An effective 1-band model for the cuprate superconductors
George Kastrinakis
Institute of Electronic Structure and Laser (IESL),
Foundation for Research and Technology - Hellas (FORTH), P.O. Box 1527, Iraklio, Crete 71110, Greece*
(Dated: Dec. 31, 2008)
Starting from the copper-oxygen Hamiltonian of the Cu02 planes, we derive analytically an
extended 1-band Hubbard Hamiltonian for the electrons on copper sites, through a canonical trans-
formation which eliminates the oxygen sites. The model sustains a variety of phases : checkerboard
0^ ' states, stripes, antiferromagnetism, local pairs and mixtures thereof. This approach may be helpful
in understanding what is so special about the CuC>2 planes, as opposed to other compounds.
O
<N
E3 ' I. INTRODUCTION
03
o
o
5h
Oh:
03
i
>
X
The starting point is the 3-band Hamiltonian, in which Cu 3d x 2_ y 2 and O 2p x y orbitals are taken into account [l[,
The creation/annihilation operators d\ a /d^ a describe electrons on the Cu02 planes, and the indices i,j run over all
lattice sites. The hopping matrix elements t a ij = t, t' and the off-diagonal Coulomb elements Vij — V, V act between
neighboring Cu and O atoms and between neighboring O atoms respectively. Ui = U, U p and e; = e<j, e p for Cu and
C/3 , O atoms.
The problem of reducing the 3-band Hamiltonian to a more amenable effective 1-band Hamiltonian has been treated
in a number of papers UH, 0, H, H, 0, II, H EH- Our goal is similar in spirit. As the holes tend to reside mostly
on the Cu atoms, we wish to incorporate in the new effective Cu 1-band Hamiltonian explicit 2-particle correlations,
i-^j , stemming from the original Hamiltonian.
We emphasize that our approach is not a large-t/ type (which was shown to be problematic [ill]), thus allowing
for double occupancy of the Cu sites. It merely eliminates the oxygen sites. To this end, we use the canonical
transformation method of Chao, Spalek and Oles (CSO) [12j |. adapted to the Hamiltonian of eq. (1) for the CuC>2
planes. Using a Hamiltonian related, but not identical, to (1), Zaanen and Oles (ZO) Q applied this canonical
transformation method to the cuprates. Besides their different Hamiltonian, ZO followed a different strategy. The
transformed Hamiltonian was separated into parts depending on the number of doubly occupied sites, and oxygen
sites explicitly appeared therein, in contrast to our approach.
, It is understood that the present method can also be applied to lattices other than the Cu02 plane.
In the following, section II contains the canonical transformation formalism. In section III we present the new
effective Hamiltonian H . Section IV contains the solution for the ground state of H along with a brief discussion on
C^) ' the phases encountered.
On '
O ■ II. CANONICAL TRANSFORMATION FORMALISM
Following CSO [I2j, we write the Hamiltonian as
H = H A + H B , (2)
H b = J2 toiA^j,* = E / '< // " / ' .
Ha = H — Hb = 22 PiH a Pi ,
with Pi being projector operators with J2j Pj = 1- Pi projects a state on the eigenstate with eigenergy Ei of the
interacting part of H Q , i.e. Ha\i >= Ei\i >. Moreover, following CSO (c.f. before eq. (6a) of ref. [Ill), we assume
that
< i\H B \j >=0 , Ei=Ej , (3)
i.e. Hb does not connect states which are energetically degenerate. This condition is further discussed below.
2
We consider the canonical transformation
H = e- lS {H A + H B )e lS
= H A + H b - i[S, H A ] + {H 5 > Ha ^ + m H 5 > H B]]n-i} ,
n=2
where the operator S is such that
H B -i[S,H A ]=0 .
(4)
(5)
This condition amounts to the elimination of the O sites from H. However, the matrix elements of H between
non-degenerate states depend implicitly on the occupation of the O sites - c.f. below. We use the notation
[[A,B]] n = [A, [A, [...,[A,B]]-
with n commutators at the right-hand side.
Substituting eq. (O into ((U yields
H = H A + J2 (n ~ 1)( p )W ' X [[g, H B ]] n ^
(6)
(7)
n=2
An expression for S is derived by substituting into ([5]) Ha and Hb from and then apply the projectors Pj from
the left and P k from the right on both sides of ([5]), thus yielding
PjH P k (l - S jk ) + iPjHoPjiPjSPk) - i(P 3 SP k )P k H P k = .
Noting that X k = P k H a P k = P k H A P k , X k are replaced by the proper energy eigenvalues E k . For Ej ^ E k eq.
leads to
PjSPk =
PjH B P k
(8)
(9)
Ej - E k
It also follows that [l2|
I'.SI'j r/> , (10)
with c an arbitrary constant, whose value is irrelevant. Then, using eqs. ©, (|10[) and J2j Pj = 1)
1{S, H L
II I 71+1
PjSPk, ~i J2 ( E * - E m)PlSPn
all k m I j=0
(11)
(12)
and the double primed summation is restricted to E km ^ E km+1 for all to, in accordance with condition above.
Substituting [[5, i?s]]„ into eq. yields
B-H A -± t^SL ± ± ^,E k ,„ } P hS P k ,S...SP h , t ,
71=2
i! I ^— ' 7*!fn — 7')!
all fc m I j=0 J V J;
This is eq. (25) of CSO. Using eq. ([9]), it can also be written as
00 //
H = H A -Y / (-1)" Pk 1 H B P k2 H B ...H B P kri+1
71=2
where
all k„
-iKn-i)! nr=i(^*-^ i+1 :
(13)
(14)
(15)
3
Expressions for the factors /„ used herein are given in Appendix A.
III. EFFECTIVE HAMILTONIAN
Carrying out the expansion to fourth order in the hopping elements t, t', the new Hamiltonian turns out to be
H= ^ c j,o- c j>o-{*i<»( 1 - nj- a )(l - ni- a ) + t lb [rij- a (l - ni-„) + (1 - nj-„)ni-„] (16)
<l,j>,a
+ti c nj l - (T ni t - <T + tidrij i - u (l - ni- a ) + t le (l - nj- a )ni- a }
- c \,ct c o,°{( 1 - n t ,-<r)(l - «i,ff)[*2o(l - Tlj ;,- CT )(l - m _ a )
<l,j;i>' ,<7
+t 2 b{rij- a (l - ni-a) + (1 - rij -0)111 - a } + t2 C nj- a ni- (T ] + t 2 dni- a (\ - n it(T )(l - nj-^ni-a
+n iiCT (l - ni- a ){t A (l - ni- a )(l - rij- a ) + t B [ni,- a (l - rij- a ) + (1 - ni-^nj-a] + tcni- a rij- a }
+ni^ni- a {t D {l - ni- a ){\ - rij- a ) + t E [ni,- a (l - nj- a ) + (1 - ni- a )nj-v] + tpni-^rij-a}}
~ J! C l<7 C J>{*2e(l - n i)(7 )(l - rii- a )
<l,j;i>, <r
+%{(! - ni-o)ni, a + rii,-a(l - n^)} + t2gn i - a n i , (T }[nj- a (l - raj.-^) + (1 - nj-„)ni-„\
+U^2n h1 n hl - A S xa Y n J^ ni -"i 1 ~ ~ n '^)
i <l,j>,<J
- Y i A sxb n jir7 nj- a [ni ia ,(l -n;_ CT /) + n i; _ CT >(l -rc/,^)] + -4sxc %>(1 -n J _ (7 )(l -ra i;<T >)(l -n/-^)}
<l,j>,(T,a'
-Ap T Y c \,A-cr c 3-° c i,° ~ A xe c \-A,<t c o,° c i-°
<l,j>,<? <l,j>,a
<i,j;l>' ,ct
{A FSa (1 - n J> )(l - ni,_ CT ) + ^FSfc [nj,<r(l - + (1 - nj,a)ni-<j] + A FBc rij^Tii-^}
- Y c \- a c \.<j c 3,a c i-c {AcMa(l - n i)(T )(l - nj,_ CT ) + A C Afb(l - n^ a )nj- a
<i,j;l>'a
+A C Mcnucr(l - Tlj - a ) + A C MdTli,crTlj-a} ■
The indices I run exclusively over the Cu lattice and rij j(7 = cJ CT Ci j(T . < j, Z > implies that j and I are nearest
neighbors (n.n.), < j, l;i > implies that j and I are second neighbors and i is the common n.n. and < j, I; i >' implies
that in addition to second neighbors j and I can be third neighbors with i the middle common n.n. The term 'empty
lattice' below refers to the case of no other electrons present than the ones hopping between initial and final positions;
we ignore the electrons in the rest of the lattice. Henceforth e = e p — a and V = 0.
We see that the only term remaining intact from the original H a is the Hubbard term on Cu sites. All hopping
terms now depend on the site occupancy. The other new terms generated include superexchangc (SX), a pair-transfer
(PT) between n.n., an exchange of electrons (XE) between n.n., a local pair- formation or pair-breaking (FB), and a
correlated motion (CM) of two electrons. The respective matrix elements are given below.
Hopping elements
1st neighbors - all the terms of order t A below contain a hopping forth and back between a Cu and an O atom.
Empty lattice, i.e. no other electron is present in the Cu sites j and I involved
t 2 AtH' f 1 1 2 1
tla ~ e + U p -6V + 3 \V(e + U p -6V) V(e + U p -7V) (e + U p - 6V)(e + U P - TV) J 5 1 ' ''
3tH /2 f 1 1
V 2 {e + U p -7V) V(e + U p -7V)
3^ f 1 3 \
+ 8(e + U p - 6V)(e + U p - 5V) { e - U + U p - 5V e-U + U p -6V)
m 4 r 1 3
(e-U + U p -6V){e-U + U p -5V) { e + U p - bV e + U p -6V
4
3t 4 9t 4
"4(e + U p - 6V)(e - U + U p - W) 2 ~ 4(e + U p - 6V) 2 (e - U + U p - 5V)
3t 4 3t 4
f _ _ _ . r , ~- 7 -
3t 4
+
(e + C/ p -6y)(e + C/ p -7^) 2 (e + U p - 6V) 2 (e + U p - TV)
it 4 f 1 1 6
8(e-f7 + f7p-4V)(e + t/p-7F) 1 e + C/ p - 5V e-U + U P -5V e + U p -6V
3t 4 f 1 1 1 f 1 3 1
+ 8(e + C/ p - 6V) t e + U p - hV + e - U + U p - hV J \ e - U + U p - W ~ e + U p - IV J
3i 4 f 1 11 2t 4
+ — | 7 ! 77 7777777 77^7 77T77 + 7"^7 7T777 77^7 77777777 > +
(e + U p -W) 2 (e-U + U P -6V) (e + U p - 6V)(e - U + U p - 6V) 2 J (e + U p - Wf
st 4 r i i
(e-U + U p - 4V)(e - U + U p - W) \ e + U p - 6V e + U p -7V
9t 4
+ -
8(e + U p -6V)(e + U p -7V)
3t 4 9t 4
{e - U + U p - W + e - U + U p - w]
A(e-U + U p - 5V) 2 {e + U P - 6V) 4{e-U + U p - 5V)(e + U P - 6V) 2
1st neighbors - an additional —a electron at either the initial site j or final site I
t 2 r i li t 2 f 2 r i
hb ~ ~2 1 e - U + Up - 5V ~ e + Up - 5V J 8^ \e - U + U p - 6V + e + U p - 7V j ^
2t 2 t' f 1 1 2 1
+ HT { 2V(e -U + Up-hV) ~ 2V(e + U p - TV) ~ (e - U + U p - 5V)(e + U p - IV) J
2t 2 t' f 1 1 2 1
+ HT \ V(e-U + U P - SV) ~ V(e + U p - 5V) ~ (e - U + U p - 6V)(e + U p - 5V) J '
1st neighbors - an additional —a electron at both initial and final sites j and I
e 3t 2 t' 2 f 1 2 1
<lc ~ e-U + Up-W 8 ty 2 (e-Z7 + C/ p -6F) V(c - U + U p - 6V) 2 J ( '
t 2 t' f 1 1 2
+ ^7- -77777 77-^7 7777 +
3 t 2V(e-U + U P -6V) 2V{e -U + U P - AV) (e - U + U p - 6V)(e - U + U p - AV)
t 4 3t 4
+ 7 77-^7 777777 +
(e - U + U p - 4V) 3 2(e-U + Up-6V) 2 (e-U + U p -W)
3t 4
+ 2(e - U + Up - 6V)(e - U + Up - AV) 2
1st neighbors - an additional — a electron at the initial site j only
3^ f 1 1 \
tld ~ 8(e - U + Up - 4V)(e - U + U p - 5V) { e + U p - hV + e - U + U p - 6V } '
3^ f 3 4 |
+ 8(e - U + Up - 6V)(e + U p - 5V) { e - U + U p - 4V + e - U + U p - 5V J
it 4 f 1 5 \
+ 8(e-U + U P -W) 2 \ e - U + Up - 5V ~ e + Up - 5V j
3^ f 1 3 \
+ 8(e - U + Up - W)(e + U p - IV) \ e - U + U p - W e - U + U p - 6V J
3^ f 1 3 1
+ 8(e - U + Up - 3V)(e - U + U p - W) \ e-U + U p -4V " e + U p - 7V J
3t 4 r i 2
4(e + U p -7V)(e -U + U p -W) { e - U + U p - 3V e + U p - 5V _
3t 4
(e - U + U p - 5V) 2 { e + U p -lV e-U + Up-SV
3t 4 f 1 3
+ 8(e + U P - 7V) 2 \ e + U p - W + e - U + U p - 5V
3t 4 r i 3
8(e - U + U p - 6V) 2 {e-U + U P -5V e + U p -W
t 4 f 1 7 1 3t 4
+ 777 77~^7 TTTTo 77~^7 777 + — "T7 777 > +
8(e-U + U p - W) 2 {e-U + U P -5V e + U p -5V J 2(e - U + U p - 5V)(e + f7 p - 5V)(e - C/ + Up - 6V)
3^ f 3 1 1
1st neighbors - an additional —a electron at the final site I only
he 4{e-U + U p - hV) 2 \e-U + U p -5V e + U p - 5V ^
3t 4 3t
+ 777 77-^7 7T777 . rT 77777 77~^7 3777 + 7 77^7 =777? 77^7 777 +
2(e-U + Up - 5V)(e + U p - W)(e - U + U P -7V) [e-U + U P -7V) 2 { e - U + U p - 5V e + U p - hV
3t 4
8(e - U + Up - W) 2 {e-U + U P -3V e - U + U p - 7V J A{e-U + U p - 5V)(e -U + U p - 3V)(e + t/ p - 7V)
9t 4 f 1 1
e-U + Up-W){e + U p -7V) {e-U + U P -AV e-U + U P -3V
3t 4 f 1
+ 777 77-^7 77777 77^7 77777 77~^7 777 +
8(e - U + Up - 4V)(e - U + U p - 3V) { e - U + U p - 6V e + U,
3t 4 f 1
(e-U + U p -AV)(e-U + U P -5V) {e-U + U P -6V e + U p - 5V
9t 4 f 1 1
+
8(e-U + U p - 6V)(e + U p - W) { e - U + U p - AV e-U + U p -5V
3t 4 f 1 3 j 3t 4
+ 777 77-^7 77777? j — T7 , Tr 777 + ^^7 777 ^ +
8(e-U + U p - W) 2 {e-U + U P -5V e + U p -5Vj 2(e - U + U p - 6V)(e - U + U p - 5V)(e + U p - 5V) '
2nd/3rd neighbors - empty lattice, i.e. no other electron is present in the Cu sites involved, j initial, i intermediate
and I final
t 4 ( 1 3
t2a = ~ 7 \ (e + Up - 5V) 2 (e + U P - TV) ~ (e + U p - 5V)(e + U P - 7V) 2 (22)
+
(e + Up - 6Vy
1
e + Up - 5V e + U p -7V
{e + Up- wy
2nd/3rd neighbors - an additional — a electron at either the initial site j or final site I
t 4
f i i 3 3 \ mi
l2h ~ _ 8(e + U P - 7V)(e + U p - W) \e + U p -W + e - U + U p - 5V ~ e - U + U p - 6V ~ e + U p -6V J ( '
t 4 r i 3 l r i i
e + U p -W e + U p -7V j {(e + Up- W){e -U + U P -6V) {e-U + U p - W){e + U P - 6V)
t 4
{e + Up - 6V + e-U + U p -6v]
8(e + U P - 5V) 2 {e + U p -6V ' e - U + U p - 6V J 4(e + U p - 6V) 2 (e - U + U p - 5V)
t 4 r i 2
4(e + U p - W)(e + U p - 6V) {e-U + U P -6V e-U + U P -5V
2nd/3rd neighbors - additional — a electron at both initial and final sites j and /
t 4 3^
<2c _ 4(e + U P - AV) 2 (e -U + U p -6V) ~ 4(e + U p - W)(e -U + U p - 6V) 2 ^ '
M 4
2(e + U p -5V) 2 (e-U + U p -5V) 4(e - U + U p - 5V)(e - U + U p - 6V) \ e + U p - AV e + U p - 5V
+
4(e + U p - AV)(e + U p - hV) \ e - U + U p - hV e-U + U p - 6V
2nd/3rd neighbors - one additional —a electron at all 3 sites, initial j, final I and intermediate i
t2d ~~4 \(e-U + U p -5V)(e-U + U p -3V) 2 ~ {e - U + U p - W) 2 (e - U + U p - 3V) ( '
1 3 2
(e - U + U p - 3V)(e - U + U p - AV) 2 (e - U + U p - 5V)(e - U + U p - AV) 2 (e-U + U p - AV f
2nd/3rd neighbors - one additional a electron at intermediate site i
= - (e + D *_6V)3 • (26)
2nd/3rd neighbors - one additional a electron at intermediate site i and one additional —a electron at either the
initial site j or final site I
t A | 1 /'
tB = —,
{ e + U p - bV + e - U + U p - bV }
8(e + U P - 6V) 2 \t + Up-bV e-U + Up- 5V J 2(e + U p - 6V)(e + U p - 5V)(e - U + U p - 5V) '
(27)
2nd/3rd neighbors - one additional er electron at intermediate site i and two additional —a electrons at both initial
site j and final site I
3t 4 t A
tc ^~2{e-U + U p - 5V) 2 (e + U p - 5V) ~ 4(e + U p - W) 2 (e -U + U p -bV) ' ^ 28 ^
2nd/3rd neighbors - two additional electrons at intermediate site i
t 4 3t 4
tD ~ ~A(e-U + U p - 5V) 2 (e + U p - 5V) ~ 4(e + U p - 5V) 2 (e - U + U p - 5V) '
2nd/3rd neighbors - two additional electrons at intermediate site i and one additional —a electron at either the
initial site j or final site I
t 4 t 4 f 3 1
tE = - 7T, , tt TT—Tr TT—Tr 7T7T-777 7T— FT 777^ { — T7 P77 +
2(e + Up - 5V)(e - U + U p - 5V)(e — U + U p — AV) 8{e-U + U p - AV) 2 \e + U p -5V e-U + U p -5V
(30)
2nd/3rd neighbors - two additional electrons at intermediate site i and two additional —a electrons at both initial
site j and final site I
t 4
tF = -{e-U + U p -AVf ■ (31)
2nd neighbors only - an additional —a electron at either the initial site j or final site I
t ^ f 1 1 ? \
2e 3 \ V(e - U + U p - 6V) V(e + U p -7V) (c - U + U p - 6V)(e + U p - TV) J ' V '
2nd neighbors only - one additional —a electron at either the initial site j or final site I and one at the intermediate
site i
_ t 2 t' f 1 1 2 1
t2f ~ 3 \V(e + U p -6V) ~ V(e-U + U P -W) + (e + U p - 6V)(e - U + U p - W) j ' ( ' '
2nd neighbors only - one additional —a electron at either the initial site j or final site I and two at the intermediate
site i
t 2 t' f 1 1
~T\V(e + U p - 5V) ~ V(e-U + U p - AV) ' (e + U p - 5V)(e - U + U p - AV)
^9 = — \t7TTTT} —-TTT-^TTT-r 7777 + 77^7 ^77777 tt I tt 7777 ^ ■ (34)
7
Other elements
Transfer of a pair to a nearest neighbor site
APT = ~2 { (e - U + U p - bV) 2 (e + U p - bV) + (e - U + U p - bV)(e + U p - bV) 2 } " (35)
Transfer of a pair to a second neighbor site =0(t 4 t' 2 ).
Formation/breaking of a pair - the pair is at site I, no other electrons at final (pair breaking) /initial (pair
formation) sites j and i
iFBa
{e + U p - bV)(e -U + U p -bV)(e + U p - W)
t 4 r 3 i
t 4 r i
+
8(e - U + U p - 3V)(e + U p - bV) {e-U + U P -4V e + U,
p -6v]
t 4
4(e + U P - 5V) 2 (e - U + U p - 4V) 4(e + U p - 5V)(e - U + U p - W) 2
Formation/breaking of a pair - with minus spin electrons at both sites i and j
t
4(e - U + U p - W) 2 {e-U + U P -5V ' e + U p -5V
Exchange of two electrons with opposite spin - nearest neighbor case
A FBc - — tT-Tt ct7T2 \ ~ TTT-Tt Fi7 + — — cT7 f ■ ( 38 )
tH' f 4 1
4(e-U + U p - 5V)(e + U P - 5V) 2 8(e-U + U p - W) 2 (e + U P - W)
t A 3t 4 t 4
A{e-U + Up-W) 2 (e + U p -6V) 4(e - U + U p - AV){e + U p - W) 2 A{e + U p -Wf
3^
4(e + U P - 6V)(e + U p - 5F)(e - (7 + U p - bV)
3 1
+
t 4
+
A{e-U + U p -5V)(e-U + U P -AV) [e + U p -6V e + U p -bV
Two electrons moving to neighboring sites - with a minus spin electron at site j
AcMb ~ 8(e - U + U p - W) 2 { e + U p -W + e -U + U p -bV< (41)
(36)
4(e + U P - 6V) 2 \ e + Up-5V e-U + U p -bV _
Formation/breaking of a pair - with a minus spin electron either at site i or at site j
t 4 ( 1 1
A FBb = j{ {e + Up _ Qvfie -U + Up-W) + (e + Up-6V)(e-U + Up- W) 2 ^ (3?)
t 4 it 4
XE ~ 2(e + U P - 5V) 2 (e - U + U p - bV) + 2(e + U P - bV)(e - U + U p - W) 2 (39)
f 1 10 )
+ 30 V{e - U + Up - 6V)(e + U p - bV) \ e + U p -bV + e - U + U p - bV J
30(e + U p - bV) 2 (e - U + U p - bV) \ e - U + U p - 6V V
Exchange of two electrons with opposite spin - second neighbor case = 0(t 4 t' 2 ).
Two electrons moving to neighboring sites - empty lattice case
r -M 4 r i_
CMa - 4U ^_ U + Up _ 5V ^ e + Up _ 5V ) + 8(e-U + U p -bV) 2 \U + e + U p -bV' ' ( '
8
t 4
+ -
8(e + U p -6V) 2 [e + U p -5V e - U + U p - W
i 4 r i 4
4(e + U p - 5V)(e - U + U p - W) \ e + U p - 6V e-U + U P -4V
Two electrons moving to neighboring sites - with a minus spin electron at site i
A ° Mc 8(e-U + U p - W) 2 { e + U p -5V + e-U + U p -W ^
t 4 f 1 3
2(e + U p -W)(e-U + U p -W)(e-U + U p -AV) 8(e - E7 + U p - W) 2 \ e - U + U p - 6V e + U p - 6V
4(e + U p - 6V){e - U + U p - W) {e-U + U P -W e-U + U P -AV
t 4 f 1 3
4(e - U + U p - 4V)(e - U + U p - 5V) { e-U + U P -3V e + U p -6V
Two electrons moving to neighboring sites - with minus spin electrons at both sites i and j
^ 3^
CMd 2(e-U + U p - W) 3 + 4(e + U P - 5V) 2 (e - U + U p - 5V) + 4(e + U P - 5V)(e -U + U p - 5V) 2 [ '
t 4 it 4
A(e-U + U p -W) 2 (e-U + U P -5V) 4(e - U + U p - W)(e - U + U p - 5V) 2 '
Superexchange between two Cu sites - one electron at site j and a minus spin electron at site I
AsXa = ~4(e - U + U p - 5V) { (e + U p - bV) 2 ~ (e-U- W) 2 } (44)
3t 4 r i i
4(e - U + U p - 5V) 2 \e+U p -5V e-U-W
Superexchange between two Cu sites - two electrons at site j and one electron at site I
AsXb = - (e + Up-Wr • (45)
Superexchange between two Cu sites - one electron at site j only, and no electron at site 1, i.e. a renormalization of
the site j energy, due to site / being empty
t 4
A S Xc = - T — Tt • (46)
(e + U p — QV) A
Typically 7]
e = 3.6eK t = 1.3eV, t' = QMeV, U = 10.5eV, U p = AeV, V = l.2eV, V = . (47)
We emphasize that higher order terms are, in principle, of similar magnitude as the terms shown. Energy level
degeneracies, due to finite 0-0 hopping, appear in fifth order of perturbation theory, restricting the present for-
mulation. Similar issues arose in the original work of CSO (l2j . One should come up with a modified procedure,
possibly including an energy diagonalization in the vicinity of the Cu atom, as in js| . Of course it is possible that the
series generated are asymptotic anyway. Then the coefficients of the terms shown can be taken as merely effective
parameters.
It is interesting that for parameter values close to the "typical" ones, factors such as
e + U p -mV , m = 5,6 , (48)
may become very small in magnitude, which yields increased values of the respective interaction amplitudes A. It
turns out that some effective hopping elements increase at least equally fast in that case, so that the ratios (A/t) e ff
9
are finite. However, this picture may be helpful in understanding why certain e.g. 2-particle processes are important
in the CuC>2 planes, as opposed to other lattices with different values of the original parameters. Otherwise put, what
is so special about the CuO-i planes.
IV. GROUND STATE OF THE HAMILTONIAN
We can treat H in the Hartree-Fock-Bogoliubov approximation, with the expectation values of four operator
products given by
< C1C2C3C4 >= d 12 d 3 4 ~ ^13^24 + ^14^23 , (49)
where =< Ci^ ai Cj_ aj > are numbers. Yet another obvious approximation is the replacement of the operators n.; CT
by their expectation values (else we would encounter expectaion values of six, instead of four, operator products).
In order to find the ground state numerically, we minimize H with a fixed total number of particles N = a n^ a .
This procedure requires a highly sophisticated optimization solver, able to handle several thousands of variables,
with adequate constraints on their values; overall a non-trivial task jl3| . In our implementation, we only looked at
non-magnetized solutions.
As a first approach, we make one further simplification, taking every fermion operator as a complex number (thus
having only 4 real numbers per lattice site - c.f. below). In short, within this approach, we obtained checkerboard
states with periods equal to 3 by 3 and 5 by 5 (not 4) lattice sites, stripes, pure antiferromagnetic states, local pairs
and mixtures thereof. E.g. an x — y anisotropic checkerboard state is a mixture with a stripe state. These were found
for filling factors n — N/V = 0.8 — 1.2 (V the system volume). The nature of the ground state is mostly determined
through the values of U and the effective interaction and hopping parameters, rather than the filling (of course the
latter dictates the values of the original Cu02 plane parameters).
For a more complete solution, we should take all <iy above as independent parameters. This amounts to 72
real numbers per lattice site (with symmetry effects taken into consideration), making the problem very demanding
computationally. The presentation of further results is postponed for a future version of this work.
Yet another route to the ground state is through the new exact variational wavefunctions which sustain superfluidity
0-
The author is indebted to Gregory Psaltakis for numerous discussions.
APPENDIX A
Here we give explicit expressions for the first few factors I n of eq. (fTS"|)
2 ( E12 E23) ' ^ ^
7 3 = i(V — + —) ■ (51)
3 V £12^23 -E12-E34 E23E34J
h = l -( ^ + ^ ^ + 1 ) , (52)
8 \ Ei 2 E23E 3i Ei 2 E 2 3E i5 E^E^E^ E 2 3E3^E^ J
I _ _L_ / 1 4 + 6 4 + 1 \ (53 ^
30 \E12E23E34Ei5 E12E23E34E5Q E12E23E45E56 E12E34E45E56 E23E34E45E56 )
1 f 1 5 10 10
h = — 1 ^t~e. 1 ( 54 )
144 \ E12E23E34E45E5Q E12E23E34E45EQ7 Ei2E 2 3E34E^Eq7 E12E23E4SE5QEQ7
5 1
+
EyiE34E4^E^Eqi i?23-£'34£ , 45-E , 56-E , 67
where E i} = E kz - E kj .
* e-mail : kast@iesl.forth.gr
[1] V. J. Emery, Phys. Rev. Lett. 58, 2794 (1987).
10
[2] F.C. Zhang and T.M. Rice, Phys. Rev. B 37, 3759 (1988).
[3] J. Zaanen and A.M. Oles, Phys. Rev. B 37, 9423 (1988).
[4] V.J. Emery and G. Reiter, Phys. Rev. B 38, 11938 (1988).
[5] M. Roger and J.M. Delrieu, Phys. Rev. B 39, 2299 (1989).
[6] H. Eskes and G.A. Sawatzky, Phys. Rev. B 44, 9656 (1991).
[7] M.S. Hybertsen, E.B. Stechel, W.M.C. Foulkes and M. Schluter, Phys. Rev. B 45, 10032 (1992).
[8] V.I. Belinicher and A.L. Chernyshev, Phys. Rev. B 49, 9746 (1994).
[9] L. F. Feiner, J. H. Jefferson, and R. Raimondi, Phys. Rev. B 53, 8751 (1996).
[10] B.E.C. Koltenbah and R. Joynt, Rep. Prog. Phys. 60, 23 (1997).
[11] A. L. Chernyshev, D. Galanakis, P. Phillips, A. V. Rozhkov, and A.-M. S. Tremblay, Phys. Rev. B 70, 235111 (2004).
[12] K.A. Chao, J. Spalek and A.M. Oles, Phys. Rev. B 18, 3453 (1978).
[13] We used the optimizer L-BFGS-B from C. Zhu, R. H. Byrd and J. Nocedal, ACM Trans. Math. Soft. 23, 550 (1997).
[14] G. Kastrinakis, arxiv e-print to appear in Jan. 2009.
