# Enhancement of low-frequency fluctuations and superconductivity breakdown in Mn-doped LaYFeAsOF superconductors

## Abstract

F NMR measurements in optimally electron-doped LaYFeMnAsOF superconductors are presented. In these materials the effect of Mn doping on the superconducting phase is studied for two series of compounds ( and ) where the chemical pressure is varied by substituting La with Y. In the series superconductivity is suppressed for Mn contents an order of magnitude larger than for the series. For both series a peak in the F NMR nuclear spin-lattice relaxation rate emerges upon Mn doping and gets significantly enhanced on approaching the quantum phase transition between the superconducting and magnetic phases. F NMR linewidth measurements show that for similar Mn contents magnetic correlations are more pronounced in the series, at variance with what one would expect for spin correlations. These observations suggest that Mn doping tends to reduce fluctuations at and to enhance other low-frequency modes. The effect of this transfer of spectral weight on the superconducting pairing is discussed along with the charge localization induced by Mn.

###### pacs:

74.70.Xa, 76.60.-k, 76.75.+i, 74.40.KbThe introduction of impurities in superconducting materials is a well known approach to test their stability for future technological applications as well as to unravel their intrinsic microscopic properties. In the cuprates the study of the staggered spin configuration around isolated spin-less impurities, as Zn, has allowed to determine how the electronic correlations evolve throughout the phase diagram.(1) When a sizeable amount of impurities is introduced they can no longer be treated as independent local perturbations, the correlation among the impurities themselves has to be considered and quantum transitions to new phases may arise.(2)

In iron-based superconductors several studies on the effect of impurities have been carried out and it has soon emerged that the behaviour may vary a lot depending on the family considered. If one concentrates on the LnFeAsOF (Ln1111) family, with Ln a lanthanide ion, one notices that diamagnetic impurities introduced by substituting Fe by Ru, cause a very weak effect both on the magnetic ()(3); (4); (5); (6) and on the superconducting ()(7); (9); (8); (10); (11) ground-state. One has to introduce almost 60% of Ru to quench either one of the two phases. On the other hand, if one considers the effect of paramagnetic impurities as Mn a much stronger effect is observed. In fact, in the optimally doped () Sm1111 the superconducting transition temperature vanishes for a Mn content around 8% (see Fig. 1). Remarkably in optimally electron-doped LaFeMnAsOF superconductors drops to zero for as low as % (12), more than an order of magnitude less than for Sm1111, and a quantum phase transition to a magnetic ground-state is observed.(13) The different behaviour of Sm1111 and La1111 against Mn impurities shows that by decreasing the lanthanide ion size decreases more slowly with and the system is driven away from a quantum critical point (QCP).

The nature of the magnetic ground-state developing at high Mn contents is still controversial.(14) In BaK(FeMnAs) superconductors neutron scattering results suggested that Mn could modify the magnetic wave-vector from to (square lattice unit cell with Fe ions at the vertexes),(15) leading to a weakening of wave pairing.(16) However, the absence of Ln1111 single crystals with a size appropriate for neutron scattering experiments makes the determination of the magnetic correlations developing upon Mn doping rather difficult for this family of superconductors.

Since in Fe-based superconductors one of the most likely pairing mechanisms involves magnetic excitations,(17) it is of major importance to investigate how the spin excitations evolve in optimally doped LaFeMnAsOF superconductors as the QCP is approached. From As nuclear spin-lattice relaxation rate measurements it was observed(13) that when vanishes for the spin correlations follow the behaviour predicted for strongly correlated electron systems close to a two-dimensional (2D) antiferromagnetic (AF) QCP.(18) In this manuscript it is shown that upon increasing the chemical pressure, by partially substituting Y for La, decreases more slowly with (Fig.1), mimicking the effect observed for Sm1111. This indicates that the different behaviour of Sm1111 and La1111 against Mn impurities has to be associated with the larger chemical pressure induced by the lanthanide ions on the FeAs planes in the former case. From F NMR measurements it is shown that besides the high frequency ( s) dynamic probed by As nuclei, an additional very low-frequency (MHz range) dynamic develops upon Mn doping and gets progressively enhanced as the QCP is approached. Furthermore, it is evidenced that if the system is driven away from the QCP by partially substituting La with Y these low-frequency dynamic gets significantly enhanced only at high Mn contents where . These results evidence that the disruption of the superconducting phase coincides with the enhancement of low-frequency fluctuations possibly competing with the ones driving superconductivity.

NMR experiments were performed on (La,Y)FeMnAsOF polycrystalline samples. Y for La substitution allows to vary the chemical pressure without introducing paramagnetic lanthanide ions which would significantly affect F NMR .(19) Two series of samples were studied, the first one with no Y and with Mn contents of 0%, 0.025%, 0.075%, 0.1%, 0.2% (referred to as LaY0), while the second one with 20% of Yttrium (LaY20 hereafter) and Mn content 0%, 0.3%, 0.5%, 10%, 20%. LaY0 samples were prepared as described in Ref. (12), while LaY20 as described in (20). All the samples were optimally electron doped with fluorine content around 11% .(21) was determined by means of Superconducting Quantum Interference Device (SQUID) zero field-cooled magnetization measurements in a 10 Oe magnetic field. The diagram of the superconducting phase, for both series of samples, as a function of Mn content is shown in Fig. 1. It is evident that the introduction of 20% of Yttrium stabilizes the superconducting phase, leading to an increase of for (22) as well as to a marked increase of from 0.2% to about 4.5%.

F NMR measurements were performed at low magnetic fields, Tesla, by using standard radiofrequency pulse sequences. The spin-lattice relaxation rate was estimated by following the recovery of nuclear magnetization after a saturation recovery sequence. The recovery was fit according to

(1) |

with M the magnetization at equilibrium. The factor is introduced to account for incomplete saturation and is a stretching exponent which indicates a distribution of . The stretching exponent was found to be 1 for K and decreased to about 0.5 at low temperature. The distribution of relaxation rates originates from the presence of different inequivalent Mn impurity configurations around F nuclei.

The temperature dependence of for both sample series is shown in Fig. 2. Below 70 K F NMR is characterized by a progressive increase upon decreasing the temperature, by a pronounced maximum around 20 K, which can be either below or above depending on the Y and Mn content (see Fig.1) and eventually by a decrease at low temperature. Since, in all samples, the increase starts well above those peaks should not be associated with dynamics which develop in the superconducting phase (e.g. vortex motions)(23) but to normal state low-energy excitations. It should be remarked that in the normal phase of Ln1111 iron-based superconductors without impurities no marked peak in have ever been reported. Only peaks in have been observed,(24) corresponding to small bumps in . Here it is noticed that those peaks are significantly enhanced by the presence of impurities suggesting that Mn tends to strengthen those low-frequency dynamics which might already be present in the pure compounds (see Fig. 2).

By performing measurements at different magnetic fields one observes that while the high temperature behaviour is only weakly field dependent the magnitude of the peak around 20 K grows by lowering the magnetic field (Fig. 3). This is exactly the behaviour expected in the presence of dynamics approaching the nuclear Larmor frequency , namely in the MHz range. If one assumes an exponential decay for the correlation function describing the fluctuations with a characteristic time , then one can write(25)

(2) |

where is the nuclear gyromagnetic ratio and the mean square amplitude of the local field fluctuations perpendicular to . In several disordered systems, including cuprates,(26) the temperature dependence of the correlation time is well accounted for by an Arrhenius law with an energy barrier and the correlation time at infinite temperature. Nevertheless, a monodispersive behaviour cannot suitably describe the broad peaks in and one rather has to consider a distribution of correlation times associated with the non-uniform distribution of Mn impurities. This corresponds to a distribution of energy barriers which, for simplicity, was taken as squared with a width centered around .(28) In order to account for the high temperature behaviour of a linear Korringa term ,(27) characteristic of metallic systems was introduced (see Fig. 2). Then can be described by the expression:(28)

(3) |

By fitting the data of the superconducting samples (% for LaY0 and % for LaY20) one notices that for LaY0 spin correlations yield a significant increase in the width of the distribution on approaching the crossover between the superconducting and magnetic phases (Fig. 4). On the other hand, for the LaY20, within the uncertainty of the fit parameters, there is no evidence for a neat increase of in the same doping range (Fig. 4). In other terms, in the LaY20 family the dynamic does not vary significantly upon increasing the Mn content up to %, suggesting that the collective coupling is still weak.

Moreover, one may notice (Fig. 2) that for %, in the LaY0 series the peak in grows significantly with Mn doping, while in the LaY20 series it remains practically unchanged. This evidences that increases progressively as , namely the strength of the local spin susceptibility in the FeAs plane gets enhanced due to the proximity to the QCP. In other terms, for similar Mn contents the spin correlations get weaker as the chemical pressure is increased by Y doping. Further support in this respect is provided by the temperature dependence of F NMR linewidth which is directly related to the amplitude of the staggered magnetization developing around the impurity.(29) As shown in Fig. 5 for similar Mn doping the F NMR linewidth is unambiguously larger in the sample without Y. The data in Fig. 5 can be fitted with a Curie-Weiss law . kHz is the temperature-independent linewidth due to nuclear dipole-dipole interaction which is assumed similar for both samples, since it is determined by F-F dipolar coupling which remains practically unchanged (the F content and the lattice parameters do not vary significantly between the two samples). The fit of the data shows that increases by a factor 3 and that increases from about 3 K to 11 K between % LaY20 and % LaY0 sample.

The observation that the magnetic correlations get depressed when La is substituted by a smaller lanthanide ion can in principle be associated with a decrease of the ratio between Coulomb repulsion and hopping integral due to the increase in the chemical pressure. However, for stripe collinear order ( or ) theoretical works(30); (31) suggest that in Ln1111 the magnetic order parameter should get enhanced on decreasing the Ln size or equivalently increasing As coordinate, exactly the opposite of what is found here. It should also be remarked that the behaviour found upon Mn-doping is the contrary of that observed in Ru-substituted Fe-based superconductors where the magnetic order is stabilized by decreasing the size of the lanthanide.(11) Hence, it is likely that upon increasing magnetic correlations different from the stripe ones develop. Giovannetti et al.,(31), in the framework of Landau free energy calculations, showed that around optimal electron doping the energy difference between the stripe and orthomagnetic phases, with a rotation of the spins, gets reduced. Hence, it might be possible that the introduction of Mn impurities could stabilize the latter type of order.

More recently Gastiasoro and Andersen (32) have considered the cooperative behaviour of paramagnetic impurities introduced in the FeAs planes of Fe-based superconductors, coupled via an RKKY interaction. They pointed out that upon increasing the Kondo-like coupling between the localized impurity and the itinerant electrons, Néel () correlations would arise and the amplitude of collinear stripe modes decrease. However, even when the coupling gets significant and Néel fluctuations enhanced the stripe spin correlations would still survive. In a real space description their results imply the development of Néel type correlations in small islands around the impurity and stripe spin arrangement at large distances from the impurity. Even if from our F NMR spectra one cannot check the validity of this model, this theoretical approach is able to explain both the weakening of the superconductivity (16) and the onset of a novel magnetic phase upon Mn doping.(13) In such a scenario the peaks in should be associated with the freezing of the spin fluctuations around Mn impurities which get more and more correlated as the QCP is approached. The theoretical model by Gastiasoro and Andersen (32) also allows to make an analogy between heavy fermion physics and the one achieved by doping Fe-based superconductors impurities. In this respect we recall that, similarly to heavy fermions, at the QCP there is a charge localization (12) suggesting a divergence of the electron effective mass. Hence, one should actually consider two possible concomitant effects which depress superconductivity: loss of spin excitations causing the pairing and/or charge localization. Once more, we remark that the behaviour achieved by Mn-doping is quite different from the one observed in Ln1111 superconductors doped with Ru spin-less impurities, where even at very high doping levels (%) the system remains metallic.(33)

An alternative explanation for the growth of low-frequancy spin fluctuations in Mn-doped Ln1111 relies on the presence of nematic fluctuations. In this respect it is interesting to observe that even nominally pure samples do show a small bump in (24); (34); (35) in the same temperature range where the peak in F NMR arises. The same low-frequency dynamic was found to affect the NMR transverse relaxation rate in Ba(FeRh)As and was tentatively associated with nematic fluctuations, possibly involving charge stripes.(34); (35) Although there is no neat evidence for these type of dynamics here, one may be tempted to relate the energy barrier probed by to the one separating the degenerate nematic phases.(36) In this framework the enhancement of the low-frequency dynamics could be associated with the pinning of those fluctuations driven by Mn.

In conclusion, the increase in the chemical pressure driven by Y for La substitution in LaYFeMnAsOF is found to lead to a less effective suppression of the superconducting ground-state by Mn doping. F NMR measurements exhibit a low-temperature peak which indicates the onset of very low-frequency dynamics with an amplitude directly related to the proximity of the compound to the QCP between superconducting and magnetic phases. Based on recent theoretical works, this behaviour could be ascribed to the enhancement of spin correlations different from stripe ones, suggesting that is depressed by the decrease in the spin fluctuations around , which are widely believed to mediate the pairing, or by the localization effect in the region close to the metal-insulator boundary.

We would like to acknowledge useful discussion with Brian Andersen, Maria N. Gastiasoro and J. Lorenzana. This work was supported by MIUR-PRIN2012 project No. 2012X3YFZ2, by the DFG through the SPP1458 in project BU887/15-1 and SFB1143 and by the Emmy-Noether program (Grant No. WU595/3-1). We acknowledge R. Wachtel, S. Müller-Litvani, and G. Kreutzer for technical support.

### References

- H. Alloul, J. Bobroff, M. Gabay, and P. J. Hirschfeld, Rev. Mod. Phys. 81, 45 (2009).
- J. S. Parker, D. E. Read, A. Kumar and P. Xiong, Europhys. Lett. 75, 950 (2006).
- M. A. McGuire, D. J. Singh, A. S. Sefat, B. C. Sales, and D. Mandrus, J. Solid State Chem. 182, 2326 (2009).
- Y. Yiu, V. O. Garlea, M. A. McGuire, A. Huq, D. Mandrus, and S. E. Nagler, Phys. Rev. B 86, 054111 (2012).
- Y. Yiu, P. Bonfá, S. Sanna, R. De Renzi, P. Carretta, M.A. McGuire, A. Huq, and S.E. Nagler, Phys. Rev. B 90, 064515 (2014).
- P. Bonfá, P. Carretta, S. Sanna, G. Lamura, G. Prando, A. Martinelli, A. Palenzona, M. Tropeano, M. Putti, and R. De Renzi, Phys. Rev. B 85, 054518 (2012).
- M. Sato and Y. Kobayashi, Solid State Commun. 152, 688 (2012).
- E. Satomi, S. C. Lee, Y. Kobayashi, and M. Sato, Journal of the Physical Society of Japan 79, 094702 (2010).
- S.C. Lee, E.Satomi, Y. Kobayashi and M. Sato, J.Phys.Soc. Jpn. 79, 023702 (2010)
- S. Sanna, P. Carretta, P. Bonfá, G. Prando, G. Allodi, R. De Renzi, T. Shiroka, G. Lamura, A. Martinelli, andM. Putti, Phys. Rev. Lett. 107, 227003 (2011).
- S. Sanna, P. Carretta, R. De Renzi, G. Prando, P. Bonfà, M. Mazzani, G. Lamura, T. Shiroka, Y. Kobayashi, and M. Sato, Phys. Rev. B 87, 134518 (2013).
- M. Sato, Y. Kobayashi, S. C. Lee, H. Takahashi, E. Satomi, and Y. Miura, J. Phys. Soc. Jpn. 79, 014710 (2010).
- F. Hammerath, P. Bonfá, S. Sanna, G. Prando, R. De Renzi, Y. Kobayashi, M. Sato, and P. Carretta, Phys. Rev. B 89, 134503 (2014).
- Maria N.Gastiasoro and B.M. Andersen, arXiv:1502.05859v1
- G. S. Tucker, D. K. Pratt, M. G. Kim, S. Ran, A. Thaler, G. E. Granroth, K. Marty, W. Tian, J. L. Zarestky, M. D. Lumsden, S. L. Bud’ko, P. C. Canfield, A. Kreyssig, A. I. Goldman, and R. J. McQueeney, Phys. Rev. B 86, 020503(R) (2012).
- R. M. Fernandes and A. J. Millis, Phys. Rev. Lett. 110, 117004 (2013).
- I. I. Mazin, D. J. Singh, M. D. Johannes, and M. H. Du, Phys. Rev. Lett. 101, 057003 (2008); K. Kuroki, S. Onari, R. Arita, H. Usui, Y. Tanaka, H. Kontani, and H. Aoki, ibid. 101, 087004 (2008); J. Zhang, R. Sknepnek, R. M. Fernandes, and J. Schmalian, Phys. Rev. B 79, 220502(R) (2009).
- A. Ishigaki and T. T.Moriya, J. Phys. Soc. Jpn. 65, 3402 (1996).
- G. Prando, P. Carretta, A. Rigamonti, S. Sanna, A. Palenzona, M. Putti, and M. Tropeano Phys. Rev. B 81, 100508(R) (2010)
- See Supplemental Material for the description of the synthesis and characterization of LaY20 samples.
- For the LaY0 samples no variation of the intensity of the F-NMR resonance line was found within the error bars, confirming that the intrinsic F content does not differ among the samples within .
- M. Tropeano, C. Fanciulli, F. Canepa, M. R. Cimberle, C. Ferdeghini, G. Lamura, A. Martinelli, M. Putti, M. Vignolo, and A. Palenzona Phys. Rev. B 79, 174523 (2009)
- L. Bossoni, P. Carretta, A. Thaler, and P. C. Canfield, Phys. Rev. B 85, 104525 (2012); A. Rigamonti, F. Borsa and P. Carretta, Rep. Prog. Phys. 61, 1367 (1998)
- F. Hammerath, U. Gräfe, T. Kühne, H. Kühne, P. L. Kuhns, A. P. Reyes, G. Lang, S. Wurmehl, B. Büchner, P. Carretta, and H.-J. Grafe, Phys. Rev. B 88, 104503 (2013).
- N. Bloembergen, E. M. Purcell, and R. V. Pound, Phys. Rev. 73, 679 (1948).
- M.-H. Julien, F. Borsa, P. Carretta, M. Horvatić, C. Berthier, and C. T. Lin Phys. Rev. Lett. 83, 604 (1999)
- C.P. Slichter in Principles of Magnetic Resonance 3 Edition, Springer, Berlin (1990)
- M. Filibian and P. Carretta, Phys. Rev. B 75, 085107 (2007).
- D. LeBoeuf, Y. Texier, M. Boselli, A. Forget, D. Colson, and J. Bobroff, Phys. Rev. B 89, 035114 (2014).
- S. Sharma, S. Shallcross, J. K. Dewhurst, A. Sanna, C. Bersier, S. Massidda, and E. K. U. Gross Phys. Rev. B 80, 184502 (2009).
- G. Giovannetti, C. Ortix, M.Marsman, M.Capone, J. van den Brink, and J. Lorenzana, Nat. Commun. 2, 398 (2011).
- M.N. Gastiasoro and B.M. Andersen, Phys. Rev. Lett. 113, 067002 (2014).
- M. Tropeano, M. R. Cimberle, C. Ferdeghini, G. Lamura, A. Martinelli, A. Palenzona, I. Pallecchi, A. Sala, I. Sheikin, F. Bernardini, M. Monni, S. Massidda, and M. Putti, Phys. Rev. B 81, 184504 (2010)
- L. Bossoni, P. Carretta, W. P. Halperin, S. Oh, A. Reyes, P. Kuhns, and P. C. Canfield, Phys. Rev. B 88, 100503 (2013).
- A. P. Dioguardi, M. M. Lawson, B. T. Bush, J. Crocker, K. R. Shirer, D. M. Nisson, T. Kissikov, S. Ran, S. L. Bud’ko, P. C. Canfield, S. Yuan, P. L. Kuhns, A. P. Reyes, H.-J. Grafe, N. J. Curro, arXiv:1503.01844
- R.M. Fernandes and J. Scmalian, Supercond. Sci. Technol. 25 084005(2012).