This action might not be possible to undo. Are you sure you want to continue?
L. Komendov´ ∗ , A. A. Shanenko, M. V. Miloˇevi´ and F. M. Peeters a s c
Departement Fysica, Universiteit Antwerpen, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium
Abstract We study the vortex proﬁles in two-gap superconductors using the extended Ginzburg-Landau theory. The results shed more light on the disparity between the eﬀective length scales in two bands. We compare the behavior expected from the standard Ginzburg-Landau theory with this new approach, and ﬁnd good qualitative agreement in the case of LiFeAs.
Key words: Two-gap superconductivity, Ginzburg-Landau theory, vortex PACS: 74.20.De, 74.70.Ad, 74.70.Xa
1. Introduction Since the ﬁrst experimental observation of the two-gap features in the superconducting magnesium diboride many researchers have tried to generalize the existing single-gap Ginzburg-Landau (GL) and microscopic (Bogoliubov-de Gennes and Gor’kov) theories to the new multi-gap materials . In the case of GL theory this was usually done via the energy functional which incorporates standard (singlegap) contributions from two order parameters, coupled together by the so-called Josephson coupling term proportional to the Josephson coupling constant γ. If no Josephson coupling is present (γ = 0), the system possesses U (1) × U (1) symmetry, clearly has two diﬀerent coherence lengths, but also two different critical temperatures . The latter contradicts the experimental observation of single critical temperature Tc in multiband materials, at which superconductivity ceases in all bands. Thus, the GL model without Josephson coupling is clearly not sufﬁcient to explain experimental data on two-gap su∗ Corresponding author. Preprint submitted to Elsevier
perconductors. The introduction of the Josephson coupling term reduces the symmetry to U (1), gives a single Tc , and can be justiﬁed from microscopic considerations . However, it also leads to the appearance of incomplete higher-order terms in the expansion parameter τ = 1 − T /Tc . In the T → Tc limit, one can omit latter terms, which results in reducing the two-gap theory to its single-gap equivalent - and a single coherence length for a two-gap superconductor . However, at lower temperatures multiple experimental, phenomenological and microscopic studies  point at the existence of two separate coherence length scales even in the coupled system. This discrepancy between above references is currently the topic of a scientiﬁc debate . In order to correctly capture all important physics in the GL domain the extended two-gap GL theory has been developed , where all the terms of the next-to-leading order are collected. Here we show that inclusion of these terms allows for diﬀerences in the characteristic length scales of the orderparameter variation, thus the system possesses again two diﬀerent eﬀective coherence lengths, albeit not entirely independent. We stress that these
21 February 2012
2. with j = 1. Theoretical formalism In this paper we apply the extended GinzburgLandau (EGL) formalism to study a single vortex. (1) and (2) in the cylindrical coordinates using the ansatz for composite vortex with integer winding number n. (7) to avoid calculating . (5) ∗ )2 (1 − τ 1 + n2 A1 /(n1 A2 ) where vj is the Fermi velocity of band j. ..e.e. (0) Similarly. Let us consider two complex order parameters at zero magnetic ﬁeld. namely ∆j (ρ. (1) and (2) is justiﬁed under two conditions. (1) + + (0) ∗(1) βj (∆j )2 ∆j (0) −K∇ where = F (∆j ) + Fj (∆j ).9]. Here A = ln(2eΓ ωD /πTc ). (2) we get for the correction gj (2) ∂ 2 gj 1 ∂gj n2 ˜ ˜ + − 2 gj = (3Ψ2 − 1)gj + F + Fj (8) j ∂ρ2 ρ ∂ρ ρ where [ ( )2 σ S 3 αY ∂Ψj ˜ F = Ψj − (Ψj − Ψj ) + 2 6Ψj + α K K ∂ρ ] ( 2 ) ∂ Ψj 1 ∂Ψj n2 3 2 (9) (3Ψj − 1) + − 2 (Ψj − Ψj ) ∂ρ2 ρ ∂ρ ρ ρj Wj2 3 χj Wj4 5 Zj Wj2 3 Ψ + Ψj + (Ψj − Ψ5 ) j α[ j α K ( )2 6n2 Ψ3 Uj Wj2 ∂Ψj ∂ 2 Ψj j Ψj − + Ψ2 + − j 2 K ∂ρ 5ρ ∂ρ2 ] [ ( )2 Ψ2 ∂Ψj Vj Wj2 ∂Ψj j + − 6Ψj + ρ ∂ρ K ∂ρ ] 2 3Ψ2 ∂Ψj n2 Ψ3 j j 2 ∂ Ψj + 3Ψj + − . The use of Eqs. i. First. Y . + Vj ∇2 (Ψ|Ψ|2 ) + 3 (4) and ˜ Fj = The deﬁnitions of the coeﬃcients α. Vj and Zj are given in Ref. K. . the ‘correction’ must be smaller than the main part of the wave function. We rewrite Eqs. we de(1) (0) mand that bulk solutions satisfy |∆j | ≪ ∆j . By procedure described in detail in Ref. j ∂ρ2 ρ ∂ρ ρ (7) (0) = 0. As a second condition. the indices 1 ↔ 2 in Eq. i. the equilibrium value of ∆j . with gij the coupling constant. it must hold τ τ ∗ with 2 2 v1 /v2 − 1 τ∗ = 1+ (v1 < v2 ). because of the gradient expansion employed during their derivation.  we obtain in the highest order α∆j + βj ∆j |∆j |2 − K∇2 ∆j In the next-to-leading order (1) α∆j (0) (1) 2βj |∆j |2 ∆j 2 (1) ∆j (0) (0) (0) (0) (1) (DOS) and nj N (0) the band-dependent DOS as A1 = λ22 − ηn1 A and A2 = λ11 − ηn2 A. (10) ∂ρ2 ρ ∂ρ ρ2 F (Ψ) = σΨ + S∇2 Ψ + Y ∇2 (∇2 Ψ) and Fj (Ψ) = ρj |Ψ|2 Ψ + χj |Ψ|4 Ψ + (3) Uj [2Ψ|∇Ψ|2 + 5 + 3Ψ∗ (∇Ψ)2 + Ψ2 ∇2 Ψ∗ + 4|Ψ|2 ∇2 Ψ]+ ] Zj [ 2|Ψ|2 ∇2 Ψ + Ψ2 ∇2 Ψ∗ . S. ϕ. The derivation and discussion of the above criterion is given in Ref. βj . 2 indexing two coupled condensates in a two-band superconductor.. with Euler constant Γ= 0.577. χj . i. Uj . we scale √ the distance to ξ = −K/α and the gap functions √ (k=0. z) (0) (6a) (6b) = gj (ρ)e inϕ . ωD is the Debye frequency and η denotes the determinant of the interaction matrix λij . with ρ the radial coordinate and ϕ the radial angle in the cylindric coordinates. we compare the results to the standard GL theory for the case of LiFeAs [8.e. z) = Ψj (ρ)einϕ . Following Ref. Besides the quantiﬁcation of diﬀerent length scales in the extended GL theory. (1) ∆j (ρ. (5) should be exchanged. . we evaluate the order parameters ∆j up to order τ 3/2 by taking (0) (1) (0) ∆j (x) = ∆j (x) + ∆j (x)..are diﬀerent from the formally deﬁned nominal coherence lengths of the two gaps in the absence of coupling. one of the recently discovered iron pnictide superconductors. where ∆j (x) ∝ τ 1/2 and ∆j (x) ∝ τ 3/2 . N (0) the total density of states 2 ˜ ˜ In derivation of the above expressions for F and Fj we already made use of Eq. from Eq. The parameters A1 and A2 are determined by the interaction matrix λij = N (0)gij . Additionally. ρj . If v1 ≥ v2 .1) ∆j to Wj = −α/βj . σ. Equation (1) then gives 1 ∂Ψj n2 ∂ 2 Ψj + − 2 Ψj = Ψ3 − Ψj . ϕ. a point singularity around which the phases of order parameters change by 2πn where n is an integer number.
384 and v1 /v2 = 0. The results shown were calculated for a single n = 1 vortex using the parameters appropriate for LiFeAs. λ22 = 0. Where this is the case.9Tc . (11) λ2 3c2 W 2 j=1.03. Most importantly. The intention is to compare the characteristic features of the extended GL results with the analysis within the standard GL theory from Ref. In Fig. Note that the information about the magnetic ﬁeld and thus about the penetration depth is absent in the present EGL formalism. i.7×1034 states/erg/cm3 (determined from N (0) value from Ref. 1(A) we show the calculated ratio of healing lengths ξ1 /ξ2 as a function of Fermi velocity and partial density of states ratios. (A) The ratio of the healing lengths ξ1 /ξ2 for LiFeAs as a function of ratios n1 /n2 and v1 /v2 at ﬁxed temperature T = 0. . according to the extended GL results the disparity of the healing lengths is expected to be quite large in the dome around n1 /n2 ≈ 1. The appropriate boundary conditions for (k) (k) vortex are ∆j (ρ → 0) → 0 and ∂ρ ∆j (ρ → ∞) → 0.384. the amplitude of each condensate disappears in the vortex centre and goes to its equilibrium value far from it. ).  for the case when ∆j are set to zero.5 K from Ref. Unfortunately. as displayed in Fig. λ12 = −0. 1(A) overlaps with region with low λ/ξ values.e. . its minimum value being 1.1 at lower temperatures. However. or type-II behavior. This occasionally means going beyond the GL validity region.2 √ 2 where W = 8π 2 Tc /7ζ(3) is a convenient energy unit. we indicate the border of the GL validity region directly in the plot by dashed line. 2.63. we used the coupling constants λ11 = 0. In Fig. the type-II region for low n1 /n2 ratio is present also in the standard GL calculation. i. which shows the ratios ξ1 /ξ2 and λ/ξ as a function of temperature and v1 /v2 for ﬁxed partial densities of states n1 /n2 = 1. This formula coincides with Eq. about preference for type-I 3 (A) (B) Fig. ) and v1 = 3 × 107 cm/s (estimate based on Ref. In Fig. (B) The estimate of ratio λ/ξ for the same situation. 3 in Ref. . where the nominal GL parameter κ1 has been ﬁxed for the ﬁrst band and the ratio n1 /n2 was only considered to aﬀect the properties of the second band. 3 similar diagrams are constructed as a function of temperature and n1 /n2 for ﬁxed v1 /v2 = 0. .high order derivatives of Ψj in later numerical evaluations. 3. the plots are made in the same parameter ranges as in Ref. Reference  also allows one to determine n1 /n2 = 1.e..642. Similar conclusions hold for the Fig. the region with great disparity of the healing lengths matches very well with the expected position of the non-monotonic intervortex interaction in the standard GL theory. Extended GL analysis of LiFeAs In what follows. Debye temperature TD = 240 K. The dashed red line shows the boundary of the strict validity region of the GL theory. To facilitate this comparison.722. When both κ1 and κ2 are allowed to change. the ratio λ/ξ stays ﬁrmly in type-II values. This discrepancy is due to the diﬀerent calculational procedure employed by Ref. 1. Since the region with the great disparity of the coherence lengths in low v1 /v2 and high n1 /n2 sector of Fig. (5). Despite that it is possible to estimate λ using formula 2 )2 ∑ 16πN (0)e2 nj vj ( (0) 1 (1) ≈ ∆j + ∆ j . the type-I region for low n1 /n2 from the standard GL theory is not reproduced. Surprisingly. in this case the most interesting portion of the parameter space is well below the expected validity region of GL theory. density of states N (0) = 2.722. Speciﬁcally. (43) from (1) Ref. we focus on a particular iron pnictide superconductor. Here. LiFeAs. 1(B) we show the (estimate of) magnetic penetration depth λ divided by the length unit ξ to get some information about the expected value of the GL parameter κ. . we conclude that this is indeed proliferable for competition between the length scales of two bands. as determined by the Eq.  using the cell volume at T = 6.061  (λ12 taken negative to get the s± gap symmetry).
Shanenko. Gurevich. and the ESF-INSTANS network. J. B 69 (2004). Rev.. B 83 (2011). p. 88 (2008). 214514.G. R. (B) Fig. 123712.. the ratio λ/ξ is always larger than 1 in the displayed region.J. p. (A) The ratio of the healing lengths ξ1 /ξ2 for LiFeAs as a function of temperature T /Tc and v1 /v2 ratio at ﬁxed n1 /n2 = 1. Rev.  A. 3.. 174509. Nekrasov. Lett. The dashed red line shows the boundary of the GL validity region. Phys. Vagov et al. The regions of reported nonmonotonic vortex-vortex interaction in standard GL correspond to the regions of large disparity between ξ1 and ξ2 in EGL where estimated GL parameter is close to unity.  H. et al.E. 014502. et al. Ibid. V. Rev. Koshelev and A. Schmalian. B 81 (2010). Rev. p. (B) The estimate of ratio λ/ξ for the same situation. B 84 (2011). p. Phys. et al.R.. Silaev and E. Kim. 538. Rev.after magnetic ﬁeld calculation is included in the EGL formalism . Dao. Physica C 456 (2007). Jpn. p. Phys.  A. (2008). 187003. M. JETP Letters.E. Chaves.. Speight. et al. 89 (2002). Babaev and M. Minimal value of ξ1 /ξ2 inside the GL validity region is about 0. B 83 (2011). 5918. Phys. thus preferring type-II-like behavior. Chem. the Belgian Science Policy (IAP). Wang. 054515. 543. Phys.C. Eskildsen. Commun. arXiv:1105.  V. 054508. et al. Babaev and M. Phys. 100502(R). Golubov. p.3756. 047005. B 83 (2011). 4 . we compared the predictions of the standard and extended GL theory for LiFeAs. et al. Phys.722. Kogan and J. Acknowledgments This work was supported by the Flemish Science Foundation (FWO-Vl)...A. Most importantly. p. Komendov´.  E.A. Soc. 094515..  M. Rev. Phys.5. While reaﬃrming the existence of two diﬀerent length scales for two order parameters. et al. 214523. 148 (2008). Phys. 180502. 064522. (A) 4. Rev. 160. The dashed red line shows the boundary of the GL validity region. Carlstr¨m. J. p..  E. a  L.5090. p. Rev. B 85 (2012).consistently with the present results. (B) The estimate of ratio λ/ξ for the same situation.H.. Lett.  I. et al. Phys. B 83 (2011). arXiv:1105. Rev. 107008. 106 (2011). Silaev. p. Phys.  M.A. Rev. p. Schmalian. et al. 90 (2003). A. Rev. Solid State Commun. Geurts. Lett. p. Rev.  X. (A)  A. More analysis is needed to make an exact comparison . References (B) Fig.  M. Kogan and J. Zhitomirsky and V. Phys. 177002. 78 (2009). B 84 (2011).384. Pitcher. Phys. Babaev. Rev. p. p. o p.. p. p. Nakamura. 2.. B 72 (2005). et al. et al. 92 (2004). p. V. Conclusion We have calculated the vortex core proﬁles in the extended Ginzburg-Landau theory for two gap superconductors. p.  H. (A) The ratio of the healing lengths ξ1 /ξ2 for LiFeAs as a function of temperature T /Tc and n1 /n2 ratio at ﬁxed v1 /v2 = 0.. Phys.G.  A.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.