1: \documentclass[aps,prb,twocolumn,superscriptaddress]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{epsfig}
4: \usepackage{float}
5:
6: \newcommand{\bfg}[1]{\mbox {\boldmath $#1$}}
7: \newcommand{\asize}[1]{\renewcommand{\arraystretch}{#1}}
8: \newcommand{\be}{\begin{equation}}
9: \newcommand{\ee}{\end{equation}}
10: \bibliographystyle{apsrev}
11:
12:
13: \begin{document}
14:
15: \title{Why the magnetic interactions in Na$_{x}$CoO$_{2}$ are 3D}
16: \author{M.D. Johannes}
17: \author{I.I. Mazin}
18: \affiliation{Code 6391, Naval Research Laboratory, Washington, D.C. 20375}
19: \author{D.J. Singh}
20: \affiliation{Condensed Matter Sciences Division, Oak Ridge National
21: Laboratory, Oak Ridge, TN 37831-6032}
22:
23: \begin{abstract}
24: The puzzle of 3D magnetic interactions in the structurally
25: 2D layered oxide Na$_{x}$CoO$_2$ is addressed using first principles
26: calculations and analysis of the exchange mechanisms.
27: The calculations agree with recent neutron
28: results, favoring AFM stacking of FM planes.
29: Superexchange via
30: direct O-O hopping and through intermediate Na $sp^2$ hybrids
31: couples each Co to its nearest and
32: six \textit{next}-nearest interplanar neighbors equally.
33: The individual exchange constants are rather 2D,
34: like the lattice itself,
35: but due to multiple c-axis exchange paths,
36: the magnetism becomes effectively 3D.
37: \end{abstract}
38:
39: \maketitle
40:
41: The layered transition metal oxide (TMO),
42: Na$_{x}$CoO$_{2}$ is attracting considerable
43: interest because of
44: its unusual magnetic and transport properties, and
45: the recently discovered
46: superconductivity in its derivative,
47: Na$_{1/3}$CoO$_{2}\cdot $4/3H$_{2}$O.
48: Experimental signatures of triplet pairing \cite{WHKO+03,AKAK+03,TWCM+} have
49: lead to suggestions that spin
50: fluctuations \cite{DJS03,MDJ+b,KIYI+,ATXH03,TFGZ+}
51: mediate the superconductivity, in analogy with other layered superconducting
52: TMO's like high-T$_c$ cuprates and Sr$_2$RuO$_4$,
53: both of which are close to magnetic instabilities, as is
54: Na$_{x}$CoO$_{2}$. \cite{DJS00,DJS03}
55: However, in contrast to those materials
56: where not only the electronic structure,
57: but also the magnetic interactions are strongly 2D,
58: recent neutron measurements of the magnon
59: dispersion \cite{SPB+04b,LMH+04} in Na$_{x\simeq 0.8}$CoO$_{2}$
60: indicate that the antiferromagnetic (AFM) exchange
61: between CoO$_2$ planes is nearly as
62: strong as the ferromagnetic (FM) in-plane exchange.
63: Since dimensionality plays a key role in unconventional superconductivity,
64: it is important to understand
65: the microscopic physics behind the apparent 3D character
66: of magnetic interactions observed in Refs.
67: \onlinecite{SPB+04b,LMH+04} and to clarify the
68: relationship of the magnetic properties of the parent compound
69: to those of the superconducting hydrate. We
70: address these issues here.
71:
72: Na$_{x}$CoO$_{2}$ has an unusual magnetic phase diagram as
73: a function of $x$, which itself poses interesting,
74: unresolved questions. For most $x$,
75: Na$_{x}$CoO$_{2}$ is a metallic paramagnet,
76: but in a very small range, around $x$=0.5, a charge-ordered,
77: insulating, and possibly
78: antiferromagnetic (AFM) region emerges \cite{MLF+04}.
79: Surprisingly, the metallic states on either side of this
80: region are quite different. For $x < 0.5$
81: the susceptibility is Pauli paramagnet like with weak $T$ dependence,
82: while for $x > 0.5$ it is Curie-Weiss like, suggesting local moments.
83: Finally, for $x$ greater than $\sim 0.75$ a spin density wave condenses,
84: \cite{TMRU+03,JSHI+03,BCS+04,JWDMP+04}
85: with clear antiferromagnetism, $T_N \sim 22K$ at $x$=0.82.
86: While this and the negative Weiss constant \cite{FCC04, JLG+03,YWNSR+03,TMRU+03}
87: suggest antiferromagnetic interactions,
88: experiments also
89: show a characteristically ferromagnetic
90: hysteresis \cite{TMRU+03} as well as predicted \cite{DJS03} in-plane
91: ferromagnetic fluctuations\cite{ATB+03}.
92:
93: These varied data are reconciled by neutron
94: scattering experiments, \cite{SPB+04b,LMH+04} which attribute the
95: $T_{M}$=22K transition to an A-type AFM ordering
96: (FM planes stacked antiferromagnetically along the $c$-axis).
97: By fitting magnon
98: dispersion curves to a linear spin wave model,
99: both groups conclude that in-plane and perpendicular
100: magnetic exchange constants are of
101: comparable magnitude, indicating magnetic isotropy,
102: despite considerable structural two-dimensionality.
103: We will show here that this
104: quasi-isotropy does not imply comparable magnitudes of the
105: nearest neighbor exchange constants in- and out-of-plane, but rather
106: unexpectedly large coupling to next nearest neighbors across the planes,
107: assisted mainly by Na $sp^2$ hybrid orbitals.
108:
109: Our first principles local density
110: approximation (LDA) calculations were done using the augmented
111: plane wave plus local orbital (APW+lo) and linearized augmented
112: planewave (LAPW) methods implemented in two codes.
113: \cite{Lnote,Wien2k}.
114: The experimental lattice parameters,
115: $a$=2.828\AA, $c$=10.94\AA, and LDA relaxed
116: oxygen height $z_{\rm O}$=0.0859
117: were used for all calculations.\cite{Onote}
118: The partial occupation of Na was treated both by the
119: virtual crystal approximation (VCA) and in supercells.
120: In the VCA, each $2d$ Wyckoff position
121: of the $P6_3/mmc$ cell was occupied by a fictitious atom, atomic number,
122: $Z=10+x$, to model a partial occupancy of $x$. We also did some
123: calculations with Na at the 2$b$ positions.
124: Supercell calculations were done at selected $x$
125: with real Na at some $2d$ or $2b$ positions.
126:
127: Prior calculations show that there is a FM solution in
128: the LDA \cite{DJS00} for Na$_{x}$CoO$_{2}$ for all $x$ in the
129: experimentally relevant range. Our
130: calculations for a tripled $x=2/3$ supercells
131: show that the A-type AFM order
132: recently revealed by neutrons is actually the preferred LDA
133: ground state.
134: AFM stacking of the ferromagnetic layers is favored by 2.3 meV/Co for
135: Na in the $2b$ site (Na on top of Co) in a $\sqrt{3}\times\sqrt{3}$ supercell,
136: and 1.7 meV/Co in the $2d$ site.
137: This is consistent with a recently measured
138: \cite{JLL04} metamagnetic AFM to
139: FM transition at the relatively low field of 8T.
140: We find very similar results in the VCA:
141: 1.4 meV/Co with Na in site $2d$ and 2.2 meV/Co with Na in site
142: $2b$. The total spin moment inside the Co APW spheres for
143: the A-type AFM ordering is only
144: $\sim $ 0.02 $\mu _{B}$ less than that in
145: the FM case which is half metallic.
146: In other words, the Co is maximally spin polarized for the doping level $x$
147: consistent with the $\sim$ 0.2 $\mu _{B} $ limit
148: from neutron \cite{SPB+04} and
149: muon spin relaxation \cite{JSHI+03} experiments
150: at $x \sim$=0.8.
151: However, note that at $x=2/3$, samples are paramagnetic; the LDA ordered
152: ground state is presumably suppressed by quantum critical fluctuations that
153: are potentially important for superconductivity. \cite{DJS03}
154:
155: % The difference between the VCA and the supercell
156: % calculations is an indication that the two terms are nearly in balance.
157:
158: We now turn to the relative magnitude of in-plane and
159: perpendicular exchange constants.
160: Since the band structure is metallic, antiferromagnetic superexchange
161: competes with ferromagnetic double exchange
162: (the former depends on hopping linearly, and the latter quadratically).
163: The large
164: in-plane dispersion leads to a net FM in-plane interaction,
165: while the net inter-plane coupling, due to the smaller $c$-axis dispersion, is
166: AFM.
167: As mentioned, nearly isotropic 3D magnetic interactions
168: are unexpected in layered compounds.
169: Still this is not inconsistent considering the large bonding-antibonding
170: splitting of the $a_{1g}$ band in LDA calculations \cite{MDJ+}
171: at large $x$.
172: At $x$=2/3, this splitting at $\Gamma$ is 0.21 eV, i.e. 15\%
173: of the full t$_{2g}$ bandwidth. If the band structure is mapped onto
174: an effective Co-only model and only nearest neighbor hopping
175: across the planes is allowed, as assumed in Refs. \onlinecite{SPB+04b}
176: and \onlinecite{LMH+04} (we will argue that this is
177: \textit{not} a good approximation), and the effective hopping amplitude
178: is $t_{\perp }$, then the $a_{1g}$ splitting at $\Gamma$ would be
179: 12$t_{\perp }$ and there would
180: be no first order splitting for the two $e_{g}^{\prime }$ bands
181: (consistent with the LDA band structure).
182: This gives $t_{\perp }\approx 15$ meV.
183: Although the $t_{2g}-t_{2g}$ in-plane hopping is
184: nearly an order of magnitude larger,
185: this still gives less anisotropy than would be anticipated
186: for a layered material.
187:
188: To proceed, we need to understand the physics of the interlayer
189: coupling.
190: The hopping must proceed,
191: via O $p$ states, which,
192: in turn, requires either direct O $p_{z}-p_{z}$ hopping
193: or hopping assisted by diffuse unoccupied Na
194: $s$ and $p$ orbitals.
195: In the latter case the details of Na placement may be important.
196: Moreover, it is significant that the energy
197: separation between the unoccupied
198: Na $3s$ and $3p$ states in Na$_{x}$CoO$_{2}$ is found to be rather small,
199: smaller in fact than the
200: corresponding band width.
201: This allows the $s$ and $p_{x,y}$ orbitals of a Na atom,
202: sitting inside an O$_{6}$ prism, to combine and form
203: bands that may be described as coming from $sp^{2}$ hybrid orbitals,
204: specifically, $h_{1}=(s-\sqrt{2}p_{y})/\sqrt{3},$ $h_{2,3}=(s\pm
205: \sqrt{3/2}p_{x}+p_{y}/\sqrt{2})/\sqrt{3}.$
206: These are asymmetric, orthonormal, and directed to
207: the midpoints of the O prism edges.
208: So the Na-assisted part of the O-O hopping goes
209: from one O to another O above it via one, two, or three
210: Na $sp^{2}$ hybrids
211: depending on how many of the three nearby Na sites are occupied.
212: Note that if only Na $s$ states were involved, the hopping
213: amplitude to a
214: second nearest neighbor O
215: in the next plane via a specific Na atom would be the
216: same as that of hopping to the O right on top. Thus Na $p$
217: participation changes the interlayer coupling
218: in an essential way.
219:
220: \begin{figure} \includegraphics[width=.30\linewidth]{noNa.ps}
221: \includegraphics[width=.316\linewidth]{d67.ps}
222: \includegraphics[width=.27\linewidth]{triple2.ps} \caption{ (color online)
223: Charge density of the
224: a$_{1g}$ bonding orbital in
225: X$_{2/3}$CoO$_2$ with X= $e^-$ (left), X= VCA Na (middle),
226: and X= real Na (right). The Na ions facilitate O-O bonding and
227: thus superexchange.
228: The VCA and supercell calculations differ in the asymmetry
229: of the $sp^2$ hybrid.} \label{charge}
230: \end{figure}
231:
232: Besides the Na assisted hopping, direct O-O hopping may
233: be important. Considering the rhombohedral
234: O site symmetry,
235: the relevant hopping is from Co $a_{1g}$ to O $p_z$.
236: Further, in Na$_x$CoO$_2$, the O atoms in
237: adjacent CoO$_2$ sheets are directly on top of each other,
238: favoring inter-layer hopping via the opposing O $p_z$ orbitals.
239: To estimate this
240: contribution we suppressed the Na-assisted hopping in three ways.
241: First, we looked at CoO$_{2}+x$ $e^{-}$ i.e., a
242: compound without Na, but with the correct number of valence
243: electrons compensated by a uniform background charge. At
244: $x=2/3,$ the resulting $a_{1g}$ bandsplitting at $\Gamma $ is 0.12 eV
245: or $\approx $70\% of the total effective coupling in
246: Na$_{2/3}$CoO$_{2}$.
247: We verified that this is independent of the amount of extra valence
248: charge in the system, by doing
249: calculations with 0.82, 0.75, and 0.4 extra $e^{-}$. Next, for $x=2/3$,
250: we put Ne in the Na sites, again adding electrons with a
251: compensating background.
252: Finally, we did calculations for
253: $\widetilde{\mathrm{Na}}_{2/3}$CoO$_{2},$
254: where $\widetilde{\mathrm{Na}}$ denotes Na with $s$ and $p$ orbitals
255: artificially shifted to higher energy. The results change
256: very little, indicating that the splitting from direct
257: O-O hopping is $\sim$ 0.12 eV.
258: Thus, the effective interplanar
259: coupling can be written as $t_{\perp }=t_{\perp }^{Na}(x)+t_{\perp }^{O}$,
260: where $t_{\perp}^{Na}(x)$ is a function of doping and
261: $t_{\perp }^{O}$ is constant, and $t_{\perp }^{Na}(x)/t_{\perp }^{O}$ varies
262: from 0.48 to 0.78 in the range $0.6\alt x\alt0.9$.
263:
264: The Na assisted hopping can be assessed using VCA or supercell calculations.
265: These differ in two ways. First, in the VCA, all Na
266: sites are occupied regardless of $x$, while in a supercell
267: and in reality
268: a portion, $1-x$ are empty.
269: The number of effective hopping paths via
270: Na, and, correspondingly,
271: the effective hopping between the planes,
272: should therefore be reduced in this proportion relative to the
273: VCA.
274: Second, the VCA nuclear charge is $Z=10+x$, so the
275: unoccupied Na $s$ and $p$ energies are moved up.
276: This will reduce the Na-O hybridization,
277: which is inversely proportional to the energy separation
278: between Na 3$s(p)$ and O 2$p$ bands.
279: This implies a artificial monotonic increase of Na-assisted hopping
280: with $x$ in the VCA.
281: These two effects are in opposite directions and should at least
282: partially cancel.
283: To assess this,
284: we performed VCA calculations for $0\leq x\leq 1$, and supercell
285: (real Na) calculations for $x$=0, 1/2, 2/3 and 1.
286: The calculated VCA $a_{1g}$ splitting at $\Gamma $ depends on $x$ non-linearly
287: in both cases; the supercell calculations give similar results with slightly
288: larger splittings.
289: (Fig. \ref{na}). We verified that the deviation from
290: this linearity for the supercell calculations
291: has the same origin as in the VCA:
292: the Na 3$s$ and 3$p$ levels shift with $x$
293: though to a lesser degree than in the VCA (the shift is
294: due to the changing Coulomb potential as a function of doping).
295: This is seen in the charge density of the
296: bonding $a_{1g}$ state at $\Gamma $,
297: (Fig. \ref{charge}).
298: The relatively weak O $p_{z}-p_{z}$ overlap
299: is augmented in the VCA by a
300: composite state in the middle of the O-O bond,
301: consisting of the three Na sp$^{2}$ hybrids;
302: removal of one of these with the
303: corresponding Na atom in the Na$_{2/3}$CoO$_{2}$
304: supercell makes the O orbitals tilt towards the remaining Na ions.
305:
306: \begin{figure}[tbp]
307: \includegraphics[width= 0.9 \linewidth]{Na.ps}
308: \caption{A comparison of VCA and supercell a$_{1g}$ splitting at
309: the $\Gamma$ point, with the same structural parameters(see text),
310: showing a monotonic increase with $x$. Calculations without Na orbitals, but with the proper valence charge show that the O-O hopping
311: contribution is independent of $x$.}
312: \label{na}
313: \end{figure}
314:
315: A somewhat counterintuitive consequence of the above analysis
316: is the fact that the nearest neighbor
317: approximation is invalid for inter-plane hopping.
318: Neither a Co-O-Na-O-Co path nor a Co-O-O-Co path needs
319: to end on the Co site directly above the one where it started.
320: There are 9 different Na-containing paths
321: connecting nearest neighbor Co ions in different
322: planes\cite{notepath}, and 3 connecting one Co with each
323: of the 6 second neighbors\cite{note2bd}.
324: For paths without Na, there are 3 that connect nearest neighbors
325: and 1 that connects second neighbors.
326: Assuming that each Na-containing path contributes $\tau$ to the
327: effective interplanar hopping amplitude
328: and each O-only path contributes $\tau ^{\prime}$, we
329: find that $t_{\perp }^{Na}(x,\mathbf{k})/t_{\perp }^{O}(\mathbf{k})=3\tau
330: x/\tau ^\prime$ and
331: \begin{equation}
332: t_{\perp }^{O}(\mathbf{k})=\tau ^{\prime }(3+2\cos \mathbf{ak+}2\cos \mathbf{%
333: bk+}2\cos \mathbf{ck}),
334: \end{equation}
335: where \textbf{a}, \textbf{b,} and \textbf{c} are respectively
336: the three projections onto
337: the $ab$ plane of the
338: vectors connecting a given Co ion with the three O above.
339: At $\Gamma ,$ $t_{\perp }^{Na}=27\tau x$, and $t_{\perp } ^{O}$=0.12
340: eV, thus we find, at $x$=0.82, $\tau=0.5$ meV and $\tau^{\prime}=1.1$
341: meV. Finally, we use the number of $\tau $ and $\tau ^{\prime }$ paths that
342: connect a Co ion to each type of neighbor to calculate $t_{c }$ and $
343: t_{c}^{\prime }$, the hopping integrals for first and second Co
344: interplanar coupling to get:
345: \begin{equation}
346: \frac{J_{c}}{J_{c}'} = \frac{t_{c}^2}{t_{c}'^2} = \left(\frac{9 \tau
347: \cdot x
348: + 3 \tau '}{3 \tau \cdot x
349: + \tau '} \right )^2 = 9
350: \end{equation}
351: \begin{figure}[tbp]
352: \includegraphics[width= 2.1 in, angle=270]{mag.ps}
353: \caption{Magnon dispersion of Na$_x$CoO$_2$ along high symmetry lines shown
354: for $x$= 0.7 and $x$=0.82. The results of two experiments in which data were
355: fit to a nearest-neighbor interaction model are compared to our dispersion
356: which includes second-nearest out-of-plane neighbors. Our parameters were
357: fit to the data of Ref. \onlinecite{SPB+04b} for $x$=0.7,
358: adjusted for Na content (see text), and used to get the
359: dispersion for $x$=0.82 \textit{without} re-fitting.}
360: \label{omega}
361: \end{figure}
362:
363: We now revisit the interpretation of the spin wave dispersion observed
364: in Refs. \onlinecite{SPB+04b} and \onlinecite{LMH+04},
365: including second neighbor assisted hopping. A
366: straightforward generalization of Eq. 2 in Ref. \onlinecite{SPB+04b} yields
367: \[
368: E =2S\sqrt{\{\tilde{\mathcal{J}}_{\perp }(0)-\mathcal{J}%
369: _{\parallel }(0)+\mathcal{J}_{\parallel }(\mathbf{q})+ (\frac{D}{2S})\}^{2}-\{\tilde{
370: \mathcal{J}}_{\perp }(\mathbf{q})\}^{2}}
371: \]
372: where, following their notation, $hkl$ is
373: the wave vector in units of the reciprocal lattice vector. $\tilde{\mathcal{J
374: }}(\mathbf{q})=\mathcal{J}_{c }(\mathbf{q})+\mathcal{J}_{c}^{\prime
375: }(\mathbf{q})$ and $\mathcal{J}_{c}^{\prime }(\mathbf{q})=2J_{c
376: }^{\prime }\cos (\pi l)[2\cos (2\pi h)+2\cos (2\pi k)+2\cos (2\pi (h+k)]$.
377: Using the $\mathbf{k}=(0,0,1/2)$ magnon energy, we can extract $J_{c
378: }=1.98$ meV and $J_{c}^{\prime }=$0.22 meV. This fits experiment
379: \cite{SPB+04b} well with $J_{\parallel }=-4$.5 meV. Bayracki $et$ $al$,
380: using nearest neighbors only, found $J_{c}=3.3$ meV and $J_{\parallel
381: }=-4.5$ meV, leading to the conclusion of magnetic isotropy. Fig. \ref{omega}
382: shows that our model produces results that are completely indistinguishable
383: from the nearest neighbor model used in Ref. \onlinecite{SPB+04b}, however,
384: the parameters are more physically reasonable and compatible with the
385: highly anisotropic electronic structure of the compound.
386:
387: Finally, we address the differences in the magnon dispersions
388: obtained in Ref. \onlinecite{SPB+04b} and Ref. \onlinecite{LMH+04},
389: both shown in Fig. \ref{omega}.
390: Because of notational differences between Ref. \onlinecite{SPB+04b} and Ref.
391: \onlinecite{LMH+04}, we extract exchange constants
392: directly from the Ref. \onlinecite{LMH+04} data, using the same formalism as
393: Ref. \onlinecite{SPB+04b}
394: and get $J_{c}$ = 6.1 eV and $J_{\parallel }=-4$.5 eV.
395: Using VCA calculations at $x$=0.7 and $x$=0.82,
396: we scale the parameters of our nearest and
397: next-nearest neighbor model and compare them with the results of Helme $et$
398: $al$.\cite{LMH+04}
399: We estimate that at $x=0.75$ $J_{c }$=1.69 meV,
400: $J_{c }^{\prime }=$0.19 meV, and $J_{\parallel }=-$4.5 meV. So,
401: the in-plane hopping remains unchanged, in agreement with experiment,
402: and the inter-planar hopping decreases by $ \approx 8\%,$
403: since less Na is available to mediate it.
404:
405: The measured dispersion\cite{LMH+04}
406: in the $z$ direction is much \textit{larger} at $x$=0.75 than at $x$=0.82.
407: Other effects, not
408: accounted for here, must therefore be operative.
409: One possibility is a different pattern of Na ordering.
410: Indeed,
411: in our $x=2/3$ supercell calculations
412: placing all the Na at the $2b$ sites instead of the $2d$ sites
413: yields a $\sim$ 35\% increase in the interlayer
414: coupling. While this number no doubt depends on the exact Na
415: arrangement, it does indicate sensitivity to the ordering of a size
416: sufficient to explain the experiment.
417: In comparing with experiment, it should be also
418: noted that Co disproportionation,
419: seen by NMR studies \cite{IRM+04}around $x$=0.7,
420: would change the spin arrangement of the
421: lattice and therefore the magnon dispersion.
422: Co ions with formal valency (4-x)$^{+}$ split proportionally into non-magnetic
423: Co$^{3+}$ and $S=1/2$ Co$ ^{4+}$.
424: The particular arrangement is unknown,
425: but the exchange interaction may be changed.
426: Note that the phase boundary of charge order is not yet clearly defined
427: and calculations \cite{JKK-L03} suggest that, near the crossover,
428: there is a region where two distinct magnetic ions exist.
429: Another consideration
430: is that the spin-wave model adopted
431: here and in Refs. \onlinecite{SPB+04b} and \onlinecite{LMH+04} assumes
432: a rigid spin moment of S=1/2,
433: which may not be a good approximation for the weak magnetism in this compound.
434: The actual moment,
435: both measured and calculated,
436: is smaller and grows with $x$.
437: Last but not least,
438: one should keep in mind that FM and AFM
439: interactions here are competing with each other
440: thereby amplifying the relative effect of doping changes on magnetism.
441:
442: In summary, we find that inter-plane coupling to second neighbors
443: plays an important role in the out-of-plane magnetism of
444: Na$_{x}$CoO$_{2}.$ Heisenberg type models including
445: only one nearest neighbor exchange across the plane are
446: insufficient;
447: exchange between the six next nearest neighbors in adjacent planes is needed.
448: Fortunately, one can estimate the ratio of the two exchange constants,
449: making it possible to extract them from experiment
450: without increasing the number of fitting parameters.
451: The resulting exchange constants provide a magnon spectrum that matches
452: experiment extremely well and yields a physically realistic
453: picture of magnetic interactions in this layered material.
454: The LDA energy difference between the FM
455: and the A-type AFM ordering for Na content $x=2/3$ is $\sim 2$
456: meV/Co, which is reasonable considering the exchange coupling deduced
457: from experimentally measured spin wave dispersion
458: (6.6 meV and 12.1 meV/Co in
459: Refs. \onlinecite{SPB+04b} and \onlinecite{LMH+04}, respectively).
460: We note that superexchange drops with the distance
461: more strongly than other coupling mechanisms,
462: such as double exchange. Thus, although the
463: inter-planar coupling is surprisingly
464: strong in Na$_x$CoO$_2$,
465: yielding three dimensional magnetic character, it would be expected
466: to be very weak in the hydrated superconducting compound.
467: It is even possible that a cross-over from AFM to FM coupling occurs
468: with hydration due to suppression of the superexchange interaction.
469: If the magnetic interactions become effectively 2D in the
470: hydrated compound, fluctuations would be enhanced,
471: possibly revealing superconductivity associated with a nearby magnetic quantum
472: critical point.
473:
474: We are grateful for helpful discussions with
475: A.T. Boothroyd, R. Jin and S. Nagler. Work
476: at Oak Ridge National Laboratory is supported by the U.S. Department of
477: Energy.
478:
479: \begin{thebibliography}{31}
480: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
481: \expandafter\ifx\csname bibnamefont\endcsname\relax
482: \def\bibnamefont#1{#1}\fi
483: \expandafter\ifx\csname bibfnamefont\endcsname\relax
484: \def\bibfnamefont#1{#1}\fi
485: \expandafter\ifx\csname citenamefont\endcsname\relax
486: \def\citenamefont#1{#1}\fi
487: \expandafter\ifx\csname url\endcsname\relax
488: \def\url#1{\texttt{#1}}\fi
489: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
490: \providecommand{\bibinfo}[2]{#2}
491: \providecommand{\eprint}[2][]{\url{#2}}
492:
493: \bibitem[{\citenamefont{{Y. Wang} et~al.}(2003)\citenamefont{{Y. Wang}, {N. S.
494: Rogado}, {R. J. Cava}, and {N. P. Ong}}}]{YWNSR+03}
495: \bibinfo{author}{\bibnamefont{{Y. Wang \it{et al}}}},
496: \bibinfo{journal}{Nature}
497: \textbf{\bibinfo{volume}{423}}, \bibinfo{pages}{425} (\bibinfo{year}{2003}).
498:
499: \bibitem[{\citenamefont{{W. Higemoto} et~al.}(2004)\citenamefont{{W. Higemoto},
500: {K. Ohishi}, {A. Koda}, {R. Kadono}, {K. Ishida}, {K. Takada}, {H. Sakurai},
501: {E. Takayama-Muromachi}, and {T. Sasaki}}}]{WHKO+03}
502: \bibinfo{author}{\bibnamefont{{W. Higemoto \it{et al}}}},
503: \bibinfo{journal}{Phys. Rev. B}
504: \textbf{\bibinfo{volume}{70}}, \bibinfo{pages}{134508}
505: (\bibinfo{year}{2004}).
506:
507: \bibitem[{\citenamefont{{A. Kanigel} et~al.}(2004)\citenamefont{{A. Kanigel},
508: {A. Keren}, {L. Patlagan}, {K. B. Chashka}, {P. King}, and {A.
509: Amato}}}]{AKAK+03}
510: \bibinfo{author}{\bibnamefont{{A. Kanigel \it{et al}}}},
511: \bibinfo{journal}{Phys. Rev.
512: Lett.} \textbf{\bibinfo{volume}{92}}, \bibinfo{pages}{257007}
513: (\bibinfo{year}{2004}).
514:
515: \bibitem[{\citenamefont{{T. Waki} et~al.}(2003)\citenamefont{{T. Waki}, {C.
516: Michioka}, {M. Kato}, {K. Yoshimura}, {K. Takada}, {H. Sakurai}, {E.
517: Takayama-Muromachi}, and {T. Takayoshi Sasaki}}}]{TWCM+}
518: \bibinfo{author}{\bibnamefont{{T. Waki \it{et al}}}},
519: \bibinfo{journal}{cond-mat/0306036} (\bibinfo{year}{2003}).
520:
521: \bibitem[{\citenamefont{{D. J. Singh}}(2003)}]{DJS03}
522: \bibinfo{author}{\bibnamefont{{D. J. Singh}}}, \bibinfo{journal}{Phys. Rev. B}
523: \textbf{\bibinfo{volume}{68}}, \bibinfo{pages}{020503}
524: (\bibinfo{year}{2003}).
525:
526: \bibitem[{\citenamefont{{M. D. Johannes}
527: et~al.}(2004{\natexlab{a}})\citenamefont{{M. D. Johannes}, {I. I. Mazin}, {D.
528: J. Singh}, and {D. A. Papaconstantopoulos}}}]{MDJ+b}
529: \bibinfo{author}{\bibnamefont{{M. D. Johannes}}},
530: \bibinfo{author}{\bibnamefont{{I. I. Mazin}}},
531: \bibinfo{author}{\bibnamefont{{D. J. Singh}}}, \bibnamefont{and}
532: \bibinfo{author}{\bibnamefont{{D. A. Papaconstantopoulos}}},
533: \bibinfo{journal}{PRL} \textbf{\bibinfo{volume}{93}}, \bibinfo{pages}{097005}
534: (\bibinfo{year}{2004}{\natexlab{a}}).
535:
536: \bibitem[{\citenamefont{{K. Ishida} et~al.}(2003)\citenamefont{{K. Ishida}, {Y.
537: Ihara}, {Y. Maeno}, {C. Michioka}, {M. Kato}, {K. Yoshimura}, {K. Takada},
538: {T. Sasaki}, {H. Sakurai}, and {E. Takayama-Muromachi}}}]{KIYI+}
539: \bibinfo{author}{\bibnamefont{{K. Ishida \it{et al}}}},
540: \bibinfo{journal}{J.
541: Phys. Soc. Jpn.} \textbf{\bibinfo{volume}{72}}, \bibinfo{pages}{3041}
542: (\bibinfo{year}{2003}).
543:
544: \bibitem[{\citenamefont{{A. Tanaka} and {X. Hu}}(2003)}]{ATXH03}
545: \bibinfo{author}{\bibnamefont{{A. Tanaka}}} \bibnamefont{and}
546: \bibinfo{author}{\bibnamefont{{X. Hu}}}, \bibinfo{journal}{Phys. Rev. Lett.}
547: \textbf{\bibinfo{volume}{91}}, \bibinfo{pages}{257006}
548: (\bibinfo{year}{2003}).
549:
550: \bibitem[{\citenamefont{{T. Fujimoto} et~al.}(2004)\citenamefont{{T. Fujimoto},
551: {G. Zheng}, {Y. Kitaoka}, {R. L. Meng}, {J. Cmaidalka}, and {C. W.
552: Chu}}}]{TFGZ+}
553: \bibinfo{author}{\bibnamefont{{T. Fujimoto \it{et al}}}},
554: \bibinfo{journal}{Phys. Rev.
555: Lett.} \textbf{\bibinfo{volume}{92}}, \bibinfo{pages}{047004}
556: (\bibinfo{year}{2004}).
557:
558: \bibitem[{\citenamefont{{M. L. Foo} et~al.}(2004)\citenamefont{{M. L. Foo}, {Y.
559: Wang}, {S. Watuchi}, {H. W. Zandbergen}, {T. He}, {R. J. Cava}, and {N. P.
560: Ong}}}]{MLF+04}
561: \bibinfo{author}{\bibnamefont{{M. L. Foo \it{et al}}}},
562: \bibinfo{journal}{Phys. Rev. Lett} \textbf{\bibinfo{volume}{92}},
563: \bibinfo{pages}{247001} (\bibinfo{year}{2004}).
564:
565: \bibitem[{\citenamefont{{T. Motohashi} et~al.}(2003)\citenamefont{{T.
566: Motohashi}, {R. Ueda}, {E. Naujalis}, {R. Tojo}, {I. Terasaki}, {T. Atake},
567: {M. Karppinen}, and {H. Yamauchi}}}]{TMRU+03}
568: \bibinfo{author}{\bibnamefont{{T. Motohashi \it{et al}}}},
569: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{67}},
570: \bibinfo{pages}{064406} (\bibinfo{year}{2003}).
571:
572: \bibitem[{\citenamefont{{J. Sugiyama} et~al.}(2003)\citenamefont{{J. Sugiyama},
573: {H. Itahara}, {J. H. Brewer}, {E. J. Ansaldo}, {T. Motohashi}, {M.
574: Karppinen}, and {H. Yamauchi}}}]{JSHI+03}
575: \bibinfo{author}{\bibnamefont{{J. Sugiyama \it{et al}}}},
576: \bibinfo{journal}{Phys. Rev.
577: B} \textbf{\bibinfo{volume}{67}}, \bibinfo{pages}{214420}
578: (\bibinfo{year}{2003}).
579:
580: \bibitem[{\citenamefont{{B. C. Sales} et~al.}(2004)\citenamefont{{B. C. Sales},
581: {R. Jin}, {K. A. Affholter}, {P. Khalifah}, {G. M. Veith}, and {D.
582: Mandrus}}}]{BCS+04}
583: \bibinfo{author}{\bibnamefont{{B. C. Sales \it{et al}}}},
584: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{70}}, \bibinfo{pages}{174419} (\bibinfo{year}{2004}).
585:
586: \bibitem[{\citenamefont{{J. Wooldridge} et~al.}(2004)\citenamefont{{J.
587: Wooldridge}, {D. M. Paul}, {G. Balakrishnan}, and {M. R. Lees}}}]{JWDMP+04}
588: \bibinfo{author}{\bibnamefont{{J. Wooldridge}}},
589: \bibinfo{author}{\bibnamefont{{D. M. Paul}}},
590: \bibinfo{author}{\bibnamefont{{G. Balakrishnan}}}, \bibnamefont{and}
591: \bibinfo{author}{\bibnamefont{{M. R. Lees}}},
592: \bibinfo{journal}{cond-mat/0406513} (\bibinfo{year}{2004}).
593:
594:
595: \bibitem[{\citenamefont{{J. L. Gavilano} et~al.}(2003)\citenamefont{{J. L.
596: Gavilano}, {D. Rau}, {B. Pedrini}, {J. Hinderer}, {H. R. Ott}, {S. Kazakov},
597: and {J. Karpinski}}}]{JLG+03}
598: \bibinfo{author}{\bibnamefont{{J. L. Gavilano \it{et al}}}},
599: \bibinfo{journal}{cond-mat/0308383} \textbf{\bibinfo{volume}{69}}, \bibinfo{pages}{100404} (\bibinfo{year}{2004}).
600:
601: \bibitem[{\citenamefont{{F. C. Chou} et~al.}(2004)\citenamefont{{F. C. Chou},
602: {J. H. Cho}, and {Y. S. Lee}}}]{FCC04}
603: \bibinfo{author}{\bibnamefont{{F. C. Chou}}}, \bibinfo{author}{\bibnamefont{{J.
604: H. Cho}}}, \bibnamefont{and} \bibinfo{author}{\bibnamefont{{Y. S. Lee}}},
605: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{70}},
606: \bibinfo{pages}{144526} (\bibinfo{year}{2004}).
607:
608: \bibitem[{\citenamefont{{A. T. Boothroyd} et~al.}(2003)\citenamefont{{A. T.
609: Boothroyd}, {R. Coldea}, {D. A. Tennant}, {D. Prabhakaran}, and {C. D.
610: Frost}}}]{ATB+03}
611: \bibinfo{author}{\bibnamefont{{A. T. Boothroyd \it{et al}}}},
612: \bibinfo{journal}{Phys. Rev.
613: Lett.} \textbf{\bibinfo{volume}{92}}, \bibinfo{pages}{197201}
614: (\bibinfo{year}{2003}).
615:
616: \bibitem[{\citenamefont{{D. J. Singh}}(2000)}]{DJS00}
617: \bibinfo{author}{\bibnamefont{{D. J. Singh}}}, \bibinfo{journal}{Phys. Rev. B}
618: \textbf{\bibinfo{volume}{61}}, \bibinfo{pages}{13397} (\bibinfo{year}{2000}).
619:
620: \bibitem[{\citenamefont{{S. P. Bayrakci}
621: et~al.}(2004{\natexlab{a}})\citenamefont{{S. P. Bayrakci}, {I. Mirebeau}, {P.
622: Bourges}, {Y. Sidis}, {M. Enderle}, {J. Mesot}, {D. P. Chen}, {C. T. Lin},
623: and {B. Keimer}}}]{SPB+04b}
624: \bibinfo{author}{\bibnamefont{{S. P. Bayrakci \it{et al}}}},
625: \bibinfo{journal}{cond-mat/0410224} (\bibinfo{year}{2004}{\natexlab{a}}).
626:
627: \bibitem[{\citenamefont{{L. M. Helme} et~al.}(2004)\citenamefont{{L. M. Helme},
628: {A. T. Boothroyd}, {R. Coldea}, {D. Prabhakaran}, {D. A. Tennant}, {A.
629: Hiess}, and {J. Kulda}}}]{LMH+04}
630: \bibinfo{author}{\bibnamefont{{L. M. Helme \it{et al}}}},
631: \bibinfo{journal}{cond-mat/0410457} (\bibinfo{year}{2004}).
632:
633: \bibitem[{Lno()}]{Lnote}
634: \bibinfo{note}{We did well converged
635: calculations with the Wien2k code \cite{Wien2k}
636: and an independent LAPW code. These gave practically identical
637: results when tested for the same system.
638: For the APW+lo calculations sphere radii of
639: 1.9 $a_0$,
640: 1.6 $a_0$ and 2.0 $a_0$ were used for Co, O and Na, respectively.
641: For the LAPW calculations the corresponding radii were
642: 1.95 $a_0$, 1.55 $a_O$ and 2.0 $a_O$.
643: }
644:
645: \bibitem[{\citenamefont{{P. Blaha} et~al.}(2002)\citenamefont{{P. Blaha}, {K.
646: Schwarz}, {G. K. H. Madsen}, {D. Kvasnicka}, and {J. Luitz}}}]{Wien2k}
647: \bibinfo{author}{\bibnamefont{{P. Blaha}}}, \bibinfo{author}{\bibnamefont{{K.
648: Schwarz}}}, \bibinfo{author}{\bibnamefont{{G. K. H. Madsen}}},
649: \bibinfo{author}{\bibnamefont{{D. Kvasnicka}}}, \bibnamefont{and}
650: \bibinfo{author}{\bibnamefont{{J. Luitz}}}, \emph{\bibinfo{title}{Wien2k}}
651: (\bibinfo{year}{2002}), \bibinfo{note}{iSBN 3-9501031-1-2}.
652:
653: % \bibitem[{\citenamefont{{J. P. Perdew} and {Y. Wang}}(1992)}]{LDA5}
654: % \bibinfo{author}{\bibnamefont{{J. P. Perdew}}} \bibnamefont{and}
655: % \bibinfo{author}{\bibnamefont{{Y. Wang}}}, \bibinfo{journal}{Phys. Rev. B}
656: % \textbf{\bibinfo{volume}{45}}, \bibinfo{pages}{13244} (\bibinfo{year}{1992}).
657:
658: \bibitem[{Ono()}]{Onote}
659: \bibinfo{note}{We verified that relaxing the O position as a function of $x$
660: has relatively little effect, and keeping the position fixed helps elucidate
661: the physics related to Na doping.}
662:
663: \bibitem[{\citenamefont{{J. L. Luo} et~al.}(2004)\citenamefont{{J. L. Luo},
664: {N.L. Wang}, { G.T. Liu}, {D. Wu}, {X.N. Jing}, {F. Hu}, and {T.
665: Xiang}}}]{JLL04}
666: \bibinfo{author}{\bibnamefont{{J. L. Luo \it{et al}}}},
667: \bibinfo{journal}{Phys. Rev.
668: Lett} \textbf{\bibinfo{volume}{93}}, \bibinfo{pages}{187203}
669: (\bibinfo{year}{2004}).
670:
671: \bibitem[{\citenamefont{{S. P. Bayrakci}
672: et~al.}(2004{\natexlab{b}})\citenamefont{{S. P. Bayrakci}, {C. Bernhard}, {D.
673: P. Chen}, {B. Keimer}, {R. K. Kermer}, {P. Lemmens}, {C. T. Lin}, {C.
674: Niedermayer}, and {J. Strempfer}}}]{SPB+04}
675: \bibinfo{author}{\bibnamefont{{S. P. Bayrakci \it{et al}}}},
676: \bibinfo{journal}{Phys. Rev.
677: B} \textbf{\bibinfo{volume}{69}}, \bibinfo{pages}{100410}
678: (\bibinfo{year}{2004}{\natexlab{b}}).
679:
680: \bibitem[{\citenamefont{{M. D. Johannes}
681: et~al.}(2004{\natexlab{b}})\citenamefont{{M. D. Johannes}, {D. A.
682: Papaconstantopoulos}, {D. J. Singh}, and {M. J. Mehl}}}]{MDJ+}
683: \bibinfo{author}{\bibnamefont{{M. D. Johannes}}},
684: \bibinfo{author}{\bibnamefont{{D. A. Papaconstantopoulos}}},
685: \bibinfo{author}{\bibnamefont{{D. J. Singh}}}, \bibnamefont{and}
686: \bibinfo{author}{\bibnamefont{{M. J. Mehl}}}, \bibinfo{journal}{Euro. Phys.
687: Lett.} \textbf{\bibinfo{volume}{68}}, \bibinfo{pages}{433}
688: (\bibinfo{year}{2004}{\natexlab{b}}).
689:
690: \bibitem[{not({\natexlab{a}})}]{notepath}
691: \bibinfo{note}{Note that if hopping was only via Na s orbitals, and not Na
692: sp$^{2}$ hybrids, this counting would be different.}
693:
694: \bibitem[{not({\natexlab{b}})}]{note2bd}
695: \bibinfo{note}{This is independent of whether the Na is at the $2b$
696: or the $2d$ sites.}
697:
698: \bibitem[{\citenamefont{{I. R. Mukhamedshin} et~al.}(2004)\citenamefont{{I. R.
699: Mukhamedshin}, {H. Alloul}, {G. Collin}, and {N. Blanchard}}}]{IRM+04}
700: \bibinfo{author}{\bibnamefont{{I. R. Mukhamedshin}}},
701: \bibinfo{author}{\bibnamefont{{H. Alloul}}},
702: \bibinfo{author}{\bibnamefont{{G. Collin}}}, \bibnamefont{and}
703: \bibinfo{author}{\bibnamefont{{N. Blanchard}}}, \bibinfo{journal}{Phys. Rev.
704: Lett.} \textbf{\bibinfo{volume}{93}}, \bibinfo{pages}{167601}
705: (\bibinfo{year}{2004}).
706:
707: \bibitem[{\citenamefont{{K. -W. Lee} et~al.}(2004)\citenamefont{{K. -W. Lee},
708: {J. Kunes}, and {W. E. Pickett}}}]{JKK-L03}
709: \bibinfo{author}{\bibnamefont{{K. -W. Lee}}}, \bibinfo{author}{\bibnamefont{{J.
710: Kunes}}}, \bibnamefont{and} \bibinfo{author}{\bibnamefont{{W. E. Pickett}}},
711: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{70}},
712: \bibinfo{pages}{045104} (\bibinfo{year}{2004}).
713:
714: \end{thebibliography}
715:
716: \end{document}
717: