Professional Documents
Culture Documents
net/publication/225137448
CITATIONS READS
103 63
6 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by R. G. Deminov on 20 May 2014.
In superconducting spin valves with the layer se- fect [9, 10, 11, 13] with TcP > TcAP (i.e., ∆Tc < 0). The
quence F1/S/F2 the superconducting transition tem- most advanced calculations within the proximity effect
perature Tc of the system can be controlled by mutual theory, which take into account the triplet components
alignment of magnetizations M1,2 of the two ferromag- of the superconducting pairing [14], demonstrate only
netic layers F1 and F2. Therefore, at a temperature T the standard switching [3, 15] with Tc monotonically in-
fixed inside the range of Tc variation, there is an oppor- creasing from the P to AP configuration [3]. Additional
tunity for switching the superconductivity on and off physical mechanisms like spin imbalance effect [9, 11] or
by reversing the magnetization direction of the F1 or magnetic domain structure [10, 12] should be recruited
F2 layer. Model calculations have shown that the tran- to explain the inverse spin-valve effect in the studied
sition temperature TcAP for the antiparallel (M1 ↑↓ M2 ) F1/S/F2-type structures.
orientation of the F1 and F2 magnetizations should be A bit earlier an unconventional spin-valve-like
higher than the transition temperature TcP for the op- S/F1/F2 structure was theoretically proposed in [16]
posite case (M1 ↑↑ M2 ) [1, 2, 3]. The situation with to control the superconducting Tc in the S layer by
this order of Tc ’s (i.e., TcP < TcAP ) is commonly referred mutual alignment of the magnetizations of the two
to as the “standard” switching (see, e.g., [4]), and the adjacent ferromagnetic layers F1 and F2. The authors
switching in this case actually occurs at temperatures of [16] argued that TcP < TcAP in their system because
T such that TcP < T < TcAP . The basic physical reason of partial cancelation of the pair-breaking exchange
for the difference ∆Tc = TcAP − TcP > 0 is partial com- fields just within the magnetic F1/F2 subsystem of the
pensation of the pair-breaking ferromagnetic exchange structure, thus predicting the standard switching as in
field, if the magnetizations of the F1 and F2 layers are the interleaved F1/S/F2 structure.
aligned antiparallel.
The S/F1/F2 structures are much less investigated
Several experimental groups have published results experimentally [8, 17], and the experiments indicate the
on superconducting spin valves of the F1/S/F2 type standard switching effect [16] with the maximal size of
[4, 5, 6, 7, 8, 9, 10, 11, 12, 13]. The experimental results about 200 mK. In this Letter we study the critical tem-
turned out to be controversial. Some studies of F1/S/F2 perature of a S/F1/F2 trilayer at arbitrary angle be-
structures have shown the standard spin-valve effect tween the in-plane magnetizations of the ferromagnetic
[4, 5, 6, 7, 8] with the maximum shift ∆Tc ≈ 41 mK layers (see Fig. l). We demonstrate that this struc-
reported for the Ni/Nb/Ni trilayer in [7]. However, ture allows not only the standard but also inverse spin-
some experiments revealed the “inverse” spin-valve ef- switching effect. Moreover, we show for the first time
1
2 Ya. V. Fominov, A. A. Golubov, T. Yu. Karminskaya, et al.
f2 (x) 1 1 0 1 2 3 4 5 6
2 kh dF
1 −1
+C2 1 cosh k̃h x + C3 1 cosh k̃h∗ x +
Fig. 2. Dependence of W (0) − W (π), Eq. (12), on the
thickness dF of the F1 layer. Positive values of this
0 0 oscillating function correspond to stronger suppression
of superconductivity at the P alignment (the standard
1 −1
switching effect), while negative values correspond to
+S2 1 sinh k̃h x + S3 1 sinh k̃h∗ x . (8)
stronger suppression of superconductivity at the AP
0 0 alignment (the inverse switching effect).
0.33
Tc ( )
0.30
Tc( )
dS / = 2.75 dF / F
= 0.73, dS / S
= 2.75
S
0.25
dF / F
= 0.73, dS / S
= 2.69
0.30
dF / F
= 0;
dF / F
= 0.40, dS / S
= 2.75
dF / F
= 0.15
0.20 dF / = 0.40, dS / = 2.66
0.27 F S
dF / = 0.73
Tc / TcS
Tc / TcS
F
0.24
dF / F
= 1.10 2 0.15
1
0.21
0.10
2 2
0.18
1 4
0.05
4
0.15
3 4
0.00
Fig. 3. Critical temperature Tc vs. the misalignment Fig. 4. Tc (α) at various dF and dS . Curve “1” coincides
angle α for various thicknesses of the F1 layer. We took with curve “3” in Fig. 3. Curves “2” and “4” demon-
h/πTcS = 6.8; all other parameters are shown in the strate the reentrant behavior, in which case the triplet
figure. In the cases dF = 0 and dF = ∞, which are spin-valve effect takes place even at T = 0.
physically equivalent (curve “1”), Tc does not depend
on α. Curves “2” and “4” correspond to the standard
and inverse switching effects, respectively. Curve “3”
to a close vicinity of either α = 0 or α = π, becoming
demonstrates the triplet spin-valve effect. The coher-
shallow and indistinguishable.
ence lengths ξS and ξF were taken equal (denoted by ξ
in the text) in order to present our main results in the Figure 4 demonstrates the possibility of reentrant
simplest possible case. At α = 0 all the curves coincide, Tc (α) dependence. In this situation the triplet spin-
since in this case the F part of the system is physically valve effect takes place even at T = 0.
equivalent to a single half-infinite F layer. 4. Discussion. The physical interpretation of the
triplet spin-valve effect can be given as follows: at the
collinear configurations, both the singlet component f0
lation of the second derivatives of W (α) at α = 0 and π and the zero-projection triplet component f3 of the
(the first ones are zero) shows that under the above as- pairing function are short-ranged (with the character-
sumption, both the collinear alignments represent local istic penetration depth of the order of kh−1 ), so that at
minima of W (α). This means that Tc (α) decreases as kh−1 ≪ dF the middle F1 layer plays a role of a shield
the configuration deviates from the P or AP alignment. separating the S layer from the ferromagnetic half-space
Therefore, Tc (α) is nonmonotonic, and the minimal Tc F2. When the angle between magnetizations declines
must be achieved at some noncollinear configuration of from the collinear configurations, the long-range triplet
magnetizations at α 6= 0, π. component f2 of the pairing function is generated [14].
The analytical results obtained at kω ≪ kh are il- Then, the S layer becomes effectively coupled by this
lustrated and extended by numerical calculations at ar- long-range triplet component to the semi-infinite fer-
bitrary relation between kω and kh . Figure 3 shows romagnetic F2 layer. The pair-breaking in the S layer
dependence of the transition temperature Tc on the an- enhances, giving rise to more effective suppression of su-
gle α between the magnetizations. We see that at small perconducting Tc . In other words, we can say that the
thicknesses dF of the middle ferromagnetic layer F1, the Tc suppression is due to “leakage” of Cooper pairs into
switching effect is standard, while at larger dF the ef- the ferromagnetic part. In this language, the generation
fect is inverse (TcP > TcAP ). Moreover, when the F1 layer of the long-range triplet component opens up an addi-
thickness is around a half of the coherence length ξ, the tional channel for this “leakage”, hence Tc is suppressed
minimal critical temperature TcTr at noncollinear orien- stronger.
tations is significantly lower that both TcP and TcAP — In order to supply the qualitative picture by more
this case corresponds to the triplet spin-valve effect. quantitative details, we can find the amplitudes of dif-
Note that depending on the parameters of the system, ferent components at the S/F1 interface in the limit of
the minimum of Tc (α), predicted analytically, can shift kω ≪ kh and large dF . This can be done analytically
Superconducting triplet spin valve 5
from the boundary conditions which produce the linear the RFBR (projects 07-02-00963-a, 09-02-12176-ofi m,
system of equations for the coefficients entering Eqs. (7)- 09-02-12260-ofi m, and 10-02-90014-Bel a), by the
(9). We find that in the limit of dF ≫ kω−1 , kh−1 , the am- NanoNed (project TCS7029), and by the Russian
plitudes of the long-range triplet components near the Federal Agency of Education (contract NK-529P).
S/F1 interface (which are given by C1 , S1 , and B) are
suppressed by the factor e−kω dF −kh dF which has a clear
physical interpretation. The long-range components are 1. L. R. Tagirov, Phys. Rev. Lett. 83, 2058 (1999).
generated from the short-range ones at the F1/F2 inter- 2. A. I. Buzdin, A. V. Vedyayev, and N. V. Ryzhanova, Eu-
face (i.e., at x = dF ), where electrons “feel” inhomoge- rophys. Lett. 48, 686 (1999); A. I. Buzdin, Rev. Mod.
Phys. 77, 935 (2005).
neous magnetization. Therefore, the long-range contri-
3. Ya. V. Fominov, A. A. Golubov, and M. Yu. Kupriyanov,
bution at the S/F1 interface is obtained as a result of
Pis’ma Zh. Eksp. Teor. Fiz. 77, 609 (2003) [JETP Lett.
a “wave” that goes from the S/F1 interface as a short-
77, 510 (2003)].
range component with the wave vector kh and returns
4. I. C. Moraru, W. P. Pratt, Jr., and N. O. Birge, Phys.
after reflection at the F1/F2 interface as a long-range
Rev. B 74, 220507(R) (2006).
component with the wave vector kω . At the same time,
5. J. Y. Gu, C.-Y. You, J. S. Jiang et al., Phys. Rev. Lett.
the self-consistency equation (3) that determines Tc , 89, 267001 (2002).
contains only the singlet short-range component. There- 6. A. Potenza and C. H. Marrows, Phys. Rev. B 71,
fore, the influence of the long-range components on Tc 180503(R) (2005).
is indirect: the long-range components influence Tc only 7. I. C. Moraru, W. P. Pratt, Jr., and N. O. Birge, Phys.
through their influence on the singlet component. We Rev. Lett. 96, 037004 (2006).
find that while the difference between W (that encodes 8. G. Nowak, H. Zabel, K. Westerholt et al., Phys. Rev. B
the information about the suppression of Tc ) for the 78, 134520 (2008).
P and AP cases is suppressed as e−2kh dF (the short- 9. A. Yu. Rusanov, S. Habraken, and J. Aarts, Phys.
range components go from the S/F1 to F1/F2 interface Rev. B 73, 060505(R) (2006).
and back), the changes in W due to noncollinear mag- 10. R. Steiner and P. Ziemann, Phys. Rev. B 74, 094504
netizations contain the same exponential, e−2kh dF . Of (2006).
course, the influence of the long-range triplet compo- 11. A. Singh, C. Sürgers, and H. v. Löhneysen, Phys. Rev. B
nents is contained in prefactors but no long-ranged ex- 75, 024513 (2007).
ponential (with kω instead of kh ) appears in W , because 12. D. H. Kim and T. J. Hwang, Physica C 455, 58 (2007).
W still originates from the short-range components. 13. P. V. Leksin, R. I. Salikhov, I. A. Garifullin et al., Pis’ma
In conclusion, we have considered a mesoscopic Zh. Eksp. Teor. Fiz. 90, 64 (2009) [JETP Lett. 90, 59
S/F1/F2 structure composed of a superconducting layer (2009)].
S, a ferromagnetic layer of arbitrary thickness F1, and 14. F. S. Bergeret, A. F. Volkov, and K. B. Efetov, Phys.
a ferromagnetic half-space F2. We have demonstrated Rev. Lett. 86, 4096 (2001); Rev. Mod. Phys. 77, 1321
that the structure exhibits different relations between (2005).
the critical temperatures in the parallel and antiparal- 15. J. Linder, M. Zareyan, and A. Sudbø, Phys. Rev. B 79,
064514 (2009).
lel configuration: both the standard (TcP < TcAP ) and
inverse (TcP > TcAP ) switching can be realized depend- 16. S. Oh, D. Youm, and M. R. Beasley, Appl. Phys. Lett.
71, 2376 (1997).
ing on the system’s parameters. At the same time, our
17. K. Westerholt, D. Sprungmann, H. Zabel et al., Phys.
main result is that TcTr at noncollinear magnetizations is
Rev. Lett. 95, 097003 (2005).
lower than both TcP and TcAP, which makes this system
18. D. A. Ivanov and Ya. V. Fominov, Phys. Rev. B 73,
a triplet spin valve. Possible experimental observation
214524 (2006).
of a nonmonotonic (like curve “3” in Fig. 3 or curves “1”
19. Ya. V. Fominov, N. M. Chtchelkatchev, and A. A. Gol-
and “3” in Fig. 4) or even reentrant (like curves “2” and
ubov, Pis’ma Zh. Eksp. Teor. Fiz. 74, 101 (2001) [JETP
“4” in Fig. 4) behavior of Tc (α) could be a signature of Lett. 74, 96 (2001)]; Phys. Rev. B 66, 014507 (2002).
existence of the long-range triplet superconducting cor- 20. M. Yu. Kuprianov and V. F. Lukichev, Zh. Eksp. Teor.
relations [14] in SF hybrid structures. Fiz. 94, 139 (1988) [Sov. Phys. JETP 67, 1163 (1988)].
We are grateful to I. A. Garifullin and A. S.
Sidorenko for discussions stimulating this study, to
O. V. Nedopekin for assistance in numerical calcula-
tions, and to M. V. Feigel’man and V. V. Ryazanov for
discussion of the results. This work was supported by