J. G. Tobin, S.-W. Yu, R. Qiao, W. L. Yang, and D. K. Shuh ...

Oxidant K edge x-ray emission spectroscopy of UF4 and UO2J. G. Tobin, S.-W. Yu, R. Qiao, W. L. Yang, and D. K. Shuh

Citation: Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films 36, 03E101 (2018);View online: https://doi.org/10.1116/1.5016393View Table of Contents: http://avs.scitation.org/toc/jva/36/3Published by the American Vacuum Society

Oxidant K edge x-ray emission spectroscopy of UF4 and UO2

J. G. Tobina)

Department of Physics and Astronomy, University of Wisconsin-Oshkosh, Oshkosh, Wisconsin 54901

S.-W. YuLawrence Livermore National Laboratory, Livermore, California 94550

R. Qiao, W. L. Yang, and D. K. ShuhLawrence Berkeley National Laboratory, Berkeley, California 94720

(Received 18 November 2017; accepted 3 January 2018; published 31 January 2018)

The K-Edge (1s) x-ray emission spectroscopy of uranium tetrafluoride and uranium dioxide were

compared to each other and to the results of a pair of earlier cluster calculations. Using a very sim-

plified approach, it is possible to qualitatively reconstruct the main features of the x-ray emission

spectra from the cluster calculation state energies and 2p percentages. Published by the AVS.https://doi.org/10.1116/1.5016393

I. INTRODUCTION

Uranium dioxide (UO2) and uranium tetrafluoride (UF4)

are isoelectronic1,2 in the limit of complete ionization (U4þ,

[Rn]5f2) and have essentially the same nearest neighbor dis-

tances.3 However, their roles in today’s highly technological

society are vastly different. Uranium dioxide is the most

commonly used nuclear fuel for the generation of electricity,

making it of great intrinsic importance.1,2,4 UO2 also exhibits

a number of scientifically interesting traits, including the

presence of covalent behavior in the U5f states.4,5 On the

other hand, its cousin uranium tetrafluoride is, in some

respects, much simpler, with behavior more along the lines

of the ionic limit.1,2 Here, we will compare the experimental

results for the O1s and F1s x-ray emission spectroscopy

(XES) of these two compounds to each other and to simu-

lated spectra derived from the cluster calculations of

Ryzhkov and coworkers.6,7

The comparison to the cluster calculation results will be

done in terms of both the original histogram occupied den-

sity of states (ODOS) and the simulated spectra that can be

generated by replacing each histogram with a scaled func-

tion. These functions are meant to simulate the conditions

that underlay the experimental results, including broaden-

ing from sources such as the whole lifetime and the mea-

surement instrument. To properly perform this task, it is

necessary to treat each component separately and then

sum, rather than summing the histograms and subsequently

broadening. Otherwise, spectral fine structure would be

irretrievably lost.

Cluster calculations have two great advantages: (1) their

simplicity of the models and (2) the use of “natural” orbitals.

These clusters are each composed of a central actinide metal

cation, surrounded by anions of the oxidant, with an excess

of electrons. The calculations are done using wave-functions

that retain the orbital nature of the electrons that will be

involved in the bonding. The result is a set of hybridized

orbitals that provide an approximation to the molecular

orbitals that hold the molecule together, along with a reten-

tion of their original atomic orbital nature.

II. EXPERIMENT

The experiments were carried out at the Advanced Light

Source, Lawrence Berkeley National Laboratory, Berkeley,

CA, and at Lawrence Livermore National Laboratory,

Livermore, CA. XES of the uranium dioxide was performed

at LLNL,8 using a spectrometer described in detail else-

where.9 The XES of the uranium tetrafluoride was done at

the ALS,1,2,10 using a spectrometer that is part of Beamline

8.11 Both samples have been extensively characterized,12–14

and the efficacy of the LLNL spectrometer has been demon-

strated with nonactinide samples.8,15

III. XES AND XPS EXPERIMENTAL COMPARISON

To begin, consider the results shown in Figs. 1 and 2. In

each case, for uranium tetrafluoride and uranium dioxide,

there is a direct comparison of the XES results with the x-

ray photoelectron spectroscopy (XPS) and the histogram

ODOS derived from the calculations of Ryzhkov and cow-

orkers.6,7 Note that while the energy steps are identical, each

of the x-axes can be shifted relative to the others.

It is of interest to note that there is good agreement

between the XPS and XES, with the suggestion that each

XES spectrum is composed of two or possibly three main fea-

tures: a strong central peak with a distinct shoulder on the left

side (lower XES energies) and possibly a less distinct shoul-

der on the right side (higher XES energies). In the case of

UF4, there may be an inflection point on the right side, sug-

gesting a very small shoulder. In the UO2, there is a tail

extending from hv¼ 525 eV to higher energies. However, nei-

ther of these higher photon energy features is as obvious and

as strongly defined as the central peak and low photon energy

(left side) shoulder in each XES and XPS spectrum. Clearly,

the ODOS histograms from the cluster calculations fall within

the manifolds defined by the XPS and XES spectra. (The

weak peak near hv¼ 680 eV is from a UO2F2 surface contam-

inant, as described in Refs. 10 and 13.) However, the question

remains: Can these histograms be converted into a simulated

a)Author to whom correspondence should be addressed; electronic mail:

[email protected]

03E101-1 J. Vac. Sci. Technol. A 36(3), May/Jun 2018 0734-2101/2018/36(3)/03E101/5/$30.00 Published by the AVS. 03E101-1

spectrum which is consistent with each XES measurement?

This issue will be addressed next.

IV. CONCEPTUALIZATION OF XES SPECTRUM

The K (1s) edge XES spectrum is the result of a transition

of an electron from an occupied 2p state into a 1s core–hole,

as shown schematically in Fig. 3. These transitions are elec-

tric dipole in nature, with Dl¼61. Generally speaking, for

each state in the manifold with 2p character, there should be

a corresponding peak in the XES spectrum, with a finite

width caused by a variety of factors, including lifetime

broadening and instrumental resolution limitations. This is

also shown schematically in Fig. 3. Our approach is to take

each occupied state in the Ryzhkov cluster calculation with

nonzero 2p character and generate an XES component peak

for it, scaling the intensity to the 2p percentage, and then

sum all of the component peaks to get an overall spectrum.

For the sake of simplicity and transparency, we will begin

by utilizing the ubiquitous Gaussian function for our compo-

nent line-shape.

V. RESULTS AND DISCUSSION: SPECTRALSIMULATION BASED UPON THE HISTOGRAMODOS

The approach here is minimalist: the goal is to find the

least complicated set of conditions that can explain the

spectral observations. In this analysis, we have made a pair

of simplifying assumptions. (1) The component peak cross

sections are all equal. (2) The component peak widths and

shapes are all the same. To begin, a Gaussian lineshape was

utilized and the full width at half maximum (FWHM) was

varied systematically, as shown in Fig. 4, for the F2p ODOS,

FIG. 1. (Color online) Comparison of the XES results for UF4 with the XPS

of UF4 from Thibaut et al. (Ref. 16) and the ODOS derived from the calcu-

lations of Ryzhkov and coworkers (Ref. 6), for a (UF8)4� cluster. The XES

peak is at hv¼ 675 eV. The XPS measurements of Teterin et al. (Ref. 6)

confirm the Thibaut result. BE (eV) is the binding energy in electron volts,

used in the XPS experiment. The photon energy in the XES experiment is

denoted with hv (eV). ODOS En (eV) is the occupied density of states

energy in electron volts for cluster calculations. This figure is similar to the

one in Ref. 10.

FIG. 2. (Color online) Following Fig. 1, this is a comparison of the XES,

XPS (Ref. 5), and cluster results for UO2. The calculations of Ryzhkov and

coworkers (Ref. 7) are based upon a (UO8)12� cluster. The XPS spectrum

was taken from Ref. 5.

FIG. 3. (Color online) XES process and resultant spectrum. See text for details.

The photon is denoted as hv.

03E101-2 Tobin et al.: Oxidant K edge XES of UF4 and UO2 03E101-2

J. Vac. Sci. Technol. A, Vol. 36, No. 3, May/Jun 2018

with the intensity scaling from the histograms in Fig. 1. The

plots in Fig. 4 show the summation of the contributions from

each component. As can be seen in Fig. 5, the best match

corresponds to FWHM near 1 eV.

Even with this very simple spectral simulation, it is possi-

ble to achieve a fairly good match, as shown in Fig. 5. One

viewpoint would be to note that the three features are

obtained in the simulated spectrum, although the shoulder on

the right side (higher XES energy) is much stronger than that

observed experimentally. There are also two winglike,

weaker shoulders, one on each side, that are not resolved

experimentally, but are consistent with the overall peak

width near the base of the 2p envelope peak. Despite the

appeal of the three feature viewpoint, a superior viewpoint

would be to concede that, experimentally, the smaller and

weaker features are being lost and only the two main strong

features, the central peak and the lower photon energy shoul-

der are resolvable. Thus, the simple theory gives us the two

main spectral features, which are resolved experimentally, as

well as the smaller winglike shoulders and the fine structure

near the peak maximum, which are not resolved experimen-

tally. In some respects, this result is not surprising, for two

reasons: (1) theory almost always resolves more fine struc-

ture than experiment; and (2) the real UF4 sample has two U

sites in its unit cell, which may have slightly different indi-

vidual spectra, which are not separable here.1–3

A similar analysis has been carried out for the O2p mani-

fold in UO2, as shown in Figs. 6 and 7. Once again, the best

visual match is near FWHM¼ 1 eV and a simulated spec-

trum is obtained, with parallel limitations as those described

above for UF4.

FIG. 4. (Color online) Simulated XES spectra from the F2p ODOS histo-

grams, utilizing a Gaussian lineshape and specified FWHM for each compo-

nent, For a Gaussian function, FWHM¼ 2(2ln2)1/2 Sigma¼ 2.355 Sigma.

FIG. 5. (Color online) Comparison of the XES spectrum of UF4 and the sim-

ulated spectrum with the component FWHM¼ 0.94 eV.

FIG. 6. (Color online) Simulated XES Spectra from the F2p ODOS histo-

grams, utilizing a Gaussian lineshape and specified FWHM for each

component.

03E101-3 Tobin et al.: Oxidant K edge XES of UF4 and UO2 03E101-3

JVST A - Vacuum, Surfaces, and Films

Once again, the best analysis may be to accept that some

of the fine structure in the simulation is lost in the experi-

ment. Then, the overall, plateaulike appearance of the simu-

lated spectrum can be explained as being consistent with the

two main features, the central peak and the lower photon

energy shoulder, which are both observed experimentally,

albeit with an overemphasis of the shoulder relative to the

central peak. However, this leaves the problem at the peak

base.

Here, the Gaussian line-shape fails to give the proper

width and tailing. This problem can be fixed, by substituting

a Lorentzian line-shape, as shown in Fig. 8. Although there

is a slight sharpening of peak tops with the Lorentzian, to a

large extent, the central part of the line-shape remains much

the same for the Gaussian and Lorentzian cases: only the

tailing at the base, with its concomitant broadening, is

strongly affected. This result suggests that lifetime broaden-

ing is the dominant effect here and the possible high energy

shoulder is merely lifetime broadening.

VI. SUMMARY AND CONCLUSIONS

It has been shown that, even with the simplifying assump-

tions of constant cross sections and component line-widths

and shapes, it is possible to construct simulated spectra from

the histogram ODOS of the Ryzhkov clusters that agree

fairly well with the experimental oxidant K Edge spectra of

UO2 and UF4. These simulations reconstruct the two main

features in all of the XES and XPS spectra: the central peak

and the lower photon energy shoulder. However, much of

the additional simulated spectral fine structure is lost, includ-

ing the details near the peak maximum. Because of the pres-

ence of smaller features on the wings in the UF4, it is not

necessary to utilize a Lorentzian lineshape, so long as the

premise of the loss of simulated fine structure is accepted.

However, for the UO2, a Lorentzian lineshape is required to

explain the tailings on either side of the peak. To improve

the quality of the match, it will be necessary to properly

include state specific matrix elements for each transition.17

ACKNOWLEDGMENTS

Lawrence Livermore National Laboratory (LLNL) is

operated by the Lawrence Livermore National Security, LLC,

for the U.S. Department of Energy, National Nuclear Security

Administration, under Contract No. DE-AC52-07NA27344.

Work at Lawrence Berkeley National Laboratory (LBNL)

(D.K.S.) was supported by the Director of the Office of

Science, Office of Basic Energy Sciences (OBES), Division

of Chemical Sciences, Geosciences, and Biosciences (CSGB),

Heavy Element Chemistry (HEC) Program of the U.S.

Department of Energy under Contract No. DE-AC02-

05CH11231. The ALS is supported by the Director of the

Office of Science, OBES of the U.S. Department of Energy at

LBNL under Contract No. DE-AC02-05CH11231. The UF4

sample was originally prepared at Oak Ridge National

Laboratory and provided to LLNL by J. S. Morrell of Y12.

1J. G. Tobin et al., Phys. Rev. B 92, 035111 (2015).2J. G. Tobin et al., Phys. Rev. B 92, 045130 (2015).3J. G. Tobin, W. Siekhaus, C. H. Booth, and D. K. Shuh, J. Vac. Sci.

Technol., A 33, 033001 (2015).4J. G. Tobin and S.-W. Yu, Phys. Rev. Lett. 107, 167406 (2011).

FIG. 7. (Color online) Comparison of the XES spectra of UO2 and the simu-

lated spectrum with component FWHM¼ 0.94 eV.FIG. 8. (Color online) Comparison of the O1s XES spectra with a composite

spectrum utilizing a Lorentzian line-shape for the component peaks, with a

component FWHM of 1 eV. The inset shows a direct comparison of

Gaussian and Lorentzian line-shapes for equivalent FWHM values. The

overall FWHM of the O1s peak was about 2 eV, and the projected instru-

mental broadening was 1.2 eV (Ref. 8).

03E101-4 Tobin et al.: Oxidant K edge XES of UF4 and UO2 03E101-4

J. Vac. Sci. Technol. A, Vol. 36, No. 3, May/Jun 2018

5S.-W. Yu, J. G. Tobin, J. C. Crowhurst, S. Sharma, J. K. Dewhurst, P.

Olalde-Velasco, W. L. Yang, and W. J. Siekhaus, Phys. Rev. B 83,

165102 (2011).6A. Yu. Teterin, Yu. A. Teterin, K. I. Maslakov, A. D. Panov, M. V.

Ryzhkov, and L. Vukcevic, Phys. Rev. B 74, 045101 (2006).7Yu. A. Teterin, K. I. Maslakov, M. V. Ryzhkov, O. P. Traparic, L.

Vukcevic, A. Yu. Teterin, and A. D. Panov, Radiochemistry 47, 215 (2005).8S.-W. Yu and J. G. Tobin, J. Electron Spectrosc. Relat. Phenom. 187, 15

(2013).9S.-W. Yu, J. G. Tobin, and B. W. Chung, Rev. Sci. Instrum. 82, 093903

(2011).10J. G. Tobin, S.-W. Yu, R. Qiao, W. L. Yang, and D. K. Shuh, “F1s x-ray

emission spectroscopy of UF4,” Prog. Nucl. Sci. Technol. (submitted).

11J. J. Jia et al., Rev. Sci. Instrum. 66, 1394 (1995).12S.-W. Yu and J. G. Tobin, J. Vac Sci. Technol., A 29, 021008 (2011).13J. G. Tobin, A. M. Duffin, S.-W. Yu, R. Qiao, W. L. Yang, C. H. Booth,

and D. K. Shuh, J. Vac. Sci. Technol., A 35, 03E108 (2017).14S.-W. Yu, J. G. Tobin, P. Olalde-Velasco, W. L. Yang, and W. J.

Siekhaus, J. Vac. Sci. Technol., A 30, 011402 (2012).15J. G. Tobin, S. W. Yu, B. W. Chung, G. D. Waddill, E. Damian, L. Duda,

and J. Nordgren, Phys. Rev. B 83, 085104 (2011).16E. Thibaut, J.-P. Boutique, J. J. Verbist, J.-C. Levet, and H. Noel, J. Am.

Chem. Soc. 104, 5266 (1982).17J. D. Ward, M. Bowden, C. T. Resch, G. C. Eiden, C. D. Pemmaraju,

D. Prendergast, and A. M. Duffin, Spectrochim. Acta, B 127, 20

(2017).

03E101-5 Tobin et al.: Oxidant K edge XES of UF4 and UO2 03E101-5

JVST A - Vacuum, Surfaces, and Films