Porous and high surface area silicon oxycarbide-based materials—A review

Materials Science and Engineering R 97 (2015) 23–49 Contents lists available at ScienceDirect Materials Science and Engineering R journal homepage: ...

Download PDF

13MB Sizes 562 Downloads 543 Views

Report

Full Text

Materials Science and Engineering R 97 (2015) 23–49

Contents lists available at ScienceDirect

Materials Science and Engineering R journal homepage: www.elsevier.com/locate/mser

Porous and high surface area silicon oxycarbide-based materials—A review Kathy Lu * Virginia Tech, 213 Holden Hall, Blacksburg, VA 24061, USA

A R T I C L E I N F O

A B S T R A C T

Article history: Available online

Silicon oxycarbide (SiOC)-based materials are a class of polymer-derived ceramics that enables the formation of a homogeneous structure at the molecular level starting from polymer precursors. In this system, oxygen and carbon atoms share bonds with silicon atoms in the amorphous network structure while elemental carbon, and possibly nanosized SiO2 and SiC nanodomains may co-exist. Because of the ﬂexibility of molecular level composition and microstructure designs, the systems can be made porous with high speciﬁc surface areas by changing the precursor compositions and the ceramization conditions. In this review, two strategies of creating porous SiOCs are discussed: conventional approach of using fugitive ﬁllers, as well as pore formation and selective removal of certain SiOC matrix compositions (such as carbon, SiO2, or SiC) at the molecular level. For the former, it includes ceramic replication of an organic template, direct foaming, and sacriﬁcial pore formers. For the latter, it includes molecular level pore formation, molecular level species removal, and SiOC porous network creation through molecular templates. Direct pore formation can be achieved by changing processing conditions, using different precursor architectures, and using different hydrosilylation agents. For SiOC porous network creation through molecular level species removal, it includes molecular level free carbon removal, molecular level SiO2 nanocluster removal, and molecular level carbon removal from SiC (and possibly BCx for SiOBC). To understand single nanometer (<10 nm) pore formation and phase separation for selective species removal, SiOC nanostructure models and composition descriptions after the pyrolysis are explained. ß 2015 Elsevier B.V. All rights reserved.

Keywords: Silicon oxycarbide Speciﬁc surface area Porous material Pore former Template Molecular level removal

Contents 1. 2.

3. 4.

Overview of high surface area SiOC materials . . . . Conventional SiOC porous material formation . . . . Ceramic replication of an organic template. 2.1. Direct foaming . . . . . . . . . . . . . . . . . . . . . . . 2.2. Sacriﬁcial pore formers. . . . . . . . . . . . . . . . . 2.3. 2.3.1. Expandable organic additives . . . . Nonexpandable organic additives . 2.3.2. Carbon template . . . . . . . . . . . . . . 2.3.3. 2.3.4. Silica template . . . . . . . . . . . . . . . . SiOC nanostructure model. . . . . . . . . . . . . . . . . . . . SiOC formation process . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

24 24 24 25 28 28 28 31 31 32 35

Abbreviations: D4H, 2,4,6,8-tetramethylcyclotetrasiloxane; DVB, divinylbenzene; HMMH, 1,1,3,3-tetramethyldisiloxane; NLDFT, non-linear density functional theory; PDMS, poly(dimethylsiloxane); PHMS, poly(hydridomethylsiloxane); PMMA, poly(methyl methacrylate); PSO, polysiloxane; Q(MH)4, tetrakis(dimethylsiloxy)silane; SiOC, silicon oxycarbide. * Tel.: +1 540 231 3225; fax: +1 540 231 8919. E-mail address: [email protected] http://dx.doi.org/10.1016/j.mser.2015.09.001 0927-796X/ß 2015 Elsevier B.V. All rights reserved.

24

5.

6.

7. 8.

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

Molecular level pore formation. . . . . . . . . . . . . . . . . . . . . . . . . Direct pore formation . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Different polysiloxane precursor architectures . . . . . . . 5.2. Different hydrosilylation agents . . . . . . . . . . . . . . . . . . . 5.3. SiOC porous network creation through atomic and molecular Atomic level bond modiﬁcation . . . . . . . . . . . . . . . . . . . 6.1. Atomic level carbon removal . . . . . . . . . . . . . . . . . . . . . 6.2. Molecular level SiO2 nanocluster removal . . . . . . . . . . . 6.3. 6.4. Molecular level SiC (BCx) removal . . . . . . . . . . . . . . . . . SiOC porous network creation through molecular templates . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.............. .............. .............. .............. species removal. .............. .............. .............. .............. .............. .............. ..............

1. Overview of high surface area SiOC materials Porous and high surface area materials have important applications as sensors, chemical reactors, electrodes, gas storage media, molecular sieves, membrane supports, lightweight structural materials, thermal insulators, bioimplants, among others. On this front, much work has been done for room temperature use materials, such as porous carbons, mesoporous silicates, metal– organic frameworks, and organic frameworks. However, many applications require the porous materials to be thermally stable and the speciﬁc surface area to be higher than a couple of hundred square meters per gram. Thermally stable and high surface area materials are seriously lacking. This can be attributed to two factors: the instability of the matrix itself (sintering, decomposition) and the instability of the pores (shrinkage, closure) with temperature increase. These challenges call for the development of new, highly porous materials with stable networks. Building up porous materials from polymer precursor molecular units and tailoring the polymer molecular structure to ceramic transformation so that high surface area and thermally stable phases can be consistently produced are of great scientiﬁc and technological signiﬁcance [1,2]. SiOC porous materials are such a unique system that can be produced from a large variety of polysiloxane (PSO) precursors and a wide range of processing conditions. A wide array of interesting SiOC microstructures can be created to offer highly porous materials with the thermal stability from a few hundred degrees Celsius to 1300 8C. From the composition point of view, SiOC is an exciting high temperature material system that enables the formation of a homogeneous amorphous-nanocrystalline mixed structure at the molecular level starting from polymer precursors containing Si– H, Si–Me, Si–Et, Si–Vy, or Si–Ph bonds. SiOC ceramics, prepared from preceramic polymer routes, can be thermochemically stable up to 1200 8C in air, and do not show compositional or microstructural changes after fairly extended periods of time at elevated temperatures. They have good oxidation resistance, possess a low coefﬁcient of thermal expansion, and exhibit good mechanical strength, creep resistance, and resistance to corrosion [3]. However, after extended time at 1200 8C or higher, the stability of SiOC deteriorates due to the evolution of porosity, the oxidation of different species, and the devitriﬁcation of the SiOC matrix [4–6]. The compositions and the structural characteristics of these materials change continuously with the precursors and the temperatures used to process the precursors and any ﬁller that might be employed. The chemical compositions may be controlled by varying the molecular architecture of organosiloxane precursors as well as the temperature and atmosphere of pyrolysis [7]. Depending on the pore forming process, the pore sizes for SiOC can vary greatly from nanometers to millimeters and lead to large ranges of porosity and speciﬁc surface area [2].

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

36 36 38 40 40 40 41 42 46 47 48 48

Polymer-derived SiOC materials have the advantages of amendable to polymer-processing techniques, homogeneity of precursors at the molecular level, relatively lower processing temperatures (compared to conventional ceramic processing methods), and possible synthesis of new compositions that can have functional properties, such as electrical conductivity and magnetic property (Fig. 1). More importantly, polymer-derived SiOCs can be formed from different polymer molecular structures so that the resulting nano- and/or micro-structures and phases can be tailored [1]. In general, there are two ranges of pores that can be created for a SiOC material. One kind of pores is in the submicron to tensthousands of microns range. This type of SiOC is generally produced using fugitive ﬁllers or other types of pore forming processes with the following approaches: ceramic replication of an organic template, direct foaming, and sacriﬁcial pore formers. The SiOC matrix may be dense or porous and serves the major function of a connecting skeleton for the porous materials. The starting polymer precursors can vary greatly but the small, singlenanometer pores (if there is any) in the SiOC matrix do not make signiﬁcant contribution to the overall porosity. This pore forming strategy can be labeled as conventional SiOC porous material formation process. The advantage of such porous materials is the easy accessibility of the pores by liquid, gas, and even other agents for different applications due to their large sizes. The shortcoming is the low mechanical properties. The other kind of porous SiOC materials has pores in the single nanometer to at most tens of nanometers range. These pores are created at two stages: volatile species formation during the pyrolysis (direct pore formation, different precursor architectures) or selective species removal after the pyrolysis based on the phase separated domains (free carbon removal, SiO2 nanocluster removal, and carbon removal from SiC). The advantages of such porous structures are their ultrahigh speciﬁc surface areas, small pore sizes, and narrow pore size distributions. However, the accessibility of such small pores can be an issue for certain applications. In this review, the above two strategies of creating porous SiOCs will be discussed. To understand single nanometer pore formation and phase separation for the selective species removal approach, the SiOC nanostructure models and composition interrelationships after the pyrolysis are explained. 2. Conventional SiOC porous material formation 2.1. Ceramic replication of an organic template To form SiOC porous materials with pores in the single micron to hundreds or even thousands of microns range, using organic template is the most institutively straightforward process. First, organic templates can be environmentally benign and cost effective. Second, organic templates can be easily removed to create pores. In certain cases, organic materials may also contain

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

25

Fig. 1. General characteristics of porous SiOC synthesis strategies and application potentials.

Si-related species and provide additional possibility of forming SiOC or SiC species to integrate with the SiOC matrix. The replica technique is widely used for the production of open-cell structures with various pore sizes, porosities, and chemical compositions. Several templates such as wood and polymeric foams can be used to prepare porous ceramics. The porous ceramics prepared from polymeric foams and ceramic slurries possess thin struts containing central pore channels. Porosities ranging from 85% to 96% can be obtained. However, there are also limitations to this approach. Owing to the inefﬁciency of inﬁltration and the back ﬂow of excess SiOC precursor liquids, the cell size generally ranges from 150 mm to a few millimeters. The precursor concentration, inﬁltration efﬁciency in-between the template, template type, and pyrolysis temperature signiﬁcantly affect the microstructure and strength of the resultant ceramics. SiOC ceramics with hierarchical pore structures are synthesized from a mixture of hydrogen-containing silicone oil and divinylbenzene using wood as the templates at different pyrolysis temperatures. These SiOC ceramics display a trimodal pore size distribution in the micro-, meso-(micropores + mesopores, 1.7–14 nm), and macro-(1–15 mm) pore ranges as shown in Fig. 2. The mesopores and micropores mainly originate from the formation of large amounts of SiC crystals and nanowires, graphite-like microcrystallites, and nanometer-scale pores of ray parenchyma cells and pits of wood biomass. The SiOC sample prepared at higher temperatures possesses the speciﬁc surface area up to 180.5 m2/g [8]. The microstructures of the above SiOC samples annealed at different temperatures show the evolution of a typical SiOC system. Fig. 3(a) and (b) are the typical TEM images taken from the samples obtained at 1000 8C and 1200 8C. Graphite-like microcrystallites (indicated by the guidelines) are embedded in the amorphous SiOC matrix phase (denoted by the circles). After annealing at 1300 8C, Fig. 3(c) shows that the graphite-like microcrystallites contain a handful of crystalline SiC (indicated by the rectangles). Free carbon also progressively involves in the reactions to form crystalline SiC. In Fig. 3(d), free carbon is minimally present. Moving away from natural organic templates, porous SiOC structures can be created based on synthetic templates. For

example, a polyurethane sponge is immersed in a polymeric suspension or precursor solution, followed by pyrolysis to produce porous SiOC ceramics with the same cell morphology as the original polyurethane sponge [9,10]. Since the synthetic templates tend to have well-controlled sizes and shapes, the large pores formed are more regular. By irradiating a mixture comprising poly(methyl methacrylate) (PMMA) microspheres and poly(dimethylsiloxane) (PDMS) with an electron beam, a highly porous micro/nano hierarchical ceramic composite is made (Fig. 4). The formation of the porous hierarchical structure is caused by the volume reduction of PMMA microporous spheres and the supposed transformation of PDMS to SiOC by electron irradiation. The pore sizes from the PMMA templates are 15 mm [11]. Because of the large pore sizes, open cellular structures, and sometimes broken struts in the porous structures through the organic template approach, ceramic replicas tend to possess larger pores and higher permeability, but suffer from low mechanical properties and residual cracks [12]. For example, the weak ceramic matrix may just possess <2 MPa overall compressive strength [13]. The random nature of the microstructures also presents signiﬁcant challenges for performance prediction. As a result, this approach is not widely practiced in the more advanced research of porous SiOCs. 2.2. Direct foaming Because of the necessary steps of liquid precursor crosslinking and polymer pyrolysis for the SiOC systems, both additives and processing variables can be tailored to generate gas bubbles and thus pores. This is the so-called direct foaming technique for porous SiOCs and Colombo et al. conducted extensive work on this topic [12–18]. Different from the process relying on organic templates, direct foaming can offer spherical pores consistently. More often, the porosity is high at 60–90% and the pores are open. The other advantage is the drastically improved porous material strength from interconnecting cells, as high as >100 MPa, which is critical for high porosity SiOC materials. To create pores, the most common approach is to blow gas during the crosslinking stage, just like blowing soap bubbles. For the polymer precursors, the

26

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

Fig. 2. N2 sorption isotherms for the SiOC samples obtained at different annealing temperatures (a), and the corresponding pore size distribution curves of these samples (b) [8]. (Reprinted from Journal of the European Ceramic Society, 34, J. M. Pan, J. F. Pan, X. N. Cheng, X. H. Yan, Q. B. Lu, and C. H. Zhang, Synthesis of hierarchical porous silicon oxycarbide ceramics from preceramic polymer and wood biomass composites, 249–56, 2014, with permission from Elsevier.)

viscosity is too high for the gas bubbles to escape. Accordingly, porous matrix is created. The other is to use organic additives that produce gases. For the gas blowing technique, a porosity of 60% and a cell density greater than 104 cells/cm3 are obtained from a mixture of GE Toshiba YR3370 PSO and a low temperature endothermic blowing agent (which has an average particle size of 12 mm and liberates 100 cm3/g CO2 at 125–165 8C) by simple compression molding and subsequent pyrolysis of the PSO [19]. The average cell

sizes are 650 mm before pyrolysis and shrink to 400 mm after pyrolysis. Increasing gas blowing pressure does not necessarily lead to larger pores. For a silicone resin (Dow Corning 217) ﬁlled with SiOC powders (Dow Corning 217 and SiOC powders with a volume ratio of 70:30), with the exterior gas pressure increase from 1 MPa to 4 MPa, the total and open porosities of the resulting porous ceramics decrease from 69.8% to 58.3% and from 58.4% to 43.9%, respectively (Fig. 5) [20]. During the porous SiOC creation, ﬁrst the Dow Corning 217 silicone resin derived SiOC ceramics

Fig. 3. TEM images of SiOC samples obtained at different temperatures (a) 1000 8C, (b) 1200 8C, (c) 1300 8C, and (d) 1400 8C [8]. (Reprinted from Journal of the European Ceramic Society, 34, J. M. Pan, J. F. Pan, X. N. Cheng, X. H. Yan, Q. B. Lu, and C. H. Zhang, Synthesis of hierarchical porous silicon oxycarbide ceramics from preceramic polymer and wood biomass composites, 249–56, 2014.)

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

27

Fig. 4. SEM images of (a) the multilayer of PMMA microspheres, (b) the multilayer of PMMA microspheres with a silicone grease coating, and (c) the porous hierarchical structure formed after electron irradiation. (d) Magniﬁed image of the wall of the porous hierarchical structure in image (c). Scale bars are 100 mm for images (a)–(c) and 10 mm for image (d) [11]. (Reprinted with permission from E. J. Lee, J. J. Kim, and S. O. Cho, Fabrication of porous hierarchical polymer/ceramic composites by electron irradiation of organic/inorganic polymers: route to a highly durable, large-area superhydrophobic coating, Langmuir, 26 (5) 3024–30. Copyright 2010, American Chemical Society).

(1200 8C, 1 h, argon) are ground to powders of a mean size of 9.1 mm. The powders are then used as ﬁllers to restrain the shrinkage during pyrolysis. With the heating rate increase from 0.25 8C/min to 3 8C/min during the crosslinking stage, the total and open porosities of the porous ceramics increase ﬁrst and then degrade. The explanation is that higher exterior gas pressure suppresses the growth of bubbles, resulting in smaller pore sizes and lower porosities. In addition, adjoining bubbles become less likely to merge. As a result, the connectivity (open porosity) of

Fig. 5. Effect of the exterior gas pressure on the total and open porosities of the porous ceramics.20 (Reprinted from Materials Letters, 66, H. Tian and Q. S. Ma, Effects of exterior gas pressure on the structure and properties of highly porous SiOC ceramics derived from silicone resin, 13–15, 2012, with permission from Elsevier.)

porous ceramics degrades with increasing exterior gas pressure. As expected, the compressive strength increases with increasing exterior gas pressure, and the average compressive strength of the porous ceramics is in the range of 3.9–14.9 MPa. The corresponding microstructure examination indicates that with the exterior gas pressure increase, the ﬁnal structure of the porous ceramics becomes more regular and equirotal [20–22]. Maximum total and open porosities are 88.2% and 72.5%, respectively, at a heating rate of 0.5 8C/min for the same system [23]. However, the compressive strength decreases progressively from the low value of 2.3 MPa to even lower value of 1.0 MPa. Based on similar ideas, the SiOC materials can be made into different shapes. Cylindrical SiOC foam ﬁlaments with a radial gradient in pore size are processed by continuous extrusion foaming of a methyl polysilsesquioxane. Upon leaving the extrusion nozzle, foaming is initiated by pressure release which causes precipitation of supersaturated CO2 from the polymer ﬁlament. Rapid cooling of the thin ﬁlaments generates a radial gradient of melt viscosity which gives rise to the formation of isotropic pore cells in the core (diameter <200 mm) and nonisotropic pore cells near the surface (shell, <20 mm). After pyrolysis at temperatures from 800 to 1400 8C, the stabilized gradient polymer foams can be converted into closed cell, gradient SiOC ceramic foams [24]. Desirably, a high compressive strength of 9 MPa and a Young’s modulus of 7 GPa at a relative density of 18% are obtained at an optimal pyrolysis temperature of 1000 8C. Different from providing external gas pressure, additives can be used to generate gas bubbles inside the PSO precursors. Since gas bubbles are generated internally, both open pores and closed pores are possible. Using poly(hydridomethylsiloxane) (PHMS) as a polymer precursor, 1,4-diazabicyclo [2.2.2] octane has been used

28

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

to release hydrogen gas for pore formation. This results in selfblowing of the polymer precursor at room temperature and thereby offers the possibility of producing SiOC foams without the need of any external blowing gas. Bubbles of hydrogen nucleate at the bottom of the container and move upwards resulting in the formation of localized colonies followed by the growth of existing bubbles at the expense of newly formed bubbles. Elongated pores oriented along the foaming direction are observed. The foam structure is retained during the pyrolysis [25]. This method of producing SiOC foams is promising since the removal of sacriﬁcial phase during pyrolysis is not required. However, it is difﬁcult to control the density of ceramic foams. In addition to relying on blowing agents, chemical reactions, and processing conditions for gas bubble generation, other unconventional sources can also be used to generate pores in PSO. Methylsilicone resin as a polymer precursor has been cured and foamed by electron beam irradiation in air prior to pyrolysis under an inert atmosphere. Methylsilicone foams are obtained without additional foaming agent when exposed to accelerated electrons with radiation doses up to 9 MGy and a dose rate of 2.8 kGy/s. During irradiation the polymer is melted and simultaneously gaseous products are formed by the methyl group oxidation and by the poly-condensation crosslinking reactions [26,27]: Si OH þ HO Si ! Si O Si þ H2 O "

(1)

Si OC2 H5 þ HO Si ! Si O Si þ C2 H5 OH "

(2)

The formed H2O and C2H5OH gases could not escape from the molten polymer and aggregate into bubbles. When the methylsilicone foams are pyrolyzed in N2 atmosphere at temperatures of 1200 and 1500 8C, an amorphous SiOC foam and a partially crystalline ceramic foam form, respectively. A porosity of 84% is achieved in the pyrolyzed foams, with cell sizes ranging from 30 to 300 mm and a density of 0.31 g/cm3. Macrocellular foams, with cell sizes from 100 mm to 600 mm and a bulk density from 0.25 to 0.58 g/cm3, are fabricated using a direct foaming approach. Precursors for polyurethane (polyols and isocyanates) in dichloromethane (with the addition of surfactants and catalysts) are used as a physical blowing agent for the system [16]. In addition, open-cell ceramic foams are obtained from a silicone resin and blown polyurethane. The density ranges from 0.1 to 0.3 g/cm3. The average pore size ranges from 200 to 400 mm. The open porosity is 80 to 90% [28]. Overall, large size pores (hundreds to thousands of microns) and open cellular structures are the dominating characteristics of such porous SiOCs. 2.3. Sacriﬁcial pore formers Even though different sacriﬁcial pore formers perform very similar functions of pore formation, the resulting pore sizes can be different depending on whether or not the additives are expandable during decomposition. The sizes of the pore formers almost always directly determines the ﬁnal pore sizes. Depending on the species used, the ﬁnal pores can be from hundreds of microns to single nanometers (Fig. 6). The most common pore formers are decomposable polymer spheres. They are often introduced into the PSO precursors before crosslinking. Other pore formers can be introduced as templates, such as highly porous carbons. Liquid PSO penetrates into the mesopores of the carbon templates. At high temperatures, the carbon template is removed by pyrolysis. Singlenanometer pores and very high speciﬁc surface areas can be produced using this approach. In addition, etchable pore formers such as SiO2 and B2O3 can be introduced and stay through the PSO crosslinking and pyrolysis process. After the SiOC formation, the

Fig. 6. Different sacriﬁcial pore forming agents.

dispersed SiO2 particles can be etched away using strong chemicals, such as a HF solution. The pore sizes are determined by the additive particle sizes and distributions and can range from single nanometers to tens of microns. 2.3.1. Expandable organic additives For the pore formation through expandable organic species approach, clean burning (little or no residues) polymers are the most common additive. Polyurethane can be added up to 50 wt% to control the morphology of the porous structure. Because of the internal pore formation nature, both open pores (cells) and closed pores (cells) can be produced. Polyurethane expansion requires heat treatment at 25–40 8C, accelerated by high speed mixing (introduction of bubbles in the solution) [3]. From a different perspective, SiOC foams can also be fabricated from a mixture of PSO and hollow microspheres (a commercial expandable product from Sundsvall (92DU40, Expancel)). During steam chest molding, PSO softens and the hollow microspheres expand, resulting in a porous microstructure. During heat treatment, the microspheres decompose and the PSO transforms to an amorphous SiOC. The porosity can be controlled from 62% to 81% by adjusting the initial packing density of expandable spheres from 0.13 to 0.22 g/cm3 [29]. As shown in Fig. 7 [30], closed-cell, microcellular SiOC ceramics with cell densities greater than 109 cells/cm3 and cells smaller than 30 mm can be obtained using the above expandable microspheres. Higher microsphere content leads to higher porosity in the microcellular ceramics. The compressive strength of the microcellular ceramics with 30% relative density (70% porosity) reaches 100 MPa, consistent with the earlier comment. If the PSO pyrolysis conditions are controlled properly, the cell walls can also be made porous. Hierarchically ordered macro/ micro porous foams are generated using PSOs and platinic acid as precursors and expandable polystyrene beads as ‘templates’. The large pore sizes are 1200 mm and the strut thickness is 200 mm. Through pyrolysis at 500 8C under a nitrogen atmosphere the templates can be removed completely, micro-porosity and thus high speciﬁc surface areas (548–611 m2/g) are generated, leading to hierarchically ordered monoliths [31]. The porous microstructures can be accessed by convective gas transport. To make more sophisticated porous structures, high temperature stable ﬁllers and decomposable ﬁllers can be used simultaneously. The pore size and porosity of closed-cell SiOC foams are investigated using expandable microspheres as a blowing agent and SiO2 powder as an inert ﬁller. PSO, an expandable microsphere, and SiO2 (0.8 mm, High Purity Chemicals, Saitama, Japan) are used as the raw materials. The addition of SiO2 inert ﬁllers leads to higher porosity and larger cell sizes due to the higher expansion allowed. All specimens containing ﬁllers have cell densities greater than 109 cells/cm3 and cell sizes smaller than 35 mm [32]. 2.3.2. Nonexpandable organic additives Rice bran has been used to make porous ceramics from a commercial methyl silicon resin, Silres 610 [(CH3SiO1.5)n], without forming cracks. The volume shrinkage resulting from polymer-toceramic conversion increases with increasing rice bran weight percentage, understandably due to the increasing amount of shrinkable pores. As shown in Fig. 8 [33], the porous ceramics

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

Fig. 7. Typical fracture surfaces of closed-cell, microcellular SiOC ceramics pyrolyzed at 1200 8C for 1 h with a heating rate of 2 8C/min in nitrogen: (a) 50 wt% expandable microspheres, and (b) 90 wt% expandable microspheres [30]. (Reprinted from Scripta Materialia, 53, Y. W. Kim, Y. J. Jin, Y. S. Chun, I. H. Song, and H. D. Kim, A simple pressing route to closed-cell microcellular ceramics, 921–25, 2005, with permission from Elsevier.)

made from 50/50 mixtures of Silres/rice bran show the highest ceramic yield (47.8%) and the highest compressive strength (2.7 MPa), but predictably the lowest porosity (30%). The highest porosity but the lowest compressive strength is obtained from 20/ 80 mixture of Silres/rice bran. With the increase of the crosslinking temperature for Silres 610, the porous SiOC ceramic compressive strength also increases linearly. However, due to the wide size distribution and random arrangement of the rice bran sacriﬁcial template, the morphology of the product porosity is random and the pore sizes produced are in the wide range of 2.7–45 mm [33]. With smaller size additives, the pore sizes can be effectively decreased. To make this work, the additive sizes must be small and the additive size distributions must be narrow. As it turns out, most of these species are not expandable during the processing. SiOC microcellular ceramic foams with pore sizes of 8 mm are produced using a commercially available methyl PSO (SR350, General Electric Silicone Products Division) and PMMA microbeads acting as a sacriﬁcial ﬁller [16,34]. In a different study, microfoams having cell sizes in the range of 10–150 mm and cell window sizes in the range of 10–50 mm are prepared using a commercial PSO precursor (MK, Wacker-Chemie GmbH, Germany) and PMMA microbeads (Altuglas BS, Altuglas International, Arkema Group, Rho (MI), Italy) of different nominal sizes (10, 25, 50, 100, and 185 mm) [12]. By mixing the silicone resin powder with PMMA microbeads of different sizes, bulk samples are formed by compaction. The PMMA beads are then eliminated by heat treatment in air at 300 8C, leaving a preceramic cellular foam. After pyrolysis at 1000–1200 8C in an inert atmosphere, macrocellular foam samples (Fig. 9(a)) possess a porosity >70 vol% and

29

Fig. 8. Effect of increasing weight percent of rice bran (RB) and the crosslinking temperature on the compressive strength of porous ceramic made from Silres 610 and rice bran blends. (a) Effect of weight percent of rice bran; (b) effect of the crosslinking temperature [33]. (Reprinted from Journal of the European Ceramic Society, 33, M. M. Hassan, T. Takahashi, K. Koyama, Preparation and characterisation of Si–O–C ceramics made from a preceramic polymer and rice bran, 1207–1217, 2013, with permission from Elsevier.)

a regular morphology comprised of spheroidal empty voids (cells) with sizes ranging from 150 mm to 1 mm. Micro-cellular foam samples (Fig. 9(b)) possess a porosity >70 vol% and a regular morphology comprised of spheroidal empty voids (cells) with sizes ranging from 1 mm to 100 mm [13]. For the SiOC ceramic foam prepared using silicon resin powder as the raw material and polyvinyl butyral powder as the poreforming agent, the compressive strength increases ﬁrst and then decreases, and the porosity decreases continually with the increase of the pyrolysis temperature from 1000 8C to 1400 8C. Compressive strength and porosity are 52.3 MPa and 72%, respectively, when the content of polyvinyl butyral and the pyrolysis temperature are 50 wt% and 1250 8C, respectively. Microstructure study reveals that the ceramic foams have three-dimensional web structures and uniform pores [35]. In another study, poly(methyl methacrylateco-ethylene glycol dimethacrylate) microbeads of 20–40 mm sizes are used as the pore forming agent along with PSO to form porous SiOC ceramic. A porosity of 27% to 88% and a cell density higher than 107 cells/cm3 are obtained [34]. The amount of pores in the cellular structures has an important effect on the stress release during pyrolysis and the integrity of the porous materials. Microcellular SiOC is produced through pyrolysis of methylsilicone resin with PMMA microbeads. The sacriﬁcial PMMA is added in the range of 0 to 80 wt%, and the pyrolysis is performed at 800–1200 8C. The volume shrinkage resulting from the polymer-to-ceramic conversion becomes stable at pyrolysis temperatures above 1000 8C, and the shrinkage value reaches

30

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

Fig. 9. SEM images of different porous morphologies obtained by pyrolysis of a silicone resin. (a) Macro-cellular open cell foam, (b) micro-cellular open cell foam [13]. (Reprinted from Journal of the European Ceramic Society, 28, P. Colombo, Engineering porosity in polymer-derived ceramics, 1389–95, 2008, with permission from Elsevier.)

about 60% with 80 wt% PMMA addition. Microcellular foams with pore diameters from 1 to 10 mm are produced in the matrix. Macro-cracks, which are generated in the products using 5–20 wt% PMMA, are prevented above 30 wt% of PMMA addition through the formation of microcellular forms with open pore sizes above 0.5 mm [36]. If porous SiOC is heated further to temperatures above 1400 8C, porous SiC can be made. While the matrix is not exactly SiOC as focused in this review, the extended use of SiOC systems to make high temperature porous SiC should be appreciated. In one example, SiC ﬁllers are used to fabricate partially interconnected, open-cell porous SiC ceramics by carbothermal reduction of PSO precursor containing PMMA microbeads, followed by sintering at as high as 1950 8C [37]. The porosity of the porous SiC ceramics can be controlled within a range of 32–64% by adjusting the sintering

temperature and SiC additive to polymer-derived SiC ratio while keeping the sacriﬁcial template content to 50 vol% [38]. The porous SiC formation can be envisioned as three steps: (i) pyrolysis of PSO at 1100 8C, which leads to the conversion of PSO to SiOC; (ii) carbothermal reduction of SiOC and C mixture at 1450 8C, which converts the mixture to a SiC ceramic; and (iii) liquid-phase sintering of the SiC using Al2O3–Y2O3–CaO as a sintering additive at 1800–2000 8C [39–41]. As Fig. 10 shows [39], SiOC transforms into SiC with large pores in the tens of microns and porous pore walls. With the increase of the Al2O3–Y2O3–CaO sintering aid, the pore walls changes into dense microstructures. While expandable and nonexpendable additives are separately discussed here, process-wise there is no restriction on using both types of additives in one system. Porous SiC ceramics with a duplex pore structure are successfully fabricated using both kinds of pore

Fig. 10. Microstructures of the porous SiC ceramics from PSO precursors, carbon black, and hollow PMMA spheres with diameters ranging from 15 to 25 mm, sintered at 1950 8C for 4 h in argon: (a) Al2O3–Y2O3–CaO 2 wt%, (b) Al2O3–Y2O3–CaO 5 wt%, (c) Al2O3–Y2O3–CaO 10 wt%, and (d) Al2O3–Y2O3–CaO 20 wt% [39]. (Reprinted from J. H. Eom, Y. W. Kim, and M. Narisawa, Processing of porous silicon carbide with toughened strut microstructure, Journal of the Ceramic Society of Japan, 118 [1377] 380–83 (2010), with permission from The Ceramic Society of Japan.)

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

formers, expandable microspheres and non-expandable PMMA spheres, thereby resulting in different pores sizes. The duplex pore structure consists of large pores derived from the expandable microspheres and small windows in the strut area replicated from the PMMA spheres. The openness of the strut area in porous ceramics can be effectively adjusted by using these two kinds of templates that result in different pore sizes [42]. 2.3.3. Carbon template At higher temperatures than that for organic pore former decomposition, highly porous carbon can be used as the template for SiOC formation through carbon removal during the pyrolysis. The pore sizes created can be orders of magnitude smaller than those from conventional pore formers discussed above (nanometers vs. hundreds of microns). Polycarbosilane as a ceramic precursor and mesoporous carbon CMK-3 as a hard template are mixed and cast into bulk samples. Reactive gases including air and ammonia are simultaneously employed to incorporate oxygen or nitrogen into SiC ceramics and to remove the carbon template and excess carbon deposits. The ordered mesoporous SiOC ceramics have reasonable speciﬁc surface areas (200–400 m2/g), pore volumes (0.4–0.8 cm3/g), and pore size distributions (4.9– 10.3 nm) [43]. Also, highly ordered mesoporous SiOC monoliths are synthesized using liquid PHMS as the starting preceramic polymer and mesoporous carbon CMK-3 as the template, crosslinking at 150 8C under humid air and subsequent pyrolysis at 1000 or 1200 8C under argon atmosphere. The carbon template is removed by thermal treatment at 1000 8C in an ammonia atmosphere. The as-prepared SiOC monoliths exhibit crack-free, ordered 2-dimentional hexagonal p6 mm symmetry with high speciﬁc surface areas of 616 to 602 m2/g (Fig. 11) [44]. Moreover, the porous SiOC monoliths possess good compressive strength and oxidation resistance. 2.3.4. Silica template Even though most studies focus on pore forming agents that can be completely removed before or during pyrolysis, in reality pores can also be generated after pyrolysis by selective etching. In this case, SiO2 can be a very desirable pore former as HF etching can effectively remove SiO2 (both amorphous and crystalline) and B2O3 (if present). Similarly, phase-separated SiO2 during pyrolysis can be removed to form pores. At least three processes can be used to introduce SiO2 into the SiOC matrix. One is direct SiO2 particle

31

dispersion into the PSO liquid before crosslinking. The second is porous SiOC surface coating of SiO2. The third is using SiO2 monosphere regular packing as a template. As an example for the ﬁrst case, we have made porous SiOC bulk ceramics with micro-/meso-size pores by pyrolysis of a commercial PSO (SPR-684 Polyramic1, Starﬁre Systems Inc.) with SiO2 nanoparticles as the ﬁller and pore-forming agent. In addition to forming pores through post-pyrolysis SiO2 etching, the SiO2 ﬁller also promotes SiOC phase separation during the pyrolysis. The ﬁller content and pyrolysis temperature have important effects on phase compositions and microstructures of porous SiOC ceramics. The resulting porous SiOC bulk ceramic has a maximum speciﬁc surface area of 822.7 m2/g and an average pore size of 2.6 nm, and consists of free carbon, SiC, and SiOC phases. Fig. 12 shows two types of SiOC regions. One is the porous regions with 10–50 nm pores. These are the porous structure resulted from the HF etching removal of SiO2 nanoparticle additives. The other is the much denser regions with nanoparticle morphology. These regions are also porous with pore sizes in the 2–5 nm range. These pores result from the HF removal of phase separated SiO2 nanodomains [45] (see Section 3 for more detailed description of SiOC nanostructures). Using porous SiOC as the framework, hierarchical meso- and macro-porosity is obtained by depositing a mesoporous SiO2 coating on the surface of an open-cell SiOC foam. The walls of the macroporous substrate are covered uniformly by a mesoporous SiO2 layer with a thickness of 1–5 mm. The mesopores exhibit a narrow size distribution and appear to be interconnected through micropores, giving rise to an interconnected 3D porous network in the coating. The uncoated SiOC foam possesses a speciﬁc surface area of 3.4 m2/g, and pores with sizes <100 nm are not detected. For the coated sample, the speciﬁc surface area is 565 m2/g, largely due to the surface area contribution from the mesoporous coating; the pore size distribution is narrow and the pores are mostly 3–5 nm [18]. Monodispersed spherical SiO2 particles can also be regularly packed to serve as a template. An incipient wetness method is used to ﬁll the interparticle voids of SiO2 spheres of several tens of nanometers. The spheres have a very narrow size distribution in the mesoscale that could be replicated as pores by pyrolysis in an inert atmosphere and subsequent HF etching. Using a pyrolysis temperature between 700 8C and 1300 8C, the speciﬁc surface area decreases from 592.1 m2/g to 433.1 m2/g [46]. After pyrolysis and

Fig. 11. (a) Photographs of the SiOC with CMK-3 as the template after 1200 8C pyrolysis, (b) TEM images of the SiOC after 1000 8C pyrolysis [44]. (Reprinted from Microporous and Mesoporous Materials, 147, X. Y. Yuan, H. L. Jin, X. B. Yan, L. F. Cheng, L. T. Hu, and Q. J. Xue, Synthesis of ordered mesoporous silicon oxycarbide monoliths via preceramic polymer nanocasting, 252–58, 2012, with permission from Elsevier.)

32

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

controlled pyrolysis of MoSi2 ﬁlled methyltriethoxysilane gel precursors at 1200 8C and 1350 8C. The MoSi2 ﬁller is mostly inert in the SiOC matrix. The ceramic samples display a high amount of residual porosity (>40%) even though the pores are mostly in the micron range [47]. 3. SiOC nanostructure model

Fig. 12. SEM cross section images of SiOC ceramic with 30 wt% SiO2 additive pyrolyzed at 1300 8C after the HF etching.

HF etching of SiOCs, highly uniform mesopores with diameters in the range of 40 nm to 50 nm can be achieved. The materials have pore volumes as high as 1.41 cm3/g (Fig. 13) [46]. From the above discussion, it is clear that SiOC is a versatile system that can accommodate different permanent ﬁllers in the SiOC matrix, as long as the interactions between the permanent ﬁllers and the pore formers do not lead to signiﬁcant structural or compositional changes. Because of the similarities in the approaches of introducing permanent ﬁllers into the SiOC matrix as those for the pore formers, they will not be discussed in detail. For example, MoSi2/SiOC ceramic components are produced by

So far, the discussion has mostly focused on pore formation through the addition of foreign species in the SiOC interconnecting matrix. The SiOC phase transformation and microstructure evolution are taken as a given without detailed examination of its composition, composition evolution, and possible pore formation. In reality, SiOC is an intriguing system that can produce a huge variety of compositions and microstructures, depending on polymer precursors and crosslinking/pyrolysis conditions. In addition, different additives can be added to the PSO precursors and mixed at the molecular level to form new porous microstructures. This kind of additives is fundamentally different from the additives used above as they have the ability of molecular level mixing with PSO precursors. The evolution of the precursors and additives, as the material changes from organic into ceramic state, occurs via solid state radical reactions with the formation of gaseous by-products such as methane and hydrogen [48]. The oxygen and some of the carbon atoms are bonded randomly to silicon in a three-dimensional covalent structure: SiO4, CSiO3, C2SiO2, C3SiO, SiC4. Therefore, excluding pores, the matrix of the system can be described as a composite formed by a SiOC matrix in which the free carbon is dispersed. The chemical composition in the resulting SiOC may be controlled by varying the species and structure of monomeric organosiloxane precursors as well as the temperature and atmosphere of the pyrolysis process. To control

Fig. 13. SEM images of SiO2 nanospheres (A), and porous SiC pyrolyzed at 700 8C (B), 1000 8C (C), and 1300 8C K (D) [46]. (Reproduced from C. Hoffmann, T. Biemelt, A. Seifert, K. Pinkert, T. Gemming, S. Spange, and S. Kaskel, Polymer-derived nanoporous silicon carbide with monodisperse spherical pores, Journal of Materials Chemistry, 22, 24841– 47, 2012, with permission of The Royal Society of Chemistry.)

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

pore formation and evolution, general SiOC nanoscale structure and phase distribution must be understood ﬁrst. The overall SiOC structural evolution process can be outlined in Fig. 14. At low pyrolysis temperatures (800–1100 8C), SiOC consists of a homogeneous network of mixed Si–C–O tetrahedra and free carbon species. The excessive free carbon may lead to improved resistance against creep deformation and the reported rise in glass transition temperature Tg, viscosity h, Young’s modulus E, and hardness H, all of which increase with increased incorporation of carbon into the glass network structure [49]. The SiOC microstructure is generally amorphous and featureless, and more often than not, dense. There are two distinct contributions at 1.61– 1.62 A˚ and 1.88–1.89 A˚ bond lengths for the ﬁrst coordination shell assigned to Si–O and Si–C atomic pairs. The average Si–O–Si bond angle values of 140 to 1448 are smaller than that found for amorphous SiO2 and decrease with the free carbon amount [50]. Because of the carbon-rich nature of PSO-derived SiOC, inbetween the SiOC clusters exists a graphitic layer composed of turbostratic carbon. Carbon bonds to either silicon or to itself, but not to oxygen. The carbon bonded to silicon is called sp3 or carbidic carbon and the carbon in graphene is sp2 or graphitic carbon. The silicon, oxygen, and carbon content inﬂuences on the SiOC composition can be understood based on the phase diagram (Fig. 15). The line connecting the SiC and SiO2 phases represents compositions in which all carbon atoms are bonded to silicon and no excess or free carbon is present. The general stoichiometry on the SiC–SiO2 line is SiCxO2(1x). Rewriting it as xSiC + (1x)SiO2 shows that the x value is a direct measure of the relative proportion between Si–C and Si–O bonds present in the glass network. SiOC glasses studied so far (roughly inside the shaded region in Fig. 15) usually contain excess carbon, their composition falls inside the C– SiO2–SiC triangle, and must be expressed as SiO2(1x)Cx + y Cfree, where x + y is the molar ratio of the carbon relative to the silicon content [50]. TH/DH ratio of 2 theoretically yields a stoichiometric SiOC glass without the residual free carbon phase. TH represents HSiO3 units, derived from HSi(OR)3, while DH stands for CH3SiHO2 units, prepared from a CH3SiH(OR)2 precursor [49]. For example, triethoxysilane HSi(–OEt)3 is TH and methyldiethoxysilane (CH3)HSi–(OEt)2 is DH [7]. At 1100–1300 8C, amorphous SiO2 nanodomains form; the microstructure of the SiOC matrix consists of clusters of SiO2 tetrahedra that are encased in mixed bonds of SiOC; these units are then embedded in a cage-like network of sp2 graphitic carbon that forms an interconnected network. The excessive carbon leads to lower thermal stability and inferior oxidation resistance [51,52]. Consequently, much attention in developing SiOC applications has focused on minimizing the over-stoichiometric amount of carbon as expressed in SiO2(1x)Cx + y Cfree. The resistance against SiO2 devitriﬁcation is due to their lower absolute enthalpy values compared to SiC, graphite, and quartz. Around the SiO2 nanodomains, SiOC mixed bonds form a 1–2 nm

33

thick interlayer. The overall composition of SiOx0 Cy0 partitions as follows: SiOx0 Cy ! SiOx0 Cð10:5x0 Þ þ Cy0 ð10:5x0 Þ

(3)

where the ﬁrst term on the right refers to the predominantly tetrahedrally bonded stoichiometric composition and the second term to the free graphitic carbon. It should be noticed that SiOx0 C(10.5x0 ) in Eq. (3) is equivalent to SiO2(1x)Cx expressed earlier. A stoichiometric SiOC sample reveals a phase separation process: SiCx O2ð1xÞ ! xSiC þ ð1xÞSiO2

(4)

which starts at 1200 8C and results in the formation of nanosized SiC precipitates ﬁnely dispersed in the amorphous SiO2-rich matrix. The phase separation into crystalline SiC and amorphous SiO2 progresses with increasing temperature [53]. Based on the assumption of the overall carbon-rich composition, the size of the SiO2 nanodomains dSiO2 and the thickness of the graphitic cell walls have been predicted as follows [54]: V SiO2 2x dSiO2 ¼ 6 1 (5) ASiOC 2 x0 V SiO2 is the effective volume per SiO2 tetrahedral, ASiOC is the average area occupied by the interfacial tetrahedra, x is deﬁned as in SiCxO2(1x). The effective width of the graphitic interface (the factor of one half accounts for two domains abutted against the same graphitic layer) dW is 2V x 2y0 dW ¼ W 1 (6) ASiOC 2 x0 Vw is the volume occupied per carbon atom in the interdomain walls. The mixed Si–O–C bonds sandwiched between the SiO2 domains and the graphitic cell walls are assumed to be isolated tetrahedra, equally populated by SiO1C3, SiO2C2, and SiO3C1. For NMR analysis, the peak positions are: SiO4 (107 ppm), SiCO3 (72 ppm), SiC2O2 (34 ppm), SiC3O (1 ppm), and SiC4 (16 ppm) [50]. For high carbon SiOC, our TEM work clearly shows that the sp2 carbon (line-like) forms a continuous network (Fig. 16(a)). At higher temperatures (1300–1450 8C), the amorphous network undergoes further phase separation, which generates two interpenetrating nano-sized networks, namely a SiO2-rich phase and a C-rich phase, the latter being comprised of b-SiC nanocrystals and graphite [48,54–57]. High resolution TEM shows that b-SiC crystal precipitates in the nanometer range at temperatures above 1250 8C (Fig. 16(b)). The SiO2 domains are surrounded by graphitic cell walls. The interface between the SiO2 domains and the graphitic cell walls contains SiC nanocrystals and mixed SiOC units, SiCxO4x, 1 < x < 4 [49,58,59]. In addition, the onset for crystallization of b-SiC in SiOC glasses is dependent on the amount of carbon. For the glasses with lower C content, 5 wt% of b-SiC crystalline phase can be detected at 1200 8C, whereas for

Fig. 14. Overall SiOC structural evolution with temperature.

34

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

Fig. 15. Si–O–C phase diagram showing the binary stable thermodynamic phases (SiO2 and SiC) at high temperatures. The composition of the stoichiometric SiOC glasses (SiCxO2(1x)) are on the SiC–SiO2 tie-line. Compositions of SiOC glasses reported in the literature are roughly plotted in the shaded area.

the carbon-rich glass the pyrolysis temperature must be increased up to 1400 8C to promote crystallization [50]. As pointed out earlier, the composition of the SiOC network depends to a large extent on the chemical natures of its polymeric precursors. Different siloxane precursors produce different stoichiometry SiOC networks. The amount of free carbon formed depends on the nature of the organic groups. A variety of organic functionalities can be included or excluded to tailor the crosslinking of the polymers. Saturated groups such as methyl lead to moderate free carbon, whereas unsaturated groups (vinyl or phenyl) lead to higher free carbon contents [7]. Vinyl groups can enable radical-initiated cure of the polymers for shape retention and bulk integrity [60]. In silsesquioxane derived ceramics, linear dependence between the amount of free carbon and the amount of phenyl groups exists. Carbon-containing species formed during the ceramization processes evolve and escape through the pores. The transient porosity created in the intermediate stages of the pyrolysis is associated with the elimination of gaseous species (CH4, H2, C6H6, etc.) [61,62]. These gas generation reactions eliminate the organic moieties in the ceramization, lead to redistribution reactions for the formation of the SiOC network, and can potentially create pores [63]. To avoid structure collapse and closing of the porous structure, a careful control of the pyrolysis heating rate is required [62]. Until now, the discussion has been evolving around the SiOC system. Due to the glass forming ability of B2O3, SiBOC can also be obtained. For the SiBOC glasses, their structure and hightemperature evolution are reported to be similar to those found for SiOC. At 1200 8C the amorphous network is formed by a random mixture of silicon and boron oxycarbide units, SiBOC samples cannot be etched. This result suggests that, at this temperature, the glassy borosilicate clusters are not yet formed and the amorphous network is still built up by a combination of mixed SiOC (SiO2C2, SiCO3) and boron oxycarbide (BOC2, BO2C) units. At higher temperatures, redistribution reactions between Si–O/B–O and Si–C/B–C bonds are active [64]. The presence of boron in the glass

Fig. 16. SiOC microstructure showing continue carbon phase (a) and crystallized SiC (b) for the SiOC samples pyrolyzed at 1300 8C.

network accelerates the phase separation into SiC4 and borosiloxane (BOSi3) units. At high temperatures (1400 and 1500 8C), the presence of boron in the SiOC glass favors the consumption of the mixed silicon (and mixed boron in SiBOC) units with a phase separation into SiC4 and borosilicate glass clusters. Boron in SiBOC glasses, besides enhancing the SiC crystallization and the borosilicate nanocluster formation, has another important effect: it promotes the graphitization of the graphene-stacking layer, leading to an increase of the thickness of carbon in the direction orthogonal to the carbon sp2 basal plane, and decreases the interlayer spacing, of the graphitic nanocrystals. The thickness of the graphitic carbon layer is 2.9 nm for SiOC and reaches 5.3 nm for SiBOC at 1500 8C [65]. Despite the large number of papers that have been published in the last 10 years on the synthesis and characterization of SiOC glasses, fundamental issues remain to be resolved in order to

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

achieve true molecular level tuning of ceramized polymer precursors, such as (i) the inﬂuence of the crosslinking and pyrolysis conditions on the ﬁnal composition and structure, (ii) the high temperature (T 1000 8C) crystallization and/or phaseseparation process, and (iii) the ability of the pyrolyzed nanostructures being modiﬁed by different chemical agents. Also, SiOC glass network structure, composition, and thermal stability after pyrolysis strongly depend on the molecular structure of the polymer precursors subjected to crosslinking and any potential additives used. There are three key factors to be examined: different precursors that have varying degrees of carbon and silicon contents, similar Si:O:C ratio precursors with different polymer architectures, and different hydrosilylation agents [50] (see the deﬁnition for hydrosilylation in Section 4). 4. SiOC formation process To generate high speciﬁc surface area SiOCs (>1000 m2/g), pores formed must be less than 10 nm, desirably 2–5 nm. PSO precursors have the ﬂexibility to be tailored during precursor mixing (molecular level homogeneity), crosslinking (polymer network formation), pyrolysis (phase separation, selective removal of network species for pore formation), and post-pyrolysis treatment (selective removal of nanodomains for pore formation), thus achieving high speciﬁc surface area (>1000 m2/g) and mesoporous (2–5 nm) microstructures. Transient micro-/meso-pores are formed during the early stages of the SiOC precursor transformation from the polymeric to the ceramic phase by the evaporation of low molecular weight molecules and oligomers, showing its maximum presence at around 600 8C. Then, the transient pores (typically less than a few nanometers in size) are dramatically reduced or completely eliminated during the pyrolysis at higher temperatures. This micro-/meso-porosity elimination is associated with signiﬁcant shrinkage and/or densiﬁcation of the evolved SiOC systems regardless of its composition [66]. To create high surface area and mesoporous SiOCs, the ﬁrst necessary condition is that the SiOC network after crosslinking must be stable (again defect formation and collapse). Conventional sol–gel process does allow porous structure formation yielding stoichiometric or nearly stoichiometric SiOC upon pyrolysis. However, without any templates or excessive carbon, the network has low resistance again defect formation, along with the problem of pore shrinkage/collapse during the pyrolysis. Because of the high tendency for drastic shrinkage or even bulk structure collapse, there is little room left for nanoscale pore size tuning. During pyrolysis up to 600 8C, the network quickly collapses and the speciﬁc surface area/porosity decreases drastically. In the high temperature pyrolysis process, there is little possibility to modify the network for new pore creation. For example, highly porous aerogels have been synthesized using methyltriethoxysilane, (bistriethoxysilyl) methane, and 1,2 (bistriethoxysilyl) ethane as the starting precursors and dried at ambient pressure, which are later pyrolyzed for the synthesis of porous SiOC glass in ﬂowing argon, at 100 ml/min. The aerogels possess very high surface areas and pore volumes; and a maximum speciﬁc surface area of 1131 m2/g with a pore volume of 1.29 cm3/g is obtained. Upon pyrolysis, however, the speciﬁc surface area is only retained at 452 m2/g with a pore volume of 0.33 cm3/g [67]. On the other hand, pyrolysis of higher carbon-containing SiOC precursors leads to microstructures that can inhibit the removal of the SiO2 domains during etching and thus surface area generation, even for microcellular foams [15]. Phenyltrimethoxysilane has been hydrolyzed or cohydrolyzed with tetramethoxysilane to make aerogels and xerogels. Porous SiC/SiOC glasses are prepared by further pyrolyzing these gels in ﬂowing argon up to 1500 8C.

35

Pyrolyzing 25 mol% phenyltrimethoxysilane-75 mol% tetramethoxysilane aerogel in argon at 1000 8C only results in porous SiOC glass of 581 m2/g speciﬁc surface area. Pyrolyzing pure phenyltrimethoxysilane gels at 1400–1500 8C produces porous SiC/C/SiOC composites with 400–500 m2/g speciﬁc surface areas [68]. Isothermal annealing of three dimensionally crosslinked polycyclic copolyorganocyclocarbosiloxanes prepared from alkoxy derivatives of cyclic organodi- and tri-silylenemethylenes in an inert atmosphere at 450, 500, and 600 8C yields preceramics. However, with increasing annealing temperature, the pore size drops to the extent that benzene molecules cannot penetrate the pores [69]. Nanoporous monolithic SiOC aerogels are synthesized by the reaction of tetraethoxysilane with dimethyldiethoxysilane via sol–gel process, supercritical drying, and pyrolysis at 1200 8C. The density and speciﬁc surface area of the resulting porous materials are only about 0.3 g/cm3 and 208.6 m2/g, respectively [70]. Fortunately, PSO precursor crosslinking and pyrolysis can avoid the network collapse problem and further strengthen the network during the high temperature ceramization process. The crosslinking process of PSOs is a crucial step in the fabrication of porous SiOC. Fundamentally, the crosslinking is achieved via condensation or free radical initiation. A speciﬁc process, hydrosilylation, can enable the formation of Si–C–C–Si or Si–C–Si bridges in the polymeric network by adding Si–H bond to multiple carbon–carbon or carbon–heteroatom bonds. Hydrosilylation of PSO is a versatile reaction that provides many beneﬁts in relation to other crosslinking mechanisms: It can occur at room temperature in the presence of a catalyst, or it can be accelerated by heat. Furthermore, the shrinkage in the ‘‘green body’’, the undesired feature usually observed in the sol–gel process, is minimized. However, the most important feature of the hydrosilylation reaction is the incorporation of carbosilane or related bridges (Si–Cx–Si) in the polymeric network. Carbosilane or related bridges allow for better control of the carbon content without decreasing the crosslinking density of the PSO networks, which also provides an option for controlled O/Si ratios. The presence of these bridges has also been responsible for the better incorporation of carbon in the oxycarbide glasses, compared with those derived from alkoxysilanes containing Si–H and/or Si–CH3 groups by the sol–gel process. In contrast to the hydrolysis-condensation reactions of alkoxysilanes commonly utilized during the sol–gel process, hydrosilylation reactions enable the formation of Si–C–C– Si or Si–C–Si bridges in the polymeric network, contributing to the generation of a greater amount of SiC4 molecular sites in the ﬁnal product [1]. The unsaturated bonds on the polymer chains dominate the crosslinking extent and the network stability. In our work, a commercial PSO was mixed with 1 wt% platinum-divinyl tetraethyl disiloxane complex (crosslinking catalyst) by ball milling for 30 min. The cross-linking process was carried in an oven at 120 8C for 6 h. After crosslinking, transparent polymer samples were obtained (Fig. 17(a)) [71]. At temperatures between 600 and 1000 8C, extensive C–H, Si–C, and Si–O bond cleavage occurs for the crosslinked PSO, leading to ceramics consisting of amorphous SiOC. Various redistribution reactions between Si–O, Si–C, and Si–H bonds occur and hydrocarbons (mainly CH4) and hydrogen escape from the system. Through heating rate and atmosphere control, this process offers great ﬂexibility in controlling the loss of low molecular weight species and their fragmentation during ceramization, and thus the stoichiometry of the SiOC matrix while retaining the shape of the crosslinked polymers [72]. At T > 1000 8C, SiOC undergoes phase separation, which leads to a multi-phase material consisting of a-SiO2 (‘a’-amorphous) and free carbon (Fig. 17(c)). With further temperature increase (1200 8C), SiC forms (this is especially the case when the C

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

36

Fig. 17. Photographs of (a) crosslinked pure PSO, (b) crosslinked PSO with SiO2 ﬁller, (c) pyrolyzed SiOC before etching, (d) pyrolyzed SiOC after etchin. The PSO is from commercial polysiloxane RD-684 Polyramic [45]. (Preparation of micro-/mesoporous SiOC bulk ceramics, J. Li, K. Lu, T. Lin, K. Shen, Journal of the American Ceramic Society, 98 [6] 1753–1761 (2015), John Wiley and Sons.)

content is not too excessive). The formation of the phase separated SiO2, SiC, and free carbon can be expressed as: Six Cy Oz ! a-SiO2 þ SiC þ C

(7)

Subsequently, carbothermal reaction of SiO2 occurs [73]: a-SiO2 þ 3C ! SiC þ 2COg

(8)

A potential side reaction is: 2SiO2 þ SiC ! 3SiOg þ COg

(9)

At >1300 8C, SiO2 may crystallize (depending again on the free C content), carbothermal reaction of SiO2 dominates, and b-SiC can be unambiguously identiﬁed: SiO2 þ 3C ! b-SiC þ 2COg

Fig. 18. Molar diagram showing the composition of the SiOC glasses. The full lines indicate the predictable domain size, dSiO2 , and the dashed lines the interdomain boundary, dW. A has the overall composition of SiC2.24O1.71, B of SiC1.18O1.71, C of SiC0.68O1.51, and D of SiC0.31O1.39, all by starting with different precursors [58]. (Preparation of ultrathin-walled carbon-based nanoporous structures by etching pseudo-amorphous silicon oxycarbide ceramics, R. Pena-Alonso, G. D. Soraru, and R. Raj, Journal of the American Ceramic Society, 89 [8] 2473–80 (2006), John Wiley and Sons.)

(10)

The exact onset temperature for the carbothermal reaction forming b-SiC depends on the carbon content [50] and the porosity (or the speciﬁc surface area) of the SiOC specimens (because in this case the loss of the carbon-containing gaseous species is facilitated). The formation of carbosilane bridges in the PSO networks during the crosslinking contributes a larger amount of carbidic sites to the ﬁnal products. An increase in the pyrolysis temperature leads to a distribution of silicon sites and crystallization proﬁles (crystallization temperature, crystallite size) that depend on the molecular architecture of the PSO precursors [1]. A commercial PSO RD-684 Polyramic1 (Starﬁre Systems Inc.) has been used by us to make crack-free bulk SiOC. This polymer precursor is C-rich and can result in as high as 86 mol% of free carbon along with >9 mol% SiO2 after pyrolysis [74]. However, it lacks well-deﬁned molecular structure and contains a combination of –H, aryl, alkyl, and vinyl groups. As a result, the SiOC network exhibits simultaneous transformation of different PSO species. Systematic study of SiOC formation from better-deﬁned PSO precursors should be able to systematically address this issue. By using PSOs with varying C-rich side groups, the ﬁnal SiOC composition in the C–SiC–SiO2 region of the phase diagram can be widely adjusted. For example, based on a predictive model and experimental work, it has been reported that 75% of the vinyl groups will stay and contribute two carbons to the ceramic; 67% of the phenyl groups will stay and contribute six carbons to the ceramic; and 60% of the methyl groups will stay and contribute one carbon to the ceramic [75]. This way different Si:O:C compositions can be produced for pore size and composition modiﬁcation. Based on the Si–O–C diagram (Fig. 18), different PSOs can be selected to tailor the SiO2 domain size, the graphitic cell wall thickness, and the SiC domain size (at higher temperatures) [58]. Lower oxygen content precursors produce smaller SiO2 domains from 1 to 6 nm,

higher silicon content precursors produce thinner cell walls of 0.35 nm to 2.1 nm. The preparation of SiOC with an O/Si ratio equal to 1 is not straightforward, although this ratio can be found in a linear polydiorganosiloxane, such as PDMS, in which pyrolysis leads to 0% residue. For even higher SiO2 content, tetraethylorthosilicate will have to be used as one of the starting materials [76]. This will correspondingly move the SiOC composition to the SiO2 corner of the C–SiC–SiO2 region (Fig. 18) [58]. Because of the more Si-rich nature, the SiO2 cluster size will depend on the oxygen content, which provides ﬂexibility to tune the pore sizes and the speciﬁc surface areas. If the oxygen content can be decreased during the PSO pyrolysis, the composition will shift to the SiC corner. This will likely decrease the pore size and offer more room to tailor the SiOC network speciﬁc surface area. With free carbon content increase, the SiOC composition will shift to the carbon corner and this will increase the wall thickness of the graphitic cells [54]. The pyrolysis heating rate will have direct inﬂuence on the phase separation and thus the resulting compositions as it directly inﬂuences polymer chain breakup and fugitive species release. A typical pyrolysis heating rate is 2–5 8C/min [77]. When the goal is to obtain thicker graphitic layers, higher carbon content PSO should be used and the crosslinking should be enhanced, the latter can be achieved by increasing the crosslinking temperature and duration as well as the catalyst amount. When the goal is to increase the phase separation, the pyrolysis peak temperature/duration and heating rate should be further increased. This may also facilitate the gaseous species escape and bond breakage. 5. Molecular level pore formation 5.1. Direct pore formation Without using any pore formers, micro- to meso-pores can be formed during the PSO precursor crosslinking and pyrolysis processes. This is because during these processes, volatile species are generated. Their escape from the PSO/SiOC matrix generates very small pores, which may be preserved for porous SiOC material

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

formation. Because of the precursor and processing condition differences, pore formation temperature varies greatly. The speciﬁc surface area is generally modest and the pore size distribution is broad. Commercially available polymethyl(phenyl)siloxane resin is pyrolyzed from 1250 to 1450 8C under vacuum with a high heating rate of 15 8C/min [78]. Pores generated by pyrolysis at different temperatures are multimodal and have wide range distributions even though they are predominantly at mesoscale (2 < d < 50 nm). The porous structure evolution is mainly determined by the carbothermal reduction progress. Three stages are demonstrated corresponding to three consecutive stages of carbothermal processes, respectively. During the ﬁrst stage (1250–1350 8C), the ceramic has relatively small speciﬁc surface areas. During the second stage (1400–1450 8C), as the carbothermal reaction proceeds to a larger extent, the speciﬁc surface area and total pore volume increase accordingly. During the third stage (above 1450 8C), the speciﬁc surface area and total pore volume decrease rapidly because of the partial sintering of PSO-derived SiC. With the increase of the pyrolysis temperature, micropores decrease from 42% to 4% due to the sintering of the SiOC matrix and the coarsening of the mesopores. In contrast, the macropore volume increases from 7% to 23%. The pyrolysis of polyborodiphenylsiloxane has been studied at temperatures up to 1600 8C in nitrogen atmosphere [79]. Boron is retained in the resulting ceramic at all temperatures. Above 1200 8C, both nanoscale b-SiC and an amorphous phase develop while at temperatures greater than 1400 8C a signiﬁcant amount of mesopores starts to appear, with speciﬁc surface area and pore volume reaching maximum values at around 1500 8C (Fig. 19) [79]. However, the porosity obtained at high pyrolysis temperatures is unstable, decreases at even higher pyrolysis temperatures, and disappears completely after an oxidizing atmosphere (50 cm3/ min of 50% air in helium) treatment at temperatures above 400 8C. The reason can be explained as follows. The mesopores are present in carbon-rich areas. At 1200 8C pyrolysis temperature, the carbon phase is largely inaccessible to the oxidizing gas as it has not yet separated out to any great extent. However, the sample pyrolyzed at 1575 8C releases a large amount of CO2 when treated at around 350 8C and such CO2 release reaches a broad maximum at around 850 8C. The CO2 release decreases when the atmosphere is switched to pure helium. The surface area of the material decreases from 290 m2/g to 165 m2/g even though the pore volume increases from 0.49 to 0.54 cm3/g. The pore size distribution broadens with the average pore size increasing from around 4 nm to 11 nm [79]. Carbon burn-off produces larger pore sizes. Commercial methyl–phenyl PSO powder with a melting range of 50–90 8C (Silres1 H44, Wacker Chemie AG) and pure methyl PSO powder

Fig. 19. Plot of BET surface area against pyrolysis temperature in an inert atmosphere [79]. (Reprinted from Microporous and Mesoporous Materials, 99, H. M. Williams, E. A. Dawson, P. A. Barnes, B. Rand, and R. M. D. Brydson, The development and stability of porosity formed during the pyrolysis of polyborodiphenylsiloxane, 261–67, 2007, with permission from Elsevier.)

37

with a melting range of 35–55 8C (Silres1 MK, Wacker Chemie AG) can be pre-crosslinked by heat treatment in air at 80 8C, 140 8C, and 200 8C, respectively. After that it is ground and ﬁlled into polypropylene/polyethylene tubes, and aged in ammonia (25%) for 5 days to increase the degree of crosslinking. After the aged sample is pyrolyzed under ﬂowing nitrogen at temperatures between 450 and 700 8C with 4 h holding, speciﬁc surface area from 419 to 534 m2/g is obtained even though hydrocarbons remain partly in the matrix because of the low pyrolysis temperatures [80]. Just as for the conventional pore formation systems, additives can be introduced into the SiOC for mesoscale porous material formation. Mesoporous SiOC/ZnO nanocomposite is prepared by simple mechanical incorporation of ZnO nanopowders into commercial vinyl-functionalized PSO (XP RV 200, Hanse Chemie, Germany) followed by pyrolysis at 700 8C for 2 h. An amorphous microporous SiOC ceramic (pore size <2 nm) forms at 700–900 8C pyrolysis temperatures with 0–20 wt% ZnO content. The SiOC ceramic obtained at 700 8C has the highest speciﬁc surface area of 398 m2/g with the pore size of <2 nm while the SiOC/20 wt% ZnO nanocomposite prepared at 700 8C shows the speciﬁc surface area of 220 m2/g with an average pore size of 4 nm [81]. The adsorption–desorption test shows the typical shape of type I isotherm, corresponding to a microporous structure. In contrast, the adsorption–desorption isotherm of the SiOC/20 wt% ZnO nanocomposite shows low adsorption in the P/Po range of 0–0.8 but increases continuously and raises steeply at P/Po = 0.8–1.0, characteristic of a type IV isotherm behavior related to mesoporous materials (Fig. 20) [81]. For the same precursor, porous SiOC/ TiO2 and SiOC/N-doped TiO2 ceramic composites can also be obtained. The speciﬁc surface areas of the as-prepared SiOC ceramic and SiOC/TiO2 and SiOC/N-doped TiO2 ceramic composites decrease with increasing pyrolysis temperatures due to the gradual collapse of small pores. Transformation from microporous SiOC ceramics (pore-size <2 nm) to mesoporous SiOC/TiO2 and SiOC/N-doped TiO2 ceramic composites (pore-size 4 nm) is observed [82]. From the above studies, it can be noticed that the speciﬁc surface area of the porous SiOC obtained through direct precursor pyrolysis is rather low; it is not a particularly useful approach to form high surface area SiOC materials. The stability of the matrix at

Fig. 20. N2 adsorption-desorption isotherms of the SiOC ceramic and of the SiOC/ 20% ZnO nanocomposite synthesized at 700 8C for 2 h in argon [81]. (Reprinted from Microporous and Mesoporous Materials, 151, M. Hojamberdiev, R. M. Prasad, K. Morita, M. A. Schiavon, and R. Riedel, Polymer-derived mesoporous SiOC/ZnO nanocomposite for the puriﬁcation of water contaminated with organic dyes, 330–38, 2012, with permission from Elsevier.)

38

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

high temperatures and the resistance to oxidation, among others, may also be questionable. However, this step is often necessary for porous SiOC formation before post-pyrolysis treatment. 5.2. Different polysiloxane precursor architectures An effective strategy for producing highly porous ceramics from preceramic polymers is mixing of precursors with different molecular architectures. Because of the different distributions in the initial silicon, carbon, and oxygen species, the gas generation locations during the initial stage of pyrolysis (due to the break-up of the polymer chains) and the resulting SiOC, SiO2, and SiC distributions (due to different molecular architectures) are different. This can lead to different pores. For example, a silicone resin, polysilsesquioxane, typically possesses a cage-like structure and has a high ceramic yield (>65 wt%) upon pyrolysis, while a linear structure PSO has a low ceramic yield (<30 wt%) after pyrolysis. Besides a very different weight loss, there are also differences in their relative shrinkage and the amount of gas generated (all much higher for the siloxane with a linear structure than for the silicone resin) [13]. When mixing and crosslinking these two precursors to form a green body, pores are generated during pyrolysis because of the different behaviors that the two polymers display. Two compounds: PHMS and 1,3,5,7-tetramethyl-1,3,5,7-tetracyclotetrasiloxane, have the same chemical formula of SiOCH4 and the same repeating units of –[CH3SiHO]n, apart from the three terminal –Si(CH3)3 units of the linear polymer. Thus, they have the

memory of the architecture of the preceramic network from which they are derived [48]. The stability of the SiO2 phase in the two series of SiOC glasses is different and is related to the structure of the precursors (Fig. 21) [48]. For the linear-derived SiOCs the amorphous SiO2 is stable like for the majority of reported SiOC glasses while for the cyclic-derived SiOCs partial crystallization of cristobalite has been observed. Thus, the presence of well-deﬁned siloxane rings in the starting polymer may have led to the SiO2-rich clusters and the nuclei for the SiO2 devitriﬁcation. SiO2, as to be explained in Section 6.3, can be removed for pore formation. Following the same concept, the SiOC pore sizes can be predesigned from the phase separated SiO2 nanodomain sizes by using different PSO architectures. PSO network has been prepared from polycyclic oligomer (1,3,5,7-tetramethylcyclotetrasiloxane), which has faster organic-to-inorganic transformation compared with the linear precursor PHMS [1]. In addition, other polysilsesquioxane systems, such as cages, ladders, and randomly structured networks, have been crosslinked (Fig. 22) [83–85]. The characteristics of polysilsesquioxanes are that the starting polymer is already highly crosslinked, which inﬂuences the crosslinking and phase separation, and leads to high ceramic yield upon pyrolysis. For the random PSO structure, the silicon, oxygen, and carbon distributions are more dispersed at the molecular level. During the polymer to ceramic transformation, the SiO2 cluster and graphene formation are depressed. Accordingly, the SiO2 cluster sizes should be small and the graphitic walls should be thin. At >1300 8C, the SiC formation should be facilitated as silicon and carbon are more intimately mixed. Thus, the high temperature stability of the

Fig. 21. Schematic representation of the microstructure of the resins with low amount of DVB (DVB/Si 50%), (a) PHMS-derived and (b) TMTS-derived; and gels with high amount of DVB, (c) PHMS-derived and (d) TMTS-derived [48]. (Reprinted from Journal of Non-Crystalline Solids, 356, P. Dibandjo, S. Dire, F. Babonneau, and G. D. Soraru, Inﬂuence of the polymer architecture on the high temperature behavior of SiCO glasses: a comparison between linear- and cyclic-derived precursors, 132–40, 2010, with permission from Elsevier.)

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

39

Fig. 22. Different molecular structure of polysilsesquioxanes [83]. (Polymer-Derived Ceramics: 40 Years of research and innovation in advanced ceramics, P. Colombo, G. Mera, R. Riedel, and G. D. Soraru, Journal of the American Ceramic Society, 93 [7] 1805–37 (2010), John Wiley and Sons.)

resulting SiOC can be more easily tuned. For the ladder and cage structures, the opposites should happen, which means larger SiO2 clusters, thicker graphitic walls, and less SiC formation. For the double-decker and partial cage structures, the SiOC microstructures should be in-between. Of course, the overall sizes of the SiO2 and SiC domains should still be in the 1–5 nm range. Different ratios of PSO precursors can be mixed to obtain the intermediates of the above SiOC microstructure extremes. Moreover, the stability of the amorphous SiO2 phase present in the SiOC ceramics can be used as a gauge for the microstructure evolution as it is strongly inﬂuenced by the molecular organization of the starting precursors. Two series of crosslinked PSOs are synthesized from a linear and a cyclic Si–H-containing siloxanes having the same chemical formula of SiCOH4. The crosslinking is achieved by hydrosilylation reaction with various amounts of divinylbenzene (DVB). The two resins are then pyrolyzed at 1400 8C. The stability of the amorphous SiO2 phase present in the SiOC ceramics is strongly inﬂuenced by the molecular organization of the starting precursors. The presence of siloxane rings in the cyclic-derived PSO decreases the stability of the amorphous SiO2 and promotes the crystallization of cristobalite [48]. For the linearderived SiOCs, the amorphous SiO2 is stable as for the majority of reported SiOC glasses. Thus, the presence of well-deﬁned siloxane rings in the starting polymer may lead to SiO2-rich clusters in the corresponding SiOC glasses [48]. Two types of PSO polymers, a linear PHMS (MW = 1900) and a cyclic cyclohydromethylsiloxane (MW = 240), are used to form SiOC. Each polymer precursor is mixed with platinum divinyltetramethyldisiloxane catalyst and DVB (technical 80%, mixed isomers) before crosslinking. The size distribution of the SiO2 clusters in the SiOC nanostructure is slightly larger for the cyclohydromethylsiloxanederived SiOCs than for the PHMS-derived ones. This observation is in agreement with the greater amount of micropores in the PHMSderived SiOC glass. Moreover, the amorphous SiO2 phase in the cyclohydromethylsiloxane-derived SiOC more readily crystallizes than the PHMS-derived glass. This difference can be explained by

assuming that the siloxane rings present in the cyclohydromethylsiloxane-derived polymer precursor serve as an initial building block for the development of SiO2 in the resulting oxycarbide glass, thereby facilitating both the formation of larger SiO2 clusters and their crystallization into cristobalite. These results reinforce the concept that the nanostructure/porosity and the properties of the resulting SiOC materials are strongly dependent on the architecture of the starting preceramic polymer [86]. It should be pointed out that the above studies are focused on the entire resulting SiOC systems. For the speciﬁc SiOC phase in a given system, it has been argued that the SiOxC4x units do not come from additional SiO2 or SiC phases. At higher temperatures, the crystallization of SiC displaces the equilibria toward the formation of SiC4 and SiO4 building units. The environment of the Si atoms in the silicon oxycarbide phase is dependent on the O/Si ratio of the glass and on the pyrolysis temperature, but not on the environment of the silicon atoms in the precursor. In other words, there is no direct memory of the structure of the organosilicon precursor in the structure of the silicon oxycarbide phase [87]. The PSO precursor architecture inﬂuence equally applies to the SiOC plus additive systems (such as Al2O3, TiO2, etc). Polydimethylsiloxane–ZrO2 nanocomposites are prepared by hydrolysis of diethoxydimethylsilane and zirconium n-propoxide in different molar ratios. Transparent, homogeneous, and non-porous xerogels are obtained with up to 70 mol% ZrO2 content. The polymer-toceramic conversion leads to the structural rearrangement of the siloxane component with the production of high speciﬁc surface area materials and pore sizes below 3 nm. The microstructures are amorphous up to 800 8C. At 1000 8C, the structural evolution of the silicon moiety produces an amorphous oxycarbide phase whereas the primary crystallization of tetragonal ZrO2 takes place, with crystallinity and crystallite sizes depending on the ZrO2 content [88]. At 1400 8C, the SiOC phase generates a mixture of amorphous SiO2 and crystalline SiC polymorphs. Such nanocomposite can be made porous by removing SiO2 (Section 6.3) even though this has not been widely practiced.

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

40

5.3. Different hydrosilylation agents Hydrosilylation, i.e. addition of Si–H bond to multiple carbon– carbon or carbon–heteroatom bonds, contributes to the network stability before the breakage of the chains by pyrolysis (Fig. 23) [89]. Hydrosilylation agents can tailor Si:O:C ratios through crosslinking modiﬁcation, either catalytically or thermally. Furthermore, the shrinkage during pyrolysis, the undesired feature usually observed in the sol–gel process, is minimized by using hydrosilylation agents. However, the most important feature of the hydrosilylation reaction is the incorporation of carbosilane or related bridges (Si–Cx–Si) in the polymeric network. Carbosilane bridges allow for better control of the carbon content without decreasing the crosslinking density of the PSO networks, which may provide an opportunity for controlled O/Si ratios. The presence of these bridges has been pointed out as being responsible for the better incorporation of carbon in the oxycarbide glasses, as compared with those derived from alkoxysilanes containing Si–H and/or Si–CH3 groups by the sol– gel process. In contrast to the hydrolysis-condensation reactions of alkoxysilanes, hydrosilylation reactions enable the formation of Si–C–C–Si or Si–C–Si bridges in the polymeric network, which contributes to the generation of a greater amount of SiC4 molecular sites in the ﬁnal product [1] for enhanced high temperature stability. In addition, since pores can be generated by carbon removal during the pyrolysis process, hydrosilylation agents can be used to adjust porous SiOC microstructures. Different hydrosilylation agents have been used to prepare PDMS networks by crosslinking polymer precursors containing terminal vinyl groups. Each reactive silicon group gives rise to a Si– Cx–Si site [1], where x = 1 or 2, with high selectivity toward the ethylene bridges: Si H þ H2 C ¼ CH Si

Ptcatalyst

!

Si CH2 CH2

Si þ Si CHðCH3 Þ Si

(11)

For PHMS, the DVB-derived samples have a ladder-type structure with DVB bridges between the siloxane chains while the 1,3,5,7-tetramethyl-1,3,5,7-tetracyclotetrasiloxane derived network is based on tetrasiloxane rings and the bridges between the rings may contain more than one aromatic ring. Additional differences include the higher reactivity of the cyclic molecule compared to the linear one and a higher degree of consumption of the Si–H bonds [90]. This will directly inﬂuence the amount of Si– Cx–Si groups and thus the resulting SiOC microstructures. Because of the participation of the hydrosilylation agent in the network crosslinking and pyrolysis processes, they can also be used to tailor the SiOC compositions and thus nanopores. For example, up to 60 wt% DVB has been added in the PHMS system for high free carbon content SiOC formation [49,91]. Based on the Si–O–C phase diagram, this can signiﬁcantly shift the ﬁnal composition and thus the SiO2 domain size as well as the graphitic cell wall thickness. In Sections 6.2 and 6.3, we will see that these will directly impact the pore sizes and porosity. For fundamental understanding, linear dihydrogensiloxane (1,1,3,3-tetramethyldisiloxane, HMMH), linear yet branched

Fig. 23. Schematic representation of hydrosilylation of vinyl compounds [89]. (Reprinted from Spectrochim Acta A, 79, A. Nyczyk, C. Paluszkiewicz, A. Pyda, and M. Hasik, Preceramic polysiloxane networks obtained by hydrosilylation of 1,3,5,7tetravinyl-1,3,5,7-tetramethylcyclotetrasiloxane, 801–08, 2011, with permission from Elsevier.)

tetrahydrogensiloxane (tetrakis(dimethylsiloxy)silane, Q(MH)4), and cyclic tetrahydrogensiloxane (2,4,6,8-tetramethylcyclotetrasiloxane, D4H) are used as hydrosilylation agents (Fig. 24) to crosslink two linear polysiloxanes with vinyl groups regularly distributed in their chains (D2V polymer with a vinyl group at every third and V3 polymer with a vinyl group at each silicon atom). The networks formed with D4H, i.e. the least reactive crosslinking agent, tend to exhibit the highest thermal stability; the networks prepared using HMMH, i.e. the most reactive crosslinking agent, show the lowest thermal stability [92], understandably by inﬂuencing the Si–O–C network connectivity. However, the direct effect on the SiOC nanostructure and phase separation has not been studied. Since different precursors directly affect phase separation, they can lead to different porosity samples after selective etching as discussed in Section 6. 6. SiOC porous network creation through atomic and molecular species removal 6.1. Atomic level bond modiﬁcation NH4OH solution aging can be used to modify the bonding characteristics in the SiOC network during the transient carbon evolution stage (during PSO precursor crosslinking yet before SiOC network stabilization) in order to generate nanometer-sized pores. PSO precursors treated in an NH4OH solution demonstrate an increase in the crosslinking degree, whereas the number of terminal Si–O bonds such as Si–OH or Si–O increases signiﬁcantly in the reference system. As a result, the SiOC glass reveals a lower free carbon content in the NH4OH-treated samples (Fig. 25) [63]. Moreover, when the NH4OH solution is added before gel formation, the free carbon content is almost negligible and is below 0.02 mol free carbon per mole silicon. The presence of less carbon in the NH4OH-aged materials (compared to a reference SiOC system) has been found to facilitate the redistribution of the siloxane bonds during pyrolysis in a longer range order (due to less physical barrier) and thus the formation of kinetically more favorable and less stable fourfold siloxane rings. The free carbon nanodomains possess a minimum size of 2.63 nm and increase with the carbon content of the obtained ceramics [63]. Consistent with what to be discussed next, in the NH4OH-treated materials, the average pore size is also increased [63,93]. Mixtures of methyldimethoxysilane and tetraethoxysilane are ceramized through acidic hydrolysis and condensation in different solvents. An oxycarbide structure is established in the mesoporous glasses after pyrolysis of the aged gels. A surface area of 275 m2/g and an average pore size of 3 nm are obtained for a 50% methyldimethoxysilane–50% tetraethoxysilane mixture after pyrolysis at 800 8C in a ﬂowing argon atmosphere. The average pore size is increased by aging the precursor gels in a NH4OH solution. The increased average pore size and the higher strength of the mesoporous gel network enhance the surface area stability (Fig. 26) [93]; in this case, speciﬁc surface areas >200 m2/g are

Fig. 24. Different hydrosilyation agents.

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

41

different distributions in the carbon nanodomains because of the addition of the aging solution. As mentioned before, the materials treated in NH4OH show increased crosslinking, whereas the number of dangling bonds increases signiﬁcantly in the microporous material taken as a reference. 6.2. Atomic level carbon removal

Fig. 25. Calculated La, graphite nanodomain size, from the Raman spectra as a function of the NH4OH concentration. Lines are plotted to guide the eye [63]. r-SiOC represents the as-obtained sol, with no addition of NH4OH solution. SiOC-30 and SiOC-60 refer to the SiOCs obtained when the NH4OH solution was added 30 or 60 min after the reaction was ﬁnished. (Surface and structural modiﬁcation of nanostructured mesoporous silicon oxycarbide glasses obtained from preceramic hybrids aged in NH4OH, A. Tamayo, F. Rubio, J. Rubio, and J. L. Oteo, Journal of the American Ceramic Society, 96 [1] 323–30 (2013), John Wiley and Sons.)

retained at 1200 8C under an argon atmosphere. For pure SiO2 and oxycarbide glasses where all the carbon groups are removed through low-temperature plasma-oxidation treatments, the thermal stability of the surface area decreases dramatically, the resulting speciﬁc surface area is still fairly low at 200 m2/g. Aging in NH4OH reduces the amount of hydrogenated silicon bonds [93]. The glass made from NH4OH-aged gels contains a lower concentration of oxycarbide [CSiO3] and [C2SiO2] species [63]. Highly mesoporous SiOC glasses are obtained from preceramic hybrid materials aged in NH4OH solutions. The surface shows amphoteric character because of the presence of carbon and SiO2 nanodomains. The dependence of the surface Lewis base constant with the concentration of the NH4OH solution suggests

During the formation of SiOC materials using different PSO precursors, carbon can be selectively removed in the SiOC network during the transient state of network formation to achieve high surface areas and single nanometer-sized pores. This aspect directly relates back to the different SiOCs produced through different PSO precursor architectures or using different hydrosilylation agents for crosslinking. In this sense, these different factors should be jointly considered when considering creation of porous SiOCs. The reason is that different SiOC precursor systems produce different C-containing radicals and different amount/ distribution of free carbon species. During the PSO pyrolysis, free carbon starts to precipitate at 500–700 8C. Different carbon distribution is related to the reaction paths during the decomposition of the hybrid network [63]. Based on the evaporative nature of the pyrolyzed species, vacuum atmosphere during pyrolysis can partially remove free carbon to form pores but is not efﬁcient enough to create large surface areas before carbon is locked in the SiOC network [78] [94]. Relying on the transient stage of PSO precursor chain breakage and radical formation, C-containing radicals can be removed by using different pyrolysis atmospheres. There are three most effective pyrolysis atmospheres. The ﬁrst is NH3, the second is water vapor, and the third is H2. Previously, carbon content has been adjusted in-situ during polysilazane pyrolysis. Increasing NH3 contents from 10 to 50 vol% in the pyrolysis atmosphere promotes carbon reduction [95]. The initial trace formation of the free carbon phase occurs through the rearrangement of the carbon atoms in the early stage of pyrolysis. Thermogravimetric analysis curves of the collected products exhibit a steep drop from 510 to 700 8C, indicating a large amount of amorphous carbon [95]. As the temperature increases, the lower accessibility of the carbon-containing centers implies an increase in the carbon rearrangement. Under NH3 atmosphere, optimal annealing temperature for carbon content adjustment is believed to be 500–550 8C according to: C þ NH3 ! HCN þ H2

(12)

For PSO systems, this might present a valuable pyrolysis window to preferentially remove free carbon for pore formation. Compared to NH3, water vapor reduces carbon through the following reaction [35]: Cðfree; sÞ þ H2 OðgÞ ! H2 ðgÞ þ COðgÞ

(13)

Apart from the removal of free carbon, water pyrolysis alters the network conﬁguration of SiOC. 29Si NMR spectral analysis indicates the reduction of carbon-rich SiC4 and Si2C2O2 units with the enrichment of SiO4 and SiCO3 units in the resultant SiOC ceramic [96]. Possible water-carbon reduction reactions include [96]:

Fig. 26. Time dependence of surface area for un-aged and NH4OH-aged glasses with 50% MDMS at 800 8C in argon atmosphere [93]. (Porous silicon oxycarbide glasses, A. K. Singh and G. G. Pantano, Journal of the American Ceramic Society, 79 [10] 2696–704 (1996), John Wiley and Sons.)

Si CH3 þ H2 O ! Si OH þ CH4

(14)

Si CH2 Si þ H2 O ! Si OH þ Si CH3

(15)

The reactions between Si–CH3/Si–CH2–Si and H2O produce Si– OH bonds, and the Si–OH bonds subsequently condense into Si–O– Si by dehydrogenation: Si OH þ HO Si ! Si O Si þ H2 O

(16)

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

42

These reactions deplete Si–C bonds and form more Si–O bonds in the resultant SiOC network. The reduction of Si–C bonds from PSOs suppresses the precipitation of carbon in the subsequent pyrolysis at higher temperatures and thus provides additional leverage for the pores. When the water pyrolysis is conducted at temperatures above 800 8C, the ceramic is undesirably oxidized and the water vapor inﬂuence on pores is minimal [97]. Through our work on PSO SPR-684 Polyramic1 pyrolysis, water injection temperature and duration have a signiﬁcant effect on the carbon loss. Weight loss of 50% through carbon removal can occur. In the 500–700 8C transient stage for SiOC network formation, holding water vapor injection for some time (3–5 h) causes more carbon loss than water injection throughout the pyrolysis process with no holding (Table 1). For the SiOC ceramic without water assisted pyrolysis, the linear shrinkage and ceramic yield for all the samples are 19–22% and 72–76%, respectively. For the SiOC ceramic with water assisted pyrolysis, the linear shrinkage and ceramic yield are 21–25% and 51–63%, respectively. The water assisted pyrolysis induces more linear shrinkage and mass loss. At the same pyrolysis temperature, the linear shrinkage of the SiOC ceramic with water assisted pyrolysis is greater than that of the SiOC ceramic without water assisted pyrolysis, while the ceramic yield shows the opposite trend. In addition, with the increase of the water injection time from 0 h to 3 h, and to 5 h at the same pyrolysis temperature (such as 1300 8C), the linear shrinkage increases from 22.7%, to 23.4%, and to 24.6%, and the ceramic yield rapidly decreases from 60.6%, to 55.5%, and to 51.4%. As explained earlier [45], the major stage of organic to inorganic transformation and weight loss for the crosslinked PSO is between 400 8C and 800 8C. During this stage, for the pyrolysis in the Ar ﬂow, C–H, Si–H, and C5 5C bond breakage in organic groups occurs, leading to SiOC ceramics consisting of amorphous SiOC with Si–O, Si–C, and Si–H bonds. In addition, various redistribution reactions between Si–O, Si–C, and C–C bonds occur. For the studied PSO, the polymer to ceramic transformation mainly occurs in the temperature range of 400–600 8C. The mass loss is due to the release of hydrocarbons (such as vinyl, methyl and propyl, etc.) and hydrogen [74], which results in the shrinkage of the preceramic. This structural change/mass loss also represents an important window to tailor the SiOC compositions and thus microstructures. For the SiOC ceramic pyrolyzed in Ar + H2O, except for the above pyrolysis reactions, water reacts with free carbon and some organic groups (such as Si–CH3, Si–H) in the preceramic, and produces Si–O–Si bonds, hydrogen, and CO2 [97,98]. This can signiﬁcantly change the subsequent ceramization and phase separation of the SiOC

Table 1 Highly porous SiOCs made in our lab with HF etching after pyrolysis. Water vapor injection temperature and time have drastic effect on the speciﬁc surface area and pore size. The samples pyrolyzed in Ar at 1100 8C are not listed in Table 1 because they are fully dense before and after HF etching. ‘‘/’’ symbols in the table denote that the experiments are not carried out at those pyrolysis conditions in this work. The average pore size and pore volume are derived from the Non-linear Density Functional Theory (NLDFT) model on the adsorption and reported in the table [101]. (Highly porous SiOC bulk ceramics with water vapor assisted pyrolysis, J. Li, K. Lu, Journal of the American Ceramic Society, 98 [8] 2357–2365 (2015), John Wiley and Sons.). Temperature (8C), Water injection holding time (h)

1100, 1200, 1300, 1300, 1300, 1400,

0 0 0 3 5 0

Speciﬁc surface area (m2/g) 1%

Average pore size (nm) 0.01

Ar

Ar + H2O

Ar

Ar + H2O

– 44.78 630.41 – – 647.26

449.88 755.38 1602.20 1835.30 2391.60 1656.32

– 1.37 2.11 – – 2.58

2.03 2.05 2.21 2.38 2.87 2.65

ceramics. Moreover, with the increase of the holding time for water injection, these reactions occur more aggressively and induce more extensive structural changes. The carbon loss is mainly through evaporative species formation. For the pyrolysis reactions in the Ar + H2O ﬂow, except for the above reactions, it can be postulated that water molecules attack some organic groups based on Eqs. ((13) and (14)) and the following reactions [98]: Si H þ H2 O ¼ Si OH þ H2 "

(17)

Si CH ¼ CH2 þ H2 O ¼ Si OH þ C2 H4 "

(18)

As a result, there is additional loss of carbon via CH4 and C2H4 evaporation and carbon oxidation into CO. At the same time, the reactions between the organic groups and H2O produce Si–OH bonds, and the Si–OH bonds are subsequently condensed into Si– O–Si through dehydrogenation reaction of Eq. (16). These reactions consume Si–C and Si–H bonds and produce more Si–O bonds in the pyrolyzed ceramics. Therefore, for the SiOC ceramic pyrolyzed in Ar + H2O, the FTIR intensity of the Si–O–Si peak is much higher. Moreover, polymer chain break-up leads to free carbon precipitation at low pyrolysis temperatures of 400–800 8C. Water then reacts with this free carbon phase and change the SiOC network (Eq. (13)) [97]. Due to water reaction with the organic groups and free carbon loss, the SiOC ceramic yield decreases and the linear shrinkage increases compared to the samples pyrolyzed only in the Ar ﬂow. Fig. 27 shows the FTIR spectra of the SiOC ceramics from the PSO after pyrolysis in Ar (Fig. 27(a)) and in Ar + H2O (Fig. 27(b)) at different temperatures before etching. Two main absorption peaks are observed around 800 cm1 and 1100 cm1 wavenumbers, which correspond to Si–C and Si–O–Si bonds [44]. There also seems to be a very weak C–C peak at 1500 cm1. With the pyrolysis temperature increase, for the SiOC ceramics pyrolyzed in the Ar ﬂow, the intensities of the peaks have no obvious change. For the SiOC ceramics pyrolyzed in Ar + H2O ﬂow, the intensity of the Si–O–Si peak slightly increases with the temperature. More importantly, at the same pyrolysis temperature, the intensity of the Si–O–Si peak is much higher for the samples pyrolyzed in the Ar + H2O ﬂow. This means that water vapor greatly favors Si–O–Si bond formation. It should be pointed out that pyrolysis of SiOC precursors in NH3 and water vapor may not directly produce porous SiOC structures. However, they inﬂuence the phase separation at higher pyrolysis temperatures. As a result, more separated phases, such as SiO2, can be removed more effectively for pore generation. Again, highly porous SiOC material generation requires consideration of several composition and processing variables. The interrelationships need to be fully understood. The third effective atmosphere to remove carbon is H2. Narisawa’s group prepared different Si–O–C(–H) ceramics with various carbon-to-hydrogen ratios by pyrolyzing a polysiloxane precursor in H2 at 800–1100 8C [99,100]. More recently, SiOC samples pyrolyzed in Ar and H2 are compared [101]. The latter sample contains a signiﬁcantly higher concentration of hydrogen, which primarily bonds to carbon. However, such low pyrolysis temperature does not lead to phase separation and thus the potential for porous SiOC material formation is low. Further work to increase the pyrolysis temperature might lead to interesting results in this regard. 6.3. Molecular level SiO2 nanocluster removal At 1000–1250 8C, the SiOC glass network contains the maximum amount of mixed SiOC units, SiCxO4x, 1 x 3. With the increase of the pyrolysis temperature, the SiOC network undergoes phase separation with a consumption of the SiOC mixed units and

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

43

Fig. 28. Shift on the SiOC composition after etching in HF solution (hypothetically from A ! B) and after further chlorine etching (hypothetically from B ! C).

Fig. 27. FTIR spectra of the un-etched SiOC ceramics from PSO after pyrolysis in (a) Ar (b) Ar + H2O at different temperatures [101]. (Highly porous SiOC bulk ceramics with water vapor assisted pyrolysis, J. Li, K. Lu, Journal of the American Ceramic Society, 98 [8] 2357–2365 (2015), John Wiley and Sons.)

an increase of SiC4 and SiO4 units. The progressive increase of SiC4 and SiO4 units results in growth of SiC nanocrystals and SiO2 nanoclusters. To obtain narrow pore size distribution, controlling the SiO2 nanodomain and SiC crystallite sizes is another key aspect. The sizes of both SiO2 and SiC phases increase with the pyrolysis temperature according to Eq. (7). Strong acid attack, such as HF, dissolves the SiO2-based nanoclusters, forms a porous structure, and allows new nanometer-sized pore formation [102]. HF etching can thus be used to remove the SiO2 nanoclusters and obtain pores of SiO2 nanocluster sizes while shifting the SiOC composition on the phase diagram from A ! B (Fig. 28): SiO2 þ 4HF ! SiF4 ðgÞ þ 2H2 OðlÞ

(19)

With excessive aqueous HF: 6HF þ SiO2 ! H2 SiF6 þ 2H2 O

(20)

It should be pointed out that free carbon is insoluble in HF solutions and will act as a chemical barrier, hindering the attack or diffusion of H+ and F ions at and within the SiOC surface. In Section 6.1, NH3 and water vapor treatments can effectively remove carbon and thus facilitate the SiO2 removal through the HF etching. Since HF-etching generates small pores in the order of single to a couple of tens of nanometers’ range, the SiO2-rich

regions must be of a similar dimension. However, it cannot be excluded that a small fraction of mixed SiOxCy units can also be etched away by HF acid. SiOC microcellular foams are prepared using a preceramic polymer (MKWacker-Chemie GmbH, Germany) and PMMA microbeads (Altuglas BS, Altuglas International, Arkema Group, Rho (MI), Italy) of nominal size of 50 mm acting as a sacriﬁcial ﬁller. Since PMMA ﬁller is used, microcellular ceramic foams with a bimodal pore size distribution are produced. After etching by a 20 vol% HF solution, an increase in one order of magnitude in speciﬁc surface areas is observed compared to the unetched samples. This speciﬁc surface area increase is accompanied by micro- and meso-pore formation from the removal of SiO2 nandomains. Higher speciﬁc surface area values (up to 65 m2/g) are reached by inducing slight phase separation, accompanied by growth of the nanodomain sizes, and by an oxidative treatment that partly removes the residual carbon [15]. Nanoporous SiOC with high surface area is prepared by pyrolysis of polymethyl(phenyl)siloxane resin (SR249) under vacuum at 1250–1350 8C followed by leaching in a HF solution. The pyrolysis temperature and etching condition have important effects on the compositions and structures of the SiOC materials. Their speciﬁc surface areas are very low at <55 m2/g before etching but increase signiﬁcantly after leaching. The speciﬁc surface area and total pore volume are as high as 1148 m2/g and 0.608 cm3/g, respectively, when the pyrolysis is carried out at 1300 8C. The pore sizes of all the samples after leaching are in the narrow range of 1–4 nm. TEM shows that the free carbon phase, SiC nanocrystallines, and SiOC ceramics wrap each other [94]. The surface properties of different SiOC glasses before and after HF etching are analyzed by means of nitrogen adsorption and inverse gas chromatography at inﬁnite dilution. Raman and Drift spectroscopies have also been used for structural characterization. The dispersive surface energy (gsd) and acid base (kA/kB) character are then compared with those found for vitreous SiO2 glass. The gsd value of the SiOC glass is higher than that of vitreous SiO2 (gsd = 58.8 mJ/m2) and decreases from 151.02 to 104.39 mJ/m2 with the increase of the carbon content. The highly ordered

44

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

graphite is attributed to the higher dispersive surface energy. After HF etching the gsd values of the SiOC glasses increase up to 200 mJ/ m2, in accordance with the elimination of the SiO2 phase and therefore the presence of a high concentration of carbon mainly as ordered graphite. Raman and Drift spectroscopies suggest that the HF etching not only removes SiO2 but poorly ordered graphite and probably b-SiC nanocrystals. The acid base properties, i.e., the kA/kB ratio, of the analyzed SiOC glasses are very close to those of vitreous silica glass or natural graphites, depending on their free carbon content, indicating that the free carbon phase contains surface active acidic groups [103]. In our work, SiOC etching has been performed by stirring an aqueous HF (20 wt%) solution at room temperature until there is no distinct mass loss. Pore sizes from 2 to 6 nm and speciﬁc surface areas from 540 to 2390 m2/g have been achieved by us. The SiOC compositions on the C–SiO2–SiC phase diagram (Fig. 28) also shift to the O– and Si-deﬁcient direction (A ! B). Compared with the SiOC ceramics pyrolyzed only in Ar [74], the similarity is that a completely amorphous structure is present at 1100 8C pyrolysis temperature. At 1200 8C and higher pyrolysis temperatures, the SiOC starts to undergo phase separation and forms SiC, carbon, and SiO2 nanodomains. In the range of 1100– 1400 8C pyrolysis temperatures, SiO2 crystal are not observed for the samples pyrolyzed in the Ar atmosphere. This difference can be understood as follows. For the SiOC ceramic pyrolyzed only in Ar, at low temperatures (1000–1200˚ C), silicon atoms share bonds with oxygen and carbon atoms in mixed silicon oxycarbide SiO4xCx (1 x 3) units [54,65], the distribution of the various silicon sites is completely random, so it shows an amorphous structure. With the increase of the pyrolysis temperature (T 1200 8C), the redistribution between Si–C and Si–O bonds results in phase separation between [SiC4] and [SiO4] units with the consumption of the mixed SiO4xCx (1 x 3) tetrahedrals, forming SiO2 nanodomains (still amorphous, not SiO2 crystals). This process also leads to the crystallization of b-SiC and free carbon. For the SiOC ceramic pyrolyzed in Ar + H2O, except for the above reactions, water reacts with the SiOC matrix and free carbon (Eqs. (13), (14) and (16)–(18)) in the temperature range of 500–700 8C, and forms more Si–O–Si bonds in the SiOC. With the increase of the pyrolysis temperature, Si–O–Si related species crystallize into SiO2 nanocrystals (such as at T = 1300 8C) [104] due to more Si– O–Si bond formation and reduced comprehensive stress [45], so the XRD patterns show the SiO2 peak (Fig. 29(a)). Fig. 29(b) shows the XRD patterns of the SiOC ceramic pyrolyzed in Ar + H2O at different temperatures after the HF etching. Compared with the XRD patterns before the HF etching (Fig. 29(a)), the SiO2 peaks disappear. The intensities of the XRD peaks for b-SiC and carbon increase after the etching for the same sample, and obviously increase with the increase of the pyrolysis temperature. Moreover, the samples pyrolyzed at 1300 8C and 1400 8C show a very weak silicon peak (Fig. 29(b)), which cannot be found before the HF etching due to the presence of other strong peaks (such as SiO2). Our understanding for this observation is as follows. During the Ar + H2O atmosphere pyrolysis, SiO2 nanodomains form due to phase separation, and the Si–O–Si bonds crystallize into SiO2 nanocrystals as well as a small amount of silicon (Eqs. (21) and (22)). Moreover, at high pyrolysis temperatures (1000–1500 8C), SiO2 and SiC in the preceramic react and produce silicon and CO gas (Eq. (9)) [7]. SiO2 nanodomains, Si–O–Si bonds, and SiO2 nanocrystals in the SiOC ceramic are etched away by the HF acid solution (Eq. (19)), so the silicon peaks become more visible, and the intensities of SiC and carbon peaks increase [104]. Without water presence during the pyrolysis, silicon cannot be produced because more excessive carbon is present to form SiC. As explained before, higher pyrolysis temperature promotes more

Fig. 29. XRD patterns of the SiOC ceramics pyrolyzed in Ar + H2O at different temperatures (a) before and (b) after etching (Note: water injection without holding time) [101]. (Highly porous SiOC bulk ceramics with water vapor assisted pyrolysis, J. Li, K. Lu, Journal of the American Ceramic Society, 98 [8] 2357–2365 (2015), John Wiley and Sons.)

extensive phase separation and produces more SiO2 nanodomains, SiC, and silicon. 2Si O Si ¼ SiO2 þ 3Si

(21)

SiO2 þ 2SiC ¼ Si þ 2CO "

(22)

The BET adsorption–desorption isotherms of the SiOC ceramics pyrolyzed in the Ar + H2O ﬂow at different temperatures after the HF etching are shown in Fig. 30. The insets in Fig. 30(a)–(d) show that the pore sizes are in the range of 2–10 nm, and the size distributions of the pores gradually increase from 2–4 nm to 2–10 nm as well as from monomodal to bimodal when the pyrolysis temperature increases from 1100 8C to 1400 8C. In addition, N2 maximum adsorbed volume gradually increases from 103.78 to 414.53 cm3/g with the increase of the pyrolysis temperature from 1100 8C to 1400 8C. N2 maximum adsorbed volume and pore volume are lower than those of samples pyrolyzed in the Ar + H2O ﬂow at the same pyrolysis temperature. For the SiOC ceramics pyrolyzed in the Ar ﬂow, with the increase of the pyrolysis temperature from 1200 8C to 1400 8C, the speciﬁc surface area of the SiOC ceramic increases from 44.48 m2/g to 647.26 m2/g; the average pore size and pore volume increase from 1.4 nm to 2.6 nm and from 0.016 cm3/g to 0.319 cm3/g, respectively. For the SiOC ceramics pyrolyzed in the Ar + H2O ﬂow without holding time at 500–700 8C, with the increase of the pyrolysis temperature from 1100 8C to 1400 8C, the speciﬁc surface

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

45

Fig. 30. N2 adsorption (solid symbols) and desorption (open symbols) isotherms for the samples obtained from the SiOC ceramics pyrolyzed in Ar + H2O at (a) 1100 8C, (b) 1200 8C, (c) 1300 8C, and (d) 1400 8C after etching, respectively. The pore size distribution (V–dV) is derived from NLDFT model on the adsorption and reported in the inset [101]. (Highly porous SiOC bulk ceramics with water vapor assisted pyrolysis, J. Li, K. Lu, Journal of the American Ceramic Society, 98 [8] 2357–2365 (2015), John Wiley and Sons.)

area of the SiOC ceramic increases from 449.88 m2/g to 1656.32 m2/g; the average pore size and pore volume increase from 2.0 nm to 2.7 nm and from 0.154 cm3/g to 0.588 cm3/g, respectively. Moreover, for the SiOC ceramics pyrolyzed in the Ar + H2O ﬂow at 1300 8C with holding time at 500–700 8C from 0 h to 5 h, the speciﬁc surface area of the SiOC ceramic increases from 1602.20 m2/g to 2391.60 m2/g; the average pore size and pore volume increase from 2.1 nm to 2.9 nm and from 1.574 cm3/g to 1.024 cm3/g, respectively. The above experimental results may be explained as follows. For the SiOC pyrolyzed only in the Ar ﬂow, during the pyrolysis of the preceramic at the temperature of 1100–1400 8C, main reactions take place as Eqs. ((7)–(10)) [105,106]. With the increase of the pyrolysis temperature, the rate of phase separation is promoted and more SiO2 nanodomains form. These nanodomains are etched away by the HF acid solution and leave behind more micro-/meso-pores. Therefore, the speciﬁc surface area of the samples shows an increasing trend with the increase of the pyrolysis temperature. Based on the XRD analysis, the phase separation mainly takes place during 1200 8C and 1300 8C. When the pyrolysis temperature is above 1300 8C, the phase separation is almost complete. Carbothermal reaction (Eq. (10)) is the main reaction and consumes the SiO2 nanodomains in the preceramic [106]. Accordingly, after the HF etching the speciﬁc surface area of the samples shows a remarkable increase from 1200 8C to 1300 8C and then only a slight increase from 1300 8C to 1400 8C. For the SiOC ceramics pyrolyzed in the Ar + H2O ﬂow, except for the above phase separation and carbothermal reactions, water reacts with organic groups and free carbon during water injection at 500–700 8C, and produces Si–O–Si bonds, hydrocarbons, hydrogen, and CO (Eqs. ((13)–(18)). Si–O–Si bonds may be etched by HF acid and leaved behind mico-/meso-pores, so the SiOC ceramic is

porous at low pyrolysis temperature (such as T = 1100 8C) after the HF etching. With the increase of the pyrolysis temperature, Si–O–Si bonds crystallize into SiO2 nanocrystals. Therefore, the samples show higher speciﬁc surface area than those of the SiOC ceramics pyrolyzed only in Ar. Moreover, with the water injection time increase, more Si–O–Si bonds are created and more free carbon is removed, so the SiOC ceramics have higher speciﬁc surface area after the HF etching. On a slightly different note, SiBOC is a parallel system to SiOC, for which the microstructure can be modiﬁed by HF etching. The presence of B in the SiOC network has been found to (i) increase the network stability (increase SiOC decomposition temperature), (ii) reduce the tendency of SiO2 crystallization, and (iii) increase the crystallization rate of SiC [65]. SiBOC precursor gels can be prepared as for the SiOC systems with additional boric acid addition [58]. SiBOC networks undergo faster high temperature rearrangement compared to B-free SiOC samples and are more easily etched and develop higher porosity and larger pore size. Boron also has an important effect on the ordering of the sp2 carbon phase and leads to thicker sp2 graphitic layers [59] [65]. The SiO4 and BO3 units, which are present in the borosilicate nanodomains, will be leached out by HF, thus forming SiBOC glass. B2O3 present in the SiO2 glass is dissolved by a HF solution with a similar reaction [107]: B2 O3 þ 6HF ! 2BF3 þ 3H2 O

(23)

Silicon and boron present in mixed oxycarbide units with at least one Si–C or B–C bond will still be stable. Indeed, Si–C and B–C bonds cannot be cleaved by HF attack and therefore act as a permanent bridge to the C-rich phase. The rate-limiting step for

46

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

the dissolution is the diffusion of the reactants/products through the porous Si(B)OC. Compared to the B-free SiOC glass, because SiBOC is more easily etched [58], it gives ﬂexibility to tailor pore sizes and speciﬁc surface areas of the SiBOC system even though the etching needs to be carefully controlled in order to obtain high speciﬁc surface areas. The possible limit of boron content incorporation needs to be further evaluated. The role of boron in promoting nanoporous structure formation can be obtained by comparing the evolution of the B-free (SiOC) and B-containing (SiBOC) samples. Since boron promotes the formation of borosilicate clusters, it can further inﬂuence the graphene formation (sp2 carbon). One possible scenario is thicker graphitic walls. It may also form BC3 sites [58] and thus inﬂuence the thermal stability of the ultrahigh surface area SiBOC microstructures. The Sto¨ber process for the preparation of ﬁne SiO2 spherical particles is a well-known method. As an interesting new extension, highly porous SiOC (Si–O–C) spheres, with diameters in the range of 100–400 nm, are synthesized through pyrolysis of sol–gelderived hybrid precursors of silsesquioxane from different organotriethoxysilanes (RTES, R = CH3, C5H11, and C6H5). The hybrid particles are transformed into dense inorganic Si–O–C spheres through pyrolysis at 1000–1300 8C in a controlled atmosphere. The spherical morphology is kept provided that the glass transition temperature of the silsesquioxane network is higher than the onset of the polymer-to-ceramic transformation. The Si–O–C spheres are stable up to 1400 8C. By HF etching of the SiO2 nanodomains present in the SiOC structure, the dense Si–O–C particles can be further engineered and transformed into highly porous Si–O–C spheres with speciﬁc surface area up to 564 m2/g and pore volume up to 0.7 cm3/g [102]. This approach can certainly be extended to include different ﬁllers such as boron, aluminum, titanium, and zirconium by adding the corresponding metal alkoxides to the starting liquid. During the pyrolysis, these metalcontaining species can be converted to metal oxides. 6.4. Molecular level SiC (BCx) removal As shown in Figs. 3 and 16, the carbon phase is dispersed in the SiOC matrix (along with SiO2 nanoclusters) at low pyrolysis temperatures (<1300 8C). At high temperatures, the carbothermal reduction of SiO2 takes place, consuming the carbon phase [1]: SiO2 ðsÞ þ CðsÞ ! SiOðgÞ þ COðgÞ

(24)

This reaction results in the evaporation of CO and SiO gaseous species. However, the above equation is only valid for low levels of free carbon in the matrix. In the presence of high carbon contents, SiC(s) and CO(g) are the major decomposition products, according to: SiOðgÞ þ 2CðsÞ ! SiCðsÞ þ COðgÞ

(25)

The above equations show that the SiO species plays an important role as an intermediate in the reduction of SiO2. For the samples with high carbon amounts, the carbothermal reduction of SiO2 will preferentially produce SiC, resulting in widespread presence of SiC precipitates (Fig. 31) [53]. In addition to using different atmospheres to remove carbon and using HF etching to remove SiO2 (B2O3), the silicon in SiC nanoclusters can be controllably removed with Cl2 gas to obtain SiOC plus graphitic carbon porous structures (Fig. 32). Carbon ﬁlms are produced from SiC by ﬂowing a gas mixture of Cl2 and H2 at 1000 8C for 20 h [108]. Highly porous carbon samples (up to 2430 m2/g) are created by heating SiC at 7.5 8C/min up to 700 8C, 800 8C, and 900 8C, respectively, under a dynamic chlorine atmosphere for 2 h, followed by a treatment in ammonia at

Fig. 31. TEM images of SiOC annealed at 1300 8C in argon. Numerous nanosized SiC precipitates are observed.

600 8C for 3 h to remove residual chlorine from the highly porous material [109]. Wet chemical conversion of phenyltrimethoxysilane to SiOC is conducted using a two-step acid/base sol–gel process in an aqueous medium. The resulting material is subsequently pyrolyzed at 700 8C, 1000 8C, and 1300 8C to obtain a non-porous SiOC ceramic. Chlorination at 700 8C and 1000 8C in a stream of 80 ml/min Cl2 and 70 ml/min argon [110] leads to carbons having speciﬁc surface areas exceeding 2000 m2/g and large micro-/meso-pore volumes up to 1.4 cm3/g. The temperature of the thermal treatment signiﬁcantly inﬂuences the carbon and ﬁnal pore structure. Pyrolysis at 700 8C and subsequent chlorination at 700 8C lead to a mainly microporous material, whereas pyrolysis at 1300 8C and subsequent chlorination at 1000 8C generate a hierarchically porous SiOC with micro- and meso-pores, respectively. Chlorine etching also allows synthesis of porous SiCN materials with 2000 m2/g speciﬁc surface area and 3–10 nm pore sizes [111]. Etching pyrolyzed polymethylsilsesquioxane and polymethylphenylsilsesquioxane with Cl2 gas at 1200 8C produces speciﬁc surface areas up to 1100 m2/g with >1 nm pores [112]. This implies that after high pyrolysis temperatures (1300–1450 8C) for SiC nanocluster formation in-between the SiOC layer and the graphitic cell wall, low temperature chlorine etching can be conducted to produce small pores [6]: SiC þ 2Cl2 ðgÞ ! SiCl4 ðgÞ þ C

(26)

The mechanisms for pore creation by chlorination of SiOC may not be as straightforward as for the chlorination of pure SiC [113] due to the presence of different species, namely the SiOC phase itself, free carbon, and SiO2 domains. Hierarchical pore formation during chlorination is due to SiC nanocrystals and the SiOC domains. Micropores are thus generated by SiC, whereas pores in the range of 1–5 nm are achieved by the removal of the silicon and oxygen atoms from the SiOC. Since the SiOC is an amorphous phase, the pore size distributions are rather broad. For SiOBC, there is a possibility of removing BCx below 600 8C due to the formation of CCl4(g), based on the chlorination of B4C

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

47

Fig. 32. SiC formation during the SiOC pyrolysis process and its removal using Cl2 afterwards.

[114]: B4 CðsÞ þ 8Cl2 ðgÞ ! 4BCl3 ðgÞ þ CCl4 ðgÞ

(27)

The structure remaining after SiO2 and SiC removal is postulated to consist of a scaffold of graphitic networks (a network of sp2 carbon) with their surfaces covered with SiCxO4x (1 < x < 4). The pores existing within such scaffolds should have been ﬁlled with SiO2 tetrahedra and SiC. SiOC tetrahedra form the transition between the pore interface and the graphitic structures, and can potentially offer high thermal stability for the high surface area materials. 7. SiOC porous network creation through molecular templates Highly porous SiOC materials can also be made by mixing fugitive polymer precursors with different SiOC precursors before crosslinking. The control of the pore sizes in the SiOC network depends on the characteristics of the precursor mixtures, the relative miscibility, the amount of the precursors used, and the degree of crosslinking before pyrolysis [13]. However, to produce stable SiOC nanoporous structures, the molecular templates must also be Si-containing polymer precursors and have the right size. Bis(triethoxysilyl)-methane and cetyltrimethylammonium chloride have been used to synthesize porous SiOC by varying a large number of synthesis parameters (pH, concentration of surfactant, dilution, crosslinking conditions) to optimize the extent of ordering of the porous phase. It is very difﬁcult to convert such samples into a SiOC glass with high porosity. By 1000 8C, the structure collapses with a complete cleavage of the Si–C bonds [115]. Additionally, the size of the polymer precursors should not be too large. PDMS liquid is mixed with two solid PSOs: poly(methyl–siloxane) and poly(methyl–phenyl–siloxane). Only very large pores (5–600 mm) are produced [116]. By changing the viscosity of the PDMS or its mass ratio in the mixed polymers, the pore sizes and porosities of the porous SiOC materials can be adjusted in the range from 10 nm to 3 mm, and 20% to 90%, respectively. However, the pore size distribution is still very large and the speciﬁc surface area is still fairly low (128 m2/g) [117] [66]. Using large size nonionic (PEO)106–(PPO)70–(PEO)106 (Pluronic(R) F127) (Fig. 33) as a templating agent, periodic mesoporous SiOC with cubic Fm3m phase is obtained from acidic solutions using 1,2-bis(triethoxysilyl) ethane as organosilane. After pyrolysis under argon ﬂow for 2 h the Fm3m phase is retained and the sample still

displays speciﬁc surface area of 260 m2/g [118,119]. For this organicinorganic hybrid framework, use of KCl during the initial mixing stage is important to obtain highly ordered material, without which the structure ordering of the products would be relatively poor. When TEOS is used as the template for pore formation, hydrolysis and condensation of TEOS can act as the leading mechanism for SiO2 (and thus pore) formation. The TEOS inorganic/PSO organic dominance in the nanometer scale is inﬂuenced by the thermal treatment required to form the SiOC phase [120]. Differently, PDMS belongs to the same PSO family and should mix with SiOC precursors at the molecular level. Molecular level phase separation should lead to preferential nanometer-sized pore formation as PDMS will be mostly decomposed during the pyrolysis. With proper selection of PDMS molecular weight (radius of gyration Rg can be used for pore size prediction) and volume fraction (to still enable stable, continuous network) to generate single nanometer pores, pores can be created by removing PDMS before the SiOC network locks in. The key is to mix the PSOs and the PDMS at the molecular level and be able to tailor the molecular scale phase separation. In our research of PSO (RD-684 Polyramic1, Starﬁre Systems Inc.) plus PDMS mixture pyrolysis, PSO:PDMS of 50:50 (wt%) leads to >500 m2/g speciﬁc surface area after 1300 8C pyrolysis and HF etching. As explained in Section 5.2, polysilsesquioxane has a cage-like structure and high ceramic yield (>65 wt%) upon pyrolysis, and can act as precursors for pore formation. A conventional PSO has a linear structure and a low ceramic yield (<30 wt%). When mixing these two precursors to form a crosslinked body, porosity is formed during pyrolysis because of the different decomposition behaviors [13]. Polysilsesquioxane can produce high enough Si–O bonds and high level of Si-H to reduce the overall amount of excess carbon. Highly porous structures can thus be produced depending on the compositions of the starting mixtures. PDMS has a linear structure with methyl side groups, whereas methylsilicone resin is a non-linear polymer containing a large amount of oligomeric methyl silsesquioxane (CH3SiO3/2). With electron bean irradiation, crosslinking in irradiated PDMS happens though C–H and C–Si bond scission in the methyl (CH3) side chain, by eliminating two hydrogen atoms only, or one methyl and one hydrogen atom, or two methyl groups. Due to these crosslinking reactions in PDMS, volatile products such as hydrogen, methane, and ethane are formed during irradiation, according to the reactions [27]: Si CH2 þ CH2 Si ! Si CH2 CH2 Si þ H2 "

Fig. 33. Structure of (PEO)106–PPO)70–(PEO)106 (Pluronic(R) F127).

(29)

Si CH2 þ Si ! Si CH2 Si þ CH4 "

(30)

Si + Si ! Si Si + C2H6 "

(31)

The evolution of these volatile species during the electron beam irradiation can also contribute to bubble formation and expansion of the molten methylsilicone, which need to be carefully controlled. The ﬂexibility of starting from different polymer precursors offers great potentials to create nanosized pores and nanodomains of

48

K. Lu / Materials Science and Engineering R 97 (2015) 23–49

different attributes. Hybrid materials have been obtained using a given concentration of tetraethylorthosilicate and PDMS. Different size and shape monolithic samples have been achieved. The hybrid materials are pyrolyzed under nitrogen atmosphere to yield porous SiOC glasses, consisting of interconnected spherical SiO2 particles and continuous open-pore channels, with gradient particle size distributions from the top (2.88 mm) to the bottom (7.83 mm) of the samples. The gradient in the particle size also leads to a gradient distribution of porosities in the SiOC glasses. Thus, pore sizes increase from 6 mm at the top evaporation face to 13 mm at the bottom of the samples, and porosities in the corresponding regions decrease from 77% to 63%, respectively. An interesting ﬁnding is that the gradient PDMS distribution also provokes gradient nanodomain sizes of the segregated free carbon in the form of graphite [121]. PHMS and vinyl-terminated PDMS of different molecular weights in the presence of tetravinyltetramethylcyclotetrasiloxane as a crosslinking enhancer have been crosslinked via a Pt catalyzed hydrosilylation reaction. The PDMS serves the double purpose of size-controlling templating agent as well as solvent in the early stage of the synthesis. During the crosslinking step, the vinylterminated PDMS is chemically bonded to the preceramic network through hydrosilylation reaction and ‘‘solidify’’. Accordingly, its removal during pyrolysis occurs through the decomposition of solid phase with the retention of formed pores. The structural and morphological evolution of the preceramic gels containing the molecular spacers shows that the pore size distribution of the resulting SiOCs depends on the molecular weight and thus the molecular volume of PDMS, assuming that the PDMS chains are entangled into spheroidal shapes. The total pore volume is related to the initial amount of templating PDMS assuming its complete decomposition during the pyrolysis [66]. As the molecular size of the fugitive PDMS is nanoscale in size, the larger than expected pore structure must be attributed to phase separation at the sol stage of the precursors. The interaction of the molecular pore former and the preceramic backbone network is critical for the resultant pore size and speciﬁc surface area. Starting from a similar precursor, PHMS crosslinked using Karstedt’s Pt catalyst and methyl-terminated PDMS leads to the formation of micro/nano-porous SiOC foam from PSO [117]. However, the PDMS templating agent, being methyl-terminated, is not covalently bonded to the cured PHMS network and the starting precursor is composed of a ‘‘hard’’ crosslinked PHMS skeleton and a ‘‘soft’’ uncrosslinked PDMS network. Only porous SiOCs with much larger pore sizes are obtained, from 10 nm to almost 3 mm. In a different study, pore sizes around 4 nm for PDMS of 9400 molecular weight and around 30–31 nm for PDMS of 155,000 molecular weight are obtained [66]. This means that the pore dimensions in the ﬁnal porous SiOC are derived from the molecular dimensions of the polydimethylsiloxane chains. This is different from the pore dimensions resulting from the starting homogeneous PHMS– PDMS solution during the crosslinking [117]. Sodium dodecyl sulfate, N-trimethoxylsilylpropyl-N,N,N-trimethylammonium chloride (TSA) (50% in methanol), and tetraethoxylsilane are mixed with the composition of TEOS– TSA–SDS–H2O = 1:0.2x:0.1:155, where x is either 0.1, 0.2, 0.3 or 0.4. Well-ordered meso-porous SiOC ceramic is synthesized by calcining the organic–inorganic systems at 600 8C under nitrogen. The obtained SiOC products exhibit high speciﬁc surface area, up to 1762 m2/g, with uniform pore size distribution. The speciﬁc surface area is only reduced to 1403 m2/g when exposed to 800 8C in air [122]. 8. Summary The small pore size, the large speciﬁc surface area, and the high temperature stability of the porous SiOC systems have great

application potentials in catalysis, reﬁnery, gas separation, sensing, electrodes, gas storage, molecular sieves, absorbents, thermal insulation, and micro-reactors. In addition, SiOC glasses possess important applications at room temperatures as substrates for microelectronic devices, blood-contact biocompatible materials, mesoporous ordered substrates, etc. In this review, fugitive ﬁllers and selective removal of different SiOC compositions are discussed as two strategies of creating porous SiOCs. The former includes ceramic replication of an organic template, direct foaming, and sacriﬁcial pore formers. The latter includes molecular level pore formation, molecular level species removal, and SiOC porous network creation through molecular templates. To understand the single nanometer pore formation and the phase separation for selective species removal, the SiOC nanostructure models and composition descriptions after the pyrolysis are explained. Direct pore formation can be achieved by changing processing conditions, using different precursor architectures, and using different hydrosilylation agents. For SiOC porous network creation through molecular level species removal, the approaches include molecular level carbon removal, molecular level SiO2 nanocluster removal, and molecular level silicon (boron) removal from SiC (BCx). The technique developed can be applied to a wide range of polymer precursors including polysilanes, polycarbosilanes, polysilazanes, polyborosilazanes, polycarbosiloxane, and polysilsesquioxanes. References [1] B.F. Sousa, I.V.P. Yoshida, J.L. Ferrari, M.A. Schiavon, J. Mater. Sci. 48 (2013) 1911–1919. [2] L. Qiu, Y.M. Li, X.H. Zheng, J. Zhu, D.W. Tang, J.Q. Wu, C.H. Xu, Int. J. Thermophys. 35 (2014) 76–89. [3] P. Colombo, Adv. Eng. Mater. 1 (1999) 203–205. [4] P. Colombo, J.R. Hellmann, D.L. Shelleman, J. Am. Ceram. Soc. 84 (2001) 2245–2251. [5] P. Colombo, J.R. Hellmann, D.L. Shelleman, J. Am. Ceram. Soc. 85 (2002) 2306–2312. [6] S.J. Widgeon, S. Sen, G. Mera, E. Ionescu, R. Riedel, A. Navrotsky, Chem. Mater. 22 (2010) 6221–6228. [7] J. Latournerie, P. Dempsey, D. Hourlier-Bahloul, J.P. Bonnet, J. Am. Ceram. Soc. 89 (2006) 1485–1491. [8] J.M. Pan, J.F. Pan, X.N. Cheng, X.H. Yan, Q.B. Lu, C.H. Zhang, J. Eur. Ceram. Soc. 34 (2014) 249–256. [9] B.V.M. Kumar, Y.W. Kim, Sci. Technol. Adv. Mater. 11 (2010) (044303-1–044303-16). [10] A.R. Studart, U.T. Gonzenbach, E. Tervoort, L.J. Gauckler, J. Am. Ceram. Soc. 89 (2006) 1771–1789. [11] E.J. Lee, J.J. Kim, S.O. Cho, Langmuir 26 (2010) 3024–3030. [12] L. Biasetto, P. Colombo, M.D.M. Innocentini, S. Mullens, Ind. Eng. Chem. Res. 46 (2007) 3366–3372. [13] P. Colombo, J. Eur. Ceram. Soc. 28 (2008) 1389–1395. [14] L. Biasetto, A. Francis, P. Palade, G. Principi, P. Colombo, J. Mater. Sci. 43 (2008) 4119–4126. [15] L. Biasetto, R. Pena-Alonso, G.D. Soraru, P. Colombo, Adv. Appl. Ceram. 107 (2008) 106–110. [16] P. Colombo, E. Bernardo, Compos. Sci. Technol. 63 (2003) 2353–2359. [17] P. Colombo, M. Modesti, J. Am. Ceram. Soc. 82 (1999) 573–578. [18] S. Costacurta, L. Biasetto, E. Pippel, J. Woltersdorf, P. Colombo, J. Am. Ceram. Soc. 90 (2007) 2172–2177. [19] Y.W. Kim, K.H. Lee, S.H. Lee, C.B. Park, J. Ceram. Soc. Jpn. 111 (2003) 863–864. [20] H. Tian, Q.S. Ma, Mater. Lett. 66 (2012) 13–15. [21] H. Tian, Q.S. Ma, Y. Pan, W.D. Liu, Ceram. Int. 38 (2012) 5039–5043. [22] H. Tian, Q.S. Ma, Y. Pan, W.D. Liu, Ceram. Int. 39 (2013) 71–74. [23] H. Tian, Q.S. Ma, Ceram. Int. 38 (2012) 2101–2104. [24] B. Ceron-Nicolat, F. Wolff, A. Dakkouri-Baldauf, T. Fey, H. Munstedt, P. Greil, Adv. Eng. Mater. 14 (2012) 1097–1103. [25] S. Nedunchezhian, R. Sujith, R. Kumar, J. Adv. Ceram. 2 (2013) 318–324. [26] R.M. da Rocha, E.A.B. Moura, J.C. Bressiani, A.H.A. Bressiani, Radiat. Phys. Chem. 79 (2010) 301–305. [27] R.M. Rocha, E.A.B. Moura, A.H.A. Bressiani, J.C. Bressiani, J. Mater. Sci. 43 (2008) 4466–4474. [28] P. Colombo, M. Griffoni, M. Modesti, J. Sol-Gel Sci. Technol. 13 (1998) 195–199. [29] Y.W. Kim, J.H. Eom, C.B. Park, W.T. Zhai, Y.T. Guo, M. Balasubramanian, J. Am. Ceram. Soc. 93 (2010) 3099–3101. [30] Y.W. Kim, Y.J. Jin, Y.S. Chun, I.H. Song, H.D. Kim, Scr. Mater. 53 (2005) 921–925. [31] M. Adam, S. Kocanis, T. Fey, M. Wilhelm, G. Grathwohl, J. Eur. Ceram. Soc. 34 (2014) 1715–1725. [32] S.H. Kim, Y.W. Kim, C.B. Park, J. Mater. Sci. 39 (2004) 3513–3515. [33] M.M. Hassan, T. Takahashi, K. Koyama, J. Eur. Ceram. Soc. 33 (2013) 1207–1217. [34] Y.W. Kim, C.M. Wang, C.B. Park, J. Ceram. Soc. Jpn. 115 (2007) 419–424. [35] H.L. Liu, M. Hu, Rare Met. Mater. Eng. 38 (2009) 369–372. [36] M. Shibuya, T. Takahashi, K. Koyama, Compos. Sci. Technol. 67 (2007) 119–124.

K. Lu / Materials Science and Engineering R 97 (2015) 23–49 [37] B.V.M. Kumar, J.H. Eom, Y.W. Kim, J. Ceram. Soc. Jpn. 119 (2011) 48–54. [38] S.H. Chae, Y.W. Kim, I.H. Song, H.D. Kim, M. Narisawa, J. Eur. Ceram. Soc. 29 (2009) 2867–2872. [39] J.H. Eom, Y.W. Kim, M. Narisawa, J. Ceram. Soc. Jpn. 118 (2010) 380–383. [40] J.H. Eom, Y.W. Kim, I.H. Song, H.D. Kim, J. Eur. Ceram. Soc. 28 (2008) 1029–1035. [41] D.H. Jang, Y.W. Kim, I.H. Song, H.D. Kim, C.B. Park, J. Ceram. Soc. Jpn. 114 (2006) 549–553. [42] I.H. Song, I.M. Kwon, H.D. Kim, Y.W. Kim, J. Eur. Ceram. Soc. 30 (2010) 2671–2676. [43] Y.F. Shi, Y. Wan, Y.P. Zhai, R.L. Liu, Y. Meng, B. Tu, D.Y. Zhao, Chem. Mater. 19 (2007) 1761–1771. [44] X.Y. Yuan, H.L. Jin, X.B. Yan, L.F. Cheng, L.T. Hu, Q.J. Xue, Microporous Mesoporous Mater. 147 (2012) 252–258. [45] J. Li, K. Lu, T. Lin, K. Shen, J. Am. Ceram. Soc. 98 (2015) 1753–1761. [46] C. Hoffmann, T. Biemelt, A. Seifert, K. Pinkert, T. Gemming, S. Spange, S. Kaskel, J. Mater. Chem. 22 (2012) 24841–24847. [47] L. Bergero, V.M. Sglavo, G.D. Soraru, J. Am. Ceram. Soc. 88 (2005) 3222–3225. [48] P. Dibandjo, S. Dire, F. Babonneau, G.D. Soraru, J. Non-Cryst. Solids 356 (2010) 132–140. [49] H.J. Kleebe, Y.D. Blum, J. Eur. Ceram. Soc. 28 (2008) 1037–1042. [50] H. Brequel, J. Parmentier, S. Walter, R. Badheka, G. Trimmel, S. Masse, J. Latournerie, P. Dempsey, C. Turquat, A. Desmartin-Chomel, L. Le Neindre-Prum, U.A. Jayasooriya, D. Hourlier, H.J. Kleebe, G.D. Soraru, S. Enzo, F. Babonneau, Chem. Mater. 16 (2004) 2585–2598. [51] C. Brewer, D. Bujalski, V. Parent, K. Su, G. Zank, J. Sol-Gel Sci. Technol. 14 (1999) 49–68. [52] G. Soraru`, D. Suttor, J. Sol-Gel Sci. Technol. 14 (1999) 69–74. [53] H.-J. Kleebe, C. Turquat, G.D. Soraru`, J. Am. Ceram. Soc. 84 (2001) 1073–1080. [54] A. Saha, R. Raj, D.L. Williamson, J. Am. Ceram. Soc. 89 (2006) 2188–2195. [55] G. Gregori, H.J. Kleebe, Y.D. Blum, F. Babonneau, Int. J. Mater. Res. 97 (2006) 710–720. [56] Q. Wei, E. Pippel, J. Woltersdorf, M. Schefﬂer, P. Greil, Mater. Chem. Phys. 73 (2002) 281–289. [57] C.G. Pantano, A.K. Singh, H.X. Zhang, J. Sol-Gel Sci. Technol. 14 (1999) 7–25. [58] R. Pena-Alonso, G.D. Soraru, R. Raj, J. Am. Ceram. Soc. 89 (2006) 2473–2480. [59] G.D. Soraru, R. Pena-Alonso, H.J. Kleebe, J. Eur. Ceram. Soc. 32 (2012) 1751–1757. [60] D.R. Bujalski, S. Grigoras, W.L.N. Lee, G.M. Wieber, G.A. Zank, J. Mater. Chem. 8 (1998) 1427–1433. [61] P. Greil, J. Am. Ceram. Soc. 78 (1995) 835–848. [62] H. Schmidt, D. Koch, G. Grathwohl, P. Colombo, J. Am. Ceram. Soc. 84 (2001) 2252–2255. [63] A. Tamayo, F. Rubio, J. Rubio, J.L. Oteo, J. Am. Ceram. Soc. 96 (2013) 323–330. [64] C. Gervais, F. Babonneau, N. Dallabonna, G. Soraru`, J. Am. Ceram. Soc. 84 (2001) 2160–2164. [65] R. Pena-Alonso, G. Mariotto, C. Gervais, F. Babonneau, G.D. Soraru, Chem. Mater. 19 (2007) 5694–5702. [66] Y. Blum, G.D. Soraru, A.P. Ramaswamy, D. Hui, S.M. Carturan, J. Am. Ceram. Soc. 96 (2013) 2785–2792. [67] P.R. Aravind, G.D. Soraru, Microporous Mesoporous Mater. 142 (2011) 511–517. [68] C.L. Liu, H.X.Z. Chen, S. Komarneni, C.G. Pantano, J. Porous Mater. 2 (1995) 245–252. [69] T.A. Babushkina, T.P. Klimova, N.V. Chizhova, N.N. Makarova, Inorg. Mater. 47 (2011) 916–918. [70] J. Feng, N. Zhao, Y.G. Jiang, J.Z. Feng, Rare Met. Mater. Eng. 41 (2012) 458–461. [71] J. Li, K. Lu, Preparation and properties of highly porous SiOC bulk ceramics, in: MA&T2014, Pittsburgh, PA, 2014. [72] E. Ionescu, H.J. Kleebe, R. Riedel, Chem. Soc. Rev. 41 (2012) 5032–5052. [73] J. Parmentier, G.D. Soraru, F. Babonneau, J. Eur. Ceram. Soc. 21 (2001) 817–824. [74] S. Martinez-Crespiera, E. Ionescu, H.J. Kleebe, R. Riedel, J. Eur. Ceram. Soc. 31 (2011) 913–919. [75] K. Eguchi, G.A. Zank, J. Sol-Gel Sci. Technol. 13 (1998) 945–949. [76] M.A. Mazo, A. Nistal, A.C. Caballero, F. Rubio, J. Rubio, J.L. Oteo, J. Eur. Ceram. Soc. 33 (2013) 1195–1205. [77] T. Lin, K. Lu, Development of conducting SiOC-based composite coating for ferritic stainless steel interconnects, in: MS&T2014, Pittsburgh, PA, 2014. [78] L.Q. Duan, Q.S. Ma, Ceram. Int. 38 (2012) 2667–2671. [79] H.M. Williams, E.A. Dawson, P.A. Barnes, B. Rand, R.M.D. Brydson, Microporous Mesoporous Mater. 99 (2007) 261–267. [80] T. Prenzel, M. Wilhelm, K. Rezwan, Microporous Mesoporous Mater. 169 (2013) 160–167. [81] M. Hojamberdiev, R.M. Prasad, K. Morita, M.A. Schiavon, R. Riedel, Microporous Mesoporous Mater. 151 (2012) 330–338.

49

[82] M. Hojamberdiev, R.M. Prasad, K. Morita, Y.F. Zhu, M.A. Schiavon, A. Gurlo, R. Riedel, Appl. Catal., B 115 (2012) 303–313. [83] P. Colombo, G. Mera, R. Riedel, G.D. Soraru, J. Am. Ceram. Soc. 93 (2010) 1805–1837. [84] R.H. Baney, Poly(phenylsilsesquioxane), in: J.E. Mark (Ed.), Polymer Data Handbook, Oxford University Press, Oxford, 1999, p. 734. [85] J.D. Torrey, R.K. Bordia, C.H. Henager, Y. Blum, Y. Shin, W.D. Samuels, J. Mater. Sci. 41 (2006) 4617–4622. [86] P. Dibandjo, S. Dire, F. Babonneau, G.D. Soraru, Glass Technol.: Eur. J. Glass Sci. Technol., A 49 (2008) 175–178. [87] P.H. Mutin, J. Am. Ceram. Soc. 85 (2002) 1185–1189. [88] S. Dire, R. Ceccato, S. Gialanella, F. Babonneau, J. Eur. Ceram. Soc. 19 (1999) 2849–2858. [89] A. Nyczyk, C. Paluszkiewicz, A. Pyda, M. Hasik, Spectrochim. Acta, A 79 (2011) 801–808. [90] P. Dibandjo, M. Graczyk-Zajac, R. Riedel, V.S. Pradeep, G.D. Soraru, J. Eur. Ceram. Soc. 32 (2012) 2495–2503. [91] Y.D. Blum, D.B. MacQueen, H.J. Kleebe, J. Eur. Ceram. Soc. 25 (2005) 143–149. [92] A. Nyczyk-Malinowska, M. Wojcik-Bania, T. Gumula, M. Hasik, M. Cypryk, Z. Olejniczak, J. Eur. Ceram. Soc. 34 (2014) 889–902. [93] A.K. Singh, G.G. Pantano, J. Am. Ceram. Soc. 79 (1996) 2696–2704. [94] L.Q. Duan, Q.S. Ma, Z.H. Chen, New Carbon Mater. 28 (2013) 235–240. [95] D. Galusek, S. Reschke, R. Riedel, W. Dressler, P. Sajgalik, Z. Lencees, J. Majling, J. Eur. Ceram. Soc. 19 (1999) 1911–1921. [96] T. Liang, Y.-L. Li, D. Su, H.-B. Du, J. Eur. Ceram. Soc. 30 (2010) 2677–2682. [97] G. Tobias, L.D. Shao, C.G. Salzmann, Y. Huh, M.L.H. Green, J. Phys. Chem. B 110 (2006) 22318–22322. [98] T. Saegusa, Macromol. Symp. 98 (1995) 719–729. [99] M. Narisawa, S. Watase, K. Matsukawa, T. Dohmaru, K. Okamura, Bull. Chem. Soc. Jpn. 85 (2012) 724–726. [100] M. Narisawa, K. Terauds, R. Raj, Y. Kawamoto, T. Matsui, A. Iwase, Scr. Mater. 69 (2013) 602–605. [101] A.H. Tavakoli, M.M. Armentrout, M. Narisawa, S. Sen, A. Navrotsky, J. Am. Ceram. Soc. 98 (2015) 242–246. [102] S. Dire, V. Tagliazucca, L. Salvadori, G.D. Soraru, J. Am. Ceram. Soc. 94 (2011) 3819–3824. [103] A. Tamayo, R. Pena-Alonso, J. Rubio, R. Raj, G.D. Soraru, J.L. Oteo, J. Am. Ceram. Soc. 94 (2011) 4523–4533. [104] J. Li, K. Lu, J. Am. Ceram. Soc. 98 (2015) 2357–2365. [105] G.D. Soraru, R. Pena-Alonso, M. Leoni, Microporous Mesoporous Mater. 172 (2013) 125–130. [106] A. Saha, R. Raj, J. Am. Ceram. Soc. 90 (2007) 578–583. [107] G.D. Soraru`, R. Pena-Alonso, M. Leoni, Microporous Mesoporous Mater. 172 (2013) 125–130. [108] H.J. Choi, H.T. Bae, M.J. McNallan, Y.H. Sohn, D.S. Lim, Surf. Coat. Technol. 204 (2009) 1018–1021. [109] Y. Korenblit, M. Rose, E. Kockrick, L. Borchardt, A. Kvit, S. Kaskel, G. Yushin, ACS Nano 4 (2010) 1337–1344. [110] A. Meier, M. Weinberger, K. Pinkert, M. Oschatz, S. Paasch, L. Giebeler, H. Althues, E. Brunner, J. Eckert, S. Kaskel, Microporous Mesoporous Mater. 188 (2014) 140–148. [111] S.H. Yeon, P. Reddington, Y. Gogotsi, J.E. Fischer, C. Vakifahmetoglu, P. Colombo, Carbon 48 (2010) 201–210. [112] C. Vakifahmetoglu, V. Presser, S.-H. Yeon, P. Colombo, Y. Gogotsi, Microporous Mesoporous Mater. 144 (2011) 105–112. [113] A. Delcamp, L. Maille, B. Ruﬁno, S. Mazerat, R. Pailler, A. Guette, P. Weisbecker, H. Plaisantin, E. Philippe, S. Loison, Surf. Coat. Technol. 205 (2010) 703–709. [114] R.K. Dash, A. Nikitin, Y. Gogotsi, Microporous Mesoporous Mater. 72 (2004) 203–208. [115] W. Toury, F. Babonneau, J. Eur. Ceram. Soc. 25 (2005) 265–270. [116] C. Vakifahmetoglu, P. Colombo, Adv. Eng. Mater. 10 (2008) 256–259. [117] J.Q. Wu, Y.M. Li, L.M. Chen, Z.B. Zhang, D. Wang, C.H. Xu, J. Mater. Chem. 22 (2012) 6542–6545. [118] A. Pauletti, S. Handjani, C. Fernandez-Martin, C. Gervais, F. Babonneau, J. Ceram. Soc. Jpn. 116 (2008) 449–453. [119] A. Pauletti, G. Moskowitz, T. Millan, C. Fernandez-Martin, C. Boissiere, C. Gervais, F. Babonneau, Pure Appl. Chem. 81 (2009) 1449–1457. [120] A. Tamayo, J. Rubio, F. Rubio, J.L. Oteo, R. Riedel, J. Eur. Ceram. Soc. 31 (2011) 1791–1801. [121] A. Tamayo, J. Rubio, R. Pena-Alonso, F. Rubio, J.L. Oteo, J. Eur. Ceram. Soc. 28 (2008) 1871–1879. [122] Q.D. Nghiem, S.J. Cho, D.P. Kim, J. Mater. Chem. 16 (2006) 558–562.

Porous and high surface area silicon oxycarbide-based materials—A review

Porous and high surface area silicon oxycarbide-based materials—A review

Recommend Documents