SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedures
  5. III. Results and Data Analysis
  6. IV. Discussion
  7. V. Conclusions
  8. Acknowledgments
  9. References

The structure and properties of polymer-derived Si–(B–)OC glasses have been shown to be significantly influenced by the boron content and pyrolysis temperature. This work determined the impact of these two parameters on the thermodynamic stability of these glasses. High-temperature oxide melt solution calorimetry was performed on a series of amorphous samples, with varying boron contents (0–7.7 at.%), obtained by pyrolysis of precursors made by a sol–gel technique. Thermodynamic analysis of the calorimetric results demonstrated that at a constant pyrolysis temperature, adding boron makes the materials energetically less stable. While the B-containing glasses pyrolyzed at 1000°C were energetically less stable than the competitive crystalline components, increasing the pyrolysis temperature to 1200°C led to their enthalpic stability. 29Si and 11B MAS nuclear magnetic resonance (NMR) spectroscopy measurements on selected samples confirmed a decrease in the concentrations of mixed Si-centered SOiC4−i and B-centered BOjC3−j bonds at the expense of formation of SiO4 and B(OSi)3 species (indicating a tendency toward phase separation) when the boron content and pyrolysis temperature increased. In light of the findings from calorimetry and NMR spectroscopy, we propose a structure–energetic relationship in Si–(B–)OC glasses.

I. Introduction

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedures
  5. III. Results and Data Analysis
  6. IV. Discussion
  7. V. Conclusions
  8. Acknowledgments
  9. References

The greater thermodynamic stability of amorphous Si–O–C and Si–C–N polymer-derived ceramics (PDCs) with respect to the mechanical mixture of their competitive crystalline components (SiO2, SiC, and C in the former, and Si3N4, SiC, and C in the latter) has been demonstrated in the literature.[1-4] Although the stabilizing factors are not completely understood and may vary from one system to another, it has been suggested that the unique and complex structural features containing clusters/nanodomains made of mixed and pure Si-centered tetrahedra (SiCiO4−i, i = 0–4, in Si–O–C PDCs and SiCiN4−i, i = 0–4, in Si–C–N PDCs)[1, 5] as well as bridging chemical bonds between the domains, which may accommodate hydrogen atoms,[3, 4] make the major contributions to the stabilization. In addition to the structural factors, a slight change in the chemistry of PDCs can significantly affect the energetics of the materials. For instance, the Si–B–C–N PDCs with higher boron content have been shown to be thermodynamically less stable.[5]

Gel-derived Si–O–C glasses are a class of Si–O–C PDCs, for which multifunctional properties such as photoluminescence,[6] lithium intercalation,[7] electrical conductivities,[8] and gas sensing[9] have been reported. Addition of boron into Si–O–C glasses increases the decomposition temperature above 1500°C, delays silica crystallization, and promotes SiC crystallization.[10, 11] The structural effects of boron have been studied by nuclear magnetic resonance (NMR) and Raman spectroscopy.[10, 12] The NMR measurements indicate that in addition to Si-centered tetrahedral units (SiCiO4−i; x = 0–4), B-centered trigonal units (BCjO3−j; x = 0–3) are formed.[10, 12] Moreover, the incorporation of boron in the graphene sheets (free-carbon phase) has been suggested.[12]

Although the effect of boron on the high-temperature behavior (delayed crystallization and decomposition) of Si–O–C glasses has been attributed to the change in both chemistry and structure of the amorphous network,[10, 12] it is not yet understood whether these effects are largely kinetic or thermodynamic in origin. The incorporation of boron into the amorphous network could cause more kinetic constraints for mobility of atoms in the structure (kinetic effect) or could lead to the formation of a structure that is energetically more stable and therefore has less tendency to undergo crystallization and decomposition (thermodynamic effect). Therefore, in this work, the enthalpy of formation of the Si–(B–)O–C compositions with varying boron contents in a range from 0 to 7.7 at.% has been studied by high-temperature oxide melt solution calorimetry. In addition, the change in the enthalpic stability with increase in pyrolysis temperature has been investigated. The thermochemical data and the structural insights obtained by NMR spectroscopy have been coupled to draw a structure–energetics correlation for glasses in the Si–(B–)O–C system.

II. Experimental Procedures

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedures
  5. III. Results and Data Analysis
  6. IV. Discussion
  7. V. Conclusions
  8. Acknowledgments
  9. References

(1) Materials and Methods

The gel without boron was prepared by hydrolysis of methyltriethoxysilane (MTES, (CH3Si(OCH2CH3)3; ABCR, Karsruhe, Germany). The three boron-containing gels were synthesized by adding a proper amount of boric acid (B(OH)3; ≥99.5% metals purity, Sigma-Aldrich, St. Louis, MO) to MTES to obtain compositions with nominal B/Si molar ratios of 0.1, 0.3, and 0.5 (detailed information on the synthesis procedure of boron-doped gels can be found elsewhere[13]). The obtained solutions were then cast in plastic tubes and left in air at room temperature for gelation. Gelation time was 2–3 d for the boron-free gel and 15, 15, and 30 d, respectively, for compositions with B/Si ratios of 0.1, 0.3, and 0.5. The recovered wet gels were ground to course powders and then dried in an oven in two steps (50°C for 10 d, 80°C for 5 d). The dried samples were then milled and placed into closed vessels. Coarse gel powders were converted into Si–O–C and Si–B–O–C glasses using a pyrolysis treatment in flowing argon (100 mL/min) at 1000°C and 1200°C (heating rate 5°C/min; dwell time at the maximum temperature 1 h).

(2) Characterization

X-ray diffraction (XRD) patterns were collected using a Bruker AXS D8 Advance diffractometer applying CuKα radiation (40 kV/40 mA).

11B MAS NMR spectra were recorded at 11.75 T on a Bruker (Karsruhe, Germany) Avance500 wide-bore spectrometer operating at 128.28 MHz, using a Bruker 4 mm probe and a spinning frequency of the rotor of 14 kHz. The spectra were acquired using a spin-echo θ – τ − 2θ pulse sequence with θ = 90° to overcome problems of probe signal. The τ delay synchronized with the spinning frequency and a recycle delay of 1 s was used. 29Si MAS NMR spectra were recorded at 7.0 T on a Bruker Avance300 wide-bore spectrometer operating at 59.62 MHz, using a Bruker 7 mm probe, a spinning frequency of 5 kHz and recycle delays of 60 s. Chemical shifts were referenced to BF3(OEt)2 for 11B and tetramethylsilane (TMS) for 29Si. Spectra were fitted using the DMFIT program.[14]

Chemical analysis of hydrogen, oxygen, and carbon was performed using ELTRA ONH-2000 and ELTRA CS 800 C/S (Haan, Germany) equipments based on the combustion techniques. Chemical analysis of silicon and boron was performed by Mikroanalytisches Labor Pascher (Remagen, Germany) applying inductively coupled plasma atomic emission spectroscopy (ICP-AES).

Thermogravimetric (TG) analysis of the ceramic samples was done using a Netzsch (Selb, Germany) 449 TG/DSC instrument under 1 bar Ar atmosphere (25°C–1200°C, heating rate 5°C/min).

(3) Calorimetry

High-temperature oxidative drop solution calorimetry was used to determine the enthalpies of formation of the Si–(B–)O–C glasses. This method is well developed[15-18] and has been applied previously to study a variety of compounds including silicon nitride and oxynitride,[19-21] other oxynitrides,[22, 23] and various groups of PDCs (Si–O–C,[2, 24] Si–C–N,[25] Si–C–N–O,[26] and Si–B–C–N[5]). Using this technique, 1–2 mg pellets, made by pressing the ceramic powders in a 1-mm die, were dropped from room temperature into molten sodium molybdate (3Na2O·4MoO3) solvent at 802°C in a custom built Tian–Calvet twin microcalorimeter[16, 17] under oxidizing atmosphere. The reaction rate of the pellet with solvent was accelerated and oxidizing conditions maintained by bubbling oxygen through the solvent at ~5 mL/min. The pellet dissolved in the solvent within an hour, and converted into oxides and CO2 gas evolved as a result of oxidation. To remove the evolved CO2, oxygen flow at 90 mL/min was used to continuously flush the headspace above the solvent inside the calorimeter. Multiple runs of 1–2 mg pellets were carried out to obtain appropriate statistics. The calorimeter was calibrated using the heat content of platinum rods, as described earlier.[19-21]

III. Results and Data Analysis

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedures
  5. III. Results and Data Analysis
  6. IV. Discussion
  7. V. Conclusions
  8. Acknowledgments
  9. References

(1) Chemistry and Structure

Chemical compositions of the samples pyrolyzed at 1000°C are given in Table 1. As the polymer precursors were prepared with different boron contents, the obtained ceramic samples SiOC, SiBOC1, SiBOC2, and SiBOC3 contain 0, 3.1, 5.7, and 7.7 at.% B, respectively. With increase in the boron content from 0 to 7.7 at.%, Si is reduced from 31.4 to 22.8 at.%, O is increased from 46.7 to 51.2 at.%, and C is decreased from 22.0 to 18.3 at.%. The residual hydrogen, which mostly remains within the microstructure due to conversion of the organic precursor into a ceramic, was measured to be <0.1 wt% (1 at.%). The TG measurements between 1000°C and 1200°C indicated a mass loss <0.2 wt%, which is within the elemental measurement errors. In addition, the isothermal TG measurement performed on one of the samples (SiOC) did not show any detectable mass loss after a dwell time of 1 h at 1200°C. Therefore, for the thermodynamic analysis, the chemical compositions of the glasses pyrolyzed at 1200°C were considered the same as those of the samples pyrolyzed at 1000°C.

Table 1. Elemental Analysis of the Si–(B–)OC Glasses
ElementSample
SiOCSiBOC1SiBOC2SiBOC3
  1. a

    wt%, at.% in parentheses. Uncertainty is two standard deviations of the mean.

Si44.5 ± 1.0 (31.4)a41.5 ± 1.0 (27.4)39.6 ± 1.0 (25.6)35.9 ± 1.0 (22.8)
B1.8 ± 0.1 (3.1)3.4 ± 0.1 (5.7)4.7 ± 0.1 (7.7)
O37.7 ± 0.5 (46.7)42.0 ± 0.5 (48.7)44.1 ± 0.5 (50.0)45.9 ± 0.9 (51.2)
C13.3 ± 0.5 (22.0)13.4 ± 0.2 (20.7)12.4 ± 0.2 (18.8)12.3 ± 0.2 (18.3)
H<0.1 (<1.0)<0.1 (<1.0)<0.1 (<1.0)<0.1 (<1.0)

Figure 1 shows that the XRD patterns of the samples pyrolyzed at 1000°C lack diffraction peaks, indicating absence of crystallinity within the microstructures, which is in agreement with previous reports.[10, 12] Although samples SiOC, SiBOC1 and SiBOC2 stay amorphous when the pyrolysis temperature is increased to 1200°C, the XRD pattern of sample SiBOC3 with the maximum boron content reveals formation of SiC nanocrystallites within the amorphous phase (see Fig. 1).

image

Figure 1. XRD patterns of samples SiOC, SiBOC1, SiBOC2, and SiBOC3 pyrolyzed at 1000°C and 1200°C.

Download figure to PowerPoint

The 29Si MAS NMR spectra of samples SiOC and SiBOC2 pyrolyzed at 1000°C and 1200°C are shown in Fig. 2. All spectra reveal a major component at −109 ppm due to tetrahedrally coordinated Si in SiO4 species in addition to the well-known mixed silicon oxycarbide units[27, 28]: SiO3C at approximately −71 ppm, SiO2C2 at approximately −35 ppm, and SiC4 at approximately −11 ppm. The presence of SiC3O units, expected at approximately −1 ppm, cannot be ruled out but would be in a trace amount if it exists. In agreement with the literature,[29, 30] there is no obvious 29Si chemical shift difference for the tetrahedrally coordinated Si units to distinguish between Si–O–Si and Si–O–B bonds when comparing the SiCiO4−i (x = 0–2) environments in samples SiOC and SiBOC2 pyrolyzed at a given temperature. The simulated spectra, presented in Fig. 2, helped quantify the concentrations of the existing species (see Table 2). Considering samples with the same compositions but different pyrolysis temperatures, the fractions of mixed silicon oxycarbide units diminish when the pyrolysis temperature is increased while the fractions of SiO4 and SiC4 units increase. This trend is more pronounced for the boron-containing glass. The same trend (elimination of mixed units at the expense of formation of pure units) holds when the boron content is increased at constant pyrolysis temperatures (see Table 2). The 11B MAS NMR spectra of sample SiBOC2 pyrolyzed at 1000°C and 1200°C are shown in Fig. 3. The presence of four components with similar quadrupolar parameters, as listed in Table 3, is characteristic of trigonal boron atoms. According to Gervais et al.,[31] the components can be assigned to the following sites: BO3 at 13.9 (12.8) ppm; BO2C at 27.0 (29.1); BOC2 at 45.9 (49.5), and BC3 at 68.4 (68.4) (peak positions out of and in parentheses belong to samples after pyrolysis at 1000°C and 1200°C, respectively). The first peak in both spectra, the major component, is indicative of trigonal B bonded to SiO4 tetrahedra (B(OSi)3 in a borosilicate network)[32] while the peaks corresponding to BO2C and BOC2 indicate that B atoms share bonds with oxygen and carbon (mixed boron oxycarbide units). It is worth mentioning that the presence of some B–O–B bonds at 1000°C cannot be completely excluded considering the signal position. The quantified fractions of the components, given in Table 3, signify the dominance of B(OSi)3 environments over mixed boron oxycarbide units, indicating that boron atoms tend to stay in a borosilicate-like environment. Comparison between the fractions of the same components at two pyrolysis temperatures reveals that the concentration of boron oxycarbide units diminishes in favor of B(OSi)3 and BC3 units when the pyrolysis temperature is increased.

image

Figure 2. Experimental and simulated 29Si MAS NMR spectra recorded on samples SiOC and SiBOC2 pyrolyzed at 1000°C and 1200°C.

Download figure to PowerPoint

Table 2. 29Si MAS NMR Characterization of Samples SiOC and SiBOC2 Pyrolyzed at 1000°C and 1200°C
Sample/pyrolysis temp.Signal intensity of SiCiO4−i (i = 0–4) units (±1%)
SiO4 δ = −109 ± 1 ppmSiCO3 δ = −71 ± 1 ppmSiC2O2 δ = −35 ± 1 ppmSiC4 δ = −11 ± 1 ppm
SiOC/1000°C5031145
SiOC/1200°C6119128
SiBOC2/1000°C5828104
SiBOC2/1200°C731089
image

Figure 3. Experimental and simulated 11B MAS NMR spectra recorded on sample SiBOC2 pyrolyzed at 1000°C and 1200°C.

Download figure to PowerPoint

Table 3. 11B MAS NMR Characterization of Sample SiBOC2 Pyrolyzed at 1000°C and 1200°C
Sample/pyrolysis temp.Signal intensity of BCjO3−j (j = 0–3) units
BO3/B(OSi)3 δ = 13.9/12.8a ± 1 ppm CQb = 2.8 ± 0.1 ηc = 0.0 ± 0.1BCO2 δ = 27.0/29.1 ± 1 ppm CQ = 2.6 ± 0.1 η = 0.3 ± 0.1BC2O δ = 45.9/49.5 ± 1 ppm CQ = 2.8 ± 0.1 η = 0.0 ± 0.1BC3 δ = 68.4/68.4 ± 1 ppm CQ = 2.9 ± 0.1 η = 0.1 ± 0.1
  1. a

    The given chemical shifts are obtained from the spectra of the samples pyrolyzed at 1000°C and 1200°C, respectively.

  2. b

    The quadrupolar coupling constant.

  3. c

    Asymmetry parameter.

SiBOC2/1000°C5923153
SiBOC2/1200°C6413148

(2) Thermodynamic Analysis

According to the results of chemical analysis, compositions SiaOcCd and SiaBbOcCd of samples SiOC, SiBOC1, SiBOC2, and SiBOC3 are located within the three-phase region SiO2–SiC–C and four-phase region SiO2–SiC–C–B2O3 of the phase diagram of the Si–B–O–C system. Thus, the aim of the calorimetric measurements in this work was to obtain the enthalpy of formation of amorphous Si–(B–)O–C PDCs with respect to a mixture of the most stable crystalline components: SiO2, SiC, and C (graphite) for sample SiOC; and SiO2, SiC, C, and B2O3 for samples SiBOC1–3. This enthalpy, ΔHf,comp of Si–(B–)O–C, is associated with the following reaction:

  • display math(1)

When a pellet is dissolved in the solvent during the calorimetric measurement, the oxidative dissolution reaction, represented by Eq. (2), has a corresponding enthalpy, which we call ΔHds of Si–(B–)O–C:

  • display math(2)

In this equation, the final state of the sample is SiO2 (cristobalite) and dissolved B2O3 (for boron-containing ceramics) and CO2 gas is evolved as a result of oxidation. The rationale and evidence for the final state of SiO2 and completeness of oxidation of carbon are explained elsewhere.[24] The measured values of ΔHds of samples SiOC–SiBOC3 are given in Table 4. These values combined with other known thermodynamic quantities were used to calculate the oxidation enthalpy at 25°C, ΔHox, the formation enthalpy from the corresponding elements, ΔHf,elem, and the enthalpy of formation from the components, ΔHf,comp, through the thermodynamic cycles given in Table 5. The enthalpy values obtained for the Si–(B–)O–C glasses are given in three right-hand columns of Table 4. In Fig. 4, the values of ΔHf,comp are plotted versus boron content. For both pyrolysis temperatures, the enthalpy of formation as a function of the boron content changes approximately linearly with a positive slope, indicating that the glasses become enthalpically less stable with increasing boron content. For samples pyrolyzed at 1000°C, while the value of ΔHf,comp for sample SiOC is exothermic [−9.9 ± 2.0 kJ (g·at.)−1], the corresponding values for the B-containing samples are endothermic [4.9 ± 2.2, 11.6 ± 2.5, and 23.8 ± 2.4 kJ (g·at.)−1], which means that the B-doped glasses pyrolyzed at 1000°C are energetically less stable than the competitive crystalline components. For identical compositions, the glasses pyrolyzed at 1200°C have enthalpies of formation more exothermic than those pyrolyzed at 1000°C, indicating that the higher the pyrolysis temperature, the more stable the Si–(B–)O–C glasses. Using the enthalpy values given in Table 4, the exothermic enthalpy changes from 1000°C to 1200°C for the samples containing 0.0, 3.1, and 5.7 at.% were obtained to be −11.6 ± 3.1, −17.7 ± 3.2, and −14.1 ± 3.6 kJ (g·at.)−1, respectively. As seen, the enthalpy changes are almost the same within errors and do not vary systematically with the boron content. Calorimetry was not performed on sample SiBOC3 pyrolyzed at 1200°C because the sample was partially crystalline.

Table 4. Enthalpies of Oxidative Drop Solution at 802°C (ΔHds), Enthalpies of Oxidation at 25°C (ΔHox), Enthalpies of Formation from Elements at 25°C (ΔHf, elem), and Enthalpies of Formation from Crystalline Components SiO2, SiC, C, and B2O3 at 25°C (ΔHf, comp) for the Amorphous Si–(B–)–OC Glasses
SamplePyrolysis temperature (°C)SiaBbOcCd (a + b + c + d = 1)ΔHds [kJ (g·at.)−1]ΔHox [kJ (g·at.)−1]ΔHf,elem [kJ (g·at.)−1]ΔHf,comp [kJ (g·at.)−1]
a b c d
  1. a

    Uncertainty is two standard deviations of the mean. The number in parenthesis is the number of experiments.

SiOC10000.3140.0000.4670.220−127.3 ± 1.8 (6)a−143.7 ± 1.8−227.8 ± 1.9−9.9 ± 2.0
SiBOC110000.2740.0310.4870.207−114.7 ± 2.0 (7)−131.7 ± 2.0−218.9 ± 2.14.9 ± 2.2
SiBOC210000.2560.0570.5000.188−107.8 ± 2.4 (7)−125.6 ± 2.4−216.6 ± 2.511.6 ± 2.5
SiBOC310000.2280.0770.5120.183−102.4 ± 2.3 (6)−120.6 ± 2.3−207.5 ± 2.423.8 ± 2.4
SiOC12000.3140.0000.4670.220−115.7 ± 2.2 (8)−132.1 ± 2.2−239.4 ± 2.3−21.5 ± 2.4
SiBOC112000.2740.0310.4870.207−96.9 ± 2.3 (8)a−113.9 ± 2.3−236.7 ± 2.4−12.8 ± 2.4
SiBOC212000.2560.0570.5000.188−93.8 ± 2.5 (8)a−111.6 ± 2.5−230.6 ± 2.6−2.5 ± 2.6
Table 5. Thermodynamic Cycles for the Determination of (a) Enthalpies of Oxidation inline image, (b) Enthalpies of Formation from the Elements inline image, and (c) Enthalpies of Formation from the Components inline image at 25°C for Amorphous Si–(B–)OC Materials
ReactionEnthalpy (ΔH)
  1. a

    Low cristobalite at 25°C, high cristobalite at 811°C.

  2. b

    Enthalpy of formation of β-SiC.

(a) Enthalpies of oxidation inline imageat 25°C
(1) SiaBbOcCd (solid, 25°C) + (4a + 3b + 4− 2c)/4 O2 (gas, 802°C) [RIGHTWARDS ARROW] a SiO2 (cristobalite, 802°C) + b/2 B2O3 (dissolved, 802°C) + d CO2 (gas, 802°C)ΔH1 = ΔHds [kJ (g·at.)−1]
(2) SiO2 (cristobalite, 25°C) [RIGHTWARDS ARROW] SiO2 (cristobalite, 802°C)ΔH2 = 50.1 kJ/mola[41]
(3) B2O3 (solid, 25°C) [RIGHTWARDS ARROW] B2O3 (dissolved, 802°C)ΔH3 = 136.9 ± 3.1 kJ/mol[5]
(4) O2 (gas, 25°C) [RIGHTWARDS ARROW] O2 (gas, 802°C)ΔH4 = 25.3 kJ/mol[41]
(5) CO2 (gas, 25°C) [RIGHTWARDS ARROW] CO2 (gas, 802°C)ΔH5 = 37.5 kJ/mol[41]
SiaBbOcCd (solid) + (4a + 3b + 4− 2c)/4 O2 (gas) [RIGHTWARDS ARROW] a SiO2 (cristobalite) + b/2 B2O3 (solid) + d CO2 (gas)inline image [kJ (g·at.)−1]
inline image = ΔH1 − aΔH2 − b/2ΔH3 + (4a + 3b + 4d − 2c)/4ΔH4 − dΔH5 
(b) Enthalpies of formation from the elements inline image at 25°C
(1) SiaBbOcCd (solid) + (4a + 3b + 4c − 2c)/4 O2 (gas) [RIGHTWARDS ARROW] a SiO2 (cristobalite) + b/2 B2O3 (solid) + d CO2 (gas)ΔH1 = inline image [kJ (g·at.)−1]
(2) Si (solid) + O2 (gas) [RIGHTWARDS ARROW] SiO2 (cristobalite)ΔH2 = −908.4 ± 2.1 kJ/mol[42]
(3) 2 B (solid) + 3/2 O2 (gas) [RIGHTWARDS ARROW] B2O3 (solid)ΔH3 = −1273.5 ± 1.4 kJ/mol[42]
(4) C (solid) + O2 (gas) [RIGHTWARDS ARROW] CO2 (gas)ΔH4 = −393.5 ± 0.1 kJ/mol[42]
a Si (solid) + b B (solid) + c/2 O2 (gas) + d C (solid) [RIGHTWARDS ARROW] SiaBbOcCd (solid)inline image [kJ (g·at.)−1]
inline image = −ΔH1 + aΔH2 + b/2ΔH3 + dΔH4 
(c) Enthalpies of formation from the components inline image at 25°C
(1) a Si (solid) + b B (solid) + c/2 O2 (gas) + d C (solid) [RIGHTWARDS ARROW] SiaBbOcCd (solid)ΔH1 = inline image [kJ (g·at.)−1]
(2) Si (solid) + C (solid) [RIGHTWARDS ARROW] SiC (solid)ΔH2 = −73.2 ± 6.3 kJ/molb[41]
(4) Si (solid) + O2 (gas) [RIGHTWARDS ARROW] SiO2 (cristobalite)ΔH3 = −908.4 ± 2.1 kJ/mol[42]
(5) 2 B (gas) + 3/2 O2 (gas) [RIGHTWARDS ARROW] B2O3 (solid)ΔH4 = −1273.5 ± 1.4 kJ/mol[42]
(4− 2c + 3b)/4 SiC (solid) + (2c − 3b)/4 SiO2 (cristobalite) + b/2 B2O3 (solid) + (4d − 4a + 2− 3b)/4 C (solid) [RIGHTWARDS ARROW] SiaBbOcCd (solid)inline image [kJ (g·at.)−1]
inline image = ΔH1 − (4a − 2c + 3b)/4ΔH2 − (2c − 3b)/4ΔH3 − b/2ΔH4 
image

Figure 4. Enthalpies of formation of samples SiOC, SiBOC1, SiBOC2, and SiBOC3 with respect to crystalline SiO2 (cristobalite), SiC, graphite, and crystalline B2O3, ΔHf, comp, plotted as a function of the boron content.

Download figure to PowerPoint

As expected, the values of ΔHf,elem are all significantly exothermic (see Table 4) because the enthalpies of formation of components SiO2, SiC, and B2O3 from elements are very exothermic, as seen in Table 5. The change in ΔHf,elem as a function of the boron content follows the same trend as explained above in the case of ΔHf,comp.

IV. Discussion

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedures
  5. III. Results and Data Analysis
  6. IV. Discussion
  7. V. Conclusions
  8. Acknowledgments
  9. References

Our understanding of the local environments of Si and B, found by analysis of the NMR spectra, combined with the previous findings on the architecture of the amorphous carbon network in Si–(B–)O–C glasses[12, 33] helps describe the likely microstructure at the nanoscale. According to the model proposed by Corriu et al.[34] for silicon oxycarbide species with a random distribution of Si–O and Si–C bonds, the probability of finding a site SiCiO4−i, P(SiCiO4−i), depends on the probability of finding Si–O, PSi–O, and Si–C, PSi–C, bonds in the oxycarbide phase

  • display math(3)

PSi–O and PSi–C depend on the O/Si ratio of the glass (PSi–O = (O/Si)/2; PSi–C = 1 − PSi–O). Accordingly, P(SiCiO4−i) for sample SiOC in this work was calculated, assuming random distribution of Si–O and Si–C bonds: P(SiO4) = 0.306; P(SiCO3) = 0.422; P(SiC2O2) = 0.217; P(SiC3O) = 0.050; and P(SiC4) = 0.005. Comparing the obtained values with the fractions of SiCiO4−i sites calculated by simulation of the 29Si MAS NMR spectra of sample SiOC pyrolyzed at 1000°C (see Table 2), it was found that the concentration of SiO4 sites is almost 70% more than the value calculated using Eq. (3). This mismatch rules out a random distribution of Si–O and Si–C bonds in this Si–O–C glass and strongly suggests heterogeneity in the silicon oxycarbide phase at the nanoscale, as widely proposed for Si–O–C PDCs.[1, 12, 35, 36] Concentration gradients of Si–O and Si–C bonds within the silicon oxycarbide phase can be suggested through formation of oxygen-rich and carbon-rich regions with a length scale of about a few tetrahedral units for each. Carbon-rich SiCiO4−i structural units are expected to form in vicinity of the free-carbon phase. Upon increasing the pyrolysis temperature to 1200°C, the region containing oxygen-rich SiCiO4−i structural units appears to grow at the expense of diminished carbon-rich SiCiO4−i tetrahedra, in accord with analysis of the NMR results in Table 2 indicating the increased concentration of SiO4 sites and the diminished concentrations of mixed silicon oxycarbide bonds. Furthermore, thickening and ordering of carbon layers in the free-carbon phase is likely to occur with the increase in pyrolysis temperature, as suggested in the previous study.[12] In case of Si–B–O–C glasses, formation of a boron-doped carbon phase and a borosilicate-like phase is possible. The partial substitution of C by B atoms in the free-carbon phase has been suggested before based on the analysis of Raman spectra of the Si–B–O–C glass.[12] This substitution is supported by the 11B MAS NMR spectra in this work, showing the signal corresponding to BC3 sites (see Fig. 3). As the major contributions of B(OSi)3 and SiO4 environments were confirmed by analysis of the 11B and 29Si MAS NMR spectra of the Si–B–O–C glass, the borosilicate-like phase is believed to consist of a mixture of oxygen-rich SiCiO4−i and oxygen-rich BCjOj−3 units in addition to a minor contribution of a mixture of carbon-rich SiCiO4−i and carbon-rich BCjOj−3 units near the boron-doped carbon layers. Whether these changes in abundances are gradational or represent two distinct regions (one containing the oxygen-rich units and one near the boundary with carbon) is not known, and, if the oxygen-rich regions are on the order of 1 nm thick, they would contain only three to four structural units, making the question of phase separation essentially moot. A structural change upon increasing pyrolysis temperature with the same pattern as explained in case of the Si–O–C glass is expected to occur for the Si–B–O–C glasses: demixing of mixed silicon oxycarbide and boron oxycarbide units, ordering of the boron-doped carbon phase and thickening of carbon layers.

The calorimetric data show that the incorporation of B into the Si–O–C glass is energetically unfavorable relative to the crystalline components. On the other hand, adding boron raises the configurational entropy, through which the entropy term makes an exothermic contribution in the Gibbs free energy of formation of the Si–B–O–C glass (ΔGf = ΔHf − TΔS). However, our previous study on Si–(B–)C–N PDCs revealed that the magnitude of the TΔS term, originating from adding B, does not exceed 1.8 kJ (g·at.)−1 at room temperature.[5] Accordingly, the enthalpy term remains the leading term for determination of the thermodynamic stability. Thus, we conclude that the thermodynamic stability of the Si–(B–)O–C glasses diminishes with the increase in boron content. Partial crystallization of sample SiBOC3 after pyrolysis at 1200°C while samples SiOC, SiBOC1, and SiBOC2 with less boron remain amorphous (see Fig. 1) can be attributed to a greater driving force for crystallization originating from the lower thermodynamic stability of sample SiBOC3, as concluded above. The diminished thermodynamic stability of the glasses with adding B can be attributed to local disordering due to the formation of B-centered trigonal units within the silicon oxycarbide phase containing Si-centered tetrahedral units. Disrupting the linkages of Si tetrahedra by introducing B trigonal units can induce less stable bonding, which would lead to an increase in the internal energy (destabilization) of the system. Furthermore, lower thermodynamic stability of borosilicate glasses with respect to amorphous SiO2 and B2O3, as reported in the literature,[37-39] supports our argument noted above because the structure of oxygen-rich silicon boron oxycarbide phase in the Si–B–O–C glasses is to some extent similar to silicon-rich borosilicate glasses.

In addition to the effect of boron on the thermodynamic stability, it was found that pyrolysis at higher temperature leads to the stabilization of the Si–B–O–C glasses and a greater stability of the Si–O–C glass. A similar trend has been recently reported for the Si–(B–)C–N PDCs.[40] As the chemistry of the glasses remains unchanged with the increase in temperature, the enthalpy change from 1000°C to 1200°C is attributed to the structural change. Indeed, pyrolysis at the higher temperature results in a lesser degree of disordering and a more favorable arrangement of the structural units. A more ordered structure at higher temperature appears, at first glance, to be inconsistent with thermodynamic principles, which would favor more disorder, a more positive enthalpy and a more positive configurationally entropy at higher temperature. This dilemma can be resolved in at least two ways. First, if the materials pyrolyzed at a lower temperature contain more hydrogen (even at low levels), the evolution of this hydrogen on heating would provide the entropic driving force, as has been argued for Si–C–N PDCs.[3] Second, if the glasses pyrolyzed at 1000°C are not fully relaxed into their most favorable structural state (have not attained equilibrium), heating at a higher temperature may allow irreversible transformation to the lowest free energy state. Moreover, relaxation of the structure due to demixing of mixed SiCiO4−i and BCjO3−j bonds, as shown by the NMR results, and coarsening of silicon oxycarbide and free-carbon domains[12] are additional factors that can contribute to the stabilization of the glasses at the higher pyrolysis temperature and eventually lead to crystallization.

V. Conclusions

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedures
  5. III. Results and Data Analysis
  6. IV. Discussion
  7. V. Conclusions
  8. Acknowledgments
  9. References

The analysis of calorimetric measurements on a set of Si–(B–)O–C polymer-derived glass compositions demonstrates that adding B into the Si–O–C glass diminishes the thermodynamic stability with respect to the competitive crystalline components. With support of NMR measurements, the diminished thermodynamic stability is suggested to be connected to the increased net internal energy of the system due to the formation of BCjO3−j trigonal units within the silicon oxycarbide phase, which could disrupt the linkages of SiCiO4−i tetrahedral units in the Si–O–C glass. The increase in the pyrolysis temperature leads to energetic stabilization of the B-containing glasses and a greater stability of the B-free Si–O–C glass.

Acknowledgments

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedures
  5. III. Results and Data Analysis
  6. IV. Discussion
  7. V. Conclusions
  8. Acknowledgments
  9. References

The authors acknowledge financial supports by the National Science Foundation, grant MWN-0907792 and the EU through the MC-ITN FUNEA, CT-264873.

References

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedures
  5. III. Results and Data Analysis
  6. IV. Discussion
  7. V. Conclusions
  8. Acknowledgments
  9. References