at SciVerse ScienceDirect

Vacuum 99 (2014) 211e215

Contents lists available

Vacuum

journal homepage: www.elsevier .com/locate/vacuum

Investigation of thickness dependence on electronic structures of ironand nickel thin films by L-edge X-ray absorption spectroscopy

Guvenc Akgul a,*, Funda Aksoy Akgul b,**, Yuksel Ufuktepe c

aBor Vocational School, Nigde University, 51700 Nigde, Turkeyb Physics Department, Nigde University, 51240 Nigde, Turkeyc Physics Department, Cukurova University, 01330 Adana, Turkey

a r t i c l e i n f o

Article history:Received 11 March 2013Received in revised form30 May 2013Accepted 2 June 2013

Keywords:MetalsThin filmsXANESElectronic structure

* Corresponding author. Tel.: þ90 (388) 3114527; f** Corresponding author. Tel.: þ90 (388) 2254217; f

E-mail addresses: guvencakgul@gmail.com (Ggmail.com (F.A. Akgul).

0042-207X/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.vacuum.2013.06.003

a b s t r a c t

We have studied the effect of the film thickness on the electronic structure of pure nickel and iron thinfilms. Series of the thin films were evaporated by e-beam evaporation on SiN substrates. The electronicstructure of the thin films was investigated using X-ray absorption near edge structure (XANES) spec-troscopy. We have showed the thickness dependent variation of the experimental branching ratio (BR)and full-width at half-maximum (FWHM) at the L3 and L2 edges for both thin films. A strong thicknessdependence of the L2,3 BR and FWHM was found. We have also focused on the deviation of L3 to L2 ratiofrom its statistical value. The average L3/L2 white-line intensity ratio was calculated to be 3.4 and 3.0 frompeak height and integrated area under each L3 and L2 peaks, respectively for iron. Additionally, a theo-retical L2,3 edge calculation for nickel was presented. The obtained results were consistent with thegeneral view of the L2,3 BR and FWHM of iron and nickel transition metals.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

The X-ray absorption near edge structure (XANES) has proved tobe a powerful technique to study thin film materials because of itschemical and elemental sensitivity and the possibility to deriveground state properties of a material. It corresponds to the elec-tronic transition from a core level to an unoccupied or hybridizedorbital and mainly reflects the unoccupied electronic states [1e3].In an X-ray absorption process, a primary core level electron isstimulated by X-ray photon and moved into an excited state with acore hole leaving behind. The system can relax via filling of the corehole by an electron from an upper energy level. This process resultsin either the radiationless emission of an Auger electron or in theemission of a fluorescence photon.

For 3d transition metals (TM), L2,3 edges of the XANES arecharacterized generally by two peaks, called as “white-lines”. Theseabsorption lines indicate from the excitations of 2p core electronsto unoccupied d states. Unique information can be obtained byexciting the 2p electrons into the unoccupied 3d valence orbitals orinto the continuum. The intensity of the white-lines denotes the

ax: þ90 (388) 3118437.ax: þ90 (388) 2250180.. Akgul), fundaaksoy01@

All rights reserved.

unoccupied 3d density of state (DOS) and the d states occupationsof 3d transition metals which have been studied systematically inthe literature [4e9]. Due to the dipole allowed transition (2p-3d), Ledge absorption spectrum has several advantages compared to Kedge spectrum of TM elements and compounds. L2 and L3 edges of3d transition elements are dominated by the large Coulomb inter-action between the 2p core hole and 3d electrons.

In the case of K edge of transition metals (1s-3d), the pre-edgepeak is mainly due to dipoleeforbidden transition from a coreelectron to 3d and leading to a weak signal in absorption spectrum[10]. In the last several years, renewed interest in X-ray absorptionspectroscopy (XAS) has demanded a better understanding of thetotal electron yield (TEY) measurements of 3d transition metals[3,6,7,11,12]. Especially, thin film structures which contain iron andnickel layers have received particular attention owing to theirtechnological importance, herein, it is crucial to determine theirelectronic structure depending on increasing the film thickness[13e15].

Branching ratio (BR) at the L2 and L3 edges is one of theimportant parameters for transition metals [16e22]. BR gives ususeful information on the electronic structure of 3d metal that re-lates to chemical bonding and characteristics [23]. The BR isstrongly influenced by electrostatic interaction (between d electronand valence hole) and spin-orbit splitting. In particular, the BR canbe used as identification the spin state of 3d transition metals [17].

Fig. 1. Normalized total electron yield (TEY) spectra of nickel films with different filmthicknesses. In each spectrum, the TEY signal before the onset of the L3 edge at 845 eVwas set to zero. The inset shows theoretical calculation for the nickel thin filmwith thethickness of 8 �A. The black dashed line demonstrates the two step-like functions forbackground subtraction of nickel film with 12 �A.

G. Akgul et al. / Vacuum 99 (2014) 211e215212

But, experimentally obtained L2,3 BR is normally not enough to giveunambiguous information about the valence band spin-orbitsplitting, the electrostatic interactions between core-hole andvalence-electron, or the spin state. The BR is ranged from 0.63 to0.70, with an average of 0.67 for low spin state and ranged from0.71 to 0.76 with an average of 0.74 for high spin state [18].Although many experiment has been performed to determine theL2,3 branching ratio but, it is not completely trivial to obtainexperimental values for this parameter [24e31]. Our aim in thiswork is to accurately determine thickness dependence of the ironand nickel L2,3 branching ratio intensity and full-width at half-maximum (FWHM) change in L2 and L3 region. We hope that ob-tained results can be one of the key requirements for the devel-opment of TM based applications.

2. Experimental methods

The experiments were performed on beamlines 8.2 and 10.1 ofthe Stanford Synchrotron Radiation Lightsource (Menlo Park, CA-USA). The beamline 8.2 is a bending-magnet beamline dedicatedto photoemission, photoabsorption spectroscopy, and XANES ex-periments within an energy range of 100e1300 eV. The beamline10.1 is a wiggler-magnet beamline and primarily used for X-rayabsorption and photoemission spectroscopies within an energyrange of 250e1200 eV. The L edge spectra of iron and nickel thinfilms were measured in TEY mode. The TEY signal in 3d transitionmetals is generated by secondary electrons following X-ray photonabsorption. The TEY spectra were collected at room temperature.The films were grown by e-beam evaporation on SiN coated Siwafers. The thicknesses of the nickel and iron layers were variedfrom 12 �A to 120 �A and from 3 �A to 112 �A respectively. The basepressure of the chamber was 1�10�10 Torr and below 1�10�9 Torrduring the evaporation. The X-ray incidence angle was controlledby a rotatable manipulator with an accuracy of about 0.5�. The filmthicknesses were determined bymonitoring the growth rate with aquartz crystal thin film monitor (Inficon). The substrates weremounted on an aluminum sample holders through a thin copperwire connected. To eliminate the effect of the substrate from theabsorption spectrum, we measured the photo-absorption of theclean SiN substrate itself immediately prior to the film deposition.In order to record the TEY spectra, the TEY signal was monitored bythe sample drain current. This was simply done by measuring thecurrent from a copper wire connected to the sample holder, whichhas been electrically isolated from the chamber. Simultaneously,the incident photon flux I0 was determined by an 80% transmissiveAu net. To avoid synchrotron dependent intensity, all TEY signalswere normalized to I0. The currents were recorded with a Keithley427 current amplifier.

Theoretical XANES amplitude for nickel L2,3 edges was obtainedby FEFF8 simulation. The FEFF8 program that is a self-consistentreal space multiple-scattering code for simultaneous calculationsof X-ray absorption spectra including polarization dependence,core hole effects and local field corrections was used to calculate L3and L2 edges. Calculations are based on all electron real spacerelativistic Green’s function formalism with no symmetry re-quirements [32].

3. Results and discussion

Fig. 1 demonstrates the evolution of the nickel L2,3 absorptionedges with varying thickness from 12 �A to 120 �A in TEY mode. Ineach spectrum, the TEY signal before the onset of the L3 edge at845 eV was set to zero, and the L3 intensity of the thickest nickelfilm was normalized to one. The main features in the spectrum aretwomaxima at 852.6 eV and 870.2 eV corresponding to the edges of

L3 and L2, respectively. The L3 and L2 lines correspond to transitionsinto nickel 3d states that produce 2p3/2 and 2p1/2 core holes. Thus,these two peaks can be explained by the electrostatic interactionbetween core-hole and valence electrons and the initial state spin-orbit splitting. The energy separation of the two peaks is 17.6 eVdue to the spin-orbit interaction of the nickel 2p core level. We canalso indicate that the each peak intensity is proportional the un-occupied states in d band.

From Fig. 1, we can determine the white-line intensity ratio (L3/L2) by measuring integrated areas under each corresponding peak.Generally, the L3 and L2 edges XANES spectra include excitationsarising from the 2p-3d transitions with contribution from the 2pstate to continuum states. To calculate L3 to L2 ratio from a preciseestimation of the peak areas, it is necessary to subtract a back-ground component from the excitation spectra to eliminate thetransitions into continuum states. Two step-like functions alignedat the maxima of the L3 and L2 edges with relative heights of 2:1(the expected intensity ratio for transitions into the L3 and L2 edges)was used for the background subtraction of the XANES spectra.Same method of background subtraction was used in the literatureby Dhesi et al. [12] for nickel films on Co/Cu (001) and by Chen et al.[33] for Fe and Co thin films. The two step-like function used for thenickel thin film of 12�A is shown in Fig.1. The boundary between theintegrated areas of the L3 and L2 sum spectra was at the just belowthe onset of the L2 edge, namely 868 eV. Additionally, a theoreticalcalculation for the nickel film with the thickness of 8 �A is alsoshown in the inset of Fig. 1 for the purpose of comparison betweenexperiment and theory. From the inset of Fig. 1, the intensity trendwith increasing film thickness confirms our theoretical calculationto be in good agreement with the experimental spectra.

Fig. 2 represents the normalized and background subtracted TEYspectra of the nickel thin films with varying thicknesses. It is clearfrom Fig. 2 that the intensity originated from the 2p3/2 to 3d tran-sition return to zero between the L3 and L2 edges. Therefore, there isa certainty in the choice of backgroundwhich can effects thewhite-line intensity ratio. After the background subtraction, we calculatedthe white-line intensity ratio (L3/L2) for each nickel thin films. Theobtained average value of L3/L2 ratio is 3.2 which is consistent withthe literature [25,27]. In fact, our previous results of total yield

Fig. 4. Branching ratio (BR) of the background subtracted intensities at the L2,3 edge asa function of iron film thickness. The drawn line is a guide to the eye. The inset showsthe BR versus inverse of the film thicknesses.

Fig. 2. Normalized and background subtracted total electron yield (TEY) spectra ofnickel films with different film thicknesses. The inset shows the intensity change of6 eV satellite feature as a function of nickel film thickness. The thicknesses of the nickelfilms in Angstrom (�A) units are given beside each curve.

G. Akgul et al. / Vacuum 99 (2014) 211e215 213

XANES on nickel L edge have also given the same white-line in-tensity ratio (L3/L2) for nickel thin films [34]. Same method of thebackground subtraction was used to calculate the white-line in-tensity ratio (L3/L2) for each thickness of the iron films as done forthe nickel films. The average value of L3/L2 ratio was calculated to be3.0 and 3.4 from integrated area and peak heights, respectivelywhich are consistent with the literature [29].

Between the L3 and L2 region, the XANES spectra have anotherpeak at 859 eV, which is due to the satellite at w6 eV above the L3peak. The 6 eV feature in the L edge spectrumwas explained by ourprevious results of total yield XANES on nickel [34]. The inset ofFig. 2 represents the evolution of the 6 eV satellite feature as afunction of the nickel film thickness. As the film thickness increasesa remarkable change in the intensity of the nickel L3 and L2 white-line peaks in Fig. 2 can be observed for the low thicknesses up to60 �A. The same behavior is noticed for the satellite feature indi-cating that a significant increase is followed by a deceleration in thegrowth rate of intensity [35]. This trend of the satellite peak with

Fig. 3. Branching ratio (BR) of the background subtracted intensities at the L2,3 edge asa function of nickel film thickness. The drawn line is only a guide to the eye. The insetshows the BR versus inverse of the film thicknesses.

increasing film thickness can be explained by noting that nickel 2psatellite peak intensity is associated with the weight of unoccupiedstates of 3d character near the EF Fermi energy and by its positionabove the EF level [36e38].

Fig. 3 shows the measured branching ratio (A(L3)/[A(L2) þA(L3)]) of the background subtracted integrated intensities at theL2,3 edge as a function of nickel film thickness. To estimate theexperimental L2,3 branching ratio intensity for nickel thin film, wehave used normalized L3 and L2 integrated peak areas [A(L)] underthe L3 and L2. The calculated area of the L3 and L2 lines would be thecorrect measure of their intensity which is proportional to the

Fig. 5. 2p spin-orbit splitting of nickel and iron with varying film thickness. Thedashed lines give the average standard value of spin-orbit splitting.

Fig. 6. Full-width at half-maximum at (a) L2 and (b) L3 edges in total electron yield as a function of nickel film thickness. The drawn lines are a guide to the eye.

Fig. 7. Full-width at half-maximum at (a) L2 and (b) L3 edges in total electron yield as a function of iron film thickness. The drawn lines are a guide to the eye.

G. Akgul et al. / Vacuum 99 (2014) 211e215214

oscillator strength of the 2p-3d spin orbit interaction. It can be seenfrom Fig. 3 that the BR changes significantly from 0.771 to 0.756 asthe film thickness increases from 12 �A to 120 �A. The theoretical BRwas calculated to be 0.73 in the literature by Thole et al. [31]. The BRof the nickel films below 60 �A shows a strong dependence on thefilm thickness. On the other hand, Fig. 3 also reveals that the BR isalmost same for thickness 90 �A and 120 �A and in agreement withthe value reported in the literature [19]. All these values of BRdetermined for different thickness of nickel thin film give a sup-portive evidence that the electronic structure of the nickel filmwiththe thickness above 60 �A does not significantly change.

We have already studied the TEY of iron thin films as a functionof film thickness and calculated X-ray attenuation length andelectron escape depth for iron [7,9]. As we have mentioned earlierBR gives information on the electronic structure of a material.Therefore, we have also demonstrated in this work how the BR ofiron changes depending on the film thickness. Fig. 4 shows themeasured BR of the background subtracted integrated intensities atthe L2,3 edge as a function of iron film thickness. It can be seen fromFig. 4 that the BR changes significantly from 0.765 to 0.715 as thefilm thickness increases from 3�A to 112�A. The thickness dependentvariation of the experimental L2,3 branching ratio in iron and nickelcan be explained by screening of the p-d electrostatic interactionwith increased delocalized valence electrons in thicker films. Infact, the branching ratio is a measure of the spin-orbit interaction inthe valence states. The variation of L2,3 branching ratio withincreasing thickness is also proportional to the strength of spin-orbit splitting.

The inset of Figs. 3 and 4 shows BR versus inverse of the filmthicknesses. The solid line in the figures is a linear curve fit to theexperimental data which assumes ideal layer by layer growth.Therefore, the growth mode of the iron and nickel films shows agood agreement with layer by layer model [39,40].

From TEY spectra, 2p spin-orbit splitting of the prepared sam-ples was measured as the peak to peak distance of L2 and L3 edgesas shown in Fig. 5. The 2p spin-orbit splitting is not significantlychanging with increasing thickness for nickel and iron. Thesefindings clearly indicate that the variation of the L2,3 branchingratio is rather originated by core-hole screening with increasedelectron delocalization in thicker films than by spin-orbit splitting.

To study the thickness dependence more quantitatively weexamined the FWHM at the L2 and L3 edges as a function of iron andnickel film thicknesses. Fig. 6 (a) and (b) shows the FWHM at the L2and L3 edges as a function of nickel coverage, respectively.

Fig. 7 (a) and (b) shows the FWHM at the L2 and L3 edges as afunction of iron coverage, respectively. It can be seen from the fig-ures the FWHMat the L edges decreases rapidlywith increasing ironand nickel film thickness. This decrease in the L2 and L3 FWHM foriron and nickel are most likely to be related to modifications in thesurface electronic states. The strong dependence of the BR of thethinner iron and nickel thin films on the film thickness supports thisidea. The measured L2 and L3 FWHM decreases from 2.8 to 2.6 and1.9 to 1.7 for nickel and from 3.8 to 2.3 and 2.7 to 1.9 for ironrespectivelydue to change in3dbandwith ironandnickel thickness.

4. Conclusions

Concluding the presented results, we have performed an X-rayabsorption experiment to confirm difference in experimentalbranching ratio and full-width at half-maximum behavior of ironand nickel thin films prepared by the e-beam evaporation. Thevalues of the BR and the FWHM determined for different thicknessof nickel and iron thin films provide a prominent evidence that theelectronic structure of the nickel and ironwith the thickness below60�A is significantly changing.We have calculated the average valueof L3/L2 ratio 3.0 and 3.4 from integrated peak area and peak heights

G. Akgul et al. / Vacuum 99 (2014) 211e215 215

respectively for iron using TEY method. Also, we have ascertainedthe thickness dependent variation of the experimental L2,3branching ratio and FWHM of iron and nickel. The 2p spin-orbitsplitting is almost constant for nickel and iron thin films. Thus,we have point out that the observed thickness dependent variationof the L2,3 branching ratio is rather originated by core-holescreening with increased electron delocalization in thicker filmsthan by spin-orbit splitting. Finally, the XANES measurement is asuitable technique to investigate the electronic structure of theinvestigated materials. This work may be the foundation for futuresurface applications using 3d transition metals.

Acknowledgments

The authors express their thank to Professor Piero Pianetta,Professor HermanWinick and the staff at the Stanford SynchrotronRadiation Lightsource (SSRL) for their excellent support, where theXANES experiments have been carried out. SSRL is supported by theDepartment of Energy (DOE), Office of Basic Energy Science. Theauthors acknowledge financial support the DOE CooperativeResearch Program for the Synchrotron-light for Experimental Sci-ence and Applications in the Middle East (SESAME).

References

[1] Stöhr J. NEXAFS spectroscopy. In Springer series in surface sciences, vol. 25.Heidelberg: Springer; 1992.

[2] Lee PA, Citrin PH, Eisenberg P, Kincaid BM. Rev Mod Phys 1981;53:769.[3] Fuggle JC, Inglesfield JE, editors. unoccupied electronic states. Topics in

applied physics, vol. 69. Berlin: Springer-Verlag; 1992.[4] Pearson DH, Ahn CC, Fultz B. Phys Rev B 1993;47(14):8471.[5] Pearson DH, Ahn CC, Fultz B. Phys Rev B 1994;50(17):12969.[6] Nakajima R, Stöhr J, Idzerda YU. Phys Rev B 1999;59:6421.[7] Ufuktepe Y, Akgül G, Lüning J. J Alloys Compounds 2005;401:193.[8] Gota S, Gautier-Soyer M, Sacchi M. Phys Rev B 2000;62:4187.[9] Akgül G, Aksoy F, Bozduman A, Ozkendir OM, Ufuktepe Y, Lüning J. Thin Solid

Films 2008;517:1000.[10] Yamamoto T. X -Ray Spectrom 2008;37:572.

[11] Waddington WG, Rez P, Grant IP, Humphreys CJ. Phys Rev B 1986;34:1467.[12] Dhesi SS, Dürr HA, van der Laan G, Dudzik E, Brookes NB. Phys Rev B 1999;60:

12852.[13] Shalygina EE, Perepelova EV, Kozlovskiĭ LV, Tamanis E, Mukasheva MA,

Shalygin AN. Tech Phys Lett 2006;33(6):456.[14] Shen J, Jenniches H, Mohan Ch V, Barthel J, Klaua M, Ohresser P, et al. Euro-

phys Lett 1998;43(3):349.[15] Park KW, Choi JH, Sung YE. J Phys Chem B 2003;107:5851.[16] Wang H, Ralston CY, Patil DS, Jones RM, Gu W, Verhagen M, et al. J Am Chem

Soc 2000;122:10 544.[17] Ralston CY, Wang H, Ragsdale SW, Kumar M, Spangler NJ, Ludden PW, et al.

J Am Chem Soc 2000;122:10553.[18] Wang H, Patil DS, Gu W, Jacquamet L, Friedrich S, Funk T, et al. J Electron

Spectrosc Relat Phenom 2001;114:855.[19] Farangis B, Nachimuthu P, Richardson TJ, Slack JL, Perera RCC, Gullikson EM,

et al. Phys Rev B 2003;67:085106.[20] Telling ND, Keatley PS, Shelford LR, Arenholz E, van der Laan G, Hicken RJ,

et al. Appl Phys Lett 2008;92:192503.[21] Bagus Paul S, Freund Hajo, Kuhlenbeck Helmut, Ilton Eugene S. Chem Phys

Lett 2008;455:331.[22] Hirsch K, Zamudio-Bayer V, Rittmann J, Langenberg A, Vogel M, Möller T, et al.

Phys Rev B 2012;86:165402.[23] Koshinoa M, Kuratab H, Isodaa S, Kobayashia T. ICR Annu Rep 2000;7:6.[24] Laskowski R, Blaha P. Phys Rev B 2010;82:205104.[25] Leapman RD, Grunes LA, Fejes PL. Phys Rev B 1982;26:614.[26] Barth J, Gerken F, Kunz C. Phys Rev B 1983;28:3608.[27] Leapman RD, Grunes LA. Phys Rev Lett 1980;45:397.[28] Fink J, Müller-Heinzerling T, Scheerer B, Speier W, Hillebrecht FU, Fuggle JC,

et al. Phys Rev B 1985;32:4899.[29] Schwitalla J, Ebert H. Phys Rev Lett 1998;80:4586.[30] Ankudinov AL, Nesvizhskii AI, Rehr JJ. Phys Rev B 2003;67:115120.[31] Thole BT, Van der Laan G. Phys Rev B 1988;38:3158.[32] Ravel B. J Synchroton Rad 2001;8:314.[33] Chen CT, Idzerda YU, Lin H-J, Smith NV, Meigs G, Chaban E, et al. Phys Rev Lett

1995;75:152.[34] Ufuktepe Y, Akgül G, Aksoy F, Nordlund D. X-ray Spectrom 2011;40:427.[35] Nietubyc R, Föhlisch A, Glaser L, Lau JT, Martins M, Reif M, et al. Phys Rev B

2004;70:235414.[36] Hillebrecht FU, Fuggle JC, Bennett PA, Zolnierek Z. Phys Rev B 1983;27:

2179.[37] Diplas S, Jørgensen S, Taftø J, Tønnessen T, Knutsen T, Lehrmann J, et al. Surf

Interface Anal 2006;38:238.[38] Vogel J, Sachhi M. Phys Rev B 1994;49:3230.[39] Huijben M, Martin LW, Chu Y-H, Holcomb MB, Yu P, Rijnders G, et al. Phys Rev

B 2008;78:094413.[40] Lambrakos SG, Cooper KP. J Mater Eng Perform 2009;18:221.