1: \documentclass[prb,superscriptaddress,twocolumn,showpacs,aps]{revtex4}
2:
3: %This next line is good for preprints
4: %\documentclass[preprint,endfloats*,prb,superscriptaddress,showpacs,aps]{revtex4}
5:
6: \usepackage{graphicx}
7: \usepackage{dcolumn}
8: \usepackage{amsmath}
9:
10:
11: \begin{document}
12:
13: \title{Neutron scattering study of crystal fields in CeRhIn$_{5}$}
14: \author{A. D. Christianson}
15: \affiliation{Los Alamos National Laboratory, Los Alamos NM 87545}
16: \affiliation{Colorado State University, Fort Collins CO 80523}
17: \email{achristianson@lanl.gov}
18: \author{J. M. Lawrence}
19: \affiliation{University of California, Irvine, CA 92697}
20: \author{P. G. Pagliuso}
21: \author{N.O. Moreno}
22: \author{J. L. Sarrao}
23: \author{J. D. Thompson}
24: \affiliation{Los Alamos National Laboratory, Los Alamos NM 87545}
25: \author{P. S. Riseborough}
26: \affiliation{Temple University, Philadelphia, PA 19122}
27: \author{S. Kern}
28: \affiliation{Colorado State University, Fort Collins CO 80523}
29: \author{E. A. Goremychkin}
30: \affiliation{Argonne National Laboratory, Argonne IL 60439}
31: \author{A. H. Lacerda}
32: \affiliation{Los Alamos National Laboratory, Los Alamos NM 87545}
33:
34: \date{\today}
35:
36: \begin{abstract}
37: Neutron scattering results for the tetragonal compound
38: CeRhIn$_{5}$ give evidence for two crystal field (CF) excitations
39: at 6.9 and 23.6 meV. The scattering can be fit assuming a set of
40: CF parameters B$^{0}_{2}$ = -1.03 meV, B$^{0}_{4}$ = 0.044 meV and
41: B$^{4}_{4}$ = 0.122 meV. To compare our results to previous work,
42: we calculate the susceptibility and specific heat for this CF
43: scheme, including a molecular field term $\lambda = $35 mol/emu to
44: account for the Kondo effect. We also include a calculation based
45: on these CF parameters that uses the non-crossing approximation to
46: the Anderson model to estimate the effect of Kondo physics on the
47: susceptibility, specific heat and neutron linewidths.
48: \end{abstract}
49:
50: \pacs{75.30.Mb 75.20.Hr 71.27.+a 71.28.+d 61.10.Ht }
51:
52: \maketitle
53:
54:
55: CeRhIn$_{5}$ crystallizes in the same tetragonal HoCoGa$_{5}$
56: structure as the heavy fermion superconductors CeIrIn$_{5}$ and
57: CeCoIn$_{5}$\cite{hegger,p1,p2}. At ambient pressure CeRhIn$_{5}$
58: undergoes a transition to an antiferromagnetic (AF) state at
59: $T_{N} =$ 3.8 K\cite{hegger,Bao}. With application of hydrostatic
60: pressure the N\'{e}el temperature remains essentially constant
61: until antiferromagnetism disappears and superconductivity appears
62: at pressures above 15 kbar\cite{hegger}. Recently, Pagliuso
63: \textit{et al.}\cite{Pagliuso} have suggested the importance of CF
64: splitting to the ground state properties of the CeMIn$_{5}$ family
65: of heavy fermion superconductors, underscoring the fact that the
66: ultimate ground state achieved by a particular member of the
67: family must grow out of the ground state crystal field doublet.
68: Thus a careful determination of both the CF splitting and
69: wavefunctions is important. To that end we have begun to directly
70: probe the CF energy level splitting in the CeMIn$_{5}$ family
71: using inelastic (IE) neutron scattering. The first step in our
72: investigations has been determining the crystal field level scheme
73: in CeRhIn$_{5}$.
74:
75: \begin{figure}[t]
76: \includegraphics[width=3.3in]{Fig1.EPS}
77: \caption{Neutron energy spectra of (a) CeRhIn$_{5}$ and (b)
78: LaRhIn$_{5}$ at an initial energy E$_{i}$ = 35 meV, at 8 K and for
79: two mean scattering angles, 20$^{\circ}$ and 100$^{\circ}$. The
80: data have been corrected for neutron absorption and the scattering
81: from the sample holder has been subtracted from the data. (c) The
82: $Q$=0 magnetic scattering, determined as described in the text, in
83: CeRhIn$_{5}$ at 8K and for three incident energies $E_{i}$.
84: \label{fig1}}
85: \end{figure}
86:
87: In CeRhIn$_{5}$, as in the other members of CeMIn$_{5}$ family,
88: the crystal field Hamiltonian in tetragonal symmetry can be
89: written
90: \begin{equation*}
91: H_{CF}=B_{2}^{0}O_{2}^{0}+B_{4}^{0}O_{4}^{0}+B_{4}^{4}O_{4}^{4}
92: \end{equation*}
93: where $O_{l}^{m}$ and $B_{l}^{m}$ are the Stevens operators and CF
94: parameters respectively. The Ce$^{3+}$ $J=5/2$ wavefunction
95: splits into three doublets, $\Gamma _{7}^{(1)}=\{\alpha \lbrack
96: \pm 5/2\rangle +\beta \lbrack \mp 3/2\rangle \}$, $\Gamma
97: _{7}^{(2)}=\{\beta \lbrack \pm 5/2\rangle -\alpha \lbrack \mp
98: 3/2\rangle \}$ and $\Gamma _{6}=[\pm 1/2\rangle $\cite{Pagliuso}.
99: \ An analysis of susceptibility and thermal expansion
100: results\cite{Takeuchi} suggested crystal field levels $\Gamma
101: _{7}^{(2)}$, $\Gamma _{7}^{(1)}$ and $\Gamma _{6}$ at $E$=0, 5.86
102: meV (68 K) and 28.43 meV (300 K) respectively, with $\beta=0.969$
103: (yielding a nearly pure $\lbrack \pm 5/2\rangle$ ground state). A
104: subsequent study\cite{Pagliuso} based on an analysis of the
105: susceptibility and specific heat suggested a similar scheme, but
106: with splittings 6 and 12 meV (70 and 140 K). In this paper we
107: report the results of an analysis of neutron scattering data for
108: CeRhIn$_{5}$ which indicate that these initial estimates are
109: nearly correct; our results have somewhat different values for the
110: splittings and a smaller value for the mixing parameter $\beta$,
111: i.e., a greater admixture of $\lbrack \mp 3/2\rangle$ into the
112: $\lbrack \pm 5/2\rangle$ ground state. To assist in comparison of
113: our results to those of Pagliuso \textit{et al.}\cite{Pagliuso}
114: and Takeuchi \textit{et al.}\cite{Takeuchi}, we report
115: calculations of the specific heat and magnetic susceptibility
116: based on our CF parameters which include the Kondo effect in an ad
117: hoc manner similar to those of refs. 5 and 6. We also present
118: more sophisticated calculations that employ the non-crossing
119: approximation (NCA)\cite{BCW,crossover} to the Anderson model in
120: order to estimate the effect of Kondo spin fluctuations on the
121: susceptibility, specific heat and IE neutron spectra.
122:
123: \begin{figure}[two]
124: \includegraphics[width=3.3in]{Fig2.EPS}
125: \caption{Temperature dependence of the magnetic part of the IE
126: neutron scattering response of CeRhIn$_{5}$ for $E_{i}$ = 35 meV.
127: The scattering dependence due to the Ce$^{3+}$ form factor has
128: been removed as in the previous figure. The data at all three
129: temperatures ((a) 8 K, (b) 70 K, and (c) 140 K) have been fit
130: simultaneously (solid lines) with a least squares fitting routine
131: to determine the crystal field parameters. The results of the
132: fitting parameters including the crystal field parameters are
133: displayed in table \ref{table1}. We have included in (a) the
134: results of the NCA calculation (dashed line).}
135: \end{figure}
136:
137: Large high quality single crystals of CeRhIn$_{5}$ and
138: LaRhIn$_{5}$ were obtained using the flux-growth
139: method\cite{hegger}. For the magnetic susceptibility and specific
140: heat measurements, single crystals were carefully prepared which
141: were free of residual In flux; in the case of the neutron
142: scattering measurements, $\sim$50 g of single crystals for both
143: CeRhIn$_{5}$ and LaRhIn$_{5}$ were powdered.
144: The neutron scattering experiments were performed in time-of-flight
145: mode using LRMECS at IPNS (Argonne National Laboratory) with experimental
146: conditions that were similar to that of an earlier report\cite{Lawrence}.
147: A key problem in our investigations was the high neutron
148: absorption of both In and Rh. In initial experiments the standard
149: LRMECS sample holder was used; however, in subsequent experiments
150: a new sample holder was employed which was designed to maintain a
151: more uniform sample thickness than the standard holder, thus
152: allowing for a more accurate absorption correction.
153: Neutron scattering spectra were collected for several different
154: incident energies ($E_{i}$) and temperatures between 8 and 140 K
155: with counting times ranging from 24 to 48 hours. To improve
156: statistics, we were able to take advantage of the nondispersive
157: nature of the CF scattering and group detectors into three bins
158: with mean scattering angle 20$^{o}$ (low $Q$), 60$^{o}$ and
159: 100$^{o}$ (high $Q$). A Vanadium standard was utilized to put the
160: scattering on an absolute scale.
161:
162: Data for CeRhIn$_{5}$ and LaRhIn$_{5}$ (measured to help identify
163: the nonmagnetic scattering in CeRhIn$_{5}$) at 8 K and $E_{i} = $
164: 35 meV for low and high Q are shown in Fig. \ref{fig1}. The data
165: were corrected for absorption assuming a uniformly thick
166: flat-plate sample and the scattering due to the empty sample
167: holder was then subtracted from the data. Direct comparison of
168: low angle scattering (where magnetic scattering is strongest) for
169: CeRhIn$_{5}$ (Fig. \ref{fig1}a) and LaRhIn$_{5}$ (Fig.
170: \ref{fig1}b) shows two additional peaks near 7 and 23 meV. In
171: particular, we determine the nonmagnetic scattering in
172: CeRhIn$_{5}$ in two ways: 1) By using the expression
173: $S_{mag}(20^{\circ})=S(Ce,20^{\circ})-f S(La,20^{\circ})$ where we
174: choose the factor $f$ as the ratio (0.75) of the total scattering
175: cross-sections $\sigma$(CeRhIn$_{5}$)/$\sigma$(LaRhIn$_{5}$). 2)
176: By determining the ratio $R=S(La,100^{\circ})/S(La,20^{\circ})$
177: for scaling the high angle nonmagnetic scattering to low
178: angle\cite{Goremychkin}. Excellent agreement with 1) is obtained
179: using $S_{mag}(20^{\circ})=S(Ce,20^{\circ})-F S(Ce,100^{\circ})/R$
180: with inclusion of an additional factor F = 1.33 to account for the
181: difference in Q-scaling of the La and Ce compounds. The value of F
182: is similar to the one used in a recent study of
183: YbXCu$_{4}$\cite{crossover}; it can be justified on the basis that
184: for high angle scattering the data are predominantly
185: single-phonon, proportional to $\sigma$, while the low angle
186: scattering contains a significant contribution from multiple
187: scattering (one elastic and one phonon) proportional to
188: $\sigma^{2}$, so that the cross section does not cancel in the
189: ratio. Results of this analysis for three different $E_{i}$ are
190: shown in Fig. 1c. The dependence of the scattering on the
191: Ce$^{3+}$ form factor has been removed in this plot, so the data
192: represent the $Q = $ 0 scattering with the assumption that the
193: crystal fields are in fact purely local and uncoupled entities.
194: The data have been truncated below 0.15$E_{i}$ (where the elastic
195: line dominates the scattering) and above 0.8$E_{i}$, where
196: statistics are small due to the $k_{f}/k_{i}$ factor. Good
197: agreement is evident for data taken at three different $E_{i}$,
198: with all data sets displaying magnetic excitations at
199: approximately 7 and 24 meV.
200:
201:
202: \begin{table}[tbp]
203: \caption{ Crystal field parameters $B_{l}^{m}$, splittings and
204: Lorentzian halfwidths $\Gamma$ of the IE excitations at four
205: temperatures for CeRhIn$_{5}$ and the wave function mixing
206: parameter $\beta$. The units of all quantities (except for
207: $\beta$, which is unitless) are meV. The reduced Chi-square for
208: the fit was $\chi^{2} = 0.69$} \label{table1}
209: \begin{tabular}{lll}
210: $B_{2}^{0}$ &$B_{4}^{0}$ &$B_{4}^{4}$\\
211: \tableline -1.03$\pm$0.02 & 0.044$\pm$0.001 &
212: 0.122$\pm$0.003 \\
213: \tableline & &\\
214: $E(\Gamma_{7}^{1})$ &$E(\Gamma_{6})$ &$\beta$\\
215: \tableline 6.9$\pm0.3$ & 23.6$\pm0.5$ & 0.80$\pm$0.02\\
216: \tableline & &\\
217: $\Gamma (8K)$ & $\Gamma (70K)$ & $\Gamma (140K)$
218: \\
219: \tableline 2.3$\pm$0.1 & 2.9$\pm$0.2 & 4.2$\pm$0.4\\
220: \tableline \\
221: \end{tabular}
222: \end{table}
223:
224:
225: In Fig. 2 we plot the $Q = $ 0 (form factor removed) magnetic
226: scattering (method 1), determined at $E_{i}= $ 35 meV, for three
227: different temperatures. We have performed a simultaneous least
228: squares fit to four datasets (8 K, 70 K and 140 K at $E_{i} = $ 35
229: meV and 8 K at $E_{i} = $ 80 meV) to determine the CF parameters.
230: The fit includes the effects of instrumental resolution. Variables
231: of the fit include $B_{2}^{0}, B_{4}^{0}, B_{4}^{4}$ and an
232: overall scale factor (which four parameters were constrained to
233: the same values for all datasets) and the Lorentzian halfwidth
234: $\Gamma$ of the IE excitations which was allowed to vary with
235: temperature. (We constrained the quasi-elastic (QE) halfwidth to
236: 1/2 $\Gamma$.) Results of the fit are shown in Table \ref{table1}
237: and plotted in Fig. 2.
238:
239:
240: To compare our results to those of Pagliuso \textit{et
241: al.}\cite{Pagliuso} and Takeuchi \textit{et al.}\cite{Takeuchi},
242: we have calculated the susceptibility and specific heat (Fig. 3).
243: The susceptibility includes a positive molecular field
244: contribution $\lambda = $ 35 mol/emu where $\lambda$ represents
245: contributions to $1/\chi$ from AF and Kondo fluctuations. At high
246: T these contribute to $1/\chi$ as $(T_{K}+T_{N})/C_{J}$; with
247: $C_{J}$ = 0.807 emu-K/mol for $J=5/2$ and $T_{N}$ = 3.8 K this
248: gives $T_{K} \sim$ 25 K. We note that this value of $k_{B}T_{K}$
249: is similar to the width of the 7 meV IE excitation at 8 K. The
250: calculation for the specific heat contains both a Schottky term
251: due to the excited levels and a Kondo doublet term\cite{Rajan}
252: with $T_{K} = $ 25 K for the ground state level, which puts the
253: calculated specific heat in the range 20-50 K in better agreement
254: with measured value -- without this, the calculated value due only
255: to the Schottky contribution is smaller by a factor of 0.8. We
256: have not attempted to fit for the effects on $C_{mag}$ and $\chi$
257: of the AF transition at 3.8 K.
258:
259: \begin{figure}[where]
260: \includegraphics[width=3.2in]{Fig3.EPS}
261: \caption{ (a) Measured anisotropic susceptibilities $\chi^{zz}$
262: (triangles) and $\chi^{xx}$ (circles) for CeRhIn$_{5}$ compared to
263: the value calculated for the CF parameters of Table \ref{table1}
264: with a molecular field contribution $\lambda =$ 35 mol/emu (solid
265: lines) and compared to the results of the NCA calculation (dashed
266: lines). (b) Magnetic specific heat compared to the value
267: calculated for a Schottky contribution from the excited levels and
268: a Kondo contribution from the ground state doublet (solid line)
269: and to the results of the NCA calculation (dashed line).}
270: \end{figure}
271:
272: A more sophisticated way to include Kondo spin fluctuations is
273: through calculation for the Anderson model. We present results
274: obtained using the non-crossing approximation
275: (NCA)\cite{BCW,crossover}. We have used a Gaussian background band
276: with density of states $N(\varepsilon
277: )=e^{-(\varepsilon/W)^{2}}/\surd \pi W$ with $W$ = 3 eV and we set
278: the 4$f$ level position at $E_{f} = 2$ eV and the spin-orbit
279: splitting at $E_{so} = 0.273$ eV, which are standard values for
280: Ce. Since the Kondo physics renormalizes the CF levels upward by
281: an amount approximately equal to the Kondo temperature the bare
282: level energies were chosen to be $E_{b}=5.3$ meV and $E_{c}=23$
283: meV which are smaller than the measured level energies. The mixing
284: parameter $\beta=0.80$ was chosen to be similar to that obtained
285: in Table 1. The hybridization was then varied until a good fit to
286: the anisotropic susceptibility was obtained for $V=0.4665$ eV. The
287: results for $S_{mag}$, $\chi$ and $C_{mag}$ are given in Figs. 2a
288: and 3.
289:
290:
291:
292: We now turn to discussion of the effect of systematic errors on
293: our conclusions. As mentioned previously, the neutron absorption
294: of In and Rh is an important consideration. Comparison of the data
295: for two different sample holders (which exhibited small
296: differences in sample thickness and distribution) indicated
297: similar results, augmenting our belief that the absorbtion
298: correction employed is correct. If the nonmagnetic background
299: subtraction is varied by varying $f$ (method 1) or $F/R$ (method
300: 2), the scattering at 7 meV is relatively unaffected but the
301: strength of the 24 meV scattering, and hence $\beta$, is affected
302: somewhat. Given the good consistency between results at different
303: $E_{i}$ and $T$ we think that our CF scheme is basically correct.
304: We were unable to observe quasi-elastic (QE) scattering, due to
305: the requirement that to obtain the resolution necessary a small
306: E$_{i}$ is required which causes the effects of neutron
307: absorption, which varies as 1/$\surd E$, to become large. In our
308: fits we constrained the QE halfwidth to half the value of the IE
309: width to prevent proliferation of fit parameters. Constraining to
310: other values (e.g. $\Gamma_{QE}=\Gamma_{IE}$) leads only to minor
311: variation in the final fits.
312:
313: Our fits to $\chi$ using the CF parameters plus a molecular field
314: term are not as good as those of Takeuchi \textit{et
315: al.}\cite{Takeuchi} or Pagliuso \textit{et al.}\cite{Pagliuso}.
316: However, their fits use a value of $\beta$ very close to unity,
317: indicating essentially no $[\mp 3/2\rangle$ admixture into $[\pm
318: 5/2\rangle$ ground state. In this case there would be no
319: observable amplitude for the $\Delta m_{z}=1$ transition to $[\pm
320: 1/2\rangle$ state at 24 meV. This cannot be correct as we clearly
321: observe this transition in the neutron scattering data. A
322: possible reason that our fits are not as good as those of refs. 5
323: and 6 is that we do not include the effect of exchange anisotropy,
324: which should only be important below 20 K. Such anisotropy can be
325: mimicked as in Pagliuso \textit{et al.}\cite{Pagliuso} through
326: inclusion of an anisotropic mean field parameter, which we have
327: chosen not to do for simplicity.
328:
329: On the other hand, the NCA calculations based on our CF scheme and
330: a Kondo temperature of order 25 K does an excellent job
331: reproducing $\chi$. However, it overestimates the width of the 7
332: meV excitation as seen in Fig. 2a and underestimates the
333: temperature of the peak in the specific heat (Fig. 3b). These
334: deviations from the data may reflect the fact that we have neither
335: included antiferromagnetic exchange, exchange anisotropy nor
336: anisotropic hybridization (i.e. different hybridization to the
337: different CF multiplets) in the NCA fits.
338:
339: In summary, we find a more significant $[\mp 3/2\rangle$ admixture
340: into $[\pm 5/2\rangle$ ($\beta=0.80$) ground state than found
341: earlier by Takeuchi \textit{et al.} ($\beta=.969$)\cite{Takeuchi}
342: or Pagliuso \textit{et al.} ($\beta\sim1$)\cite{Pagliuso}. The
343: resulting CF level parameters provide reasonable fits to both the
344: magnetic susceptibility and specific heat with the inclusion of a
345: mean field parameter and a Kondo doublet respectively. In
346: addition, NCA fits the susceptibility remarkably well with some
347: deficiencies in both the specific heat and neutron scattering
348: linewidths. Taken together the NCA calculations and the fits to
349: specific heat and susceptibility all indicate a $T_{K} \sim$ 25 K.
350: We note that the ordered moment $g \mu _{B} \langle J_{z} \rangle
351: = $ 0.92$\mu _{B}$ deduced for $\beta =$0.80 is substantially
352: larger than the value 0.36$\mu_{B}$ needed to fit the diffraction
353: pattern in the ordered state\cite{Bao}. For our estimate of
354: $T_{K}$ the moment is reduced by the Kondo physics at temperatures
355: $T > T_{N}$.
356:
357:
358: We acknowledge useful discussions with W. Bao and R.J. McQueeney.
359: Work at UC Irvine was supported by UCDRD funds provided by the
360: University of California for the conduct of discretionary research
361: by the Los Alamos National Laboratory and by the UC/LANL Personnel
362: Assignment Program. Work at Los Alamos and Argonne was performed
363: under the auspices of the DOE. The work at Temple University was
364: supported by the Department of Energy, award DE-Fg02-01ER45827.
365:
366: \begin{references}
367:
368:
369:
370: \bibitem{hegger}H. Hegger, C. Pettrovic, E.G. Moshopoulou, M.F.
371: Hundley, J.L. Sarrao, Z. Fisk, and J.D. Thompson, Phys. Rev. Lett.
372: \textbf{84}, 4986 (2000).
373:
374: \bibitem{p1}C. Petrovic, R. Movshovich, M. Jaime, P.G. Pagliuso,
375: M.F. Hundley, J.L. Sarrao, Z. Fisk, and J.D. Thompson, Europhys.
376: Lett. \textbf{84}, 354 (2001).
377:
378: \bibitem{p2}C. Petrovic, P.G. Pagliuso, M.F. Hundley, R.
379: Movshovich, J.L. Sarrao, J.D. Thompson, and Z. Fisk, J. Phys.:
380: Condens. Matter \textbf{13} (2001) L337-L342.
381:
382: \bibitem{Bao} W. Bao, P. G. Pagliuso, J. L. Sarrao, J. D.
383: Thompson, Z. Fisk, J. W. Lynn and R. W. Erwin, Phys. Rev. B
384: \textbf{62}, R14621 (2000); \textbf{63}, 219901(E) (2001).
385:
386: \bibitem{Pagliuso} P.G. Pagliuso, N.J. Curro, N.O. Moreno, M.F.
387: Hundley, J.D. Thompson, J. L. Sarrao and Z. Fisk, Physica B
388: (PHYSB:23607) to be published.
389:
390: \bibitem{Takeuchi} T. Takeuchi, T. Inoue, K. Sugiyama, D. Aoki,
391: Y. Tokiwa, Y. Haga, K. Kindo and Y. Onuki, Journ. Phys. Soc. Japan
392: \textbf{70}, 877 (2001).
393:
394: \bibitem{BCW} N. E. Bickers, D. L. Cox and J. W. Wilkins, Phys. Rev. B \textbf{36}, 2036 (1987).
395:
396: \bibitem{crossover} J. M. Lawrence, P. S. Riseborough, C. H. Booth, J. L.
397: Sarrao, J. D. Thompson and R. Osborn, Phys. Rev. B \textbf{63},
398: 054427 (2001).
399:
400: \bibitem{Lawrence} J. M. Lawrence, R. Osborn, J. L. Sarrao and Z. Fisk, Phys. Rev. B \textbf{59}, 1134
401: (1999).
402:
403: \bibitem{Goremychkin} E. A. Goremychkin and R. Osborn, Phys. Rev.
404: B \textbf{47}, 14280 (1993).
405:
406: \bibitem{Rajan} V. T. Rajan, Phys. Rev. Lett. \textbf{51}, 1983 (308).
407:
408:
409:
410:
411: \end{references}
412:
413: \end{document}
414: