- Email: [email protected]
Antiferromagnetism and superconductivity: Determination of the Cu spin–spin relaxation time T2 and the spin–lattice relaxation time T1 in Gd1.5Ce0.5Sr2Cu2RuO10
Solid State Communications 149 (2009) 1702–1705
Contents lists available at ScienceDirect
Solid State Communications journal homepage: www.elsevier.com/locate/ssc
Antiferromagnetism and superconductivity: Determination of the Cu spin–spin relaxation time T2 and the spin–lattice relaxation time T1 in Gd1.5 Ce0.5 Sr2 Cu2 RuO10 H.A. Blackstead a,∗ , W.B. Yelon b , M.P. Smylie a a
Physics Department, University of Notre Dame, Notre Dame, IN 46556, USA
b
Missouri University of Science and Technology, Materials Research Center and Department of Chemistry, Rolla, MO 65401, USA
article
info
Article history: Received 4 June 2009 Accepted 12 June 2009 by F. De la Cruz Available online 23 June 2009 PACS: 74.20.Mn 74.25.Ha 74.70Pq 76.50+g
abstract Gd1.5 Ce0.5 Sr2 Cu2 RuO10 exhibits antiferromagnetic resonance at 23.9 GHz for applied fields less than 1000 Oe with a spin–spin relaxation time T2 of approximately 0.45 ns, and with a spin–lattice relaxation time T1 of at least 320 µs. Since in the homologue, Eu1.5 Ce0.5 Sr2 Cu2 RuO10 , the Ru atoms evidently fail to exhibit magnetic order, the antiferromagnetic resonance must arise from the cuprate planes. In other homologues, the cuprate planes are known to order ferromagnetically and are stacked in an antiferromagnetic configuration. The large value of T1 suggests that phonon mediation plays no role in high temperature superconductivity. In addition, the presence of ferromagnetic cuprate planes is inconsistent with spin-fluctuation models of high temperature superconductivity. © 2009 Elsevier Ltd. All rights reserved.
Keywords: A. Superconductivity A. Antiferromagnetism
1. Introduction The origin of the anomalous magnetization observed in fieldcooled measurements of the superconducting ruthenocuprates such as Gd1.5 Ce0.5 Sr2 Cu2 RuO10 (O10) has been widely attributed to ferromagnetic order, or canted antiferromagnetism, of the Ru in the RuO2 layer [1]. In the closely related O8 compounds such as GdSr2 RuCu2 O8 , the anomalous magnetization has been attributed to a ferromagnetic component on the Ru sites, but several neutron scattering studies have failed to show any evidence for such an origin, reporting, instead, simple antiferromagnetic Ru order. The situation in the O10 compounds is further complicated by the body centered unit cell which leads to a very long c-axis. Fig. 1 shows that antiferromagnetic superexchange interactions of the Ru spins in the O10 compounds are geometrically frustrated by the presence of the fluorite {(Gd, Ce)2 O2 } block which introduces a shift of (a/2, a/2) in the layer stacking. Neutron diffraction measurements by Kuz’micheva et al. [2] failed to detect magnetic order in Nd1.4 Ce0.6 Sr2 Cu2 RuO10 and more recent extensive measurements by Lynn [3] et al. failed to find Ru antiferromagnetic order and also specifically excluded ferromagnetic order of the Ru in
Eu1.5 Ce0.5 Sr2 Cu2 RuO10 in the temperature range of interest here. Since the Y homologue with Nb replacing Ru also exhibits similar magnetic behavior [4], and closely related YSr2 Cu2 NbO8 does as well [5], attention is necessarily focused on the cuprate (CuO2 ) planes. Neutron diffraction studies [6,7] have shown these to be magnetically ordered in the closely related material YSr2 Cu2 RuO8 . As it happens, the magnon energy gap of the ordered Cu spins in the ruthenocuprates is sufficiently small that antiferromagnetic resonance is easily detected at conventional microwave frequencies. In some cases, antiferromagnetic resonance can also be observed in cuprates [8]. In the following, we present the results of such measurements as a function of the rf power level at a temperature of 100 K, substantially above the ∼40 K superconducting transition temperature. These data show that the cuprate plane antiferromagnetic mode is surprisingly easy to saturate, indicating a long spin–lattice relaxation time, T1 . In addition, the spin–spin relaxation time T2 is quite short, leading to a broad fielddependent response. The Gd spins exhibit paramagnetic resonance which shows no saturation effects in the range of rf power levels employed. The Gd resonance is superimposed on the broad Cu response, and is used to calibrate the intensity of Cu resonance. 2. Experimental details
∗ Corresponding address: Physics Department, University of Notre Dame, 225 Nieuland Science Hall, 46556 Notre Dame, IN, USA. Tel.: +574 631 7078. E-mail address: [email protected] (H.A. Blackstead). 0038-1098/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ssc.2009.06.020
The procedure for sample preparation is described elsewhere [9], and follows conventional solid state reaction procedures. The microwave spectrometer employed has novel features
H.A. Blackstead et al. / Solid State Communications 149 (2009) 1702–1705
1703
Fig. 1. Schematic structure of Gd1.5 Ce0.5 Sr2 Cu2 RuO10 , here 1/2 of the unit cell is shown. Replacement of Gd in GdSr2 Cu2 RuO8 with the fluorite block (Gd, Ce)2 O2 results in a similar structure in which the antiferromagnetic Ru superexchange is frustrated, and it fails to exhibit long-ranged order. The Cu order ferromagnetically and are stacked antiferromagnetically. The Cu superexchange interactions across the RuO2 and SrO layers are not frustrated for this configuration.
which were used to advantage in these measurements. One of these features is that the spectrometer (see Fig. 2) does not utilize magnetic field modulation. Field modulation is employed with a lock-in amplifier to extract the derivative of the lineshape as a function of the applied field. For resonances with large linewidths, very small signals would result. Since we did not use this technique, it was possible to measure signals with very large linewidths. In addition, the power reflected from the samplebearing cavity was amplified using a low-noise solid state amplifier with a gain of nearly 30 db. An attenuator following the klystron source was used to control the power level incident on the cavity. Another attenuator placed after a directional coupler, and before the amplifier, was adjusted to compensate for changes in incident power levels. Thus, the point-contact diode detector, except for the lowest power levels utilized, was driven by a nearly constant power level. This is of some importance, since such detectors have a square-law response, and a constant power level biases them to a reproducible sensitivity. Additional circuitry (not shown) was employed to ‘‘lock’’ the klystron frequency to the resonant frequency of the sample-bearing TE101 rectangular cavity [10]. This feature ensures that only changes in χ ’’ will be detected. The polycrystalline sample was mounted in the center of the bottom of the cavity, using a small quantity of silicone grease. For the data reported here, the dc magnetic field H was applied perpendicular to the very uniform cavity magnetic field (Hrf ), in the usual configuration for magnetic resonance. Similar Cu signals were detected with H ||Hrf , but in that case the paramagnetic Gd resonance was very small, and it was not useful as a calibration signal. 3. Analysis of the data The resonant response (aside from an amplitude including Hrf2 ) was characterized using a form given by Dyson [11,12] for conducting materials, including the term in the denominator which involves the rf field intensity and leads to saturation at high rf power levels: y (H ) =
1 + αγ T2 (H − H0 ) 1+γ
2T 2 2
(H − H0 )2 + γ 2 T1 T2 Hrf2
.
(1)
Here γ is the gyromagnetic ratio given by γ = (g µB )/h, T1 and T2 are the spin–lattice and spin–spin relaxation times, respectively,
Fig. 2. Simplified block diagram of the K-band microwave spectrometer. The directional coupler serves to separate the incident power from the power reflected from the resonant cavity. The power level was set with the attenuator in series with the klystron source, and an approximately constant power level was provided to the amplifier by adjusting the compensating attenuator following the directional coupler.
and α = 0.63 is a parameter of the fit to the data expressing the line-asymmetry. The signal resulting from an experiment is proportional to Hrf2 y (H ); saturation of the response occurs for large rf fields. At resonance, the lineshape y(H = H0 ) is: y ( H = H0 ) =
1 1+γ
2T
2 1 T2 Hrf
.
(2)
Note that y(H ) is maximum for H > H0 , a consequence of the asymmetry of the Dysonian lineshape function. Because the Gd relaxation times are both small, the Gd paramagnetic response does not saturate. It follows that if γ 2 T1 T2 Hrf2 > 1, the signal intensity, normalized by the Gd signal, will be substantially reduced at high power levels [13]. Usually in metals, this condition is difficult to satisfy owing to the small sizes of both T1 and T2 . For sufficiently low power levels the resonance signal is not a function of T1 , and T2 is independently determined by the linewidth, see Figs. 3 and 4. As −db
a function of attenuator setting (in db), Hrf = Hrf 0 10 20 . The product T1 Hrf2 0 was found by fitting signal intensities as a function of this parameter, see Fig. 5. An upper limit for Hrf 0 was determined from the cavity radiation quality factor ‘‘Q ’’, the input rf power level, and the cavity and waveguide dimensions [14,15]. The result of this evaluation is that Hrf 0 ≤ 2.0 Oe. It follows from the results shown in Fig. 5 and the estimated value for Hrf that T1 ≥ 320 µs, while from the linewidth, T2 was found to be ∼0.45 ns. 4. Discussion We have recently shown in a sequence of materials which contain Ru4+ or Ru5+ in octahedral coordination with six oxygen, that the Ru are esr silent in both paramagnetic and magnetically ordered configurations [16]. These results exclude Ru as the source of the resonance discussed here. The results of the recent neutron diffraction experiments on two materials, Eu1.5 Ce0.5 Sr2 Cu2 RuO10 and Y1.5 Ce0.5 Sr2 Cu2 NbO10 are of critical importance to the interpretation of the results of the experiments presented here.
1704
H.A. Blackstead et al. / Solid State Communications 149 (2009) 1702–1705
Fig. 3. Resonance signal with 50 db attenuation at 100 K (black dots). In addition the fits using the summed resonance response functions for the Cu and Gd spins are shown (red line). The Gd response (green line) peaks near µ0 H = 0.9 T, while the Cu response (blue line) peaks near 0.08 T. The Cu spin–spin relaxation time T2 is approximately 0.45 ns. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Lynn et al., using low absorption cross-section 153 Eu, found no evidence for Ru magnetic ordering in Eu1.5 Ce0.5 Sr2 Cu2 RuO10 . Magnetic measurements [4] on Y1.5 Ce0.5 Sr2 Cu2 NbO10 , a material
which has neither a strong neutron absorber such as Gd or Eu, nor magnetic Ru, show magnetic order, which necessarily must be attributed to the cuprate planes. Magnetic resonance studies [4] of Y1.5 Ce0.5 Sr2 Cu2 NbO10 find behavior similar to that reported here, at a lower temperature. Thus, even if the RuO2 layer orders [17] (perhaps two-dimensionally), it has no impact on our results. We have concluded that, in all of these ruthenocuprates, the cuprate planes are magnetically ordered in an antiferromagnetic configuration in which ferromagnetic planes are stacked with alternating magnetization directions. This configuration is almost invisible to neutron diffraction, because the magnetic order only enhances nuclear reflections, without leading to additional reflections. Because the moment is small, (∼0.4 µB ) the Cu contribution to these peaks is small. The magnetic resonance data are unambiguous that the magnetic ordering is antiferromagnetic [18] while the system exhibits pronounced metamagnetism which has been widely misinterpreted as ferromagnetism. This metamagnetism most likely results from the relative rotation of the magnetic layers as driven by an applied magnetic field. These results have strong implications for the origin of superconductivity of the cuprates. If there is one pairing mechanism, then it seems very unlikely that it has anything to do with either magnetic spin fluctuations or phonons. This conclusion is a direct consequence of the magnetically ordered cuprate planes and the very weak Cu spin–lattice coupling these measurements find. In addition, the relatively small size of T2 indicates that the Cu spins are strongly coupled. In a broader context, superconductivity in the ruthenocuprates seems to occur in systems in which there are ferromagnetic layers separated by either metallic or oxide layers in an antiferromagnetic configuration. The presence of Ru, either antiferromagnetically ordered, as in GdSr2 Cu2 RuO8 , or as in Eu1.5 Ce0.5 Sr2 Cu2 RuO10 , has little impact on the superconducting transition temperature, but does raise the magnetic transition temperature of the cuprate planes in both cases, as compared to Nb.
Fig. 4. Microwave resonance signals as a function of input power. The Cu antiferromagnetic resonance peaks in intensity at low applied fields, near µ0 H = 0.085 T, while the Gd paramagnetic resonance is seen near µ0 H = 0.9 T. The Gd resonance was used to normalize the several signals. The signal for 50 db attenuation has a noticeable noise component.
H.A. Blackstead et al. / Solid State Communications 149 (2009) 1702–1705
1705
Acknowledgements H. A. B. and W. B. Y. thank the U. S. Department of Energy (MISCON Grant No. DE-FG0290ER45427) for generous past support, and thank I. Felner for providing the sample material and D. B. Pulling and P. J. Beeli for technical assistance. References
Fig. 5. Cu antiferromagnetic signal level intensity as a function of input power. These data (red dots) are described by Eq. (2) for the intensity at resonance (solid line). The Cu spin–lattice relaxation time T1 was found to be approximately 320 µs. The spin–spin relaxation time was found from the linewidth.
It remains to be determined if this is due to localized Ru magnetism or to the availability of electronic d-states, which are not present in closed shell Nb5+ . Recently, Anderson [19] has suggested that boson mediated pairing may not be needed for high temperature superconductivity. While this is a paradigm-changing proposal, our findings, which pose severe limits for both spin-fluctuation and phonon pairing schemes, are consistent with such a scenario.
[1] I. Felner, U. Asaf, Y. Levi, O. Millo, Phys. Rev. B 55 (1997) R3374. [2] G.M. Kuz‘micheva, A.I. Kubakov, I.E. Kostyleva, S.A. Lachenkov, Zh. Neorg. Khim 51 (2006) 1949 (Zhuranal Neorganicheskoi Khimii 51 (2006) 2070). [3] J.W. Lynn, Y. Chen, Q. Huang, S.K. Goh, G.V.M. Williams, J.L. Tallon, Phys. Rev. B 76 (2007) 014519. [4] H.A. Blackstead, W.B. Yelon, M.P. Smylie, Q. Cai, J. Lamsal, V.P.S. Awana, S. Balamurugan, E. Takayama-Muromachi, (unpublished). [5] H.A. Blackstead, W.B. Yelon, M. Kornecki, M.P. Smylie, Q. Cai, J. Lamsal, V.P.S. Awana, H. Kishan, S. Balamurugan, E. Takayama-Muromachi, Phys. Rev. B 75 (2007) 140514(R). [6] H.A. Blackstead, W.B. Yelon, M. Kornecki, M.P. Smylie, Q. Cai, J. Lamsal, V.P.S. Awana, H. Kishan, S. Balamurugan, E. Takayama-Muromachi, Phys. Rev. B 76 (2007) 094507. [7] W.B. Yelon, Q. Cai, Jagat Lamsal, H.A. Blackstead, M. Kornecki, V.P.S. Awana, H. Kishan, S. Balamurugan, E. Takayama-Muromachi, J. Appl. Phys. 101 (2007) 09G104. [8] H.A. Blackstead, J.D. Dow, M. Osada, M. Kakihana, Phys. Rev. B 76 (2007) 024522. [9] I. Felner, U. Yaron, U. Asaf, T. Kröner, V. Breit, Phys. Rev. B 49 (1994) 6903. [10] C.P. Poole Jr, Electron Spin Resonance, Interscience Publishers, 1967, 264. [11] F.J. Dyson, Phys. Rev. 98 (1955) 349. [12] C.P. Poole Jr, Electron Spin Resonance, Interscience Publishers, 1967, 814. [13] If the term in the denominator of the line-shape function which varies as Hrf2
[14] [15] [16]
[17] [18] [19]
could be neglected, the overall signal intensity would increase with Hrf2 as the Gd response does. C.P. Poole Jr, Electron Spin Resonance, Interscience Publishers, 1967, 302. The loaded Q (Ql ) was found to be ∼1500, and the maximum power input to the cavity was ∼250 mW. H.A. Blackstead, W.B. Yelon, M. Kornecki, M.P. Smylie, P.J. McGinn, Q. Cai, B.W. Benapfl, S.D. Knust, J Mag. Mag. Mat. (in press) http://dx.doi.org/10.1016/ j.jmmm.2009.05.048. A.C. Mclaughlin, I. Felner, V.P.S. Awana, Phys. Rev. B 78 (2008) 094501. These authors claim that the Ru exhibit long-range magnetic order near 160 K. E.A. Turov, Physical Properties of Magnetically Ordered Crystals, Academic Press, 1965, 151. P.W. Anderson, Science 316 (2007) 1705.
Contents lists available at ScienceDirect
Solid State Communications journal homepage: www.elsevier.com/locate/ssc
Antiferromagnetism and superconductivity: Determination of the Cu spin–spin relaxation time T2 and the spin–lattice relaxation time T1 in Gd1.5 Ce0.5 Sr2 Cu2 RuO10 H.A. Blackstead a,∗ , W.B. Yelon b , M.P. Smylie a a
Physics Department, University of Notre Dame, Notre Dame, IN 46556, USA
b
Missouri University of Science and Technology, Materials Research Center and Department of Chemistry, Rolla, MO 65401, USA
article
info
Article history: Received 4 June 2009 Accepted 12 June 2009 by F. De la Cruz Available online 23 June 2009 PACS: 74.20.Mn 74.25.Ha 74.70Pq 76.50+g
abstract Gd1.5 Ce0.5 Sr2 Cu2 RuO10 exhibits antiferromagnetic resonance at 23.9 GHz for applied fields less than 1000 Oe with a spin–spin relaxation time T2 of approximately 0.45 ns, and with a spin–lattice relaxation time T1 of at least 320 µs. Since in the homologue, Eu1.5 Ce0.5 Sr2 Cu2 RuO10 , the Ru atoms evidently fail to exhibit magnetic order, the antiferromagnetic resonance must arise from the cuprate planes. In other homologues, the cuprate planes are known to order ferromagnetically and are stacked in an antiferromagnetic configuration. The large value of T1 suggests that phonon mediation plays no role in high temperature superconductivity. In addition, the presence of ferromagnetic cuprate planes is inconsistent with spin-fluctuation models of high temperature superconductivity. © 2009 Elsevier Ltd. All rights reserved.
Keywords: A. Superconductivity A. Antiferromagnetism
1. Introduction The origin of the anomalous magnetization observed in fieldcooled measurements of the superconducting ruthenocuprates such as Gd1.5 Ce0.5 Sr2 Cu2 RuO10 (O10) has been widely attributed to ferromagnetic order, or canted antiferromagnetism, of the Ru in the RuO2 layer [1]. In the closely related O8 compounds such as GdSr2 RuCu2 O8 , the anomalous magnetization has been attributed to a ferromagnetic component on the Ru sites, but several neutron scattering studies have failed to show any evidence for such an origin, reporting, instead, simple antiferromagnetic Ru order. The situation in the O10 compounds is further complicated by the body centered unit cell which leads to a very long c-axis. Fig. 1 shows that antiferromagnetic superexchange interactions of the Ru spins in the O10 compounds are geometrically frustrated by the presence of the fluorite {(Gd, Ce)2 O2 } block which introduces a shift of (a/2, a/2) in the layer stacking. Neutron diffraction measurements by Kuz’micheva et al. [2] failed to detect magnetic order in Nd1.4 Ce0.6 Sr2 Cu2 RuO10 and more recent extensive measurements by Lynn [3] et al. failed to find Ru antiferromagnetic order and also specifically excluded ferromagnetic order of the Ru in
Eu1.5 Ce0.5 Sr2 Cu2 RuO10 in the temperature range of interest here. Since the Y homologue with Nb replacing Ru also exhibits similar magnetic behavior [4], and closely related YSr2 Cu2 NbO8 does as well [5], attention is necessarily focused on the cuprate (CuO2 ) planes. Neutron diffraction studies [6,7] have shown these to be magnetically ordered in the closely related material YSr2 Cu2 RuO8 . As it happens, the magnon energy gap of the ordered Cu spins in the ruthenocuprates is sufficiently small that antiferromagnetic resonance is easily detected at conventional microwave frequencies. In some cases, antiferromagnetic resonance can also be observed in cuprates [8]. In the following, we present the results of such measurements as a function of the rf power level at a temperature of 100 K, substantially above the ∼40 K superconducting transition temperature. These data show that the cuprate plane antiferromagnetic mode is surprisingly easy to saturate, indicating a long spin–lattice relaxation time, T1 . In addition, the spin–spin relaxation time T2 is quite short, leading to a broad fielddependent response. The Gd spins exhibit paramagnetic resonance which shows no saturation effects in the range of rf power levels employed. The Gd resonance is superimposed on the broad Cu response, and is used to calibrate the intensity of Cu resonance. 2. Experimental details
∗ Corresponding address: Physics Department, University of Notre Dame, 225 Nieuland Science Hall, 46556 Notre Dame, IN, USA. Tel.: +574 631 7078. E-mail address: [email protected] (H.A. Blackstead). 0038-1098/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ssc.2009.06.020
The procedure for sample preparation is described elsewhere [9], and follows conventional solid state reaction procedures. The microwave spectrometer employed has novel features
H.A. Blackstead et al. / Solid State Communications 149 (2009) 1702–1705
1703
Fig. 1. Schematic structure of Gd1.5 Ce0.5 Sr2 Cu2 RuO10 , here 1/2 of the unit cell is shown. Replacement of Gd in GdSr2 Cu2 RuO8 with the fluorite block (Gd, Ce)2 O2 results in a similar structure in which the antiferromagnetic Ru superexchange is frustrated, and it fails to exhibit long-ranged order. The Cu order ferromagnetically and are stacked antiferromagnetically. The Cu superexchange interactions across the RuO2 and SrO layers are not frustrated for this configuration.
which were used to advantage in these measurements. One of these features is that the spectrometer (see Fig. 2) does not utilize magnetic field modulation. Field modulation is employed with a lock-in amplifier to extract the derivative of the lineshape as a function of the applied field. For resonances with large linewidths, very small signals would result. Since we did not use this technique, it was possible to measure signals with very large linewidths. In addition, the power reflected from the samplebearing cavity was amplified using a low-noise solid state amplifier with a gain of nearly 30 db. An attenuator following the klystron source was used to control the power level incident on the cavity. Another attenuator placed after a directional coupler, and before the amplifier, was adjusted to compensate for changes in incident power levels. Thus, the point-contact diode detector, except for the lowest power levels utilized, was driven by a nearly constant power level. This is of some importance, since such detectors have a square-law response, and a constant power level biases them to a reproducible sensitivity. Additional circuitry (not shown) was employed to ‘‘lock’’ the klystron frequency to the resonant frequency of the sample-bearing TE101 rectangular cavity [10]. This feature ensures that only changes in χ ’’ will be detected. The polycrystalline sample was mounted in the center of the bottom of the cavity, using a small quantity of silicone grease. For the data reported here, the dc magnetic field H was applied perpendicular to the very uniform cavity magnetic field (Hrf ), in the usual configuration for magnetic resonance. Similar Cu signals were detected with H ||Hrf , but in that case the paramagnetic Gd resonance was very small, and it was not useful as a calibration signal. 3. Analysis of the data The resonant response (aside from an amplitude including Hrf2 ) was characterized using a form given by Dyson [11,12] for conducting materials, including the term in the denominator which involves the rf field intensity and leads to saturation at high rf power levels: y (H ) =
1 + αγ T2 (H − H0 ) 1+γ
2T 2 2
(H − H0 )2 + γ 2 T1 T2 Hrf2
.
(1)
Here γ is the gyromagnetic ratio given by γ = (g µB )/h, T1 and T2 are the spin–lattice and spin–spin relaxation times, respectively,
Fig. 2. Simplified block diagram of the K-band microwave spectrometer. The directional coupler serves to separate the incident power from the power reflected from the resonant cavity. The power level was set with the attenuator in series with the klystron source, and an approximately constant power level was provided to the amplifier by adjusting the compensating attenuator following the directional coupler.
and α = 0.63 is a parameter of the fit to the data expressing the line-asymmetry. The signal resulting from an experiment is proportional to Hrf2 y (H ); saturation of the response occurs for large rf fields. At resonance, the lineshape y(H = H0 ) is: y ( H = H0 ) =
1 1+γ
2T
2 1 T2 Hrf
.
(2)
Note that y(H ) is maximum for H > H0 , a consequence of the asymmetry of the Dysonian lineshape function. Because the Gd relaxation times are both small, the Gd paramagnetic response does not saturate. It follows that if γ 2 T1 T2 Hrf2 > 1, the signal intensity, normalized by the Gd signal, will be substantially reduced at high power levels [13]. Usually in metals, this condition is difficult to satisfy owing to the small sizes of both T1 and T2 . For sufficiently low power levels the resonance signal is not a function of T1 , and T2 is independently determined by the linewidth, see Figs. 3 and 4. As −db
a function of attenuator setting (in db), Hrf = Hrf 0 10 20 . The product T1 Hrf2 0 was found by fitting signal intensities as a function of this parameter, see Fig. 5. An upper limit for Hrf 0 was determined from the cavity radiation quality factor ‘‘Q ’’, the input rf power level, and the cavity and waveguide dimensions [14,15]. The result of this evaluation is that Hrf 0 ≤ 2.0 Oe. It follows from the results shown in Fig. 5 and the estimated value for Hrf that T1 ≥ 320 µs, while from the linewidth, T2 was found to be ∼0.45 ns. 4. Discussion We have recently shown in a sequence of materials which contain Ru4+ or Ru5+ in octahedral coordination with six oxygen, that the Ru are esr silent in both paramagnetic and magnetically ordered configurations [16]. These results exclude Ru as the source of the resonance discussed here. The results of the recent neutron diffraction experiments on two materials, Eu1.5 Ce0.5 Sr2 Cu2 RuO10 and Y1.5 Ce0.5 Sr2 Cu2 NbO10 are of critical importance to the interpretation of the results of the experiments presented here.
1704
H.A. Blackstead et al. / Solid State Communications 149 (2009) 1702–1705
Fig. 3. Resonance signal with 50 db attenuation at 100 K (black dots). In addition the fits using the summed resonance response functions for the Cu and Gd spins are shown (red line). The Gd response (green line) peaks near µ0 H = 0.9 T, while the Cu response (blue line) peaks near 0.08 T. The Cu spin–spin relaxation time T2 is approximately 0.45 ns. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Lynn et al., using low absorption cross-section 153 Eu, found no evidence for Ru magnetic ordering in Eu1.5 Ce0.5 Sr2 Cu2 RuO10 . Magnetic measurements [4] on Y1.5 Ce0.5 Sr2 Cu2 NbO10 , a material
which has neither a strong neutron absorber such as Gd or Eu, nor magnetic Ru, show magnetic order, which necessarily must be attributed to the cuprate planes. Magnetic resonance studies [4] of Y1.5 Ce0.5 Sr2 Cu2 NbO10 find behavior similar to that reported here, at a lower temperature. Thus, even if the RuO2 layer orders [17] (perhaps two-dimensionally), it has no impact on our results. We have concluded that, in all of these ruthenocuprates, the cuprate planes are magnetically ordered in an antiferromagnetic configuration in which ferromagnetic planes are stacked with alternating magnetization directions. This configuration is almost invisible to neutron diffraction, because the magnetic order only enhances nuclear reflections, without leading to additional reflections. Because the moment is small, (∼0.4 µB ) the Cu contribution to these peaks is small. The magnetic resonance data are unambiguous that the magnetic ordering is antiferromagnetic [18] while the system exhibits pronounced metamagnetism which has been widely misinterpreted as ferromagnetism. This metamagnetism most likely results from the relative rotation of the magnetic layers as driven by an applied magnetic field. These results have strong implications for the origin of superconductivity of the cuprates. If there is one pairing mechanism, then it seems very unlikely that it has anything to do with either magnetic spin fluctuations or phonons. This conclusion is a direct consequence of the magnetically ordered cuprate planes and the very weak Cu spin–lattice coupling these measurements find. In addition, the relatively small size of T2 indicates that the Cu spins are strongly coupled. In a broader context, superconductivity in the ruthenocuprates seems to occur in systems in which there are ferromagnetic layers separated by either metallic or oxide layers in an antiferromagnetic configuration. The presence of Ru, either antiferromagnetically ordered, as in GdSr2 Cu2 RuO8 , or as in Eu1.5 Ce0.5 Sr2 Cu2 RuO10 , has little impact on the superconducting transition temperature, but does raise the magnetic transition temperature of the cuprate planes in both cases, as compared to Nb.
Fig. 4. Microwave resonance signals as a function of input power. The Cu antiferromagnetic resonance peaks in intensity at low applied fields, near µ0 H = 0.085 T, while the Gd paramagnetic resonance is seen near µ0 H = 0.9 T. The Gd resonance was used to normalize the several signals. The signal for 50 db attenuation has a noticeable noise component.
H.A. Blackstead et al. / Solid State Communications 149 (2009) 1702–1705
1705
Acknowledgements H. A. B. and W. B. Y. thank the U. S. Department of Energy (MISCON Grant No. DE-FG0290ER45427) for generous past support, and thank I. Felner for providing the sample material and D. B. Pulling and P. J. Beeli for technical assistance. References
Fig. 5. Cu antiferromagnetic signal level intensity as a function of input power. These data (red dots) are described by Eq. (2) for the intensity at resonance (solid line). The Cu spin–lattice relaxation time T1 was found to be approximately 320 µs. The spin–spin relaxation time was found from the linewidth.
It remains to be determined if this is due to localized Ru magnetism or to the availability of electronic d-states, which are not present in closed shell Nb5+ . Recently, Anderson [19] has suggested that boson mediated pairing may not be needed for high temperature superconductivity. While this is a paradigm-changing proposal, our findings, which pose severe limits for both spin-fluctuation and phonon pairing schemes, are consistent with such a scenario.
[1] I. Felner, U. Asaf, Y. Levi, O. Millo, Phys. Rev. B 55 (1997) R3374. [2] G.M. Kuz‘micheva, A.I. Kubakov, I.E. Kostyleva, S.A. Lachenkov, Zh. Neorg. Khim 51 (2006) 1949 (Zhuranal Neorganicheskoi Khimii 51 (2006) 2070). [3] J.W. Lynn, Y. Chen, Q. Huang, S.K. Goh, G.V.M. Williams, J.L. Tallon, Phys. Rev. B 76 (2007) 014519. [4] H.A. Blackstead, W.B. Yelon, M.P. Smylie, Q. Cai, J. Lamsal, V.P.S. Awana, S. Balamurugan, E. Takayama-Muromachi, (unpublished). [5] H.A. Blackstead, W.B. Yelon, M. Kornecki, M.P. Smylie, Q. Cai, J. Lamsal, V.P.S. Awana, H. Kishan, S. Balamurugan, E. Takayama-Muromachi, Phys. Rev. B 75 (2007) 140514(R). [6] H.A. Blackstead, W.B. Yelon, M. Kornecki, M.P. Smylie, Q. Cai, J. Lamsal, V.P.S. Awana, H. Kishan, S. Balamurugan, E. Takayama-Muromachi, Phys. Rev. B 76 (2007) 094507. [7] W.B. Yelon, Q. Cai, Jagat Lamsal, H.A. Blackstead, M. Kornecki, V.P.S. Awana, H. Kishan, S. Balamurugan, E. Takayama-Muromachi, J. Appl. Phys. 101 (2007) 09G104. [8] H.A. Blackstead, J.D. Dow, M. Osada, M. Kakihana, Phys. Rev. B 76 (2007) 024522. [9] I. Felner, U. Yaron, U. Asaf, T. Kröner, V. Breit, Phys. Rev. B 49 (1994) 6903. [10] C.P. Poole Jr, Electron Spin Resonance, Interscience Publishers, 1967, 264. [11] F.J. Dyson, Phys. Rev. 98 (1955) 349. [12] C.P. Poole Jr, Electron Spin Resonance, Interscience Publishers, 1967, 814. [13] If the term in the denominator of the line-shape function which varies as Hrf2
[14] [15] [16]
[17] [18] [19]
could be neglected, the overall signal intensity would increase with Hrf2 as the Gd response does. C.P. Poole Jr, Electron Spin Resonance, Interscience Publishers, 1967, 302. The loaded Q (Ql ) was found to be ∼1500, and the maximum power input to the cavity was ∼250 mW. H.A. Blackstead, W.B. Yelon, M. Kornecki, M.P. Smylie, P.J. McGinn, Q. Cai, B.W. Benapfl, S.D. Knust, J Mag. Mag. Mat. (in press) http://dx.doi.org/10.1016/ j.jmmm.2009.05.048. A.C. Mclaughlin, I. Felner, V.P.S. Awana, Phys. Rev. B 78 (2008) 094501. These authors claim that the Ru exhibit long-range magnetic order near 160 K. E.A. Turov, Physical Properties of Magnetically Ordered Crystals, Academic Press, 1965, 151. P.W. Anderson, Science 316 (2007) 1705.