VOLUME 78, NUMBER 25 PHYSICAL REVIEW LETTERS 23 JUNE 1997 Pseudogap and Superconducting Gap in the Electronic Raman Spectra of Underdoped Cuprates R. Nemetschek, M. Opel, C. Hoffmann, P. F. Müller, and R. Hackl Walther Meissner Institut, Bayerische Akademie der Wissenschaften, D-85748 Garching, Germany H. Berger and L. Forró Ecole Polytechnique Fédérale de Lausanne, Ecublens, CH-1015 Lausanne, Switzerland A. Erb and E. Walker DPMC, Université de Genève, CH-1211 Genève, Switzerland (Received 6 February 1997) Raman spectra of YBa 2 Cu 3 O 72x and Bi 2 Sr 2 sCa 0.62 Y 0.38 dCu 2 O 81d with T c > 0.65T max c in the underdoped regime of the phase diagram are studied as a function of temperature and polarization. At B 2g sxyd symmetry a reduction of spectral weight by 10% for frequencies less than 700 cm 21 or 15k B T c is found below approximately 200 K. Below T c , a superconducting gap opens up which closely resembles that observed at higher doping levels. It is compatible with d x 2 2y 2 pairing and its amplitude 2D 0 can be estimated to be 8k B T c . [S0031-9007(97)03387-5] PACS numbers: 74.25.Gz, 74.72. – h, 78.30.Er The doping-temperature phase diagram of all CuO 2 sys- tems shows the proximity of magnetically ordered, metal- lic, and superconducting phases, and suggests that these phases are all interrelated. The undoped parent compounds are antiferromagnetically (AF) ordered Mott-Hubbard in- sulators, and magnetic fluctuations are observed along with superconductivity at low doping levels. Therefore, the evolution of the metallic phase from the AF insulating phase attracted a great deal of attention. Recently, gaplike features in the electronic spectrum of underdoped cuprates have been inferred from or directly observed in measure- ments of the specific heat [1], the dc [2], and infrared (ir) [3] conductivity, and in angle-resolved photoemission (ARPES) experiments [4] at characteristic temperatures T p well above T c . Although several ideas have been proposed which can lead to a qualitative understanding of this phe- nomenon [5–12], at present the interpretation of, and the relationship between, these results is not settled. In part, this is related to not yet understood discrepancies in the experimental data obtained by different methods. In addi- tion, basically all experimental techniques used so far are not similarly sensitive to both the formation of the pseudo- gap at T p and the transition to the superconducting phase at T c which is expected to develop a full, though most likely, strongly k-dependent gap. In this Letter, we want to contribute new information ob- tained from inelastic light-scattering experiments as both types of gaps can clearly be identified in the spectra. So far, only the superconducting gap could be found by Raman scattering [13]. All observations pointed to a strong anisotropy in k space and a superconducting order parameter which is most compatible with d x 2 2y 2 pairing [13,14]. In the normal state the electronic response is also related to carrier properties [13]. Then, anomalies such as a pseudogap which are observed in the conductivity [3] should manifest themselves in the Raman spectra. In ad- dition to what is found in ir experiments, Raman scattering weighs out different parts of the Fermi surface (FS) for dif- ferent light polarizations, and therefore anisotropies may be studied. The purpose of this paper is to demonstrate the direct observation of the pseudogap and, at lower tem- peratures, that of the superconducting gap in the Raman spectra of underdoped cuprates, and to discuss the depen- dences on polarization, energy, and temperature. The experiments were performed in back-scattering ge- ometry using a standard Raman setup with excitation at 476 nm, a spot size of 50 3 150 mm 2 , and laser powers between 0.3 and 4 mW. The polarizations of the incom- ing and outgoing photons were always parallel to the CuO 2 planes. All symmetries refer to a tetragonal point group. We studied single crystals of Bi 2 Sr 2 sCa 0.62 Y 0.38 dCu 2 O 81d (Bi2212) and YBa 2 Cu 3 O 72x (Y123). Bi2212 was pre- pared in ZrO crucibles. The resistively determined T c (midpoint) is 57 K, the width is DT c ≠ 5 K. The Y123 crystal was grown in BaZrO 3 [15], annealed for 100 h at 784 ± C in 1 bar oxygen, and subsequently quenched. Ac- cording to the calibration by Lindemer et al., this ther- mal treatment results in an oxygen content of 6.5 [16]. The magnetically measured transition temperature is T c ≠ 53.5 K with DT c ≠ 3 K. It is important to note that in Y123 the actual T c relevant to the Raman experiment can be higher by several degrees due to the illumination with intense light [17]. Therefore a higher approximate value for T c is given in the figures. The underdoped materials exhibit an important differ- ence to samples with optimal or higher doping levels in that features related to properties of the carriers can be observed only at B 2g polarization. At the other two symmetries, A 1g and B 1g , the spectra show hardly any temperature depen- dence [18]. Therefore we will discuss here B 2g data only. 0031-9007y 97y 78(25) y4837(4)$10.00 © 1997 The American Physical Society 4837