A Δ-self-consistent-field study of the nitrogen 1s binding energies in carbon nitrides

  • Published on

  • View

  • Download

Embed Size (px)


  • A -self-consistent-field study of the nitrogen 1s binding energies in carbon nitridessa Johansson and Sven Stafstrm Citation: The Journal of Chemical Physics 111, 3203 (1999); doi: 10.1063/1.479662 View online: http://dx.doi.org/10.1063/1.479662 View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/111/7?ver=pdfcov Published by the AIP Publishing Articles you may be interested in A tiered approach to Monte Carlo sampling with self-consistent field potentials J. Chem. Phys. 135, 184107 (2011); 10.1063/1.3660224 Self-consistent-field calculations of core excited states J. Chem. Phys. 130, 124308 (2009); 10.1063/1.3092928 The convergence of complete active space self-consistent-field energies to the complete basis set limit J. Chem. Phys. 123, 074111 (2005); 10.1063/1.1999630 Self-consistent field, ab initio molecular orbital and three-dimensional reference interaction site model study forsolvation effect on carbon monoxide in aqueous solution J. Chem. Phys. 112, 9463 (2000); 10.1063/1.481564 Core ionization energies of carbonnitrogen molecules and solids J. Chem. Phys. 111, 9678 (1999); 10.1063/1.480300

    This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 09 Dec 2014 10:04:38


  • A D-self-consistent-field study of the nitrogen 1 s binding energiesin carbon nitrides

    sa Johanssona) and Sven Stafstromb)Department of Physics and Measurement Technology, IFM, Linkoping University,S-581 83 Linkoping, Sweden

    ~Received 14 December 1998; accepted 24 May 1999!

    Binding energies of the N 1s level in hard and elastic CNx films are investigated by means oftheoretical studies of model molecules. Results for the model systems are obtained fromab initioHartreeFock calculations, where the core electron binding energies are determined using the deltaself-consistent-field method. The theoretical results are compared with experimental x-rayphotoelectron spectroscopy data in order to understand the microstructure of the CNx films. Bothsp2 and sp3 bonded nitrogen have been identified. The presence of nitrogen bonded to graphiteedges is also likely. 1999 American Institute of Physics.@S0021-9606~99!30931-4#


    Carbon based materials have been shown to exist in anumber of forms in addition to the well known graphite anddiamond phases. The most intensively studied examples ofthese novel forms are fullerenes and carbon nanotubes. Ma-terials such as amorphous carbon, hydrogenated amorphouscarbon and nitrogenated carbon have attracted a great deal ofattention recently since they have shown very interesting me-chanical properties. In particular, it was shown that hardnessand elasticity can be combined in nonstoichiometric CNxthin films prepared from sputtering of a graphite target in anitrogen atmosphere.1 High resolution transmission electronmicrographs~HRTEM! of this material show a graphitelikematerial with a large content of curved graphite layers andinterlayer crossings. The elasticity and hardness of the mate-rial could be rationalized as originating from the curvedgraphite layers and the crosslinkings, respectively.

    Even though the overall structural properties of the CNxmaterial can be observed in HRTEM studies, we still lack theknowledge about the nature of the chemical bonds in thematerial and the role played by nitrogen for the structure ofthe material. Useful information concerning the differentchemical environments to a particular type of atom can beobtained from x-ray photoelectron spectroscopy~XPS!. In-dependent XPS studies of the N 1s core level report twomajor peaks at 398.1398.4 eV and 400.3400.7 eV.26 Thesmall difference between these data could be attributed todifferent growing temperatures and different nitrogen con-tent in the films. The interpretation of these peaks are inaccord, the low binding energy peak originates from nitrogenin an environment ofsp3 carbons and the high binding en-ergy peak corresponds to nitrogen in asp2 carbon environ-ment. Here the chemical shift is defined as the difference inbinding energy of two such peaks. In general, binding ener-gies depend upon both initial and final state effects. Initialstate effects include the charge on the atom from which the

    core electron is ejected and the charge on the other atoms,while final state effects result from the electron redistributionafter ionization. Thus the chemical shift reflects differencesin these effects due to differences in the chemical surround-ing of the particular atom. In the case of thesp2 and sp3

    peaks discussed above, the chemical shift is partially due todifferent charges on the nitrogens in the two systems. Theshift may also depend on the fact that thesp3 system is moreclosely packed which gives rise to a stronger relaxation andconsequently a lower binding energy in that case.

    In two recent reports, high resolution N 1s XPS spectraof the same CNx material as introduced in Ref. 1 werepresented.5,6 In addition to the two N 1s peaks discussedabove, this spectrum showed a high binding energy peak at402.6 eV. This peak has also been observed before2 in asimilar type of material and was in that case assigned tonitrogen bound to oxygen. In recent experiments, however,very little oxygen is present in the sample and it might bethat other chemical environments to nitrogen also contributeto the peak in this energy region.6 Another new peak is ob-served at 399.0 eV. This feature is much weaker than thepreviously reportedsp2 and sp3 peaks but becomes moredominating in the spectrum as the take off angle of the pho-toelectrons was changed from normal exit to almost grazingdirections. This is a clear indication that the origin of thispeak can be associated with nitrogen present at the surface ofthe material. Likewise, the low binding energy peak corre-sponding to nitrogen in asp3 carbon environment is hardlypresent at grazing take off angles. Thus, thesp3 content isvery low at the surface, instead there are other types of bondsformed between carbon and nitrogen.

    As a complement to the XPS results, Raman spectros-copy studies have also been performed on the samples dis-cussed in the previous paragraph.5 These data show clearindications of C[N triple bonds. Thus, it is important tolocate the contribution from this type of nitrogen in the XPSspectrum. Together with the somewhat unclear interpretationof the high binding energy peak in the N 1s spectrum, this

    a!Electronic mail: asajo@ifm.liu.seb!Electronic mail: sst@ifm.liu.se


    32030021-9606/99/111(7)/3203/6/$15.00 1999 American Institute of Physics

    This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 09 Dec 2014 10:04:38

  • calls for a systematic study of the chemical shifts of nitrogenin various carbon environments.

    In this article we present results from calculations of theN 1s binding energies. Based on these results we can per-form a direct analysis of the XPS results discussed above, inparticular the origin of the N 1s peaks at 399.0 and 402.6 eV,by direct comparison of the experimental and calculatedchemical shifts. The nitrogen atoms were studied in carbonenvironments of different chemical nature and different size.The calculations were performed at theab initio HartreeFock level of theory and using the delta self-consistent-field(D-SCF! technique7 to obtain the chemical shifts betweenthe different systems. The methodology is presented brieflyin Sec. II followed by a presentation and analysis of theresults in Sec. III.


    The nitrogen containing clusters~molecules! that wehave studied are shown in Figs. 14 below. The systems aregrouped together according to the chemical nature of thecarbon atom~s! attached to nitrogen, i.e.,sp3, sp2 and sp1

    type of carbon~s!. For the graphitelike systems (sp2), wehave also studied the binding energies of nitrogen located atthe edge of the cluster. The varying size of the clusters ineach group were introduced to obtain information about thesize dependence of the binding energies. Indeed, this is a bigeffect and it must be stressed that the calculations were per-

    formed on model systems. However, as shown below thebasis set used allows for accurate calculations on fairly largemolecules and calculated chemical shifts are in close agree-ment with known experimental data.

    The idea of studying a series of structures of varying sizeis twofold. First, by careful extrapolation of the results forthe finite systems we can obtain a good estimate of the N 1sbinding energies in the type of solids we are interested in.Second, the evolution of the binding energies with increasingsize of the system provides a very clear separation of initialand final state contributions to the chemical shifts.8 Thisseparation allows us to identify how one or the other of thesetwo contributions is affected by the different chemical envi-ronment.

    The calculations were performed at theab initioHartreeFock level, using a split valence basis set of con-tracted Gaussians~6-31G!.9 To test the influence of the basisset, some calculations were also performed using a largerbasis set~see below!.

    The DSCF method7 is very well known and has alreadysince the 1970s been an effective approach to the determina-tion for XPS binding energies. The core electron bindingenergy is defined as the energy difference between theground state of the neutral system and the core hole state ofthe ionic system. In theD-SCF method, both the initial andfinal states are determined as SCF wave functions


    (n) #. ~1!

    For CSCF(n21) , i.e., then21 electron wave function of the

    final state, the electrons are attracted by the core hole. Theattraction energy as well as the relaxation of the final state,i.e., screening of the core hole, must be included to get rea-

    FIG. 1. Structures with nitrogen bonded tosp3 hybridized carbon~left! andsp2 hybridized carbon~right!.

    FIG. 2. Structures with nitrogen bonded to two carbons~nitrogen bonded tosp2 hybridized carbon!.

    3204 J. Chem. Phys., Vol. 111, No. 7, 15 August 1999 . Johansson and S. Stafstrom

    This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: On: Tue, 09 Dec 2014 10:04:38

  • sonable values for the binding energies. TheD-SCF methodtakes this into account in contrast to calculations based uponKoopmans theorem where only the initial state contributionsare considered.

    Due to the use of relatively small model systems of theactual material and since the calculations were performedusing a fairly small basis set, the values of the binding ener-gies cannot be compared directly with the experimental bind-ing energies. However, since the discussion below is based

    on differences in binding energies, the errors introduced byperforming calculations with small basis sets and on modelsystems are systematic and thus cancel to a large extent.

    As a test of the accuracy of the model we have per-formed calculations of binding energies of small systems forwhich it is possible to extend the basis set and for whichexperimental data are available. In Table I are shown the N1s binding energies of pyrrole and pyridine. In the first twocolumns are the values obtained in this study with two dif-ferent basis sets, but with the same geometry. The third andfourths columns show experimental and calculated valuesfrom other peoples work. The calculated N 1s chemicalshift between these two systems are in very close agreementwith the experimental value for all types of calculations. Inparticular, it seems as if the 6-31G basis set is particularlygood, a result that justifies the use of small basis sets incalculations of chemical shifts. Consequently, accurateD-SCF calculations of fairly large molecules can be per-formed, which is the approach taken in the present study.


    Focusing first on the two major features in the N 1sspectrum, namely the peaks assigned to nitrogen bound tosp3 and sp2 hybridized carbon with experimental bindingenergies of 398.1 and 400.6 eV, respectively.6 Since thebinding energy is strongly size dependent~see below!, it isimportant to consider systems of similar size in the calcula-tion of the difference in the 1s binding energies of these twonitrogens. The model system for nitrogen in asp3 carbonenvironment is N~C~CH3)3)3 , see Fig. 1~a!. We obtain abinding energy of 405.2 eV for the N 1s level in this system.For nitrogen bound tosp2 hybridized carbon we use themodel system NC15H10, see Fig. 1~b!. The N 1s bindingenergy in this system is 407.7 eV.~Note that the nitrogen inthis case is situated in the interior of a graphitelike systemand is threefold coordinated tosp2 carbon atoms.! Thesemodel systems are of the same size and represent the correctnearest and second nearest neighbor environment to nitro-gen. Using the binding energy of the nitrogen in an environ-ment ofsp3 carbons as reference, we obtain a shift of 2.5 eVtowards higher binding energies for nitrogen in asp2 envi-ronment. The corresponding experimental shift given in Ref.6 is 1.9 eV for samples prepared at 100 C and 2.5 eV forsamples prepared at 350 500C. Thus, our calculatedchemical shift is in close comparison for this type of struc-ture and gives full support for the assignments of the twomajor features in the N 1s spectrum.

    TABLE I. N(1s) binding energies~eV! for pyridine, Fig. 2~a!, and pyrrole,Fig. 2~a!, DSCF calculated with different basis sets.


    Binding energy

    6-31G/6-31G 6-31G/TZP Exp. Papera

    pyridine 406.81 404.21 404.94b 405.09pyrrole 407.96 405.68 406.15c 406.26

    aSee Ref. 11.bSee...