Upload
yuan-zhou
View
213
Download
1
Embed Size (px)
Citation preview
Can we explain the coexistence of ferromagnetism and
superconductivity based on two parameters model?
Yuan Zhoua,*, Jun Lia, Chang-De Gonga,b
aNational Key Laboratory of Solid State Microstructures, Department of Physics, Nanjing University, Nanjing 210093, ChinabDepartment of Physics, The Chinese University of Hongkong, Hongkong, China
Received 11 May 2004; received in revised form 1 September 2004; accepted 7 September 2004 by P. Sheng
Available online 18 September 2004
Abstract
A possible explanation about the coexistence of ferromagnetism (FM) and superconductivity (SC) based on a two parameters
mean field model in a two-dimensional system is discussed. The key feature of this model is that there are two independent
parameters which are responsible for ferromagnetism and superconductivity, respectively. We point out that the coexisting FM
and s-wave pairing SC state is energetically not favorable among all possible state. We generalize the two parameter model to
include the coexistence of FM with p-wave SC. We find that the phase diagram is not consistent with what experimentally
discovered in UGe2.
q 2004 Elsevier Ltd. All rights reserved.
PACS: 74.10.Cv; 75.50.Cc; 75.10.Lp
Keywords: A. Superconductors; D. Electronic band structure
1. Introduction
It is a long-time period for people to believe that the
singlet s-wave pairing superconductivity (SC) and ferro-
magnetism (FM) are mutually exclusive in the case of spin
singlet pairing because of Pauli paramagnetic limit. But
interests in this aspect are revived due to a recent discovery
of the coexistence of itinerant FM and SC in UGe2 [1,2] and
other compounds, such as URhGe [3] and ZnZr2 [4]. These
experiments show that the same electrons are responsible for
both the FM and SC. UGe2 is a heavy-fermion compound
with highly anisotropic structure, however, its 5f electrons
are more itinerant than that in many other heavy-fermion
systems. Below the Curie temperature, Both UGe2 and
URhGe contain zigzag chains of nearest-neighbour uranium
0038-1098/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ssc.2004.09.012
* Corresponding author. Tel.: C86 2583594678; fax: C86
2583595535.
E-mail address: [email protected] (Y. Zhou).
ions [1,3]. An ab initial calculation pointed out that the
quasi-two-dimensional Fermi surface has weak dispersion
along ky [5]. It means that the structure of these compounds
are anisotropic two-dimensional plane system [6], the inter-
plane distance is much larger than the lattice constants (a,b)
within the plane, or more precisely it is equivalent to a 2Ce-
dimension system, (where 0!e!1), in which the thermal
fluctuation has been suppressed. So far as the conduction
problems are concerned, including the superconductivity,
the system can be described by a 2-D mean field model with
appropriately selected parameters for considering the 2Ce
dimensionality effect. In general, high pressure would favor
the choice of a more symmetric cell [2]. Another important
experimental result about the superconducting critical
temperature is its non-monotonic pressure dependence in
coexistent phase [1]. Recently, Karchev et al. [7] proposed
the first three-dimensional mean field model where s-wave
SC and FM are caused by two independent coupling
constants. Though it seems to be questionable to derive this
model from the original 4-fermion Hamiltonian [Eq. (1) in
Solid State Communications 132 (2004) 507–511
www.elsevier.com/locate/ssc
Y. Zhou et al. / Solid State Communications 132 (2004) 507–511508
[7]] by means of the saddle point approximation [8].
Furthermore, the stability of their coexistent solution in
comparison with other solutions is doubtful [9]. On the other
hand, there are experimental evidences from FM exchange
band splitting and specific heat to preclude the spin singlet
pairing and support the spin-triplet p-wave pairing in UGe2.
Theories based on spin-fluctuation mediated p-wave pairing
[10] are applied to explain the UGe2 compound [6,11–15].
Motivated by these progresses in the field of unconven-
tional ferromagnetic superconductivity, we believe more
works within the two parameters model should be
performed. But the question is that what the symmetry of
the SC pairing can be tolerated in the ferromagnetic SC in
the frame of mean field theory and two parameters model.
We will present numerical study with more realistic two-
dimensional (2D) band structure instead of the three-
dimensional free electron energy. We show that the
coexisting FM and s-wave pairing SC state has highest
ground state energy among all possible solutions. Further-
more, we generalize the two parameters model to the case of
spin triplet pairing. We find that the coexisting FM and p-
wave paring SC phase is the most stable one. Unfortunately,
the parameter phase diagram of this coexistence phase is not
consistent with the experimental finding in UGe2 system.
This might be due to the unrealistic isotropic dispersion to
be used. By using a more realistic tight-binding anisotropic
2D band structure, we will find that the magnetization
dependence of the SC energy gap is non-monotonic, which
may explain the experimental result discovered in UGe2[16].
Fig. 1. The total energy vs. g with s-wave pairing. Solid line for the
FM phase; dot line for the coexistence phase and dash line for pure
SC phase at half-filling case. The total energy of the coexistence
phase is always highest among three possible phase, while the total
energy of the FM phase is smaller than that of the pure SC phase
unless g is large enough.
2. Theory and results for s-wave pairing
The mean-field Hamiltonian with s-wave pairing and
itinerant FM order is [7]
Hseff Z
Xks
ekscCkscks K
Xk
ðD�cKkYck[CDcCk[cCKkYÞ
C1
2JM2 C
jDj2
g; (1)
where eksZekGðJM=2Þ (sZG1 for spin-up and spin-
down, respectively), 2kZK2tðcos kxCcos kyÞKm for 2D
system with the nearest-neighbor hopping t. The SC energy
gap is DZghcKkYck[i. The ferromagnetic exchange J and
SC coupling g are independent to each other in this model.
The magnetization of the system defined by MZKhSZ iZ1=2ðhnYiK hn[iÞ and the chemical potential m determined by
the total particle number per site NZhnYiChn[i should be
solved jointly with the gap equation
1
gZ
1
2
XEkO
JM2
1
Ek
: (2)
The particle distribution functions at zero temperature are
found to be
hnk[iZ1
21K
ek
Ek
� �q Ek K
JM
2
� �; (3)
and
hnkYiZ 1K1
21C
ek
Ek
� �q Ek K
JM
2
� �; (4)
where EkZffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffie2kC jDj2
q: The total energy is
hHiZXEk!
JM2
ek CXEkO
JM2
ek 1Kek
Ek
� �K
JM2
2K
D2
g
CmN: (5)
Solutions of the above three self-consistent equations are
corresponding to the free electron state (DZ0, MZ0), the
ferromagnetic state (DZ0, Ms0), the BCS state (Ds0,
MZ0) and the coexisting SC and FM state (Ds0, Ms0).
From the point of view of the total energy hHi, the free
electron state is eliminated. The relative stabilities of the
remaining three phases could be obtained by comparing
the total energy as plotted in Fig. 1. The total energy of the
coexistence phase is higher than that of the two single
phases for all values of J and g. On the other hand, the
normal FM phase and the BCS phase compete with each
other. In mean-field theory, non-zero magnetization is
available only when JOJc. The FM phase is stable for
small enough g at JOJc, otherwise, the BCS phase is the
ground state. The (J, g) phase diagrams at three different
electron fillings are plotted in Fig. 2. So that it is clear that
the two parameters model with tight-binding electron band
in the s-wave pairing channel cannot explain the coexistence
of SC and FM. The possibility of other pairing symmetry
Fig. 2. Phase diagram in the condition of s-wave pairing, solid line
for NZ1.0 case; dotted line for NZ0.95 case and dash line for NZ0.9 case. The system prefer to FM phase unless the SC coupling
constant g is large enough.
Y. Zhou et al. / Solid State Communications 132 (2004) 507–511 509
should be considered to understand the coexistence
phenomena within the two parameters model.
3. Theory and results for p-wave pairing
We generalize the above two-parameter model to the
case of triplet p-wave pairing. Due to a large spin-splitting
of Fermi surfaces, the possibility of pairing opposite
momentum states with anti-parallel spins is removed [3],
and only equal-spin pairing (ESP) should be considered.
The effective Hamiltonian with ESP is
Hpeff Z
Xks
eksCCksCks K
1
2
Xks
ðD�s ðkÞcKkscks
CDðkÞcCkscCKksÞ:
The p-wave SC energy gap is defined as DsðkÞZP
k 0 Vðk;
k 0ÞhcKk 0sck 0si; where the interaction is assumed as V(k,k 0)ZggkKk 0 with gkKk 0ZcosðkxKk 0xÞCcosðkyKk 0yÞ: The anti-
symmetric p-wave energy gap is taken as DsðkÞZDsðsin
kxC i sin kyÞ; a form analogous to that in the A1 phase of
He3[17]. Hence, the four self-consistent equations for the
magnetization M, chemical potential m and the SC order
parameters (D[, DY) are derived as
M ZK1
4
Xks
seksffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e2ks C jDsj2ðsin2 kx Csin2 kyÞ
q0B@
1CA; (6)
N ZXk
1K1
2
Xks
eksffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffie2ks C jDsj
2ðsin2 kx Csin2 kyÞ
q ; (7)
1
gZ
1
2
Xk
sin2 kxffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffie2ks C jDsj
2ðsin2 kx Csin2 kyÞ
q ; (8)
where sZG1 for spin-up and spin-down channel, respect-
ively. The total energy with p-wave pairing is
hHiZXks
ekhnksiK1
4
!Xks
jDsj2ðsin2 kx Csin2 kyÞffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e2ks C jDsj2ðsin2 kx Csin2 kyÞ
q0B@
1CA
KJM2
2CmN: (9)
If the magnetization equal to zero, it refers to the pure SC
phase.
There exist two types of equal spin pairing in the FM
phase due to spin rotation symmetry breaking. The ESP SC
gap in spin-down (which is taken as the majority spin
polarization direction) is generally larger than that in the
spin-up channel except the half-filled band. The chemical
potential is zero in this special case, so that ek[ZKekC~QY
with ~QZ ðp;pÞ: Thus the gap equation in the two ESP
channels are identical. When FM disappears, the two type of
the SC will be coincident. The results are shown in Fig. 3.
The SC energy gap increases with the increasing g, while the
magnetization decreases, which means that the SC pairing
will suppress the magnetization. On the other hand, the
increasing of J, i.e. the strong magnetization will also
depress the SC energy gap as shown in Fig. 4, the SC energy
gap will disappear if magnetization is large enough. The FM
wishes the spin to be in same direction, while SC prefers to
the equal amount of spin-up or spin-down pairing for
lowering the total energy. The competition between SC and
FM is the reason of the depression of each other. Is the
coexistence phase stable in the p-wave pairing case? We can
calculate the total energy of the coexistence phase and
compare with that of the FM phase and the pure SC phase
with the same parameters. The Fig. 5 shows that the total
energy of the p-wave type coexistence phase is the lowest
one among the pure SC phase, FM phase and the
coexistence phase, which means the coexistence phase
could be established. The two parameter (J, g) p-wave
pairing model tolerate the coexistence phase, but shows only
monotonous behavior of SC energy gap vs. g unless g/J have
some anomalous behaviors under the pressure. If instead of
the isotropic 2D tight-binding hand, we use more realistic
anisotropic dispersion given by Sandeman et al. [6]: ekZKa cos kxKb cos kx cos kyKg cosð2kxÞKd cosð3kxÞ with
aZ1, bZ0.7, gZ0.03, dZK0.03. Through the same
calculations as given in (6)–(9), the SC energy gap versus
Fig. 3. The magnetization and SC energy gap vs. g with p-wave pairing. (a) NZ1.0, JZ8.0 case, DYZD[; (b) NZ0.9, JZ8.8 case, DYOD[
until MZ0. The SC energy gap increases with the increasing g, while magnetization decreases down to zero.
Fig. 4. The SC energy gap with p-wave pairing vs. the
magnetization with different g at half-filling case.
Fig. 5. The total energy vs. g with p-wave pairing at NZ1.0 case. It
shows that the total energy of coexistence phase is always smaller
than that of the pure SC phase with p-wave pairing in 2D case.
Y. Zhou et al. / Solid State Communications 132 (2004) 507–511510
Fig. 6. The SC energy gap with p-wave spin-down pairing vs.
magnetization with anisotropic band structure given in the text. NZ0.72, solid line for gZ1.2, dash line for gZ1.0, and dotted line for
gZ0.8 [16].
Y. Zhou et al. / Solid State Communications 132 (2004) 507–511 511
magnetization plot shows non-monotonic behavior as
shown in Fig. 6, i.e. a two parameters model in an
anisotropic 2D system with p-wave pairing is a possible
choice to describe the coexistence of FM and SC in UGe2within the mean field theory. Other choice beyond the two
parameters model, for example the mean field one parameter
model may also give reasonable coexistence phase with
p-wave pairing.
4. Conclusion
In conclusion, the coexistence of FM and SC in the two-
dimensional system was investigated in the case of s-wave
or p-wave pairing based on the two parameters mean field
model. We find that the order parameters of the FM and the
SC depress each other in both cases. By comparing the total
free energy, we show that a s-wave type coexistence is
always unstable, while a p-wave type coexistence is
possible. Our results on the p-wave pairing fail to show
the non-monotonous behaviors of the SC energy gap as
experimental data shown, which means that the two-
parameters model with isotropic two-dimensional band
structure is not adequate to describe the coexistence of FM
and SC discovered in UGe2.
References
[1] S.S. Saxena, et al., Nature 406 (2000) 587.
[2] A. Huxley, et al., Phys. Rev. B 63 (2001) 144519.
[3] D. Aoki, et al., Nature 413 (2001) 613.
[4] C. Pfleiderer, et al., Nature 412 (2001) 58.
[5] A.B. Shick, W.E. Pickett, Phys. Rev. Lett. 86 (2001) 300.
[6] K.G. Sandeman, G.G. Lonzarich, A.J. Schofield, Phys. Rev.
Lett. 90 (2003) 167005.
[7] N.I. Karchev, K.B. Blagoev, K.S. Bedell, P.B. Littlewood,
Phys. Rev. Lett. 86 (2001) 846.
[8] Y.J. Joglekatr, A.H. MacDonald, Phys. Rev. Lett. 92 (2004)
199705.
[9] Y. Zhou, J. Li, C.-D. Gong, Phys. Rev. Lett. 91 (2003) 069701.
[10] D. Fay, J. Appel, Phys. Rev. B 22 (1980) 3173.
[11] T.R. Kirkpatrick, D. Belitz, T. Vojta, R. Narayanan, Phys.
Rev. Lett. 87 (2001) 127003.
[12] P. Monthoux, D.J. Scalapino, Phys. Rev. Lett. 72 (1994) 1874.
[13] Z. Wang, W. Mao, K. Bedell, Phys. Rev. Lett. 87 (2001)
257001.
[14] P. Monthoux, G.G. Lonzarich, Phys. Rev. B. 59 (1999) 14598.
[15] R. Roussev, A.J. Millis, Phys. Rev. B 63 (2001) 140504.
[16] Y. Zhou, J. Li, C.D. Gong, Chin. Phys. Lett. 21 (2004) 1805.
[17] P.W. Anderson, et al., Phys. Rev. 118 (1960) 1442 Phys. Rev.
Lett. 30 (1973) 1108.