1: %%
2: %% Revised manuscript for final publication in Nature
3: %% 5 December 2001; Publication date 13 December 2001
4: %% 13 December corrected caption Fig2.
5: %% Conversion of 7 November version to Revtex PRL style
6: %% Same version as PDF --- R.W.Hill 13/12/01
7:
8: \documentclass[twocolumn,showpacs,prl,aps,floats,amssymb]{revtex4}
9: %\documentclass[twocolumn,showpacs,prl,aps,amssymb]{revtex4}
10: \usepackage{graphicx}
11:
12: %\usepackage[dvipdfm,backref=none,colorlinks=false,bookmarks=true,bookmarksopen=true,bookmarksnumbered=false,breaklinks=true,hyperfigures=false]{hyperref}
13: %\usepackage[backref=none,colorlinks=false,bookmarks=false,bookmarksopen=false,bookmarksnumbered=false,breaklinks=true,hyperfigures=false]{hyperref}
14: %\hypersetup{
15: % pdftitle={Breakdown of Fermi-liquid theory in a cuprate superconductor},
16: % pdfauthor={R.W. Hill, Cyril Proust, Louis Taillefer, P. Fournier and R.L. Greene},
17: % pdfsubject={Nature article},
18: % pdfkeywords={Unconventional superconductivity, heat transport, electron doped, high temperature superconductor},
19: % pdfcreator={},
20: % pdfproducer={}
21: %}
22:
23:
24: \begin{document}
25:
26: \title{Breakdown of Fermi-liquid theory in a cuprate superconductor}
27:
28: \author{R.W. Hill,
29: Cyril Proust$^\dagger$ and Louis Taillefer}
30:
31: \address{Canadian Institute for Advanced Research}
32:
33: \address{
34: Department of Physics, University of Toronto, Toronto,
35: Ontario M5S 1A7, Canada}
36:
37: \author{P. Fournier$^\ddagger$ and R.L. Greene}
38:
39: \address{Center for Superconductivity Research\\ Department of Physics,
40: University of Maryland, College Park, Maryland 20742, USA}
41:
42: \date{December 13, 2001}
43:
44:
45: \begin{abstract}
46:
47: The behaviour of electrons in solids is remarkably well described by Landau's Fermi-liquid
48: theory, which says that even though electrons in a metal interact they can still be treated
49: as well-defined fermions, called ``quasiparticles''.
50: At low temperature, the ability of quasiparticles
51: to transport heat is strictly given by their
52: ability to transport charge, via a universal relation known as the Wiedemann-Franz law,
53: which no material in nature has been known to violate.
54: High-temperature superconductors have long been thought to fall outside the realm of
55: Fermi-liquid theory, as suggested by several anomalous properties, but this has yet to
56: be shown conclusively.
57: Here we report on the first experimental test of the Wiedemann-Franz law in a
58: cuprate superconductor, (Pr,Ce)$_2$CuO$_4$.
59: Our study reveals a clear departure from the universal law and provides
60: compelling evidence for the breakdown of Fermi-liquid theory in high-temperature
61: superconductors.
62:
63: \end{abstract}
64:
65: \pacs{74.70.Tx, 74.25.Fy}
66:
67: \maketitle
68:
69: \noindent
70: {\bf INTRODUCTION}\\
71:
72: Landau's Fermi-liquid theory is the definitive theory of electrons in metals \cite{Landau}, or
73: more generally fermions in condensed matter, and a major landmark of 20th century physics.
74: For example, it is the necessary foundation for the theory of superconductivity by Bardeen, Cooper and Schrieffer
75: (BCS) \cite{BCS}.
76: In essence, it says that
77: even in the presence of
78: interactions the low-energy excitations of a system of mobile
79: fermions can still be described in terms of
80: well-defined
81: fermionic particles, called ``quasiparticles'', with charge $e$, spin $\frac{1}{2}$ and
82: a mass $m^{\star}$, the latter being renormalized by interactions.
83: The Wiedemann-Franz (WF) law is one of the basic properties of a Fermi liquid, reflecting
84: the fact that the ability of a quasiparticle to transport heat is the same
85: as its ability to
86: transport charge, provided it cannot lose energy through collisions.
87: Reported as an empirical observation by Wiedemann and Franz in 1853
88: \cite{WFL}, the law states that
89: the heat conductivity $\kappa$ and
90: the electrical conductivity $\sigma$ of a
91: metal are related by a universal constant:
92: \begin{equation}
93: \frac{\kappa}{\sigma T} = \frac{\pi^2}{3} ~ (\frac{k_B}{e})^2 \equiv L_0
94: \end{equation}
95: where $T$ is the absolute temperature,
96: $k_B$ is Boltzmann's constant and $L_0=
97: 2.45 \times 10^{-8}$~W~$\Omega$~K$^{-2}$
98: is Sommerfeld's value for the
99: Lorenz ratio
100: $L \equiv \kappa/\sigma T$ \cite{Sommerfeld}.
101:
102: The linear power of temperature in Eq.~(1) comes from the linear temperature dependence
103: of the fermionic specific heat, through the
104: relation between heat transport and heat capacity. In kinetic theory,
105: \begin{equation}
106: \kappa = \frac{1}{3} ~ c ~ v~ l
107: \end{equation}
108: where $c$ is the specific heat of the carriers, $v$ is their average velocity and $l$ their mean free
109: path.
110: Provided the mean free path between collisions
111: is independent of temperature, as it is for the strictly elastic processes that operate at $T \to 0$,
112: the heat conduction has the same $T$ dependence as the specific heat. So for electrons,
113: it is also linear in $T$, and given by Eq.~(1).
114: By contrast, boson-like excitations such as phonons or magnons give rise to a
115: low-temperature specific heat and a heat conductivity as $T \to 0$ which
116: are proportional to $T^3$.
117:
118: Theoretically, electrons are predicted to obey the WF law at $T \to 0$
119: in a very wide range of environments \cite{Kearney}:
120: in both three dimensions or two dimensions (but not strictly
121: in one dimension), for any strength of disorder and interaction
122: (in the scaling theory of disorder and interaction) \cite{Castellani},
123: scattering \cite{Chester} and magnetic field \cite{Smrcka}.
124: Experimentally, the WF law does appear to be universal at $T \to 0$:
125: no material has been reported
126: to violate
127: it. (Note that a departure from Eq.~(1)
128: has in fact been reported for silver \cite{Pobell}, but is in
129: contradiction with other measurements \cite{Cu1,Cu2},
130: including our own
131: \cite{Ag-test}.)
132: This is true not only of simple metals like copper \cite{Cu1,Cu2},
133: but also of systems with strong
134: electron correlations -- such as the heavy fermion compounds CeAl$_3$ \cite{Ott},
135: CeCu$_6$ \cite{Amato} and UPt$_3$ \cite{Suderow} --
136: or highly anisotropic conduction --
137: such as the quasi-2D system
138: Sr$_2$RuO$_4$ \cite{Tanatar} (a ruthenate which is isostructural to the cuprate La$_2$CuO$_4$),
139: the quasi-1D organic conductor (TMTSF)$_2$ClO$_4$ \cite{Belin}, or
140: the 2D electron gas in a MOSFET \cite{Pepper}.
141: Even in the presence of so-called
142: non-Fermi liquid behaviour, where the electrical
143: resistivity is seen to deviate from the standard $T^2$ dependence,
144: the WF law still holds at $T \to 0$,
145: as observed in CeNi$_2$Ge$_2$ \cite{Kambe}.
146: All this strongly suggests that the ground state
147: of every metal investigated thus far is a Fermi liquid. Of course, if the metallic
148: state gives way to a superconducting state
149: at low temperature, then the WF law is entirely violated, as the charge is no longer
150: transported by fermions but by Cooper pairs, which carry no entropy.
151:
152: In 1986, high-temperature superconductivity was discovered in a class of oxides called cuprates,
153: and
154: fifteen years hence the fundamental question about these materials still remains:
155: are they host to a new state of matter?
156: In particular, are they Fermi liquids in their ground state, once superconductivity is removed
157: (either by application of a magnetic field or by doping)?
158: Anderson proposed early on
159: \cite{Anderson} that the cuprates are indeed fundamentally different from other
160: metals in that they are Mott insulators.
161: When a small number of electrons or holes are doped into the CuO$_2$ planes of their crystal
162: structure, these materials
163: can conduct charge reasonably well and also form a superconducting state.
164: It is believed that the
165: basic excitations of a doped Mott insulator are not fermionic quasiparticles
166: as in other metals -
167: with charge, spin and heat all carried by one and the same particle.
168: Instead, the electron is thought to ``fractionalize'' into a neutral
169: spin-$\frac{1}{2}$
170: fermion called a ``spinon'' and a spinless charge-$e$
171: boson called a ``holon'', or a ``chargon'' \cite{Kivelson}, a phenomenon
172: called ``spin-charge'' separation.
173:
174: Even though this bold proposal for a fundamentally new state of matter continues to be at the heart
175: of several current theories of the cuprates \cite{Fisher},
176: there has been little experimental evidence for
177: such spin-charge separation. On the contrary, for example,
178: a BCS/Fermi-liquid theory of $d$-wave quasiparticles
179: appears to work rather well in the superconducting state of
180: hole-doped cuprates, at least at low energy and near
181: optimal doping
182: \cite{Millis}.
183: Conceptually, a simple way to investigate the possibility of spin-charge
184: separation
185: is to measure the properties
186: of charge transport and compare them to those of either spin or entropy
187: transport.
188: In this Article, we report on the first test of this kind, where both the
189: charge conductivity $\sigma$ and the heat conductivity $\kappa$ of a cuprate
190: superconductor are measured in the normal state
191: at $T \to 0$. {\sl We observe a good, metallic-like nearly temperature-independent
192: charge transport at low temperature without the corresponding
193: heat transport expected from the WF law}.
194: The normalized
195: Lorenz number $L/L_0$ goes from being much
196: less than unity at $T \to 0$ to being greater than unity above 0.2 K.
197: This is highly suggestive of spin-charge separation
198: and a compelling demonstration that the ground state of cuprates
199: can fall outside the realm of Fermi-liquid theory.\\
200: \\
201: {\bf THE MATERIAL}\\
202:
203: The cuprate material used for this study is Pr$_{2-x}$Ce$_x$CuO$_{4-y}$ (PCCO), with a doping concentration
204: of $x=0.15$, which is roughly
205: the optimum value for maximizing the critical temperature for superconductivity, where
206: $T_c \simeq$~20 K.
207: The addition of $x$ Ce atoms on Pr sites adds $x$ electrons to the CuO$_2$ planes of the parent
208: insulating compound, the Mott insulator Pr$_2$CuO$_4$.
209: PCCO is the electron-doped analog of the hole-doped
210: material La$_{2-x}$Sr$_x$CuO$_{4-y}$ (LSCO), wherein the addition of $x$ Sr atoms on La sites
211: removes $x$ electrons from (adds $x$ holes to)
212: the CuO$_2$ planes.
213: In LSCO, the optimum doping is also at approximately $x=0.15$, with a maximum $T_c \simeq$~40 K.
214: For reasons that are not clear, both materials have critical temperatures that are significantly
215: lower than those of several other cuprates, such as YBa$_2$Cu$_3$O$_{7-\delta}$ (Y-123),
216: Bi$_2$Sr$_2$CaCu$_2$O$_{8+\delta}$ (Bi-2212)
217: and Tl$_2$Ba$_2$CuO$_{6+\delta}$ (Tl-2201), which all have a maximal $T_c$ around 90 K.
218: Nevertheless, all cuprates share
219: the same basic crystal structure, namely that of a stack of CuO$_2$ planes, and the
220: same generic dependence on doping.
221: The technical advantage of LSCO and PCCO is that
222: the magnetic field needed to destroy superconductivity is correspondingly lower, with upper critical
223: field values of $\simeq$~50 T and $\simeq$~10 T, respectively. Because these fields can be achieved in the
224: laboratory, it is possible to investigate their low-temperature transport properties in the normal
225: state, such as was done for charge conductivity
226: by Boebinger {\it et al.}
227: on LSCO \cite{Boebinger} and by Fournier {\it et al.} on PCCO \cite{Fournier}.\\
228: \\
229: {\bf CRYSTALS}\\
230:
231: The single crystal of PCCO used in this study was grown with a flux technique described elsewhere
232: \cite{Growth1,Growth2}, using a Al$_2$O$_3$
233: crucible and an oxygen reduction treatment at 1000 $^{\circ}$C.
234: The concentration of cerium was fixed at $x=0.15$.
235: Contacts were made with silver epoxy, diffused at 500 $^{\circ}$C for 1 hour, and were used to measure
236: both electrical resistivity and thermal conductivity.
237: The electrical resistance of contacts was typically
238: $\simeq 1~\Omega$.
239: The crystal was in the shape of a platelet, 37 $\mu$m thick and 720 $\mu$m wide, with 1.0 mm
240: separation between voltage contacts.
241: The superconducting transition temperature $T_c$ was obtained from three different measurements:
242: as the end of the drop in resistivity, the end of the diamagnetic drop in susceptibility
243: and the onset
244: of a small peak in thermal conductivity.
245: In all cases, $T_c = 20$~K to within less than $\pm 0.5$~K.
246: The width of the transition is 6~K in resistivity, 3~K in magnetisation and about 1~K in
247: thermal conductivity. The broad resistive transition, seen both as a function of temperature and
248: magnetic field, is most likely due to an inhomogeneous distribution of oxygen near the surface.
249: The narrower transition
250: obtained in thermal conductivity suggests that this inhomogeneity is confined to a negligible volume
251: of the sample, and therefore does not affect bulk measurements such as heat transport.
252: Note that all results presented in this paper were reproduced on
253: several other single crystals grown in similar conditions, with comparable $T_c$ and resistivity.\\
254:
255: \begin{figure}[b]
256: \centering
257: \resizebox{\columnwidth}{!}{\includegraphics*{Fig1_final.eps}}
258: \vspace{4pt}
259: \caption{Electrical resistivity of PCCO vs temperature
260: for a current in the basal plane at different values of magnetic field applied
261: normal to the plane. {\it Upper inset}: zoom of low temperature dependence at 13 T and
262: 14 T, the latter being shifted upwards by 0.8~$\mu \Omega$~cm for clarity. Lines
263: are fits to the data below 1.0~K. {\it Lower inset}:
264: Resistivity vs magnetic field at $T=0.3$~K, showing that
265: the crystal has fully completed its transition to the normal state by 13.5~T.
266: The line is a linear fit
267: to the negative magnetoresistance beetween 13.5 and 15~T.}
268: \label{Fig2}
269: \end{figure}
270:
271: \noindent
272: {\bf CHARGE TRANSPORT}\\
273:
274: The temperature dependence of the electrical resistivity is shown in Fig.~\ref{Fig2}
275: for different magnetic fields applied
276: parallel to the $c$-axis of the tetragonal crystal structure.
277: The current is made to flow in the CuO$_2$ planes,
278: where the conduction is some
279: 10$^4$ times better than
280: perpendicular to the planes (at room temperature).
281: Above 25~K, the temperature dependence is not linear as in most optimally-doped cuprates,
282: but follows a power law $a +
283: b T^{\alpha}$ with
284: $\alpha \simeq 1.7$,
285: and has a magnitude characteristic of a reasonably good metal: $a \simeq 25~\mu \Omega$~cm
286: and $\rho$(300~K) $\simeq$ 140 ~$\mu \Omega$~cm.
287: The effect of a magnetic field is to rapidly
288: suppress the superconducting transition initially, but then more slowly at high fields, giving
289: rise to a curve $T_c(H)$ that has strong positive curvature.
290: From the
291: magnetoresistance at 0.3 K shown in the lower inset, one can see
292: that the entire sample has completed its
293: transition to the normal state by 13.5 T.
294: Note, however, that the bulk of the sample is no longer superconducting above 8 T, the field
295: beyond which the low-temperature
296: heat transport ceases to evolve (see Fig.~\ref{Fig3} below).
297:
298: The overall behaviour of the resistivity in PCCO
299: is similar to that of hole-doped cuprates when overdoped.
300: Indeed, when doped to have a $T_c$ of 15~K, Tl-2201 has a resistivity which
301: goes as $T^{1.75}$ with values of
302: 10~$\mu \Omega$~cm at low temperature
303: and 180~$\mu \Omega$~cm at room temperature, and a resistive critical field with a strong upward curvature
304: reaching a value of 16 T at $T \to 0$
305: \cite{Mackenzie}.
306: One difference is the slight upturn seen at low temperature in PCCO and not in Tl-2201.
307: The resistivity in the normal state (at 14 T) increases by about 10\% in going from
308: 8~K down to
309: 0.3~K (see Fig.~\ref{Fig2}).
310: Similar but much larger upturns were observed
311: in Nd$_{2-x}$Ce$_x$CuO$_4$
312: (NCCO) \cite{Harus,Fournier2} and PCCO \cite{Fournier2} at lower carrier concentrations,
313: and attributed to weak localization.
314: In support of this interpretation is the suppression of the upturn by a transverse
315: magnetic field, also observed in our samples (see inset of Fig.~\ref{Fig2}).
316: This
317: negative magnetoresistance
318: eventually flattens to a constant beyond 25 T, equal to 25.2~$\mu \Omega$~cm at 0.5 K \cite{Hill}.
319:
320: In summary, once the superconducting order has been completely suppressed throughout the sample,
321: namely in a magnetic field larger
322: than 13.5 T, the charge transport of this cuprate material at low temperature
323: is that of a fairly
324: good quasi-two-dimensional
325: metal, with a residual resistivity $\rho(T \to 0) = \rho_0 = 26.8~
326: \mu \Omega$~cm in 14 T,
327: which corresponds
328: to a conductivity
329: per CuO$_2$ plane of 60~$e^2/h$ ({\it i.e.} $k_F l \simeq 60$).\\
330: \\
331: \noindent
332: {\bf HEAT TRANSPORT}\\
333:
334: Fifteen years after the discovery of high-temperature superconductivity,
335: very little is known about the normal state
336: of cuprates at low temperature.
337: The ability to suppress superconductivity in optimally-doped PCCO
338: with a field of only 14 T
339: provides a rare opportunity to shed light on the fundamental nature of the electron system
340: in these
341: materials. In the present study, this
342: was done by measuring the transport of heat
343: as $T \to 0$. The thermal conductivity $\kappa(T)$ of PCCO is shown in Fig.~\ref{Fig3}
344: at temperatures below
345: 0.3~K, for heat flowing parallel to the CuO$_2$ planes.
346: The data is plotted as $\kappa/T$ vs $T^2$ in order to easily separate the contributions of phonons
347: and electrons, using the fact that
348: phonons are expected
349: to have a heat
350: conductivity
351: which varies as $T^3$ when their mean
352: free path reaches the size of the crystal at sufficiently
353: low temperature. A straight line extrapolating to zero in $\kappa/T$ vs $T^2$ is
354: indeed the asymptotic ($T \to 0$) behaviour
355: observed in insulators \cite{Thacher}, including the Mott insulator
356: YBa$_2$Cu$_3$O$_6$ ({\it i.e.} Y-123 without
357: any doped carriers) \cite{Taillefer}. This is also the behaviour of conventional ($s$-wave)
358: superconductors.
359: The absence of a residual
360: linear term $\kappa_0/T$, defined as the value of $\kappa/T$ as $T \to 0$,
361: means that there are no mobile fermionic
362: excitations at low temperature. In insulators, this is due
363: to the lack of electronic carriers, while in superconductors it is because electrons
364: have all formed Cooper pairs at $T << T_c$. The heat
365: is then conducted entirely by phonons.
366: In contrast, a residual linear term
367: is observed
368: when fermionic excitations are present, as in the following three
369: instances:
370: 1) in metals, with a residual linear term given precisely by the WF law;
371: 2) in the vortex state of
372: superconductors, where a magnetic field induces a residual linear term
373: by generating
374: delocalized quasiparticles
375: \cite{Boaknin2};
376: 3) in $d$-wave superconductors, where zero-energy quasiparticles are induced by impurity scattering
377: \cite{Durst}.\\
378: \begin{figure}[t]
379: \centering
380: \resizebox{\columnwidth}{!}{\includegraphics*{Fig2_final.eps}}
381: \vspace{4pt}
382: \caption{Thermal conductivity of PCCO for a heat current in the basal
383: plane, plotted as $\kappa/T$ vs $T^2$,
384: at different values of the magnetic
385: field applied normal to the plane.
386: The solid line is a linear fit to the zero-field data below
387: 130 mK.
388: The dashed line shows the behaviour of a Fermi-liquid with the residual resistivity
389: $\rho_0$ of this sample, calculated as the sum of a constant electronic
390: linear term via Eq.~(1) and a phonon conductivity
391: given by the zero-field data.
392: }
393: \label{Fig3}
394: \end{figure}
395:
396: \noindent
397: {\bf Superconducting state}\\
398:
399: In the absence of a magnetic field,
400: there is only a very small residual linear term in the thermal conductivity of PCCO.
401: From data on four different crystals, we find $\kappa_0/T = 0.03 - 0.07$~ mW K$^{-2}$ cm$^{-1}$.
402: This is considerably smaller than the
403: residual linear term observed in three hole-doped cuprates, at
404: optimal doping:
405: $\kappa_0/T$ = 0.14, 0.15 and 0.11 mW K$^{-2}$ cm$^{-1}$, in
406: Y-123 \cite{Taillefer,Chiao}, Bi-2212 \cite{Chiao2,Aubin} and LSCO \cite{Taillefer2},
407: respectively.
408: Within BCS theory applied to a $d$-wave superconductor, this fermionic residual heat conduction
409: is expected, arising from zero-energy quasiparticles
410: induced by impurity scattering near the nodes in the $d_{x^2-y^2}$ gap function \cite{Durst}.
411: In the case of Bi-2212, the
412: excellent quantitative agreement between theory and experiment has been viewed as a strong
413: validation of a Fermi-liquid quasiparticle picture for the superconducting state
414: \cite{Millis,Chiao2},
415: at least at optimal doping and low energies.
416:
417: Several tentative explanations
418: may be offered for the weakness of the residual
419: linear term in PCCO.
420: The first possibility
421: is the localization of $d$-wave quasiparticles,
422: investigated by several authors \cite{Lee}.
423: It was invoked recently as a possible cause for the
424: absence of a residual linear term in YBa$_2$Cu$_4$O$_8$
425: (Y-124) \cite{Hussey}. However, it
426: is not clear why localization should occur in Y-124 and not in Y-123,
427: given that these two cuprates have comparable
428: structure and properties.
429: It seems more
430: plausible in PCCO, because of the highly two-dimensional nature of its conduction --
431: some 100 times more anisotropic than either Y-124 or LSCO.
432: On the other hand, Bi-2212 is equally anisotropic, yet shows delocalized quasiparticles.
433: A second scenario is the absence of nodes in the gap structure of PCCO.
434: This
435: would eliminate
436: any
437: residual linear term (at least in zero magnetic field). The symmetry of the order parameter
438: need not be $s$-wave, but could
439: have the form $d+ix$, where $x$ is a subdominant component of either $s$ or $d_{xy}$ symmetry.
440: The possibility of a $T=0$ transition in the order parameter from $d$ to $d+ix$
441: as a function of doping has been
442: raised in the context of a
443: possible quantum critical point in the phase diagram of cuprates \cite{Sachdev}.
444: For PCCO, however,
445: this would seem to be in conflict with evidence for a rather pure $d_{x^2-y^2}$ symmetry as
446: determined in a recent tricrystal experiment \cite{Tsuei}. The other difficulty
447: is that, as we shall see below,
448: a magnetic field normal to the CuO$_2$ planes never induces a residual
449: linear term, implying that the minimum gap would have to be a sizable fraction of the maximum
450: gap ({\it i.e.} $x$ comparable to $d$). This would have to be reconciled with the penetration
451: depth data \cite{lambda1,lambda2}.
452: A third possibility, proposed recently by Granath {\it et al}. \cite{Granath}, is the removal
453: of the nodes as a result of static stripe ordering.\\
454: \\
455: \noindent
456: {\bf Normal state}\\
457:
458: As seen in Fig.~\ref{Fig3},
459: the basic effect of a magnetic field is to rapidly increase the thermal conductivity
460: of PCCO (at finite temperature).
461: Given that the phonon conductivity cannot exceed
462: its value at zero field (where the phonon mean free path has reached its maximum value), this increase
463: must be due
464: to excitations of electronic origin.
465: The field-induced conduction (over and above the zero-field curve)
466: grows rapidly (and approximately linearly) with field at low fields
467: to eventually saturate at
468: high fields, where $\kappa$ is absolutely constant above 8 T.
469: This is compelling evidence that the thermodynamic upper
470: critical field in this sample is at most
471: 8 T and the bulk of the crystal is in the normal state beyond that.
472: Indeed, in all known superconductors, whether $s$-wave or $d$-wave, the electronic
473: thermal conductivity is always affected by the magnetic field when in the vortex state,
474: and seen to saturate above $H_{c2}$ (see for example Ref.~\cite{Belin2}).
475: \begin{figure}[t]
476: \centering
477: \resizebox{\columnwidth}{!}{\includegraphics*{Fig3_final.eps}}
478: \vspace{4pt}
479: \caption{Comparison of charge conductivity $\sigma(T)=1/\rho(T)$, plotted
480: as $L_0/\rho(T)$ (open triangles), and electronic
481: heat conductivity $\kappa_e$, plotted as
482: $\kappa_e/T$ (full circles), as a function of temperature in the normal state at $H=13$~T.
483: The electronic contribution to the heat conduction
484: is the difference between the measured $\kappa$(13 T) and the phonon contribution
485: $\kappa_{ph}$(13 T), estimated in the text.
486: In a Fermi liquid, the curve of
487: $\kappa_e/T$ would lie precisely on top of the data for
488: $L_0/\rho$. Below 0.15~K, $\kappa_e \sim T^{3.6}$.
489: }
490: \label{Fig4}
491: \end{figure}
492: \noindent
493: Such an $H_{c2}$ agrees reasonably well
494: with the criterion of zero resistance, achieved at about 6~T, and also
495: agrees with the $H_{c2}$ values reported previously in thin films of PCCO \cite{Fournier,Alff}
496: and NCCO \cite{Osofsky,Alff} at $x=0.15$.
497: A roughly linear growth of $\kappa$ with field is also seen,
498: for example, in the borocarbide superconductor
499: LuNi$_2$B$_2$C \cite{Boaknin2},
500: however with a major difference: in LuNi$_2$B$_2$C, the field induces a
501: linear term at $T \to 0$.
502: Furthermore, this finite residual linear term ultimately
503: reaches a value in the normal state given precisely
504: by the WF law. In striking contrast, {\sl a magnetic field, even as high as
505: 13~T, almost twice $H_{c2}$, never induces a residual linear term in PCCO.}
506: Given the known value of the electrical
507: resistivity at $T \to 0$ and 13~T, namely $\rho_0 = 26.8~ \mu \Omega$~cm, the WF law predicts
508: a residual linear term $\kappa_0/T = L_0/\rho_0 = 0.91$~mW~K$^{-2}$~cm$^{-1}$, as drawn in
509: Fig.~\ref{Fig3}.
510: (Note that weak localization is not expected to alter the WF law appreciably
511: \cite{Castellani},
512: as verified experimentally in
513: layered graphene \cite{graphene}.)
514:
515: This complete violation of the WF law shows that the non-superconducting
516: ground state of this cuprate material
517: is not a Fermi liquid. {\sl
518: The charge carriers responsible for the good electrical conduction
519: do not display the expected fermionic heat transport}.
520:
521: It is instructive to consider the effect of
522: increasing the temperature above zero. Contributions to the heat transport from the electron system
523: ($\kappa_e$) and from the ionic lattice ($\kappa_{ph}$) will add to give the measured thermal
524: conductivity: $\kappa(T) = \kappa_e(T) + \kappa_{ph}(T)$.
525: These two systems are coupled. It is unambiguous that phonons are scattered by electrons
526: since the crystal conducts heat better at zero field than at 13 T
527: beyond 0.6~K.
528: In other words, as
529: the temperature is raised, inelastic scattering processes become increasingly
530: effective
531: at limiting the phonon
532: heat flow, and we expect the dominant process to be electron-phonon scattering.
533:
534: Given this observed electron-phonon coupling, the total phonon scattering
535: in field is the sum of whatever scatters phonons in zero field plus the
536: added scattering that comes from the field-induced electronic excitations.
537: The thermal resistivity
538: of phonons due to electron scattering is given by
539: $W_e^{ph} = B T^{-2}$ \cite{Berman}, where the constant $B$ depends on the strength of the electron-phonon coupling, the electronic density of states, and various other factors. This means
540: that the phonon conductivity at low $T$ is given by
541: $\kappa_{ph} = \kappa(0) [ 1 + \kappa(0) W_e^{ph} ]^{-1}$, where $\kappa(0)$ is the
542: measured zero-field conductivity, assumed to be essentially all phononic.
543: %%%% REVISION: corrected spelling of "essentially" %%%%%%%%%%
544: We can put a reasonable lower bound on the electron-phonon coupling
545: parameter
546: $B$ by requiring that the resulting $\kappa_e/T$ does not
547: decrease with $T$. Adjusting $B$ such that $\kappa_e/T$ is constant
548: above 0.3~K gives the curve shown in Fig.~3 for $H = 13$~T (full circles).
549: The same can be done for other fields with the resulting values of $\kappa_e/T$ = 0.6,
550: 1.2 and 1.9 mW~K$^{-2}$~cm$^{-1}$ and corresponding values of $B$ = 0.22, 0.4, and 0.68
551: K$^3$~m~W$^{-1}$ for $H$ = 2, 4, and 8 (or 13) T, respectively. (In
552: comparison, electron-phonon coupling in copper gives $B = 4-8$~K$^3$~m~W$^{-1}$
553: \cite{Berman}.)
554: Note the self-consistency of this analysis
555: whereby $B$ scales with the value of $\kappa_e/T$ (above 0.3 K);
556: in other words, by increasing the field -- thereby increasing the
557: density of electronic excitations -- one increases {\it both} the electronic heat
558: conduction
559: and the electron scattering of phonons.
560:
561: From
562: Fig.~\ref{Fig4}, we see that $\kappa_e(T) / T$ exceeds the value of $L_0/\rho$
563: appropriate for a Fermi liquid.
564: In other words,
565: the low-energy excitations of the electron system
566: violate the WF law not only at $T \to 0$, where
567: $\kappa_e(T) / T \ll L_0/\rho_0$, but also at finite temperatures, where
568: $\kappa_e(T) / T > L_0/\rho$ (above 0.2~K).
569: It is important to appreciate that qualitatively
570: the latter violation is independent of our choice of $B$, in the sense that
571: even if $B$ is set to zero, $\kappa_e/T$ still exceeds $L_0/\rho$, in this case by
572: 30\% at 0.25~K. We stress that $B=0$ is unphysical, as it would
573: lead to a
574: {\it negative} $\kappa_e/T$ for $T > 0.6$~K.
575: %%%%%%%% REVISION %%%%%%%%%%
576: It is also independent of our assumption that $\kappa(0)$ is entirely phononic;
577: indeed if some part of $\kappa(0)$ is in fact electronic, then the violation turns out to be
578: even more pronounced. In other words, while there is no accurate way to extract the electronic
579: contribution at 13 T, the fact that the WF law is violated is independent of our model for the phonon
580: conduction and the electron-phonon scattering. The result $\kappa_e(T) / T \simeq
581: 2 L_0/\rho_0$ above 0.3~K is obtained in several crystals, even though $\rho(T_c)$ varies in
582: the range
583: 20 to 40~$\mu \Omega$~cm.
584:
585: We have investigated the possibility that our measurement gradually ceases to detect the
586: electronic heat current as electrons fall out of thermal equilibrium with the phonon
587: bath when the temperature approaches $T=0$. One can imagine this happening
588: when the electrical contact resistance between the heater and the sample
589: is too large and the heat is predominantly carried by the phonons. This
590: might then explain the drop in $\kappa_e/T$ measured below 0.3 K or so (see Fig.~\ref{Fig4}).
591: Two facts lead us to rule out electron-phonon decoupling as the main mechanism for the
592: drop. First, identical data is obtained by sending heat directly through the electron system
593: using photons (as in Ref.~\cite{Decoupling}). Secondly, a similar drop is observed in samples
594: of optimally-doped (La,Sr)$_2$CuO$_4$, where we have achieved electrical contact resistances that
595: are 100 times lower than in our PCCO samples, namely ~ 10~m$\Omega$.
596: We therefore believe this low-temperature drop is intrinsic and not
597: specific to PCCO.\\
598: %%%%%%%%%%%%%%%%%%%%%%%%
599: \\
600: {\bf CONCLUSION}\\
601:
602: The cuprate PCCO is the first material to violate the Wiedemann-Franz law.
603: The breakdown of this robust signature of Fermi-liquid theory suggests that the fundamental
604: entities that carry heat, charge (and spin) in the cuprates are not the usual
605: Landau
606: quasiparticles. The fact that the electron system conducts heat in a way which is largely
607: unrelated to the way it carries charge at low temperature points to the existence of
608: neutral excitations responsible for much of $\kappa_e$. If these neutral excitations
609: are fermions, {\it i.e.} spinons, they should contribute a constant term to $\kappa_e/T$,
610: as may well be the case for $T > 0.3$~K.
611: The drop seen at the very lowest temperatures
612: then remains to be explained, and the possibility of spinon localization should be considered.
613:
614: The present study is currently limited to a single carrier concentration, near
615: optimal doping, in the whole phase diagram. It will be interesting to investigate
616: the evolution with doping of this emerging picture. If one speculates
617: that the WF law is satisfied
618: in overdoped cuprates, where Fermi-liquid theory is generally assumed to hold, our results
619: on optimally-doped PCCO suggest that a violation of the law is a property of
620: underdoped cuprates. A comprehensive study will shed light on the gradual breakdown of
621: Fermi-liquid theory in the cuprates as the carrier concentration is reduced towards
622: the Mott insulating state.\\
623: \\
624: \noindent
625: {\bf Acknowledgements}\\
626:
627: We thank C. Lupien, E. Boaknin, M. Sutherland, M. Chiao,
628: R. Gagnon, J. Brooks, L. Balicas
629: and B. Brandt
630: for their invaluable help in various
631: aspects of the measurements.
632: This work was supported by the Canadian Institute for Advanced
633: Research and funded by NSERC of Canada.
634: LT ackowledges the support of a
635: Premier's Research Excellence Award from the Government of Ontario.
636: The work in Maryland was supported by the NSF Division of Condensed
637: Matter Physics.\\
638: \\
639: Correspondence should be addressed to L.T. (Louis.Taillefer@utoronto.ca).\\
640: \\
641: \noindent
642: $^\dagger$ Present address:
643: Laboratoire National des Champs Magn\'etiques Puls\'es,
644: 143 Avenue de Rangueil, BP 4245, 31432 Toulouse, Cedex 04, France. \\
645: $^\ddagger$ Present address:
646: D\'epartement de Physique, Universit\'e de Sherbrooke, Sherbrooke, Qu\'ebec J1K 2R1, Canada.
647:
648:
649: \begin{references}
650:
651: \bibitem{Landau} Landau, L.D.
652: The theory of a Fermi liquid.
653: {\it Soviet Physics JETP} {\bf 3}, 920-925 (1957).
654:
655: \bibitem{BCS} Bardeen, J., Cooper, L.N. \& Schrieffer, J.R.
656: Theory of superconductivity.
657: {\it Phys. Rev.} {\bf 108}, 1175-1204 (1957).
658:
659: \bibitem{WFL} Wiedemann G. \& Franz R. {\it Annl. Phys.} {\bf 89}, 497-532
660: (1853).
661:
662: \bibitem{Sommerfeld} Ashcroft, N.W. \& Mermin, N.D. {\it Solid State Physics}
663: (HRW, Philadelphia, 1976), p.~322.
664:
665: \bibitem{Kearney} Kearney, M.J. \& Butcher, P.N.
666: Thermal transport in disordered systems.
667: {\it J. Phys. C: Solid State Phys.} {\bf 21}, L265-L270 (1988).
668:
669: \bibitem{Castellani} Castellani, C., DiCastro, C., Kotliar, G., Lee, P.A. \& Strinati, G.
670: Thermal conductivity in disordered interacting-electron systems.
671: {\it Phys. Rev. Lett.} {\bf 59}, 477-480 (1987).
672:
673: \bibitem{Chester} Chester, G.V. \& Tellung, A.
674: The law of Wiedemann and Franz.
675: {\it Proc. Phys. Soc.} {\bf 77}, 1005-1013 (1961).
676:
677: \bibitem{Smrcka} Smrcka, L. \& Streda, P.
678: Transport coefficients in strong magnetic fields.
679: {\it J. Phys. C: Solid State Phys.} {\bf 10}, 2153-2161 (1977).
680:
681:
682: \bibitem{Pobell}
683: Gloos, K., Mitschka, C., Pobell, F. \& Smeibidl, P.
684: Thermal conductivity of normal and superconducting metals.
685: {\it Cryogenics} {\bf 30}, 14-18 (1990).
686:
687: \bibitem{Cu1} Anderson, A.C., Peterson, R.E. \& Robichaux, J.E.
688: Thermal and electrical conductivity of Ag and Pt below 1~K.
689: {\it Phys. Rev. Lett.}
690: {\bf 20}, 459-461 (1968).
691:
692: \bibitem{Cu2} Rumbo, E.R.
693: Transport properties of very pure copper and silver below 8.5~K.
694: {\it J. Phys. F: Metal Phys.} {\bf 6}, 85-98 (1976).
695:
696: \bibitem{Ag-test} Hill, R.W. \& Boaknin, E. {\it Personal communication}.
697:
698: \bibitem{Ott} Ott, H.R., Marti, O. \& Hulliger, F.
699: Low temperature thermal conductivity of CeAl$_3$.
700: {\it Solid State Commun.} {\bf 49}, 1129-1131 (1984).
701:
702: \bibitem{Amato} Amato, A. {\it et al}.
703: Transport properties of CeCu$_6$ at very low temperature.
704: {\it J. Magn. Magn. Mater.} {\bf 63 \& 64}, 300-302 (1987).
705:
706: \bibitem{Suderow} Suderow, H. {\it et al.}
707: Thermal conductivity and gap structure of the superconducting phases of UPt$_3$.
708: {\it J. Low Temp. Phys.} {\bf 108}, 11-30 (1997).
709:
710: \bibitem{Tanatar} Tanatar, M.A. {\it et al.}
711: Thermal conductivity of superconducting Sr$_2$RuO$_4$ in oriented magnetic fields.
712: {\it Phys. Rev. B.} {\bf 63}, 064505-1-7 (2001).
713:
714: \bibitem{Belin} Belin, S. \& Behnia, K.
715: Thermal conductivity of superconducting (TMTSF)$_2$ClO$_4$: evidence
716: for a nodeless gap. {\it Phys. Rev. Lett.} {\bf 79}, 2125-2128 (1999).
717:
718: \bibitem{Pepper} Syme, R.T., Kelly, M.J. \& Pepper, M. Direct measurement of the thermal conductivity
719: of a two-dimensional electron gas. {\it J. Phys.: Condens. Matter} {\bf 1}, 3375-3380 (1989).
720:
721: \bibitem{Kambe} Kambe, S. {\it et al}. Spin-fluctuation mediated thermal conductivity around
722: the magnetic instability of CeNi$_2$Ge$_2$. {\it J. Low Temp. Phys.} {\bf 117}, 101-112 (1999).
723:
724: \bibitem{Anderson} Anderson, P.W.
725: The resonating valence bond state in La$_2$CuO$_4$ and
726: superconductivity.
727: {\it Science} {\bf 235}, 1196-1198 (1987).
728:
729: \bibitem{Kivelson} Kivelson, S.A., Rokhsar, D.S. \& Sethna, J.P.
730: Topology of the resonating valence bond state: Solitons and high-T$_c$ superconductivity.
731: {\it Phys. Rev. B} {\bf 35}, 8865-8868 (1987).
732:
733: \bibitem{Fisher} Senthil, T. \& Fisher, M.P.A.
734: Z$_2$ gauge theory of electron fractionalization in strongly correlated systems.
735: {\it Phys. Rev. B} {\bf 62}, 7850-7881
736: (2000); {\sl and references therein}.
737:
738: \bibitem{Millis} Orenstein, J. \& Millis, A.J. Advances in the physics of high-temperature
739: superconductivity. {\it Science} {\bf 288}, 468-474 (2000).
740:
741: \bibitem{Boebinger} Boebinger, G.S. {\it et al}.
742: Insulator-to-metal crossover in the normal state of La$_{2-x}$Sr$_x$CuO$_4$ near optimum doping.
743: {\it Phys. Rev. Lett.} {\bf 77}, 5417-5420 (1996).
744:
745: \bibitem{Fournier} Fournier, P. {\it et al}.
746: Insulator-metal crossover near optimal doping in Pr$_{2-x}$Ce$_x$CuO$_4$: Anomalous normal-state low
747: temperature resistivity.
748: {\it Phys. Rev. Lett.} {\bf 81}, 4720-4723 (1998).
749:
750: \bibitem{Growth1} Peng, J.L., Li, Z.Y. \& Greene, R.L.
751: Growth and characterization of high-quality single crystals of
752: R$_{2-x}$Ce$_x$CuO$_{4-y}$ (R=Nd,Sm).
753: {\it Physica C} {\bf 177}, 79-85 (1991).
754:
755: \bibitem{Growth2} Brinkmann, M. {\it et al}.
756: Crystal growth of high-$T_c$ superconductors
757: Pr$_{2-x}$Ce$_x$CuO$_{4+\delta}$ with substitutions of Ni and Co for Cu.
758: {\it J. Crystal Growth} {\bf 163}, 369-376 (1996).
759:
760: \bibitem{Mackenzie} Mackenzie, A.P. {\it et al}. Resistive upper critical field of
761: Tl$_2$Ba$_2$CuO$_6$ at low temperatures and high magnetic fields.
762: {\it Phys. Rev. Lett.} {\bf 71}, 1238-1241 (1993).
763:
764: \bibitem{Harus} Harus, G.I {\it et al}.
765: Two-dimensional weak localization effects in high-temperature
766: superconductor Nd$_{2-x}$Ce$_x$CuO$_{4-\delta}$.
767: {\it Soviet JETP} {\bf 116}, 1-12 (1999).
768:
769: \bibitem{Fournier2} Fournier, P. {\it et al.} Anomalous saturation of the
770: phase coherence length in underdoped Pr$_{2-x}$Ce$_x$CuO$_4$ thin films.
771: {\it Physical Review B} {\bf 62}, R11993-R11996 (2000).
772:
773: \bibitem{Hill} Hill, R.W. {\it et al}. {\it Personal communication}.
774:
775: \bibitem{Thacher} Thacher, P.D.
776: Effect of boundaries and isotopes on the thermal conductivity of LiF.
777: {\it Phys. Rev.} {\bf 156}, 975-988 (1967).
778:
779: \bibitem{Taillefer} Taillefer, L. {\it et al}.
780: Universal heat conduction in YBa$_2$Cu$_3$O$_{6.9}$.
781: {\it Phys. Rev. Lett.} {\bf 79}, 483-486 (1997).
782:
783: \bibitem{Boaknin2} Boaknin, E. {\it et al}. Highly anisotropic gap function in
784: borocarbide superconductor LuNi$_2$B$_2$C. {\it Phys. Rev. Lett.} accepted
785: Nov 2001.
786:
787: \bibitem{Durst} Durst, A.C. \& Lee, P.A. Impurity-induced quasiparticle transport and universal-limit
788: Wiedemann-Franz violation in $d$-wave superconductors. {\it Phys. Rev. B} {\bf 62}, 1270-1290 (2000).
789:
790: \bibitem{Chiao} Chiao, M. {\it et al}. Quasiparticle transport in the vortex state of
791: YBa$_2$Cu$_3$O$_{6.9}$. {\it Phys. Rev. Lett.} {\bf 82}, 2943-2946 (1999).
792:
793: \bibitem{Chiao2} Chiao, M. {\it et al}.
794: Low-energy quasiparticles in cuprate superconductors: A quantitative analysis.
795: {\it Phys. Rev. B} {\bf 62}, 3554-3558
796: (2000).
797:
798: \bibitem{Aubin} Behnia, K. {\it et al}. Features of heat conduction
799: in organic and cuprate superconductors. {\it J. Low Temp. Phys.} {\bf 117}, 1089-1098 (1999).
800:
801: \bibitem{Taillefer2} Taillefer, L. \& Hill, R.W. Heat transport in high-temperature superconductors.
802: {\it Physics in Canada}, {\bf 56}, 237-240 (2000).
803:
804: \bibitem{Lee} Senthil, T. {\it et al}. Quasiparticle transport and localization in high-$T_c$
805: superconductors. {\it Phys. Rev. Lett.} {\bf 81}, 4704-4707 (1998); {\sl and references therein}.
806:
807: \bibitem{Hussey} Hussey, N.E. {\it et al}.
808: Absence of residual quasiparticle conductivity in the underdoped
809: cuprate YBa$_2$Cu$_4$O$_8$. {\it Phys. Rev. Lett.} {\bf 85}, 4140-4143 (2000).
810:
811: \bibitem{Sachdev} Vojta, M., Zhang, Y. \& Sachdev, S.
812: Quantum phase transitions in d-wave superconductors.
813: {\it Phys. Rev. Lett.} {\bf 85}, 4940-4943 (2000).
814:
815: \bibitem{Tsuei} Tsuei, C.C. \& Kirtley, J.R.
816: Phase-sensitive evidence for d-wave pairing symmetry in electron-doped cuprate superconductors.
817: {\it Phys. Rev. Lett.} {\bf 85}, 182-185 (2000).
818:
819: \bibitem{lambda1} Kokales, J.D. {\it et al}.
820: Microwave electrodynamics of electron-doped cuprate superconductors.
821: {\it Phys. Rev. Lett.} {\bf 85}, 3696-3699 (2000).
822:
823: \bibitem{lambda2} Prozorov, R. {\it et al}.
824: Evidence for nodal quasiparticles in electron-doped cuprates from penetration depth measurements.
825: {\it Phys. Rev. Lett.} {\bf 85}, 3700-3702 (2000).
826:
827: \bibitem{Granath} Granath, M. {\it et al}.
828: Nodal quasiparticles and coexisting orders in striped superconductors.
829: {\it Phys. Rev. Lett.} {\bf 87}, 167011-1-4 (2001).
830:
831: \bibitem{Belin2} Belin, S. {\it et al}.
832: Probing the upper critical field of $\kappa$-(BEDT-TTF)$_2$Cu(NCS)$_2$.
833: {\it J. Supercon.} {\bf 12}, 497-500 (1999).
834:
835: \bibitem{Alff} Kleefisch, S. {\it et al}.
836: Possible pseudogap behavior of electron doped high-temperature superconductors.
837: {\it Phys. Rev. B} {\bf 63}, 100507-100510 (2001).
838:
839: \bibitem{Osofsky} Osofsky, M.S. {\it et al}.
840: Temperature dependence of the upper critical magnetic field in BiSrCuO and NdCeCuO.
841: {\it Proceedings of the 10th anniversary HTS workshop} (World Scientific,
842: Singapore, 1996), pp.~284-287.
843:
844: \bibitem{graphene} Bayot, V. {\it et al}.
845: Evidence for weak localization in the thermal conductivity of a
846: quasi-two-dimensional electron system. {\it Phys. Rev. Lett.} {\bf 65}, 2579-2582 (1990).
847:
848: \bibitem{Berman} Berman, R. {\it Thermal conduction in solids}
849: (Clarendon, Oxford 1976), p.~145
850:
851: %%%%%%%% REVISION %%%%%%%%%%
852:
853: \bibitem{Decoupling} Gutsmiedl, P., Probst. C. \& Andres, K.
854: Low temperature calorimetry using an optical heating method.
855: {\it Cryogenics} {\bf 31}, 54-57 (1991).
856:
857: %%%%%%%%%%%%%%%%%%%%%%%%%
858:
859: \end {references}
860:
861: \end{document}
862:
863: