1: \documentclass[prb,preprint]{revtex4}
2: %\documentclass[prl,twocolumn]{revtex4}
3: \usepackage{graphicx}
4: \hbadness 5000
5: \newcommand{\bbb}{Bi$_{2}$Sr$_{2}$Ca$_{2}$Cu$_{3}$O$_{8}$}
6: \newcommand{\bb}{Bi$_{2}$Sr$_{2}$CaCu$_{2}$O$_{8}$}
7: \newcommand{\bs}{Bi$_{2}$Sr$_{2}$CuO$_{6}$}
8: \newcommand{\hTc}{high-$T_c$}
9:
10: \begin{document}
11: \title{Doping Dependence of the Redistribution of Optical Spectral Weight in Bi$_{2}$Sr$_{2}$CaCu$_{2}$O$_{8+\delta}$}
12: \author{F. Carbone, A.B. Kuzmenko, H.J.A. Molegraaf,
13: E. van Heumen, V. Lukovac, F. Marsiglio and D. van der Marel}
14: \affiliation{ Departement de Physique de la Mati\`{e}re
15: Condens\'{e}e, Universit\'{e} de Gen\`{e}ve, 24 Quai
16: Ernest-Ansermet, CH-1211 Geneva 4, Switzerland \\}
17: \author{K. Haule, G. Kotliar}
18: \affiliation{Departement of Physics, Rutgers University, Piscataway,
19: NJ 08854, USA\\}
20: \author{H. Berger,S. Courjault}
21: \affiliation{\'{E}cole Polytechnique Federale de Lausanne,
22: Departement de Physique, CH-1015 Lausanne, Switzerland \\}
23:
24: \author{P.H. Kes,M. Li}
25: \affiliation{Kamerlingh Onnes Laboratory, Leiden University, 2300 RA
26: Leiden, The Netherlands \\}
27:
28: \date{\today}
29:
30: \begin{abstract}
31:
32: We present the ab-plane optical conductivity of four single crystals
33: of Bi$_{2}$Sr$_{2}$CaCu$_{2}$O$_{8+\delta}$ (Bi2212) with different
34: carrier doping levels from the strongly underdoped to the strongly
35: overdoped range with $T_c$=66, 88, 77, and 67 K respectively. We
36: focus on the redistribution of the low frequency optical spectral
37: weight (SW) in the superconducting and normal states. The
38: temperature dependence of the low-frequency spectral weight in the
39: normal state is significantly stronger in the overdoped regime. In
40: agreement with other studies, the superconducting order is marked by
41: an increase of the low frequency SW for low doping, while the SW
42: decreases for the highly overdoped sample. The effect crosses
43: through zero at a doping concentration $\delta$=0.19 which is
44: slightly to the right of the maximum of the superconducting dome.
45: This sign change is not reproduced by the BCS model calculations,
46: assuming the electron-momentum dispersion known from published ARPES
47: data. Recent Cluster Dynamical Mean Field Theory (CDMFT)
48: calculations based on the Hubbard and t-J models, agree in several
49: relevant respects with the experimental data.
50:
51: \end{abstract}
52:
53: \maketitle
54:
55: \section{INTRODUCTION}\label{intro}
56:
57: One of the most puzzling phenomena in the field of high temperature
58: superconductivity is the doping dependence of the electronic
59: structure of the cuprates. Several experiments report a conventional
60: Fermi Liquid behavior on the overdoped side of the superconducting
61: 'dome' \cite{Proust,arpes1,specheat,deutscher}, while the enigmatic
62: 'pseudogap phase' is found in underdoped
63: samples\cite{NMR,ARPES,optics}. In the underdoped and optimally
64: doped regions of the phase diagram it has been shown for bi-layer
65: Bi2212 \cite{hajo,santander,comment} and tri-layer Bi2223
66: (Bi$_{2}$Sr$_{2}$Ca$_{2}$Cu$_{3}$O$_{10}$) \cite{io} that the
67: superconductivity induced low frequency Spectral Weight (SW)
68: increases when the system becomes superconducting. This observation
69: points toward a non BCS-like pairing mechanism, since in a BCS
70: scenario the superconductivity induced SW transfer would have the
71: opposite sign. On the other hand, in Ref. \onlinecite{deutscher} a
72: fingerprint of more conventional behavior has been reported using
73: optical techniques for a strongly overdoped thin film of Bi2212: the
74: SW redistribution at high doping has the opposite sign with respect
75: to the observation for under and optimal doping.
76:
77: It is possible to relate the SW transfer and the electronic kinetic
78: energy using the expression for the intraband spectral weight $W$
79: via the energy momentum dispersion n$_k$ of the conduction electrons
80: \cite{kinsw}
81: %
82: \begin{equation}\label{equation1}
83: W(\Omega_c,T) \equiv {\int}^{\Omega_c}_0 \sigma_1(\omega,T)d\omega =
84: \frac{\pi e^2 a^2}{2\hbar^2V}<-\hat{K}>,
85: \end{equation}
86: %
87: \noindent where $\sigma_1(\omega,T)$ is the real part of the optical
88: conductivity, $\Omega_c$ is a cutoff frequency, $a$ is the in-plane
89: lattice constant, $V$ is the volume of the unit cell and
90: $\hat{K}\equiv -a^{-2}\sum_k \hat{n}_k
91: \partial^2\epsilon_k/\partial k^2$. The operator $\hat{K}$ becomes the exact kinetic
92: energy $\sum_k \hat{n}_k \epsilon_k$ of the free carriers within the
93: nearest neighbor tight-binding approximation. It has been shown, in
94: Refs. \onlinecite{dirk, frank}, that even after accounting for the
95: next nearest neighbor hopping parameter the exact kinetic energy and
96: $<-\hat{K}>$ approximately coincide and follow the same trends as a
97: function of temperature. According to Eqn. (1), the lowering of
98: $W(\Omega_c)$ implies an increase of the electronic kinetic energy
99: and vice-versa. In this simple scenario a decrease of the low
100: frequency SW, when the system becomes superconducting, would imply a
101: superconductivity induced increase of the electronic kinetic energy,
102: as it is the case for BCS superconductors.
103:
104: In the presence of strong electronic correlations this basic picture
105: has to be extended to take into account that at different energy
106: scales materials are described by different model Hamiltonians, and
107: different operators to describe the electric current at a given
108: energy scale \cite{eskes, rozenberg}. In the context of the Hubbard
109: model, Wrobel {\em et al.} pointed out\cite{wrobel} that if the
110: cutoff frequency $\Omega_c$ is set between the value of the exchange
111: interaction $J \simeq 0.1$ eV and the hopping parameter $t \simeq
112: 0.4$ eV then $W(\Omega_c)$ is representative of the kinetic energy
113: of the holes within the t-J model in the spin polaron approximation
114: and describes the excitations below the on-site Coulomb integral $U
115: \simeq 2$ eV not involving double occupancy, while $W(\Omega_c > U)$
116: represents all intraband excitations and therefore describes the
117: kinetic energy of the full Hubbard Hamiltonian. A numerical
118: investigation of the Hubbard model within the dynamical cluster
119: approximation\cite{meier} has shown the lowering of the full kinetic
120: energy below $T_c$, for different doping levels, including the
121: strongly overdoped regime. Experimentally, this result should be
122: compared with the integrated spectral weight where the cutoff
123: frequency is set well above $U = 2$ eV in order to catch all the
124: transitions into the Hubbard bands. However, in the cuprates this
125: region also contains interband transitions, which would make the
126: comparison rather ambiguous.
127:
128: Using Cluster Dynamical Mean Field Theory (CDMFT) on a 2$\times$2
129: cluster Haule and Kotliar \cite{kotliar} recently found that, while
130: the total kinetic energy decreases below $T_c$ at all doping levels,
131: the kinetic energy of the holes exhibits the opposite behavior on
132: the two sides of the superconducting dome: In the underdoped and
133: optimally doped samples the kinetic energy of the holes, which is
134: the kinetic energy of the t-J model, increases below $T_c$. In
135: contrast, on the overdoped side the same quantity decreases when the
136: superconducting order is switched on in the calculation. This is in
137: agreement with the observations of Ref.\onlinecite{deutscher} as
138: well as the experimental data in the present paper. The good
139: agreement between experiment and theory in this respect is
140: encouraging, and it suggests that the t-J model captures the
141: essential ingredients, needed to describe the low energy excitations
142: in the cuprates, as well as the phenomenon of superconductivity
143: itself.
144:
145: %\subsection{}\label{sec3}
146: %
147: The Hubbard model and the t-J model are based on the assumption that
148: strong electron electron correlations rule the physics of these
149: materials. Based on these models an increase of the low frequency SW
150: in the superconducting state was found in the limit of low doping
151: \cite{wrobel} in agreement with the experimental results
152: \cite{hajo,io}. The optical conductivity of the t-J model in a
153: region of intermediate temperatures and doping near the top of the
154: superconducting dome has been recently studied using CDMFT
155: \cite{kotliar}. The CDMFT solution of the t-J model at different
156: doping levels suggests a possible explanation for the fact that the
157: optical spectral weight shows opposite temperature dependence for
158: the underdoped and the overdoped samples. It is useful to think of
159: the kinetic energy operator of the hubbard model, at large U as
160: composed of two physically distinct contributions representing the
161: superexchange energy of the spins and the kinetic energy of the
162: holes. The superexchange energy of the spins is the result of the
163: virtual transitions across the charge transfer gap, thus, the
164: optical spectral weight integrated up to an energy below these
165: excitations is representative only of the kinetic energy of the
166: holes. The latter contribution to the total kinetic energy was found
167: to decrease in the underdoped regime while it increases above
168: optimal doping, as observed experimentally. This kinetic energy
169: lowering is however rather small compared to the lowering of the
170: superexchange energy.
171: %This larger energy scale is most likely responsible for the high
172: %T$_c$ of these materials \cite{philips}.
173: Upon overdoping the kinetic energy of the holes increases in the
174: superconducting state, while the larger decrease of the
175: super-exchange energy makes superconductivity favorable with a still
176: high value of $T_c$. In the CDMFT study of the t-J model, a stronger
177: temperature dependence of $W(T)$ is found in the overdoped side.
178: This reflects the increase in Fermi Liquid coherence with reducing
179: temperature.
180:
181:
182: In the present paper we extend earlier experimental studies of the
183: temperature dependent optical spectral weight of Bi2212 by the same
184: group\cite{hajo,comment} to the overdoped side of the phase diagram,
185: {\em i.e.}with superconducting phase transition temperatures of 77 K
186: and 67 K. We report a strong change in magnitude of the temperature
187: dependence in the normal state for the sample with the highest hole
188: doping, and we show that the kink in the temperature dependence at
189: $T_c$ changes sign at a doping level of about 19 percent, in
190: qualitative agreement with the report by Deutscher et al.
191: \cite{deutscher}.
192:
193: \section{EXPERIMENT AND RESULTS}\label{sec1}
194:
195: In this paper we concentrate on the properties of single crystals of
196: Bi2212 at 4 different doping levels, characterized by their
197: superconducting transition temperatures. The preparation and
198: characterization of the underdoped sample (UD66K), an optimally
199: doped crystal (Opt88) and an overdoped sample (OD77) with $T_c$'s of
200: 66, 88 K and 77 K respectively, have been given in Ref.
201: \onlinecite{hajo}. The crystal with the highest doping level (OD67)
202: has a $T_c$ of 67 K. This sample has been prepared with the
203: self-flux method. The oxygen stoichiometry of the single crystal has
204: been obtained in a PARR autoclave by annealing for 4 days in Oxygen
205: at 140 atmospheres and slowly cooling from 400 °C to 100 °C. The
206: infrared optical spectra and the spectral weight analysis of samples
207: UD66 and OpD88 have been published in Refs.
208: \onlinecite{hajo,comment}. The phase of $\sigma(\omega)$ of sample
209: OD77 has been presented as a function of frequency in a previous
210: publication\cite{MarelNature03}. In the present manuscript we
211: present the optical conductivity of samples OD77 and OD67 for a
212: dense sampling of temperatures, and we use this information to
213: calculate $W(\Omega_c,T)$. The samples are large ($4\times4\times
214: 0.2$ mm$^{3}$) single crystals. The crystals were cleaved within
215: minutes before being inserted into the optical cryostat. We measured
216: the real and imaginary part of the dielectric function with
217: spectroscopic ellipsometry in the frequency range between 6000 and
218: 36000 cm$^{-1}$ (0.75 - 4.5 eV). Since the ellipsometric measurement
219: is done at a finite angle of incidence (in our case 74$^\circ$), the
220: measured pseudo-dielectric function corresponds to a combination of
221: the ab-plane and c-axis components of the dielectric tensor. From
222: the experimental pseudo-dielectric function and the published c-axis
223: dielectric function of Bi-2212 \cite{tajima} we calculated the
224: ab-plane dielectric function. In accordance with earlier results on
225: the cuprates \cite{bozovic,io} and with the analysis of
226: Aspnes\cite{aspnes}, the resulting ab-plane dielectric function
227: turns out to be very weakly sensitive to the c-axis response and its
228: temperature dependence. In the range from 100 to 7000 cm$^{-1}$
229: (12.5 - 870 meV) we measured the normal incidence reflectivity,
230: using gold evaporated {\em in situ} on the crystal surface as a
231: reference.
232: \begin{figure}[ht]
233: \centerline{\includegraphics[width=8cm,clip=true]{Rdop.eps}}
234: \caption{Reflectivity spectra of Bi2212 at selected temperatures for different doping levels, described in the text.}
235: \label{fig1}
236: \end{figure}
237:
238: The infrared reflectivity is displayed for all the studied doping
239: levels in Fig. \ref{fig1}. The absolute reflectivity increases with
240: increasing doping, as expected since the system becomes more
241: metallic. Interestingly, the curvature of the spectrum also changes
242: from under to overdoping; this is reflected in the frequency
243: dependent scattering rate as has been pointed out recently by Wang
244: {\em et al.} \cite{TimuskNature}. In order to obtain the optical
245: conductivity in the infrared region we used a variational routine
246: that simultaneously fits the reflectivity and ellipsometric data
247: yielding a Kramers-Kronig (KK) consistent dielectric function which
248: reproduces all the fine features of the measured spectra. The
249: details of this approach are described elsewhere
250: \cite{Kuzmenko04,io}. All data were acquired in a mode of continuous
251: temperature scans between 20 K and 300 K with a resolution of 1 K.
252: Very stable measuring conditions are needed to observe changes in
253: the optical constants smaller than 1\%. We use home-made cryostats
254: of a special design, providing a temperature independent and
255: reproducible optical alignment of the samples. To avoid spurious
256: temperature dependencies due to adsorbed gases at the sample
257: surface, we use a Ultra High Vacuum UHV cryostat for the
258: ellipsometry in the visible range, operating at a pressure in the
259: $10^{-10}$ mbar range, and a high vacuum cryostat for the normal
260: incidence reflectivity measurements in the infrared, operating in
261: the $10^{-7}$ mbar range.
262:
263: In Fig. \ref{fig2} we show the optical conductivity of the two
264: overdoped samples of Bi2212 with $T_c$ = 77 K and $T_c$ = 67 K at
265: selected temperatures. Below 700 cm$^{-1}$ one can clearly see the
266: depletion of the optical conductivity in the region of the gap at
267: low temperatures (shown in the inset). The much smaller absolute
268: conductivity changes at higher energies, which are not discernible
269: at this scale, will be considered in detail below.
270:
271: \begin{figure}[ht]
272: \centerline{\includegraphics[width=8cm,clip=true]{sig77.eps}
273: \includegraphics[width=8cm,clip=true]{sig67.eps}}
274: \caption{In-plane optical conductivity of slightly overdoped ($T_c$= 77 K, left panel) and
275: strongly overdoped ($T_c$= 67 K, right panel) samples of Bi2212 at selected temperatures.
276: The insets show the low energy parts of the spectra.}
277: \label{fig2}
278: \end{figure}
279:
280: One can see the effect of superconductivity on the optical constants
281: in the temperature dependent traces, displayed in Fig. \ref{fig4},
282: at selected energies, for the two overdoped samples. In comparison
283: to the underdoped and optimally doped samples \cite{hajo,io} where
284: reflectivity is found to have a further increase in the
285: superconducting state at energies between 0.25 and 0.7 eV, in the
286: overdoped samples reflectivity decreases below $T_c$ or remain more
287: or less constant. In the strongly overdoped sample one can clearly
288: see, for example at 1.24 eV, that at low temperature $\epsilon_1$
289: increases cooling down, opposite to the observation on the optimally
290: and underdoped samples. These details of the temperature dependence
291: of the optical constants influence the integrated SW trend as we
292: will discuss later in the text.
293:
294: \begin{figure}[ht]
295: \centerline{\includegraphics[width=9cm,clip=true]{fig77re.eps}
296: \includegraphics[width=9cm,clip=true]{fig4.eps}}
297: \caption{Leftmost (rightmost) two columns: reflectivity and
298: dielectric function of sample OD77 (OD67)
299: as a function of temperature for selected photon energies.
300: The corresponding photon energies are indicated in the panels.
301: The real (imaginary) parts of $\epsilon(\omega)$ are indicated
302: as closed (open) symbols.}
303: \label{fig4}
304: \end{figure}
305:
306: \section{DISCUSSION}\label{sec2}
307: \subsection{Spectral weight analysis of the experimental data}
308:
309: As it is discussed in our previous publications
310: \cite{Kuzmenko04,io,comment}, using the knowledge of both $\sigma_1$
311: and $\epsilon_1$ we can calculate the low frequency SW without the
312: need of the knowledge of $\sigma_1$ below the lowest measured
313: frequency. When the frequency cut off of the integral is chosen to
314: be lower than the charge transfer energy (around 1.5 eV), the SW is
315: representative of the free carrier kinetic energy in the t-J model
316: \cite{wrobel,io,kotliar}. In this paper we set the frequency cut-off
317: at 1.25 eV and compare the results with the predictions of BCS
318: theory and CDMFT calculations based on the t-J model. In Fig.
319: \ref{fig7} we show a comparison between $W(T)$ for different samples
320: with different doping levels. One can clearly see that the onset of
321: superconductivity induces a positive change of the SW(0-1.25 eV) in
322: the underdoped sample and in the optimally doped one\cite{hajo}; in
323: the 77 K sample no superconductivity induced effect is detectable
324: for this frequency cut off and in the strongly overdoped sample we
325: observe a decrease of the low frequency spectral weight. In the
326: righthand panel of Fig. \ref{fig7} we also display the derivative of
327: the integrated SW as a function of temperature. The effect of the
328: superconducting transition is visible in the underdoped sample and
329: in the optimally doped sample as a peak in the derivative plot; no
330: effect is detectable in the overdoped 77 K sample, while in the
331: strongly overdoped sample a change in the derivative of the opposite
332: sign is observed.
333:
334: \begin{figure}[ht]
335: \centerline{\includegraphics[width=9cm,clip=true]{fig7_fin.eps}
336: \includegraphics[width=9cm,clip=true]{fig9_fin.eps}}
337: \caption{Left panel: spectral weight $W(\Omega_c,T)$ for $\Omega_c$ = 1.24 eV, as a function of temperature for different doping levels.
338: Right panel: the derivative ($\frac{-\partial W(\Omega_c,T)}{\partial T}$) as a function of temperature for different doping levels.
339: For the derivative curves the data have been averaged in 5 K intervals in order to reduce the noise.}
340: \label{fig7}
341: \end{figure}
342:
343: The frequency $\omega_p^*$ for which $\epsilon_1(\omega_p^*)$ = 0
344: corresponds to the eigenfrequency of the longitudinal oscillations
345: of the free electrons for $k\rightarrow 0$. $\omega_p^*$ can be
346: read off directly from the ellipsometric spectra, without any
347: data-processing. The temperature dependence of $\omega_p^*$ is
348: displayed in Fig. \ref{fig8}. The screened plasma-frequency has a
349: red shift due to the bound-charge polarizability, and the interband
350: transitions. Therefore its temperature dependence can be caused by a
351: change of the free carrier spectral weight, the dissipation, the
352: bound-charge screening, or a combination of those. This quantity and
353: its derivative as a function of temperature can clarify whether a
354: real superconductivity-induced change of the plasma frequency is
355: already visible in the raw experimental data. In view of the fact
356: that the value of $\omega_p^*$ is determined by several factors, and
357: not only the low frequency SW, it is clear that the SW still has to
358: be determined from the integral of Eq. \ref{equation1}. It is
359: perhaps interesting and encouraging to note, that in all cases which
360: we have studied up to date, the temperature dependences of $W(T)$
361: and $\omega_p^{*}(T)^2$ turned out to be very similar.
362:
363: \begin{figure}[ht]
364: \centerline{\includegraphics[width=9cm,clip=true]{plasma1.eps}
365: \includegraphics[width=9cm,clip=true]{dplasma.eps}}
366: \caption{Left panel: Screened plasma frequency as a function of temperature for different doping levels.
367: Right panel:Derivative as a function of temperature, ($-\partial \omega_p/\partial T$), of the Screened plasma frequency for different doping levels.}
368: \label{fig8}
369: \end{figure}
370: One can clearly see in the underdoped and in the optimally doped
371: sample that superconductivity induces a blue shift of the screened
372: plasma frequency. A corresponding peak is observed at $T_c$ in the
373: derivative plots. In the 77 K sample no effect is visible at $T_c$
374: while the 67 K sample shows a red shift of the screened plasma
375: frequency. The behavior of the screened plasma frequency also seems
376: to exclude the possibility that a narrowing with temperature of the
377: interband transitions around 1.5 eV is responsible for the observed
378: changes in the optical constants. If this would be the case then one
379: would expect the screened plasma frequency to exhibit a
380: superconductivity-induced shift in the same direction for all the
381: samples.
382:
383: \subsection{Predictions for the spectral weight using the BCS model}
384: In order to put the data into a theoretical perspective, we have
385: calculated $W(T)$ in the BCS model, using a tight-binding
386: parametrization of the energy-momentum dispersion of the normal
387: state. The parameters of the parametrization are taken from ARPES
388: data \cite{golden}. The details of this calculation are discussed in
389: the Appendix. Because in this parametrization both $t'$ and $t''$
390: are taken to be different from zero, the spectral weight is not
391: strictly proportional to the kinetic energy. Nonetheless for the
392: range of doping considered here, $W$ follows the same trend as the
393: actual kinetic energy, as has been pointed out previously by some of
394: us\cite{marel_hvar2002}. Results for the
395: $t-t^{\prime}-t^{\prime\prime}$ model are shown in Fig. \ref{fig9}.
396: We do wish to make a cautionary remark here, that a sign change as a
397: function of doping is not excluded {\em a priori} by the BCS model.
398: However, in the present case this possibility appears to be excluded
399: in view of the state of the art ARPES results for the
400: energy-momentum dispersion of the occupied electron bands. One can
401: see that for all considered doping levels, W decreases below $T_c$,
402: thus BCS calculations fail reproducing the temperature dependence in
403: the underdoped and optimally doped samples.
404:
405: \begin{figure}[ht]
406: \centerline{\includegraphics[width=8cm,clip=true]{fig8_n.eps}}
407: \caption{Low frequency SW as a function of temperature calculated for the same doping levels experimentally measured.}
408: \label{fig9}
409: \end{figure}
410:
411:
412: \subsection{Superconductivity induced transfer of spectral weight: experiment and cluster DMFT calculations}
413: %
414: In order to highlight the effect of varying the doping
415: concentration, we have extrapolated the temperature dependence in
416: the normal state of $W(\Omega_c,T)$ of each sample to zero
417: temperature, and measured it's departure from the same quantity in
418: the superconducting state, also extrapolated to T=0: $\Delta
419: SW_{sc}\equiv W(T=0)-W_n^{ext.}(T=0)$ In Fig. \ref{ddq1} the
420: experimentally derived quantities are displayed together with the
421: recent CDMFT calculations of the t-J\cite{kotliar} model and those
422: based on the BCS model explained in the previous subsection. While
423: the BCS-model provides the correct sign only for the strongly
424: overdoped case, the CDMFT calculations based on the t-J model are in
425: qualitative agreement with our data and the data in Ref
426: \onlinecite{deutscher}, insofar both the experimental result and the
427: CDMFT calculation give $\Delta SW_{sc}>0$ on the underdoped side of
428: the phase diagram, and both have a change of sign as a function of
429: doping when the doping level is increased toward the overdoped side.
430: The data and the theory differ in the exact doping level where the
431: sign change occurs. This discrepancy may result from the fact that
432: for the CDMFT calculations the values $t'=t''=0'$ were adopted. This
433: choice makes the shape of the Fermi surface noticeably different
434: from the experimentally known one, hence the corresponding
435: fine-tuning of the model parameters may improve the agreement with
436: the experimental data. This may also remedy the difference between
437: the calculated doping dependence of $T_c$ and the experimental one
438: (see righthand panel of Fig. \ref{ddqns}). We also show, in Fig.
439: \ref{ddq2}, the doping dependence of the plasma frequency and
440: effective mass compared to the CDMFT results. One can see that a
441: reasonable agreement is achieved for both quantities.
442: %
443: \begin{figure}[ht]
444: \centerline{\includegraphics[width=9cm,clip=true]{Dopdepqu_sc.eps}}
445: \caption{Doping dependence of the superconductivity induced SW changes: experiment vs. theory. Two theoretical calculations are presented:
446: d-wave BCS model and CDMFT calculations in the framework of the t-J model.}
447: \label{ddq1}
448: \end{figure}
449: %
450: \begin{figure}[ht]
451: \centerline{\includegraphics[width=9cm,clip=true]{Dopdepqu_ns.eps}
452: \includegraphics[width=9cm,clip=true]{dome.eps}}
453: \caption{Left panel: Comparison between the experimental and the theoretical $W(T)$ in the normal state for different doping levels.
454: Right panel: comparison between the 'dome' as derived from theory and the experimental one.}
455: \label{ddqns}
456: \end{figure}
457: %
458:
459:
460: \subsection{Normal state trend of the spectral weight}
461: %
462: The persistence of the $T^2$ temperature dependence up to energies
463: much larger than what usually happens in normal metals has been
464: explained in the context of the Hubbard model \cite{calvani},
465: showing that electron-electron correlations are most likely
466: responsible for this effect. Indeed, experimentally we observe a
467: strong temperature dependence of the optical constants at energies
468: as high as 2 eV. In most of the temperature range, particularly for
469: the samples with a lower doping level, these temperature
470: dependencies are quadratic. Correspondingly, $W(T)$ also manifests a
471: quadratic temperature dependence. For sample OD67 the departure from
472: the quadratic behavior is substantial; the overall normal state
473: temperature dependence at this doping is also much stronger than in
474: the other samples. We compare, in Fig. \ref{ddqns}, the temperature
475: dependence of the SW with the predictions of the Hubbard model.
476:
477: In Fig. \ref{ddqns} the experimental $W(T)$ is compared to the CDMFT
478: calculations for the same doping concentration. Since the $T_c$
479: obtained by CDMFT differs from the experimental one, (see Fig.
480: \ref{ddqns}) it might be more realistic to compare theory and
481: experiment for doping concentrations corresponding to the same
482: relative $T_c$'s. Therefore we also include in the comparison the
483: CDMFT calculation at higher doping level, at which $T_c/T_{c,max}$
484: corresponds to the experimental one (see the right panel of Fig.
485: \ref{ddqns}). We see that the experimental and calculated values of
486: $W(T)$ are in quantitative agreement for the temperature range where
487: they overlap. It is interesting in this connection, that the
488: curvature in the opposite direction, clearly present in all CDMFT
489: calculations, may actually be present in the experimental data, at
490: least for the highly doped samples. These observations clearly call
491: for an extension of the experimental studies to higher temperature
492: to verify whether a cross-over of the type of temperature dependence
493: of the spectral weight really exists, and to find out the doping
494: dependence of the cross-over temperature. The experimental data, as
495: mentioned before, show a rapid increase of the slope of the
496: temperature dependence above optimal doping. This behavior is
497: qualitatively reproduced by the CDMFT calculations.
498:
499: One can calculate the normal state kinetic energy of the charge
500: carriers and its temperature dependence starting from the
501: tight-binding dispersion relation neglecting the correlation
502: effects. In this context, one can find a stronger temperature
503: dependence of the normal state SW when the chemical potential
504: approaches the van Hove singularity. Extrapolating the experimental
505: bandstructure beyond x=0.22 we estimate that this would happen at a
506: doping level as high as 0.4 in Bi2212 \cite{golden}. This offers an
507: alternative scenario for the normal state temperature dependence,
508: although the role of the van Hove singularity has to be explored in
509: further detail. We also point out that as a result of crossing this
510: singularity one can get a SW increase in the superconducting state
511: within the BCS model. In the CDMFT calculations presented in Fig.
512: \ref{ddq1} the SW temperature dependence in the normal state is a
513: pure correlation effect, since in this calculation the van Hove
514: singularity is located at exactly half filling, far away from the
515: experimentally considered doping levels.
516:
517: \begin{figure}[ht]
518: \centerline{\includegraphics[width=10cm,clip=true]{Dopdepqu2.eps}}
519: \caption{Comparison between the calculated plasma frequency and effective mass and the experimental values.}
520: \label{ddq2}
521: \end{figure}
522:
523:
524:
525: \section{CONCLUSIONS}
526:
527: In conclusion, we have studied the doping dependence of the optical
528: spectral weight redistribution in single crystals of Bi2212, ranging
529: from the underdoped regime, $T_c$ = 66 K to the overdoped regime,
530: $T_c$ = 67 K. The low frequency SW increases when the system becomes
531: superconducting in the underdoped region of the phase diagram, while
532: it shows no changes in the overdoped sample $T_c$ = 77 K and
533: decreases in the $T_c$ = 67 K sample. We compared these results with
534: BCS calculations and CDMFT calculations based on the t-J model. We
535: show that the latter are in good qualitative agreement with the data
536: both in the normal and superconducting state, suggesting that the
537: redistribution of the optical spectral weight in cuprates
538: superconductors is ruled by electron-electron correlations effects.
539:
540: \section*{ACKNOWLEDGMENTS}
541:
542: We are grateful to T. Timusk, N. Bontemps, A.F. Santander-Syro, J.
543: Orenstein, and C. Bernhard for stimulating discussions. This work
544: was supported by the Swiss National Science Foundation through the
545: National Center of Competence in Research "Materials with Novel
546: Electronic Properties-MaNEP".
547:
548: \section{APPENDIX}
549:
550: The pair formation in a superconductor can be described by a
551: spatial correlation function $g(r)$ which has a zero average in
552: the normal state and a finite average in the superconducting
553: state. Without entering into the details of the mechanism itself
554: responsible for the attractive interaction between electrons, one
555: can assume that an attractive potential V(r) favors a state with
556: enhanced correlations in the superconducting state. In the
557: superconducting state the interaction energy differs from the
558: normal state by:
559: \begin{equation}
560: <H^i>_s - <H^i>_n=\int dr^3g(r)V(r)
561: \end{equation}
562: With some manipulations one can relate the correlation function to
563: the gap-function and the single particle energy. The result is a BCS
564: equation for the order parameter, with a potential which can be
565: chosen to favor pairing with d-wave symmetry. The simplest approach
566: is to use a simple separable potential which leads to an order
567: parameter of the form, $\Delta_k = \Delta_0(T) (\cos{k_x} -
568: \cos{k_y})/2$. The temperature dependence of $\Delta_0(T)$ can then
569: be solved as in regular BCS theory. We have done this for a variety
570: of parameters \cite{frank}, and find that $\Delta_0(T)/\Delta_0(0) =
571: \sqrt{1 - (t^4+t^3)/2}$ gives a very accurate result (for either
572: s-wave or d-wave symmetry), where $t \equiv T/T_c$. Then, for
573: simplicity, we adopt the weak coupling result that $\Delta_0(0) =
574: 2.1 k_B T_c$. Finally, even in the normal state, the chemical
575: potential is in principle a function of temperature (to maintain the
576: same number density); this is computed by solving the number
577: equation, $n = ({2 \over N})\sum_k n_k$, where
578: $$
579: n_k = 1/2 - (\epsilon_k - \mu) {[1 - 2f(E_k)] \over 2E_k}
580: $$
581: where $E_k \equiv \sqrt{(\epsilon_k - \mu)^2 + \Delta_k^2}$ at each
582: temperature for $\mu$ for a fixed doping. Once these parameters are
583: determined, one can calculate the spectral weight sum, $W$, for a
584: given band structure. We use:
585: $$
586: \epsilon_k = -2t*(cos(k_x) + cos(k_y)) + 4t^\prime*cos(k_x)cos(k_y)
587: - 2t^{"}*(cos(2k_x) + cos(2k_y)).
588: $$
589: In this expression $\delta$ is the hole doping, and $\Delta_0$ is
590: the gap value calculated as $2.1K_BT_c$, t = 0.4 eV, t$^\prime$ =
591: 0.09 eV and t$^{\prime\prime}$ = 0.045 eV. The dispersion is taken
592: from ARPES measurements \cite{golden}; for simplicity we have left
593: out the bi-layer splitting and the constant. The spectral weight sum
594: is given by
595: $$
596: W = \sum_k {\partial_k^2 \epsilon_k \over \partial k_x^2} n_k.
597: $$
598: Results are plotted in Fig. 6 for the doping levels of the samples
599: used in the experiments. These calculations clearly show that BCS
600: theory predicts a lowering of the spectral weight sum in the
601: superconducting phase; this is in disagreement with the experimental
602: results in the underdoped and optimally doped samples. Moreover,
603: there is no indication of a change of sign of the
604: superconductivity-induced SW changes in this doping interval within
605: the BCS formalism. Note, however, that preliminary calculations
606: indicate that the van Hove singularity can play a role at much
607: higher doping levels (not realized, experimentally), and that in
608: theory a sign change in the anomaly can occur even within BCS
609: theory.
610:
611: \begin{thebibliography}{100}
612: %
613: \bibitem{Proust} C. Proust, E. Boaknin, R.W. Hill, L. Taillefer, A.P. Mackenzie. Phys. Rev. Lett. {\bf 89}, 147003 (2002).
614: \bibitem{arpes1} Z.M. Yusof, B.O. Wells, T. Valla, A.V. Fedorov, P.D. Johnson, Q. Li, C. Kendziora, S. Jian, D.G. Hinks. Phys. Rev. Lett. {\bf 88}, 167006 (2002).
615: \bibitem{specheat} A. Junod, A. Erb, C. Renner. Physica C {\bf 317}, 333 (1999).
616: \bibitem{deutscher} G. Deutscher, A. F. Santander-Syro, and N.
617: Bontemps, Phys. Rev. B, {\bf 72} 092504 (2005).
618: \bibitem{NMR} R.E. Walstedt, Jr and W.W. Warren, Science {\bf 248}, 1082 (1990).
619: \bibitem{ARPES} M.R. Norman, H. Ding, M. Rendeira, J.C. Campuzano, T. Yokoya, T. Takeuchi, T. Takahashi, T. Mochiku, K. Kadowaki, P. Guptasarma, D.G. Hinks. Nature {\bf 392}, 157 (1998).
620: \bibitem{optics} A.V. Puchkov, D.N. Basov, and T. Timusk, J. Phys.: Condens. Matter {\bf
621: 8}, 10049 (1996).
622: \bibitem{hajo} H. J. A. Molegraaf, C. Presura, D. van der Marel, P.H. Kes, M. Li. Science {\bf 295}, 2239
623: (2002).
624: \bibitem{santander} A. F. Santander-Syro, R.P.S. Lobo, N. Bontemps, Z. Konstatinovic, Z.Z. Li, H. Raffy. Europhys. Lett. {\bf 62}, 568 (2003).
625: \bibitem{comment} A. B. Kuzmenko, H. J. A. Molegraaf, F. Carbone and D. van der Marel
626: Phys. Rev. B. {\bf 72} 144503 (2005).
627: \bibitem{io} F. Carbone, A.B. Kuzmenko, H.J.A. Molegraaf, E. van Heumen, E. Giannini, D. van der Marel. submitted to Phys. Rev.
628: B. cond-mat/0603737
629: \bibitem{kinsw} P. F. Maldague, Phys Rev B {\bf 16}, 2437 (1977).
630: \bibitem{dirk} D. van der Marel, H.J.A. Molegraaf, C. Presura, I. Santoso. \emph{Concepts in electron correlations}, edited by A. Hewson and V. Zlatic, Kluwer (2003).
631: \bibitem{frank} F. Marsiglio. Phys. Rev. B {\bf 73}, 064507 (2006).
632: \bibitem{eskes} H. Eskes, A. M. Oles, M.B.J. Meinders, W. Stephan. Phys. Rev. B {\bf 50}, 17980 (1994).
633: \bibitem{rozenberg} M. J. Rozenberg, G. Kotliar, H. Kajueter. Phys. Rev. B {\bf 54}, 8452 (1996).
634: \bibitem{wrobel} P. Wrobel, R. Eder and P. Fulde, J. Phys. Cond. Matt., {\bf 15}, 6599 (2003).
635: \bibitem{meier} Th.A. Maier, M. Jarrell, A. Macridin, C. Slezak. Phys. Rev. Lett. {\bf 92}, 027005 (2004).
636: \bibitem{kotliar} K. Haule and G. Kotliar cond-mat/0601478.
637: \bibitem{MarelNature03} D. van der Marel, H.J.A. Molegraaf, J. Zaanen, Z. Nussinov, F. Carbone, A. Damascelli, H. Eisaki, M. Greven, P.H. Kes, M. Li. Nature {\bf 425} 271 (2003).
638: %\bibitem{berger} .... {\em et al.} {\bf 425} (2003).
639: \bibitem{tajima} S. Tajima, G.D. Gu, S. Miyamoto, A. Odagawa, N. Koshizuka. Phys. Rev. B {\bf 48} 16164 (1993).
640: \bibitem{bozovic} I. Bozovic, Phys. Rev. B {\bf 42}, 1969 (1990).
641: \bibitem{aspnes} D.E. Aspnes, J. Opt. Soc. Am. {\bf 70}, 1275 (1980).
642: \bibitem{TimuskNature} J. Wang, T. Timusk, G.D Dung. Nature {\bf
643: 427} 714 (2004)
644: \bibitem{Kuzmenko04} A. B. Kuzmenko, Rev. Sci. Instr. {\bf 76} 083108 (2005).
645: \bibitem{golden} J. Fink, S. Borisenko, A. Kordyuk, A. Koitzsch, J.
646: Geck, V. Zabalotnyy, M. Knupfer, B. Buechner, H. Berger.
647: cond-mat/0512307 (2006).
648: %\bibitem{timusk} J. Hwang {\em et al.} Phys. Rev. B {\bf 69} 094520 (2004).
649: %\bibitem{Kuzmenko03} A. B. Kuzmenko {\em et al.} Phys Rev Lett. {\bf 91} 037004 (2003).
650: \bibitem{marel_hvar2002} D. van der Marel, H. J. A. Molegraaf, C. Presura and I. Santoso, in "Concepts in electron correlation", Edited by A. Hewson and V.
651: Zlatic, Kluwer (2003), p 7-16.; cond-mat/0302169.
652: \bibitem{calvani} A.Toschi, M. Capone, M. Ortolani, P. Calvani, S. Lupi, C.
653: Castellani. Phys. Rev. Lett. {\bf 95} 097002 (2005)
654: %\bibitem{campuzano} M. R. Norman, M. renderia, H. Ding, J.C. Campuzano. Phys. Rev. B {\bf 52} 615 (1985).
655: %\bibitem{ioprb} F. Carbone {\em et al.} in preparation.
656: %\bibitem{ando} Y. Ando {\em et al.} Phys. Rev. Lett {\bf 93} 267001 (2004).
657: %\bibitem{BenfattoCM06}
658: %L. Benfatto, J.P. Carbotte, and F. Marsiglio, cond-mat/0603661.
659:
660: \end{thebibliography}
661:
662: \end{document}
663: