DaltonTransactions

Dynamic Article Links

Cite this: Dalton Trans., 2011, 40, 6023

www.rsc.org/dalton PAPER

Photoemission study on the valence band of a b-FeSi2 thin film usingsynchrotron radiation†

Hitoshi Fujimoto,*a Koji Ogawa,b Kenichi Takarabe,c Haruhiko Udono,d Harue Sugiyama,b Junpei Azuma,b

Kazutoshi Takahashib and Masao Kamadab

Received 27th December 2010, Accepted 31st March 2011DOI: 10.1039/c0dt01832e

Resonant and constant-initial state photoemission spectroscopies using synchrotron radiation wereapplied to investigate the valence-band electronic structure of a semi-conducting b-type iron-disilicide(b-FeSi2) thin film. The results clearly indicated that the component elements, iron (Fe) and silicon (Si),contribute differently to the valence band features; the Fe 3d orbitals mainly concentrate in the topregion of the valence band while the Si 3s and 3p orbitals spread over the wide region of the valenceband. The b-FeSi2 thin film showed a typical p-type semi-conducting nature with a work function of4.78 eV. The b-FeSi2 film showed the Fe M1VV Auger lines around the kinetic energy of 88 eV. It wouldbe expected from these observations that there exist strong interactions between iron and silicon atomsin the b-FeSi2 film resulting in orbital mixing and band formation.

Introduction

Ecologically and environmentally friendly semiconductor b-typeiron-disilicide (b-FeSi2) has been attracting much attention as anew material for silicon based devices.1 Especially, an observation2

of the electroluminescence at the wavelength of 1.5 mm has acceler-ated studies on this material for both fundamental and applicativeaspects. Detailed information of the electronic structure is essentialto understand the nature of the material, but is still missing.

Many experimental studies have been performed to investigatethe band gap nature, and have reported that b-FeSi2 has a directgap of 0.80–0.95 eV.1,3 It has been also reported that an indirectgap is 0.7–0.78 eV,4 which is lower than the direct one. Theelectronic structure of b-FeSi2 has been investigated by theoreticalcalculations, suggesting that b-FeSi2 has an indirect-type bandstructure.5 It has been also pointed out that minor deformationsturn the gap into a direct one.6 In order to discuss the band gapnature, detailed characterization is needed for both the valenceand conduction bands.

In the present study, we investigate the valence band electronicstructure of a thin b-FeSi2 film by photoelectron spectroscopy

aFaculty of Science, Kumamoto University, 2-39-1 Kurokami, Kumamoto,860-8555, Japan. E-mail: fuji@sci.kumamoto-u.ac.jp; Fax: +81-96-342-3320; Tel: +81-96-342-3383bSynchrotron Light Application Centre, Saga University, 1 Honjo, Saga,840-8502, JapancOkayama University of Science, Ridai, 700-0005, JapandFaculty of Engineering, Ibaraki University, 4-12-1 Nakanarusawa, Hitachi,Ibaraki, 316-8511, Japan† Electronic supplementary information (ESI) available: Core-level UPSspectra for Fe 3p and Si 2p, and the spectral differences for on- and off-resonance and for before and after resonance at the Fe 3p core excitation.See DOI: 10.1039/c0dt01832e

(UPS) with using synchrotron radiation. The valence band featuresare characterised from the results obtained by the resonantphotoemission and constant initial state (CIS) photoemissionspectroscopies.

Experimental

UPS measurements were carried out at the Saga University beamline BL13 in the Saga Light Source.7 Synchrotron radiation withthe critical energy of 1.9 keV was provided from the 1.4 GeV ring,and was monochromatized using a plane grating monochromator.UPS spectra were obtained with the photon energy (hn) regionof 40 to 130 eV using the hemi-spherical electron analyser,OMICRON EA125 HR. The resolution of the electron analyser,which depends on the pass energy of photoelectrons, is of about0.06 eV for all measurements in this study. The base pressure waskept at about 3 ¥ 10-8 Pa during the measurements.

Thin films of b-FeSi2 with thickness of about 150 nm weredeposited on glass substrates using a facing target dc sputteringmethod with the same conditions as reported previously.8 Atransmission electron microscope observation and a fluorescenceX-ray analysis showed that the grown sample consisted of fineFeSi2 crystalline particles with a 3–5 nm radius. The sample wastransferred to the measurement system in air. A gold foil wasalso put near the sample in order to calibrate the energy. Theclean surface was obtained by argon (Ar) ion bombardment at0.5 keV under a sputtering gas pressure of 3 ¥ 10-4 Pa for 10 h.An Auger electron spectrum showed that surface contaminationssuch as oxygen and carbon were removed effectively by sputteringwith this condition, but Ar atoms were implanted into the filmeven with soft sputtering conditions like here. The implantedAr atoms remained after flushing the sample at about 250 ◦C;

This journal is © The Royal Society of Chemistry 2011 Dalton Trans., 2011, 40, 6023–6027 | 6023

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

New

Yor

k U

nive

rsity

on

25/1

0/20

14 0

3:42

:50.

View Article Online / Journal Homepage / Table of Contents for this issue

therefore, the following UPS measurements were performed withthe sputtered surface at the above condition without any furthertreatments.

The work function of the electron-energy analyser was deter-mined to be 4.59 eV by measuring the Fermi edge of the gold foilat hn = 21 eV, where the second order light was used for calibrationof the excitation energy. The photon energies of the excitation lightwere calibrated with using the work function and the Fermi edgeposition of the gold foil.

Results and discussion

The wide range UPS spectra measured at hn = 120 and 130 eVare shown in Fig. 1, where the corresponding features of Augerelectrons and photoelectrons are combined by red and blue lines,respectively. The work function of the b-FeSi2 film was estimatedfrom the cut-off position of the secondary electrons to be 4.78eV, which is nearly the same as the value of 4.7 eV obtained bypreliminary measurements using the retarding-field type electron-energy analyser and a hydrogen discharge lamp.9 The energydifference between the ionization threshold and the Fermi levelindicates that the b-FeSi2 film used in this study is a typical p-typesemiconductor considering the band gap of about 0.7 eV.4 Thecore-level spectra for the Fe 3p and Si 2p levels are presented inthe figures of the ESI.† The photoelectrons from the core levels areobserved at the binding energies of 54 eV and 100 eV for Fe 3p andSi 2p, respectively, which are consistent with the reported values.10

The Auger electrons are also found around the kinetic energiesof 44 eV and 88 eV for Fe M23VV and Si L23VV , respectively.Additionally, it was reported that Fe 3s core levels would beobserved at the binding energy of 90 eV.10 A weak structure canbe seen in favourable at the electron binding energy of about 92eV (see Fig.1 (b) in the ESI), and this might correspond to theFe 3s core level with consideration of the results on the resonantand CIS spectra as discussed later. X-Ray crystallography showedthat a unit cell of the orthorhombic b-FeSi2 crystal consists of 48atoms and both Fe and Si atoms occupy two non-equivalent sitesin the unit cell.11 Therefore, the observed core-level peaks mightconsist of two components due to any difference in the valences

Fig. 1 Wide range UPS spectra of b-FeSi2 observed at hn = 120 and130 eV. Red and blue lines combine the corresponding features of Augerelectrons and photoelectrons, respectively.

of the elements. However, we could not recognise them within theresolution of the spectrometer used.

The valence-band UPS spectra in the hn regions of the Fe3pand Si 2p core-level excitations are shown in Fig. 2(a) and (b),respectively. Three features marked by A–C are observed at 0.5,6.6 and 11 eV, respectively, and these features show differenthn dependences around the Fe 3p core-level excitation: resonantbehaviours are observed in the region of the feature A and Bwith the Fe 3p core-level excitation. This shows that these spectralfeatures have some contributions from the Fe. On the otherhand, however, the features B and C become less pronouncedas compared to the structure A in the hn region of the Si 2pcore-level excitation. It would be thought that this is due to hndependences of the ionization cross-sections and kinetic energydependences of the electron escape depths, besides the fact thatthe sub-shells of the component elements contribute differentlyto the feature A and to the others. According to the calculationresults on the atomic sub-shell photo-ionization cross sections,12

the cross-section ratios of the Si 3s and 3p orbitals to the Fe3d ones decrease about 70% and 90% from hn = 40.8 eV to 80eV, respectively. In addition to these observations, the electronescape depth shows a minima around 0.4 nm at about 40 eV inthe electron kinetic energy, and increases steeply and graduallyin the lower and higher energy regions, respectively.13 Secondaryelectrons raise the background, and also contribute the intensitiesof the high-binding energy structures, especially in the low hnregion. Moreover, the result of the elemental analysis along adepth direction by the X-ray photoelectron spectroscopy showedthat there exists a silicon-rich oxide layer of about 2 nm thickin the air-exposed surface of the as-prepared film.14 It is difficultto discuss quantitatively; however, the intensity decreases of thefeatures B and C are obviously larger than those of the atomic sub-shell ionization cross-sections. From these facts and an assumptionof Si termination occurred on the Ar-sputtered surfaces, thesespectral observations could be explained as follows: the featureA has a relatively large contribution from Fe 3d orbitals, and thestructure B has contributions from both Fe and Si orbitals. The Siatoms in the surface region affect strongly on the spectral featuresB especially in the hn region around the Fe 3p core-level excitation.The origins of the spectral structures and the features marked byred triangles in Fig. 2(b), which alter their binding energy withhn, will be discussed later. It should be noted with interest thatresonance at the Fe 3p core excitation affects the features A orB differently, of which intensity loses or gains, respectively. Thisis more clearly seen in the figures of the ESI, where the spectraldifferences are shown for just on-resonance and after-resonanceand for before and after resonance, respectively.

In order to confirm how each element contributes valence-bandUPS spectral features, the CIS spectra are plotted in Fig. 3(a)and (b) for the hn regions of the Fe 3p and Si 2p core-levelexcitations, respectively. The total photoemission yield is alsoshown in the figures. It is obvious that the wide region of thevalence band is affected along the Fe 3p core-level excitationshowing the contribution of Fe 3d and/or 4s sub-shell orbitals.The CIS spectra clearly show the anti-resonance at the topregion of the valence band but the Fano-like resonance atthe other region across the Fe 3p core-level excitation. Itwas reported from model calculations for Cr, Ni, Cu andZn that the anti-resonance dip can be observed in the main 3d

6024 | Dalton Trans., 2011, 40, 6023–6027 This journal is © The Royal Society of Chemistry 2011

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

New

Yor

k U

nive

rsity

on

25/1

0/20

14 0

3:42

:50.

View Article Online

Fig. 2 hn dependence of the valence-band UPS spectra near the Fe 3p(a) and Si 2p (b) core-level ionization threshold. The base line of eachspectrum is shifted vertically by the proportional amount to hn. Thespectra measured at hn corresponding to Fe 3p and Si 2p core excitationsare depicted by blue and purple curves, respectively. It might be thoughtthat the UPS spectrum of hn = 91.9 eV (red curve) is just taken at the Fe3s core-level excitation, and that the features marked by red triangles areoriginated from the Fe M1VV Auger line (see text).

Fig. 3 CIS spectra in the hn regions of (a) the Fe 3p and Si 2p (b) core-levelexcitations. The total yield is also plotted in the figure (blue curve). It mightbe thought that the intensity change at the hn region shown with a red circleis due to the Fe 3s core-level excitation.

This journal is © The Royal Society of Chemistry 2011 Dalton Trans., 2011, 40, 6023–6027 | 6025

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

New

Yor

k U

nive

rsity

on

25/1

0/20

14 0

3:42

:50.

View Article Online

band and the Auger intensity does not show such dip profile.15

The results were applied to discuss the CIS spectra of FeSi.16

Recently, the self-consistent full-potential linearized augmentedplane wave method was applied to investigate the electronicstructure of b-FeSi2, and the band structure and the densityof states were reported.17 In the case of b-FeSi2, the electronconfigurations of both initial and final states are not confirmedyet, and the reliability of the localized atomic model or the bandmodel is also not inspected; however, many experimental datahave been discussed within a framework of the band model,which gives successful results.5 Therefore, we will discuss belowthe photoemission results on b-FeSi2 with using the band modeland examine the applicability of the band model to the presentsystem.

The band calculations showed that the bands in the regionbetween -13.5 and -5 eV relative to the Fermi energy consistmainly of a Si 3s character with some amount of Fe 4s and 3pcontributions, and that Fe 3d states concentrate in the region nearthe Fermi energy and they strongly hybridize with the Si 3p states.17

It was also shown that the Fe 3d sub-shells contribute the densityof states in the top region above -5 eV of the valence band, whereasthe Si contributions extend widely in the region down to -15 eVof the valence band with relatively small amounts.17 According tothese results on the band calculation and the photoemission inten-sity along the core-level excitation, it would be concluded that theUPS spectral feature A consists mainly of the Fe 3d contributions.Contrary to this, the spectral structure B changes its intensitystrongly across the Fe 3p core excitation; however, the density ofstates has almost no contribution from Fe in this binding energyregion. The most plausible assignment to this feature may be a two-hole state of the Fe 3d states or a satellite structure of the main Fe3d states, which accidentally overlaps with the Si 3s and 3p statesdiscussed above. It is more difficult to give a confident assignmentfor the third feature C. There exists only a small contributionfrom Si 3s sub-shells to the density of states in this binding energyregion. It was also reported that the scraped polycrystalline sampleof FeSi2 does not show such a spectral structure in the X-rayphotoelectron spectra and Ar atoms used for the sputter processgive a similar feature at this binding energy region.18 From theseevidences, we would tentatively assign the feature C as the Si 3s orAr 3p band. It should be noted that the Si 2p core-level excitationdoes not affect strongly on the intensity of the valence band, butdoes give a small effect on the whole valence band, especially inthe hn region above 100 eV, which is consistent with the bandstructure and the density of states as reported recently.17

Fig. 4 compares a spectral profile of the Auger peak observedat 100 eV in the electron kinetic energy, which is the same spectralfeature as those marked by the red triangles in Fig. 2(b). It isinteresting to note that the sharp Auger peak can be observed athn below the Si 2p core-level excitation and the Auger profile so-obtained is obviously different from that of a Si L23VV Auger linemeasured for a Si (001) surface at hn = 130 eV. Moreover, the Augerline of b-FeSi2 observed at the higher hn shows a similar profileto that of the Si L23VV Auger line, which overlaps with the sharpone. It should be also noticed that the top region of the valenceband changes its photoemission intensity in the hn region, wherethe sharp Auger line can be observed (marked by a red circle inFig. 3). From these results and the fact that the Fe 3s core-level canbe only excited in this hn region, the sharp Auger peak observed

below hn = 100 eV would be assigned to the Fe M1VV Augerline, and the structure in the CIS spectra around 92 eV would beoriginated from the Fe 3s core-level excitation. Interesting evidenceabout the valence electronic structure of b-FeSi2 would come fromthis assignment. The first step of the Fe M1VV Auger processis a photo-excitation of the Fe 3s core electrons to the vacantlevel, which should be optically allowed. The band calculationshowed that the bottom of the conduction band has nearly nocontribution from p-type orbitals of Fe atoms, but does have muchcontribution from Si s- and p-type orbitals.17 It could be thoughtthat the destination state of the first step is the hybridized levelbetween Fe 3d and Si 3s and/or 3p orbitals. The result will alsosupport an applicability of the band model to b-FeSi2.

Fig. 4 Auger peak observed at several photon energies. The Si L23VVAuger line at hn = 130 eV is also shown by a red curve.

Conclusions

The resonant and constant initial-state photoemission spectro-scopies have been applied to investigate the valence band electronicstructure of a b-FeSi2 thin film. The results were discussed on theband structure and the density of states.

Three features are observed at 0.5, 7.5 and 11 eV relative to theFermi energy in the valence band UPS spectra. These are assignedfrom the resonance and CIS spectra to the bands originated fromthe Fe 3d sub-shells, the two-hole states of Fe 3d states, and the Si3s and/or Ar 3p states, respectively. In addition, the wide regionof the valence band has a small but obvious contribution fromSi states. It is consistent with the results of the band calculations,showing that the strong hybridization occurs between Fe and Siand the band model would be applied to b-FeSi2.

The b-FeSi2 film showed a change in the Auger line profile atthe photon energy of the Si 2p core-level excitation: the sharpAuger line is observed before the Si 2p core-level excitation, butthe broad line overlaps with the sharp one after the excitation.The broad Auger line has a similar line shape and kinetic energyto the Si L23VV Auger line of a Si (001) surface, and is assigned tothe Si L23VV Auger line of b-FeSi2. The sharp one could originatefrom the Fe 3s core-level excitation, and could be assigned to theFe M1VV Auger line. The strong hybridization between Fe and Simay relate a reason why the Fe M1VV Auger line and the resonanceat the Fe 3s core-level excitation can be observed in b-FeSi2.

Notes and references

1 M. C. Bost and J. E. Mahan, J. Appl. Phys., 1985, 58, 2696; M. C. Bostand J. E. Mahan, J. Appl. Phys., 1988, 64, 2034.

6026 | Dalton Trans., 2011, 40, 6023–6027 This journal is © The Royal Society of Chemistry 2011

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

New

Yor

k U

nive

rsity

on

25/1

0/20

14 0

3:42

:50.

View Article Online

2 D. Leong, M. Harry, K. J. Reeson and K. P. Homewood, Nature, 1997,387, 686.

3 Z. Yang, K. P. Homewood, M. S. Finney, M. A. Harry and K. J. Reeson,J. Appl. Phys., 1995, 78, 1958; M. Ozvold, V. Bohac, V. Gasparık,G. Leggieri, S. Luby, A. Luches, E. Majkova and P. Mrafko, ThinSolid Films, 1995, 263, 92; D. H. Tassis, C. L. Mitsas, T. T. Zorba, M.Angelakeris, C. A. Dimitriadis, O. Valassiades, D. I. Siapkas and G.Kiriakidis, Appl. Surf. Sci., 1996, 102, 178; M. Rebien, W. Henrion, U.Muller and S. Gramlich, Appl. Phys. Lett., 1999, 74, 970; H. Udono,I. Kikuma, T. Okuno, Y. Masumoto, H. Tajima and S. Komuro, ThinSolid Films, 2004, 461, 182.

4 C. Giannini, S. Lagomarsino, F. Scarinci and P. Castrussi, Phys. Rev.B: Condens. Matter, 1992, 45, 8822; A. B. Filonov, D. B. Migas, V.L. Shaposhnikov, N. N. Dorozhkin, G. V. Petrov, V. E. Borisenko, W.Henrion and H. Lange, J. Appl. Phys., 1996, 79, 7708; K. Takakura, N.Hiroi, T. Suemasu, S. F. Chichibu and F. Hasegawa, Appl. Phys. Lett.,2002, 80, 556.

5 N. E. Christensen, Phys. Rev. B: Condens. Matter, 1990, 42, 7148; R.Eppenga, J. Appl. Phys., 1990, 68, 3027; S. Eisebitt, J.-E. Rubensson,M. Nicodemus, T. Boske, S. Blugel, W. Eberbardt, K. Radermacher,S. Mantl and G. Bihlmayer, Phys. Rev. B: Condens. Matter, 1994, 50,18330; A. B. Filonov, D. B. Migas, V. L. Shaposhnikov, V. E. Borisenko,W. Henrion, M. Rebien, P. Stauss, H. Lange and G. Behr, J. Appl. Phys.,1998, 83, 4410; V. N. Antonov, O. Jepsen, W. Henrion, M. Rebien, P.Stauß and H. Lange, Phys. Rev. B: Condens. Matter, 1998, 57, 8934;S. J. Clark, H. M. Al-Allak, S. Brand and R. A. Abram, Phys. Rev.B: Condens. Matter, 1998, 58, 10389; L. Miglio, V. Meregalli and O.

Jepsen, Appl. Phys. Lett., 1999, 75, 385; E. G. Moroni, W. Wolf, J.Hafner and R. Podloucky, Phys. Rev. B: Condens. Matter, 1999, 59,12860.

6 L. Miglio and G. Malegori, Phys. Rev. B: Condens. Matter, 1995, 52,1448; K. Yamaguchi and K. Mizushima, Phys. Rev. Lett., 2001, 86,6006.

7 H. Sugiyama, K. Takahashi, K. Ogawa, J. Azuma and M. Kamada,Proc. International Conf. On Synchrotron Radiation Instrumentation(2009 Melbourn) AIP Conf. Proc., 2010, 1234, 685.

8 K. Takarabe, H. Doi, Y. Mori, K. Fukui, Y. Shim, N. Yamamoto,T. Yoshitake and K. Nagayama, Appl. Phys. Lett., 2006, 88,061911.

9 Unpublished data. The resolution of the system is less than 0.1 eV.10 F. Sirotti, M. De Santis and G. Rossi, Phys. Rev. B: Condens. Matter,

1993, 48, 8299.11 Y. Dusausoy, J. Protas, R. Wandji and B. Roques, Acta Crystallogr.,

Sect. B: Struct. Crystallogr. Cryst. Chem., 1971, 27, 1209.12 J. J. Yeh and I. Lindau, At. Data Nucl. Data Tables, 1985, 32, 1.13 C. R. Brundle, J. Vac. Sci. Technol., 1974, 11, 212.14 Unpublished data.15 L. C. Davis and L. A. Feldkamp, Phys. Rev. B, 1981, 23, 6239.16 A. Kakizaki, H. Sugawara, I. Nagakura, Y. Ishikawa, T. Komatsubara

and T. Ishii, J. Phys. Soc. Jpn., 1982, 51, 2597.17 Z. J. Pan, L. T. Zhang and J. S. Wu, Mater. Sci. Eng., B, 2006, 131,

121.18 V. Kinsinger, I. Dezsi, P. Steiner and G. Langouche, J. Phys.: Condens.

Matter, 1990, 2, 4955.

This journal is © The Royal Society of Chemistry 2011 Dalton Trans., 2011, 40, 6023–6027 | 6027

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

New

Yor

k U

nive

rsity

on

25/1

0/20

14 0

3:42

:50.

View Article Online