1: \documentclass[prl,twocolumn]{revtex4}
2: %%\input{tcilatex} ,twocolumn
3: %\input{tcilatex}
4:
5:
6: %\documentclass[showpacs,preprintnumbers,preprint]{revtex4}
7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8: \usepackage{amsfonts}
9: \usepackage{amsmath}
10: \usepackage{amssymb}
11: \usepackage{graphicx}
12:
13: \setcounter{MaxMatrixCols}{10}
14: %TCIDATA{OutputFilter=LATEX.DLL}
15: %TCIDATA{Version=4.10.0.2363}
16: %TCIDATA{LastRevised=Wednesday, September 20, 2006 10:15:29}
17: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
18: %TCIDATA{Language=American English}
19:
20:
21: \begin{document}
22:
23: \title{Evidence for two fluids in the ground state of cuprates}
24: \author{Amit Keren}
25: \author{ Amit Kanigel}
26: \altaffiliation[Current Address: ]{Department of Physics, University of Illinois at Chicago, IL 60607}
27: \author{Galina Bazalitsky}
28: \affiliation{Technion - Israel Institue of Technology.}
29: \pacs{}
30:
31: \begin{abstract}
32: We report charge density measurements, using NMR, in the superconducting
33: compound (Ca$_{x}$La$_{1-x}$)(Ba$_{1.75-x}$La$_{0.25+x}$)Cu$_{3}$O$_{y}$,
34: which has two independent variables $x$ (family) and $y$ (oxygen). For
35: underdoped samples we find the rate at which holes are introduced into the
36: plane upon oxygenation to be family-independent. In contrast, \emph{not} all
37: carriers contribute to either antiferromagnetic or superconducting order
38: parameters. This result is consistent with a two fluid phenomenology or
39: intrinsic mesoscopic inhomogeneities in the bulk. We also discuss the impact
40: of weak-chemical-disorder on $T_{c}$.
41: \end{abstract}
42:
43: \date{\today }
44: \maketitle
45:
46: Proper counting of holes in the cuprates is essential for understanding
47: their properties. Since they have a small ratio of coherence length to mean
48: free path, they are considered clean superconductors where the carrier
49: density $n$ is the same as\ the superconducting carrier density $n_{s}$ in
50: the zero temperature limit \cite{Tallon}. In contrast, modern experiments in
51: the superconducting part of the phase diagram consistently find strong
52: inhomogeneities in these materials \cite{InhomExp,KerenSSC03}. Moreover,
53: several theories of cuprate superconductivity are based on two fluids
54: comprised of: hole-poor and hole-rich regions \cite{EmeryPhysicaC}, bosons
55: and fermions \cite{Boson}, hot and cooled electrons \cite{Pines}, etc.. This
56: contradiction leads to the question: do all holes participate in the
57: superconducting order parameter? Addressing this question requires a
58: simultaneous measurement of $n$ and $n_{s}$. The in-plane $^{63}$Cu(2)
59: nuclear quadrupole resonance (NQR) parameter $\nu _{Q}$ is a direct measure
60: of $n$. $n_{s}$ can be extracted from the penetration depth. In this work we
61: compare the two numbers for different superconducting families. We find that
62: $n_{s}$ is \emph{not} a universal fraction of $n$, and provide a simple
63: relation between $n_{s}$ and $y$, from parent to overdoped samples. The data
64: support two fluids-based theories for the entire phase diagram, but
65: constrain the possible division between the fluids.
66:
67: Our study is done on the (Ca$_{x}$La$_{1-x}$)(Ba$_{1.75-x}$La$_{0.25+x}$)Cu$%
68: _{3}$O$_{y}$ (CLBLCO) compound where each value of $x$ is a superconducting
69: family with its own maximum $T_{c}$ ($T_{c}^{max}$) ranging between $58$ and
70: $80$~K, as shown in Fig.~\ref{TcandUni}(a) \cite{OferToBe}. By choosing a
71: sample with a particular $x$ and oxygen level $y$ one can control $n$ and $%
72: n_{s}$ independently. This allows determination of the rates at which
73: oxygenation produces carriers and carriers turn superconducting. We evaluate
74: the average $\nu _{Q}$ form $^{63}$Cu(2) nuclear magnetic resonance (NMR)
75: measurements. In passing, we estimate the chemical disorder from the width, $%
76: \Delta \nu _{Q}$, of the $\nu _{Q}$ distribution. $n_{s}$ is obtained from
77: previous muon spin rotation ($\mu $SR) measurements \cite%
78: {KerenSSC03,OferToBe}.
79:
80: The ability of NMR to determine carrier density is based on the fact that $%
81: ^{63}$Cu, with its spin $3/2$ nuclei, is directly coupled to charge degrees
82: of freedom via the electric field gradient (EFG), and $\nu _{Q}$ is a
83: measure of this coupling. $\nu _{Q}$, in turn, depends linearly on the hole
84: density \cite{Asayama} according to
85: \begin{equation}
86: \nu _{Q}=An+{}\nu _{Q}^{0}\text{,} \label{nuqvsn}
87: \end{equation}%
88: where $A$ and $\nu _{Q}^{0}$\ are doping-independent, but, in principle,
89: could be family-dependent. This linear dependence was demonstrated for
90: various compounds such as Y123 \cite{Yasuoka}, La$_{2-x}$Sr$_{x}$CuO$_{4}$
91: (La$214$) \cite{Zheng}, and HaBa$_{2}$CuO$_{4+\delta }$ (Ha124) \cite%
92: {Gippius}. Therefore, Eq. \ref{nuqvsn}\ and the ability of NMR to detect the
93: in-plane copper [Cu(2)] $\nu _{Q}$ selectively will allow us to determine
94: the evolution of the in-plane carrier concentration and the width of its
95: distribution.
96:
97: \begin{figure}
98: \begin{center}
99: \includegraphics[width=7.5cm]{TcAndUni.EPS}
100: \end{center}
101: \caption{(a) The superconducting part of the (Ca$_{x}$La$_{1-x}$)(Ba$%
102: _{1.75-x}$La$_{0.25+x}$)Cu$_{3}$O$_{y}$ phase diagram \protect\cite{clblco}.
103: (b) Normalized critical temperatures plotted as a function of $K(x)\Delta y$%
104: , including the glass $T_{g}$ temperature, and corrected N\'{e}el
105: temperature $T_{N}^{cor}$ (see text) \protect\cite{OferToBe}. Inset, the
106: normalized muon spin relaxation rates taken from Ref.~\protect\cite%
107: {KerenSSC03} as a function of oxygen level $y$ for the $x=0.1$ and $0.4$
108: CLBLCO samples.}
109: \label{TcandUni}
110: \end{figure}
111:
112:
113: %\begin{figure}[tbp]
114: %\begin{center}
115: %\includegraphics[
116: %natheight=8in,
117: %natwidth=7in,
118: %height=4in,
119: %width=3.0in
120: %]{TcAndUni.EPS}
121: %\end{center}
122: %\caption{(a) The superconducting part of the (Ca$_{x}$La$_{1-x}$)(Ba$%
123: %_{1.75-x}$La$_{0.25+x}$)Cu$_{3}$O$_{y}$ phase diagram \protect\cite{clblco}.
124: %(b) Normalized critical temperatures plotted as a function of $K(x)\Delta y$%
125: %, including the glass $T_{g}$ temperature, and corrected N\'{e}el
126: %temperature $T_{N}^{cor}$ (see text) \protect\cite{OferToBe}. Inset, the
127: %normalized muon spin relaxation rates taken from Ref.~\protect\cite%
128: %{KerenSSC03} as a function of oxygen level $y$ for the $x=0.1$ and $0.4$
129: %CLBLCO samples.}
130: %\label{TcandUni}
131: %\end{figure}
132:
133: The measurements were done on powder samples fully enriched with $^{63}$Cu.
134: Their preparation is described in Ref.~\cite{clblco}. The oxygen content was
135: measured by double iodometric titration. The accuracy of this method in the
136: enriched CLBLCO is about 0.01. We measured between five and seven different
137: samples for each $x$ in the normal state at 100~K. The most overdoped sample
138: is a non superconducting $x=0.1$ compound. The NMR measurements were done by
139: sweeping the field in a constant applied frequency $f_{\text{app}}=77.95$%
140: ~MHz, using a $\pi /2$ - $\pi $ echo sequence. The echo signal was averaged $%
141: 100,000$ times and its area evaluated as a function of field. The full
142: spectrum of the optimally doped $x=0.4$ sample ($y=7.156$) is shown in the
143: inset of Fig.~\ref{Optimalfit}.
144:
145: The Cu spin Hamiltonian can be written as \cite{Slichter}:
146: \begin{equation}
147: \mathcal{H}/h=-\nu _{l}\mathbf{I\cdot (1+K)\cdot \hat{H}}+\frac{\nu _{Q}}{6}%
148: [3\mathbf{I}_{z}^{2}-\mathbf{I}^{2}+\eta (\mathbf{I}_{x}^{2}-\mathbf{I}%
149: _{y}^{2})], \label{H}
150: \end{equation}%
151: where $\nu _{l}=(^{63}\gamma /2\pi )H$, $\hat{\mathbf{H}}$ is a unit vector
152: in the direction of the field, $\mathbf{K}$ is the shift tensor, and $\eta $
153: is the asymmetry parameter of the EFG. In the absence of magnetic field,
154: there is only one transition frequency given by $f=\nu _{Q}\sqrt{1+\eta /3}$%
155: , so $\nu _{Q}$ cannot be separated from $\eta $, and the use of the
156: magnetic field is essential. This field, applied in the direction $\theta $
157: and $\phi $ with respect to the principal axis of the EFG, lifts this
158: degeneracy, and three transition frequencies $\nu _{m}(H,\theta ,\varphi )$
159: are expected: a center line which corresponds to the $1/2\rightarrow -1/2$
160: transition ($m=0$), and two satellites which correspond to the $%
161: 3/2\rightarrow 1/2$ ($m=1$) and $-1/2\rightarrow -3/2$ ($m=-1$) transitions.
162: Expressions for $\nu _{m}(H,\theta ,\varphi )$ up to second order
163: perturbation theory in $\nu _{Q}$ for completely asymmetric EFG and shift
164: tensors are given in Ref.~\cite{Taylor}. In a powder spectrum, where $\theta
165: $ and $\varphi $ are integrated out, each one of the $m=-1$ and $m=1$
166: transitions contributes one peak, and the $m=0$ transition generates two
167: peaks provided that $\eta <1$ \cite{Taylor,KerenPRB98}. In principle, the
168: bigger $\nu _{Q}$ is, the further the peaks are away from each other.
169: Similarly, the wider the peaks are, the broader the distribution of $\nu
170: _{Q} $. In CLBLCO as in YBCO, the middle peak is from the Cu(1) which has $%
171: \eta \sim 1$. This peak is labeled $1$ in the inset of Fig.~\ref{Optimalfit}%
172: . The other four peaks are associated with the plane Cu(2) and are labeled $%
173: 2 $.
174:
175:
176: \begin{figure}
177: \begin{center}
178: \includegraphics[width=9cm]{OptimalFit.EPS}
179: \end{center}
180: \caption{NMR spectra of $^{63}$Cu at $T=100$~K in optimally doped CLBLCO
181: samples with varying $x$. The inset shows the full spectrum of the $x=0.4$
182: compound including contributions from Cu(1) and Cu(2). The main figure zooms
183: in on the Cu(2) contribution (note the three axis breakers). The position of
184: the Cu(2) peaks are shown by dotted lines.}
185: \label{Optimalfit}
186: \end{figure}
187:
188: A zoom on the main features of the Cu(2) signal of all four optimally doped
189: samples is depicted in Fig.~\ref{Optimalfit} (note the three axis breakers).
190: The evolution of the main peaks as $x$ increases is highlighted by the
191: dotted lines. It is clear that as $x$ decreases the peaks move away from
192: each other. This means that $\nu _{Q}$ at optimal doping is a decreasing
193: function of $x$. A more interesting observation is the fact that there is no
194: change in the width of the peaks, at least not one that can easily be
195: spotted by the naked eye. This means that the distribution of $\nu _{Q}$ is $%
196: x$-independent and that there is no difference in the weak-chemical-disorder
197: (WCD) between the optimally doped samples of the different families. By WCD
198: we mean disorder that is not strong enough to \textquotedblleft wipe
199: out\textquotedblright\ the contribution of the Cu(2) in its vicinity from
200: the spectrum. Thus, WCD is not relevant to the variation of $T_{c}^{max}$
201: between the different families. As we shall see, this conclusion is
202: supported by more rigorous analysis.
203:
204: Although we limit our conclusion to WCD, it is important to place a few
205: limitations on the possible existence of strong chemical disorder. The
206: difference in ionic size between La and Ba in the Ba layer creates local
207: lattice distortions. These impact the apical oxygen O(4) position, which
208: contributes to $\nu _{q}$. However a simple calculation based on Eqs. 4 and
209: 5 in Ref.~\cite{Chmaissem} shown that the probability of finding an O(4) out
210: of its ideal place increases with $x$. Therefore, if the O(4) displacement
211: were important for chemical disorder we would have expected the $x=0.4$
212: family to have a lower $T_{c}$ than the $x=0.1$, in contrast to observation.
213: In addition, magic angle spinning (MAS) Ca NMR found only one line. This
214: ensures the existence of only one Ca site. Also, since Ca and Ba have the
215: same valance, and the total amount of La is fixed, doping is done only by
216: the chain layer oxygen, which is relatively remote from the CuO$_{2}$ layer.
217: Therefore, chemical disorder is expected to be minimal. Overall, we cannot
218: find evidence for the existence or importance of strong chemical disorder.
219:
220: The evolution of the main peaks for $x=0.4$ as a function of $y$ is shown in
221: Fig.~\ref{x04lines}. Here, as $y$ increases the peaks moves away from each
222: other, \textit{i.e.}, $\nu _{Q}$ increases as a function of doping as
223: expected from Eq.~\ref{nuqvsn}. Similar data for the $x=0.1$ family can be
224: found in Ref.~\cite{KanigelToBe}.
225:
226: \begin{figure}
227: \begin{center}
228: \includegraphics[width=9cm]{x04Lines.EPS}
229: \end{center}
230: \caption{NMR spectra of $^{63}$Cu(2) at $T=100$~K in CLBLCO samples with $%
231: x=0.4$ and varying $y.$ The positions of the peaks are shown by dotted lines.
232: }
233: \label{x04lines}
234: \end{figure}
235:
236: In order to analyze these data more precisely we must fit them to a
237: field-swept powder spectrum $P(H)$ which is given by%
238: \begin{equation}
239: P_{m}(H)\propto \sum_{m}\int M^{2}\delta \lbrack f_{\text{app}}-\nu
240: _{m}(H,\theta ,\varphi )]d\Omega \label{trans_cond}
241: \end{equation}%
242: where $M$ is a matrix element. Usually one can extract all the averaged
243: Hamiltonian parameters from the position of the peaks in the spectra \cite%
244: {KerenPRB98}. However, we are interested in both the parameters and their
245: distribution. Therefore, we must account for the entire line shape. For this
246: purpose we simulate the line by using a grid in the $(\theta ,\phi )$ space,
247: calculate the frequency $\nu _{m}(H,\theta ,\varphi )$ for every $m$, field,
248: and point on the grid, and add 1 to a histogram of $H$ when one of the
249: frequencies $\nu _{m}$ equals $f_{\text{app}}$. The matrix elements are
250: taken as unity. This numerical simulation approximates $P(H)$ in Eq.~\ref%
251: {trans_cond}. To account for the peak widths we assume that the main
252: contribution to this width, of the order of a few MHz, is from a
253: distribution in $\nu _{Q}$, since the quadrupole interaction is the only
254: interaction of such magnitude in the system. Consequently, we added to the
255: numerical evaluation of Eq.~\ref{trans_cond} a loop over $200$ values of $%
256: \nu _{Q}$ drawn from a normal distribution with a width $\Delta \nu _{Q}$.
257: We also added the contribution of the chain site, with $\eta \sim 1$. We
258: searched for the best fit to the data by $\chi ^{2}$ minimization using a
259: simplex code \cite{NumericalRec}. The result for the optimal doped samples
260: is shown as the solid line in Fig.~\ref{Optimalfit}.
261:
262: The fit is not perfect, mostly because the spectrum is not symmetric. Such
263: an asymmetric spectrum is a result of correlations between different \textit{%
264: a priory} random parameters in the Hamiltonian \cite{HasseJS00}. For
265: example, if the shift tensor $K$ is also non uniform in the sample but its
266: values are correlated with the values of $\nu _{Q}$, the spectrum could be
267: non symmetric. Attempts to take this kind of effect into account failed due
268: to the enormous increase in computing time. We continue the discussion based
269: on the best fit we could practically achieve. For a discussion on the error
270: bars evaluation see Ref.~\cite{KanigelToBe}. We could not determine, $K_{z}$%
271: , $K_{x}$ and $K_{y}$ very accurately, and found very small $\eta $ for all
272: the samples, ranging from 0 to 0.1, in agreement with various estimates for
273: Y123 \cite{eta}.
274:
275: The results for fitted $\nu _{Q}$ and $\Delta \nu _{Q}$ are shown in Fig.~%
276: \ref{NuqDnuq} (a) and (b), respectively. From Fig.~\ref{NuqDnuq}(a) it is
277: clear that $\nu _{Q}$ grows linearly with doping in the underdoped side of
278: the phase diagram, in agreement with other compounds \cite%
279: {Yasuoka,Zheng,Gippius}, and Eq.~\ref{nuqvsn}. It is also clear that $\nu
280: _{Q}^{0}$ in this equation is $x$-dependent, but this could be attributed to
281: NQR base line properties. In contrast, the behavior in the overdoped side of
282: the $x=0.1$ is surprising since $\nu _{Q}$ saturates. As pointed out before
283: in Ref.~\cite{KanigelToBe}, the added holes on the overdoped side no longer
284: go into the planes.
285:
286: \begin{figure}[tbp]
287: \begin{center}
288: \includegraphics[width=9cm]{NuqDnuq.EPS}
289: \end{center}
290:
291: \caption{(a) Nuclear quadrupole resonance frequncy of $^{63}$Cu(2) in all
292: the CLBLCO samples extracted from the NMR spetra as described in the text.
293: (b) The width of the nuclear quadrupole resonance frequency distribution of $%
294: ^{63}$Cu(2) in all the CLBLCO samples extracted from the NMR spectra as
295: described in the text.}
296: \label{NuqDnuq}
297: \end{figure}
298:
299: The most interesting finding is that within the experimental error, the
300: slope of $\nu _{Q}(x,y)$ in the underdoped side is $x$-independent, as
301: demonstrated by the parallel solid lines, and $\partial \nu
302: _{Q}(x,y)/\partial y=7.2(1.4)$ MHz/Oxygen. This means that the rate at which
303: holes are introduced into the CuO$_{2}$ planes, $\partial n/\partial y$, is
304: a constant independent of $x$ or $y$ in the underdoped region. Using further
305: the ubiquitous assumption that the optimal hole density, $n^{opt}$, at
306: optimal oxygenation, $y^{opt}$, is universal, we conclude that $n(x,y)$ is a
307: function only of $\Delta y=y-y^{opt}$. This result was predicted previously
308: \cite{Chmaissem} based on bond valance calculations.
309:
310: An implication of this conclusion is that the Presland \textit{et al.}
311: formula~\cite{PreslanPhysicaC91} $T_{c}/T_{c}^{\max }=1-82.6(n-0.16)^{2}$
312: could not be correct for CLBLCO. For example, the $x=0.1$ and $x=0.4$
313: samples with $T_{c}=0$ have different $\Delta y$, and according to our
314: experiment different $n$. However, the Presland formula predicts the same $n$%
315: .
316:
317: If the conversion from oxygen to holes is universal, but different families
318: begin to superconduct at different $y$ values (see Fig.~\ref{TcandUni}), it
319: means that the rate at which holes turn into superconducting carriers is
320: family-dependent. To demonstrate this concept further we focus on the region
321: of the CLBLCO phase diagram where $T_{c}$ grows linearly with doping, as
322: emphasized by the solid lines in Fig.~\ref{TcandUni}(a). This behavior is
323: unique to CLBLCO, which does not show a $1/8$ dip or plateau in $T_{c}$. We
324: determine $n_{s}$ from $\mu $SR measurement at $T\rightarrow 0$. The $\mu $%
325: SR relaxation rate $\sigma $ is a measure of the density of superconducting
326: carriers, and $\sigma /\sigma ^{opt}=n_{s}/n_{s}^{opt}$. In the inset of
327: Fig.~\ref{TcandUni}(b)\ we depict $\sigma /\sigma ^{opt}$, taken from Ref.~%
328: \cite{KerenSSC03}, as a function of $y$ for two extreme families with $x=0.1$
329: and $0.4$. In this particular region $\partial n_{s}/\partial y$ is a
330: constant, which is family-dependent and denoted hereafter by $K(x)$. Using
331: the universality of $\partial n/\partial y$ we find that $\partial
332: n_{s}/\partial n=K(x)(\partial n/\partial y)^{-1}$, which varies between
333: families. This, again, rules out the possibility that $n_{s}=n$ in all
334: CLBLCO samples.
335:
336: In light of previous work a stronger conclusion could in fact be extended to
337: the entire phase diagram of CLBLCO. In Fig.~\ref{TcandUni}(b), taken from
338: Ref.~\cite{OferToBe}, we depict $T_{N}^{cor},T_{g}$ and $T_{c}$, normalized
339: by $T_{c}^{\max }$ of each family. $T_{N}^{cor}$ is the N\'{e}el temperature
340: after the contribution from anisotropies have been divided out. The scaled
341: variable $K(x)\Delta y$, with $K=0.77,$ $0.67$, $0.54$, $0.47$, for the $%
342: x=0.1$ to $0.4$ respectively, collapsed the entire phase diagram into a
343: single function. However, no interpretation was given for $K(x)$. The
344: present experiment suggests that $\partial n_{s}/\partial y=K(x)$ over the
345: entire phase diagram, and that $n_{s}$ should be considered as carriers
346: participating in both superconducting and antiferromagnetic order parameters.
347:
348: As for the line width, examination of Fig.~\ref{NuqDnuq}(b) reassures us of
349: our previous intuition that $\Delta \nu _{Q}$ is not changing between the
350: different families, especially for the optimally doped samples. This becomes
351: obvious when considering $\Delta \nu _{Q}$ of YBCO$_{7}$ at $T=100$~K, which
352: also has $\nu _{Q}=31$ MHz, but $\Delta \nu _{Q}=0.5$~MHz \cite{OferPRB06}.
353: The changes in $\Delta \nu _{Q}$ between $x=0.4$ and $x=0.1$ are minute
354: compared to the changes between $x=0.1$ and YBCO$_{7}$. Yet $x=0.4$ has a $%
355: T_{c}^{max}$ very similar to YBCO$_{7}$. Thus, if YBCO is considered
356: disorder free, the difference in WCD between $x=0.4$ and $x=0.1$ either does
357: not exist or is not relevant. The same conclusion was reached by MAS Ca NMR
358: \cite{MarchandThesis}.
359:
360: In summary, we find that the rate at which holes are doped into the planes
361: when oxygen is added, $\partial n/\partial y$, is identical in all families
362: in the superconducting underdoped region. In part of this region, the rate
363: at which holes contribute to the condensate $\partial n_{s}/\partial
364: n\propto K(x)$ is a family-dependent constant. Based on this and previous
365: findings, we conclude that in CLBLCO not all oxygens contribute holes to the
366: superconducting and antiferromagnetic order parameters. We also show that
367: weak-chemical-disorder, as determined from $\Delta \nu _{Q}$, is not playing
368: a role in the $T_{c}^{max}$ variations. This reinforces previous conclusions
369: that the changes in $T_{c}^{max}$ between different CLBLCO families is
370: caused only by variation in the in-plane coupling constant $J$ \cite%
371: {OferToBe,KanigelPRL02}. Finally, since $\nu _{q}$ is not changing in the
372: overdoped side as the oxygen level increases, in this side holes are not
373: added to the planes. The open question in this study is which physical or
374: chemical property of CLBLCO sets $K(x)$.
375:
376: This work was funded by the Israeli Science Foundation. We are grateful to
377: Arkady Knizhnik for his help with the Iodometric titration and to Assa
378: Auerbach for helpful discussions.
379:
380: \begin{thebibliography}{99}
381: \bibitem{Tallon} J. L. Tallon, J. R. Cooper, S. H. Naqib, and J. W. Loram,
382: Phys. Rev. B \textbf{73}, 180504(R) (2006).
383:
384: \bibitem{KerenSSC03} A. Keren, A. Kanigel, J. S. Lord, A. Amato, Solid State
385: Commun. \textbf{126}, 39 (2003). A. Kanigel, A. Keren, A. Knizhnik, O.
386: Shafir, Phys. Rev. B \textbf{71}, 224511 (2005).
387:
388: \bibitem{InhomExp} J. M. Tranquada \textit{et al}, Nature \textbf{375}, 561
389: (1995); Ch. Niedermayer, C. Bernhard, T. Blasius, A. Golnik, A. Moodenbaugh,
390: and J. I. Budnick, Phys. Rev. Lett. 80, 3843 (1998); C. Howald, P. Fournier,
391: and A. Kapitulnik, Phys. Rev. B \textbf{64}, 100504(R) (2001); S. H. Pan
392: \textit{et al.,} Nature \textbf{413}, 282 (2001); M. Vershinin \textit{et
393: al.,} Science \textbf{303}, 1995 (2004).
394:
395: \bibitem{EmeryPhysicaC} V. J. Emery and S. A. Kivelson, Physica C \textbf{209%
396: }, 597 (1993);
397:
398: \bibitem{Boson} E. Altman and A. Auerbach Phys. Rev. B \textbf{65}, 104508
399: (2002); J. Ranninger, J.M. Robin, and M. Eschrig, Phys. Rev. Lett. \textbf{74%
400: }, 4027 (1995); R. Friedberg and T.D. Lee, Phys. Rev. B \textbf{40}, 6745
401: (1989).
402:
403: \bibitem{Pines} D. Pines, cond-mat/0404151.
404:
405: \bibitem{OferToBe} R. Ofer \textit{et al., }cond-mat/0606695.
406:
407: \bibitem{Asayama} K. Asayama \textit{et al.,} Prog. Nuc. Mag. Reso. Spec.
408: \textbf{28}, 221 (1996).
409:
410: \bibitem{Yasuoka} H.Yasuoka, in Spectroscopy of Mott Insulator and
411: Correlated Metals, edited by A. Fujimori and Y. Tokura, Solid State
412: Sciences, Vol. 119 (Springer-Verlag, Berlin, 1995), p. 213.
413:
414: \bibitem{Zheng} G.-q. Zheng \textit{et al.}, J. Phys. Soc. Jpn. \textbf{64},
415: 2524 (1995).
416:
417: \bibitem{Gippius} A. A. Gippius \textit{et al.}, Physica C \textbf{276}, 57
418: (1997).
419:
420: \bibitem{clblco} D. Goldschmidt, A. Knizhnik, Y. Direktovitch, G. M.
421: Reisner, and Y. Eckstein, Phys. Rev. B\textbf{\ 49}, 15928 (1994).
422:
423: \bibitem{Slichter} C. P. Slichter, Principles of Magnetic Resonance, Harper
424: and Row, New York, (1963).
425:
426: \bibitem{Taylor} J. F. Baugher, P.C. Taylor, T. Oja and P.J. Bray, J. of
427: Chem. Phys. \textbf{50}, 4914 (1969).
428:
429: \bibitem{KerenPRB98} A. Keren \textit{et al.,} Phys. Rev. B \textbf{57},
430: 10745 (1998).
431:
432: \bibitem{MarchandThesis} S. Marchand, Ph.~D. thesis, Universite Paris 6.
433:
434: \bibitem{KanigelToBe} A. Kanigel and A. Keren\textit{.,} cond-mat/0606475.
435:
436: \bibitem{Chmaissem} O. Chmaissem, Y. Eckstein, C. G. Kuper, Phys. Rev. B%
437: \textbf{\ 63}, 174510 (2001).
438:
439: \bibitem{PreslanPhysicaC91} M. R. Preslan \textit{et al.}, Physica C \textbf{%
440: 176,} 95 (1991).
441:
442: \bibitem{NumericalRec} W. H. Press \textit{et al.}
443: \newblock{numerical
444: Recipes in C}, Cambridge University Press (1988).
445:
446: \bibitem{HasseJS00} J. Haase, C. P. Slichter, R. Stern, C. T. Milling and D.
447: G. Hinks, J. Supercond. \textbf{13}, 723 (2000).
448:
449: \bibitem{eta} C. H. Pennington, D. J. Durand, D. B. Zax, C. P. Slichter, J.
450: P. Rice, and D. M. Ginsberg, Phys. Rev. B \textbf{37}, R7944 (1988); T.
451: Shimizu \textit{et al.}, Journal of the Physical Society of Japan, \textbf{57%
452: }, 2494 (1988).
453:
454: \bibitem{KanigelPRL02} A. Kanigel \textit{et al., }Phys. Rev. Lett. \textbf{%
455: 88}, 137003 (2002).
456:
457: \bibitem{OferPRB06} R. Ofer, S. Levy, A. Kanigel, and A. Keren, Phys. Rev. B
458: \textbf{73}, 012503, (2006).
459:
460: \bibitem{StollUnKnown} E. P. Stoll and P. F. Meier, Private Communication.
461: \end{thebibliography}
462:
463: \end{document}
464: