pss standard-reprint webversion - uantwerpenematweb.cmi.ua.ac.be/emat/pdf/1965.pdf · nitrogen...

reprint

Part of Topical Section onFundamentals and Applications of Diamond and Nanocarbons

Aberration-corrected microscopyand spectroscopy analysis ofpristine, nitrogen containingdetonation nanodiamond

Stuart Turner*,1, Olga Shenderova2, Fabiana Da Pieve1, Ying-gang Lu1, Emrah Yücelen3,4,Jo Verbeeck1, Dirk Lamoen1, and Gustaaf Van Tendeloo1

1 EMAT, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium2 International Technology Centre, 8100-120 Brownleigh Drive, Raleigh, NC 27617, USA3 FEI Company, Europe NanoPort, Achtseweg Noord 5, 5651 GG Eindhoven, The Netherlands4National Centre for HREM, Delft University of Technology, Lorentzweg 1, 2628 CJ Delft, The Netherlands

Received 4 July 2013, revised 21 August 2013, accepted 21 August 2013Published online 11 September 2013

Keywords detonation nanodiamond, HRTEM, EELS, DFT

*Corresponding author: e-mail [email protected], Phone: 0032 32653572, Fax: 0032 32653257

Aberration-corrected transmission electron microscopy, elec-tron energy-loss spectroscopy, and density functional theory(DFT) calculations are used to solve several key questionsabout the surface structure, the particle morphology, and thedistribution and nature of nitrogen impurities in detonationnanodiamond (DND) cleaned by a recently developed ozonetreatment. All microscopy and spectroscopy measurements areperformed at a lowered acceleration voltage (80/120 kV),allowing prolonged and detailed experiments to be carried outwhile minimizing the risk of knock-on damage or surfacegraphitization of the nanodiamond. High-resolution TEM(HRTEM) demonstrates the stability of even the smallestnanodiamonds under electron illumination at low voltage and isused to image the surface structure of pristine DND. Highresolution electron energy-loss spectroscopy (EELS) measure-

ments on the fine structure of the carbon K-edge ofnanodiamond demonstrate that the typical p� pre-peak in factconsists of three sub-peaks that arise from the presence of,amongst others, minimal fullerene-like reconstructions at thenanoparticle surfaces and deviations from perfect sp3 coordi-nation at defects in the nanodiamonds. Spatially resolved EELSexperiments evidence the presence of nitrogen within the coreof DND particles. The nitrogen is present throughout thewhole diamond core, and can be enriched at defect regions.By comparing the fine structure of the experimental nitrogenK-edge with calculated energy-loss near-edge structure(ELNES) spectra from DFT, the embedded nitrogen is mostlikely related to small amounts of single substitutional and/orA-center nitrogen, combined with larger nitrogen clusters.

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction Nanodiamond synthesized by deto-nation of carbon-based explosives (detonation nanodiamondor DND) continues to fascinate the scientific communitywith its versatile properties and plethora of potentialapplications. Nanodiamond is already used for CVD growthof larger diamond structures, polishing and hardening ofmetals, and plastics [1]. Many exciting new applications arealso being put forward, like graphitized nanodiamond as abattery electrode and supercapacitor material [2–4]. Thewide band gap in diamond (Eg� 5.4–5.6 eV) can also beadopted for unique applications like UV-blocking and

the enhancement of the field emission properties of cathodesin e.g., flat panel displays [1, 5].

As a result of its seemingly inherent nitrogen doping,DND is also known to exhibit photoluminescent properties.These properties allow DND particles to be used as opticalmarkers for biological imaging and drug delivery, even inits as-synthesized form (only non-diamond carbon removalsteps undertaken) [6, 7], without irradiation to producevacancies. In native DND, these properties are related to(rare) nitrogen-related centers (NV centers) formed bynitrogen impurities embedded during synthesis as well as to

Phys. Status Solidi A 210, No. 10, 1976–1984 (2013) / DOI 10.1002/pssa.201300315

Invited Articleapplications and materials science

a

statu

s

soli

di

www.pss-a.comph

ysi

ca

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

a luminescence originating from amorphous carbon andother defects. Intermittent luminescence from single NV�

centers in as-synthesized 5 nm DNDs was recentlymeasured, paving the road for large-scale production ofluminescent particles [8, 9]. Further work showed that thenative photoluminescence from NV� and NV0 centers couldbe enhanced in DND by a suitable post-synthesis irradiationand illumination procedure [10]. However, a great deal ofuncertainty remains about the position and nature of thenitrogen impurities within as‐synthesized DND. EPRmeasurements on native DND species synthesized from agraphite/RDX (cyclotrimethylenetrinitramine) precursor in-dicated a concentration of only 1–2 ppm of paramagnetic N(or close to 1 atom in a 30 nm particle) [11] and 0.02 ppm ofparamagnetic N in DND produced from a trinitrotoluene(TNT)\RDX mixture [7]. Recent EPR measurements haveshown high concentrations of 1 at.% NV centers combinedwith the same amount of Ns (P1 center) in sintered DND[12], in good agreement with previous spatially resolvedelectron energy‐loss spectroscopy (STEM‐EELS) measure-ments performed by our and other groups indicating morethan 2 at.% of N in DND [13, 14]. XPS measurements alsodemonstrated a far higher concentration of tetrahedrallyincorporated nitrogen present in DND [15]. The fact thatEPR measurements on native DND were not able to measurehigh N concentrations leads to the conclusion that most ofthe nitrogen in native DND must be non‐paramagnetic andpresent in an optically inactive form (such as large nitrogenclusters or enrichment of nitrogen at intrinsic defects in thenanodiamonds [10]) or in a form that is undetectable by EPR(like charge‐neutral A‐centers). The main difficulties indetecting and distinguishing between the various possibleforms of nitrogen in nanodiamonds are the small scale of thenanodiamond structures, the low spectroscopically detect-able nitrogen doping and the difficulties of producing andmeasuring DND samples with clean diamond crystal facets.The latter aspect is particularly important since nitrogen (ifmeasured) could also be present in the graphitic shell.

Being able to produce nanodiamond with a maximal sp3

diamond fraction (and a minimal non-diamond graphiticcarbon shell) is a major research challenge [1]. Even thoughmany cleaning and etching methods have been developedto remove the non-diamond shell from nanodiamonds,signals from C–C, C��C, CHx, C–O, and C��O species arefrequently measured by several spectroscopic and diffractiontechniques including electron energy-loss spectroscopy onDND [16]. In the past this led to the development of adiamond core – non-diamond shell model that was widelyaccepted [17]. However, previous experiments carried outby our group showed that diamond particles with a minimalsurface graphitic shell could be produced by severe acidtreatments [13]. A detailed surface analysis could not beperformed, as any clean diamond particle surface is extremelyprone to carbon contamination and graphitization at the highacceleration voltages used [18]. These experiments indicatedthat all transmission electron microscopy and spectroscopy ofdiamond nanoparticles is far from trivial [16].

In this work we set out to study clean DND, obtainedthrough a recently developed ozone-treatment [19, 20],using a state-of-the-art aberration corrected microscopeoperated at lowered acceleration voltages (80 and 120 kV).Low voltage microscopy has previously been used togreat success in imaging beam sensitive materials likenanotubes, graphene, meteoritic nanodiamond, and hexago-nal BN [21–23], and should allow for prolonged experimentsto be performed while minimizing the risk of knock-ondamage or surface graphitization of the nanodiamond [24].The surface structure of the clean nanodiamonds is furtherinvestigated through the fine structure of the carbon K-edgeby high resolution EELS with an electron monochromator.

By combining high resolution imaging with spatiallyresolved EELS, the nitrogen embedding in single nano-diamonds will be studied. Finally, we will compare theexperimental nitrogen K-edge fine structure signatureobtained from individual nanodiamonds to calculated spectraobtained from density functional theory (DFT) calculationsin an attempt to determine the coordination of the nitrogenimpurities in native DND.

2 Experimental2.1 Sample preparation The detonation soot that

was processed in the ozone treatment reactor was a productof the detonation of a TNT\RDX mixture (40/60wt.%) withice as a cooling medium. Prior to the treatment in theozone purification reactor, the soot was pre-purified of metalcontaminants by washing with acids at ambient temperature,followed by a washing in DI water and drying. Subsequent-ly, the nanodiamond sample was ozone purified at 200 8Cover 72 h in a fluidized bed reactor, as described in theliterature [20].

2.2 (Scanning) transmission electronmicroscopyand electron energy-loss spectroscopy High resolu-tion and spatially-resolved EELS experiments were carriedout on an FEI Titan 80–300 “cubed” microscope fitted withan aberration-corrector for the imaging lens and the probeforming lens, a monochromator and a GIF quantum energyfilter for spectroscopy. High-resolution TEM (HRTEM)experiments were carried out at 80 kV acceleration voltage.The Cs was set to þ15mm for the experiment and themonochromator was excited to provide a 0.3 eV energyspread of the illumination for maximum informationtransfer. HRTEM image simulations were carried out usingthe JEMS software package.

The monochromated EELS experiments were performedat 80 kV using a convergence semi-angle a of �18mrad,a collection semi-angle b of �108mrad, a dispersion of0.03 eV per pixel and at an energy resolution of �250meV.The STEM-EELS experiments were performed at 120 kVusing a convergence semi-angle a of �18mrad, a collectionsemi-angle b of �60mrad, an energy dispersion of 0.3 eVper pixel and an energy resolution of approximately 1 eV.The EELS data was fitted with reference spectra for diamondand amorphous carbon to generate the maps in Fig. 4. The

Phys. Status Solidi A 210, No. 10 (2013) 1977

www.pss-a.com � 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Invited

Article

nitrogen map was generated by integration of the intensityunder the N K-edge after standard power-law backgroundsubtraction in digital micrograph. Prior to N mapping, theEELS data was rebinned by a factor of 4 in the energy-direction to improve the signal to noise ratio of the map.When plotted, the core-loss spectra were backgroundsubtracted, aligned using the maximum of the p� peak (at285 eV) and normalized in intensity to their maxima. EELSquantification was also performed using the digital migro-graph software package, using 25 eV wide energy windowsfor the C–K and N–K edges.

2.3 Energy-loss near-edge structure (ELNES)calculations using density functional theory Ab initioDFT-GGA calculations using the PBE parametrization havebeen made using the all electron code WIEN2k, basedon the full potential linearized augmented plane wavemethod [25, 26]. The following values for the relevantparameters were used: radius of the muffin tin¼ 1.35, RKmax

(basis cutoff, i.e., the product of the smallest muffin tinradius in the systems and the length of the maximumK-vector of the interstitial plane wave basis)¼ 6.5;Gmax¼ 14.0. The converged k-mesh for the full Brillouinzone consisted of 2000 k-points for ground state calculationsand 750 k-points for the 2� 2� 2 supercells used tocalculate the ELNES with the core hole. No orientationdependence of the EELS spectra was considered, i.e., thecross-sections were averaged over all possible directions ofthe scattering vector with respect to the crystal. All structureshave been optimized allowing for a change of internalfractional atomic coordinates with default tolerances(convergence on total energy of 2meV). In some cases aninitial elongation of the N–C bond was set by hand and thesystem was subsequently allowed to relax, according tothe procedure adopted by other authors [27]. C and NK-edgeELNES calculations were performed on supercells usingthe full core hole approximation, i.e., introducing a frozenhole in the excited orbital and smearing the charge uniformlyin space. The calculated ELNES spectra were by broadenedby 1 eV to accommodate for instrumental broadening.

3 Results3.1 (High resolution) TEM imaging An overview

TEM image of the DND material used for this study isdisplayed in Fig. 1. The DND particles have primary particlesizes ranging between approximately 2 and 20 nm, and aregenerally agglomerated into larger structures. The crystalstructure of the particles can be determined from the insetelectron diffraction pattern; only rings corresponding to thediamond crystal structure (Fd-3m, Space Group No. 227)are present. No graphite or n-diamond [28] reflections arevisible, an important indication for a high purity of the DNDmaterial. The agglomeration of the DND particles is typicalfor nanosized materials, and is mainly due to van der Waalsattractive forces [29].

A HRTEM image of typical DND particles in the ozonetreated sample is presented in Fig. 2a. To acquire this image

the electron microscope was operated at 80 kV with thethird-order spherical aberration (Cs) tuned to þ15mm. Tofurther extend the information limit of the microscope, whichis normally limited by chromatic aberration, the monochro-mator of the microscope was excited to achieve an energyspread of the order of 0.3 eV for the illuminating electronbeam. By extending the information transfer of themicroscope in this way, intricate details of the nanodiamondsurface become visible (see Fig. 2a).

The DND particle is imaged along the [011] zone axisorientation, allowing {111} and {100}-type surface facetsand S¼ 3 {111} twin boundaries to be imaged edge-on.Several coherent S¼ 3 {111} twin boundaries run throughthe imaged particle, as well as a fivefold twin structure(indicated by white arrows). This type of twin is typical forDND as it has the lowest defect energy of all diamonddefects [30]. This results from the fact that this twin type onlydemands a small deviation from the ideal lattice, namelyonly in the fourth neighbor lattice positions. The abundanceof these defects has been attributed to the increased growthrate of twinned nanodiamonds in comparison to untwinnednanodiamond under detonation synthesis conditions [30].Although the presence of multiple coherent {111} twins wasapparent from earlier conventional HRTEM work, severalfeatures remained unclear until now. The first is the surfacestructure, which in this sample exhibits a minimal presenceof graphitic carbon. Secondly, the details of the multiple twinboundaries are now clearly imaged; no indication for anenrichment of amorphous carbon or nitrogen clusters at thetwin boundaries can be derived from the HRTEM image.

The nanoparticles show mainly {111}-type surfacefacets with some {100} truncation, a result of the lowsurface energy of these low-index facets [31]. To illustratethe stability of the nanodiamond surfaces under electronillumination at 80 kV, the same particle is imaged after 3min

Figure 1 Overview bright-field TEM image of ozone-treatedDND. The DND particle size ranges from approximately 2 to20 nm. The electron diffraction ring pattern (inset) proves thediamond crystal structure of the nanoparticles.

1978 S. Turner et al.: Microscopy and spectroscopy analysis of pristine, nitrogen containing DND

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-a.com

ph

ysic

a ssp stat

us

solid

i a

of continuous electron illumination in Fig. 2b. It is clear thatno major changes in surface structure of the diamond particlehave occurred over the illumination time, which is in strongcontrast to structural changes seen under 200 kV electronillumination [18]. A typical (100) truncation is imaged inmore detail in Fig. 2c. Under the imaging conditions used(Cs¼þ15mm, 4 nm underfocus) the atomic columns areimaged in white, as evidenced by the simulation of atruncated diamond particle in Fig. 2c. The combination of

the HRTEM image and image simulation allows to concludethat the surfaces (both the {111} and {100} type surfaces)are extremely clean in the ozone-treated sample. In theregion imaged in Fig. 2c, no graphitic surface material can beidentified.

To investigate the possibility of detecting substitutionalnitrogen atoms or vacancies by HRTEM imaging, extra imagesimulations were performed (not displayed). It is important topoint out that all atomic columns demonstrate similar imagecontrast in the experimental image. Image simulations usingthe truncated diamond particle showed no significant changein column contrast when either a vacancy or a substitutionalnitrogen atomwere inserted into a carbon column. This meansthat even at extended information transfer, HRTEM cannot beused to locate individual vacancies or substitutional nitrogenatoms and another technique (EELS) will be needed to detectnitrogen in the particles.

3.2 Electron energy-loss spectroscopy3.2.1 Carbon K-edge ELNES It is well known

that the fine structure of EELS edges is related to thelocal symmetry and environment of the excited atomicspecies [32, 33]. Therefore, the shape of the carbon K-edgefine structure should provide clues about the form inwhich carbon is present in the nanodiamond particles. Weperformed EELS experiments at high energy resolution(monochromator excited to provide an energy resolutionof 0.25 eV) to investigate the carbon K-edge fine structure.The carbon K-edge was acquired from several cleannanodiamond particles simultaneously to achieve the bestpossible signal to noise ratio.

The C–K EELS spectrum in Fig. 3 is dominated by the1s!s� feature starting at 290 eV, which is indicative ofcarbon in an sp3 coordination. As is typical for DND, a small

Figure 2 Stability under electron beam illumination; (a) HRTEMimage of DND particles exhibiting coherent S¼ 3 {111} twinboundaries (indicated by white arrows). (b) The same particles after3min of electron beam illumination. (c) HRTEM image of atruncated nanodiamond surface, showing a minimal graphiticpresence. (d) Image simulation performed for Cs¼ 15mm, Df¼�4 nm. The right side has the projected potential overlaid; the whiteimage contrast peaks at the atomic positions.

Figure 3 Experimental carbon K-edge fine structure (ELNES)acquired with monochromatic electron illumination. The s� featurestarting at 290 eV is typical for diamond. The p� contributionaround 285 eV consists of 3 pre-peaks, at 282.7 eV (A), 285 eV (B),and 286.6 eV (C).



Invited

Article

pre-peak at 285 eV is also present, which is absent for bulkdiamond. This 1s!p� feature is indicative of carbon ina lower or distorted sp3 coordination. As a result of theimproved energy resolution provided by the electronmonochromator, the data shows a splitting of the p�

contribution into three pre-peaks, at 282.7 eV (A), 285 eV(B), and 286.6 eV (C). These three peaks have beenexperimentally observed in UNCD diamond films and ininterstellar nanodiamond by EELS [34, 35], in DND usingEXAFS [36] and have been theoretically predicted by tight-binding molecular dynamics simulations [37] and DFT [36].Peak A has been observed in amorphous carbon anddiamond surfaces and is associated with mid-gap stateslocalized at dangling bonds, which can be stable underUHV conditions, and distorted sp3 sites. Peaks B and C

are attributed to unoccupied p� antibonding states of sp2

coordinated carbon, most likely arising from a fullerene-likesurface reconstruction and deviation from the ideal sp3

coordinated carbon symmetry at defect sites.

3.2.2 Core-loss EELS spectrum imaging To imagethe distribution of non-diamond carbon in individualnanodiamond particles and simultaneously confirm thepossible presence of nitrogen impurities within theseparticles, spatially resolved EELSmeasurements were carriedout. STEM-EELS results acquired on a multiply twinnednanodiamond particle are displayed in Fig. 4. The HRTEMimage of the multi-twin (taken at 120 kV acceleration) againshows the presence of simpleS¼ 3 coherent twin boundariesand lamellar multiply twinned regions, probably containinga large degree of other defects [10]. The particle surfaceappears to be nearly non-sp3 carbon free.

EELS data were acquired using the so-called spectrumimaging technique; in this technique the electron probe isscanned over the sample and an EELS spectrum is acquiredat each scan position. For the nanodiamond particle inFig. 4, a 53� 67 pixel spectrum image was acquired at 2Åintervals, recording the carbon K-edge (285 eV) and nitrogenK-edge (403 eV) simultaneously [38]. sp3 and non-sp3

carbon maps were generated from the acquired spectraby fitting references for amorphous carbon and diamondto the acquired carbon K-edge. The nitrogen map wasgenerated by integrating the intensity under the background-subtracted nitrogen K-edge at 403 eV. The generated mapsare displayed in Fig. 4c (sp3 C), Fig. 4d (non-sp3 C) andFig. 4e (N).

The sp3 carbon map clearly indicates the size of thenanodiamond core, which is almost equal to the particlesize from the ADF image. Although only minimal graphiticmaterial was visible in the HRTEM image, the non-sp3 Cmap clearly follows the surface contours of the nano-diamond particle. The origin of this signal is thereforeattributed mainly to some remaining fullerene-like sp2

material and dangling bonds, which can be stable underthe UHV conditions in the microscope, at the surface ofthe nanodiamond particle. A significant non-diamond Csignal is also measured from both the coherent S¼ 3 twinboundary (example indicated by A) as well as the lamellartwinned regions (example indicated by B). This signalagain arises from sp2 or distorted sp3 coordinated carbon,indicating that even the coherent twin boundaries candeviate from their ideal symmetry, and contain some non-sp3 coordinated carbon [39]. The nitrogen K-edge signalis mapped in Fig. 4e. As the nitrogen content in thenanodiamond particle is low, the elemental N map displaysa relatively high noise level. It is however clear that thesignal is localized in the diamond core of the particle, inaccordance with previous measurements [10]. The lineprofile in Fig. 4g, taken over the regions indicated in Fig. 4d,confirms the presence of nitrogen in the diamond core.

Interestingly, an enrichment of nitrogen at the multiplytwinned region is also apparent from the map. The nitrogen

Figure 4 Spatially resolved EELS analysis of a multiply-twinnednanodiamond; (a) HRTEM image of a typical DND particleshowing the presence of a single coherent twin indicated by A andmultiple parallel twin regions indicated by B. (b) ADF-STEMimage of the same particle; the white rectangle depicts the STEM-EELS scanned region (53� 67 pixel spectrum image). (c) sp3 Cmap, (d) non-sp3 C map, (e) N map, (f) N K-edge spectra fromregion 1 and 2 indicated in (b). The N content in region 1 is2.7� 0.2 at.%; defected region 2 shows a clear enrichment ofnitrogen (3.8� 0.2 at.%). (g) Intensity profiles for non-sp3 C(dashed line), sp3 C (black line), and N (bars) taken over theregion indicated by the arrow (d) (5 px integration width).



ph

ysic

a ssp stat

us

solid

i a

K-edge signal summed over two regions, region 1 being thenon-defected diamond region indicated in Fig. 4b and region2 being the multiply twinned region, are plotted in Fig. 4f.The spectra have been aligned to their continuum level toallow easy comparison. It is clear that both regions containsignificant amounts of nitrogen, however, the multiplytwinned area is nitrogen enriched as evidenced by theenhanced signal. The nitrogen content in both regions wasquantified using standard EELS quantification methods (seeSection 2). The average N content in pristine diamond coreregion 1 is 2.7 at.%. In defected region 2, the nitrogencontent is more than 1 at.% higher (3.8 at.%). On the otherhand, no clear nitrogen enrichment was measured at coherenttwin boundary A (not displayed).

3.2.3 Nitrogen K-edge ELNES As discussed earlier,the EELS edge fine structure is related to the local density ofstates, and is therefore sensitive to the local symmetry andenvironment of the excited atomic species. In experimentsperformed by our own and other groups, the shape and onsetposition of the nitrogen K-edge led to conclude that nitrogenimpurities were tetrahedrally embedded in the nanodiamondparticles under study [13, 14].

Untreated and smoothed experimental nitrogen K-edge fine structures from a diamond core region and theexperimental carbon K-edge of a nanodiamond particle areplotted in the left panel of Fig. 5. The nitrogen K-edge showsa clear triangular shape, with three defined peaks B, C, and Dand a less pronounced pre-peak A, which varies in size

depending upon the sample position where the data wasacquired. This nitrogen ELNES is similar in other DNDsamples, even in samples with very high nitrogen doping[14]. Furthermore, the nitrogen and carbon (diamond)K-edge fine structures are clearly similar; peaks B, C, and Dare all three present in both spectra. In the case of thecarbon K-edge, pre-peak A at 285 eV is related to non-sp3

hybridized carbon [33]. The large degree of similaritybetween the N and C ELNES indicates that nitrogen isembedded in an environment that is highly similar to that ofcarbon in diamond, being a tetrahedral setting.

In order to provide more detailed information on thelocal environment of the nitrogen impurities in diamond,we performed DFT calculations to obtain the nitrogenK-edge ELNES for four different nitrogen centers, being

(i) Single substitutional nitrogen (Ns or P1 center)(ii) Two neighboring embedded nitrogen atoms (A-center)(iii) The charge neutral nitrogen-vacancy center (NV0)(iv) The negatively charged nitrogen-vacancy center (NV�)

The DFT calculated spectra, performed in the GGAapproximation on 2� 2� 2 supercells in the full core holeapproximation using the code WIEN2k, are plotted inFig. 5 [38]. A first impression is that all four calculated N–Kedge spectra look similar. This is a result of the similarcoordination of nitrogen coordination in all four cases. Eventhough the spectra look similar, small differences can bemade out between them. The spectra for both NV centers arequasi-identical, and are visually different from the experi-mental nitrogen K-edge. Even though all the calculatedspectra have been broadened by 1 eV to take into accountinstrumental broadening in the experimental data, the spectrafor NV� and NV0 are narrower and less defined than theexperimental spectra. This is a confirmation of other datapresent in the literature that shows that the nitrogen presentin as-synthesized nanodiamond is not present in the form ofNV� or NV0 centers in sizeable amounts [10]. Significantnumbers of these defects can only be generated in DNDafter application of proper vacancy creation and sinteringtreatments [1, 12, 40].

The spectrum for single substitutional nitrogen Ns ismore triangular in shape, similar to the experimentalnitrogen K-edge ELNES, reproducing features B, C, and Das well as a pre-peak to the main feature B. This pre-peak isa p� contribution that arises from bond length elongationof one of the C–N bonds by approximately 30%, a result ofthe energy relaxation step in the calculation, which is ingood agreement with the literature [27]. Owing to thiselongation, the coordination of the substitutional nitrogenatom is more planar with respect to the other 3 bondingcarbon atoms, giving rise to the p� contribution [41]. Aswas the case for the calculated NV center spectra, theNs spectrum is less broad than the experimental nitrogenK-edge.

The calculated spectrum for the A-center model (2neighboring, substitutional nitrogen atoms) provides the best

Figure 5 Comparison of the experimental (left) and DFTcalculated (right) nitrogen and carbon K-edge fine structures.(a) Nanodiamond C–K ELNES. (b) Raw N–K ELNES fromnanodiamond. (c) 1.5 eV smoothed N–K ELNES from nano-diamond. Peak B corresponds to 406 eV for the nitrogen K-edgesand to 292 eV for the carbon K-edge. (d) Single substitutionalnitrogen Ns (e) A-center (2N), (f) NV

0, (g) NV�.



Invited

Article

match with the experiment. Peaks at B, C, and D arereproduced correctly, with peak positions that fit well withthe experimental ones. Pre-peak A, which is present in theexperimental ELNES, is however not reproduced by theA-center model. This result differs from earlier ELNESsimulation work using multiple-scattering theory, in whichpre-peak A was reproduced [41].

4 Discussion This work concentrates on the morphol-ogy and surface structure of pristine DND and thedistribution and nature of nitrogen in as-synthesizednanodiamond powder. Powders that were previously treatedin an ozone process show clean surfaces with a minimalgraphitic (non-sp3) surface shell.

Many nanodiamond particles show the presence ofcoherent twin defects in HRTEM images due to the fast out-of-equilibrium synthesis conditions [30]. The improvedresolution offered by aberration-corrected microscopy, incombination with imaging at lower acceleration voltagesbelow the C knock-on damage threshold allows the particlesto be imaged over prolonged periods, without graphitizationof the nanodiamond surface, as evidenced by the data inFig. 2a, b. Owing to this added surface stability, any detectednon-sp3 material arises from the material itself and is not animaging artifact. HRTEM imaging confirms the presence ofmainly {111}-type surface facets in the DND material, withsome {100} type truncation present. No significant amountsof {110} surface planes are present, confirming that thehigh-energy, fast detonation synthesis favors formation ofrapid growth {111} and to a lesser extent {100} surfaces,as well as low-energy defects like S¼ 3 {111} twinboundaries [30].

Even though HRTEM imaging can be used to visualizethe surface structure of the nanodiamond particles, thepresence of nitrogen or a small amount of non-sp3

coordinated carbon at e.g., coherent boundaries cannoteasily be imaged. The presence of non-sp3 coordinatedcarbon was confirmed through three distinct C K-edge pre-peak EELS signatures in Fig. 3. Image simulations withsingle nitrogen atoms substituting for carbon yielded nodiscernable image contrast changes, meaning that detectingnitrogen by only imaging is not practically viable. The sameis true for a distortion of the perfect sp3 coordination atthe coherent interfaces. Therefore, spatially resolved EELSwas used not only to locate the nitrogen impurities in theDND particles, but also to determine the local nitrogenenvironment. As the carbon K-edge and nitrogen K-edgeare acquired simultaneously in the experiments, the localcoordination of the carbon atoms in the diamond particlescan be mapped in the form of sp3 and non-sp3 carbon maps.Combining the sp3, non-sp3 and nitrogen maps (Fig. 4)precisely shows the nitrogen embedding. The nitrogen signalclearly originates from the diamond regions (high sp3

signal), confirming our earlier conclusions that nitrogen isembedded in the nanodiamond “core”. It is important to notethat even in regions where no large amounts of non-sp3

carbon are detected from HRTEM images (e.g., at the

coherent twin boundaries), a clear non-sp3 signal ismeasured through the presence of a C–K edge pre-peak at285 eV in EELS. This pre-peak must then largely originatefrom dangling bonds, fullerene-like carbon and distorted sp3

sites that are present at the nanodiamond surface orboundaries. Interestingly, an enrichment of nitrogen at themultiply twinned region is also observed, in good agreementwith previous measurements [10].

Finally, we performed detailed DFT calculations todetermine the nature of the nitrogen embedded in thenanodiamond cores. Previous measurements as well ascomparison of the nitrogen ELNES in heavily N-dopednanodiamond with references had already pointed in thedirection of tetrahedral embedding of the nitrogen atomsinto the diamond lattice [13, 14]. In Fig. 5, DFT ELNEScalculations appear to rule out the presence of significantamounts of NV0 or NV� centers in the diamond cores ofas-synthesized nanodiamonds. The results from singlesubstitutional nitrogen and the A-center (two neighboringembedded nitrogen atoms) do provide an interestinginsight into the predominant nitrogen embedding. The Ns

spectrum clearly shows that the pre-peak present in mostexperimental spectra can easily arise from bond elongationdue to the nitrogen embedding, and does therefore nothave to be the result of e.g., nitrogen at the surface ofthe nanodiamond particles. Overall, the A-center spectrumbest resembles the experimental data, even though peak Ais not present in the calculated spectrum. Peak A waspresent in A-center simulations for natural diamond byBrydson et al. [41] based upon multiple scattering theory.The discrepancy is thought to arise from the energyminimization/structure relaxation steps which were per-formed in our simulations, but not in the work byBrydson et al., or possibly due to artificial surface effectsthat arise from the multiple scattering approach. However,as pointed out in a previous publication, A-centers can bedetected by optical methods or by EPR in the form ofthe Aþ center and are not measured in large amounts inas-synthesized DND [10]. It is to be expected that theELNES signature for a cluster of 3, 4 or more tetrahedrallycoordinated nitrogen atoms would be highly similar to theA-center spectrum. This type of “larger” nitrogen clustersneed not be optically active, and could very well entaila large degree of local bond elongation, giving rise tothe pre-peak A, as was the case for the Ns spectrum.Unfortunately, calculating spectra for larger structurescontaining 3 or 4 nitrogen atoms is computationallyexpensive. Multiple scattering simulations should be ableto compute the spectrum for larger nitrogen clusters, andcould be performed in future work.

To summarize, a large amount (up to �4 at.%) ofnitrogen in these samples is embedded in the nitrogen core,mostly in a tetrahedral coordination. The embedding is likelyto be a mix of some small amounts of single substitutionaland/or A-center nitrogen atoms, combined with larger 3,4and more atom clusters. The nitrogen is enriched at defectiveboundaries, and in some cases at coherent boundaries [10].



ph

ysic

a ssp stat

us

solid

i a

5 Conclusions Aberration-corrected electron micros-copy allowed us to solve several remaining key questionsabout the (surface) structure, morphology and the presenceof nitrogen impurities in DND.We were able to image clean,stable nanodiamond facets at true atomic resolution for thefirst time – evidencing that clean nanodiamond facets arefeasible, and that the core–shell model for nanodiamondis not necessarily representative for materials that haveundergone sufficiently harsh surface cleaning treatments.Core-loss EELS spectrum imaging experiments prove thepresence of nitrogen in DND particles, spread out withinthe “core” of the particles and in some cases enriched atdefective regions as well as at twin boundaries, with amaximal nitrogen content around 4 at.%. DFT calculationsof the Ns, A, NV

0, and NV� center N–K edge ELNESsignatures were performed. By comparing the fine structureof the experimental nitrogen K-edge with the calculatedspectra, the embedded nitrogen is shown to likely be a mixsome small amounts of single substitutional and/or A-centernitrogen, combined with larger nitrogen clusters.

Acknowledgements The authors acknowledge helpful dis-cussions with Dr. I. Vlasov. The authors acknowledge support fromthe EuropeanUnion under the Framework 7 program under a contractfrom an Integrated Infrastructure Initiative (Reference 262348ESMI). GVT acknowledges the ERC grant COUNTATOMS.S. T. gratefully acknowledges financial support from the Fund forScientific Research Flanders (FWO). The microscope used in thisstudy was partially financed by the Hercules Foundation. D. L andF. D. P. gratefully acknowledge financial support from the GOAproject “XANES meets ELNES” of the research fund of theUniversity of Antwerp, Belgium. Computing time was provided bythe University of Antwerp (calcUA), Vlaams Supercomputercentrum(VSC) and the Hercules foundation.

References

[1] M. Baidakova and A. Vuĺ, J. Phys. D 40, 6300 (2007).[2] D. Pech, M. Brunet, H. Durou, P. Huang, V. Mochalin, Y.

Gogotsi, P.-L. Taberna, and P. Simon, Nature Nano 5, 651(2010).

[3] Y. Z. Jin, Y. J. Kim, C. Gao, Y. Q. Zhu, A. Huczko, M. Endo,and H. W. Kroto, Carbon 44, 724 (2006).

[4] I. Kovalenko, D. G. Bucknall, and G. Yushin, Adv. Funct.Mater. 20, 3979 (2010).

[5] O. A. Shenderova and D. M. Gruen, UltrananocrystallineDiamond, second ed. (William Andrew Publishing, Oxford,2012).

[6] B. R. Smith, M. Niebert, T. Plakhotnik, and A. V. Zvyagin, J.Lumin. 127, 260 (2007).

[7] O. A. Shenderova, I. I. Vlasov, S. Turner, G. Van Tendeloo, S.B. Orlinskii, A. A. Shiryaev, A. A. Khomich, S. N. Sulyanov,F. Jelezko, and J. Wrachtrup, J. Phys. Chem. C 115, 14014(2011).

[8] C. Bradac, T. Gaebel, N. Naidoo, M. J. Sellars, J. Twamley,L. J. Brown, A. S. Barnard, T. Plakhotnik, A. V. Zvyagin, andJ. R. Rabeau, Nature Nano 5, 345 (2010).

[9] B. R. Smith, D. W. Inglis, B. Sandnes, J. R. Rabeau, A. V.Zvyagin, D. Gruber, C. J. Noble, R. Vogel, E. Osawa, and T.Plakhotnik, Small 5, 1649 (2009).

[10] I. I. Vlasov, O. Shenderova, S. Turner, O. I. Lebedev, A. A.Basov, I. Sildos, M. Rähn, A. A. Shiryaev, and G. VanTendeloo, Small 6, 687 (2010).

[11] S. B. Orlinskii, R. S. Bogomolov, A. M. Kiyamova, B. V.Yavkin, G. M. Mamin, S. Turner, G. Van Tendeloo, A.Shiryaev, I. I. Vlasov, and O. Shenderova, Nanosci. Nano-technol. Lett. 3, 63 (2011).

[12] A. A. Soltamova, I. V. Iĺin, F. M. Shakhov, S. V. Kidalov,A. Y. Vuĺ, B. V. Yavkin, G. V. Mamin, S. B. Orlinskii, andP. G. Baranov, JETP Lett. 92, 102 (2010).

[13] S. Turner, O. I. Lebedev, O. Shenderova, I. I. Vlasov, J.Verbeeck, and G. Van Tendeloo, Adv. Funct. Mater. 19, 2116(2009).

[14] V. Pichot, O. Stephan, M. Comet, E. Fousson, J. Mory, K.March, and D. Spitzer, J. Phys. Chem. C 114, 10082(2010).

[15] S. Talapatra, P. G. Ganesan, T. Kim, R. Vajtai, M. Huang, M.Shima, G. Ramanath, D. Srivastava, S. C. Deevi, and P. M.Ajayan, Phys. Rev. Lett. 95, 097201 (2005).

[16] J. L. Peng, S. Bulcock, P. I. Belobrov, and L. A. Bursill, Int. J.Mod. Phys. B 15, 4071 (2001).

[17] A. Krüger, F. Kataoka, M. Ozawa, T. Fujino, Y. Suzuki, A. E.Aleksenskii, A. Y. Vuĺ, and E. �Osawa, Carbon 43, 1722(2005).

[18] K. Iakoubovskii, K. Mitsuishi, and K. Furuya, Nanotechnol-ogy 19, 155705 (2008).

[19] G. Cunningham, A. M. Panich, A. I. Shames, I. Petrov, and O.Shenderova, Diam. Relat. Mater. 17, 650 (2008).

[20] O. Shenderova, A. Koscheev, N. Zaripov, I. Petrov, Y.Skryabin, P. Detkov, S. Turner, and G. Van Tendeloo, J. Phys.Chem. C 115, 9827 (2011).

[21] M. S. Rhonda, F. C. Matthew, R. H. Philipp, M. O. D. A.Conel, and R. N. Larry, Astrophys. J. Lett. 738, L27(2011).

[22] N. Alem, R. Erni, C. Kisielowski, M. D. Rossell, W. Gannett,and A. Zettl, Phys. Rev. B 80, 155425 (2009).

[23] Ç. Ö. Girit, J. C. Meyer, R. Erni, M. D. Rossell, C.Kisielowski, L. Yang, C.-H. Park, M. F. Crommie, M. L.Cohen, S. G. Louie, and A. Zettl, Science 323, 1705(2009).

[24] A. V. Krasheninnikov and K. Nordlund, J. Appl. Phys. 107,071301 (2010).

[25] M. Nelhiebel, P. H. Louf, P. Schattschneider, P. Blaha, K.Schwarz, and B. Jouffrey, Phys. Rev. B 59, 12807 (1999).

[26] K. Schwarz, P. Blaha, and G. K. H. Madsen, Comp. Phys.Commun. 147, 71 (2002).

[27] S. A. Kajihara, A. Antonelli, J. Bernholc, and R. Car, Phys.Rev. Lett. 66, 2010 (1991).

[28] B. Wen, J. J. Zhao, and T. J. Li, Int. Mater. Rev. 52, 131(2007).

[29] Y. Min, M. Akbulut, K. Kristiansen, Y. Golan, and J.Israelachvili, Nature Mater. 7, 527 (2008).

[30] T. L. Daulton, D. D. Eisenhour, T. J. Bernatowicz, R. S.Lewis, and P. R. Buseck, Geochim. Cosmochim. Acta 60,4853 (1996).

[31] Z. L. Wang (ed.), Characterization of Nanophase Materials(Wiley-VCH Verlag GmbH, Weinheim, Berlin, 2001).

[32] R. Brydson, in: Nanocharacterisation (The Royal Society ofChemistry, Cambridge, 2007), p. 94.

[33] R. F. Egerton (ed.), Electron Energy-Loss Spectroscopy in theElectron Microscope (Springer, New York, Dordrecht,Heidelberg, London, 2011).



Invited

Article

[34] J. Birrell, J. E. Gerbi, O. Auciello, J. M. Gibson, D. M. Gruen,and J. A. Carlisle, J. Appl. Phys. 93, 5606 (2003).

[35] A. J. L. Garvie, Meteorit. Planet. Sci. 41, 667 (2006).[36] J.-Y. Raty, G. Galli, C. Bostedt, T. W. van Buuren, and L. J.

Terminello, Phys. Rev. Lett. 90, 037401 (2003).[37] F. Cleri, P. Keblinski, L. Colombo, D. Wolf, and S. R.

Phillpot, Europhys. Lett. 46, 671 (1999).[38] I. I. Vlasov, S. Turner, G. Van, Tendeloo, and A. A. Shiryaev,

in: Ultananocrystalline Diamond, 2nd ed. (William AndrewPubl., Oxford, 2012), p. 291.

[39] R. Erni, B. Freitag, P. Hartel, H. Müller, P. Tiemeijer, M. VanDer Stam, M. Stekelenburg, D. Hubert, P. Specht, andV. Garibay-Febles, Microsc. Microanal. 12, 492 (2006).

[40] P. G. Baranov, A. A. Soltamova, D. O. Tolmachev, N. G.Romanov, R. A. Babunts, F. M. Shakhov, S. V. Kidalov, A.Y. Vul’, G. V. Mamin, S. B. Orlinskii, and N. I. Silkin, Small7, 1533 (2011).

[41] R. Brydson, L. M. Brown, and J. Bruley, J. Microsc. 189, 137(1998).



ph

ysic

a ssp stat

us

solid

i a

pss standard-reprint webversion - uantwerpenematweb.cmi.ua.ac.be/emat/pdf/1965.pdf · nitrogen...

Documents