matrix obtained at different pyrolysis temperatures. The ceramic foam obtained at 12501C pyrolysis under N2 atmosphere exhibited features typical of amorphous carbon, with a broad D band (B1330 cm1 ), more intense than the G band (B1580 cm1 ), and a small 2D band centered around 2660 cm1 , see Fig. 3 (right, (a)). As the pyrolysis temperature increased, besides a separation of the G-peak into two separate maxima at 1580 cm1 (actual G-band) and 1620 cm1 (D0 -band), all the peaks narrowed. While the intensity of the G-band in￾creased continuously, the intensity of the D-band decreased. This indicates an enhancement in the ordering of carbon toward graphite, with increasing pyrolysis temperature.55,56 Raman data demonstrate the permanence of carbon in the structure at high pyrolysis temperature, in accordance with EELS. In Fig. 4, the results of the HRTEM, EDXS, and EELS an￾alyses are reported. Figure 4(a) shows a TEM overview of the sample pyrolyzed at 14001C in N2. The EELS (top right) shows that the matrix phase surrounding the nanowires consisted of silicon carbide, containing some percent oxygen, and of graphi￾tic carbon (hence, it was a not completely crystallized SiOC ma￾terial). The EELS (middle right) and the EDX profile (bottom right) were taken from the nanowires, both indicating that the nanowires contained only silicon and nitrogen. The quantifica￾tion of the spectra evidenced that the nanowires consisted of pure Si3N4. Obviously, the nitrogen originated from the flowing N2 gas present in the pyrolysis tube, as confirmed by the fact that no Si3N4 phase was observed when Ar was used as pyrolysis atmosphere (see later). Similar investigations on samples heat treated at lower temperature indicated that nitrogen was always present only in the nanowires. The HRTEM image of the silicon nitride nanowires, shown in Fig. 4(b), shows that they had a perfect single crystalline silicon nitride structure, without defects (as confirmed also by the selected area electron diffraction pat￾tern—see inset), which was identical over the entire nanowire. Both the HRTEM image and the SAED pattern suggest that the nanowires grew along the [111] direction. Figure 4(c) shows a lattice plane imaging of an agglomeration of nanoparticles in the vicinity of nanowires, revealing that these particles contain graphitic regions (together with amorphous carbon distributed in the sample, as supported by Raman investigations) and sil￾icon carbide, in accordance with EELS data. As mentioned before, XRD results indicated that the b-Si3N4 phase became more visible with increasing pyrolysis tempera￾ture, while the Si2N2O phase gradually disappeared. This result, in combination with TEM data, implies that b-Si3N4 formed via a phase transformation of the oxynitride phase. Weimer et al. 57 have in fact shown that the control of the intermediate phase, being either carbon-rich (Si–O–C) or nitrogen-rich (Si–O–N), dictates the formation of a-Si3N4 or b-Si3N4 phase, respectively, although the factors controlling the b-Si3N4 formation are very complex.58 In order to understand the growth mechanism of the nanowires, microstructural investigations were carried out both by SEM and HRTEM, but attempts to identify transition metal￾containing compounds on the tips and roots of the nanowires were not successful. Consequently, the iron silicide phase, whose presence is evident from XRD data, is believed to be randomly distributed within the matrix phase. It was shown that SiO and CO are the main gaseous species that form during the pyrolysis of a similar polysiloxane precursor at temperatures 410001C, and that the partial pressure of both of the gases increases with increasing pyrolysis temperature up to 14001C.59 Therefore, it can be assumed that SiO gas reacted with N2 together with the free carbon, to nucleate Si3N4 crystallites according to the pro￾posed reaction 57 3SiOðgÞ þ 3CðsÞ þ 2N2ðgÞ ! Si3N4ðsÞ þ 3COðgÞ (1) The nucleation thus occurred via a vapor–solid (VS) mecha￾nism, following reaction (1), when the concentration of SiO and Fig. 5. Scanning electron microscope micrographs taken from the fracture surfaces of sample PMPS–FeCl2–ADA pyrolyzed under Argon: (a and b) at 12501C (c and d) at 13501C, and (e and f) at 14001C. Insets show high-resolution images of the nanowires. April 2010 Growth of 1D Nanostructures in Porous Polymer-Derived Ceramics 963