Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

C 1s Peak of Adventitious Carbon Aligns to the Vacuum Level: Dire Consequences for Material's Bonding Assignment by Photoelectron Spectroscopy

C 1s Peak of Adventitious Carbon Aligns to the Vacuum Level: Dire Consequences for Material's... DOI:10.1002/cphc.201700126 Communications C1sPeak of Adventitious Carbon Aligns to the Vacuum Level:Dire Consequences for Material’sBonding Assignment by Photoelectron Spectroscopy [a] Grzegorz Greczynski* and Lars Hultman The C1ssignal from ubiquitous carbon contamination on sam- dition is not fulfilled, the surface acquirespositive potential, ples formingduring air exposure, so called adventitious carbon which decreases the kinetic energy of escaping photoelec- (AdC) layers, is the most common binding energy (BE)refer- trons, and in consequence leads to the apparent shift of all ence in X-ray photoelectron spectroscopy studies. We demon- core level peaks towards higherBE; the phenomenon com- strate here, by using aseries of transition-metal nitride films monly referred to as charging. Since the specimen’s charging with differentAdC coverage, that the BE of the C1speak E state is not known apriori, the problem with correct BE refer- varies by as much as 1.44 eV.This is afactor of 10 more than encing arises for the vast majority of samples. The natural zero the typical resolvable difference between two chemical states of the BE scale exists only for specimens, in which the density of the same element, which makes BE referencingagainst the of states (DOS) exhibits awell-definedcut-off at the Fermi C1speak highly unreliable. Surprisingly,wefind that C1sshifts energy E,the so-called Fermi edge, as is the case for metals in correlate to changes in sample work function @ ,such that which high conductivity ensures Fermi level alignment be- SA the sum E þ @ is constant at 289.50: 0.15 eV,irrespective of tween the sample and the spectrometer.All other samples SA materials system andair exposure time, indicatingvacuum that lack an internal BE reference present aseriouschallenge, level alignment.This discovery allowsfor significantly better which is reflected by alarge spread of reportedBEvalues for [8] accuracyofchemical state determination than offered by the the same chemical state. Some examples include TiO with conventional methods. Our findings are not specific to nitrides the reported Ti 2p and O1speak positions varying from 3/2 and likely apply to all systemsinwhich chargetransfer at the 458.0 to 459.6 eV,and from 529.4 to 531.2 eV,respectively.In AdC/substrate interface is negligible. asimilar way, forSi N ,Si2pand N1speaks have been report- 3 4 ed at BE varyingfrom 100.6 to 102.1 eV,and 397.4 to 398.6 eV, [8] X-ray photoelectron spectroscopy (XPS) is an essential analyti- respectively. It is highly disturbingthat after more than 50 cal tool in surfacescience and materials research, providing in- years of development, the BE of constituting elements in many formation about surfacechemistryand composition. The first technologically relevant materials is accessed with an accuracy observation of chemical shifts between Cu atoms in metallic that is not better than the magnitude of typical large chemical [1] and oxidized state, followed by areport on aS 1s peak split shifts of the order of 1eV, much larger than the instrument [2] in the photoelectron spectrum of sulfur atoms in thiosulfate, resolution at 0.1 eV (or less), whichmakes the bonding assign- shortly after,carbon atoms in 1,2,4,5-benzenetetracarboxylic ment ambiguous,often leading to an arbitrary spectra inter- [3] acid, and the whole range of N-containing organic mole- pretation and contradicting results. [4] cules, laid grounds for chemical analysis by electron spectros- The situation is worsened by the fact that the use of anatu- [5,6] copy (ESCA). The unambiguous bondingassignment relies, ral BE reference such as the Fermi edge in the case of conduct- however, on the correctmeasurement of binding energy (BE) ing samples, is not at all common.This is again reflected by values. This is often anontrivial task because of the lack of an the spread of reportedBEvalues, not as large as for insulators, [7] internal BE reference. During the XPS experiments the nega- yet significant enough to often prevent correct bonding as- tive chargecontinuously removed from the surfaceregion as signment.For example, in the case of transition-metal (TM) ni- aresult of aphotoelectric effect has to be replenished with trides,whichexhibit pronouncedDOS at the Fermi level and, asufficiently high rate to preserve charge neutrality.Ifthis con- hence,metallic-like conductivity,reported BE values forcore level signals often differ by more than 1eV; the Ti 2p core 3/2 level of TiNvaries from 454.77 to 455.8 eV,whereas the posi- [a] Dr.G. Greczynski,Prof.L.Hultman Thin Film Physics Division, Department of Physics (IFM) tion of the N1speak changes by 0.9 eV for TiN, and 1.2 eV for Linkçping University ZrN, MoN, and NbN. SE-581 83 Linkçping(Sweden) It has become acommon procedure to use the C1ssignal E-mail:grzgr@ifm.liu.se from the adventitious carbon (AdC) layer presentonthe vast Supporting Information and the ORCID identification number(s) for the author(s) of this article can be found under: majority of surfaces following air exposure, asa BE reference. http://dx.doi.org/10.1002/cphc.201700126. To calibrate the BE scale the C@C/C@Hpeak of AdC is deliber- T 2017 The Authors. Published by Wiley-VCH Verlag GmbH &Co. KGaA. ately set at 284.0–285.2 eV and all core-level spectra are This is an openaccessarticleunder the termsofthe Creative Commons [9] alignedaccordingly. The method wasfirst proposed by Sieg- Attribution-NonCommercial License,which permits use, distribution and [6] bahn et al. in the early days of XPS applications and was orig- reproduction in any medium,provided the original work is properly cited, andisnot used for commercialpurposes. inally based on the observation that the AdC layer is present ChemPhysChem 2017, 18,1507 –1512 1507 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications on all air-exposed surfaces with the C1sline as it appeared 289.50@@ .Weshow that this approachresults in aconsidera- SA constantat285.0 eV,which made it an ideal candidate for BE bly better accuracyofchemical state determination as com- [10] referencing. Soon after,however,the claim was dropped, as pared with the status quo. it becameclear that the C1sBEinpractice varies with the The ubiquitous nature of AdC hasbeen analyzed in detail by [18] thickness of the hydrocarbon layer by as much as 0.6 eV for Pd Barr et al., who concluded that it predominantly consists of [11, 12] and Au substrates. In the review of existing literature pub- polymerichydrocarbon species (C@C/C@H), together with lished in 1982, Swift concluded that “although the use of C1s aminor component (10–30% of the total signalintensity) due electrons from adventitious carbon layers is often aconvenient to carbooxides containing C@O@Cand O@C=Obonds. Indeed, methodofenergy referencing, interpretationofbinding aset of C1score level spectra acquired from (TM)N surfaces in [9] energy data obtained should be treated with caution”. In the the as-received state (see Figure 1) reveals that carbon is pres- following years, problems with using the C1speak for BE refer- ent in several chemical states almost on every surface ana- encing accumulated. For example, Werrett et al. reported in- lyzed. In all cases, however,the spectra are dominated by the consistentresults when referencing to C1sofAdC during stud- ies of oxidized Al-Si alloys, which was due to the oxidation of [13] AdC, whereas Gross et al. showed that the Au4f signal from gold particles deliberately deposited on amorphous SiO pro- [14] vides more reliable BE reference than C1s. More recent ex- amplesindicate that the issue of correct referencing of XPS [15, 16] spectra remains unresolved, which contrasts with the fact that the method based on adventitiouscarbon is widely adopted. Our literature review shows that in 58 of the first 100 top- cited papers dealing with XPS studies of magnetron sputtered films published between 2010 and 2016 in peer-reviewed jour- [17] nals, C1sofAdC was used as aBEreference, whereas, alarm- ingly,the remaining papers lack information about any refer- encing methodused. Withinthe first group, the C1speak was set quite arbitraryatthe BE, varying from 284.0 to 285.2 eV (here we disregard two extreme cases of 283.0 and 298.8 eV). This seriousinconsistency easily accounts for the large spread of reported BE values for the same chemical species (see exam- ples above),and contradicts the notionofthe BE reference, which per definition should be associated with one single energy value (as was originally intended in ref. [6]). Here, we examine the reliability of using AdC for XPS BE ref- erencing by measuring the position of the C1speak for aseries of TM nitride thin-film layers that exhibit awell-defined Figure 1. C1sXPS spectraofadventitious carbon obtained from as-received Fermi edge cut-off serving as an internal BE reference. Meas- air-exposed (ca. 10 min.) polycrystalline (TM)N thin films, where TM= Mo, V, W, Ti,Cr, Nb, Ta,Zr, and Hf, grownbymagnetron sputtering on Si(001) sub- urements are performed as afunction of the AdC layer thick- strates. ness, which scales with the air exposure time. We show that the BE of the C1speak of AdC measured with respect to the Fermi edge E varies by as much as 1.44 eV,from 284.08eVin aliphatic carbon C@C/C@Hpeak, whereas C@O@Cand O@C=O the case of MoN to 285.52 eV for aHfN sample. This is afactor contributions presentathigherBEappear in much lower con- of 10 more than the typical resolvable difference between two centrations. Clearly,there is asubstantial change in the C1s chemicalstates of the same element, which makes the energy spectra appearance depending on the (TM)N studied. Not only referencing against the C1speak of AdC highly unreliable. do the number of component peaks change (e.g.,noO@C=O Moreover,wedemonstrate that the position of the C1speak peak is observed in the present case of WN andMoN), but, of AdC closely follows changes in sample work function @ , more importantly,the BE of the dominant C@C/C@Hpeak, SA assessed here by ultraviolet photoelectronspectroscopy (UPS), measured with respecttothe Fermi level of the spectrometer F F in such away that the sum E þ @ is essentially constant at E ,exhibits large variation:from 284.08 eV in the case of MoN SA B B 289.50: 0.15 eV,which corresponds welltothe gas-phase BE surfaceto285.52 eV for HfN, as summarized in Table 1. The value of longeralkaneslowered by the intermolecular relaxa- 1.44 eV change in the position of the C1speak is certainly dis- tion energy.This indicates that C1saligns to the vacuum level turbing, as one would clearlyexpect the BE of carbonspecies E ,and implies that its BE is steered by the sample work present in the same chemical state to be independent of the VAC function. Clearly,the C1sofAdC cannot be used for reliable BE underlying substrate, especiallyifused for referencingXPS referencing of XPS spectra in aconventionalway,unless acom- spectra. plementary measurement of @ is performed and C1sisset at SA ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1508 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications shown in Figure 3(a). Interestingly,even though Table 1. Binding energies relativetoFermi level E for all component peaks in C1s B F there is acertain variation in C1s E for each materi- spectratogether with work functionvalues @ obtained from polycrystalline (TM)N SA als system (rather random and not exceeding 0.5 eV), thin films in the as-received state, where TM= Mo, V, W, Ti,Cr, Nb,Ta, Zr,and Hf. alargespread in BE of the C1speak observed for (TM)N C1sBErelative to Fermi level, E DBE DBE Work function samples in the as-received state persists for layers [eV] C @C C @C [eV] C@O C@C O@C=O C@C that weresputter-cleaned and subsequentlyexposed C@C/C@HC@O O@C=O to ambient atmosphere for time periods varying TiN284.52 286.24 289.06 1.72 4.54 4.90 from 10 minutes to 7months. Thus, we can conclude VN 284.15 285.96 288.51 1.81 4.36 5.16 CrN 284.60 286.14 288.56 1.54 3.96 4.83 that the template dependence of C1sBEtakes place ZrN 285.49 287.21 289.54 1.72 4.05 4.09 irrespectiveofthe amount of accumulatedadventi- NbN 284.76 286.52 289.18 1.76 4.42 4.65 tious carbon. We note also that changes in the BE’s MoN 284.08 285.76 –1.68 – 5.35 of the intrinsic core level signals (metal and nitrogen HfN285.52 287.17 289.75 1.65 4.23 4.00 TaN285.08 286.75 289.39 1.67 4.31 4.41 peaks) during prolonged air exposure are lower than WN 284.22 285.73 –1.71 – 5.23 0.1 eV. To address the issue of C1sshifts, we first obtain areliable evaluation of the charging state of the The C1scontribution due to C@Oalso shifts from sample to actual (TM)N film. To do this, we record DOS in the vicinity of sample, from 285.76 eV for as-received MoN to 287.21 for ZrN the Fermi edge (Fermilevel cut-off). Electrons close to E pos- (1.45 eV difference), essentiallyfollowing the C@C/C@Hpeak, sess the highest kinetic energy of all excited photoelectrons so that the relative BE difference D(C @C )isnearly con- (essentially equal to hn @ @ ), which results in relatively long C@O C@C SA [20] stant at 1.70: 0.13 eV (cf. Table 1). The BE of the O@C=Opeak mean free path l,from to 18 to 24 a. In consequence, the does not follow the shifts observed for all other C1scontribu- tions, which is best seen by comparingthe C1sspectra record- ed from TiNand CrN surfaces, see Figure 1, and varies from 288.51 eV for VN to 289.75 eV for HfN (1.24 eV difference). Some C1sspectra (TiN, ZrN) possessalso an extra contribution at significantly lower BE (282.0–282.5 eV), which is assigned to [19] carbideformation during film growth, and as such is of minor importance for this work. The amount of AdC that accumulates on the surface of (TM)N films exhibits asteady increase with the air exposure time, as illustrated in Figure 2, in which surface Cconcentra- tions are plottedfor all nitride samples in the time span from 10 minutes to 7months. Even though the percentage amount of AdC varies somewhat between different samples, the accu- mulationrate is essentially the same and amountstoca. 5at% per decade. The corresponding evolution of E of the domi- nant C@C/C@HC 1s peak of AdC with air exposure time is Figure 3. a) Binding energy of the C@C/C@Hpeak in the C1sspectra of ad- ventitious Creferenced to Fermi level E ,b)work function obtained by UPS from the secondaryelectron cut-off @ ,and c) C1sBEreferencedto SA Vacuum level E for aset of polycrystalline (TM)N thin films, where VAC Figure 2. Surface carbon concentrationsplottedasafunction of air expo- TM= Mo, V, W, Ti,Cr, Nb,Ta, Zr,and Hf, grown by magnetronsputtering on sure time for polycrystalline (TM)N thin films, where TM= Mo, V, W, Ti,Cr, Si(001) substrates. The dashed curves in (a) and (b) are only for eye guiding Nb, Ta,Zr, and Hf, grown by magnetron sputtering on Si(001) substrates. to emphasize the symmetry between the plots. ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1509 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications Figure 4. The portionofthe valence bandspectra in the close vicinity of the Fermi level E indicating the Fermi level cut-offfor as-receivedpolycrystalline (TM)N thin films,where TM= Mo, V, W, Ti,Cr, Nb, Ta,Zr, and Hf, grownonSi(001)substrates:a)asmeasured (referencingto E ), and b) aligned by usingthe commonprocedure of referencing to C1speak of adventitious carbon set at 284.5 eV. XPS probingdepth, given by 3V l,well exceeds the thickness To gain more insight into the energylevel alignment at the of the AdC layer,which is in the range 4.5–9 a.This, together AdC/(TM)N interface,weperformmeasurements of sample with the fact that adventitious carbon being awide band gap work function @ in the same instrument;that is, without SA [18] materialdoes not possess DOS near E , implies that the spec- breaking the vacuum. As summarized in Figure 3(b), in which tral intensity in this region is solely determined by the TM(N). sample work functionisplotted for all TM(N) layers in the Figure 4(a) shows the Fermi level cut-off for all TM(N) samples order of an increasing TM mass, andfor variousamounts of air in the as-received state, as measured.Inall cases, the rapid exposure time, @ exhibits large apparent variations, that in SA drop in DOS coincides with “0” of the BE scale, which is indica- the case of as-received samples range from 4.00 eV for HfN to tive of aFermi level alignment between sample andthe spec- 5.35 eV for MoN. More importantly,a direct comparison to the trometer. This proves that agood electrical contact is estab- E values shown in Figure 3(a) reveals that the trend in work lished to the instrumentand excludes any possibility of charg- functionclosely correlates to that observed for the C1speak ing in the (TM)N layer. of AdC, such that the sum E þ @ remains constantfor all SA The fact that C1sshifts (cf. Figure 1) while the Fermi edge samples, irrespective of air exposure time at 289.50: 0.15 eV from the underlying (TM)N film appears at “0” eV (Figure 4(a)) (see Figure3 (c)). This implies that C1saligns to the vacuum [21, 22] clearly indicates decoupling of the measured energy levels of level E , rather than to the Fermi level, as is implicitly as- VAC adventitiouscarbon from the Fermi level of the underlying sumed when using this peak for BE referencing. Hence,the po- substrate and, hence,spectrometer.The implications for BE ref- sition of the C1speak measuredwith respect to E is steered erencing that employs the C1speak are severe. If,ascommon- by the substrate work function, which disqualifies this signal as ly practiced,one would align all recorded spectra by setting areliable reference, unless acomplementary measurement of the C@C/C@Hpeak of AdC at 284.5 eV,the highest portion of @ is performed and spectra are alignedtoC 1s set at SA the valence band spectra recordedfrom (TM)N appearsas 289.50@@ eV.The positionofthe C1sC@C/C@Hpeak refer- SA shown in Figure4 (b). Contrafactory,somespecimens (TiN, VN) enced to E ,289.50 eV,corresponds very well with the gas- VAC would exhibit no DOS at E despite their metalliccharacter, phase value of 290.15 eV measured for longer alkanes by Pir- [23] whereas for other films (HfN, ZrN, and TaN) such calibration of eaux et al., compensated for the intermolecular relaxation the BE scale resultsina non-zeroDOS above the Fermi level. energy due to electronic and atomic polarization of the neigh- These examples demonstrate that the common procedure of boringmolecules surrounding the core hole, which is typically [24] referencing to the C1slevel set at the arbitrary chosen BE of the order of 1–3 eV. value within the range, 284.0–285.2 eV,isnot justified because The vacuum level alignment is characteristic of aweakinter- it leads to unphysical results. The latter is not realized if deal- action at the interface to the substrate and is regularly ob- ing with core level spectra,inwhich case shifts in peak posi- servedfor organic films deposited on metals by using ex-situ tions by :1eVdonot lead to such clear contradictions. techniques (e.g. spin coating)inthe absence of both charge [25] transfer across the interface and interface dipole formation. ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1510 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications Figure 5. Schematic illustration of the energy level alignment at the interface between adventitious carbonlayer and a) the low work function substrate, and b) the high work function substrate. For all tested samples the sum of E and @ is constant,which is indicativeofvacuum level alignment. SA Such contacts remainwithin the Schottky–Mott limit, in which E þ @ is constantat289.50: 0.15 eV,indicating alignment SA the electronic levels of the adsorbate are determined by the to the vacuum level. Thus, the position of the C1speak from [26] work function of the substrate. As amatter of fact, the pro- AdC layer is decoupled from the instrument Fermi level and is [18] cess of AdC adsorption is also classified as physical, because steeredbythe sample work function, and as such cannot be the principal species(hydrocarbons) are not chemically reac- used for reliable BE referencing of XPS spectra.Apossible tive and can be readily desorbed by agentle anneal in remedy here is acomplementary measurement of @ and ref- SA [27] vacuum. In the presentcase, the potentialinteraction be- erencing to C1sset at 289.50@@ ,which, as we demonstrate, SA tween AdC speciesand (TM)N film is further suppressed by the yields consistent results for the whole series of TM nitrides, ir- presenceofa native oxide layer. respective of air exposure time. Conclusions from this work are Our findings are schematically summarized in Figure 5, in not limited to nitridesand likely apply to all substrates that ex- which the relevant energy levels and criticalparameters are in- hibit weak interaction towards AdC. dicatedfor (a) alow work function sample, and (b) ahigh work functionsample. Independent of @ we find that the Fermi SA Acknowledgements level cut-off of (TM)N aligns with that of the spectrometer (which is established duringthe calibration procedure), where- F The authors most gratefully acknowledge the financial supportof as the BE of C1sfrom adventitiouscarbon E closely follows F the VINN Excellence Center Functional Nanoscale Materials the changes in @ .Since DE ffi D@ ,the position of the C1s SA SA F (FunMat) Grant 2005-02666,the Swedish GovernmentStrategic peak with respect to the vacuum level, E þ @ ,remains con- SA Research AreainMaterials Science on Functional Materials at stant at 289.50: 0.15 eV.This agrees with acommon-sense Linkçping University (FacultyGrant SFO-Mat-LiU 2009-00971), notion of constant energy levels associated with Catoms pres- the Knut and Alice Wallenberg Foundation Grant 2011.0143, and ent in the same chemical environment and provides grounds the aforsk Foundation Grant 16-359. for more reliable referencing of the XPS spectra. In conclusion, we established by using aseries of TM nitride thin-film layers covered with afew monolayers of adventitious Conflict of interest carbon (AdC), that the BE of the C1speak of AdC measured with respecttothe Fermi edge E depends on the substrate, The authors declare no conflict of interest. and varies from 284.08 eV for MoN to 285.52 eV for the HfN sample in the as-received state. This wide spread in C1speak Keywords: analytical methods · binding energy · surface positionisindependent of the time samples are exposed to analysis · surface chemistry · X-ray photoelectron spectroscopy ambient atmosphere, hence of the AdC layer thickness. This disturbing result shows that the commonly used referencing [1] E. Sokolowski, C.Nordling, K. Siegbahn, Phys. Rev. 1958, 110,776. of XPS spectra against the C1speak of AdC is unreliable. More- [2] S. Hagstrçm, C. Nordling, K. Siegbahn, Z. Phys. 1964, 178,439. over,wedemonstrate that the C1ssignal closely follows the [3] G. Axelson,U.Ericson, A. Fahlman,K.Hamrin, J. Hedman, R. Nordberg, variation of sample work function @ ,such that the sum C. Nordling,K.Siegbahn, Nature 1967, 213,70. SA ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1511 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications [4] R. Nordberg, R. G. Albridge, T. Bergmark, U. Ericson, A. Fahlman,K. [16] M. Jacquemin, M. J. Genet, E. M. Gaigneaux,D.P.Debecker, ChemPhys- Hamrin, J. Hedman, G. Johansson, C. Nordling, K. Siegbahn, B. Lindberg, Chem 2013, 14,3618. [17] Accordingtothe Scopus Database searchfor “XPS” and“magnetron Nature 1967, 214,481. [5] A. Fahlman,K.Hamrin, J. Hedman, R. Nordberg, C. Nordling, K. Sieg- sputtering” articles published during 2010–2016,asof2016-11-24. bahn, Nature 1966, 210,4. [18] T. L. Barr, S. Seal, J. Vac. Sci. Technol. A 1995, 13,1239. [19] G. Greczynski, S. Mr#z, L. Hultman, J. M. Schneider, Appl. Surf. Sci. 2016, [6] K. Siegbahn, C. Nordling, A. Fahlman,R.Nordberg, K. Hamrin, J. 385,356. Hedman, G. Johansson, T. Bergmark, S.-E. Karlsson, I. Lindgren, B. Lind- [20] S. Tanuma, C. J. Powell, D. R. Penn, Surf. Interface Anal. 2011, 43,689. berg, ESCA—Atomic, Molecule and Solid State Structure Studied by Means [21] H. Ishii,E.Sugiyama, E. Ito, K. Seki, Adv.Mater. 1999, 11,605. of ElectronSpectroscopy,Almqvist &Wiksells Boktryckeri, Uppsala, [22] H. D. Hagstrum, Surf.Sci. 1976, 54,197. Sweden, 1967. [23] J. J. Pireaux, S. Svensson, E. Basilier,P.-g.Malmqvist, U. Gelius, R. Cauda- [7] J. B. Metson, Surf. Interface Anal. 1999, 27,1069. no, K. Siegbahn, Phys. Rev.A 1976, 14,2133. [8] NIST X-ray Photoelectron Spectroscopy Database, Version 4.1 (National [24] M. Lçgdlund, G. Greczynski,A.Crispin,T.Kugler,M.Fahlman,W.R.Sala- Institute of Standardsand Technology,Gaithersburg, 2012), http://srda- neck, Photoelectron Spectroscopy of Interfaces for Polymer-Based Electron- tA.nist.gov/xps/. Accessed: 2016-11-22. ic Devices, in Conjugated Polymer and Molecular Interfaces:Scienceand [9] P. Swift, Surf.Interface Anal. 1982, 4,47. Technology for Photonic and Optoelectronic Application (Eds.: W. R. Sala- [10] G. Johansson, J. Hedman, A. Berndtsson,M.Klasson, R. Nilsson, J. Elec- neck,K.Seki, A. Kahn, J. J. Pireaux), Marcel Dekker, New York, 2001. tron Spectrosc. Relat. Phenom. 1973, 2,295. [25] S. Braun, W. R. Salaneck,M.Fahlman, Adv. Mater. 2009, 21,1450. [11] S. Kohiki, K. Oki, J. Electron Spectrosc. Relat. Phenom. 1984, 33,375. [26] E. H. Rhoderick, R. H. Williams, Metal-Semiconductor Contacts,Clarendon [12] S. Kinoshita, T. Ohta, H. Kuroda, Bull. Chem. Soc. Jpn. 1976, 49,1149. Press, Oxford, 1988. [13] C. R. Werrett, A. K. Bhattacharya, D. R. Pyke, Appl. Surf. Sci. 1996, 103, [27] G. Greczynski, L. Hultman, Appl. Phys. Lett. 2016, 109,211602. [14] Th. Gross, M. Ramm, H. Sonntag,W.Unger,H.M.Weijers, E. H. Adem, Surf. InterfaceAnal. 1992, 18,59. Manuscript received:February 6, 2017 [15] A. P8lisson-Schecker,H.J.Hug, J. Patscheider, Surf. Interface Anal. 2012, Accepted manuscript online:March 10, 2017 44,29. Version of record online:April 11,2017 ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1512 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Chemphyschem Pubmed Central

C 1s Peak of Adventitious Carbon Aligns to the Vacuum Level: Dire Consequences for Material's Bonding Assignment by Photoelectron Spectroscopy

Chemphyschem , Volume 18 (12) – Apr 11, 2017

Loading next page...
 
/lp/pubmed-central/c-1s-peak-of-adventitious-carbon-aligns-to-the-vacuum-level-dire-2KzCOUN50c

References (24)

Publisher
Pubmed Central
Copyright
© 2017 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA.
ISSN
1439-4235
eISSN
1439-7641
DOI
10.1002/cphc.201700126
Publisher site
See Article on Publisher Site

Abstract

DOI:10.1002/cphc.201700126 Communications C1sPeak of Adventitious Carbon Aligns to the Vacuum Level:Dire Consequences for Material’sBonding Assignment by Photoelectron Spectroscopy [a] Grzegorz Greczynski* and Lars Hultman The C1ssignal from ubiquitous carbon contamination on sam- dition is not fulfilled, the surface acquirespositive potential, ples formingduring air exposure, so called adventitious carbon which decreases the kinetic energy of escaping photoelec- (AdC) layers, is the most common binding energy (BE)refer- trons, and in consequence leads to the apparent shift of all ence in X-ray photoelectron spectroscopy studies. We demon- core level peaks towards higherBE; the phenomenon com- strate here, by using aseries of transition-metal nitride films monly referred to as charging. Since the specimen’s charging with differentAdC coverage, that the BE of the C1speak E state is not known apriori, the problem with correct BE refer- varies by as much as 1.44 eV.This is afactor of 10 more than encing arises for the vast majority of samples. The natural zero the typical resolvable difference between two chemical states of the BE scale exists only for specimens, in which the density of the same element, which makes BE referencingagainst the of states (DOS) exhibits awell-definedcut-off at the Fermi C1speak highly unreliable. Surprisingly,wefind that C1sshifts energy E,the so-called Fermi edge, as is the case for metals in correlate to changes in sample work function @ ,such that which high conductivity ensures Fermi level alignment be- SA the sum E þ @ is constant at 289.50: 0.15 eV,irrespective of tween the sample and the spectrometer.All other samples SA materials system andair exposure time, indicatingvacuum that lack an internal BE reference present aseriouschallenge, level alignment.This discovery allowsfor significantly better which is reflected by alarge spread of reportedBEvalues for [8] accuracyofchemical state determination than offered by the the same chemical state. Some examples include TiO with conventional methods. Our findings are not specific to nitrides the reported Ti 2p and O1speak positions varying from 3/2 and likely apply to all systemsinwhich chargetransfer at the 458.0 to 459.6 eV,and from 529.4 to 531.2 eV,respectively.In AdC/substrate interface is negligible. asimilar way, forSi N ,Si2pand N1speaks have been report- 3 4 ed at BE varyingfrom 100.6 to 102.1 eV,and 397.4 to 398.6 eV, [8] X-ray photoelectron spectroscopy (XPS) is an essential analyti- respectively. It is highly disturbingthat after more than 50 cal tool in surfacescience and materials research, providing in- years of development, the BE of constituting elements in many formation about surfacechemistryand composition. The first technologically relevant materials is accessed with an accuracy observation of chemical shifts between Cu atoms in metallic that is not better than the magnitude of typical large chemical [1] and oxidized state, followed by areport on aS 1s peak split shifts of the order of 1eV, much larger than the instrument [2] in the photoelectron spectrum of sulfur atoms in thiosulfate, resolution at 0.1 eV (or less), whichmakes the bonding assign- shortly after,carbon atoms in 1,2,4,5-benzenetetracarboxylic ment ambiguous,often leading to an arbitrary spectra inter- [3] acid, and the whole range of N-containing organic mole- pretation and contradicting results. [4] cules, laid grounds for chemical analysis by electron spectros- The situation is worsened by the fact that the use of anatu- [5,6] copy (ESCA). The unambiguous bondingassignment relies, ral BE reference such as the Fermi edge in the case of conduct- however, on the correctmeasurement of binding energy (BE) ing samples, is not at all common.This is again reflected by values. This is often anontrivial task because of the lack of an the spread of reportedBEvalues, not as large as for insulators, [7] internal BE reference. During the XPS experiments the nega- yet significant enough to often prevent correct bonding as- tive chargecontinuously removed from the surfaceregion as signment.For example, in the case of transition-metal (TM) ni- aresult of aphotoelectric effect has to be replenished with trides,whichexhibit pronouncedDOS at the Fermi level and, asufficiently high rate to preserve charge neutrality.Ifthis con- hence,metallic-like conductivity,reported BE values forcore level signals often differ by more than 1eV; the Ti 2p core 3/2 level of TiNvaries from 454.77 to 455.8 eV,whereas the posi- [a] Dr.G. Greczynski,Prof.L.Hultman Thin Film Physics Division, Department of Physics (IFM) tion of the N1speak changes by 0.9 eV for TiN, and 1.2 eV for Linkçping University ZrN, MoN, and NbN. SE-581 83 Linkçping(Sweden) It has become acommon procedure to use the C1ssignal E-mail:grzgr@ifm.liu.se from the adventitious carbon (AdC) layer presentonthe vast Supporting Information and the ORCID identification number(s) for the author(s) of this article can be found under: majority of surfaces following air exposure, asa BE reference. http://dx.doi.org/10.1002/cphc.201700126. To calibrate the BE scale the C@C/C@Hpeak of AdC is deliber- T 2017 The Authors. Published by Wiley-VCH Verlag GmbH &Co. KGaA. ately set at 284.0–285.2 eV and all core-level spectra are This is an openaccessarticleunder the termsofthe Creative Commons [9] alignedaccordingly. The method wasfirst proposed by Sieg- Attribution-NonCommercial License,which permits use, distribution and [6] bahn et al. in the early days of XPS applications and was orig- reproduction in any medium,provided the original work is properly cited, andisnot used for commercialpurposes. inally based on the observation that the AdC layer is present ChemPhysChem 2017, 18,1507 –1512 1507 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications on all air-exposed surfaces with the C1sline as it appeared 289.50@@ .Weshow that this approachresults in aconsidera- SA constantat285.0 eV,which made it an ideal candidate for BE bly better accuracyofchemical state determination as com- [10] referencing. Soon after,however,the claim was dropped, as pared with the status quo. it becameclear that the C1sBEinpractice varies with the The ubiquitous nature of AdC hasbeen analyzed in detail by [18] thickness of the hydrocarbon layer by as much as 0.6 eV for Pd Barr et al., who concluded that it predominantly consists of [11, 12] and Au substrates. In the review of existing literature pub- polymerichydrocarbon species (C@C/C@H), together with lished in 1982, Swift concluded that “although the use of C1s aminor component (10–30% of the total signalintensity) due electrons from adventitious carbon layers is often aconvenient to carbooxides containing C@O@Cand O@C=Obonds. Indeed, methodofenergy referencing, interpretationofbinding aset of C1score level spectra acquired from (TM)N surfaces in [9] energy data obtained should be treated with caution”. In the the as-received state (see Figure 1) reveals that carbon is pres- following years, problems with using the C1speak for BE refer- ent in several chemical states almost on every surface ana- encing accumulated. For example, Werrett et al. reported in- lyzed. In all cases, however,the spectra are dominated by the consistentresults when referencing to C1sofAdC during stud- ies of oxidized Al-Si alloys, which was due to the oxidation of [13] AdC, whereas Gross et al. showed that the Au4f signal from gold particles deliberately deposited on amorphous SiO pro- [14] vides more reliable BE reference than C1s. More recent ex- amplesindicate that the issue of correct referencing of XPS [15, 16] spectra remains unresolved, which contrasts with the fact that the method based on adventitiouscarbon is widely adopted. Our literature review shows that in 58 of the first 100 top- cited papers dealing with XPS studies of magnetron sputtered films published between 2010 and 2016 in peer-reviewed jour- [17] nals, C1sofAdC was used as aBEreference, whereas, alarm- ingly,the remaining papers lack information about any refer- encing methodused. Withinthe first group, the C1speak was set quite arbitraryatthe BE, varying from 284.0 to 285.2 eV (here we disregard two extreme cases of 283.0 and 298.8 eV). This seriousinconsistency easily accounts for the large spread of reported BE values for the same chemical species (see exam- ples above),and contradicts the notionofthe BE reference, which per definition should be associated with one single energy value (as was originally intended in ref. [6]). Here, we examine the reliability of using AdC for XPS BE ref- erencing by measuring the position of the C1speak for aseries of TM nitride thin-film layers that exhibit awell-defined Figure 1. C1sXPS spectraofadventitious carbon obtained from as-received Fermi edge cut-off serving as an internal BE reference. Meas- air-exposed (ca. 10 min.) polycrystalline (TM)N thin films, where TM= Mo, V, W, Ti,Cr, Nb, Ta,Zr, and Hf, grownbymagnetron sputtering on Si(001) sub- urements are performed as afunction of the AdC layer thick- strates. ness, which scales with the air exposure time. We show that the BE of the C1speak of AdC measured with respect to the Fermi edge E varies by as much as 1.44 eV,from 284.08eVin aliphatic carbon C@C/C@Hpeak, whereas C@O@Cand O@C=O the case of MoN to 285.52 eV for aHfN sample. This is afactor contributions presentathigherBEappear in much lower con- of 10 more than the typical resolvable difference between two centrations. Clearly,there is asubstantial change in the C1s chemicalstates of the same element, which makes the energy spectra appearance depending on the (TM)N studied. Not only referencing against the C1speak of AdC highly unreliable. do the number of component peaks change (e.g.,noO@C=O Moreover,wedemonstrate that the position of the C1speak peak is observed in the present case of WN andMoN), but, of AdC closely follows changes in sample work function @ , more importantly,the BE of the dominant C@C/C@Hpeak, SA assessed here by ultraviolet photoelectronspectroscopy (UPS), measured with respecttothe Fermi level of the spectrometer F F in such away that the sum E þ @ is essentially constant at E ,exhibits large variation:from 284.08 eV in the case of MoN SA B B 289.50: 0.15 eV,which corresponds welltothe gas-phase BE surfaceto285.52 eV for HfN, as summarized in Table 1. The value of longeralkaneslowered by the intermolecular relaxa- 1.44 eV change in the position of the C1speak is certainly dis- tion energy.This indicates that C1saligns to the vacuum level turbing, as one would clearlyexpect the BE of carbonspecies E ,and implies that its BE is steered by the sample work present in the same chemical state to be independent of the VAC function. Clearly,the C1sofAdC cannot be used for reliable BE underlying substrate, especiallyifused for referencingXPS referencing of XPS spectra in aconventionalway,unless acom- spectra. plementary measurement of @ is performed and C1sisset at SA ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1508 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications shown in Figure 3(a). Interestingly,even though Table 1. Binding energies relativetoFermi level E for all component peaks in C1s B F there is acertain variation in C1s E for each materi- spectratogether with work functionvalues @ obtained from polycrystalline (TM)N SA als system (rather random and not exceeding 0.5 eV), thin films in the as-received state, where TM= Mo, V, W, Ti,Cr, Nb,Ta, Zr,and Hf. alargespread in BE of the C1speak observed for (TM)N C1sBErelative to Fermi level, E DBE DBE Work function samples in the as-received state persists for layers [eV] C @C C @C [eV] C@O C@C O@C=O C@C that weresputter-cleaned and subsequentlyexposed C@C/C@HC@O O@C=O to ambient atmosphere for time periods varying TiN284.52 286.24 289.06 1.72 4.54 4.90 from 10 minutes to 7months. Thus, we can conclude VN 284.15 285.96 288.51 1.81 4.36 5.16 CrN 284.60 286.14 288.56 1.54 3.96 4.83 that the template dependence of C1sBEtakes place ZrN 285.49 287.21 289.54 1.72 4.05 4.09 irrespectiveofthe amount of accumulatedadventi- NbN 284.76 286.52 289.18 1.76 4.42 4.65 tious carbon. We note also that changes in the BE’s MoN 284.08 285.76 –1.68 – 5.35 of the intrinsic core level signals (metal and nitrogen HfN285.52 287.17 289.75 1.65 4.23 4.00 TaN285.08 286.75 289.39 1.67 4.31 4.41 peaks) during prolonged air exposure are lower than WN 284.22 285.73 –1.71 – 5.23 0.1 eV. To address the issue of C1sshifts, we first obtain areliable evaluation of the charging state of the The C1scontribution due to C@Oalso shifts from sample to actual (TM)N film. To do this, we record DOS in the vicinity of sample, from 285.76 eV for as-received MoN to 287.21 for ZrN the Fermi edge (Fermilevel cut-off). Electrons close to E pos- (1.45 eV difference), essentiallyfollowing the C@C/C@Hpeak, sess the highest kinetic energy of all excited photoelectrons so that the relative BE difference D(C @C )isnearly con- (essentially equal to hn @ @ ), which results in relatively long C@O C@C SA [20] stant at 1.70: 0.13 eV (cf. Table 1). The BE of the O@C=Opeak mean free path l,from to 18 to 24 a. In consequence, the does not follow the shifts observed for all other C1scontribu- tions, which is best seen by comparingthe C1sspectra record- ed from TiNand CrN surfaces, see Figure 1, and varies from 288.51 eV for VN to 289.75 eV for HfN (1.24 eV difference). Some C1sspectra (TiN, ZrN) possessalso an extra contribution at significantly lower BE (282.0–282.5 eV), which is assigned to [19] carbideformation during film growth, and as such is of minor importance for this work. The amount of AdC that accumulates on the surface of (TM)N films exhibits asteady increase with the air exposure time, as illustrated in Figure 2, in which surface Cconcentra- tions are plottedfor all nitride samples in the time span from 10 minutes to 7months. Even though the percentage amount of AdC varies somewhat between different samples, the accu- mulationrate is essentially the same and amountstoca. 5at% per decade. The corresponding evolution of E of the domi- nant C@C/C@HC 1s peak of AdC with air exposure time is Figure 3. a) Binding energy of the C@C/C@Hpeak in the C1sspectra of ad- ventitious Creferenced to Fermi level E ,b)work function obtained by UPS from the secondaryelectron cut-off @ ,and c) C1sBEreferencedto SA Vacuum level E for aset of polycrystalline (TM)N thin films, where VAC Figure 2. Surface carbon concentrationsplottedasafunction of air expo- TM= Mo, V, W, Ti,Cr, Nb,Ta, Zr,and Hf, grown by magnetronsputtering on sure time for polycrystalline (TM)N thin films, where TM= Mo, V, W, Ti,Cr, Si(001) substrates. The dashed curves in (a) and (b) are only for eye guiding Nb, Ta,Zr, and Hf, grown by magnetron sputtering on Si(001) substrates. to emphasize the symmetry between the plots. ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1509 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications Figure 4. The portionofthe valence bandspectra in the close vicinity of the Fermi level E indicating the Fermi level cut-offfor as-receivedpolycrystalline (TM)N thin films,where TM= Mo, V, W, Ti,Cr, Nb, Ta,Zr, and Hf, grownonSi(001)substrates:a)asmeasured (referencingto E ), and b) aligned by usingthe commonprocedure of referencing to C1speak of adventitious carbon set at 284.5 eV. XPS probingdepth, given by 3V l,well exceeds the thickness To gain more insight into the energylevel alignment at the of the AdC layer,which is in the range 4.5–9 a.This, together AdC/(TM)N interface,weperformmeasurements of sample with the fact that adventitious carbon being awide band gap work function @ in the same instrument;that is, without SA [18] materialdoes not possess DOS near E , implies that the spec- breaking the vacuum. As summarized in Figure 3(b), in which tral intensity in this region is solely determined by the TM(N). sample work functionisplotted for all TM(N) layers in the Figure 4(a) shows the Fermi level cut-off for all TM(N) samples order of an increasing TM mass, andfor variousamounts of air in the as-received state, as measured.Inall cases, the rapid exposure time, @ exhibits large apparent variations, that in SA drop in DOS coincides with “0” of the BE scale, which is indica- the case of as-received samples range from 4.00 eV for HfN to tive of aFermi level alignment between sample andthe spec- 5.35 eV for MoN. More importantly,a direct comparison to the trometer. This proves that agood electrical contact is estab- E values shown in Figure 3(a) reveals that the trend in work lished to the instrumentand excludes any possibility of charg- functionclosely correlates to that observed for the C1speak ing in the (TM)N layer. of AdC, such that the sum E þ @ remains constantfor all SA The fact that C1sshifts (cf. Figure 1) while the Fermi edge samples, irrespective of air exposure time at 289.50: 0.15 eV from the underlying (TM)N film appears at “0” eV (Figure 4(a)) (see Figure3 (c)). This implies that C1saligns to the vacuum [21, 22] clearly indicates decoupling of the measured energy levels of level E , rather than to the Fermi level, as is implicitly as- VAC adventitiouscarbon from the Fermi level of the underlying sumed when using this peak for BE referencing. Hence,the po- substrate and, hence,spectrometer.The implications for BE ref- sition of the C1speak measuredwith respect to E is steered erencing that employs the C1speak are severe. If,ascommon- by the substrate work function, which disqualifies this signal as ly practiced,one would align all recorded spectra by setting areliable reference, unless acomplementary measurement of the C@C/C@Hpeak of AdC at 284.5 eV,the highest portion of @ is performed and spectra are alignedtoC 1s set at SA the valence band spectra recordedfrom (TM)N appearsas 289.50@@ eV.The positionofthe C1sC@C/C@Hpeak refer- SA shown in Figure4 (b). Contrafactory,somespecimens (TiN, VN) enced to E ,289.50 eV,corresponds very well with the gas- VAC would exhibit no DOS at E despite their metalliccharacter, phase value of 290.15 eV measured for longer alkanes by Pir- [23] whereas for other films (HfN, ZrN, and TaN) such calibration of eaux et al., compensated for the intermolecular relaxation the BE scale resultsina non-zeroDOS above the Fermi level. energy due to electronic and atomic polarization of the neigh- These examples demonstrate that the common procedure of boringmolecules surrounding the core hole, which is typically [24] referencing to the C1slevel set at the arbitrary chosen BE of the order of 1–3 eV. value within the range, 284.0–285.2 eV,isnot justified because The vacuum level alignment is characteristic of aweakinter- it leads to unphysical results. The latter is not realized if deal- action at the interface to the substrate and is regularly ob- ing with core level spectra,inwhich case shifts in peak posi- servedfor organic films deposited on metals by using ex-situ tions by :1eVdonot lead to such clear contradictions. techniques (e.g. spin coating)inthe absence of both charge [25] transfer across the interface and interface dipole formation. ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1510 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications Figure 5. Schematic illustration of the energy level alignment at the interface between adventitious carbonlayer and a) the low work function substrate, and b) the high work function substrate. For all tested samples the sum of E and @ is constant,which is indicativeofvacuum level alignment. SA Such contacts remainwithin the Schottky–Mott limit, in which E þ @ is constantat289.50: 0.15 eV,indicating alignment SA the electronic levels of the adsorbate are determined by the to the vacuum level. Thus, the position of the C1speak from [26] work function of the substrate. As amatter of fact, the pro- AdC layer is decoupled from the instrument Fermi level and is [18] cess of AdC adsorption is also classified as physical, because steeredbythe sample work function, and as such cannot be the principal species(hydrocarbons) are not chemically reac- used for reliable BE referencing of XPS spectra.Apossible tive and can be readily desorbed by agentle anneal in remedy here is acomplementary measurement of @ and ref- SA [27] vacuum. In the presentcase, the potentialinteraction be- erencing to C1sset at 289.50@@ ,which, as we demonstrate, SA tween AdC speciesand (TM)N film is further suppressed by the yields consistent results for the whole series of TM nitrides, ir- presenceofa native oxide layer. respective of air exposure time. Conclusions from this work are Our findings are schematically summarized in Figure 5, in not limited to nitridesand likely apply to all substrates that ex- which the relevant energy levels and criticalparameters are in- hibit weak interaction towards AdC. dicatedfor (a) alow work function sample, and (b) ahigh work functionsample. Independent of @ we find that the Fermi SA Acknowledgements level cut-off of (TM)N aligns with that of the spectrometer (which is established duringthe calibration procedure), where- F The authors most gratefully acknowledge the financial supportof as the BE of C1sfrom adventitiouscarbon E closely follows F the VINN Excellence Center Functional Nanoscale Materials the changes in @ .Since DE ffi D@ ,the position of the C1s SA SA F (FunMat) Grant 2005-02666,the Swedish GovernmentStrategic peak with respect to the vacuum level, E þ @ ,remains con- SA Research AreainMaterials Science on Functional Materials at stant at 289.50: 0.15 eV.This agrees with acommon-sense Linkçping University (FacultyGrant SFO-Mat-LiU 2009-00971), notion of constant energy levels associated with Catoms pres- the Knut and Alice Wallenberg Foundation Grant 2011.0143, and ent in the same chemical environment and provides grounds the aforsk Foundation Grant 16-359. for more reliable referencing of the XPS spectra. In conclusion, we established by using aseries of TM nitride thin-film layers covered with afew monolayers of adventitious Conflict of interest carbon (AdC), that the BE of the C1speak of AdC measured with respecttothe Fermi edge E depends on the substrate, The authors declare no conflict of interest. and varies from 284.08 eV for MoN to 285.52 eV for the HfN sample in the as-received state. This wide spread in C1speak Keywords: analytical methods · binding energy · surface positionisindependent of the time samples are exposed to analysis · surface chemistry · X-ray photoelectron spectroscopy ambient atmosphere, hence of the AdC layer thickness. This disturbing result shows that the commonly used referencing [1] E. Sokolowski, C.Nordling, K. Siegbahn, Phys. Rev. 1958, 110,776. of XPS spectra against the C1speak of AdC is unreliable. More- [2] S. Hagstrçm, C. Nordling, K. Siegbahn, Z. Phys. 1964, 178,439. over,wedemonstrate that the C1ssignal closely follows the [3] G. Axelson,U.Ericson, A. Fahlman,K.Hamrin, J. Hedman, R. Nordberg, variation of sample work function @ ,such that the sum C. Nordling,K.Siegbahn, Nature 1967, 213,70. SA ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1511 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications [4] R. Nordberg, R. G. Albridge, T. Bergmark, U. Ericson, A. Fahlman,K. [16] M. Jacquemin, M. J. Genet, E. M. Gaigneaux,D.P.Debecker, ChemPhys- Hamrin, J. Hedman, G. Johansson, C. Nordling, K. Siegbahn, B. Lindberg, Chem 2013, 14,3618. [17] Accordingtothe Scopus Database searchfor “XPS” and“magnetron Nature 1967, 214,481. [5] A. Fahlman,K.Hamrin, J. Hedman, R. Nordberg, C. Nordling, K. Sieg- sputtering” articles published during 2010–2016,asof2016-11-24. bahn, Nature 1966, 210,4. [18] T. L. Barr, S. Seal, J. Vac. Sci. Technol. A 1995, 13,1239. [19] G. Greczynski, S. Mr#z, L. Hultman, J. M. Schneider, Appl. Surf. Sci. 2016, [6] K. Siegbahn, C. Nordling, A. Fahlman,R.Nordberg, K. Hamrin, J. 385,356. Hedman, G. Johansson, T. Bergmark, S.-E. Karlsson, I. Lindgren, B. Lind- [20] S. Tanuma, C. J. Powell, D. R. Penn, Surf. Interface Anal. 2011, 43,689. berg, ESCA—Atomic, Molecule and Solid State Structure Studied by Means [21] H. Ishii,E.Sugiyama, E. Ito, K. Seki, Adv.Mater. 1999, 11,605. of ElectronSpectroscopy,Almqvist &Wiksells Boktryckeri, Uppsala, [22] H. D. Hagstrum, Surf.Sci. 1976, 54,197. Sweden, 1967. [23] J. J. Pireaux, S. Svensson, E. Basilier,P.-g.Malmqvist, U. Gelius, R. Cauda- [7] J. B. Metson, Surf. Interface Anal. 1999, 27,1069. no, K. Siegbahn, Phys. Rev.A 1976, 14,2133. [8] NIST X-ray Photoelectron Spectroscopy Database, Version 4.1 (National [24] M. Lçgdlund, G. Greczynski,A.Crispin,T.Kugler,M.Fahlman,W.R.Sala- Institute of Standardsand Technology,Gaithersburg, 2012), http://srda- neck, Photoelectron Spectroscopy of Interfaces for Polymer-Based Electron- tA.nist.gov/xps/. Accessed: 2016-11-22. ic Devices, in Conjugated Polymer and Molecular Interfaces:Scienceand [9] P. Swift, Surf.Interface Anal. 1982, 4,47. Technology for Photonic and Optoelectronic Application (Eds.: W. R. Sala- [10] G. Johansson, J. Hedman, A. Berndtsson,M.Klasson, R. Nilsson, J. Elec- neck,K.Seki, A. Kahn, J. J. Pireaux), Marcel Dekker, New York, 2001. tron Spectrosc. Relat. Phenom. 1973, 2,295. [25] S. Braun, W. R. Salaneck,M.Fahlman, Adv. Mater. 2009, 21,1450. [11] S. Kohiki, K. Oki, J. Electron Spectrosc. Relat. Phenom. 1984, 33,375. [26] E. H. Rhoderick, R. H. Williams, Metal-Semiconductor Contacts,Clarendon [12] S. Kinoshita, T. Ohta, H. Kuroda, Bull. Chem. Soc. Jpn. 1976, 49,1149. Press, Oxford, 1988. [13] C. R. Werrett, A. K. Bhattacharya, D. R. Pyke, Appl. Surf. Sci. 1996, 103, [27] G. Greczynski, L. Hultman, Appl. Phys. Lett. 2016, 109,211602. [14] Th. Gross, M. Ramm, H. Sonntag,W.Unger,H.M.Weijers, E. H. Adem, Surf. InterfaceAnal. 1992, 18,59. Manuscript received:February 6, 2017 [15] A. P8lisson-Schecker,H.J.Hug, J. Patscheider, Surf. Interface Anal. 2012, Accepted manuscript online:March 10, 2017 44,29. Version of record online:April 11,2017 ChemPhysChem 2017, 18,1507 –1512 www.chemphyschem.org 1512 T 2017 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim

Journal

ChemphyschemPubmed Central

Published: Apr 11, 2017

There are no references for this article.