1: \documentclass[twocolumn,prb,superscriptaddress,showpacs,preprintnumbers]{revtex4}
2: %\documentclass[preprint,prb,tighten,superscriptaddress,showpacs]{revtex4}
3: %\documentclass[preprint,tighten,endfloats*,prb,superscriptaddress,showpacs]{revtex4}
4:
5: \usepackage{graphicx}
6: \usepackage{amsmath,graphicx,dcolumn}
7: \usepackage[usenames]{color}
8: %\usepackage{datetime,fmtcount}
9:
10: \begin{document}
11:
12: \title{Quantifying structural damage from self-irradiation in a plutonium superconductor}
13:
14: \author{C. H. Booth}
15: \affiliation{Chemical Sciences Division,
16: Lawrence Berkeley National Laboratory,
17: Berkeley, California 94720, USA}
18: \author{E. D. Bauer}
19: \affiliation{Materials Physics and Applications Division, Los Alamos National Laboratory,
20: Los Alamos, New Mexico 87545, USA}
21: \author{M. Daniel}
22: \affiliation{Chemical Sciences Division,
23: Lawrence Berkeley National Laboratory,
24: Berkeley, California 94720, USA}
25: \author{R. E. Wilson}
26: \altaffiliation[Permanent address: ]{Chemistry Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439-4831, USA}
27: \affiliation{Chemical Sciences Division,
28: Lawrence Berkeley National Laboratory,
29: Berkeley, California 94720, USA}
30: \author{J.~N.~Mitchell}
31: \affiliation{Materials Science and Technology Division, Los Alamos National Laboratory,
32: Los Alamos, New Mexico 87545, USA}
33: \author{L. A. Morales}
34: \affiliation{Materials Science and Technology Division, Los Alamos National Laboratory,
35: Los Alamos, New Mexico 87545, USA}
36: \author{J. L. Sarrao}
37: \affiliation{Materials Physics and Applications Division, Los Alamos National Laboratory,
38: Los Alamos, New Mexico 87545, USA}
39: \author{P. G. Allen}
40: \affiliation{Materials Science and Technology Division, Lawrence Livermore
41: National Laboratory, Livermore, California 94550, USA}
42:
43:
44: \date{\today}
45: %\date{\today, \xxivtime}
46:
47: %\preprint{LBNL-61766}
48:
49: \begin{abstract}
50: The 18.5 K superconductor PuCoGa$_5$ has many unusual properties, including
51: those due to damage induced by self-irradiation. The superconducting
52: transition temperature decreases sharply with time,
53: suggesting a radiation-induced Frenkel defect concentration
54: much larger than predicted by current radiation damage theories. Extended
55: x-ray absorption fine-structure measurements demonstrate that while the local
56: crystal structure in fresh material is well ordered, aged material is
57: disordered much more strongly than expected from simple defects, consistent
58: with strong disorder throughout the damage cascade region. These data
59: highlight the potential impact of local lattice distortions relative to defects
60: on the properties of irradiated materials and underscore the need for more
61: atomic-resolution structural comparisons between radiation damage experiments
62: and theory.
63:
64: \end{abstract}
65:
66: \pacs{71.27.+a, 74.70.Tx, 61.80.-x, 61.10.Ht}
67: %EXAFS and XANES in condensed matter, 61.10.Ht
68: %Structure irradiation effects on, 61.80.–x
69: %Strongly correlated electron systems, 71.27.+a
70: %Heavy-fermion solids superconductivity, 74.70.Tx
71:
72: \maketitle
73:
74: %*****************************************************************************
75: %*****************************************************************************
76: \section{Introduction}
77: Plutonium is arguably the most complex and least understood of all
78: elements, ultimately due to the propensity of
79: its 5$f$-electrons to simultaneously reside in bonding and non-bonding
80: electronic states. Theoretical models of $\delta$- and $\alpha$-Pu, for instance, show
81: promise for explaining this complex
82: behavior,\cite{Shim07,Dai03,Savrasov01} but assume a homogeneous crystalline structure despite
83: unavoidable self-irradiation damage, which can significantly alter magnetic and
84: electronic properties. Superconductors
85: provide a path for elucidating radiation damage effects\cite{Sweedler79} since their
86: properties are especially vulnerable
87: to atomic-level structural disorder. Here, we report extended x-ray absorption
88: fine-structure (EXAFS) measurements on the PuCoGa$_5$ superconductor\cite{Sarrao02} that
89: demonstrate the local structure of aged
90: material is damaged at least an order of magnitude faster than theoretical predictions focusing on Frenkel defects indicate.\cite{Wolfer00}
91: These results explain the sharp reduction of the superconducting critical temperature, $T_c$, with time
92: and underscore the need for improved radiation damage models
93: relevant not only for understanding plutonium superconductors,
94: but also for Pu metal and other radioactive materials.
95:
96: In addition to changing fundamental properties in elemental Pu and PuCoGa$_5$, radiation damage
97: affects many aspects of science and industry, for example, in semiconductors, nuclear power
98: generation and its associated waste disposal, and the aging nuclear stockpile. Consequently,
99: radiation damage has been studied for over a century, and intensely since World War II. Although
100: experimental measures have provided ample verification of long-range ($>$10 nm) structural effects
101: due to radiation damage, atomic-resolution descriptions of the damage induced in a crystal have
102: relied almost exclusively on theoretical calculations. This reliance has been predictably dangerous.
103: Recent\cite{Farnan07,Farnan01} nuclear magnetic resonance (NMR) experiments reveal that radiation
104: damage accumulates about 5 times faster in zircons and other ceramics than calculations indicate,
105: calling into question both the theoretical calculations and the viability of nuclear waste
106: containment schemes. Experimental studies have not generally kept pace
107: with theoretical treatments of fully relaxed damage cascade structures at the atomic level. Damage
108: cascade calculations have not been quantitatively compared to
109: any atomic-resolution experiments until very recently,\cite{Farnan07,Farnan01} and
110: never to our knowledge in intermetallics\cite{AverbackNote} or technologically
111: important materials such as $\delta$-Pu.
112: Measurements of such structural
113: changes are vital to verify and improve these theories, and for comparison to more complex theories of dislocation loops, He bubble formation, and volume expansion.
114:
115: Because large regions of a given sample remain in an undamaged state while damaged regions no longer have translational symmetry, traditional scattering techniques such as Rietveld analysis of powder diffraction data have not generated a quantitative measure of the damage fraction of a material. Local probes that treat ordered and disordered regions on an equal footing should be able to provide a more detailed, atomic-level description of the damage. NMR is one such technique, although it does not easily lend itself to a direct structural interpretation. The EXAFS technique employed here provides another local probe and has the advantage that it gives radial pair-distance distribution information around a specific atomic species, since it relies on the backscattering of a photoelectron from a core excitation.
116:
117: To investigate structural radiation damage effects at the atomic level in
118: plutonium, we employ superconducting PuCoGa$_5$ (Fig. \ref{Tc_fig}). The
119: unusual properties of this 18.5 K superconductor, i.e., a nearly ten-fold
120: higher $T_c$ than any other heavy-fermion related intermetallic and an
121: electronic structure that bears a strong resemblance to $\delta$-Pu, have been
122: well characterized.\cite{Sarrao02,Joyce03,Curro05} Superconducting properties
123: that change with time due to plutonium $\alpha$-decay include a decrease in
124: $T_c$ by $\sim$0.2 K per month (Fig. \ref{Tc_fig}), the upper critical
125: field $H_{c2}\approx 70$ T, and large critical current density
126: $J_c > 10^4$ A/cm$^2$ for $T>0.9 T_c$,\cite{Sarrao02} likely caused by
127: damage-induced scattering and pinning centers. In addition, the impurity
128: scattering rate inferred from NMR experiments follows the observed change in
129: $T_c$.\cite{Curro05}
130:
131: %*****************************************************************************
132: \begin{figure}
133: \includegraphics[width=3.5in,trim=0 0 0 0]{Tc2.eps}
134: \caption{(Color online)
135: The superconducting transition temperature ($T_c$) as measured by magnetometry
136: as a function of the $\alpha$-decays per atom for this study (samples {\bf A}
137: and {\bf B}) and in the literature
138: (1 yr=4.9$\times10^{-6}$~$\alpha$-decays per atom for these samples).
139: A linear decay
140: of $T_c$ (\textcolor{red}{---}) starting from about 19.0 K and decaying at a
141: rate of about 5.4$\times10^{5}$ K per $\alpha$-decay per atom is also shown for
142: reference.
143: The inset shows the crystal structure of PuCoGa$_5$.\cite{Sarrao02}
144: }
145: \label{Tc_fig}
146: \end{figure}
147: %*****************************************************************************
148:
149: Although a modern theoretical
150: treatment of radiation damage in PuCoGa$_5$ is currently not available, we
151: utilize calculations on the structurally similar $\delta$-Pu system for
152: comparison.\cite{Sarrao02} In these models, the
153: $\alpha$-particle generated by the decay of a $^{239}$Pu nucleus has about
154: 5 MeV of energy and ballistically generates $\sim$300 Frenkel defect pairs over
155: a distance of nearly a micron.\cite{Wolfer00} Most of the damage,
156: however, is done by the recoiling $^{235}$U nucleus with 86 keV, which
157: produces $\sim$2300 Frenkel pairs. A typical generated damage cascade extends over
158: nearly 10 nm, with a defect volume fraction of about 3\%.
159: It is important to note that only the effect of Frenkel-type defects are
160: considered in this view of radiation damage; however, these
161: models form the basis of most modern theories such as molecular dynamics
162: theories (for example, see Ref. \onlinecite{Valone04}). In fact, lattice
163: relaxation can occur during the intermediate time-scales after these
164: displacement events, and have been explored in various metallic systems with
165: molecular dynamics and kinetic Monte Carlo techniques. These latter
166: calculations show that the effective number of defects is reduced by as much
167: as an order of magnitude within only a few picoseconds due to additional
168: defect migration.\cite{Rubia99} These values should not change substantially
169: in PuCoGa$_5$, and in fact a rough TRIM code\cite{TRIM} calculation (which
170: does not include the longer-time scale defect migration) using default values
171: generates similar damage rates and ranges to the models above.
172: We therefore expect far fewer than the initial $N_\textrm{D}\sim$2600 pair defects to survive the defect migration per
173: $\alpha$-decay. Using an $\alpha$-decay rate $\lambda_\alpha\approx3.4\times10^{-5}$ per Pu per year
174: from the samples discussed below, the expected upper-limit damage fraction
175: $f_\textrm{tot}=2 N_\textrm{D}\lambda_\alpha/7\approx2.5\%$ after one year,
176: with 1/7$^\textit{th}$ of the atoms being Pu. $T_c$ is reduced by 50\% after
177: about 3 years (Fig. \ref{Tc_fig}).
178: These damage estimates therefore indicate that, including the expected
179: additional defect migration, $f_\textrm{tot}(T_c/T_{c0}=50\%) \ll 7.5$\%. This
180: estimate is inconsistent with what one expects in a short coherence length
181: superconductor:
182: Within the Bogoliubov-de Gennes (BdG) formalism with strong scattering and a
183: coherence length $\xi_0\approx$ 2.1 nm\cite{Sarrao02}), the {\it lower limit}
184: (strong scattering limit) damage fraction is
185: $f_\textrm{BdG}(T_c/T_{c0}=50\%) \approx$15\%.\cite{Franz97} This value is consistent with recent
186: studies of Ce$_{1-x}$La$_x$CoIn$_5$\cite{Petrovic02} and
187: CeCoIn$_{5-x}$Sn$_x$\cite{Daniel05b} where the substitutions are mostly within
188: the superconducting planes, and therefore likely produce strong scattering.
189: The critical damage fraction should be higher when defects are randomly
190: distributed, as with radiation damage.
191:
192: This paper continues with a description of the sample characterization and other experimental methods (Sec. \ref{Methods}), details of the data analysis and results (Sec. \ref{Results}), a discussion of the implications of these results (Sec. \ref{Discussion}) and a concluding summary (Sec. \ref{Conclusion}).
193:
194: %*****************************************************************************
195: %*****************************************************************************
196: \section{Experimental Methods}
197: \label{Methods}
198:
199: Two PuCoGa$_5$ samples were synthesized from Ga flux.\cite{Sarrao02} At the time of the
200: most recent x-ray measurements, one was about 3 years old (sample {\bf A}), and the
201: other was about 1 year
202: old (sample {\bf B}). Superconducting critical temperatures were measured in a Quantum Design Magnetic Properties Measurement System as the point at which diamagnetism was observed in 10 Oe (Fig. \ref{Tc_fig}). The isotopic content
203: is the same for both samples with the main radioactivity coming from
204: 93.93\% $^{239}$Pu,
205: 5.85\% $^{240}$Pu, and
206: 0.12\% $^{241}$Pu.
207: The accumulated dose has been shown to be a reasonable indicator
208: of how sample properties change with time,\cite{Jutier05} and
209: we report the sample age in units of $\alpha$-decays per atom (1 yr=4.9$\times10^{-6}$~$\alpha$-decays per atom in the formula unit). We do not
210: use the more common ``displacements per atom'' (dpa) unit, because it
211: generally assumes the number of displaced atoms (Frenkel pairs) is known
212: from theoretical models, and we show below that this assumption
213: may not be correct.
214:
215: %*****************************************************************************
216: \begin{figure}
217: \includegraphics[width=3.5in]{ks.eps}
218: \caption{(Color online)
219: EXAFS data $k^3\chi(k)$ vs. $k$ from the (a) Pu $L_\textrm{III}$ edge, (b) the Co $K$ edge, and (c) the Ga $K$ edge, for a fresh ({\bf ---}, $0.2\times10^{-6}$~$\alpha$-decays per atom) and an aged ({\bf \textcolor{green}{---}}, 14.5$\times10^{-6}$~$\alpha$-decays per atom) sample.}
220: \label{ks_fig}
221: \end{figure}
222: %*****************************************************************************
223:
224: %*****************************************************************************
225: \begin{figure}
226: \includegraphics[width=3.5in]{rs.eps}
227: \caption{(Color online)
228: Fourier transform (FT) of the $k^3\chi(k)$ data of the fresh sample ({\bf ---}) in Fig. \ref{ks_fig}, together with a fit to these data (\textcolor{red}{{\bf -~-~-}}) . The fit quality is such that the fit is difficult to distinguish from
229: the data. The outer envelope
230: is $\pm$ the transform amplitude and the inner modulating line is the real part of
231: the complex transform. The Pu edge data (a) are transformed between
232: 2.5-16.0 \AA$^{-1}$, Gaussian broadened by 0.3 \AA$^{-1}$, and are fit between 2.0 and 5.0 \AA{}. The Co edge data (a) are transformed between
233: 2.5-11.0 \AA$^{-1}$, Gaussian broadened by 0.3 \AA$^{-1}$, and are fit between 2.0 and 5.5 \AA{}. The Ga edge data (a) are transformed between
234: 2.5-15.0 \AA$^{-1}$, Gaussian broadened by 0.3 \AA$^{-1}$, and are fit between 1.6 and 5.0 \AA{}. }
235: \label{rs_fig}
236: \end{figure}
237: %*****************************************************************************
238:
239: The material was triply contained for EXAFS experiments
240: using epoxy- and indium-sealed kapton windows, and placed into a LHe flow
241: cryostat at $T\leq$30 K. Sample {\bf A} was initially measured (measurements with $\leq 4 \times 10^{-6}$~$\alpha$-decays per atom) as a single crystal in fluorescence mode using 30-element Ge detectors, with the data corrected for dead time and self absorption.\cite{Booth05a} Otherwise the samples were ground for the EXAFS experiments and passed through a
242: 32 $\mu$m sieve, with about 8 mg of this powder mixed with dried boron
243: nitride and packed into an aluminum frame. Transmission and fluorescence data agree quantitatively. This sample mass resulted in a change in absorption
244: across the Pu $L_\textrm{III}$ edge $\Delta \mu_a$ of $\sim$0.5 absorption
245: lengths, whereas $\Delta \mu_a \sim$0.3 and $\Delta \mu_a \sim$2.2 absorption
246: lengths across the Co and Ga $K$ edges, respectively.
247: EXAFS spectra were collected at the
248: Stanford Synchrotron
249: Radiation Laboratory on beamlines 10-2 and 11-2 over a three year period, at
250: the Pu $L_\textrm{III}$, Co $K$, and Ga $K$ edges, generally using a
251: half-tuned, double crystal Si(220) monochromator. The monochromator resolution
252: was adjusted such that it was well below the
253: core-hole lifetime at a given edge.
254: The data were analyzed using standard
255: procedures.\cite{Li95b} In particular, the embedded atom absorption $\mu_0$
256: was determined using a cubic spline with between 4 and 6 knots over the data
257: range, which was typically about 1 keV above the absorption threshold, $E_0$, as determined by the position of the half-height of the absorption change at the edge.
258: The data were fit in $r$-space using the RSXAP package\cite{RSXAP} with theoretical scattering functions generated by FEFF7.\cite{FEFF7}
259:
260: %*****************************************************************************
261: \begin{table*}
262: \caption{
263: EXAFS fit results for the Pu $L_\textrm{III}$ and the Co $K$ edges on a fresh (0.2$\times10^{-6}$~$\alpha$-decays per atom) sample of PuCoGa$_5$. Fit and transform ranges are listed in Fig. \ref{rs_fig}. All single-scattering peaks
264: within the fit range are included in these fits. Multiple scattering was only included in the Ga edge fits to avoid errors originating from peak overlap, but the results from these scattering paths are in themselves unreliable and so are not reported. Coordination numbers, $N$, are held fixed to the nominal structure. $S_0^2$, $\Delta E_0$, and $R(\%)$ are 0.89(5), -4.7(3) eV
265: and 3.6\% for the Pu edge, 0.85(5), 5(1) eV, and 6.1\% for the Co edge, and 0.85(3), -1.7(4), and 6.9\% for the Ga edge,
266: respectively. The number of free parameters in the fits are 10, 14, and 15 for the Pu, Co , and Ga edges, respectively, and are far below the number of independent data points as given by Stern's rule.\cite{Stern93} The quoted error on each quantity is the greater of that obtained by comparisons
267: to standard materials,\cite{Li95b} repeated measurements, and a Monte-Carlo method.\cite{Lawrence01} The nature of the Ga local environment required constraints on several parameters to obtain meaningful fits, as indicated. Constraints on the pair distances in the Ga edge fits assume a tetragonal 115 structure such that Ga(1) sits at the center of the Pu $ab$ face and a plane of Co atoms splits a plane of Ga(2).
268: }
269: \begin{ruledtabular}
270: \begin{tabular}{lcccccc}
271: & $N$ & $\sigma^{2}$ (\AA$^2$) & $R$ (\AA) & $R_\textrm{diff}$\cite{Sarrao02} (\AA) &
272: $\Theta_\textrm{cD}$ (K) & $\sigma_\textrm{stat}^2$ (\AA$^2$) \\
273: \colrule
274: Pu-Ga(1)/Ga(2) & 12
275: & 0.00174(9) & 2.97(1) & 2.993 & 330(20) & -0.0001(2) \\
276: Pu-Co & 2
277: & 0.0019(4) & 3.38(2) & 3.393 & 420(30) & 0.0002(2)\\
278: Pu-Pu & 4
279: & 0.0016(2) & 4.21(2) & 4.232 & 270(20) & 0.0004(2)\\
280: Pu-Ga(2) & 24
281: & 0.0038(4) & 5.16(3) & 5.165 & 270(20) & 0.0012(4)\\
282: \\
283: Co-Ga(2) & 8
284: & 0.0019(1) & 2.45(1) & 2.471 & 250(20) & 0.0004(2)\\
285: Co-Pu & 2
286: & 0.005(3) & 3.38(5) & 3.393 & & \\
287: Co-Co & 4
288: & 0.007(7) & 4.18(8) & 4.232 & & \\
289: Co-Ga(1) & 8
290: & 0.006(1) & 4.5(1) & 4.524 & & \\
291: Co-Ga(2) & 16
292: & 0.0030(6) & 4.84(2) & 4.901 & & \\
293: \\
294: Ga(1)-Pu & 4 & 0.0022(2)\footnotemark[1] & 2.97(1)\footnotemark[2]\footnotemark[7] & 2.993\\
295: Ga(1)-Ga(2) & 8 & 0.0053(2)\footnotemark[3] & 2.95(1)\footnotemark[4] & 2.993\\
296: Ga(1)-Ga(1) & 8 & 0.0051(2)\footnotemark[5] & 4.20(2)\footnotemark[6]\footnotemark[7]\footnotemark[8] & 4.232\\
297: Ga(1)-Co & 8 & 0.0023(5) & 4.51(1) & 4.524\\
298: Ga(2)-Co & 2 & 0.017(2) & 2.43(1)\footnotemark[8] & 2.471\\
299: Ga(2)-Ga(2) & 1 & 0.0026(2) & 2.45(1)\footnotemark[8] & 2.552\\
300: Ga(2)-Ga(2) & 4 & 0.0053(2)\footnotemark[3] & 2.97(1)\footnotemark[2] & 2.993\\Ga(2)-Ga(1) & 2 & 0.0053(2)\footnotemark[3] & 2.95(1)\footnotemark[4] & 2.993\\Ga(2)-Pu & 8 & 0.0022(2)\footnotemark[1] & 2.95(1)\footnotemark[4] & 2.993\\
301: Ga(2)-Ga(2) & 4 & 0.0051(2)\footnotemark[5] & 3.94(2) & 3.933\\
302: Ga(2)-Ga(2) & 5 & 0.0051(2)\footnotemark[5] & 4.20(2)\footnotemark[6] & 4.232\\
303: \end{tabular}
304: \end{ruledtabular}
305: \footnotetext[1]{$^{-f}$like symbols held equal}
306: \footnotetext[7]{$r_\textrm{Ga(1)-Ga(1)}=\sqrt{2}r_\textrm{Ga(1)-Pu}$}
307: \footnotetext[8]{$r_\textrm{Ga(2)-Ga(2)}=\sqrt{4r_\textrm{Ga(2)-Co}^2-r_\textrm{Ga(1)-Ga(1)}^2}$}
308: \label{fit_tbl}
309: \end{table*}
310: %*****************************************************************************
311:
312: %*****************************************************************************
313: %*****************************************************************************
314: \section{Analysis and Results}
315: \label{Results}
316:
317: Figure \ref{ks_fig} shows an example of the normalized oscillations in the absorption above each measured edge as a function of $k$, the photoelectron wave vector, for the fresh (0.2$\times 10^{-6}$~$\alpha$-decays per atom) and 3 year old (14.5$\times 10^{-6}$~$\alpha$-decays per atom) samples. A Fourier transform (FT) of such data (Fig. \ref{rs_fig})
318: produces peaks in the amplitude as a function of the distance $r$ from the absorbing atomic species corresponding to neighboring atoms. For instance, the dominant peak in the Pu edge data (Fig. \ref{rs_fig}a) corresponds to the 12 Pu-Ga neighbors at $\sim$3.0 \AA.
319: Note that the atomic scattering functions generate complicated lineshapes, causing shifts in the peak positions from the actual structure that are well
320: reproduced by calculations using FEFF7.\cite{FEFF7} Information about the local
321: structure is therefore obtained by fitting these calculated scattering functions
322: to these data. The results of fits to data from fresh material that has not undergone a significant amount of $\alpha$-decay are reported in Fig. \ref{rs_fig}
323: and Table \ref{fit_tbl}. The fit quality is excellent, and all the measured
324: pair distances agree well with diffraction results. We note that the local
325: environment around Ga is more complicated than around Pu or Co, and therefore
326: constraints were necessary to reduce the number of fit parameters while still
327: obtaining high quality fits. Unfortunately, constraints add an unknown amount
328: of systematic error that is not reflected in the estimated errors, and we
329: ascribe discrepancies between diffraction and EXAFS results from the Ga $K$-edge
330: data to this source. Where the data are of sufficient quality, the temperature
331: dependence of the mean-squared displacements of the pair distances,
332: $\sigma^2$'s, were obtained and are well described by a
333: correlated-Debye model\cite{Beni76} with reasonable values of the
334: correlated-Debye temperatures, $\Theta_\textrm{cD}$'s, (for comparison to
335: $\delta$-Pu, see Ref. \onlinecite{Nelson03}), and no
336: evidence of static disorder from the fitted offsets, $\sigma_\textrm{stat}^2$'s.
337:
338: Significant radiation damage effects are readily apparent in the raw data with
339: a marked decrease in the overall amplitude of the spectra as samples are aged
340: up to 3 years (Fig. \ref{rsall_fig}). Preliminary fits to the data from the
341: aged samples showed that the decrease in amplitude in the data is due both to
342: a decrease in the overall scale factor $S_0^2$ and an increase in each atom
343: pair's $\sigma^2$ with age. This situation indicates that there are at least
344: three distinct regions within the aged samples: virtually undamaged, strongly
345: damaged where the distance widths $\sigma_\textrm{s}^2$ are large enough that
346: the local structure no longer contributes to the EXAFS amplitude, and mildly
347: damaged where the $\sigma_\textrm{m}^2$'s only allow for a weak contribution.
348: The latter regions may exist, for instance, on the edges of strongly damaged
349: regions. We therefore describe $f_\textrm{tot}$ as due to the sum of the
350: strongly damaged fraction $f_\textrm{s}$ and the mildly damaged fraction
351: $f_\textrm{m}$. Since EXAFS amplitudes $A\sim1/\sigma$, as long as
352: $\sigma_\textrm{m}$ is large enough,
353: $f_\textrm{tot}\approx 1- S_0^{2\prime}(t)/S_0^2(0)$, where $S_0^{2\prime}(t)$
354: is obtained from fits where $\sigma^2(t)$ are fixed at $\sigma^2(0)$. That is,
355: the total damage fraction in aged samples can be estimated by fixing
356: most fitting parameters to those obtained from fresh-sample fits, and
357: estimating $f_\textrm{tot}$ by the change in amplitude as given by $S_0^2$.
358: These damage fractions are shown in Fig. \ref{dam_fig} and demonstrate strong
359: damage production at apparently different rates for each atomic species.
360:
361: %*****************************************************************************
362: \begin{figure}
363: \includegraphics[width=3.5in]{rs_all.eps}
364: \caption{(Color online)
365: FT of the $k^3\chi(k)$ EXAFS data from the
366: (a) Pu $L_\textrm{III}$, (b) Co $K$, and the (c) Ga $K$ edges are shown
367: for sample ages about one year apart. Samples have accumulated doses of 0.2$\times 10^{-6}$~$\alpha$-decays per atom (---), 5.4$\times 10^{-6}$~$\alpha$-decays per atom (\textcolor{red}{---}), 10.2$\times 10^{-6}$~$\alpha$-decays per atom (\textcolor{blue}{---}), and 14.5$\times 10^{-6}$~$\alpha$-decays per atom (\textcolor{green}{---}). Transform ranges are
368: between 2.5-16.0 \AA$^{-1}$, 2.5-10.0 \AA$^{-1}$, and 2.5-14.5 \AA$^{-1}$,
369: respectively, all Gaussian broadened by 0.3 \AA$^{-1}$.}
370: \label{rsall_fig}
371: \end{figure}
372: %*****************************************************************************
373:
374: Using this estimate of $f_\textrm{tot}$, $f_\textrm{s}$ is approximated by performing a fit where the $\sigma^2(t)$'s are
375: no longer constrained, in which case
376: $f_\textrm{s}\approx(S_0^2(t)-S_0^{2\prime}(t))/S_0^2(0)$.
377: Correlations between the $S_0^2$ and $\sigma^2$ parameters are more difficult
378: to control in such a procedure, but fits to Pu edge data indicate that between 60-80\% of the damage is due to the strongly damaged regions and
379: $\sigma_\textrm{m}$ is enhanced by $\sim$0.010-0.015~\AA$^2$ over the well-ordered regions for all of the aged samples measured.
380: Similar results are obtained from the Co and Ga edge data.
381:
382: %*****************************************************************************
383: \begin{figure}
384: \includegraphics[width=3.5in,trim=0 0 0 0]{raddam.eps}
385: \caption{(Color online)
386: The total fraction of displaced atoms as a function of the
387: number of $\alpha$-decays per atom, as seen from
388: each atomic species.
389: Error bars are based on reproducibility. Absolute errors are about $\pm$5\%.
390: The line shows percolation behavior based on the
391: first two aged Pu data points for sample {\bf A}, and the hashed area gives the theoretical range of damage up to the amount expected in the absence of defect migration.
392: }
393: \label{dam_fig}
394: \end{figure}
395: %*****************************************************************************
396:
397: %*****************************************************************************
398: %*****************************************************************************
399: \section{Discussion}
400: \label{Discussion}
401:
402: The accumulated damage after one year in sample {\bf A} is measured as between 20-30\% from the point of view of each of the constituent atoms in PuCoGa$_5$. This value is an order of magnitude higher than theoretical
403: estimates that only account for defects, or closer to two orders of magnitude if one allows for defect migration
404: after defect formation.\cite{Rubia99} It is important to note at this point that the damage enhancement, while possibly indicating more defects than expected, may also indicate significantly more defect-induced lattice strain and
405: distortions. In fact, this amount of damage is so large that, assuming a
406: $\sim$10 nm damage cascade as expected for $\delta$-Pu, \textit{every} atom in
407: the cascade is displaced from its equilibrium position. Modern molecular
408: dynamics calculations should, in principle, generate significant lattice
409: distortions around defects induced by radiation damage, but this effect has not,
410: to our knowledge, been reported in the literature for intermetallics. In
411: addition, the observed damage does not proceed at as fast a rate as
412: extrapolated from lower accumulated doses. This disagreement is very
413: likely due to self-annealing\cite{Fluss04} caused by the
414: room-temperature storage of the samples between measurements. Self-annealing is also likely the cause of the differences between the samples (sample {\bf B} appears to be damaging at a somewhat slower rate), as the exact history of the storage conditions for each sample then becomes important. In addition, differences between the constituent atomic species indicate different defect production or migration rates. These issues should be studied in other materials to further explore the role of self-annealing and atomic-species effects.
415:
416: Another way of
417: describing radiation damage is to consider that a decay event generates enough
418: heat within the damage cascade that the material locally melts and then
419: rapidly resolidifies, thus quenching disorder into the cascade region
420: from the high temperature state
421: and potentially also creating a distribution of competing structural phases.
422: This picture is similar to the ``thermal spike'' model as originally proposed
423: by Seitz,\cite{Seitz49} although it proved to be calculationally
424: intractable.\cite{Kinchin55}
425: Applying molecular dynamics after (or during) Frenkel defect production seems
426: to combine the relevant aspects of these two schemes, and could, in principle,
427: generate the sort of distortions measured here in PuCoGa$_5$. Simulated EXAFS
428: data could be calculated from the results of future molecular
429: dynamics calculations on PuCoGa$_5$ for a direct comparison of the efficacy of
430: such models.
431:
432: These data largely explain the fast reduction of $T_c$ in PuCoGa$_5$, with
433: $T_c/T_{c0} \approx 50$\% when $f_\textrm{tot} \approx 40\%$. This value is now
434: greater than the lower-limit damage fraction of
435: $f_\textrm{BdG}(T_c/T_{c0}=50\%) \approx$15\%,\cite{Franz97} as expected since
436: not all damaged regions will create strong scattering. In addition, it is
437: likely that defects outside the superconducting planes scatter more weakly
438: than defects on the in-plane sites.\cite{Petrovic02,Daniel05b} Decreasing the
439: effective scattering strength would increase the necessary impurity densities
440: in the theory.
441:
442: Also shown in Fig. \ref{dam_fig} is the prediction of a cubic percolation model.
443: The time axis is chosen such that the model
444: agrees with the first aged Pu edge data points for sample {\bf A}. Because of the strong damage that occurs within a cascade, the superconducting
445: fraction is likely more closely related to the fraction of the
446: material that exists within the volume between the edges of damage cascades,
447: which is better described by the percolation model depicted in
448: Fig. \ref{dam_fig}. According to this extrapolation one might expect
449: superconductivity to cease between 3.5 and 4 years for these samples
450: ($f_\textrm{tot}\sim0.8$), roughly consistent with the data in
451: Fig. \ref{Tc_fig}, although no samples have yet been observed to become
452: non-superconducting at this time. This
453: simplification, of course, doesn't account for any proximity effects, which
454: should increase this time period, or any increased impurity scattering, which
455: would decrease this time.
456:
457: The emerging physical picture of PuCoGa$_5$ is one where the recoiling U nucleus
458: generates much more damage than expected based on models of
459: elemental Pu. This damage is likely dominated by near-neighbor lattice
460: distortions that extend into the second coordination sphere or beyond,
461: possibly generating local distributions of impurity phases. This damage
462: is so severe that it encompasses all the atoms in a given damage cascade. The
463: effective damage rate is slowed as the material anneals at room temperature.
464: Photoemission results indicate that the 5$f$ electrons have both local and itinerant character in PuCoGa$_5$.\cite{Joyce03} The partially localized $f$ electrons in the well-ordered material are likely further localized in the damaged regions, in analogy to $\delta$-Pu.\cite{McCall06} This view is consistent with x-ray absorption near-edge measurements on aged PuCoGa$_5$ samples.\cite{Booth06} Radiation damage therefore probably creates non-superconducting material both due to enhanced localization and strong defect scattering in these regions, although a proximity
465: effect could still allow some superconductivity. Annealed
466: areas within a damage cascade probably would not be superconducting
467: due to their limited size, unless they reach the edge of a cascade.
468:
469: These data support and extend the conclusion of
470: Farnan \textit{et al.}\cite{Farnan07,Farnan01}
471: that radiation damage occurs at a much faster rate than current theoretical
472: predictions. In their work, the measured damage production rate is about 5 times
473: higher than the theoretical prediction for the unrelaxed defect production rate
474: in zircons. In the present work, we find a production rate at least 10 times faster
475: than the prediction in an intermetallic, and observe
476: deviations from a percolation model that we ascribe to annealing effects. The
477: role of annealing should be carefully considered in studies of zircons and related
478: potential nuclear-waste storage materials.
479:
480: %*****************************************************************************
481: %*****************************************************************************
482: \section{Conclusion}
483: \label{Conclusion}
484:
485: Local structure data on samples of PuCoGa$_5$ demonstrate a well-ordered local lattice structure that agrees with the long-range average structure obtained by diffraction measurements. After the sample has aged long enough to accumulate a significant total number of $\alpha$-decays, the local structure exhibits strong disorder, primarily through a reduction in the amplitude of the EXAFS oscillations, but also in the pair-distance distribution variances $\sigma^2$'s. This disorder affects between 20-30\% of sample {\bf A} after one year ($\sim4\times10^{-6}$~$\alpha$-decays per atom), followed by a somewhat slower damage accumulation rate. Theoretical estimates that only account for defects predict at least an
486: order of magnitude less damage, not including the damage-reducing factors of
487: self-annealing and defect migration. These data help explain the fast reduction
488: of the superconducting transition temperature both in terms of defect
489: scattering and a simple percolation model.
490:
491: These results underscore the need for more local structure studies of radiation damage in general
492: and especially in PuCoGa$_5$. Only through better theoretical models and
493: atomic-level probes can we understand the detailed
494: electronic and structural properties of damaged regions and how they couple
495: to superconductivity. In particular, direct comparisons between damage
496: cascade structures and local structure measurements should be pursued.
497: Such studies will have ramifications
498: not only for understanding superconductivity in
499: PuCoGa$_5$, but also the unusual properties of $\delta$-Pu, and the field of radiation damage in general.
500:
501: %*****************************************************************************
502: %*****************************************************************************
503: \section*{Acknowledgments}
504:
505: We thank M. Fluss, M. Graf, A. Kubota, L. Soderholm, J. Thompson and W. Wolfer for enlightening discussions and
506: W.-J. Hu for assistance in loading one of the plutonium samples. Supported
507: by the
508: U.S. Department of Energy (DOE) under Contract No. DE-AC02-05CH11231.
509: X-ray absorption data were collected at the Stanford Synchrotron Radiation
510: Laboratory, a national user facility operated by Stanford University on
511: behalf of the DOE/OBES. Work at Los Alamos was performed under the auspices of the U. S. DOE.
512:
513: \bibliography{../bib/bibli}
514:
515: \end{document}
516: