bulk-sensitive xas characterization of light elements...

Microchemical Journal 71(2002) 221–230

0026-265X/02/$ - see front matter� 2002 Elsevier Science B.V. All rights reserved.PII: S0026-265XŽ02.00014-0

Bulk-sensitive XAS characterization of light elements: from X-rayRaman scattering to X-ray Raman spectroscopy

Uwe Bergmann *, Pieter Glatzel , Stephen P. Cramera,b, b a,b

MS 6-2100, 1 Cyclotron Rd., Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USAa

Department of Applied Science, University of California, Davis, CA 95616, USAb

Accepted 10 December 2001

Abstract

X-Ray absorption spectroscopy(XAS) is a powerful tool for element-specific characterization of local structureand chemistry. Although application of XAS in the hard X-ray region is now routine, the soft X-ray region(containinglight-element K-edges) presents a number of experimental problems. Most of the difficulties, including surfacesensitivity, restricted sample environments, and radiation damage, stem from the submicron path lengths of soft X-rays andyor electrons. X-Ray Raman scattering(XRS) provides a means for obtaining the information content of softX-ray spectra while maintaining the experimental benefits of hard X-ray techniques. In the XRS process, an incidentphoton is inelastically scattered and part of its energy is transferred to excite an inner shell electron into an unoccupiedstate. Under the dipole approximation, the resulting features are identical to the corresponding XAS spectrum. In thepast, the extremely low cross-section of XRS has made this technique impractical, but intense new X-ray facilitiesand improvements in X-ray optics have helped to put XRS on the brink of becoming a routine spectroscopic tool. Atpresent, high-quality X-ray Raman spectra can be obtained in minutes to hours. X-Ray Raman spectroscopy thusrepresents a hard X-ray alternative to conventional XAS techniques in the study of systems with light elements,including C, N and O. In this review we describe the technique, present examples of recent work, and discuss theprospects for the future.� 2002 Elsevier Science B.V. All rights reserved.

Keywords: Bulk-sensitive XAS characterization; Light elements; X-Ray Raman scattering; X-Ray Raman spectroscopy; InelasticX-ray scattering

1. Introduction

Intense and tunable synchrotron radiation(SR)sources have helped X-ray absorption spectroscopy(XAS) flourish over the past several decades. It is

*Corresponding author. Tel.:q1-510-486-6094; fax:q1-510-486-5664.

E-mail address: [email protected](U. Bergmann).

now a widely applied probe of local molecularand electronic structure. XAS is usually dividedinto X-ray absorption near edge structure(XANES) and extended X-ray absorption finestructure (EXAFS) regions (although the termXAFS is now employed to describe both phenom-ena). In this paper, we emphasize the use of X-ray Raman scattering (XRS) to obtain

222 U. Bergmann et al. / Microchemical Journal 71 (2002) 221–230

bulk-sensitive XANES on low Z(-10) materials.However, the underlying mechanisms also hold forthe EXAFS regime and for higher Z materials.Light-atom XANES spectra are commonly

measured in transmission, Auger yield, electronyield, sample photocurrent or fluorescence excita-tion modesw1x. Each method has its advantagesand limitations. Due to the submicron path lengthsof soft X-rays, transmission measurements requirevery thin samples that are sometimes difficult toprepare. For example, the absorption length ofcarbonaceous material above its K edge energycan be less than 0.1mm. The samples should alsobe transversely homogeneous, and limited sensitiv-ity restricts experiments to relatively concentratedsystems. To simplify sample preparation, the vari-ous electron detection methods are often employed.However, these methods have probe depths of lessthan 50 A and are thus surface sensitivew2x. In˚fact they might provide information about an oxidecoating or sorbed atoms rather than about the bulksample. Fluorescence yield probes deeper into thesample, but it can suffer from artifacts due to‘saturation effects’ in concentrated systemsw3x orfrom variations in fluorescence yield across theabsorption edgew4x.These considerations show that there is a large

class of systems and experimental conditionswhere the bulk properties are difficult to probe byconventional XANES methods. This class includesheterogeneous concentrated compounds, reactivematerials, liquids, and systems under extreme pres-sure or temperature. Higher energy X-ray probeshave several advantages for such samples, includ-ing ‘bulk’ (;mm) sensitivity and less stringentrequirements on the sample environment. Equip-ment such as flow tubes, furnaces, in situ chambersand high-pressure cells have all been employed inthe 5–10 keV range. X-Ray Raman scattering is atechnique that can retain all of the experimentaladvantages of hard X-ray measurements, whilestill revealing the special information that is con-tained in the soft X-ray absorption spectra.

2. A brief history

«I would like to call attention to observations of B. B.Ray w5–7x, which have not, however, been confirmed byothers w8,9x. They concern the passage of Rontgen rays¨

through very thin layers, e.g., of carbon, which resulted inthe formation of a ‘very weak, broad, diffuse’ line on thelong wave-length side of the primary line, which ‘appearedto have a more or less definite edge on its short wave side’.

(Sommerfeld to Compton, 1936w10x).Immediately after the observation of Raman

scattering in the visible region, researchers beganlooking for the X-ray version of this effect, whichwas first mentioned by Smekal in 1923w11x. Davisand Mitchell w12x reported the observation ofshifted lines when analyzing scattered Mo Ka1radiation from graphite. One of the observed linesappeared 279 eV below the elastic Rayleigh scat-tering line, seemingly corresponding to inelasticscattering involving the promotion of a C 1selectron to an empty state. However, two subse-quent studiesw13,14x could not reproduce thisresult, and the explanation by Ehrenbergw13x,assigning the observed signal to Lb fluorescencefrom a small uranium contamination, was probablycorrect. The putative ‘Raman’ line had been muchtoo strong. However, without a tunable source, thetest to discriminate fluorescence from inelasticscattering was not trivial.The hunt continued. In 1950, Das Gupta noted

in a letter to Nature that ‘The new type ofscattering would probably be better detected whenthe involved beam is more or less monochromatic’w15x. Starting with the work on polystyrenew15ax,several reports confirmed the observation of theeffect w16–18x. It was also Das Gupta who, quiteappropriately in our opinion, used the name Sme-kal–Raman modified X-ray scatteringw15ax. Still,in the mid-1960s, there was at least one studyreporting a negative resultw19x. Finally, the theo-retical work by Mizuno and Ohmuraw20x andexperiments by Suzukiw21x clearly established theclose connection between XRS and XAS. In fact,it is argued in the later literaturew22x that Suzuki’sexperiments on elements from Be to C were thefirst to show unambiguous X-ray Raman spectra.The first demonstrations of XRS for spectro-

scopic applications appeared when SR becamemore widely available. Using 8–9 keV X-rays,Tohji and Udagawa were able to obtain an EXAFS-like XRS spectrum from graphite with 6 eVresolutionw22,23x. Schulke et al. studied interband¨

223U. Bergmann et al. / Microchemical Journal 71 (2002) 221–230

Fig. 1. (Top) An X-ray photon with incident energyE is in-0

elastically scattered with final energyE , providing a fractionf

of its energy to excite a core electron into an empty bound orcontinuum state.(Bottom) Rayleigh, Compton and Ramanscattering profiles from graphite(note the log scale). X-axisshows incident photon energy at fixed analyzer setting.

transitions and core excitations in oriented graphitewith 0.8 eV resolution at;8 keV. More recently,several groups have applied XRS to look at the Band N K-edges of hexagonal boron nitridew24x,the K-edge of Li metalw25x, the C K-edge ofasphaltenesw26,27x, and the O K-edgew28,29x,and valence transitionsw30,31x of water. Thepotential of XRS has clearly been recognized, andthird-generation SR facilities such as ESRF, APSand SPring-8 have programs to build dedicatedinelastic scattering beamlines. Below, we will show(with a chemical emphasis) what is possible today.

3. XRS theory

A comparison of the XRS process with otherscattering events is shown in Fig. 1. In general, anincident photon with energyE is scattered with0

energy lossDE, yielding a photon with final energyE . Varying DEsE yE by changing the incidentf 0 f

energy at a fixed analyzer energy set toE resultsf

in a spectrum with three main components(Fig.1). The strongest feature is the elastic Rayleighscattering peak atDEs0. Next is the broad Comp-ton spectrum from inelastic scattering on valenceelectrons. Finally, riding on the Compton back-ground is the XRS spectrum containing the X-rayRaman events, where energy transfer,DE, resultsin the excitation of a core electron. The Ramancross-section is clearly much smaller than thosefor Rayleigh or Compton scattering.The graphite spectrum shown in Fig. 1 was

taken at beamline X-25 at the National Synchro-tron Light Source, with a total resolution of 2.5eV FWHM (see quasi-elastic peak). The transitionprobability for XRS,w, is described byw22x:

3 4 2 2Ž . Ž . Ž .ws 4p e h y m v v 1qcosuµ ∂i j2) ) ) )Ž . Ž .=-f exp iqr i) d E yE yh vyvŽ .f 0 i j

where- f N and Ni) are the final and initial statewave functions,v andv are incident and scatteredi j

X-ray frequencies,u is the scattering angle,q isthe momentum transfer, andr is the distance fromthe nucleus. The integrand is significant only whenr- core hole radius. Whenq r<1, the dipoleapproximation is valid and(also usingNk N(Nk N)i j

the above equation becomesw22x:5 4 2 2 2 2µ ∂ws (64p e h)y(m c ) (1qcosu) sin (uy2)±2-i±r±f)±

where the matrix element is the same as for dipoleX-ray absorptionw20x.To illustrate the excellent agreement between

XRS in the dipole limit and conventional XANES,Fig. 2 compares a 1 eV resolution spectrum takenon graphite in XRS mode compared to a 0.15 eVelectron yield spectrum taken at the AdvancedLight Sources in Berkeleyw32x. The spectra alsoshow that 1 eV resolution reveals most of thedetailed spectral features.When qr41, the appearance of the spectrum

can change because of the contribution from non-dipole terms. This effect has been observed in the


Fig. 2. Comparison of graphite K-edge XANES taken by XRS(top) with 1 eV FWHM resolution and by conventional XASin electron yield mode(bottom) at 0.15 eV FWHM resolution.The analyzer energy was set toE s6.46 keV, at an incidentf

intensity of some 10 photonsys.13

Fig. 3. High-resolution large-acceptance XRS spectrometer.Each crystal has a diameter of 8.9 cm and an 86 cm radius ofcurvature. The Si crystal analyzers are operated very close tobackscattering in(4,4,0) or (6,6,0) reflection order, at 6.46keV or 9.7 keV, respectively.

spectrum of Li metalw25x. In the dipole limit XRSdirectly corresponds to conventional XAS.

4. Instrumentation

In order to be practical, an XRS experimentrequires an intense monochromatic X-ray sourceand a large-acceptance high-resolution analyzersystem. To resolve spectral features at an absorp-tion edge, the combined monochromatoryanalyzerenergy resolution should be;1 eV or better. Flatcrystals such as Si, Ge or diamond are well suitedas monochromators for a collimated SR beam, butthey are not efficient as analyzers for the scatteredradiation. In recent years, the analyzers whereusually built with cylindrically or spherically bentGe or Si crystals.Our instrument is based on a Johannw33x type

spectrometer using spherically curved crystals. Theinstrument provides good energy resolution andrelatively large angular acceptance. With 8.9 cmdiameter Si or Ge crystals, the smallest possiblespherical bending radius is slightly less than 1 m,yielding a solid angle of;6=10 of 4p sr. Wey4

have built a multi-crystal device employing eight

such analyzers on intersecting Rowland circlesw34x (Fig. 3). At a Bragg angle,u , of approxi-B

mately 888 the spectrometer captures a solid angleof 0.5% of 4p sr. The resolution is;0.3 eV at6.46 keV using Si(4,4,0) reflections or;0.5 eVat 9.7 keV using Si(6,6,0) reflections. With furtheroptimization, a total resolution of 0.5 to 0.7 eVshould be obtainable with our device; even betterresolution can be obtained by systems withgrooved analyzer crystalsw35x.The best energy for an XRS experiment is

sometimes determined by avoiding potential over-lap of XRS and Compton scattering near its max-imum. Other factors can include the momentumtransferq, absorption edges in the sample, and thefact that the Compton cross-section stronglyincreases with increasing energy. The Comptonwavelength shift is Dl sl9yl s0.0243(1yC 0

cosu)A, where l and l9 are the incident and0˚

scattered wavelengths; andu the scattering angle.This corresponds to an energy shift ofDE sC


Fig. 4. XRS K-edge spectra showing extreme cases of aromatic(solid) and aliphatic hydrocarbons(dashed). The benzene(C H ) spectrum was taken at a sample temperature of 10 K,6 6

the n-octacosane(C H ) spectrum was taken at room28 58

temperature.

Fig. 5. Two-minute K-edge XRS raw spectrum of diamond.Incident flux ;3=10 photonsys, energy resolution 1 eV13

FWHM, E s6.46 keV.f

E wDl y(l qDlC)x, which is strongly dependent0 C 0

on the incident energy. At largeE (corresponding0

to small l ) the energy shift,DE , can be very0 C

large, resulting in unwanted background in theenergy range of XRS. Some typical numbers forus90 areDE s63.6 eV atE s5.73 keV;DE so

C 0 C

112.7 eV atE s7.64 keV; andDE s175.5 eV at0 C

E s9.55 keV. The Compton spectrum shown in0

Fig. 1 is further broadened because of the rangeof scattering angles in our apparatus(see below).Very intense X-ray sources are needed for XRS

experiments, and the selected beamlines producedflux densities of up to;10 photonsysyeV. Such13

high doses of radiation create substantial sampledamage, and exposure times have to be kept shortto avoid artifacts in the spectra. This was guaran-teed by operating the beamline monochromator atAPS beamline 18ID in continuous scanning mode,where a XANES scan typically took 2–3 minbefore the sample was automatically repositionedto a fresh spot. Depending on the data qualityrequired and XRS intensity, 5–20 of such scanswere averaged per spectrum. Unwanted back-ground scattering was minimized by using a small

detector aperture and employing a liquid nitrogencooled low-noise Ge detector.

5. Survey of recent results

We begin the examples by comparing the C K-edge XRS for aromatic(benzene) and aliphatic(n-octacosane) samples(Fig. 4). The sharp lowenergy peak in the benzene spectrum is a 1s–p*transition – these can only occur in systems withp bonding. The higher energy features()290 eV)in both spectra are 1sys* transitions. The energyof these resonances varies inversely with bondlength. Thus, the 1sys* benzene feature at 302eV is shifted to higher energy from the samefeature in n-octacosane. A detailed discussion ofthis correlation can be found in Stohr’s bookw1x.¨The higher energy 1sys* transitions are oftenreferred to as ‘shape resonances’, because it isargued that the excited state is stabilized againstimmediate decay by a hump in the potential shape.An extreme example of shape resonances domi-nating a spectrum is the K-edge of diamond shownin Fig. 5. The raw spectrum shown here onlyrequired a total scanning time of 2 min! The muchlower background contributes to the remarkable


Fig. 6. O K-edge water taken in XRS mode at 9.7 keV usinga Si (4,0,0) monochromator at a flux of;3=10 photonsys12

and Si(6,6,0) analyzers. At a resolution of 1 eV FWHM thepre-edge feature is clearly visible.

data quality. As was seen for benzene, the char-acteristic diamond features are in good agreementwith conventional XANESw1x.These C K-edge spectra clearly demonstrate that

XRS on concentrated carbonaceous systems is notonly feasible, but practical for routine bulk studies.Such work has recently been started on asphaltenesand a series of aromatic and aliphatic hydrocarbonmodel compounds and their mixturesw26,27x.Asphaltenes and coals represent only one class ofsystems that benefit from XRS analysis. There arenumerous other carbonaceous systems where XRSis an essential alternative or valuable complementto conventional XANES.Liquids are another class of samples that are

troublesome(but not impossible) for soft XAS.Variation of pressure andyor temperature is espe-cially difficult and in some cases impossible. XRSspectra of the XANES and EXAFS O K-edge forwater and ice have been reported with 2 eVresolution by Borwon et al.w28x. Interestingly,these results were published before the first ’con-ventional’ surface XANES using total ion yieldand total electron yield detection on liquid watermicrojets w36x. This indicates the experimentaldifficulties conventional XAS encounters in suchwork (see alsow37x).Fig. 6 shows a more recent XRS spectrum of

the O K-edge of water at room temperatureobtained with 1 eV resolutionw29x. The overallshape agrees well with the data of Bowron et al.,but the better resolution reveals a pre-edge featureat 534.5 eV. This important feature is interpretedby theory as due to the donor asymmetric breakingof hydrogen bondsw37x.With a K-edge at only;60 eV, Li is another

element that is extremely difficult to study byconventional XANES (see w38x and referencestherein). Lithium has many important chemicaland material science-related applications, e.g. inbatteries and glasses, and a bulk-sensitive probe isessential. Due to the small energy transfer, theXRS signal of Li is in principle large, but thereare two limiting factors. First, Compton scatteringcan be very strong in the range ofDEs50y100eV (see Fig. 1). Second, to obtain the full Li XRSsignal, a large scattering volume is required, sinceeach Li atom has a small scattering amplitude.

Even in concentrated Li systems this is oftenprevented by competition with much stronger scat-tering and absorption by the heavier elements inthe sample. To illustrate the proportion of Comptonbackground and XRS signal, the raw spectrum ofLiOHØH O is shown in Fig. 7. Also shown is the2

curve fitted for Compton background subtraction.Compton scattering contributes the largest part ofthe signal, approximately 10 times more than XRSin the XANES region. At such a low energytransfer the XRS signal will always be on therising side of the Compton spectrum. The decisionas to what energy to use for Li XRS is, therefore,case-dependent and has to include choice ofq,scattering length of the compound and ComptonyXRS ratio. Fig. 8 compares XRS spectra fromthree Li compounds with tetrahedral O coordina-tion. In the spectra two features, one at 60–62 eVand one at 65–68 eV are resolved. The study byTsuji w38x on Li halides (octahedral symmetry)shows two features at similar energies shifting tohigher energies with increasing ligand electrone-gativity. Theoretical work on understanding theobserved features is currently underway.Although our own studies have focused on

energy losses of 50–600 eV, corresponding to


Fig. 7. Li K-edge raw spectrum of LiOHØH O showing large2

Compton scattering background. Calculated curve is used tosubtract Compton background. The experiment was done withan Si 1,1,1) monochromator and Si(6,6,0) analyzers yieldinga total energy resolution of 1.7 eV FWHM at an energy ofE sf9.7 keV. The incident flux was some 10 photonsys.13

Fig. 8. Comparison of Li K-edge XRS spectra at 1.7 eVFWHM from different tetrahedrally coordinated compounds.Top to bottom: LiOHØH O, Li CO , Li SiO .2 2 3 4 4

second row element K-edges, there is substantialinterest in experiments at smaller energy losses.For example, Hayashi et al.w31x used XRS toderive the complete optical spectrum of liquidwater. Fig. 9(courtesy of Hayashi et al.) showsthe optical oscillator strength of liquid water com-pared with that of different phases of waterobtained by reflection and electron energy lossstudies. Such measurements and related data atlarge q w39x are crucial to support simulationsaimed to understand the behavior of charged par-ticles in liquid water.Another example of UV-like spectroscopy with

hard X-rays is the study ofp™p* transitions inbenzene ringsw40x. At even smaller energy losses,the sample excitations correspond to localizedvibrational modes or delocalized phononsw41–44x. Halcoussis et al.w45x observed O–H and O–Ostretching frequencies in liquid normal and deuter-ated water at 0.4 ant 0.3 eV resolution respectively.Fig. 10 (reprint of Fig. 1 from w45x) shows thespectra at an energy transfer form 200–550 meVcorresponding to wave numbers of 1613–4436cm . The signal rate was less than 1 photonys!y1

One way to enhance signal rates is to exploitresonance effects. This is well known in UV-visible Raman spectroscopy, a technique often usedto study dilute species. There is an analogous X-ray resonance Raman effect, which has been wide-ly utilized in recent years. Examples include thestudy of electronic structure in Mott insulatorsw46x, charge transfer excitations in NiOw47x andNd CuO w48x and quadrupolar transitions in Gd2 4

w49x. X-Ray resonance Raman scattering is arapidly expanding field, but beyond the scope ofthis survey.

6. Summary and outlook

In this review we have focussed on XRS as abulk-sensitive hard X-ray method for obtaininglight element XANES spectra. These experimentshave become possible by combining high resolu-tion, large-acceptance X-ray optics with third gen-eration SR sources. With incident fluxes of;10 photonsys, XRS signal rates of up to and13

10 photonsys at concentrated carbonaceous com-4

pounds with Compton background rates between10 and 10 photonsys (highest for Li compounds)3 4

are observed. Sample radiation damage can be animportant problem at such high incident fluxes.


Fig. 9. Comparison of oscillator strength distribution of liquid water obtained by inelastic X-ray scattering with that of gas phasewater obtained by EELS. Inset: High-resolution(0.5 eV) inelastic X-ray scattering data of liquid water compared to data of gasphase water(EELS), amorphous ice(EELS), and hexagonal ice(reflection study). Closed squares(black solid line): raw(smoothed)inelastic X-ray scattering data.

Hence, future improvements should address largeranalyzer solid angle and better energy resolutionas much as brighter SR sources. One proposedinstrument w50x could increase efficiency by atleast a factor of 10.Anticipated fourth-generation sources will be

many orders of magnitude brighter than currentfacilities. In some cases this might be used tofacilitate vibrational XRS, while in other casestime-resolved XRS can be envisioned. Raman andSmekal would be surely amazed at what the laserhas done for UV-visible inelastic scattering, andwe expect the future of XRS to be equallysurprising.

Acknowledgments

We thank Dr Andreas Freund for lending us thediamond crystal and the beamline staff of X-25and 18ID for support during the experiments. Thisresearch was supported by the National Institutesof Health, grants 44891-5, GM 44380 and GM-48145, and by the Department of Energy, Officeof Biological and Environmental Research. TheNational Synchrotron Light Source and theAdvanced Photon Source are supported by theDepartment of Energy, Office of Basic EnergySciences. Beamline X-25 is supported in part bythe U.S. National Institutes of Health, National


Fig. 10. Inelastic X-ray scattering spectra of the intermolecularOH and OD stretching modes of H O, D O and an H O–D O2 2 2

mixture with an equimolar HyD composition(HyDs1:1). Thespectra are vertically separated, for illustration. Intensities aregiven in arbitrary units. For the H O spectrum, a typical count2

rate at 400 meV was approximately 0.4–0.5 photonsys.

Center for Research Resources, and by the U.S.Department of Energy Office of Biological andEnvironmental Research. BioCAT is a NationalInstitutes of Health-supported Research CenterRR-08630.

References

w1x J. Stohr, NEXAFS spectroscopy, Springer-Verlag, Ber-¨lin, New York, 1992.

w2x M. Abbate, J.B. Goedkopp, F.M.F. de Groot, et al., Surf.Interface Anal. 18(1992) 65.

w3x J. Goulon, C. Goulon-Ginet, R. Cortes, J.M. Dubois, J.Phys. 43(1982) 539.

w4x F.M.F. de Groot, M.-A. Arrio, P. Sainctavit, C. Cartier,C.T. Chen, Physica B 84(1995) 208–209.

w5x B.B. Ray, Zeitschrift fuer Physik 66(1930) 261.w6x R.C. Majumdar, Nature 127(1931) 92.w7x S. Bhargava, J.B. Muckerjie, Nature 127(1931) 273.w8x A.J. O’Leary, Phys. Rev. 37(1931) 873.w9x J.M. Cork, Phys. Rev. 37(1931) 1555.

w10x A. Sommerfeld, Phys. Rev. 50(1936) 38.w11x A. Smekal, Naturwissenschaften 11(1923) 873.w12x B. Davis, D.P. Mitchell, Phys. Rev. 32(1928) 331.w13x W. Ehrenberg, Zeitschrift fuer Physik 53(1928) 234.w14x D. Coster, I. Nitta, W.J. Thijssen, Nature 124(1929)

230.

w15x K. Das Gupta, Nature 166(1950) 563.w15ax K. Das Gupta, Phys. Rev. Lett. 3(1959) 38.w16x K. Das Gupta, Phys. Rev. 128(1962) 2181.w17x K. Das Gupta, Phys. Rev. Lett. 13(1964) 338.w18x A. Faessler, P. Muhle, Phys. Rev. Lett. 17(1966) 4.¨w19x R.J. Weiss, Phys. Rev. 140(1965) A1867.w20x Y. Mizuno, Y. Ohmura, J. Phys. Soc. Jpn. 22(1967)

445.w21x T. Suzuki, J. Phys. Soc. Jpn. 22(1967) 1139.w22x K. Tohji, Y. Udagawa, Phys. Rev. B(Condensed Matter)

39 (1989) 7590.w23x K. Tohji, Y. Udagawa, Phys. Rev. B(Condensed Matter)

36 (1987) 9410.w24x N. Watanabe, H. Hayashi, Y. Udagawa, K. Takeshita,

H. Kawata, Appl. Phys. Lett. 69(1996) 1370.w25x M.H. Krisch, F. Sette, C. Masciovecchio, R. Verbeni,

Phys. Rev. Lett. 78(1997) 2843.w26x U. Bergmann, O.C. Mullins, S.P. Cramer, Anal. Chem.

72 (2000) 2609.w27x U. Bergmann, H. Groenzin, O.C. Mullins, P. Glatzel, J.

Fetzer, S.P. Cramer, submitted(2001).w28x D.T. Bowron, M.H. Krisch, A.C. Barnes, J.L. Finney,

A. Kaprolat, M. Lorenzen, Phys. Rev. B 62(2000)R9223.

w29x U. Bergmann, P. Wernet, P. Glatzel, A. Nilsson, L.G.M.Petterson, S.P. Cramer, in preparation(2001).

w30x H. Hayashi, N. Watanabe, Y. Udagawa, C.C. Kao, J.Chem. Phys. 108(1998) 823.

w31x H. Hayashi, N. Watanabe, Y. Udagawa, C.C. Kao,Proceedings of the National Academy of Sciences ofthe United States of America 97(2000) 6264.

w32x S. Anders, J. Diaz, J.W. Ager, R.Y. Lo, D.B. Bogy,Appl. Phys. Lett. 71(1997) 3367.

w33x H.H. Johann, Zeitschrift fur Physik 69(1931) 185.¨w34x U. Bergmann, S.P. Cramer, A High-Resolution Large-

Acceptance Analyzer for X-ray Fluorescence andRaman Spectroscopy, SPIE-The International Societyfor Optical Engineering, San Diego, California, 1998,p. 198.

w35x C. Masciovecchio, U. Bergmann, M. Krisch, G. Ruocco,F. Sette, R. Verbeni, Nuclear Instruments & Methods inPhysics Research Section B-Beam Interactions WithMaterials and Atoms 111(1996) 181.

w36x K.R. Wilson, B.S. Rude, T. Catalano, et al., J. Phys.Chem. B 105(2001) 3346.

w37x S. Myneni, et al., submitted for publication(2001).w38x J. Tsuji, K. Kojima, S. Ikeda, H. Nakamatsu, T. Mukoy-

ama, K. Taniguchi, J. Synchrotron Radiat. 8(2001)554.

w39x N. Watanabe, H. Hayashi, Y. Udagawa, Bull. Chem.Soc. Jpn. 70(1997) 719.

w40x H. Hayashi, N. Watanabe, Y. Udagawa, C.C. Kao, J.Electron Spectroscopy and Related Phenom. 114(2001)933.


w41x F. Sette, G. Ruocco, M. Krisch, et al., Phys. Rev. Lett.75 (1995) 850.

w42x F. Sette, G. Ruocco, M. Krisch, C. Masciovecchio, R.Verbeni, U. Bergmann, Phys. Rev. Lett. 77(1996) 83.

w43x G. Ruocco, F. Sette, M. Krisch, U. Bergmann, C.Masciovecchio, R. Verbeni, Phys. Rev. B-Condens.Matter 54(1996) 14892.

w44x G. Ruocco, F. Sette, U. Bergmann, et al., Nature 379(1996) 521.

w45x C. Halcoussis, T. Abdul-Redah, H. Neumann, G. Mon-aco, C.A. Chatzidimitriou-Dreismann, 2000, ESRFNewsletter, October 17.

w46x M.Z. Hasan, E.D. Isaacs, Z.X. Shen, et al., Science 288(2000) 1811.

w47x C.C. Kao, W.A.L. Caliebe, J.B. Hastings, J.M. Gillet,Phys. Rev. B 54(1996) 16361.

w48x J.P. Hill, C.C. Kao, W.A.L. Caliebe, et al., Phys. Rev.Lett. 80 (1998) 4967.

w49x M.H. Krisch, C.C. Kao, F. Sette, W.A. Caliebe, K.Hamalainen, J.B. Hastings, Phys. Rev. Lett. 74(1995)4931.

w50x U. Bergmann, R. Frahm, TDR XFEL workshop series‘Methods and Instrumentation for the XFEL’, J. Has-tings and Th. Tschentscher, eds.(2001) 52.

bulk-sensitive xas characterization of light elements...

Documents