Raman light scattering study and microstructural analysis of epitaxial films of the electron-doped superconductor La_{2-x}Ce_{x}CuO_{4}
M. Rahlenbeck, M. Wagenknecht, A. Tsukada, D. Koelle, R. Kleiner, B. Keimer, C. Ulrich
RRaman light scattering study and microstructural analysisof epitaxial films of the electron-doped superconductor La − x Ce x CuO M. Rahlenbeck , M. Wagenknecht , , A. Tsukada , D. Koelle , R. Kleiner , B. Keimer , and C. Ulrich , , Max-Planck-Institut f¨ur Festk¨orperforschung, Heisenbergstraße 1, D-70569 Stuttgart, Germany Physikalisches Institut - Experimentalphysik II, Universit¨at T¨ubingen,Auf der Morgenstelle 14, D-72076 T¨ubingen, Germany NTT Basic Research Laboratories, NTT Corporation,3-1 Morinosato-Wakamiya, Atsugi, Kanagawa 243-0198, Japan University of New South Wales, School of Physics, 2052 Sydney, New South Wales, Australia and The Bragg Institute, Australian Nuclear Science and Technology Organization, Lucas Heights, NSW 2234, Australia (Dated: October 28, 2018)We present a detailed temperature-dependent Raman light scattering study of optical phononsin molecular-beam-epitaxy-grown films of the electron-doped superconductor La − x Ce x CuO closeto optimal doping ( x ∼ . T c = 29 K and x ∼ . T c = 27 K). The main focus of this workis a detailed characterization and microstructural analysis of the films. Based on micro-Ramanspectroscopy in combination with x-ray diffraction, energy-dispersive x-ray analysis, and scanningelectron microscopy, some of the observed phonon modes can be attributed to micron-sized inclusionsof Cu O. In the slightly underdoped film ( x ∼ . O modes and others that can beassigned to the La − x Ce x CuO matrix show pronounced softening and narrowing upon cooling below T ∼ T c . Based on control measurements on commercial Cu O powders and on a comparison to priorRaman scattering studies of other high-temperature superconductors, we speculate that proximityeffects at La − x Ce x CuO /Cu O interfaces may be responsible for these anomalies. Experiments onthe slightly overdoped La − x Ce x CuO film ( x ∼ .
1) did not reveal comparable phonon anomalies.
PACS numbers:
Electron-doped high-temperature superconductors ex-hibit a substantially lower maximum transition temper-ature and a narrower doping range of superconductivitythan their hole-doped counterparts. The origin of thisasymmetry of the phase diagram is still incompletely un-derstood, in part due to the complex materials physicsof the electron-doped cuprates. It has been demon-strated, for instance, that the elaborate annealing pro-cedure required to stabilize superconductivity in bulksingle crystals of Nd − x Ce x CuO (NCCO), one of themost widely studied electron-doped cuprates, generates asmall amount of an epitaxially intergrown impurity phasethat profoundly affects the magnetic properties.[1] Sim-ilar observations have been reported for some epitaxialthin films of Pr − x Ce x CuO (PCCO) grown by pulsed-laser deposition.[2] The pervasiveness of such inclusionsand their influence on various physical properties is onlybeginning to be explored, but the wide variation of thetransport characteristics of thin films of nominally identi-cal composition [3] suggests that they may be quite com-mon.Since Raman scattering is both a powerful spectro-scopic probe of superconductivity in the cuprates andan excellent diagnostic tool for impurity phases, wehave carried out a high-resolution Raman scatteringstudy of molecular beam epitaxy (MBE)-grown films ofLa − x Ce x CuO (LCCO), the compound that exhibitsboth the highest critical temperature and the widestdoping range for superconductivity among all electron-doped cuprates.[4–6] LCCO crystalizes in the so-called T (cid:48) -structure, which does not include apical oxygen ions,in contrast to the T -structure found in hole-doped mem- bers of the Ln CuO family. The stability of the T (cid:48) -structure depends on the radius of the lanthanide( Ln ) ions.[4] For Ln = La, the T (cid:48) -structure is un-stable in bulk form, but can be stabilized by epitax-ial growth on SrTiO substrates.[4] Depending on thegrowth conditions, LCCO can also crystallize in the T -structure, but superconductivity is observed only in the T (cid:48) -structure.[7, 8] The superconducting state is stableover a wider doping range (0 . ≤ x ≤ .
22) than inother electron-doped compounds such as NCCO, and themaximum transition temperature ( T c ∼
30 K) is foundat a lower doping level ( x = 0 .
09, compared to 0.15 forNCCO).[6, 9]In the hole-doped cuprates, numerous Raman scat-tering experiments have elucidated the magnitude,anisotropy, and doping dependence of the superconduct-ing energy gap, ∆, either directly via electronic Ra-man scattering [10] or indirectly via phonon anoma-lies induced by the electron-phonon interaction.[11, 12]In electron-doped superconductors, electronic Ramanscattering [13, 14] has also yielded valuable informa-tion about the magnitude and anisotropy of ∆, butsuperconductivity-induced phonon anomalies have thusfar not been reported. Since electronic Raman scatteringis difficult in films, we have used the latter method inan attempt to gain insight into the energy gap of LCCO.We indeed observe phonon anomalies at temperaturesclose to the superconducting transition temperature T c ,but find that micron-sized Cu O inclusions present in allsamples complicate their interpretation.The Raman scattering experiments were performedon two 900 nm thick LCCO films that had been the a r X i v : . [ c ond - m a t . s up r- c on ] M a y basis of prior transport experiments.[15] The films hadbeen deposited epitaxially on [001]-oriented SrTiO sub-strates, yielding c-axis oriented films, in two independentfabrication runs using MBE from pure metal sources,as described elsewhere.[4] After growth, they were an-nealed in vacuum at 578 ◦ C and 559 ◦ C for 100 and 90minutes, respectively, in order to remove residual api-cal oxygen ions.[16] The Ce concentrations were adjustedto x ∼ .
08 (corresponding to electron concentrationsslightly less than optimum doping) and x ∼ . T c ∼
29 K and 27 K, respectively.The widths of the resistivity transitions at T c were be-low 1 K for both films. For comparison, we used oneplatelet-like Cu O single crystal with c -axis perpendic-ular to the surface, and two commercial Cu O powdersamples. Powder (I) had a purity of 99 .
9% with maxi-mal impurity content of 0 . . . . . . . + /Kr + mixed-gas laser for excitation. The laser beam was focused ona ∼ µ m spot on the sample surface with an inci-dent power of less than 10 mW, in order to avoid sampleheating. The scattered light was analyzed by a DILORXY triple grating spectrometer using a nitrogen-cooledcharge-coupled-device (CCD) camera. For higher spatialresolution we used a “micro” setup, where the sampleswere mounted on the cold finger of a horizontal helium-flow cryostat. The spectra were taken in backscatteringgeometry using the linearly polarized 532 . × (10 × ) microscopeobjective to a ∼ µ m ( ∼ µ m) diameter spot on thesample surface, with an incident laser power of less than1 mW. The scattered light was analyzed by a JobinYvonLabRam single grating spectrometer equipped with anotch filter and a Peltier-cooled CCD camera. For eachRaman spectrum an additional calibration spectrum of anearby argon or neon line was measured in order to accu-rately determine the frequency and linewidth of the dif-ferent phonons. For data analysis, all phonon peaks werefitted to Voigt profiles, which result from a convolutionof the Lorentzian phonon lineshape with the Gaussianshaped instrumental resolution ( ∼ − full width athalf maximum (FWHM) for the macro setup).Figure 1 shows the Raman spectrum of the slightlyunderdoped LCCO film at temperature T = 30 K in thez(xx)z polarization configuration. Here, we use the Porto FIG. 1: Raman spectra at 30 K of (top) the slightly under-doped LCCO film ( T c ∼
29 K) in z(xx)z polarization con-figuration for 514 . Opowder sample (99.9 %), and (bottom) the Cu O single crys-tal each for 532 . O spectra scaledfor clarity.) notation l(ij)m, where l and m denote the direction of in-cident and scattered light and (ij) their polarization, re-spectively. Both the slightly underdoped and the slightlyoverdoped LCCO films show comparable Raman spectra.We detect four main modes at 153.5 cm − (mode A),213.5 cm − (mode B), 308.5 cm − (mode C), and a dou-ble mode at ∼ − (mode D). A mode at ∼ − , which would correspond to an apical oxygen vibra-tion in the T -structure of La CuO (Ref. [11]), is not ob-served, indicating the absence of apical oxygen residuals.A group theoretical analysis of the phonon modes of the T (cid:48) -structure (tetragonal space group I /mmm ( D h ))[18] yields A g + B g + 2 E g Raman-active modes. To ourknowledge, Raman data on T (cid:48) -LCCO have not yet beenreported. We therefore use data on related T (cid:48) -lanthanidecuprates such as NCCO for comparison [11, 19, 20]. Theenergies of modes B and C are close to, but somewhatlower than those of the A g and B g vibrations, respec-tively, of the Nd and out-of-plane oxygen atoms in T (cid:48) -NCCO.[20] A downward shift of the LCCO modes withrespect to NCCO would be in line with a continuous fre-quency evolution previously observed in the series of lan-thanide cuprates [20].The assignment of modes A and D is more difficult,because Raman spectra of other cuprates with the T (cid:48) -structure do not exhibit modes with similar energies.We have therefore investigated possible contaminationby impurity phases, including Cu O which has been pre-viously observed as a contaminant in NCCO.[18] Figure1 shows a comparison to reference spectra of compressedCu O powder sample I (99.9 % purity) and the Cu Osingle crystal for 532.0 nm laser excitation. Based onthis comparison and on earlier data on Cu O [21, 22],
FIG. 2: X-ray Θ-2Θ-scan of the slightly underdoped LCCOfilm at room temperature (Cu K α with λ = 1 .
54 ˚A). TheBragg reflections of SrTiO and gold are due to the substrateand gold contacts on top of the film. modes A and D can be identified with the infrared-allowed Γ (1)15 and Γ (2)15 modes of Cu O.[23] Due to res-onance effects, these modes can become Raman-activewith high intensity.[24, 25] The much stronger intensityof the Γ (1)15 mode in the Cu O powder compared to theCu O single crystal can be attributed to powder averag-ing over all possible polarization geometries. Mode A inthe LCCO films also shows high intensity, suggesting anisotropic orientation of the Cu O impurity phase. Theshift of ∼ − of mode A with respect to the Γ (1)15 mode of Cu O may be a consequence of stress imposedby the LCCO matrix. Note that modes with energiesroughly comparable to those of modes B and C are alsopresent in the Cu O reference spectra, but the formermode is shifted by ∼ . − with respect to mode B,and the latter mode is extremely weak. We will showbelow that these modes likely originate from LCCO.In order to directly characterize the chemical com-position of the LCCO films, we used high-intensity x-ray diffraction (XRD). Figure 2 shows a Θ-2Θ scan ofthe slightly underdoped LCCO film at room tempera-ture, plotted on a logarithmic intensity scale. Based ona comparison with the calculated Bragg angles for the T (cid:48) -structure [4], we clearly identify the expected [001]-oriented T (cid:48) -structure as the main phase. AdditionalXRD pole figure measurements (not shown here) also re-veal the Bragg peaks (103) and (110). This indicatestwo T (cid:48) -LCCO minority phases with different growth di-rections, which are stabilized by a well-known, acciden-tal match between the in-plane and out-of-plane latticeparameters ( c/a ≈ T -structure inclusions. The XRD mea-surements also confirm the presence of a Cu O impurityphase, as well as a trace amount of La O (Fig. 2). Theintensities of the main Cu O Bragg reflections are aboutfour orders of magnitude below those of LCCO, but theirratios indicate random orientation of the Cu O crystal-
FIG. 3: SEM image of the slightly underdoped LCCO film.The upper inset shows a partly etched surface area. lites, in contrast to the epitaxially oriented LCCO ma-trix. Taking powder averaging of the Cu O Bragg reflec-tions into account, the XRD data imply that the volumefraction of Cu O is only about an order of magnitudebelow the one of LCCO.The microstructure and the local variation of the chem-ical composition of the LCCO films were further ana-lyzed by scanning electron microscopy (SEM) and en-ergy dispersive x-ray (EDX) analysis. The microstruc-ture of both LCCO films appeared comparable. Figure3 shows a SEM image of the slightly underdoped LCCOfilm. About 18% of the surface area is uniformly fleckedwith particles of ∼ µ m diameter. While the surfacebackground exhibits the composition ratio (La+Ce):Cu ∼ . ± . T (cid:48) -LCCO, the particles werefound to be centers of strongly enhanced Cu content, withthe statistically averaged ratio (La+Ce):Cu ∼ . ± . O impurityphase. The high and uniform surface coverage with par-ticles supports the picture of an isotropically orientedCu O impurity phase of (cid:38)
10% in the LCCO films. Inaddition, we observed line structures oriented along the ab crystal axes, which exhibit a slightly enhanced Cucontent. These structures may originate from structuraldefects or from the T (cid:48) -LCCO minority phases. In theinset of figure 3 we show a surface area of the slightlyunderdoped LCCO film, which was partly etched by Arion bombardment. We find that both the particles andthe line structures reach deeply into the film, suggestingtheir formation during the crystal growth. The concen-trations of impurity atoms beyond the constituent ele-ments of LCCO were below the EDX detection limit.In order to relate the Raman spectra of Fig. 1 tothe microscopic observations of Fig. 3, we employed themicro-Raman setup. Figure 4a shows a line scan of theintensity of mode A in the slightly underdoped LCCOfilm. The intensity varies strongly on a length scale of ∼ µ m, which is comparable to the size of the parti-cles in Fig. 3. This underscores the assignment of thismode to the Cu O impurity phase. The intensities of
FIG. 4: (a) Line scan of the intensities of modes A and B ofthe underdoped LCCO film taken in the micro-Raman setupwith 50 × microscope objective and λ = 532 . T = 30 K before and after sputtering of ∼
250 nm,and (c) SEM images before and after sputtering. modes B and C, on the other hand, depend only weaklyon the measuring position (Fig. 4a), supporting the con-clusion that they do not originate from Cu O, but fromthe T (cid:48) -LCCO host material. We used argon ion sput-tering under vacuum in an attempt to remove the Cu Oparticles. A comparison of SEM images of the surfacebefore and after removal of ∼
250 nm (Fig. 4c) confirmsthat the particles are stuck deeply inside the LCCO ma-trix. While it is thus not possible to remove the particles,sputtering still reduces both the volume fraction of Cu Oand the intensity of mode A (Fig. 4b). Modes B and C,on the other hand, are nearly unaffected by sputtering.Having obtained a thorough understanding of the mi-crostructure and phase composition of the LCCO films,we now focus on anomalies in the temperature depen-dence of the different optical modes at the superconduct-ing transition temperature T c . Figure 5 shows the Ra-man spectra of the slightly underdoped LCCO film from10 K to 300 K in z(xx)z polarization for 514 . T ∼
150 K. A similar activation wasalso observed in the Cu O reference samples (not shownhere), again confirming their common origin. In contrast,the intensities of modes B and C exhibit a much weakertemperature dependence. Figure 6a shows the tempera-ture dependence of the frequency and FWHM of modeA. The solid lines are the result of fits to the data above T c , using an expression based on anharmonic phonon- FIG. 5: Raman spectra of the slightly underdoped LCCO filmat temperatures 10 K ≤ T ≤
300 K in z(xx)z polarizationconfiguration for 514 . ∼ . phonon interactions [26, 27]. For simplicity we assumeda symmetric decay into two product modes, which leadsto the following expressions for the phonon frequency ω ph and FWHM Γ ph : ω ph ( T ) = − A (cid:18) aexp ( (cid:126) ω / k B T ) − (cid:19) + ω , Γ ph ( T ) = Γ a (cid:18) aexp ( (cid:126) ω / k B T ) − (cid:19) + Γ b , where A and Γ a are positive constants and a correctsfor terms arising from nonsymmetric phonon decay pro-cesses. Γ b represents the temperature-independent partof Γ ph .While mode A follows nearly perfectly the expressionfor anharmonic decay in the normal state, we observesignificant deviations from this behavior below T ∼ T c .The frequency softens by ∼ . − upon cooling below T c , and the temperature dependence of the linewidth ex-hibits a change in slope in the same temperature range,which corresponds to a narrowing of ∼ . − . Thedeviations from the anharmonic behavior are illustratedby the shaded areas in figure 6a. In contrast to modeA, the temperature dependence of the parameters char-acterizing modes B and C differs substantially from thestandard anharmonic behavior (Figs. 6b and c). In par-ticular, their frequencies increase continuously with in-creasing T , opposite to the behavior expected from an- FIG. 6: Temperature dependence of the frequency and FWHM of modes A, B, and C in the underdoped LCCO film (see Fig.1). The triangular points represent a second independent data set at low temperatures. The solid lines are the result of fits tothe data points above T c according to the theory of anharmonic phonon decay [26, 27] (see text for details). The shaded areasindicate deviations from the anharmonic temperature dependence below ∼ T c .FIG. 7: Temperature dependence of the frequency andFWHM of modes A and C in the slightly overdoped LCCOfilm ( T c ∼
27 K). The 487 .
986 nm laser line was used forexcitation. For solid lines see caption of figure 6. harmonicity. Below T c , however, they exhibit softeningand narrowing of the same magnitude as the one observedfor mode A. Remarkably, both the anomalous normal-state behavior of mode C and the renormalization of thephonon frequencies below T c are absent in the slightlyoverdoped LCCO film (Fig. 7), suggesting that thesefeatures are controlled by the doping level of LCCO.For comparison we have measured the temperaturedependence of the magnetic susceptibility and phononfrequencies of the two commercial Cu O powders. Al-though pure Cu O is nonmagnetic,[28] the susceptibili-ties of both powders exhibit pronounced low-temperatureCurie tails due to magnetic impurities (Fig. 8). The sus-ceptibility of one of the two samples even exhibits a small
FIG. 8: Temperature dependence of the magnetic suscepti-bility for different magnetic fields (FC = field cooled, ZFC= zero field cooled), and frequency of the Γ (1)15 mode of twocompressed Cu O powder samples with different purity levels. anomaly at low magnetic fields and T ∼ O.[29] The overall temperature depen-dence of the frequency of the intense Γ (1)15 mode shownin Fig. 8 is consistent with the one expected for anhar-monic decay. In powder sample I, however, the modeabruptly hardens by ∼ . − between T ∼
15 Kand the base temperature of 5 K (Fig. 8a), while pow-der sample II shows a small hint of a softening (Fig.8b). These phonon anomalies are surprising, becauselow-temperature structural instabilities have not been re-ported for Cu O.[30] Moreover, they are apparently un-correlated with features in the magnetic susceptibility.The phonon anomaly therefore likely arises from sample-specific defects or impurities. Note that the anomaly ofthe Cu O vibration in powder sample I with higher puritylevel is smaller in magnitude and of opposite sign thanthe one exhibited by mode A in the underdoped LCCOfilm (which was fabricated from ultrapure metal sourcesand is therefore much less affected by magnetic impuri-ties than the commercial powders), and that it occurs ata lower temperature. Nonetheless, the Raman data onCu O do indicate that the low-temperature behavior ofthe Γ (1)15 mode is quite sensitive to microstructural de-tails. This may provide clues to the origin of the anoma-lous low-temperature behavior of mode A in the LCCOfilm, which we had identified with the Γ (1)15 mode of theCu O inclusions.In summary, all of the Raman-active phonons observedin the underdoped LCCO film (including modes B andC that are likely due to the LCCO host material aswell as mode A, which likely arises from Cu O inclu-sions) exhibit anomalies in their temperature dependencethat are reproducible and clearly outside the experimen-tal error bars at a temperature that is consistent with T c . In discussing these observations, we first ignore theCu O inclusions and consider the standard picture ofsuperconductivity-induced phonon anomalies which hasbeen established based on Raman data on hole-dopedhigh-temperature superconductors.[11, 31, 32] Accordingto this theory, optical phonons with energies higher thantwice the superconducting gap, 2∆, harden below T c ,and their linewidths increase due to an enhanced den-sity of states above the gap. Conversely, phonons withenergies below 2∆ are expected to soften. Deviationsfrom this behavior are predicted only close to 2∆.[31]A microscopic formulation of this theory yields a quan-titative description of electronic Raman scattering andsuperconductivity-induced phonon self-energy anomaliesin hole-doped YBa Cu O y .[12] The superconductingenergy gap LCCO is known from prior transport mea-surements on the same films that we have investigatedby Raman light scattering.[15] Tunneling characteristics(one of which is reproduced in the inset of Fig. 5) showcoherence peaks at an energy of 9 meV, which implies2∆ ∼
75 cm − . Since the energies of all of the Raman-active phonons we have discussed are far above 75 cm − ,the standard model predicts a weak hardening and broad-ening below T c , in complete contrast to our observations.Since the standard theory of superconductivity-induced phonon renormalization fails to account for theobservations displayed in Fig. 6, we are forced to con-sider more unconventional scenarios. A possible explana-tion of the softening and narrowing of the optical modesbelow T c would be a second gap with magnitude in ex-cess of 2∆ ∗ ∼
100 meV that opens at a temperatureclose to T c or is at least affected by the superconductingphase transition. In this case, all of the Raman-activeoptical modes would be located below this threshold en- ergy, and the observed softening could be explained by astraightforward application of the theory of phonon self-energy to this high-energy gap. The narrowing of thephonon linewidths would then be a direct consequenceof the reduced number of relaxation channels due to theloss of spectral weight below 2∆ ∗ . Angle-resolved pho-toemission spectroscopy [33, 34] and optical spectroscopy[35, 36] experiments on underdoped NCCO have indeedyielded evidence of a high-energy “pseudogap” that opensup below a temperature T ∗ > T c . Both T ∗ and themagnitude of ∆ ∗ were found to decrease with increas-ing doping level, [34, 37], so that T ∗ cuts the super-conducting phase boundary at optimal doping.[36] Inthe superconducting regime of NCCO, ∆ ∗ ∼
100 meV (cid:29) ∆.[34] Although no evidence has yet been reportedof a similar phenomenon in LCCO, it is thus conceivablethat interplay between superconductivity and the pseu-dogap (which may in turn be related to the presence ofantiferromagnetic order [34]) could explain the unusualsuperconductivity-induced phonon self-energy anomalieswe have observed. However, nearly identical anomaliesexhibited by mode A (which, as we have argued, probablyoriginates in the Cu O inclusions) point to a more com-plex picture in which proximity and/or inverse proximityeffects at the LCCO/Cu O interface are also involved. Itis possible, for instance, that charge transfer across theinterface leads to the formation of magnetic Cu ionsand induces cooperative magnetism at the boundariesof the Cu O inclusions, which in turn enhances antifer-romagnetic order and the pseudogap in LCCO. An in-vestigation of phonon anomalies close the interface byRaman spectroscopy with spatial resolution comparableto the superconducting coherence length is well beyondour current experimental capabilities, but may becomepossible in the future based on advances in near-field op-tics. Based on our observations, it appears worthwhile tostudy such effects systematically using other experimen-tal methods.In addition to its scientific interest, the technical as-pects of our study are of general relevance for the investi-gation of electron-doped high-temperature superconduc-tors. Inclusions of impurity phases such as lanthanideoxides [1, 2] and Cu O are hard to avoid during synthe-sis and continue to be present in state-of-the-art crys-tals and films. They are also difficult to detect based onstandard x-ray diffraction, either because epitaxial inter-growth leads to Bragg peak positions that are similar tothose of the host phase [1, 2] or because powder averag-ing greatly reduces the intensity of the impurity Braggreflections. Note, in particular, that the Cu O inclusionsin our LCCO films required an experimental setup witha ratio of LCCO Bragg intensities to the noise floor of ∼ (Fig. 2), which goes beyond the typical diagnos-tics run on thin films. We have shown that both electronmicroscopy and micro-Raman spectroscopy are powerful,complementary diagnostic tools (Fig. 4). As we haveseen, a thorough understanding of impurity inclusions isimportant not only for transport, but also for spectro-scopic measurements.We thank Y. Kuru, B. Bohnenbuck, C. Busch, M. Kon-uma, E. Br¨ucher, M. Schaloske, R.K. Kremer, and A.Schulz for experimental support, and M. Cardona, D. Manske, and R. Zeyher for fruitful discussions. This workwas supported by the Deutsche Forschungsgemeinschaft(DFG) through project Kl930/11. [1] P.K. Mang, S. Larochelle, A. Mehta, O.P. Vajk, A.S. Er-ickson, L. Lu, W.J.L. Buyers, A.F. Marshall, K. Prokes,and M. Greven, Phys. Rev. B
0, 094507 (2004)[2] G. Roberge, S. Charpentier, S. Godin-Proulx, P. Rauwel,K.D. Truong, P. Fournier, J. Cryst. Growth
11, 1340(2009)[3] For a review, see N.P. Armitage, P. Fournier, and R.L.Greene, preprint arXiv:0906.2931[4] M. Naito and M. Hepp, Jpn. J. Appl. Phys. , L485(2000)[5] A. Sawa, M. Kawasaki, H. Takagi, Y. Tokura, Phys. Rev.B
6, 014531 (2002)[6] Y. Krockenberger, J. Kurian, A. Winkler, A. Tsukada,M. Naito, and L. Alff, Phys. Rev. B , 060505(R) (2008)[7] M. Naito, A. Tsukada, T. Greibe, and H. Sato, Proc.SPIE , 140 (2002)[8] A. Tsukada, H. Yamamoto, and M. Naito, Phys. Rev. B , 174515 (2006)[9] A. Tsukada, Y. Krockenberger, M. Noda, H. Yamamoto,D. Manske, L. Alff, and M. Naito, Solid State Commun. , 427 (2005)[10] For a review, see T.P. Devereaux and R. Hackl, Rev.Mod. Phys.
9, 175 (2007)[11] C. Thomsen, in
Light Scattering in Solids VI, TopicsAppl. Phys. , edited by M. Cardona and G. G¨untherodt(Springer, 1991), Vol. 68, p. 285[12] M. Bakr, A.P. Schnyder, L. Klam, D. Manske, C.T. Lin,B. Keimer, M. Cardona, and C. Ulrich, Phys. Rev. B ,064505 (2009)[13] G. Blumberg, A. Koitzsch, A. Gozar, B.S. Dennis, C.A.Kendziora, P. Fournier, and R.L. Green, Phys. Rev. Lett. , 107002 (2002)[14] M.M. Qazilbash, A. Koitzsch, B.S. Dennis, A. Gozar,Hamza Balci, C.A. Kendziora, R.L. Greene, and G.Blumberg, Phys. Rev. B
2, 214510 (2005)[15] M. Wagenknecht, D. Koelle, R. Kleiner, S. Graser, N.Schopohl, B. Chesca, A. Tsukada, S.T.B. Goennenwein,and R. Gross, Phys. Rev. Lett. , 227001 (2008); M.Wagenknecht, M. Rahlenbeck, D. Koelle, R. Kleiner, A.Tsukada, S.T.B. Goennenwein, and R. Gross, submittedto Phys. Rev. B; M. Rahlenbeck, unpublished data[16] M. Wagenknecht,
Korngrenz-Tunnelspektroskopie amelektronendotierten Kupratsupraleiter La − x Ce x CuO ,Ph.D. thesis, T¨ubingen University, 2008[17] T. Greibe, MBE growth of superconducting thin films , Technical report, NTT Basic Research Laboratories,2001[18] V.G. Hadjiev, I.Z. Kostadinov, L. Bozukov, E. Dinolova,and D.M. Mateev, Solid State Commun. , 1093 (1989)[19] S. Sugai, T. Kobayashi, and J. Akimitsu, Phys. Rev. B , 2686 (1989)[20] E.T. Heyen, R. Liu, M. Cardona, S. Pi˜nol, R.J. Melville,D. McK. Paul, E. Mor´an, and M.A. Alario-Franco, Phys.Rev. B , 2857 (1991)[21] Y. Petroff, P.Y. Yu, and Y.R. Shen, Phys. Rev. B ,2488 (1975)[22] Y. Petroff, P.Y. Yu, and Y.R. Shen, Phys. Rev. Lett. ,1558 (1972)[23] K. Huang, Z. Phys. , 213 (1963)[24] Y. Petroff, J. Phys. (Paris) Colloq. C3, , 277 (1974)[25] P.Y. Yu, Y.R. Shen, and Y. Petroff, Solid State Commun. , 973 (1973)[26] J. Men´endez and M. Cardona, Phys. Rev. B , 2051(1984)[27] V.G. Hadjiev, X. Zhou, T. Strohm, and M. Cardona,Phys. Rev. B , 1043 (1998)[28] H.L. Schlafer and G. Gliemann, Basic principles of Lig-and Field Theory (Wiley, New York 1969) p. 120[29] Ch. Chen, L. He, L. Lai, H. Zhang, J. Lu, L. Guo, andY. Li, J. Phys.: Condens. Matter , 145601 (2009)[30] A. Werner and H. D. Hochheimer, Phys. Rev. B , 5929(1982)[31] R. Zeyher and G. Zwicknagl, Z. Phys. B - CondensedMatter , 175 (1990)[32] T.P. Devereaux, Phys. Rev. B , 10287 (1994)[33] N.P. Armitage, D.H. Lu, C. Kim, A. Damascelli, K.M.Shen, F. Ronning, D.L. Feng, P. Bogdanov, and Z.-X.Shen, Phys. Rev. Lett. , 147003 (2001)[34] H. Matsui, T. Takahashi, T. Sato, K. Terashima, H. Ding,T. Uefuji, and K. Yamada, Phys. Rev. B , 224514(2007)[35] Y. Onose, Y. Taguchi, K. Ishizaka, and Y. Tokura, Phys.Rev. Lett. , 217001 (2001)[36] A. Zimmers, J.M. Tomczak, R.P.S.M. Lobo, N. Bon-temps, C.P. Hill, M.C. Barr, Y. Dagan, R.L. Greene, A.J.Millis, and C.C. Homes, Europhys. Lett. , 225 (2005)[37] Y. Dagan, M.M. Qazilbash, and R.L. Greene, Phys. Rev.Lett.9