High-Purity Boron Nitride Nanotubes via High-Yield HydrocarbonSolvent ProcessingMahmoud S. Amin,† Bennett Atwater,† Robert D. Pike,‡ Kurt E. Williamson,§ David E. Kranbuehl,‡

and Hannes C. Schniepp*,†

†Department of Applied Science, ‡Department of Chemistry, and §Department of Biology, The College of William & Mary,Williamsburg, Virginia 23185, United States

*S Supporting Information

ABSTRACT: Boron nitride nanotubes (BNNTs) are highly interesting for structural,thermal, and biomedical applications but have, thus, far been limited due to a lack of high-yield methods to produce a purified material. In particular, previous attempts have failed tofully remove h-BN due to its chemical similarity with the BNNTs. Here, we propose a mild,high-yield method that removes >99% of the h-BN impurities without any apparent loss ordamage to the BNNTs. This is in contrast to all prior methods, which have used aggressivechemical, mechanical, or thermal methods and, therefore, suffered from low yields andsignificant damage to the tubes. Our method relies on exposure to heptane at a moderatetemperature of T = 90 °C in a pressure vessel to detach h-BN impurities from the BNNTs. We showed that X-ray diffractionand Raman spectroscopy are very effective to quantitatively detect the degree of h-BN removal.

■ INTRODUCTION

Interest in boron nitride nanotubes (BNNTs) has growndramatically over the past two decades because of their uniquemechanical and thermal properties.1,2 Like carbon nanotubes,BNNTs display exceptional strength, an axial Young’s modulusof up to 1.3 TPa,3 and thermal stability in air up to 800 °C.4−13

Despite being an electrically insulating semiconductor with a5−6 eV bandgap,14−16 BNNTs have been suggested to featurea surprisingly high thermal conductivity of 3000 W m−1 K−1.17

These properties make BNNTs a highly promising material forthermal management applications in high power electronicsand photonics. They are also a strong candidate for hydrogenstorage,18,19 structural reinforcement of composites andalloys,4,20 water treatment and desalination,21 and deep UVoptoelectronics.22 Realization of these promising applicationshas been hindered by significant challenges in BNNT synthesisand purification, which has proven far more difficult comparedto carbon nanotubes (CNTs).23,24 In particular, the mostimportant BNNT production methods all yield tubes with asubstantial amount of impurities; most notably, these tubescontain hexagonal boron nitride (h-BN), which has provenvery challenging to remove. Here, we report for the first time amethod that virtually completely removes h-BN impuritieswithout damaging the BNNTs, thus achieving a near-perfectyield, representing an important step toward the scalableproduction of high-quality BNNTs. Our claims are supportedby two spectroscopic methods we developed to quantitativelyassess h-BN removal. Such spectroscopic techniques character-ize a macroscopic region of the sample and are, thus, moreinformative and more quantitative than imaging of selectedareas.Of the many BNNT synthesis methods in the literature, only

four produce tubes at the gram scale:1,25 high-temperature/

pressure synthesis (HTP),26 hydrogen-assisted BNNT syn-thesis (HABS),27 extended-pressure inductively-coupled plas-ma synthesis (EPIC),28 and ball milling/annealing.29,30 Allcurrently known production methods produce tubes with asignificant amount of impurities. The most common impuritiesare boron, its oxides and different forms of boron nitride,which typically amount to 30−60% by weight in as-receivedBNNTs.31 These impurities significantly degrade the proper-ties of BNNTs and hinder the achievement of the anticipatedoutstanding properties of pure tubes.13,24 To better realize thefull potential of BNNTs, research groups have investigated anumber of purification techniques.13,24,25,32−37

Methods for removing boron and boron oxides areknown13,23,31,34−37 and have been reasonably effective. Themajor problem is removing boron nitride (BN) impurities,which occur as nanoscale and microscale particles. Micron-sized BN impurities have been successfully removed using apolymer wrapping method;34 however, the same method hasproven ineffective in removing nanoscale BN impurities.34

Nanoscale BN impurities are extremely challenging to remove,as they are very similar to BNNTs in terms of their chemistryand bonding affinity. This makes it very difficult to developselective chemical approaches.28−30,38−43 The methods thathave been developed to selectively remove impurities in CNTshave proven ineffective for BNNTs.23 Previous attempts ofpurifying BNNT have included the use of acids, surfacemodification, mechanical methods (sonication and ball mill-ing), and oxygen treatments at temperatures >700°C.13,24,44−46 However, acid treatment has shown to damage

Received: May 1, 2019Revised: September 26, 2019Published: October 11, 2019

Article

pubs.acs.org/cmCite This: Chem. Mater. 2019, 31, 8351−8357

© 2019 American Chemical Society 8351 DOI: 10.1021/acs.chemmater.9b01713Chem. Mater. 2019, 31, 8351−8357

Dow

nloa

ded

via

CO

LG

OF

WIL

LIA

M &

MA

RY

on

Febr

uary

24,

202

0 at

17:

05:5

4 (U

TC

).Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

and erode the boron nitride nanotubes’ structure.24 A similarproblem was reported for carbon nanotubes where nitric acidremoves impurities at the expense of damaging the tubes.47

Surface modification using polymers followed by sonicationand centrifugation displays limitations similar to polymerwrapping in that it damages the tubes as well as producing lowyields.37 Sonication and other mechanical methods have beenproposed to separate large micron-sized impurities and isolatepure BNNT;23 however, these methods significantly damagepristine BNNT tubes and break apart the tube networkstructure.23 Mechanical methods also suffer from low yields,especially when tuned to reach high levels of purity.23,35 High-temperature oxygen treatments conducted up to temperatureswhere BNNTs start to decompose are reported to havevariable purity and low yields of 10−20% purified BNNT.13

Due to the inability of these different methods to remove BNimpurities, a new method is needed to overcome thesedeficiencies.Here, we demonstrate a novel, nondestructive method for

the removal of boron nitride impurities including h-BN fromBNNTs using heptane at a moderate temperature. Ourapproach focuses on tubes commercially manufactured usingthe HTP method, which produces a large fraction of high-quality tubes featuring 2−5 walls and some of the smallestdiameters in the range 5−10 nm.24,26 The HTP tubes are alsoamong the longest ones available, about 1 μm in length.24,26

Further, HTP has the advantage of producing tubes of highcrystallinity and with few defect sites.24 Our new processdescribed here represents a significant advancement in BNNTpurification, producing undamaged BNNTs using methodsthat are easily scalable in an economically viable way.Moreover, we describe two instrumental techniques to detectthe presence of h-BN impurities and to verify their removalbeyond the qualitative results obtained by imaging. Wedemonstrate that the h-BN peaks in both nonresonanceRaman and X-ray diffraction (XRD) are very effective indetecting h-BN impurities. Limited by the signal-to-noise ratioof our measurements, our method removed >99% of the h-BN.We further used FTIR to confirm the removal of h-BN. Acombination of our novel, moderate-temperature purificationprocedure using a hydrocarbon with techniques to detectimpurities creates a clear path toward achieving and verifyingBNNT purity on an application scale.

■ EXPERIMENTAL METHODSBNNT Purification. The starting material was “purified BNNT”

(BNNT LLC, Newport News, VA), which was produced based on thehigh-temperature and -pressure (HTP) methods. According to themanufacturer, the purified BNNT material contained <1% boronimpurities. We mixed 50 mg of this material with 30 mL of heptane(Sigma-Aldrich, Raleigh, NC) in a 45 mL Ace pressure tube (AceGlass Incorporation, Vineland, NJ). The system was heated andmaintained at 90 °C for 5 h in an oil bath. The system was cooled toroom temperature before the pressure tube was opened. Heptane wasseparated from the BNNT sample by decantation. The purifiedBNNT was dried in a vacuum oven under <10 mbar at 250 °Covernight. The decanted heptane solution containing the suspendedBN impurities was concentrated in a vacuum oven at a pressure <10mbar and T = 60 °C for further investigation.

FESEM Imaging. Field emission scanning electron microscopy(FESEM) was carried out using a Hitachi S-4700 FESEM equippedwith a secondary electron detector; the samples were run at anacceleration voltage of 10 kV with a 7.5−8.0 mm working distance.For the sample preparation, the BNNTs were dispersed inisopropanol, pipetted onto graphite tape, spin-dried, and sputter-coated (Anatech LTD, Hummer 6.2) with a ≈2 nm layer of gold/palladium to prevent charging. The obtained coating based on thepressure of the coating system, 0.08 mbar, was relatively rough, as canbe seen in all higher-resolution SEM images.

TEM. Transmission electron microscopy (TEM) was carried outusing a Philips CM 10 TEM at 80 kV. Micrographs were obtainedusing an AMT XR80S-B 8MP camera. The samples were prepared bydrop casting a BNNT dispersion onto lacey 300 mesh carbon TEMgrids.

XRD Analysis Method. XRD analysis was performed using aBruker APEX DUO diffractometer equipped with an APEX II CCDDetector and a microfocus copper Kα source (wavelength λ = 1.54Å). The samples were prepared by compressing 20 mg of BNNTmaterial in discs with a diameter of 3 mm. To address possibleinhomogeneities of the BNNT samples, different locations on theBNNT disc were measured.

Raman Spectroscopy. Raman data was collected using aRenishaw inVia dispersive Raman spectrometer using 514 nm withan excitation power of 40 mW and a 100× objective with a 0.90 NA.The samples for the as-received BNNT and purified BNNT wereprepared by compressing the materials into 3 mm diameter discs forhigher density and a better Raman signal. Sample preparation for theresidual impurities was done by drop casting the concentratedheptane solution on a silicon wafer. The Raman spectrum had a broadbackground over the range of 500−7000 cm−1 due to fluorescence.This background was removed using a second-order polynomial fit.

Figure 1. (a, b) As-received BNNT; yellow arrows in (a) pointing to impurities. (c−e) FESEM (c, d) and TEM (e) of BNNT after purification. (f)Purification residue (FESEM).

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.9b01713Chem. Mater. 2019, 31, 8351−8357

8352

FTIR. Fourier transform infrared spectroscopy (FTIR) measure-ments were made using a Shimatzu IRTracer-100 FTIR spectropho-tometer with a reflection ATR accessory on the as-received BNNT,the heptane-purified BNNT, and the h-BN. An average of 32 runswith a resolution of 4 cm−1 was used. The spectra shown hererepresent the absorbance (common logarithm of the intensity ratio ofincident to transmitted light). To account for different amounts ofmaterials tested, the spectra were normalized for their maximum to be1.0.

■ RESULTS AND DISCUSSIONField emission scanning electron microscopy (FESEM) andtransmission electron microscopy (TEM) were first conductedbefore the purification procedure (Figure 1a,b, respectively).This as-received material consisted of a network of tubesextending microns in length.24 This structure is generallyunderstood to be a result of the BNNT growth process, wheregrowth originates from droplets of pure boron contained infullerene-like BN cages (“nanococoons”), acting as seeds forthe growth of bundles of tubes.39,48 These tubes ultimatelyterminate when they run into another nanococoon, thusleading to a network structure.49 The larger impurities attachedto the BNNTs have been proposed to be remnants of the BNnanococoons and BN nanoparticles.39,48 As indicated in Figure1a by yellow arrows, these impurities form aggregates on thetube surfaces and nodes.Subsequently, the BNNTs were cleaned by our low-

pressure/moderate-temperature solvent method, whereBNNT was added to a pressure tube containing heptane andsubsequently heated for 5 h at 90 °C. The heptane-cleanedBNNT material was collected and imaged by FESEM (Figure1c,d) and TEM (Figure 1e). As evident in these micrographs,the heptane-cleaned tubes, microns in length, were virtuallyfree of impurities on the surface, suggesting that the cleaningprocedure has been highly successful. We note that the surfaceroughness observed in the higher-magnification FESEM imagein Figure 1d is not due to residual contamination but reflectsthe imperfect sputter coating applied for imaging purposesonly. The FESEM and TEM images after purification show noevidence of broken tubes, damage to the tube surface, ordamage to the tube network. The TEM image (Figure 1e)suggests that a small amount of impurities remained. Thisresidual impurity is most likely due to small amounts of variousforms of BN, boron oxide, and boron. Based on theseobservations, we suggest that virtually all of the boron nitridetube network remains intact and undamaged, suggesting a verygentle purification with a very high yield. Measurements of therecovered mass after heptane purification indicated a weightloss of 6−12%. Analyzing the decanted cleaning residuecollected after drying the heptane filtrate provided importantinsight into the composition of the material removed from thenanotubes via our purification method. FESEM analysis(Figure 1f) revealed that the cleaning residue consisted ofsuspended, micron-size particles with features hinting at astacked surface structure. More FESEM and TEM images atvarying magnifications are shown in the Supporting Informa-tion.To further understand the cleaning process, we carried out

X-ray diffraction (XRD) of the BNNT material before andafter cleaning (Figure 2). The as-received BNNT X-ray pattern(black curve) had a pronounced series of peaks in the range 2θ= 20−30°, corresponding to d-spacings in the range 4.4−3.0 Å,where interlayer spacings for h-BN and other 2D materials aretypically found.50 The spectrum featured a second series of

peaks in the range 2θ = 40−55°, corresponding to d-spacingsin the range 2.2−1.7 Å, likely representing the intralayerspacings between the atoms of the in-plane honeycomb lattice.The first peak series was analyzed using the Lorentzian peakfitting, with the main peak at 25.33° and a sharp subpeak onthe right (26.69°) with full widths at half maximum (FWHM)of 2.30 and 0.18°, respectively. To help interpret the measuredBNNT peak positions, we overlaid h-BN peaks based onliterature values in blue.50 We found a remarkable agreementbetween the position of the sharp shoulder at 26.69° of the as-received BNNT and the h-BN [0002] peak at 2θ = 26.63°.The presence of this h-BN peak demonstrates for the first timethat HTP-produced BNNT contains a significant amount of h-BN.When we carried out XRD on the heptane-purified BNNT

(Figure 2, red curve), we first found that the main peak wasvirtually at the same position (25.42°), albeit slightly narrower(FWHM 1.82°). Most interestingly, however, the sharpsubpeak at 26.69° was completely absent, which suggeststhat our cleaning procedure virtually completely removed h-BN impurities from the sample. To quantify the degree of h-BN removal, we performed a more substantial analysis of thespectra (see the Supporting Information). Our analysisrevealed no detectable peak at 26.69°; the detection limitwas given by the general noise level of the normalized XRDspectrum in this spectral region, about 0.5 intensity units(RMS noise). The main peak of our XRD spectra wasnormalized to 1000 intensity units; the height of the 26.69° h-BN peak of the unpurified sample was 284.1 in the same units(see Table 1). Since this peak was reduced to <0.5 in the

Figure 2. XRD pattern of as-received BNNTs (black curve) showed apeak at 25.33° with a shoulder at 26.69°. The shoulder positionmatches a known peak of the h-BN XRD pattern (blue lines), shownin more detail in the inset. In the XRD pattern of the purified BNNT(red curve), the shoulder is completely absent.

Table 1. XRD Fitting Parameters

sample ID peakpeak position

(deg)FWHM(deg)

height(a.u.) R2

as-receivedBNNT

main 25.33 2.30 1000.0 0.94h-BN 26.69 0.18 284.1

purified BNNT main 25.42 1.82 1000.0 0.98

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.9b01713Chem. Mater. 2019, 31, 8351−8357

8353

purified spectrum, our purification procedure reduced the h-BN signal by at least 99.8%. Interestingly, the spectra betweenas-received and purified materials looked very similar in the 2θ= 40−55° range. This indicates that the in-plane lattice of thepurified BNNT was not significantly altered by our purificationprocedure, which suggests that our cleaning procedure doesnot damage the tubes. Consequently, the XRD data shows thatour cleaning procedure suppresses the h-BN content <0.2%without harming the tubes.We further analyzed the effectiveness of our purification

technique to remove h-BN using Raman spectroscopy. Thepredicted bandgap of BNNTs is 5.5 eV, which wouldcorrespond to a photon wavelength of 225 nm; the bandgapof h-BN is 5.9 eV, corresponding to a wavelength of 210 nm. Arecent, rigorous study has shown that BNNTs show a peak at≈1370 cm−1 when studied using resonant Raman spectrosco-py, carried out at λ = 229 nm.51 In nonresonant Ramanspectroscopy, however, no characteristic peaks in this spectralregion are present. For h-BN, in contrast, nonresonant Ramanspectroscopy does show a pronounced peak at 1364.6 cm−1.51

Consequently, nonresonant Raman spectroscopy is a powerfultechnique to discriminate h-BN from BNNT. In ourexperiments, we used nonresonant Raman spectroscopy withan excitation wavelength of λ = 514 nm.Figure 3a features the Raman spectra we obtained from as-

received BNNT (black curve) and purified BNNT (red curve).

The spectrum of the as-received BNNT has one pronouncedpeak, located at 1368.5 cm−1, while the purified BNNTfeatured no peak in this spectral region. Since BNNTs do notshow peaks in nonresonant Raman spectroscopy, this stronglysuggests that the 1368.5 cm−1 peak observed in the as-receivedmaterial was entirely due to the presence of h-BN. In contrast,the purified BNNT material featured no Raman peaks,suggesting that our purification procedure removed virtuallyall h-BN materials. Both findings, the presence of the h-BNpeak in the as-received materials as well as the completeabsence of the h-BN peak in the purified material are inexcellent agreement with our XRD results, indicating that ourcleaning procedure leads to a near-perfect removal of h-BNfrom the BNNTs.Furthermore, we did a Raman analysis of a stock h-BN

sample (Sigma-Aldrich) and of the purification residue, asshown in the respective blue and green curves in Figure 3b,featuring pronounced peaks at 1364.6 and 1367.4 cm−1,respectively. Table 2 lists the corresponding peak positions andfull widths at half maximum (FWHM) determined viaLorentzian peak fitting as well as the corresponding R2 values.The observed h-BN peak at 1364.6 cm−1 was in excellentagreement with the literature.51 The peak position and peak

width of the purification residue were in close agreement withthe peak observed in the as-received material. We note that incontrast to the BNNT Raman samples, which were preparedby pressing the material into solid disks, the residue samplesonly contained a small amount of the material, as they wereprepared by drop casting. Yet, the observed Raman signal fromthe residue was equal to the strength of the as-received BNNT.This confirms that our purification protocol indeed removedthe h-BN impurities from BNNTs leaving behind concentratedh-BN impurities in the residue. Interestingly, the peak positionof the residue (1367.4 cm−1) was slightly blue-shifted withrespect to the peak of perfect h-BN (1364.6 cm−1). Nemanichet al. have established quantitatively that the Raman peak shiftand peak broadening depend on the h-BN particle size.52

Based on their findings, the blue-shift we observed for theremoved h-BN corresponds to a particle size on the order of 10nm.52 Importantly, the latter directly demonstrates that ourpurification method is capable of removing the mostchallenging impurities: nanoscale h-BN.As shown in Figure 4, we further carried out FTIR of as-

received (black curve) and purified (red curve) BNNTs. For

comparison, we also took a spectrum from stock h-BN (bluecurve). Before purification (black), the BNNT sample featuredtwo asymmetric, broad bands in the ranges 500−800 and 800−1600 cm−1 of similar height and relatively complex shape,featuring approximate widths of 200 and 500 cm−1 (FWHM),respectively. Peaks in these two bands have been attributedout-of-plane buckling and in-plane optical phonon modes,respectively.32 After purification (red), the spectrum lookeddramatically simpler and cleaner: aside from some backgroundclose to their bases; both peaks were more symmetric andlooked more like single peaks, featuring much less structure.High-quality fits (R2 = 0.98) were obtained using pureLorentzian distributions, indicating high sample homogeneity.After purification, the peaks were substantially narrower, 50.0

Figure 3. Raman spectra. (a) As-received (black) and purified (red)BNNT featuring a peak in the 1366 cm−1 region. (b) The residue(green) overlaid by h-BN (blue).

Table 2. Raman FWHM and Peak Position

sample peak position (cm−1) FWHM (cm−1) R2

h-BN 1364.6 ± 0.1 12.5 0.99residue material 1367.4 ± 0.2 26 0.97as-received BNNT 1368.5 ± 0.2 36 0.98

Figure 4. FTIR spectra of purified (red), as-received (black) BNNTsand h-BN (blue). The spectrum features two main peaks, out-of-planebuckling modes near 800 cm−1 and in-plane vibrational modes near1350 cm−1.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.9b01713Chem. Mater. 2019, 31, 8351−8357

8354

and 119.1 cm−1 (FWHM), respectively, with peak positions at783.6 and 1327.3 cm−1.We find it particularly interesting to compare the FTIR

spectra of as-received and purified BNNTs to the h-BNspectrum (blue). The 500−800 cm−1 peak of the as-receivedspectrum (black) is extremely similar to the h-BN spectrum inthis range: not only the peak position, peak width, and generalpeak shape are very similar, even some of the minute subpeaksare closely aligned. This strongly suggests that the as-receivedsample contained h-BN, which is fully in line with our previousfindings. In the purified BNNT spectrum (red), this broadpeak with its characteristics is virtually completely absent. Inthe 800−1600 cm−1 band, the situation is similar. The peakshape of the as-received material (black) is similar to the h-BNspectrum (blue) in the 1300−1600 cm−1 range butsignificantly broader and stronger in the 800−1300 cm−1

range with a lower peak position, 1244 vs 1300 cm−1. Thissuggests that the as-received BNNT material contains asubstantial amount of impurities other than h-BN. In thespectrum of the purified material (red), in contrast, thiscomplexity on both sides of the 1327 cm−1 main peak isabsent. This indicates a much purer material from h-BN, andthe other impurities have been successfully removed. Inparticular, our FTIR results are fully in line with our XRD andRaman data and further corroborate our claims that ourpurification method removes h-BN.We compared our results to the recent findings of Harrison

et al., who used FTIR to determine the h-BN concentration inBNNT samples. They analyzed samples of known h-BNconcentration and found that the height ratios of theabsorbance peaks at 800 and 1350 cm−1 were proportionalto the h-BN concentration.32 Harrison et al. also found that theposition of the 800 cm−1 peak shifted to higher wave numberswith increasing purity. Qualitatively, our results and claims arein full agreement with their findings: the peak ratio decreasedfrom 1.01 for the as-received material to 0.48 for the purifiedmaterial, indicating significant purification. We also observed apeak shift from 765.7 cm−1 for the as-received material to783.6 cm−1 in the purified material. However, we found thatthe FTIR absorbance spectra of our as-received BNNT and h-BN materials looked substantially different from theirs in termsof peak positions, relative peak heights, and peak shapes.Therefore, we believe that their method to quantify h-BNconcentrations is not directly applicable to our materials. Thesubstantial difference in the materials was further corroboratedby comparing our XRD spectra to theirs.The fact that the h-BN impurities were removed using very

mild conditions of heptane in a pressure tube at 90 °C whilethe BN tubes remain undamaged may first seem surprising.Previous reports of purification procedures involved muchharsher treatments that damaged the BNNTs, such as strongacids and superacids,47 strong sonication and mechanicalforces,31,35 or oxidation at 750 °C13,23 and higher temper-atures.23,31,34,35

To explain the purification mechanism (Figure 5a), wepropose two hypotheses. One hypothesis focuses onweakening the adhesion of the h-BN attached to the BNNTsurface (Figure 5b). Since h-BN is planar and the BNNTs arecurved, we assume that there is a cleavage space at the h-BN/BNNT interface, into which heptane molecules can diffuse. Forcarbon nanotubes, literature reports have shown that smallhydrocarbon molecules can move in such confined spacesparticularly effectively and rapidly;53 this enhanced hydro-

carbon mobility has been explained based on surfacediffusion.53 Importantly, our purification temperature of 90°C is very close to the boiling point of heptane, 98 °C. MonteCarlo simulations of the vapor pressure of several hydro-carbons from methane to octane in confined spaces show anincreasing pressure as confinement dimensions decrease.54

Correspondingly, the heptane molecules trapped in the h-BN/BNNT cleavage spaces may exert significant pressure on theconfining walls, thus driving h-BN sheets and BNNTs apartwith forces large enough to overcome van der Waals adhesion.Since the contact areas of the sheet-vs.-tube geometry arerelatively small, not much force may be needed. Moreover, thepresence of the molecules in the space significantly reduces vander Waals forces.55

A second hypothesis involves diffusion of heptane into theinterior of the BNNTs (Figure 5c), which has beenexperimentally observed for small solvent molecules.45 Andas described, nanoconfinement of heptane inside the tubewould result in an increase in its vaporization pressure. Thispressure may cause small changes in the tube diameters,thereby releasing attached surface impurities. Here, h-BN,which is bonded to the tube surface due to the chemicalsimilarity, is removed by the breathing of the BNNT tubes dueto the entrance and vaporization of the heptane. Ourpreliminary estimates based on the in-plane stiffness of h-BN, however, suggested that this radial breathing is relativelysmall. Nevertheless, calculations by others have shown thatCNTs with kinks, bends, and local deformations can deformmuch more than the extremely high axial strength of the tubeswould suggest.56,57 Assuming that BNNTs share theseproperties and that confinement of heptane in the inside ofthe tube significantly increases the vapor pressure, the removalof BN impurities via this breathing mechanism is plausible. Itmay even be a combination of the exterior and interior heptanelocation mechanisms causing the removal of h-BN impurities.Although molecular removal mechanisms are hypotheses, ourexperimental facts demonstrate that heptane at 90 °C in apressure tube causes h-BN particles to be released andsuspended in the solvent while the tubes remain undamaged.Longer-chain hydrocarbons showed some effect but not aspronounced as with heptane, and higher temperatures wereneeded. Pentane worked but was not used because of its higher

Figure 5. Heptane purification process (a) and two proposedmechanisms (b, c). The first proposed mechanism (b) assumes thatheptane enters the confining space between the tube surfaces and theh-BN sheets, thus lowering van der Waals forces and loosening h-BNattachment. (c) The second hypothesis assumes that heptane entersthe inside of the tube, where it causes radial tube swelling at increasedpressures, leading to disengagement of the h-BN from the surface.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.9b01713Chem. Mater. 2019, 31, 8351−8357

8355

volatility. Isopropanol, on the other hand, did not work. At thispoint, it is not obvious which property of heptane makes it soeffective in removing h-BN from BNNT. More investigationwill be needed, including molecular dynamics to determineatomic-scale effects, to fully understand the mechanism andfurther improve the purification process.

■ CONCLUSIONSIn summary, we introduced a mild, high-yield method to purifyBNNTs produced by the important HTP method. Althoughno prior study had demonstrated effective removal of h-BN,our approach removed ≥99.8% of the h-BN and a significantamount of other BN impurities. Our claims are supported byFESEM images, TEM images, as well as XRD, Raman, andFTIR spectra. The latter three spectroscopic techniquescharacterize a macroscopic region of the sample and are,thus, more representative than imaging selected areas usingelectron microscopy. XRD was particularly useful to quantifythe h-BN content, since h-BN featured a distinctly sharp peak.The FESEM and TEM images of heptane-purified BNNTsshowed tubes with a virtually clean surface and no damage tothe tube network. Our nonaggressive, nondestructive purifica-tion method using heptane at the moderate temperature T =90 °C is a significant advancement over all previously reportedmethods relying on aggressive treatments, including sonication,mechanical methods, strong acids, or oxidation at hightemperatures >700 °C. The later techniques introducesubstantial damage to the tubes and result in a low yield.13,23

It is expected13,23,31 that the BNNTs purified by our scalable,mild method will exhibit better structural and thermalproperties and, thus, have the potential to make nano-composites with significantly improved properties with awide range of potential applications.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.chemma-ter.9b01713.

FESEM and TEM of BNNTs before and afterpurification, quantification of the sensitivity and h-BNdetection limit of XRD, and purified BNNT FTIRabsorbance spectrum with Lorentzian fit (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: schniepp@wm.edu.ORCIDRobert D. Pike: 0000-0002-8712-0288Hannes C. Schniepp: 0000-0003-4645-9469Author ContributionsThe manuscript was written through the contributions of all ofthe authors. All of the authors have given approval to the finalversion of the manuscript.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSH.C.S. and D.E.K. acknowledge support from the Center ofInnovative Technology under CRCF award MF17-032-AMand from the Office of Naval Research under award number

N00014-18-1-2264. The authors acknowledge BNNT LLC forthe BNNT materials. This material is based upon worksupported by the National Science Foundation under GrantNos. DMR-1352542 and DMREF-1534428.

■ ABBREVIATIONSh-BN, hexagonal boron nitride; BNNT, boron nitride nano-tube; HTP, high-temperature pressure method for BNNTsynthesis; XRD, X-ray diffraction; FWHM, full width halfmaximum

■ REFERENCES(1) Chopra, N. G.; Luyken, R. J.; Cherrey, K.; Crespi, V. H.; Cohen,M. L.; Louie, S. G.; Zettl, A. Boron Nitride Nanotubes. Science 1995,269, 966−967.(2) Lee, C.; Bhandari, S.; Tiwari, B.; Yapici, N.; Zhang, D.; Yap, Y.Boron Nitride Nanotubes: Recent Advances in Their Synthesis,Functionalization, and Applications. Molecules 2016, 21, 922.(3) Gong, W.; Geng, H.; Qu, S.; Lu, W. Molecular Dynamics Studyon Radial Deformation of Armchair Single-Walled Boron NitrideNanotubes. Appl. Phys. Express 2017, 10, No. 105001.(4) Golberg, D.; Bando, Y.; Huang, Y.; Terao, T.; Mitome, M.;Tang, C.; Zhi, C. Boron Nitride Nanotubes and Nanosheets. ACSNano 2010, 4, 2979−2993.(5) Ghassemi, H. M.; Lee, C. H.; Yap, Y. K.; Yassar, R. S. Real-TimeFracture Detection of Individual Boron Nitride Nanotubes in SevereCyclic Deformation Processes. J. Appl. Phys. 2010, 108, No. 024314.(6) Terrones, M.; Romo-Herrera, J. M.; Cruz-Silva, E.; Lopez-Urías,F.; Munoz-Sandoval, E.; Velazquez-Salazar, J. J.; Terrones, H.; Bando,Y.; Golberg, D. Pure and Doped Boron Nitride Nanotubes. Mater.Today 2007, 10, 30−38.(7) Vaccarini, L.; Goze, C.; Henrard, L.; Hernandez, E.; Bernier, P.;Rubio, A. Mechanical and Electronic Properties of Carbon andBoron−Nitride Nanotubes. Carbon 2000, 38, 1681−1690.(8) Chen, Y.; Zou, J.; Campbell, S. J.; Caer, G. L. Boron NitrideNanotubes: Pronounced Resistance to Oxidation. Appl. Phys. Lett.2004, 84, 2430−2432.(9) Verma, V.; Jindal, V. K.; Dharamvir, K. Elastic moduli of a boronnitride nanotube. Nanotechnology 2007, 18, No. 435711.(10) Chang, C. W.; Fennimore, A. M.; Afanasiev, A.; Okawa, D.;Ikuno, T.; Garcia, H.; Li, D.; Majumdar, A.; Zettl, A. Isotope Effect onthe Thermal Conductivity of Boron Nitride Nanotubes. Phys. Rev.Lett. 2006, 97, No. 085901.(11) Xiao, Y.; Yan, X. H.; Cao, J. X.; Ding, J. W.; Mao, Y. L.; Xiang, J.Specific Heat and Quantized Thermal Conductance of Single-WalledBoron Nitride Nanotubes. Phys. Rev. B 2004, 69, No. 205415.(12) Simpson, J. A. Introduction to the Galactic Cosmic Radiation.In Composition and Origin of Cosmic Rays; Springer, 1983; pp 1−24.(13) Marincel, D. M.; Adnan, M.; Ma, J.; Bengio, E. A.; Trafford, M.A.; Kleinerman, O.; Kosynkin, D. V.; Chu, S. H.; Park, C.; Hocker, S.J.; Fay, C. C.; et al. Scalable Purification of Boron Nitride Nanotubesvia Wet Thermal Etching. Chem. Mater. 2019, 31, 1520−1527.(14) Cohen, M. L.; Zettl, A. The Physics of Boron NitrideNanotubes. Phys. Today 2010, 63, 34−38.(15) Rubio, A.; Corkill, J. L.; Cohen, M. L. Theory of GraphiticBoron Nitride Nanotubes. Phys. Rev. B 1994, 49, 5081−5084.(16) Blase, X.; Rubio, A.; Louie, S. G.; Cohen, M. L. Stability andBand Gap Constancy of Boron Nitride Nanotubes. Europhys. Lett.1994, 28, 335−340.(17) Lei, H.; Shi, Q.; Chen, J. Research on BNNTs/Epoxy/SiliconeTernary Composite Systems for High Thermal Conductivity. IOPConf. Ser.: Mater. Sci. Eng. 2018, 381, No. 012076.(18) Wu, X.; Yang, J.; Hou, J. G.; Zhu, Q. Deformation-Induced SiteSelectivity for Hydrogen Adsorption on Boron Nitride Nanotubes.Phys. Rev. B 2004, 69, No. 153411.(19) Tang, C.; Bando, Y.; Ding, X.; Qi, S.; Golberg, D. CatalyzedCollapse and Enhanced Hydrogen Storage of BN Nanotubes. J. Am.Chem. Soc. 2002, 124, 14550−14551.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.9b01713Chem. Mater. 2019, 31, 8351−8357

8356

(20) Wang, J.; Lee, C. H.; Yap, Y. K. Recent Advancements in BoronNitride Nanotubes. Nanoscale 2010, 2, 2028.(21) Hilder, T. A.; Gordon, D.; Chung, S.-H. Salt Rejection andWater Transport Through Boron Nitride Nanotubes. Small 2009, 5,2183−2190.(22) Watanabe, K.; Taniguchi, T.; Kanda, H. Direct-BandgapProperties and Evidence for Ultraviolet Lasing of Hexagonal BoronNitride Single Crystal. Nat. Mater. 2004, 3, 404−409.(23) Chen, H.; Chen, Y.; Yu, J.; Williams, J. S. Purification of BoronNitride Nanotubes. Chem. Phys. Lett. 2006, 425, 315−319.(24) Tiano, A. L.; Park, C.; Lee, J. W.; Luong, H. H.; Gibbons, L. J.;Chu, S. H.; Applin, S.; Gnoffo, P.; Lowther, S.; Kim, H. J.; Danehy, P.M.; Inman, J. A.; Jones, S. B.; Kang, J. H.; Sauti, G.; Thibeault, S. A.;Yamakov, V.; Wise, K. E.; Su, J.; Fay, C. C. Boron Nitride Nanotube:Synthesis and Applications. Proc. SPIE 2014, 9060, 906006.(25) Kim, K. S.; Kim, M. J.; Park, C.; Fay, C. C.; Chu, S. H.;Kingston, C. T.; Simard, B. Scalable Manufacturing of Boron NitrideNanotubes and Their Assemblies: A Review. Semicond. Sci. Technol.2016, 32, No. 013003.(26) Smith, M. W.; Jordan, K. C.; Park, C.; Kim, J. W.; Lillehei, P.T.; Crooks, R.; Harrison, J. S. Very Long Single- and Few-WalledBoron Nitride Nanotubes via the Pressurized Vapor/CondenserMethod. Nanotechnology 2009, 20, No. 505604.(27) Kim, K. S.; Kingston, C. T.; Hrdina, A.; Jakubinek, M. B.; Guan,J.; Plunkett, M.; Simard, B. Hydrogen-Catalyzed, Pilot-ScaleProduction of Small-Diameter Boron Nitride Nanotubes and TheirMacroscopic Assemblies. ACS Nano 2014, 8, 6211−6220.(28) Fathalizadeh, A.; Pham, T.; Mickelson, W.; Zettl, A. ScaledSynthesis of Boron Nitride Nanotubes, Nanoribbons, and Nano-cocoons Using Direct Feedstock Injection into an Extended-Pressure,Inductively-Coupled Thermal Plasma. Nano Lett. 2014, 14, 4881−4886.(29) Kim, J.; Lee, S.; Uhm, Y. R.; Jun, J.; Rhee, C. K.; Kim, G. M.Synthesis and Growth of Boron Nitride Nanotubes by a Ball Milling−Annealing Process. Acta Mater. 2011, 59, 2807−2813.(30) Chen, Y.; Gerald, J. F.; Williams, J. S.; Bulcock, S. Synthesis ofBoron Nitride Nanotubes at Low Temperatures using Reactive BallMilling. Chem. Phys. Lett. 1999, 299, 260−264.(31) Maguer, A.; Arenal, R.; Jaffrennou, P.; Cochon, J. L.; Bresson,L.; Doris, E.; Mioskowski, C.; Loiseau, A. Purification of Single-Walled Boron Nitride Nanotubes and Boron Nitride Cages. J.Nanosci. Nanotechnol. 2007, 7, 3528−3532.(32) Harrison, H.; Lamb, J. T.; Nowlin, K. S.; Guenthner, A. J.;Ghiassi, K. B.; Kelkar, A. D.; Alston, J. R. Quantification of HexagonalBoron Nitride Impurities in Boron Nitride Nanotubes via FTIRSpectroscopy. Nanoscale Adv. 2019, 1, 1693−1701.(33) Martinez Rubi, Y.; Jakubek, Z. J.; Chen, M.; Zou, S.; Simard, B.Quality Assessment of Bulk Boron Nitride Nanotubes for AdvancingResearch, Commercial, and Industrial Applications. ACS Appl. NanoMater. 2019, 2, 2054−2063.(34) Zhi, C.; Bando, Y.; Tang, C.; Honda, S.; Sato, K.; Kuwahara,H.; Golberg, D. Purification of Boron Nitride Nanotubes throughPolymer Wrapping. J. Phys. Chem. B 2006, 110, 1525−1528.(35) Vieira, S. M. C.; Carroll, D. L. Purification of Boron NitrideMultiwalled Nanotubes. J. Nanosci. Nanotechnol. 2007, 7, 3318−3322.(36) Koi, N.; Oku, T.; Inoue, M.; Suganuma, K. Structures andPurification of Boron Nitride Nanotubes Synthesized from Boron-Based Powders with Iron Particles. J. Mater. Sci. 2008, 43, 2955−2961.(37) Choi, J.-H.; Kim, J.; Seo, D.; Seo, Y.-S. Purification of BoronNitride Nanotubes via Polymer Wrapping. Mater. Res. Bull. 2013, 48,1197−1203.(38) Loiseau, A.; Willaime, F.; Demoncy, N.; Hug, G.; Pascard, H.Boron Nitride Nanotubes with Reduced Numbers of LayersSynthesized by Arc Discharge. Phys. Rev. Lett. 1996, 76, 4737−4740.(39) Arenal, R.; Stephan, O.; Cochon, J.-L.; Loiseau, A. Root-Growth Mechanism for Single-Walled Boron Nitride Nanotubes inLaser Vaporization Technique. J. Am. Chem. Soc. 2007, 129, 16183−16189.

(40) Yeh, Y.-W.; Raitses, Y.; Koel, B. E.; Yao, N. Stable Synthesis ofFew-Layered Boron Nitride Nanotubes by Anodic Arc Discharge. Sci.Rep. 2017, 7, No. 3075.(41) Saito, Y.; Maida, M.; Matsumoto, T. Structures of BoronNitride Nanotubes with Single-Layer and Multilayers Produced byArc Discharge. Jpn. J. Appl. Phys. 1999, 38, 159−163.(42) Yu, D. P.; Sun, X. S.; Lee, C. S.; Bello, I.; Lee, S. T.; Gu, H. D.;Leung, K. M.; Zhou, G. W.; Dong, Z. F.; Zhang, Z. Synthesis of BoronNitride Nanotubes by Means of Excimer Laser Ablation at HighTemperature. Appl. Phys. Lett. 1998, 72, 1966−1968.(43) Boulanger, L.; Andriot, B.; Cauchetier, M.; Willaime, F.Concentric Shelled and Plate-Like Graphitic Boron Nitride Nano-particles Produced by CO2 Laser Pyrolysis. Chem. Phys. Lett. 1995,234, 227−232.(44) Xu, X. G.; Gilburd, L.; Bando, Y.; Golberg, D.; Walker, G. C.Defects and Deformation of Boron Nitride Nanotubes Studied byJoint Nanoscale Mechanical and Infrared Near-Field Microscopy. J.Phys. Chem. C 2016, 120, 1945−1951.(45) Kleinerman, O.; Adnan, M.; Marincel, D. M.; Ma, A. W.;Bengio, E. A.; Park, C.; Chu, S. H.; Pasquali, M.; Talmon, Y.Dissolution and Characterization of Boron Nitride Nanotubes inSuperacid. Langmuir 2017, 33, 14340−14346.(46) Terao, T.; Zhi, C.; Bando, Y.; Mitome, M.; Tang, C.; Golberg,D. Alignment of Boron Nitride Nanotubes in Polymeric CompositeFilms for Thermal Conductivity Improvement. J. Phys. Chem. C 2010,114, 4340−4344.(47) Hu, H.; Zhao, B.; Itkis, M. E.; Haddon, R. C. Nitric AcidPurification of Single-Walled Carbon Nanotubes. J. Phys. Chem. B2003, 107, 13838−13842.(48) Bartnitskaya, T. S.; Oleinik, G. S.; Pokropivnyi, A. V.;Pokropivnyi, V. V. Synthesis, Structure, and Formation Mechanismof Boron Nitride Nanotubes. J. Exp. Theor. Phys. Lett. 1999, 69, 163−168.(49) Altoe, M. V. P.; Sprunck, J. P.; Gabriel, J.-C. P.; Bradley, K.Nanococoon Seeds for BN Nanotube Growth. J. Mater. Sci. 2003, 38,4805−4810.(50) Yu, B.; Xing, W.; Guo, W.; Qiu, S.; Wang, X.; Lo, S.; Hu, Y.Thermal Exfoliation of Hexagonal Boron Nitride for EffectiveEnhancements on Thermal Stability, Flame Retardancy and SmokeSuppression of Epoxy Resin Nanocomposites via Sol−Gel Process. J.Mater. Chem. A 2016, 4, 7330−7340.(51) Arenal, R.; Ferrari, A. C.; Reich, S.; Wirtz, L.; Mevellec, J. Y.;Lefrant, S.; Rubio, A.; Loiseau, A. Raman Spectroscopy of Single-WallBoron Nitride Nanotubes. Nano Lett. 2006, 6, 1812−1816.(52) Nemanich, R. J.; Solin, S. A.; Martin, R. M. Light ScatteringStudy of Boron Nitride Microcrystals. Phys. Rev. B 1981, 23, 6348−6356.(53) Holt, J. K. Fast Mass Transport Through Sub-2-NanometerCarbon Nanotubes. Science 2006, 312, 1034−1037.(54) Singh, S. K.; Sinha, A.; Deo, G.; Singh, J. K. Vapor-Liquid PhaseCoexistence, Critical Properties, and Surface Tension of ConfinedAlkanes. J. Phys. Chem. C 2009, 113, 7170−7180.(55) Parsegian, V. A. Van Der Waals Forces: A Handbook forBiologists, Chemists, Engineers, and Physicists; Cambridge UniversityPress: New York, 2005.(56) Chopra, N. G.; Benedict, L. X.; Crespi, V. H.; Cohen, M. L.;Louie, S. G.; Zettl, A. Fully Collapsed Carbon Nanotubes. Nature1995, 377, 135−138.(57) Zhu, C.; Guo, W.; Yu, T. X.; Woo, C. H. Radial Compressionof Carbon Nanotubes: Deformation and Damage, Super-Elasticityand Super-Hardness. Nanotechnology 2005, 16, 1035−1039.

Chemistry of Materials Article

DOI: 10.1021/acs.chemmater.9b01713Chem. Mater. 2019, 31, 8351−8357

8357