Thermal Conductivity of Carbon/Boron Nitride Heteronanotube and Boron Nitride Nanotube Buckypapers: Implications for Thermal Management Composites

To date, there has been limited reporting on the fabrication and properties of macroscopic sheet assemblies (specifically buckypapers) composed of carbon/boron nitride core–shell heteronanotubes (MWCNT@BNNT) or boron nitride nanotubes (BNNTs). Herein we report the synthesis of MWCNT@BNNTs via a facile method involving Atmospheric Pressure Chemical Vapor Deposition (APCVD) and the safe h-BN precursor ammonia borane. These MWCNT@BNNTs were used as sacrificial templates for BNNT synthesis by thermal oxidation of the core carbon. Buckypaper fabrication was facilitated by facile sonication and filtration steps. To test the thermal conductivity properties of these new buckypapers, in the interest of thermal management applications, we have developed a novel technique of advanced scanning thermal microscopy (SThM) that we call piercing SThM (pSThM). Our measurements show a 14% increase in thermal conductivity of the MWCNT@BNNT buckypaper relative to a control multiwalled carbon nanotube (MWCNT) buckypaper. Meanwhile, our BNNT buckypaper exhibited approximately half the thermal conductivity of the MWCNT control, which we attribute to the turbostratic quality of our BNNTs. To the best of our knowledge, this work achieves the first thermal conductivity measurement of a MWCNT@BNNT buckypaper and of a BNNT buckypaper composed of BNNTs not synthesized by high energy techniques.

Table S2: Summary of BNNT , BCNT and CNT+BNNT macroscopic sheet assemblies reported in the literature, sorted according to the method of fabrication.The dimensions of nanotubes in the assemblies are compiled, along with the macroscopic assemblies' processing steps and overall thickness and density.Electron Energy Loss Spectroscopy

X-ray Photoelectron Spectroscopy
Figure S9: XPS High Resolution C1s spectra of MWCNTs and MWCNT@BNNTs, deconvoluted according to the data shown in Table S4 below.The two spectra are very similar, with no obvious change in the C1s of the MWCNT@BNNT to suggest significant covalent interactions between the MWCNT core and h-BN shell.Oxidised carbon species appear to be reduced in the MWCNT@BNNT.However oxygen in still present in the MWCNT@BNNT bonded to the h-BN, as evidenced in the B1s and N1s spectra of Fig. 3.
Table S4: Summarised description of synthetic peaks fitted to the high resolution XPS C1s spectrum of Nanocyl NC7000 MWCNTs and MWCNT@BNNTs as shown in Figure S9 above.Line shapes for peaks are Gaussian-Lorentzian, with the exception of the asymmetric modified Lorentzian LF lineshape for fitting sp 2 carbon due to its XPS final state effects. [12]ond identification is referenced from [12] and. [13]CNT Table S5: Summarised description of synthetic peaks fitted to the high resolution XPS B1s and N1s spectrum of MWCNT@BNNTs as shown in Fig. 3G,H.High resolution peaks were fitted in CasaXPS with a GL-T lineshape for the B-N bonds [14] and GL lineshapes for other bonds.Bond identifications are referenced from.[15] B1s N1s Bond  Table S7: Summary of FTIR absorption peaks recorded from spectra of MWCNT@BNNT, BNNT (700 • C), BNNT (900 • C) and reference h-BN powder from 3M™ (shown in Figure S12B).Note that the peak at ca. 2360 cm −1 is a result of atmospheric and/or adsorbed CO 2 .

UV-Vis Spectroscopy
FTIR Frequencies (cm −1 ) h-BN 3M™ MWCNT@BNNT BNNT (700   S8 below for spectra analysis.If B and N had become primarily covalently incorporated into the graphitic lattice of the MWCNT, an increase in defect density would be detected in the Raman spectra of the MWCNT@BNNT as an increase in the I D /I G ratio relative to that of the pristine MWCNTs.However, I D /I G of MWCNT@BNNT Raman spectra do not exhibit such an increase, thereby substantiating that this synthesis technique favours h-BN deposition as a shell around the MWCNT core, the latter which maintains it graphitic lattice structural integrity.Figure S14: (A) Raman spectra collected from (i) a reference crystalline commercial 3M™ h-BN sample and (ii-iii) BNNTs synthesised by sacrificial templating.Laser wavelength used for (i-ii) was 532 nm, but 632.8 nm for (iii) to test fluorescence suppression. [20](B) A magnification of the E 2g mode for (i) reference h-BN powder and (iii) the BNNTs (laser λ 632.8 nm) for clearer comparison of peak position and FWHM.FWHM are calculated from Lorentz peak fittings.Besides fluorescence suppression, the BNNT Raman spectra, and particulary the E 2g mode, is also likely influenced by anharmonic effects such as temperature dependence. Without accounting for such effects, and according to Nemanich et al. [24] a smaller crystalline size (L a ) of h-BN results in broader FWHM of the E 2g mode and an up-shift in peak frequency.
Figure S16: Electrical resistivity measurements of buckypapers reveal an increased resistivity of the MWCNT@BNNT buckypaper relative to the MWCNT, Annealed MWCNT and * buckypapers (Annealed: treated to 1000 • C in inert atmosphere as a control).The * buckypaper was manufactured by coating a pre-fabricated MWCNT buckypaper with h-BN instead of fabricating a buckypaper with MWCNT@BNNT powder.Buckypaper density variation is not considered to be a contributing factor to the noticeable MWCNT@BNNT buckypaper resistivity increase.

Figure S2 :
Figure S2: Illustration of CNT sample setup inside the reactor, i.e. layer of CNT on quartz slide.

2 -Figure S4 :
Figure S4: Example HRTEM images of BNNTs synthesised here by sacrificial templating.These images evidence the disordered structure of the BNNT walls.The achieved turbostratic BNNT wall structure is magnified in (B), showing the disordered h-BN lamellae and a broad (002) spacing of ca.0.36 nm.

Figure S5 :
Figure S5: TEM images depicting BNNTs with different tip structures, e.g.closed vs open ended.

Figure
Figure S6: (A) Diameter spread and (B) Wall thickness distributions of MWCNT, MWCNT@BNNT and BNNTs as measured from HRTEM images.

Figure S7 :
Figure S7: MWCNT@BN specimen observed by (A)Dark Field STEM with the nanotube tip mapped by (B) EELS spectral imaging.Complete tip encapsulation by BN is illustrated in (C).EELS spectra are extracted from the coordinates labelled 1-5 (D), indicating the h-BN composition of the outer shell.An atypical appendage is observed at coordinate 5. Increasing intensity of B-K edge σ * peak relative to π * (outer shell to inner core) indicates 360 • h-BN encapsulation (F).

Figure S8 :
Figure S8: Example of MWCNT@BNNTs observed by (A) Dark Field STEM and (B) EELS spectral imaging, which reveals a BN shell around a C core.Elements detected, including oxygen, are summarised in (C) the correspond EELS map spectrum.Fine edge structures of B/N and C exhibit π * and σ * features typical of h-BN and graphitic C.

Figure
Figure S10: (A-C) UV-Vis spectra of raw Nanocyl NC7000 MWCNT, and synthesised MWCNT@BNNT and BNNTs.(D-F) Corresponding Tauc plot with extrapolated optical band gap values shown.Given that the multiple different band gaps extrapolated from the Tauc plot of MWCNT@BNNT can be individually matched to both the MWCNT and h-BN components, the van der Waals hybridisation of these nanotubes is further corroborated.
Figure S13: Raman spectra of Nanocyl NC7000™ MWCNTs, annealed MWCNTs (to 1000• C in Ar/H 2 ), and MWNCT@BNNT.Six spectra were taken from each sample using a 532 nm wavelength laser.D, G and 2D band positions are highlighted.See TableS8below for spectra analysis.If B and N had become primarily covalently incorporated into the graphitic lattice of the MWCNT, an increase in defect density would be detected in the Raman spectra of the MWCNT@BNNT as an increase in the I D /I G ratio relative to that of the pristine MWCNTs.However, I D /I G of MWCNT@BNNT Raman spectra do not exhibit such an increase, thereby substantiating that this synthesis technique favours h-BN deposition as a shell around the MWCNT core, the latter which maintains it graphitic lattice structural integrity.

Figure S17 :
Figure S17: Schematics of the SThM setup.The Wheatstone bridge operates at AC frequency of 91 kHz to reduce the 1/f noise of the probe, with the probe power provided both by the AC voltage and DC offset.

Table S3 :
Quantitative analysis of EELS spectral images.Atomic ratios are presented relative to carbon areal density.Percent contents are presented based on absolute areal density comparisons.

Table S6 :
[19]arised description of synthetic peaks fitted to the high resolution XPS B1s and N1s spectrum of BNNTs as shown in Fig.3I,J.High resolution peaks were fitted in CasaXPS with a GL-T lineshape for the B-N bonds and GL lineshapes for other bonds.Bond identifications are referenced from.[15]-[19]

Table S8 :
Measured average D,G and Raman peak parameters (Raman shift, FWHM) and intensity ratios of I D /I G and I 2D /I G for MWCNTs, Annealed MWCNTs and MWCNT@BN.Errors are standard deviations.