Pinettes 1993

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

~ ) Solid State Communications, Vol. 85, No. 7, pp. 565-567, 1993. 0038-1098/9356.00+.

00
Printed in Great Britain. Pergamon Press Ltd

A N T I F E R R O M A G N E T I C I N S T A B I L I T I E S O F T H E HUBBARD M O D E L ON T H E T R I A N G U L A R
LATTICE

C. Pinettes, C. Lacroix

Laboratoire Louis N6el, C.N.R.S., Avenue des Martyrs, B.P. 166, Grenoble Cedex 9, France

(received 19 October 1992 by P. Burlet, in revised form 1st December 1992)

The Hubbard model is studied on the triangular lattice using Hartree Fock approximation.
Due to the frustration of the lattice many antiferromagnetic states can be stabilized,
depending on the parameters. We study the stability of a mixed structure in which
magnetic and non-magnetic sites are present : such a structure can be obtained when U is
not too large as a consequence of frustration. It is shown that the anisotropy also plays an
important role for stabilizing this mixed structure.

1. I N T R O D U C T I O N uniform and the excess of electrons or holes are localized at


the vortex sites, which are nonmagnetic sites. Finally it
Frustration is frequently present in magnetic systems, should be mentioned that Ikawa and Ozaki 14 listed the
both in disordered systems and in compounds. In the various Hartree-Fock states of this model using group
second case frustration occurs either due to the theory arguments.
crystallographic structure or due to the long-range In this paper we are mainly interested in the case of
interactions in metallic systems. large uniaxial anisotropy : in this case, close to the M-NM
In frustrated systems several unusual properties are instability the Hubbard model can mapped on an S=I
observed at low temperature : spin-glass behaviour in Blume-Capel modellS; the study of this S=I model is much
disordered systems, helicoidal or antiphase structures in simpler than the starting model and it leads to many
magnetic rare-earth compounds, "order by disorder" when different ordered states (magnetic and mixed states),
the ground state is degenerate at T f 0 ° K ~ ...This paper is depending on the parametersl6.. In this paper we study the
concerned with a different case : in itinerant frustrated stability of some of the states obtained in ref.16 in the
systems, close to the magnetic instability, a new ground Hubbard model.
state has been observed in which magnetic and non-
magnetic sites are present : this is the case in several 2. T H E M O D E L
Laves phase compounds (TiFe22, ThMn23, DyMn24,5) but
also CeSb 6 could be another example of this mixed We consider the one-band Hubbard model with 1st
magnetic structure. and 2 nd neighbour hoppings,t and t',in the triangular
The RMn2 compounds can be considered as typical lattice •
examples of the effect of frustration in itinerant systems H=H0+V
close to a magnetic-nonmagnetic (M-NM) instability :
-the Laves phase structure is strongly frustrated. In fact it
has been shown that the Heisenberg model does not order Ho=t ~ c~oc~o+t' ]~ <'oCko (I)
in this structure7, 8. i,j, o i,k, 0
-the Mn lattice is close to the M-NM instability : below a
critical Mn-Mn distance de, Mn atoms remain nonmagnetic, V U L ni ~ ni, (2)
while above de Mn magnetic moments increase with the i
Mn-Mn distance4.
The interaction V will be treated in the local Hartree-
-the anisotropy on the Mn sites is very large 9, and it will be Fock approximation :
shown below that anisotropy is necessary to stabilise the
mixed state.
The Hubbard model on the triangular lattice can be 2 C oOo-; Cio÷ nicoc,o
used to describe the competition between these three effects I, 0 , 0 1, °
(frustration,M-NM instability,anisotropy) : in such a + ~ ~ (I Sil 2_ n~) (3)
lattice, mixed states can be stabilized because they suppress
or decrease the fruslration of the magnetic interactions
It has been already recognized by several authors that where Si is the selfconsistent magnetic moment on site i,
the Hubbard model on the triangular lattice shows
interesting behaviours : Krishnamurthy et al. 10, and oo~a' denote the Pauli matrices :
Jayaprakash et al. ll studied the half-filled case in a mean- S'-~= ~ (C+° <3o'-~'-" Ci O') (4)
field approximation; three different magnetic states can be O, 0'
stabilized, depending on the value of U. Machida et al.12 ni is the charge on site i :
and Fujita et al. 13 did an extensive study of the Hartree-
Fock ground states, including multi-Q structures : when ni = ~ c+o ci o (5)
°
the band filling is different from I electron per site, the
ground state can be described as a vortex-lattice state when In the case of large anisotropy we restrict to uniaxial
U is large enough. In such a state the charge is non- states and Si is replaced in eq.3 by a scalar parameter :
565
566 ANTIFERROMAGNETIC INSTABILITIES Vol. 85, No. 7
- 2 antiferromagnetic collinear phases defined by :
Ql=(0,2rcH3), p.i=[t cos(QrRi), ni=n (phase I)
Q2=(rL0), t.ti=~(cos(Q2"Ri)+sin(Q2"Ri)), ni=n (phase
II)
- 2 mixed phases :
Q3=(2rc/3,2x/q3), (phase III) and Q4=(2n/3,0), (phase
IV) with the same distributions : txi=21.t/'~/3sin(Q.Ri) and
ni=n0+4/3 (nl-n0) sin2(Q-Ri).
- the non-magnetic phase V
- t h e isotropic antiferromagnetic phase VI which is also
defined by Q3. This phase will be compared with the mixed
phase III to study the influence of the anisotropy.
We did not include the ferromagnetic phase in the
Hartree-Fock approximation this phase is stable for
U>l/p(Ef). In a triangular lattice the density of states
G becomes infinite at E=-2t (if t'=0) . For a band-filling
n=0.5 the Fermi level is equal to -2t (for t>0); thus it is
expected that close to n--0.5 as well as close to n=0 and 2
(because the density is finite at the band edges) ,
ferromagnetism will be favoured. However as this paper is
concerned with the effect of frustration on the
antiferromagnetic states, this ferromagnetic phase will not
be discussed in the following.

3. D I S C U S S I O N O F T H E R E S U L T S

3.1 Magnetic instability at small U


Fig.1 : Ordered phases considered in our model. See the Comparison of the different values of Uc(Q) indicates
text for the description of these states. the wave vector of the magnetic state which is stabilized at
small U. Of course several states correspond to the same Q
value (vI and III for example) and energy calculations are
~i = 91-"Z ¢~ ni ~ (6) necessary to obtain a complete phase diagram.
Figure 2 indicates which instability occurs at the
At large U/W (W=bandwidth) magnetic states are smallest value of U as a function of t'/t and n. In particular
usually stabilized; mixed states are expected to occur for the mixed phases III and IV can appear for any value of the
intermediate values of U/W, just above the critical value Uc band-filling. In the next paragraph we discuss more
of the M-NM instability. In the Hartree-Fock precisly the case t'=0, 0.9<n<1.7.
approximation Uc depends only on the Q-value of the
wave-vector caracterizing the magnetic state : 3.2 Stability of the mixed phase
1 = Z f(Ek)
f(.__._
- ~k +___Q) Comparison of the energies of the different phases
Uc(Q) k Ek+Q - Ek (7) requires to solve the selfconsistent equations for the
magnetic moment I.ti and the local charge ni (eq.5, 6).
where fie) is the Fermi function, and £k is the one-electron Figure 3 shows the resulting phase diagram for t'=0 and
band energy : ek = 2t ( coskx + 2 coskx/2 cos-~/3ky/2) + 2t' 0.9<n<1.7 where a mixed phase is possible : it is found
( coskyq-3+ 2 cos3kx/2 cos'~-Jky/2 ). that phase III is stable against the magnetic phases I and II
For a given value of the band filling, the magnetic in a limited range of concentration (1.25<n<1.45) and for
instability occurs when U is larger than the smallest Uc(Q). U/W smaller than a critical value above which all sites are
We have restricted our study to a small number of
commensurate states shown on figure 1 :
U/W

I or lI /

1.6

1.2
A
n ....

0.8
V
0.4 0 I I I
0.9 1.I 1.3 1.5 n

O.i -0.5 0 0.5 1


Fig.3 : The stability regions of different phases for t'=0.
t'/t Numbers correspond to phases in fig.1. The different lines
indicate the variation of the Uc(Q) : - - - - : Q = Q 1;
.... : Q=Q2; - - - : Q=Q3; . . . . . . . : Q=Q4. The
Fig.2 : Phases that appear at the smallest value of U as a stability region of phase III is limited by the solid line For
function of t'/t and n. These phases axe caracterized by their larges U/W the antiferromagntic phases I or II are
wave vectors. ~ Q 1; I ~ Q2; I Q3; ~ Q 4. stabilized.
Vol. 85, No. 7 ANTIFERROMAGNETIC INSTABILITIES 567

M with the mixed state III. Comparison of the energy of these


I I two phases indicates that the isotropic phase VI is always
II
0.6 lower in energy than phase III : in the absence of any
anisotropy the mixed phase is not stable; however as
shown in ref.12 and 13 more complicated structures are
then obtained which also show some charge transfer
0.4 between magnetic and nonmagnetic sites.
The effect of anisotropy can be taken into account by
adding to the hamiltonian H, a term V' :
0.2 U,__D~(Ci+~ci$+ci~ci$} 2,
• 2 D>0 (8)
1
In the local Hartree-Fock approximation this term writes:
1 2 U/W V' = 2D • Sx c~1"ci'l'+ c+'['ell"- D E { Sx ) 2 (9)
i 2 i
Fig.4 : Variation of the magnetic moment M per site as a where Six is the selfconsistent x component of Si (eq. 4).
function of U/W for n=1.35. The bold line indicates The instabilities along the z and x directions occur
transition from phase III to I. The inset shows the now at different values of U : for the same Q value, Siz
variations of the charge difference between magnetic (nl) becomes different from zero for U>Uc(Q), but Six can be
and non-magnetic (no) sites in the mixed phase III for different from zero only if U>Uc(Q)+4D. Thus one can
different n where phase III is stable. conclude that for 1.25<n<1.45 the mixed structure III will
be stable as soon as D~0 for Uc(Q3)<U<Uc(Q3)+4D.
Anisotropic magnetic structures (different from phase VI)
might be stable only for larger U values.
magnetic. Figure 4 shows the magnetic moment per site 4. CONCLUSION
and the charge transfer between magnetic and nonmagnetic
sites in this region. For n=1.35 the mixed phase is stable in We have shown that the mixed phase can be stable in
a large U region (0.4<U<1.7) and the charge transfer can the triangular lattice if (i) U is not too large and (ii) there is
be quite large as in the vortex lattice state studied in ref.12 a large anisotropy. These conditions are realized in the
and 13. RMn2 compounds• Of course our model provides only a
It can also be seen on figure 3 that the critical values qualitative description of the RMn2 since we have not
Uc(Q) of the different phases are very close to each other : considered realistic crystallographic and electronic
in a frustrated lattice several magnetic states corresponding structures. Recently Uhl and al. 17 reported an ab-initio
to different Q values are very close in energy. Small calculation for ThMn2 in the ordered phase; the results
distortions can lift the "quasi-degeneracy" of these states reproduce correctly the mixed magnetic phase of this
and reduce the charge transfer. compound•
Finally this paper conf'trras the results obtained within
the S=I modelI536 : in principle a correspondance
3.3 I n f l u e n c e of the anisotropy between the parameters of the two models could be made.
Both phases III and VI have the same critical U As this S=I model is much simpler it can also be used to
value : Uc(Q3). If anisotropy vanishes, phase VI competes find the magnetic structures of the Laves phase lattice•

REFERENCES

1. J. Villain, R. Bidaux, J. P. Carton, R. J. Conte, J. 10. H. R. Krishnamurthy, C. Jayaprakash, S. Sarker, W.


Phys. (Paris), 41, 1263 (1980) Wenzel, Phys. Rev. Lett. 64, 950 (1990)
2. P. J. Brown, J. Dtportes, B. Ouladdiaf, to be published 11. C. Jayaprakash, H. R. Krishnamurthy, S. Sarker, W.
3. J. Dtportes, R. Lemaire, B. Ouladdiaf, E. Roudaut, F. Wenzel, Europhys. Lett. 15, 265 (1991)
Sayetat, J. Magn. Magn. Mater. 70, 191 (1987) 12. K. Machida, M. Fujita, Phys. Rev. B42, 2673 (1990),
4. K. Yoshimura, M. Shiga, Y. Nakamura, J. Phys. Soc. K. Machida, Surf. Science, 242, 206 (1991)
Jap., 55, 3585 (1986) 13. M. Fujita, T. Nakanishi, K. Machida, Phys. Rev.,
5. C. Ritter, S. H. Kilcoyne, R. Cywinski, J. Phys. B45.2190 (1992)
Condens. Matter, 3,727 (1991) 14. A. Ikawa, M. Ozaki, J. Phys. Condens. Matter, 4,
6. J. Rossat-Mignod, P. Burlet, H. Bartholin, O. Vogt, R. 4039 (1992)
Lagnier, J. Phys. C 13,6381 (1980) 15. R. Ballou, C. Lacroix, M. D. Nunez-Regueiro, Phys.
7. P. W. Anderson, Phys. Rev. 102, 1008 (1956) Rev. Lett. 66, 1910 (1991)
8. J. N. Reimers, Phys. Rev. B45. 7287 (1992) 16. M. D. Nunez-Regueiro, C. Lacroix, R. Ballou, Phys.
9. R. Ballou, J. Deportes, R. Lemaire, Y. Nakamura, B. Rev. B 46, 990 (1992)
Ouladdiaf, J. Magn. Magn. Mater. 70, 129 (1987) 17. M. Uhl, L. Sandratskii, J. Kiibler, To be published

You might also like