Fundamental Discovery of New Phases and Direct Conversion of … · 2015. 11. 9. · formation. It has been conjectured that liquid carbon may exist as a thermodynamically stable

Fundamental Discovery of New Phases and DirectConversion of Carbon into Diamond and hBN intocBN and Properties

JAGDISH NARAYAN and ANAGH BHAUMIK

We review the discovery of new phases of carbon (Q-carbon) and BN (Q-BN) and address criticalissues related to direct conversion of carbon into diamond and hBN into cBN at ambienttemperatures and pressures in air without any need for catalyst and the presence of hydrogen. TheQ-carbon and Q-BN are formed as a result of quenching from super undercooled state by usinghigh-power nanosecond laser pulses.We discuss the equilibrium phase diagram (P vs T) of carbon,and show that by rapid quenching, kinetics can shift thermodynamic graphite/diamond/liquidcarbon triple point from 5000 K/12 GPa to super undercooled carbon at atmospheric pressure inair. Similarly, the hBN-cBN-Liquid triple point is shifted from 3500 K/9.5 GPa to as low as2800 Kand atmospheric pressure. It is shown that nanosecond laser heating of amorphous carbonand nanocrystalline BN on sapphire, glass, and polymer substrates can be confined to melt in asuper undercooled state. By quenching this super undercooled state, we have created a new state ofcarbon (Q-carbon) and BN (Q-BN) from which nanocrystals, microcrystals, nanoneedles,microneedles, and thin films are formed depending upon the nucleation and growth times allowedand the presence of growth template. The large-area epitaxial diamond and cBN films are formed,when appropriate planarmatching or lattice matching template is provided for growth from superundercooled liquid. The Q-phases have unique atomic structure and bonding characteristics asdetermined by high-resolution SEM and backscatter diffraction, HRTEM, STEM-Z, EELS, andRaman spectroscopy, and exhibit new and improvedmechanical hardness, electrical conductivity,and chemical and physical properties, including room-temperature ferromagnetism and enhancedfield emission. The Q-carbon exhibits robust bulk ferromagnetism with estimated Curietemperature of about 500 K and saturation magnetization value of 20 emu g�1. We have alsodeposited diamond on cBN by using a novel pulsed laser evaporation of carbon and obtainedcBN/diamond composites, where cBN acts as template for diamond growth. Both diamond andcBN grown from super undercooled liquid can be alloyed with both p- and n-type dopants. Thisprocess allows carbon to diamond and hBN to cBN conversions and formation of usefulnanostructures and microstructures at ambient temperatures in air at atmospheric pressure onpractical and heat-sensitive substrates in a controlled way without need for any catalysts andhydrogen to stabilize sp3 bonding for diamond and cBN phases.

DOI: 10.1007/s11661-015-3312-7� The Minerals, Metals & Materials Society and ASM International 2016

JAGDISH NARAYAN, Distinguished Professor, and ANAGHBHAUMIK, Ph.D. Student, are with the Department of MaterialsScience and Engineering, Centennial Campus, North Carolina StateUniversity, Raleigh, NC 27695-7907. Contact e-mails: [email protected]; [email protected]

Manuscript submitted November 9, 2015.Article published online January 13, 2016

Jagdish Narayan is the John C. C. Fan Family DistinguishedUniversity Professor in the Department of Materials Science andEngineering at North Carolina State University. After graduating witha B. Tech. (Distinction and First Rank) from IIT, Kanpur, in 1969,Narayan established an extraordinary academic record at the Univer-sity of California, Berkeley, by finishing M.S. (1970) and Ph.D. (1971)

JAGDISH NARAYAN

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 47A, APRIL 2016—1481

RETRACTEDART

ICLE

http://crossmark.crossref.org/dialog/?doi=10.1007/s11661-015-3312-7&domain=pdf

http://crossmark.crossref.org/dialog/?doi=10.1007/s11661-015-3312-7&domain=pdf

I. INTRODUCTION

This review is divided into two parts: In the first part,we cover the formation of Q-carbon, its properties andconversion into diamond. In the second part, we focuson the formation Q-BN, its properties and conversioninto cBN. Both carbon into diamond and hBN into cBNconversions are in the form of nanocrystals, microcrys-tals, nanoneedles, microneedles, and large-area sin-gle-crystal films on a variety of practical substrates,including heat-sensitive polymers. The discovery ofQ-carbon and Q-BN and formation of diamond andcBN and their composites represent a major break-through in the field of materials science and engineering.

Direct conversion of carbon into diamond at ambientpressures and lower temperatures is scientifically chal-lenging with immense technological significance.[1–3]

Conversion of carbon, one of the most abundantmaterials in the Earth’s crust, into most preciousmaterial diamond has been a cherished goal of thescientists all over the world for the longest time.Diamond is one of the most desirable materials withmany applications ranging from abrasives, protectivecoatings, and biomedical applications to superior dia-mond electronics, photonics, and display devices. Con-ventional bulk processing involves high pressures andtemperatures,[1] and chemical vapor deposition for thinfilms requires high temperatures in the presence ofhydrogen.[4] These requirements lead to low productionvolumes and high costs. More recently, anotherapproach for the formation of nanodiamond from SiChas been reported at temperatures ~(1273 K) 1000 �Cunder flowing hydrogen and chlorine gases at ambientpressures.[5] Here we show that a direct conversion ofcarbon into diamond can occur in air at ambienttemperatures and pressures without any need for cata-lysts and hydrogen to stabilize sp3 diamond bonding. Inaddition, ferromagnetism in bulk carbon representsanother scientific challenge with profound impact onmagnetic storage, sensors, data processing, and biomed-ical applications. These two challenging problems aresolved by the discovery of new state of carbon, referred

to as Q-carbon hereafter. In addition, the Q-carbonexhibits enhanced electrical conductivity (of semicon-ducting and metallic nature), enhanced field emission,and super high hardness. The Q-carbon is formed asresult of quenching of super undercooled liquid carbonat the atmospheric pressure, which is found to bemagnetic and it plays a critical role in pure diamondformation. It has been conjectured that liquid carbonmay exist as a thermodynamically stable phase at highpressures and temperatures near the cores of Uranusand Neptune planets and contribute to their mag-netism.[1] Thus, our finding can also explain ferromag-netism in our planetary system and formation of naturaldiamonds among other things. According to the equi-librium (P vs T) phase diagram (Figure 1),[1] graphite,diamond, liquid, and vapor are thermodynamicallystable forms of carbon. At low pressures, graphiteconverts into vapor above around 4000 K. According tothe phase diagram, diamond synthesis from liquidcarbon will require even higher temperatures andpressures as the graphite/diamond/liquid carbon triplepoint occurs at 5000 K/12 GPa, where 1 GPa = 9869Atm. Currently, diamond powders are synthesized bygraphite to diamond conversion at high pressures andtemperatures. Graphite can be transformed into dia-mond above about 2000 K at 6 to 10 GPa using a liquidmetal (iron) catalyst which is used for commercialsynthesis of diamond (Figure 1).[1]

In the thermodynamically stable forms of carbon,graphite, diamond, liquid, and vapor, we introduceamorphous carbon with some sp3 content and superundercooled state of liquid carbon (Figure 1).[1] Thiscan be accomplished by nanosecond laser melting ofamorphous carbon, where undercooled state is at about4000 K, some 1000 K below the melting point ofgraphite.[6] The dotted extension to 4000 K in the phasediagram from the liquid-diamond-graphite triple pointat 5000 K represents this super undercooled state, whichupon quenching results in the formation of Q-carbon at

Fig. 1—Carbon phase hdiagram (P vs T) following Bundy et al.[1]

which has amorphous diamondlike carbon melting at 4000 K atambient pressures (dotted green line).

degrees in a record time of two years. His pioneering research ondefects and diffusion and novel materials laid foundations formaterials processing needed for systems ranging from nano to microand macroscale. Narayan’s work has been cited over 21,500 times withan h-index of over 72, and he has published nine books and more than500 papers in scholarly journals. His honors include the 2014 NorthCarolina Science (highest civilian honor of the state of North Carolina)Award, the O. Max Gardner Award (highest UNC System honor),Holladay Medal and R.J. Reynolds Prize (North Carolina State’shighest honors for excellence in research, teaching, and extension),Acta Materialia Gold Medal and Prize given for pioneering contribu-tions and leadership in materials science worldwide, ASM GoldMedal, TMS RF Mehl Gold Medal, U.S. Department of EnergyOutstanding Research Award, and three IR-100 awards in addition toFellow Honors from two academies and seven professional societies.Professor Narayan’s research has been duly recognized by theAmerican Institute of Physics (AIP) in this year’s Nobel Prize inPhysics on Blue Light Emitting Diodes (LEDs) made from GalliumNitrides (III-nitrides) based materials. The AIP has singled outNarayan’s highly cited paper (J. Appl. Phys. 87, 965 (2000) with over1,140 citations) on the development of GaN-based materials used inthe Nobel Laureates’ work.

1482—VOLUME 47A, APRIL 2016 METALLURGICAL AND MATERIALS TRANSACTIONS A

RETRACTEDART

ICLE

a temperature slightly lower than 4000. Since the moltenstate of carbon is found to be metallic,[1] the carbonatoms can be fully packed with shorter C-C bond lengthand mass density and hardness exceeding those ofdiamond. By quenching this super undercooled state, weare able to form Q-carbon, from which nanodiamonds,microdiamonds and thin films are formed dependingupon the growth template and times allowed fornucleation and growth. The new state of carbon(Q-carbon) has a very high fraction of sp3-bondedcarbon and the rest sp2, and is expected to possess novelphysical, chemical, mechanical and catalytic properties.There is a considerable reduction in volume, when theas-deposited amorphous carbon is melted in the under-cooled state and quenched as Q-carbon. Most interest-ing of all is that Q-carbon exhibits ferromagnetism atand above room temperature. The formation of cubicdiamond phase can occur if a sufficient time is allowedfor homogeneous nucleation from the Q-carbon. Thesubstrates which are planar and lattice matched withcubic diamond such as sapphire and copper aid theepitaxial nucleation of diamond.[2,3]

The primary focus of the first part of this paper is onnanosecond laser melting of amorphous carbon films onsapphire, glass, and polymer substrates. The irradiationof these films with ArF Excimer laser pulses (wavelength193 nm or photon energy of 6 eV and pulse duration of20 ns) leads to confinement of laser energy and selectivemelting of amorphous carbon films. These undercoolingvalues for amorphous carbon are considerably higherthan those achieved during melting of crystalline carbonsuch as HOPG samples, which did not lead to diamondformation. Our results show that under nanosecondpulsed laser melting, amorphous carbon can lead to ahighly undercooled state which can be quenched into anew state of Q-carbon from which nanodiamonds,microdiamonds, and large-area single-crystal diamondthin films are formed. We are also able to create nano-and microneedles which stand to make a profoundimpact on biomedical applications. The Q-carbon alsoexhibits novel properties, including room-temperatureferromagnetism (RTFM), and enhanced hardness andfield emission. It should be emphasized that the presenceof similar sp3 fraction in the as-deposited amorphouscarbon films does not result in RTFM. The nanodia-monds nucleate from the undercooled state, whichprovide a seed for microdiamond and other relatedstructures.

II. EXPERIMENTAL METHODS

The amorphous carbon films were deposited onsapphire (c-plane) and glass substrates by using KrFlaser (Pulse duration = 25 ns, wavelength = 248 nm,energy density = 3.0 J cm�2) to a thickness of 50 to500 nm. These films were characterized by TEM andRaman and found to be amorphous containing Ramansignature on sapphire (DLC broad peak = 1580 cm�1)with estimated sp3 fraction varying from 20 to 50 pct.The Raman spectra for films on glass substrates,contained D (1349 cm�1) and G (1580 cm�1) peaks

with considerably less sp3 around 20 to 25 pct. The filmson sapphire contained a single broad peak centered on1580 cm�1 with sp3 fraction over 40 pct. These filmswere irradiated in air with ArF laser pulses (Pulseduration = 20 ns, wavelength = 193 nm, energy den-sity = 0.3 to 0.6 J cm�2). These films were character-ized by high-resolution scanning electron microscopy,electron backscatter diffraction (EBSD) with character-istic diamond Kikuchi patterns, transmission electronmicroscopy, X-ray diffraction, and Raman spec-troscopy. High-resolution SEM and EBSD measure-ments were carried out using FEI Verios 460L SEM andFEI Quanta 3D FEG FIB-SEM for phase identificationand determination of grain orientation. Aberration-cor-rected STEM-FEI Titan 80 to 300 and JEOL-2010STEM/TEM were used electron energy loss spec-troscopy (EELS) with resolution of 0.15 eV andhigh-resolution TEM (point-to- point TEM resolution0.18 nm; STEM-Z resolution 0.08 nm with informationlimit of 0.06 nm). Magnetic measurements were per-formed in magnetic fields up to 1 T in an Ever CoolQuantum Design PPMS system with a base temperatureas low as 10 K. Asylum Research MFP-3D InfinityAFM was employed for MFM and EFM imaging. ForMFM scans, silicon probe with 50 nm Co-Cr-coatedtips were used, and Pt-Ti-coated cantilevers were usedfor EFM measurements. The MFM image was takenwith a delta height of 50 nm and was flattened to zerothorder to improve its contrast.

III. RESULTS AND DISCUSSION

A. Modified Carbon Phase Diagram

According to the equilibrium (P vs T) phase dia-gram,[1] graphite, diamond, liquid and vapor are ther-modynamically stable forms of carbon. At lowpressures, graphite converts into vapor above around4000 K. According to the phase diagram, diamondsynthesis from liquid carbon will require even highertemperatures and pressures as the graphite/diamond/liquid carbon triple point occurs at 5000 K/12 GPa.Consistent with the phase diagram, diamond can exist inthe interiors of the outer planets (Uranus and Neptune)and Earth’s mantle, where pressure/temperature are600 GPa/7000 K and 135 GPa/3500 K, respectively.Because of the high binding and activation energy oftransformation, carbon polymorphs can exist metasta-bly well into a P–T region, where a different phase isthermodynamically stable. As an example, diamondsurvives indefinitely at room temperature, where gra-phite is the stable form.In the thermodynamically stable forms of carbon,

graphite, diamond, liquid, and vapor, we introduceamorphous carbon with some sp3 content and superundercooled state of liquid carbon. This can be accom-plished by nanosecond laser melting of amorphouscarbon, where undercooled state is at about 4000 K,some 1000 K below the melting point of graphite. Thistemperature is estimated from model calculations byusing Singh and Narayan (SLIM) code.[6] By quenching


RETRACTEDART

ICLE

this super undercooled state, we are able to formnanodiamond and microdiamond by controlling thenucleation and growth times. The super undercooledstate upon rapid quenching leads to formation of newstate of carbon (Q-carbon) with novel physical, chem-ical, mechanical, and catalytic properties.

Our results clearly show that diamond can be formedat ambient pressure in air from super undercooled stateof carbon, produced by nanosecond pulsed laser meltingof amorphous carbon. Thus, amorphous state of car-bon, laser parameters, and film substrate determine thetemperature distribution and undercooling and play acritical role in the nucleation and growth of diamond.[6]

By scaling with the melting point of carbon, we estimatethe undercooling in carbon to be as high as 1000. Thisundercooling shifts amorphous carbon/diamond/liquidcarbon triple point to 4000 K or lower at ambientpressures. This is rather a drastic change fromgraphite/diamond/liquid carbon triple point at 5000 K/12 GPa in the equilibrium phase diagram in Figure 1.This shift of amorphous carbon/diamond/liquid carbontriple point to 4000 K or lower at ambient pressuresleads to modification[3] of equilibrium carbon phasediagram by Bundy et al.[1] to accommodate diamondprocessing under super undercooled state of carbon. Atthese transition temperatures, Gibbs free energy ofhighly undercooled liquid equals that of metastable di-amond phase which is quenched and retained at roomtemperature.

This transformation from super undercooled state ofcarbon into diamond, which can occur at 4000 K andambient pressures, modifies the equilibrium carbonphase diagram, as shown in Figure 1. The extension ofthe phase diagram (P vs T) is based upon Simonequation, where P = P0+ a[(T/Tr)

c � 1], where weused P0 = 10�4 GPa at Tr = 4000 K from our exper-iments and the experimental data of Bundy et al. toestimate a and c parameters. From the Simon equation,dP=dTð ÞTr

¼ ac=Tr was determined and the results are

compared with the value estimated from Clau-sius-Clapeyron equation, where dP=dTð ÞTr

¼ DHm=

TrDVð Þ, where DV is the change in volume from superundercooled carbon into diamond.[3]

Figure 2 shows variation of Gibbs free energy as afunction of temperature for graphitic carbon (Gg), liquidcarbon (Gliq), and diamond (Gd) near our ambienttemperature of processing[7]. According to this freeenergy diagram, amorphous Q-carbon is formed at thehighest undercooling Tq, and diamond is nucleated atslightly higher temperature of Td.

B. Formation of Quenched-in (Q) Carbon and Diamond

The super undercooled carbon layer is formed nearthe film–substrate interface, which can break into acellular (filamentary) structure upon quenching leadingto formation of a new state of carbon.[8] The formationof a cellular structure results from interfacial instabilityat the solid–liquid interface, driven by either solutesegregation or strain.[9] Figure 3(a) illustrates thegrowth of microdiamonds from the Q-carbon filament

which contained nanodiamond nuclei. This new statequenched from super undercooled carbon, referred asQ-carbon hereafter, has a matrix of mostly sp3-bonded(>75 pct) amorphous carbon in which nanocrystallitesof diamond are embedded, as shown later. As measuredby the AFM step shrinkage, the Q-carbon is consider-ably denser than amorphous carbon. The numberdensity of diamond nanocrystallites depends upon thetime available for crystal growth before quenching.Diamond nucleation occurs by a homogeneous nucle-ation and growth process, as shown in Figure 3(b),where diamond crystallites range in size 2 to 8 nm. Thisnucleation and growth can be optimized to cover theentire region with microdiamonds, as shown inFigure 3(c). Figure 3(d) shows the formation of alarge-area single-crystal diamond thin film when anepitaxial template was provided. This also shows thepath for integrating diamond thin films on siliconsubstrates for integrating functionality in next-genera-tion solid-state devices.The formation of nanodiamond can occur as a result

of homogeneous nucleation of diamond phase from thehighly undercooled pure carbon. For a homogenousnucleation of diamond from highly undercooled state ofpure carbon, the Gibbs free energy of diamond nuclei(DGT) consists of gain in volume energy (DGV) andexpense of surface free energy (DGS) terms and can bewritten as

DGT ¼ DGV þ DGS; ½1�

DGT ¼ �4

3pr3

qMm

DHm

TmDTu þ 4pr2rS; ½2�

where ‘r’ is the radius of diamond nucleus,(q=Mm½DH�m=Tm ½DT�u) is the gain in free energy forthe formation of diamond nucleus from the undercooled

Fig. 2—Gibbs free energy as a function of temperature for graphiticcarbon (Gg), liquid carbon (Gliq), and diamond (Gd). The amorphousQ-carbon is formed as a result of quenching from Tq and diamondsnucleate at Td.


RETRACTEDART

ICLE

state, DTu is the undercooling from Tm to Tr (temper-ature of nucleation), q is the solid diamond density, Mm

is the molar mass, DHm is the latent heat of melting, Tm

is the melting point of carbon, and rS is the averagesurface free energy between diamond nuclei and under-cooling carbon liquid.

The maximum of DGT* corresponds to the diamondreaction barrier at a critical size of r*, where

r� ¼ 2rSTmMm

DHmDTuq; ½3�

DG�T ¼ 16prS3T2

mM2m

3DH2mDT

2uq

2: ½4�

Rate of nucleation (I) is govern by

I ¼ A expDG�

T

KTr½5�

where Tr ¼ Tm � DTu, A = n(kT/h) exp (�DFA/kT),I = number of diamond nuclei cm�3 s�1, n = numberdensity of atoms, and DFA is free energy of activationacross the liquid–solid interface. Our calculated values

for 5 and 10 nm diamond observed crystallite size leadto A values of 1025 and 1024 cm�3 s�1, respectively. Itshould be mentioned that in the presence of epitaxialtemplate during growth from liquid, it is possible tocreate large-area single-crystal diamond films directly.

C. Formation of Diamond Nano- and Microneedles

When carbon melts around 4000 K in a highlyundercooled state near the carbon film/sapphire interface,there is a reduction in volume or shrinkage. Since thecarbon in the molten state is metallic, therefore, carbonatoms assume a closed packed structure, which can bequenched into the Q-carbon. This shrinkage and internalmelting result in the formation of bubbles which burstout and single-crystal diamond microneedles and nano-needles grow out of these areas, depending on the size ofthe bubble. The formation of two to three microns longmicroneedles will require growth velocities of the order of5 to 10 ms�1 with estimated 250 to 500 ns melt lifetime,which is only possible under crystal growth from meltunder highly undercooled state of carbon.[10]

The diamond structure determination and phaseidentification have been carried out using ElectronBackscatter Diffraction (EBSD), also known as

Fig. 3—Formation of Q-carbon, nano-, micro-, and large-area diamonds after laser irradiation; (a) formation of microdiamonds from the fila-ments toward the edge of 0.6 J cm�2 sample; (b) homogeneous nucleation of diamond (2 to 8 nm diameter) after laser irradiation at energy den-sity of 0.55 J cm�2; (c) microdiamonds covering entire area (with some nanodiamonds on the top); and (d) large-area single-crystal diamond thinfilm on Cu/TiN/Si(100) epitaxial template.


RETRACTEDART

ICLE

backscatter Kikuchi diffraction (BKD), by using fieldemission scanning electron microscope. This is a pow-erful technique to determine the crystal structure ofnano- and microdiamonds (with resolution of 10 nm)and their relative orientations with respect to thesubstrate. In the EBSD, a stationary electron beamstrikes a tilted crystalline sample and the diffractedelectrons form a pattern on a fluorescent screen. Thispattern is characteristic of the crystal structure andorientation of the sample region from which it wasgenerated. It provides the absolute crystal orientationwith sub-micron resolution. The results are shown inFigure 4 from a microdiamond with SEM micrograph,characteristic diamond Kikuchi pattern and orientationrelationship of the diamond with respect to the substratenormal. The microdiamonds grow here in the form ofmicroneedles up to two microns in length.

By providing a planar or lattice matching template, itis possible to grow large-area, single-crystal epitaxial

thin films as a result of crystal growth from superundercooled melt. This will be discussed in a separatesection in the end covering both diamond and cBNepitaxy. These microneedles grow out of the Q-carbonformed near the sapphire interface through the carbonover layers which must be molten during the growth ofthese needles. The formation of such long microneedlescan occur only in the liquid phase, where diffusivity is ofthe order of 10�4 cm2 s�1. The residual amorphouscarbon can be etched away by hydrogen and oxygenplasma, leading pristine diamond structure. Figure 4(a)shows SEM of microneedles and corresponding Ramanresults are shown in Figure 4(b), where only very sharpRaman diamond peak at 1136 cm�1 along with sharpsubstrate peak are observed. The inset shows EBSDwith the characteristic diamond Kikuchi pattern. Wepropose that the formation of microneedles occurs byrapid explosive recrystallization, where nanodiamondsnucleate from the Q-carbon and grow rapidly byliquid-mediated explosive recrystallization[11] to lengthsof the order of microns. The length of a nanoneedle or amicroneedle can be determined by growth velocity in theliquid phase, which is given by v = KsTm/(((Ds)0.5qL),Ks is thermal conductivity of sapphire(5.65 W m�1

K�1), Tm is melting point of carbon (4000 K), D is thethermal diffusivity (1.0 9 10�6 m2 s�1), q is mass den-sity of liquid carbon (3.5 g cm-3), and L is latent heat ofcarbon (8000 J g�1)6. Substituting these values, a roughestimate of the growth velocity is obtained to be 2 to3 ms�1, giving the length of microneedle one to twomicrons.

D. Raman Spectroscopy of Q-carbon and Diamond

Figure 5(a) shows Raman results from Q-carbon onsapphire substrate after a single laser pulse of ArF laser(in the outer regions of energy density of 0.6 J cm�2). TheRaman spectrum contained a diamond peak at1333 cm�1 with a broad peak around 1350 cm�1 and asmall peak at 1140 cm�1, associated with strained sp2

carbon at the interface. A Voigt profile containingconvolution of both Gaussian and Lorentzian profilesfits best to the acquired Raman spectrum. The peakpositions are fixed at 1140, 1333, and 1580 cm�1.Evaluation of the ratio of integrals helps us to obtain a76 to 81 pct sp3 and the rest sp2 fraction.[12] The S1 and S2peaks of the spectrum belong to sapphire substrate,whose contribution was subtracted in sp3 estimates, andare not shown in Figure 6(a) to avoid confusion. A slightup shift of the primary Raman peak is related toquenched-in strains, and a bump at 1140 cm�1 in theQ-carbon spectrum is characteristic of sp2 bonded carbonat the interfaces in nanodiamond. Figure 5(b) showsRaman results from microdiamond on sapphire substrateafter a single laser pulse of ArF laser (energy density(0.6 J cm�2). A sharp diamond peak at 1331.54 cm�1

along with sapphire peaks S1 and S2 (at 1360 and1375 cm�1) and small G peak of residual unconvertedamorphous graphite are observed. The Raman shift (Dx)is related to Dx (in cm�1) = 2.2 ± 0.10 cm�1 GPa�1

Fig. 4—Formation of diamond microneedles: (a) SEM micrographand (b) Raman spectrum with the diamond peak at 1336 cm�1. Theinset in (a) shows electron backscatter diffraction pattern of dia-mond.


RETRACTEDART

ICLE

along the [111] direction, Dx (in cm�1) = 0.73 ±0.20 cm�1 GPa�1 along the [100] direction, and Dx (incm�1) = 3.2 ± 0.23 cm�1 GPa�1 for the hydrostaticcomponent.[13] The biaxial stress in thin films can bedescribed as a combination of two-thirds hydrostatic andone-third uniaxial stress. The biaxial stress can beestimated using r ¼ 2lðð1þ mÞ=ð1� mÞÞ:Da:DT, where lis shear modulus,m is Poisson’s ratio, Da is the change inthermal coefficient of expansion and DT is the change intemperature.

E. HRTEM and EELS from Q-Carbon and Diamond

The HRTEM and EELS characterization of Q-car-bon and diamond were carried out using the

cross-section sample containing nano- and microdia-monds on sapphire. Figure 6(a) is a cross-sectionHRTEM image from a diamond microcrystallite, wherethe h110i cross section has two sets of {111} planes andh110i columns of diamond are clearly imaged. Thecharacteristic h110i diamond selected-area-diffractionpattern is included in the inset. This diamond film hasgrown epitaxially with underlying (0001) substrate withthe following epitaxial relationships: h111i dia is alignedh0001i sapphire in the direction normal to the film, andin plane of the film, it has h110i dia aligned withh�2110i sapphire and h112i dia aligned with h�1010i ofsapphire. Figure 6(b) shows HRTEM image from theQ-carbon near the sapphire substrate, where it hasmostly amorphous structure with a few nanodiamondsembedded into it, which show {111} planes aligned with(0001) planes of the sapphire substrate. The diffractionpattern shown in the inset shows (�2110) pattern with(0001) and (�1101) planes parallel to the film–substrateinterface. To investigate the details of bonding charac-teristics, EELS studies were carried out using aberra-tion-corrected STEM-FEI Titan 80 to 300 with anenergy resolution of 0.15 eV. Figure 6(c) is an EELSspectrum from diamond, showing characteristic sp3 (r*)bonding. The spectrum contains a sharp edge at 288 eVwith a peak at 292 eV, corresponding to sp3 (r*)bonding, which is a signature EELS spectrum fordiamond. The characteristic EELS spectrum from theQ-carbon is shown in Figure 6(d), which has a slopingedge at 285 eV with a broad peak at 292 eV. From theVoigt profile fit of the EELS spectrum, the sp3 wasestimated to be about 80 pct and the rest sp2, which isconsistent with Raman results from the Q-carbon, and isalso shown in Figure 6(d). The peak position for r* andp* edges are fixed at 292 and 285 eV, respectively. Allthe other fitted peaks are not shown to avoid confusion.

F. Diamond Deposition on Heat-Sensitive PolymerSubstrates

By using ArF pulsed Excimer laser, heating can becontrolled spatially and temporally in such a way thatwhile amorphous carbon films are melted, substratesstay close to ambient temperatures. Thus, diamond thinfilms can be formed on heat-sensitive plastics andpolymer substrates. Figure 7(a) shows SEM micrographof microdiamonds and nanodiamonds with averagegrain size of 30 and 500 nm, respectively, on ahigh-density polyethylene (HDPE) substrate. Since thesediamond grains are formed during rapid quenching, thegrains contain high-index facets unlike low-index facetsproduced during CVD processes. The high-index facetshave been shown to be catalytically more active andprovide nucleation sites for subsequent diamondgrowth. The Raman spectrum, shown in Figure 7(b),contains characteristic diamond peak at 1335 cm�1 andHDPE peak at 1464 cm�1 after the laser treatment. TheRaman spectrum from the as-deposited amorphousdiamondlike carbon films contains a broad peak around1350 cm�1, from which sp3 was extracted to be 45 pct.

Fig. 5—Raman characterization using laser wavelength (633 nm) ofQ-carbon and diamond: (a) Raman spectra from three different re-gions of Q-carbon with embedded diamond, showing diamond peakat 1331 cm�1 and diamondlike carbon peaks at 1140 and 1580 cm�1.The sp3 fraction derived is in the range of 75 to 85 pct, The S1 andS2 peaks belong to sapphire; (b) Raman spectra from microdiamondregion and the as-deposited DLC film with sp3 fraction (52 pct).


RETRACTEDART

ICLE

G. Ferromagnetism in Q-Carbon

The Q-carbon samples with sapphire substrateswere investigated for ferromagnetic properties by usingPhysical Property Measurements System (PPMS).The results showed room-temperature ferromagnetism(RTFM) in Q-carbon, as shown in Figure 8, where M vsH curve has finite coercivity (~150 Oersted) at roomtemperature. It is interesting to note that saturationmagnetization in the M–H plots decreases only slightlywith temperature, and coercivity increases only slightlyfrom ~150Oersted at 300 K to ~200Oersted at 10 K. Thesaturation magnetization Ms(T) is given by Ms(T) =Ms(0) tanh [(Ms(T)/Ms(0))/(T/Tc)], where Ms(0) is satu-ration magnetization at zero degree K and Tc is Curietemperature. Using this equation, a decrease by 10 pct inMs(T) at 300 K gives a rough estimate for the Curietemperature of Q-carbon about 500 K. From the satura-tion magnetization value of 20 emu g�1, the magnetic

moment in the Q-phase is estimated to be 0.4 lB (Bohrmagneton) per atom. Controlled samples with onlydiamond, diamondlike carbon, and sapphire substrateshowed only diamagnetism, which was subtracted in theplot, as shown in Figure 8. The magnetic force micro-scopy measurements (in the inset) from a triple pointshow a granular (about 50 nm diameter) magneticcontrast in the Q-carbon and a dark contrast fromdiamagnetic diamond. The granular features reflect thedistribution of sp3- and sp2-bonded carbon, where theamorphous grains have mostly sp3- with sp2-bondedcarbon at the boundaries. The diamond nucleates pref-erentially at the triple points of Q-carbon. The magneticcontrast from MFM was compared carefully with EFMcontrast to distinguish it from electric field contrast fromthe surface. The origin of magnetism may be related tounpaired spins generated by mixing of sp3 and sp2

bonding in the Q-carbon. This is the first direct evidence

Fig. 6—High-resolution TEM and EELS from diamond and Q-carbon from cross-section TEM sample with diamond on sapphire using FEI Ti-tan: (a) h110i HRTEM image showing individual columns of atoms with resolution of 0.18 nm with inset h110i electron diffraction pattern; (b)HRTEM from Q-carbon showing amorphous structures and some nanodiamonds with inset showing SAD pattern of sapphire; (c) EELS spec-trum from microdiamond (energy resolution 0.15 eV) showing a sharp edge at 288 eV with a peak at 292 eV; and (d) EELS spectrum fromQ-carbon has a sloping edge at 285 eV with a broad peak at 292 eV.


RETRACTEDART

ICLE

for bulk intrinsic ferromagnetism in carbon without thepresence of hydrogen or any other impurities. Ferromag-netism in carbon has been observed near the surface inproton- and carbon ion-irradiated HOPG, where itsorigin was linked to p-electrons and hydrogen-defectcomplexes.[14] Ferromagnetism has been predicted theo-retically[15] in bulk carbon containing 50 pct sp2- and50 pct sp3-hybridized carbon. The structure of thisferromagnetic phase is predicted to be rhombic type(a = 0.2608 nm, b = 0.3961 nm, c = 0.5289 nm witha = b = c = 90�), which is intermediate between thegraphite and diamond structure with mass density of2.9 g cm�3). However, in our experiments, amorphouscarbon with 50 pct sp2 and 50 pct sp3 is found to bediamagnetic. This amorphous carbon turns magneticonly when it is melted under super undercooled state andquenched to form Q-carbon. Experimentally, ferromag-netism has been also observed[16] in polymerized state ofrhombohedral C60, where ferromagnetic order resultsfrom unpaired spins with magnetic moment estimated as0.4 lB per atom, in agreement with our value. It should bepointed out that no ferromagnetism was observed inpristine fullerene or depolymerized C60 samples. It hasbeen observed that under high pressures (~17 GPa) and

ambient temperatures,[17] about 50 pct of p bonds ingraphite convert into r bonds. It would be interesting tosee if the unpaired electrons associated with p bonds leadto ferromagnetic order under high pressures, where massdensity is higher.

H. Enhanced Field Emission, Electrical conductivity, andHardness of Q-Carbon

Surface potential measurements were carried out toassess the field emission characteristics of the Q-carbonphase. The surface potential measurements were carriedout using Kelvin Probe Force Microscopy (KPFM) onthe Q-carbon filaments, which were embedded intodiamondlike carbon. The results shown in Figure 9show a lower surface potential, compared to diamond-like carbon up to 40 meV, indicating higher fieldemission potential compared to diamondlike carbonwhich already has higher field emission properties. It isinteresting to note that surface potential decrease atnanodiamonds, which are embedded into the Q-carbonfilaments, is lower than Q-carbon values. These resultson KPFM surface potential measurements are consis-tent with secondary electron emission contrast in theSEM images, showing significantly enhanced contrastfor the Q-carbon filaments. The shrinkage at theQ-carbon filaments was found to be about 35 to40 nm, indicating that the mass density of Q-carbon isconsiderably higher than that of the as-depositedamorphous carbon.[8]

The resistivity of carbon films was measured beforeafter the laser treatment, where there are embeddedfilaments. Preliminary results showed a significantdecrease in resistivity after the laser treatments insamples with embedded filaments. The decrease in

Fig. 7—Formation of diamond on HDPE: (a) SEM micrograph and(b) characteristic Raman spectrum before and after laser annealingwith 0.8 J cm�2, containing diamond peak at 1335 cm�1 and HDPEpeak (P) at 1464 cm�1.

Fig. 8—Bulk ferromagnetism in carbon: M–H curves at differenttemperatures from sample containing Q-carbon filaments with ahigher magnification inset at 100 K. The insets also show diamag-netic behavior before the laser treatment, and MFM measurementsfrom a triple point with a granular (about 50 nm diameter amor-phous sp3 carbon) magnetic contrast in the Q-carbon and a darkcontrast from diamagnetic crystalline diamond.


RETRACTEDART

ICLE

resistivity with increasing temperature showed charac-teristics semiconducting behavior up to 125 K and asemiconductor-to metal transition. Depending upon thequenching rate, we can control the semiconductor stateand preserve it at room temperature. The Q-carbon isformed as a result quenching from the melt, which isexpected to be metallic. During melting, there is over 10pct decrease in volume, and upon quenching, thestructure is amorphous with high sp3 fraction and therest sp2 fraction. Thus, Q-carbon in this state isamorphous and exhibits semiconducting behavior.Two resistivity transitions observed at 125 K (semicon-ductor to metal) and 185 K (metal to semiconductor)may be part of Q-carbon characteristics. Further studiesare underway to unravel the contributions of competingscattering mechanisms (variable range hopping, elec-tron–electron and electron–phonon scattering) as afunction of temperature.[10]

The formation Q-carbon occurs as a result of quench-ing of the super undercooled state of liquid carbon. Theliquid carbon has metallic bonding[1] which promotesthe close packing of carbon atoms with atomic radiusof 0.070 nm. Upon quenching, these carbon atomsarrange themselves mostly into fourfold coordination(sp3-bonded carbon, covalent radius 0.077 nm) and therest into threefold coordination (sp2-bonded carbon).The relative fraction for sp3-bonded carbon varies from70 to 85 pct. Assuming a homogeneous mixture, theeffective radius is calculated to be 0.075 nm. Since bulkmodulus (B in GPa) varies Nc/4 (1972 � 220I)d�3.5,where Nc is the coordination number, I is the iconicityparameter accounting for the charge transfer, and d iscovalent radius in (A).[18,19] Considering the predomi-nance of first nearest neighbors in the amorphousstructure of Q-carbon, a rough estimate can lead to

lead to a few percent higher hardness than diamond.The hardness of Q-carbon filaments was measured byusing nanoindentor measurements. These filaments ofQ-carbon were embedded in the diamondlike carbon.The relative hardness of the Q-carbon filaments wasmeasured to be in the range of 35 GPa compared to21 GPa for the diamondlike carbon. Thus, Q-carbon isharder by over 60 pct compared to diamondlike carbon,suggesting that hardness of Q-carbon may exceed that ofdiamond as a result of shorter average C-C bond lengthof Q-carbon. The Q-carbon phase was also found to bequite stable as a function of temperature. The Ramanspectra as function of second laser pulse at 0.2, 0.4, 0.5,and 0.6 J cm�2 showed almost no change up to 0.4J cm�2, indicating that Q-carbon is quite stable againstthermal heating. Only after 0.5 and 0.6 J cm�2, whentemperatures exceed 4000 K, there are observablechanges in the Raman spectra. The estimated temper-atures using Singh and Narayan model (SLIM) calcu-lations after 0.2, 0.4, and 0.5 J cm�2 are 1330 K,2665 K, and 3330 K, respectively.[6]

I. Direct Conversion of hBN Into cBN

A direct conversion of hexagonal boron nitride(h-BN) into phase-pure cubic boron nitride (cBN) hasbeen accomplished by nanosecond pulsed laser meltingat ambient temperatures and atmospheric pressure inair.[20] According to the P–T phase diagram, thetransformation from hBN into cBN under equilibriumprocessing can occur only at high temperatures andpressures, as the hBN-cBN-Liquid triple point isat 3500 K/9.5 GPa or 3700 K/7.0 GPa with a recenttheoretical refinement.[11,21,22] Using nonequilibriumnanosecond laser melting, we have created super under-cooled state and shifted this triple point to as low as2800 K and atmospheric pressure. The rapid quenchingfrom super undercooled state leads to formation of anew phase, named Q-BN. The cBN phase is nucleatedfrom Q-BN depending upon the time allowed fornucleation and growth. Thus, we obtain a directconversion of hBN into cBN by controlling the kineticsof transformation at ambient temperatures and atmo-spheric pressure in air by nanosecond pulsed laserannealing. We present detailed characterization of hBNand cBN layers by using Raman spectroscopy, high-resolution scanning electron microscopy, electronbackscatter diffraction, HRTEM, and electron energyloss spectroscopy, and discuss the mechanism of forma-tion of crystalline c-BN. We have also depositeddiamond on cBN and obtained cBN/diamond compos-ites, where cBN acts as template for diamond growth.

IV. INTRODUCTION

Boron nitride exists in four polymorphs, namely,hexagonal (hBN), rhombohedral (rBN), wurtzitic(wBN), and cubic (cBN) zinc-blende structures. Out ofthese, hBN (Point group = D6h, Space group = P63/mmc) and cBN (Space group= Fd3m) have generatedtremendous scientific and technological interests due to

Fig. 9—KPFM measurements, showing negative surface potentialassociated with Q-carbon compared to the matrix of diamond anddiamondlike carbon which already have low surface potential.


RETRACTEDART

ICLE

their analogous structures and properties with graphiteand diamond, respectively. The cBN (lattice constant =0.361 nm) and diamond (0.356 nm) have the numberdensity of atoms 1.69 9 1029 m�3 and 1.77 9 1029 m�3,respectively. This fact coupled with strong interatomicpotentials leads to the highest hardness (~7500 kg mm�2

for cBN and ~10000 kg mm�2 for diamond) and stiff-ness (Young’s modulus ~700 GPa for cBN and~1000 GPa for diamond). The thermal conductivity ofcBN (12 W cm�1 K�1) is close to that of diamond(20 W cm�1 K�1) which has the highest value of all theknown materials, including copper (4 W cm�1 K�1).The coefficient of friction of cBN and diamond is thelowest (less than 0.1) of all the known materials.However, cBN coatings have certain advantages overdiamond due to higher oxidation resistance than dia-mond because of protective boron oxide layers. Due toless reactivity even at high temperatures with ferrousalloys, cBN coatings are ideally suited for machiningthese alloys. The electrical properties of cBN are quitesimilar to those of diamond, where cBN has Johnson(for high-power devices) and Keyes (for integratedcircuits) figures of merit 8200 and 32 compared to 1for silicon. However, unlike diamond films which can bedoped reliably with p-type dopants only, cBN can bedoped with both n- and p-type dopants and it can haveboron oxide as the insulating layer needed for solid-statedevices. In view of the above properties, cBN anddiamond represent Holy Grail for solid-state devicesand systems, ranging from high-power devices to cuttingtools and biomedical applications.[23]

Here, we review direct conversion of hBN into cBN,which is analogous to carbon to diamond conversion, bya similar nanosecond pulsed laser melting at ambienttemperatures and atmospheric pressure in air in accor-dance with curve 3 of the phase diagram (Figure 10).This extension for BN phase diagram is very similar tothe one proposed for carbon.[12,13] Similar to Q-carbon,we have created Q-BN, which is formed as a result of

quenching from super undercooled state from whichphase-pure cBN is grown in the form of single-crystalnanodots, microcrystals, nanoneedles, microneedles,and large-area films. We have also grown diamond onthe top of these structures to create epitaxial cBN/dia-mond and diamond/cBN composites. We discuss thesimilarities and differences between hBN to cBN andcarbon to diamond conversions and accompanyingphase transformations.

A. BN Synthesis and Phase Diagram

According to the BN phase diagram (Figure 10), thehBN-cBN-Liquid triple point occurs at 3500 K/9.5 GPaor at 3700 K/7.0 GPa according to a recent theoreticalrefinement.[11,21,22] Thus, phase-pure cBN can be syn-thesized only near the triple point at high temperaturesand pressures. Recently, phase-pure cBN has beensynthesized at slightly lower pressures and temperatures(1700 K/5.5 GPa) by using a lattice matching diamondsubstrate for cBN growth from melt.[24] The cBNprocessing at high pressures and temperature involvesexpensive steps with limited yield. In addition, synthesisand scale-up processing of phase-pure cBN by bothPVD and CVD methods are still very challenging, onlyup to ~85 pct cBN content has been achieved.[23,25] ThePVD methods close to equilibrium lead to the formationof hBN, and the formation of cBN requires highlynonequilibrium processing requiring energetic ion bom-bardment, localized stress, and high concentration ofdefects.[26–28] These observations tend to favor Corriganand Bundy phase diagram (curve 1 in Figure 10) show-ing hBN as the stable phase and cBN metastable in theentire temperature range at ambient pressures.[11] Ther-modynamic calculations based upon BN melting andscaling of specific heats Cp

0(T) to high temperatures bySolozhenko et al.[21,22] have suggested curve 2, wheretriple point is shifted to lower pressures but highertemperatures and made cBN to be the stable phase 0 to1600 K and hBN beyond that. In view of the multiplefitting parameters in Cp

0(T) extrapolation to highertemperatures for calculating enthalpies and entropies(Table I), the accuracy of free energy values to betterthan 5 pct is questionable. It should be pointed out thatwhile the free energies of formation of cBN and hBN arelarge, the differences between the two are small less than4 pct. The PVD methods for cBN synthesis involvesignificant levels of ion bombardment during growth ina narrow window of parameters. The yields are verylimited due to sputtering losses from the ion bombard-ment, and CVD methods are not as well established asthose for diamond synthesis.[10] In the case of diamond,thermal and CVD techniques are fairly well established,although these processes occur at high temperatures inthe presence of hydrogen, and are very energy intensivewith limited yield.

V. RESULTS AND DISCUSSION

Hexagonal BN was deposited onto c-sapphire sub-strates at room temperature using ArF laser (pulse

Fig. 10—BN phase diagram showing P–T phase space for stabilityof hBN, cBN, and L (liquid BN), dotted lines due to Bundy[11] andrecent modifications,[21,22] and dash-dot line extension for superundercooling.[8]


RETRACTEDART

ICLE

duration = 20 ns, wavelength = 193 nm, energydensity = 3.0 J cm�2) in laser MBE chamber under avacuum of 3E�08 torr. The as-deposited films werenanocrystalline with grain size ~25 nm. These films wereirradiated using pulsed ArF laser having energy densitiesof 0.3 to 1.0 J cm�2. Figure 11(a) shows the formationof super undercooled BN (referred as Q-BN) fromwhich nano- and micro-cBN crystallites are formed. Theas-deposited hBN is in the form of nanocrystalline hBN,which is melted at atmospheric pressure in air at theestimated temperature of 2800 K. The Q-BN is formednear the sapphire interface, similar to Q-carbon, whichcan break into filamentary structure through interfacialinstability.[15] The Q-BN can be converted intonanocrystalline films, as shown in Figure 11(b), and

large-area h111i platelets and single-crystal thin films (asshown Figures 11(c) and (d)).The formation of nanoneedles and microneedles of

cBN is illustrated in Figure 12(a); it is interesting to notethat some of these microneedles are over two micronslong. The mechanism of formation of nanoneedles andmicroneedles is illustrated in Figure 12(b), where inter-facial instability in super undercooled BN leads toformation of periodic features of the order of 90 nm,which coalesce to form larger size microneedles. Alarge-area single-crystal thin films are formed in themiddle for the laser beam, where (0001) sapphiresubstrate is providing a template for (111) growth ofdiamond. There is 100 pct conversion of hBN intophase-pure cBN in the form of nanodots, nanorods,

Fig. 11—High-resolution SEM micrographs: (a) Q-BN and formation of nano- and microcrystalline cBN; (b) fully converted hBN into cBNnanocrystallites; (c) formation CBN platelets; and (d) large-area cBN films.

Table I. Thermodynamic Parameters

A. Specific heat coefficientsh-BN: a = 53.63023, b = 68.87958, c = 36927.910c-BN: a = 46.83548, b = �11.66081, c = 66261.937

B. Formation enthalpy DH0f298

h-BN = �250.6 ± 2.1 kJ m�1 and c-BN = �263.2 ± 2.3 kJ m�1

Phase Transformation DH0298 (kJ/m) DS0

298 (J/km) DG0298 (kJ/m)

C. Free energy of transformation at 298 K, DG0298K

h-BN fi c-BN �16.2 ± 3.0 �8.24 ± 0.11 �13.7 ± 3.0Graphite fi diamond +1.85 �3.38 +2.26


RETRACTEDART

ICLE

microcrystalline thin films, and large-area single-crystalcBN thin films. These structures are phase-pure cBNwith 100 pct conversion from hBN into cBN.

The cBN structure determination and phase identifi-cation have been also carried out using ElectronBackscatter Diffraction (EBSD), also known as

backscatter Kikuchi diffraction (BKD), by using fieldemission scanning electron microscope. The results areshown in Figure 13(a) from a nano-cBN with SEMmicrograph, characteristic cBN Kikuchi pattern andorientation relationship of the cBN with respect to thesubstrate normal. The nano-cBN grows here in the form

Fig. 12—High-resolution SEM micrographs: (a) Formation cBN nanoneedles and cBN microneedles up to three microns in length; (b) mecha-nism of initial stages of formation of nanostructures and evolution of nanoneedles; and (c) formation large-area flat cBN films.

Fig. 13—(a) Electron backscatter diffraction pattern from the small nanorod of cBN (shown as red dot) showing the inset Kikuchi pattern andthe orientation relationship; and (b) EBSD pattern from a microneedle of cBN having a length of about 10 microns.


RETRACTEDART

ICLE

of microneedles up to two microns in length. Thesemicroneedles grow out of the Q-BN formed near thesapphire interface through the BN over layers which mustbe molten during the growth of these needles. Theformation of such long microneedles can occur only inthe liquid phase, where diffusivity is of the order of10�4 cm2 s�1. The growth of a large-area cBN is illus-trated in Figure 13(b) with a corresponding EBSDpattern. The inset shows EBSD with the characteristiccBN Kikuchi pattern. We propose that the formation ofmicroneedles occurs by rapid explosive recrystallization,where nano-cBN nucleates from the Q-BN and growsrapidly by liquid-mediated explosive recrystallization[29]

to lengths of the order of microns. The length of ananoneedle or a microneedle can be determined bygrowth velocity in the liquid phase, which is given byv = KsTm/(((Ds)0.5qL), Ks is thermal conductivityof sapphire(5.65 W m�1 K�1), Tm is melting pointof BN (2800 K), D is the thermal diffusivity(1.0 9 10�6 m 2s�1), q is mass density of liquid cBN(3.45 g cm�3), and L is latent heat of cBN (1800 J g�1).[6]

Substituting these values, a rough estimate of the growthvelocity is obtained to be 5 to 10 ms�1, giving the lengthof microneedle about one to two microns.

Figure 14 shows Raman spectrum using 532 nmexcitation wavelength from hBN before laser annealingwhich has sharp E2g peak at 1370 cm�1, characteristic ofhigh-quality hBN. From high-resolution X-ray diffrac-tion, the average grain size was determined to be 25 nm,in agreement with high-resolution SEM data. After asingle laser pulse of ArF laser (in the outer regions ofenergy density of 0.6 J cm�2). The Raman spectra fromtwo different regions (1&2) contain all characteristic TOand LO peaks of cBN, where there is very small peak ofhBN (region 1) and is completely gone in region 2 (lowercurve), showing a complete conversion of hBN into

phase-pure cBN. Table II shows a detailed comparisonof our experimental results of first-order Raman peakswith previous experimental results from second-orderpeaks[30] and theoretical ab-initio calculations of Ramanactive modes in cBN.[31] Such a fine structure in theRaman spectra, besides well known TO(C) at 1060 cm�1

and LO(C) at 1310 cm�1, is indicative of very high-qual-ity structure of cBN. The Raman shift (Dx) is related tostrain along the [111] and [100] directions.[13] The biaxialstress in thin films can be described as a combination oftwo-thirds hydrostatic and one-third uniaxial stress. Thebiaxial stress can be estimated using r ¼ 2lð1þ mÞ=ð1� mÞ:Da:DT, where l is shear modulus, m is Poisson’sratio, Da is the change in thermal coefficient of expan-sion, and DT is the change in temperature.The HRTEM and EELS characterization of cBN and

Q-BN were carried out using the cross-section TEMsamples prepared by focused ion beam (FIB) thinningtechniques. The HRTEM results show epitaxial growthof cBN films on (0001) sapphire substrate. Figure 15(a)is a cross-section HRTEM image from cBN thin films,where the h110i cross section has two sets of {111}planes and h110i columns of cBN are clearly imaged.The characteristic h110i diamond selected-area-diffrac-tion pattern is included in the inset, showing first-order(111) and (002) spots. The HRTEM shows cBN to befree from defects, such as dislocations, twins, andstacking faults. These films have shown h111i orienta-tion aligned with h0001i sapphire substrate with in-planeepitaxial alignment as h110i cBN aligned with h�2110iof sapphire and h112i cBN aligned with h�1010i ofsapphire. Figure 15(b) shows HRTEM image from theQ-BN, where it has mostly amorphous structure with afew nanocrystallites embedded into it, as shown byarrows. To investigate the details of bonding character-istics EELS studies were carried out using aberra-tion-corrected STEM-FEI Titan 80 to 300 with anenergy resolution of 0.15 eV. Core loss electron energyloss spectroscopy (EELS) is a powerful technique todetermine the structure and bonding characteristics insolids, where the electrons are excited from core levels tothe unoccupied states and thereby providing a directway of measuring electronic transitions and structure.An intense sub-angstrom electron probe EELS in

Fig. 14—Raman spectra collected using 532 nm excitation wave-length showing characteristic E2g peak at 1370 cm�1 from phase-pure hBN; and characteristic TO(C) peak at 1066 cm-1 and LO(C)peak at 1311 cm�1 along with other second-order peaks as shown inTable I.

Table II. Experimentally Observed Raman Active

Vibrational Modes of c-BN

OpticalBranch

Theory(cm�1)[18]

Experiment(cm�1)[17]

Our Experiment(cm�1)

TO(X) 900 900 902TO(K) 910 915 918TO(Q) 945 940 948TO(W) 965 970 971TO(Q) 1000 1000 998TO(C) 1035 1055 1066LO(K) 1075 1085 1075LO(L) 1140 1135 1142LO(C) 1285 1305 1311

The theoretical and experimental values are obtained from ab-initiocalculations[18] and previously reported values,[17] respectively.


RETRACTEDART

ICLE

conjunction with HRTEM is used to evaluate variousband edge structures of B and N in c-BN and Q-BN,formed after laser annealing process. Figures 15(c) and(d) represent EELS spectrum of c-BN and Q-BNrespectively. The insets represent B K-edge of c-BNand Q-BN. The c-BN formed after laser annealing hasits characteristic B K-edge at 197.5 and 216 eV, which isin accordance with the previously reported results.[32,33]

There is a complete absence of the 216 eV peak fromw-BN due to its different density of states as comparedto c-BN.[32] There are also characteristic N K-edges at408 and 415 eV, which show that the cBN formed afterlaser annealing technique is pristine. Hexagonal BN andw-BN have their characteristic B K-edge at 192 eVwhich is completely absent in the EELS spectra of laserannealed cBN. BN is a partially covalent material owing

to the different electronegativity values of B and Natoms. In h-BN, there occur r states on its basal planewith weak interlayer p states thereby having character-istic r*and p* edges in EELS spectrum. There are both pedge (at 192 eV) and r edge (at 197 eV) in h-BN andw-BN which are affected by the strong attractiveinteraction between the core hole and excited elec-tron.[32] Interestingly, it is found that the EELS spec-trum of Q-BN is uniquely different from both h-BN andc-BN. It has B K-edge peaks at 195 and 204 eV whichcorrespond to r edges.[32] The Q-BN is supposed to havea higher sp3 fraction than h-BN but lower than that ofc-BN which causes the above-mentioned shift in the redge peaks. The intensity of higher energy band edge Npeaks is lower owing to the reduced thickness of theQ-BN layer.

Fig. 15—HRTEM and EELS from phase-pure cBN and Q-BN: (a) h110i cross-section showing defect-free cBN and corresponding h110iselected-area-diffraction pattern; (b) HRTEM from Q-cBN which is mostly amorphous with embedded nanocrystallites of cBN, and the insetdiffraction pattern shows h�2110i oriented sapphire substrate; (c) EELS spectrum from phase-pure cBN showing characteristic B K-edges at197.5 and 216 eV and N K-edges at 408 and 415 eV; and (d) EELS spectrum from Q-BN showing characteristic B K-edges at 195 and 204 eVcorresponding to r bonding.


RETRACTEDART

ICLE

A. Growth of Large-Area Single-Crystal Thin Films ofcBN and Diamond and Composites

We have created large-area epitaxial thin films ofdiamond and cBN on (0001) sapphire substrate byepitaxial growth from super undercooled liquid. Byproviding a planar matching (0001) sapphire epitaxialtemplate for diamond as well as cBN growth from superundercooled liquid, we have obtained large-area sin-gle-crystal thin films through domain matching epitaxyparadigm, where large misfit is accommodated byintegral multiple matching of lattice planes across the

film/substrate interface.[34] Table III provides a sum-mary of domain matching epitaxy parameters to accom-modate large lattice misfit between diamond/sapphireand cBN/sapphire interfaces.[34] Figure 16(a) shows alarge-area single-crystal h111i diamond thin film whichhas grown epitaxially on (0001) sapphire substrate witha following epitaxial relationships: h111i dia //h0001isapphire out of the plane and in-plane h110idiamond // h�2110i sapphire. According to the domainepitaxy paradigm, 19 {110} half-planes match with 20(�2110) planes of sapphire to accommodate the planar

Table III. Diamond and cBN Epitaxy on Sapphire (0001)

Epitaxial relationsh111i Diamond or cBN // h0001i Sapphireh110i Diamond or cBN // �2110

� �Sapphire

h112i Diamond or cBN // 10�10� �

SapphireDomain matching epitaxy via matching of planesa) DME of h111i Diamond on h0001i SapphireIn plane: 19 (2d110) Diamond = 20 (d�2110) Sapphireb) DME of h111i cBN on h0001i SapphireIn plane: 15 (d110) c-BN = 16 (d�2110) Sapphire

Fig. 16—(a) Formation of large-area epitaxial h111i diamond thin film on (0001) sapphire substrate; and (b) h111i cBN thin film on (0001) sap-phire substrate. The inset EBSD patterns show h111i orientations and the TEM results (Figs. 7(a) and 15(a)) confirm diamond and cBN epitaxyon (0001) sapphire.


RETRACTEDART

ICLE

lattice misfit. Figure 16(b) shows a large-area sin-gle-crystal h111i cBN thin film which has grownepitaxially on (0001) sapphire substrate with a followingepitaxial relationships: h111i cBN // h0001i sapphire outof the plane and in-plane h110i cBN // h�2110i sapphireand h112i cBN// h�1101i sapphire. According to thedomain epitaxy paradigm, 15 {110} half-planes matchwith 16 (�2110) planes of sapphire to accommodate theplanar lattice misfit.[34] These results on epitaxial growthare fully consistent with cross-section TEM results,shown in Figure 7(a) for diamond and Figure 15(a) forcBN. Since diamond and cBN are grown from superundercooled liquid, they can be doped with both n- andp-type dopants. We have also created cBN/diamondepitaxial composites by growing diamond films on ourphase-pure cBN films. The diamond films were depos-ited by pulsed laser evaporation of amorphous carbontarget using KrF (248 nm laser wavelength) laser withenergy density 3 to 4 J cm�2, Pulse duration 25 ns, and10 Hz with oxygen partial pressure of 0.2 torr. TheRaman results are shown in Figure 17, where there arecBN phase-pure cBN peaks and strong diamond peak at1333 cm�1.

VI. SUMMARY

hBN into cBN Conversion and Properties: We haveshown that nanocrystalline hBN can be converteddirectly into phase-pure cBN at ambient temperatureand atmospheric pressure in air. By using nanosecondlaser pulses, we create super undercooled molten stateof BN which is quenched into a new state of BN,which we have named Q-BN, analogous to theformation of Q-carbon. High-quality phase-purecBN is formed from Q-BN, and by controlling thenucleation and growth of cBN, we can form nan-odots, microcrystallites, nanoneedles, microneedles,and large-area thin films. Large-area epitaxial cBNthin films are formed in the presence of planar (likesapphire) or lattice matching (like copper) substrates

which provide a template during growth from superundercooled liquid. The atomic structure of hBN,cBN, and Q-BN and bonding characteristics havebeen studied by high-resolution, SEM, HRTEM,STEM, EELS, and Raman spectroscopy. We havealso created cBN/diamond epitaxial composites withprofound implications for high-speed machining andnext-generation microelectronic, photonic, and powerdevices. The preliminary studies have shown Q-BN tobe ferromagnetic, and exhibit superior field emission,hardness and optical and electronic properties, anal-ogous to the properties of Q-carbon. The details ofthese properties of Q-BN and cBN will be reportedshortly.Carbon into Diamond Conversion and Properties: Wehave created a novel phase of carbon by nanosecondlaser melting and quenching carbon from the superundercooled state in the form of thin films orfilaments. This Q-carbon has amorphous structurewith some embedded nanocrystalline diamonds whosenumber density is determined by the nucleation timeavailable for growth. The Raman spectrum from theQ-carbon shows a very large fraction of sp3 (75 to85 pct)-bonded carbon, along with diamond peak at1333 cm�1 and sp2-bonded peaks at 1140 and1580 cm�1. Nanodiamonds, microdiamonds, nano-needles, microneedles, and thin films are readilyformed from the Q-carbon depending upon the timeallowed for growth during the quenching period.Large-area epitaxial diamond thin films are formed inthe presence of planar (like sapphire) or latticematching (like copper) substrates during growth fromsuper undercooled liquid. The Q-carbon is found tobe ferromagnetic at room temperature and above,whose origin is linked to the p-electrons. Thisferromagnetism can explain now magnetism in earth’smantle and other planets with liquid and quenched-inQ-carbon. It also solves the mystery of formation ofnatural diamonds in the planets. This method allowssynthesis and processing of microdiamonds, micro-needles, and nanostructures of diamonds rather inex-pensively on practical and heat-sensitive substrates fora variety of applications ranging from abrasivepowders, novel catalytic properties, smart displays,myriads of biomedical, and microelectronic andnanoelectronic applications with a maximum impacton humankind

ACKNOWLEDGMENTS

Part of this research was presented at the 2015MS&T Symposium, honoring Professor J. C. M. Lifor his 90th birthday celebrations. We are grateful toFan Family Foundation Distinguished Chair Endow-ment for Professor J. Narayan, and this research waspartly funded by the National Science Foundation. Weare also very pleased to acknowledge technical helpand useful discussions with John Prater, Jim LeBeau,Weizong Xu, Roger Narayan, and Jerry Cuomo.

Fig. 17—Raman spectra from cBN/diamond epitaxial composites,showing characteristic cBN and diamond peaks.


RETRACTEDART

ICLE

REFERENCES1. F.P. Bundy, W.A. Bassett, M.S. Weathers, R.J. Hemley, H.U.

Mao, and A.F. Goncharov: Carbon N. Y., 1996, vol. 34,pp. 141–53.

2. J. Narayan, V.P. Godbole, and C.W. White: Science, 1991,vol. 252, pp. 416–18.

3. J. Narayan and A. Bhaumik: APL Materials, vol. 3, 2015, pp.100702-1-100702-11 and Two US Patents pending.

4. J.C. Angus and C.C. Hayman: Science, 1988, vol. 241, pp. 913–21.5. Y. Gogotsi, S. Welz, D.A. Ersoy, and M.J. McNallan: Nature,

2001, vol. 411, pp. 283–87.6. R.K. Singh and J. Narayan: Mater. Sci. Eng. B, 1989, vol. 3,

pp. 217–30.7. A.Y. Basharin, V.S. Dozhdikov, A.V. Kirillin, I.Y. Lysenko, M.A.

Turchaninov, and L.R. Fokin: Tech. Phys. Lett., 2010, vol. 36,pp. 559–62.

8. J. Narayan and A. Bhaumik: J. Appl. Phys., 2015, vol. 118,pp. 215303-1–215303-12.

9. J. Narayan: J. Appl. Phys., 1981, vol. 52, pp. 1289–93.10. J. Narayan and A. Bhaumik: COSSMS, DOI:10.1016/j.cossms.

2016.01.001.11. F.R. Corrigan and F.P. Bundy: J. Chem. Phys., 1975, vol. 63,

pp. 3812–20.12. J. Narayan and A. Bhaumik: Sci. Adv. (in press).13. S. Prawer and R.J. Nemanich: Philos. Trans. A. Math. Phys. Eng.

Sci., 2004, vol. 362, pp. 2537–65.14. H. Ohldag, T. Tyliszczak, R. Hohne, D. Spemann, P. Esquinazi,

M. Ungureanu, and T. Butz: Phys. Rev. Lett., 2007, vol. 98,p. 187204.

15. A.A. Ovchinnikov and I.L. Shamovsky: J. Mol. Struct.THEOCHEM, 1991, vol. 251, pp. 133–40.

16. T.L. Makarova, B. Sundqvist, R. Hohne, P. Esquinazi, Y.Kopelevich, P. Schraff, V. Davydov, L.S. Kashevaova, and A.V.Rakhmanina: Nature, 2001, vol. 413, pp. 716–18.

17. W.L. Mao, H.-K. Mao, P. Eng, T. Trainor, M. Newville, K.-C.Kao, D. Heinz, J. Shu, Y. Meng, and R.J. Hemley: Science, 2003,vol. 302, pp. 425–27.

18. A.K. Sharma and J. Narayan: Int. Mater. Rev., 1997, vol. 142,pp. 137–54.

19. A.Y. Liu and M.L. Cohen: Science, 1989, vol. 245, pp. 841–42.20. J. Narayan and A. Bhaumik: APL Mater. (submitted).21. V.L. Solozhenko: High Press. Res., 1995, vol. 13, pp. 199–214.22. V.L. Solozhenko, V.Z. Turkevich, and W.B. Holzapfel: J. Phys.

Chem. B, 1999, vol. 103, pp. 2903–05.23. C.B. Samantaray and R.N. Singh: International Materials Reviews,

2005, vol. 50, pp. 313–44.24. C. Chen, Z. Wang, T. Kato, N. Shibata, T. Taniguchi, and Y.

Ikuhara: Nat. Commun., 2015, vol. 6, pp. 6327-1–6.25. P.B. Mirkarimi, K.F. McCarty, and D.L. Medlin: Mater. Sci.

Eng. B, 1997, vol. 21, pp. 47–100.26. D.J. Kester and R. Messier: J. Appl. Phys., 1992, vol. 72,

pp. 504–13.27. J. Narayan, H. Wu, and R.D. Vispute: J. Electron. Mater., 1996,

vol. 25, pp. 143–49.28. J. Narayan: Mater. Sci. Eng. B, 1997, vol. 45, pp. 30–35.29. S. Reich, A.C. Ferrari, R. Arenal, A. Loiseau, I. Bello, and J.

Robertson: Phys. Rev. B, 2005, vol. 71, pp. 205201-1–12.30. K. Karch and F. Bechstedt: Phys. Rev. B, 1997, vol. 56,

pp. 7404–15.31. Y. Meng, H.-K. Mao, P.J. Eng, T.P. Trainor, M. Newville, M.Y.

Hu, C. Kao, J. Shu, D. Hausermann, and R.J. Hemley: Nat.Mater., 2004, vol. 3, pp. 111–14.

32. D.G. McCulloch, D.W.M. Lau, R.J. Nicholls, and J.M. Perkins:Micron, 2012, vol. 43, pp. 43–48.

33. M. Yoshimoto, K. Yoshida, H. Maruta, Y. Hishitani, H.Koinuma, S. Nishio, M. Kakihana, and T. Tachibana: Nature,1999, vol. 399, pp. 340–42.

34. J. Narayan and B.C. Larson: J. Appl. Phys., 2003, vol. 93,pp. 278–85.


RETRACTEDART

ICLE

http://dx.doi.org/10.1016/j.cossms.2016.01.001

http://dx.doi.org/10.1016/j.cossms.2016.01.001

Documents

Fundamental Discovery of New Phases and Direct Conversion of … · 2015. 11. 9. · formation. It has been conjectured that liquid carbon may exist as a thermodynamically stable