Interplay between Hubbard interaction and charge transfer energy in three-orbital

Emery model
Yan Peng1 and Mi Jiang1, 2, ∗
1
School of Physical Science and Technology, Soochow University, Suzhou 215006, China
2
Jiangsu Key Laboratory of Frontier Material Physics and Devices, Soochow University, Suzhou 215006, China
We use the numerically unbiased determinant quantum Monte Carlo (DQMC) method to
systematically investigate the three-orbital Emery model in the normal state in a wide range of
local interactions, charge transfer energy, and doping levels. We focus on the influence of the
onsite Hubbard Udd and charge transfer energy scale ϵp on the electronic properties via the orbital
occupancies, local moments, spin correlations, and spectral properties. Rich features of the orbital-
arXiv:2509.11028v1 [[Link]-el] 14 Sep 2025

resolved local and momentum-dependent spectra are revealed to associate with the possible Zhang-
Rice singlet (ZRS) breakdown reflected by the peak splitting near the Fermi level in the heavily
overdoped regime. Moreover, the pseudogap features at small charge transfer energy scale (relevant
to cuprates) are shown to diminish at larger ϵp , which implies the weakening or absence of the
pseudogap in the infinite-layer nickelates. Besides, an optimal value of ϵp is identified for maximizing
the antiferromagnetic (AFM) spin correlations. Our large-scale simulations provide new insights on
the well-established Emery model, particularly in the regime of heavily overdoped and/or large
charge transfer energy scale.

I. INTRODUCTION

Cuprate high-temperature superconductors (SC) have

remained a subject of extensive research since their
discovery in the 1980s1 . Other emergent phenomena
such as pseudogap, stripe phase, and strange metal
behavior2–5 render its underlying physical mechanism
even more elusive. Due to the quasi-two-dimensional
structure of Cu-O planes and strong local interaction
on Cu sites, an effective single-orbital model which
originates from a more involved three-orbital Emery
model has been proposed to explain the low-energy
physics dominated by the well-known Zhang-Rice singlet
(ZRS)6 . However, the omission of oxygen degrees
of freedom makes the regime of its validity unclear,
especially based on the experimental fact that the
cuprates are charge-transfer insulators (CTI) rather than
Mott-Hubbard insulators (MHI) in essence 7 . Recent
experimental and theoretical studies on cuprates have FIG. 1. A schematic illustration of a Cu-dx2 −y2 orbital and
also challenged the applicability of the single-orbital its four nearest-neighbor O-px/y orbitals. Red (blue) color
model in the overdoped regime8–10 . Hence, the three- indicates positive (negative) phase factor. The unit cell is
orbital Emery model11 explicitly including 3dx2 −y2 , 2px , outlined by the dashed box.
and 2py orbitals in Cu-O plane, is more close to the
realistic physical picture without assuming the existence
of ZRS in advance.
correlation regime and casted doubt on the validity
The three-orbital model has been applied to investigate of the ZRS approximation when the charge transfer
cuprate SC since its discovery and already been energy enters into the large regime. Owing to
shown to be plausible framework. For example, the improved computational capabilities, in recent
previous work12,13 uncovered the strong suppression years, there is growing support for the emergence
of the antiferromagnetic state upon doping by both of superconductivity4 , pseudogap2,3 , and density wave
exact diagonalization (ED) and dynamical mean- orders5,16,17 within the three-orbital model. Another key
field theory (DMFT). Contemporaneously, Guerrero et motivation arises from the recently discovered infinite-
al.14 used the constrained-path Monte Carlo method layer nickelate high-Tc superconductors18 , which have a
to demonstrate that the d-wave pairing correlations larger charge transfer energy19–21 than cuprates and are
dominate the extended s-wave. Medici et al.15 studied closer to a competing regime of Mott-Hubbard versus
the doping asymmetry of ZRS using DMFT. They charge- transfer dominance22,23 , in spite of the claimed
pointed out that the cuprates are in an intermediate importance of other orbitals like interstitial s orbital24–26 .
 2

In this work, we adopt the determinant quantum x̂ and ŷ are the in-plane unit cell basis vectors. Similarly,
Monte Carlo (DQMC) method to systematically ξjj ′ = 1 for j ′ = j + x̂2 + ŷ2 or j ′ = j − x̂2 − ŷ2 and ξjj ′ = 0
investigate the Emery model in the normal state in a wide for j ′ = j + x̂2 − ŷ2 or j ′ = j + x̂2 − ŷ2 . Both the unit cell
range of local interactions, charge transfer energy, and and the sign of the hopping amplitude are illustrated in
doping levels in a lattice size larger than before15,27–29 . Figure 1. Due to gauge invariance, this choice of signs is
In particular, except for a broad range of the Cu local not unique as mentioned in other studies30–32 .
interaction Udd , we extend the O site energy ϵp from the
The interaction term Û describes the onsite repulsion
conventional charge transfer regime relevant for cuprates
on the holes of dx2 −y2 or px/y orbitals. In the limit
to the Mott-Hubbard regime in the Zaanen-Sawatzky-
of tpp = 0, the amplitude of Udd together with the
Allen scenario7 . Besides, we push the doping level to
charge transfer energy ∆ = ϵp − ϵd significantly affect the
highly overdoped regime.
insulating behavior of the ground state at half-filling33,34 .
This paper is organized as follows: Section II presents
For Udd < ∆ it is a Mott-Hubbard insulator whereas for
the three-orbital Emery model and the basic principle
Udd > ∆ it is a charge transfer insulator7 . Cuprates fall
of DQMC methodology. Then in Section III, we first
into the latter category.
analyze the density distribution and spectral functions
to investigate the quasiparticle behavior and band According to the canonical parameter set5,35 , we
renormalization. Subsequently, we examine the magnetic rescale our parameter set as tpd = 1.0, tpp = 0.4, ϵd = 0
correlations to capture the essential collective behavior and Udd from 4.0 to 8.0, ϵp from 2.0 to 6.0 for convenience
under different Udd and ϵp combinations. Finally, so that tpd = 1.0 serves as the energy unit. In the hole
Section IV summarizes our work. language, the half-filling is defined as ⟨ntot ⟩ = 1 and hole
(electron) doping corresponds to ⟨ntot ⟩ > 1 (< 1).
We remark that the omission of Upp alleviates the
II. MODEL AND METHOD sign problem, allowing us to access larger lattice sizes
up to 8 × 8 (and even 12 × 12 not shown here) than
The three-orbital Emery model involving the Cu- previous works32,35 at reasonably low temperatures.
3dx2 −y2 , O-2px , and O-2py orbitals, with all onsite Unless otherwise specified, the results shown below are
interactions taken into account, reads as for 8 × 8 lattice and the inverse temperature βtpd = 10.0.
Apart from the investigation on the interplay between
Ĥ = Eˆs + K̂ pd + K̂ pp + Û Udd and ϵp , we also examined the role of Upp in a smaller
lattice size, which was found to not obviously affect the
X X p
Eˆs = (ϵd − µ) n̂diσ + (ϵp − µ) n̂jσ
iσ jσ
spin-spin correlations35 . Nonetheless, we will show that
the low energy excitations obtained from the spectral
ˆ†
X
K̂ pd
= tij
pd (diσ p̂jσ + h.c.) functions show strong dependence on Upp in the hole
⟨ij⟩σ doping regime.
X †′
K̂ pp = tjj
pp (p̂jσ p̂j ′ σ + h.c.)
To fully take into account all the energy scales on
⟨jj ′ ⟩σ the equal footing, we use the well established numerical
X X technique of finite temperature determinant Quantum
Û = Udd n̂i↑ n̂i↓ + Upp n̂j↑ n̂j↓ , (1) Monte Carlo (DQMC)36 . As a celebrated computational
i j method, DQMC provides a numerically unbiased solution
in the presence of strong correlations.
where Ê s represents the onsite energies of 3dx2 −y2 orbital
at site i and 2px/y orbital of site j, where n̂diσ (n̂pjσ ) is In order to deal with the ill-posed problem caused
the hole density operator for the d (p) orbital, whose on- by the inversion of the fermionic imaginary-time Green
site energy is ϵd (ϵp ). The chemical potential µ controls function to obtain the spectral function via
the total occupancy. The kinetic energy terms K̂ pd and
+∞
dω e−ωτ A(k, ω)
Z
K̂ pp describe the nearest-neighbor (NN) Cu-O and O-
G(k, τ ) = (3)
O hoppings, denoted by ⟨ij⟩ or ⟨jj ′ ⟩, in corresponding −∞ 2π 1 + e−βω
order. Specifically, we choose the hole language so that
dˆ†iσ (dˆiσ ) creates (annihilates) a hole with spin σ on a we adopt the maximum entropy analytic continuation
d orbital at site i. The same applies to the p-orbital′ method (MaxEnt) to extract the least biased spectral
operators. Besides, the hopping integrals tij pd and tpp
jj
function from all the feasible solutions37 . We examine
take the convention as both the density of states (DOS) and the orbital- and
tij ηij momentum-resolved spectral function Aα (k, ω). Based
pd = tpd (−1)
′
on the equal-time or the time-dependent Green function,
tjj
pp = tpp (−1)
ξjj ′
(2) substantial physical quantities, such as the spin-spin
correlation and the single-particle spectral functions,
with the phase convention ηij = 1 for j = i+ x̂2 or j = i− ŷ2 allow us to access a thorough and comprehensive
and ηij = 0 for j = i − x̂2 or j = i + ŷ2 , where the vectors understanding of the Emery model.
 3

2.00
1.75 (a) p = 2.0
III. RESULTS

A. Orbital-resolved density distribution

1.50
We start by the variation of the total density ⟨ntot ⟩ 1.25
1.00

ntotal
with µ, Udd , and ϵp . As illustrated in Fig. 2(a), for
small ϵp = 2.0, due to the dynamical hopping through
the Oxygen sublattice and the finite temperature, the
0.75
system is insufficient to open an energy gap, indicating 0.50 Udd=4.0
a metallic behavior at half-filling. In this situation, Udd 0.25 Udd=6.0
always suppresses the total filling, which has a maximal Udd=8.0
effect near half-filling, reflecting the natural role of Udd . 0.002.0 1.5 1.0 0.5 0.0 0.5 1.0
In contrast, a larger ϵp = 6.0 induces clear gaps shown 2.00
in Fig. 2(b), implying the enlarged effective Udd with
1.75 (b) p = 6.0
increasing ϵp .
In order to further reveal the electron-hole 1.50
asymmetry38 , Fig. 3 shows the mutual dependence 1.25
between ⟨nCu ⟩ and ⟨nO ⟩ with varying Udd or ϵp . The
1.00

ntotal
gray dashed line denotes the half-filling case. We
emphasize that here ⟨nO ⟩ is the sum of the hole densities 0.75
of Ox and Oy orbitals. The same applies to the O 0.50
spectra discussed in the following section. The panel (a)
indicates that the turning point at half-filling becomes 0.25
sharper and the slope on the electron doping side grows 0.00 1 0 1 2 3 4 5
quickly, which signifies preferential doping onto Cu.
Obviously, as a direct result of larger ϵp , the critical
⟨nO ⟩ of the turning point shifts to smaller values and the FIG. 2. Total filling ⟨ntot ⟩ versus the chemical potential µ for
Cu–O density asymmetry becomes more pronounced. In fixed ϵp = 2.0 (a) or =6.0 (b) with varying Udd .
panel (b), at a fixed large ϵp = 6.0, similar to Fig. 2,
Udd has almost no effect on the charge distribution on
the electron-doping side and pushes more holes into
O orbitals. These behaviors vividly demonstrate the Fermi level lie the ZRS and UHB, respectively, despite
influence of Udd or ϵp on the charge distribution in the the UHB exhibits a two-peak structure. Limited by the
three-orbital model. Specifically, Udd prohibits double still high temperature scale, the ZRT and LHB band is
occupancy on the Cu orbitals, while a lower ϵp than thermally broadened and merge into the background.
Udd allows holes to avoid energy penalties by occupying When holes are doped into the system, the broad O
the O orbitals, hence leading to a strong asymmetry band and UHB both shift to the right linearly with hole
between electron and hole doping. doping, whereas the ZRS remains close to the Fermi level.
On the other hand, electron doping shifts the O band
and UHB to the left and rapidly suppresses the ‘ZRS’
B. Orbital-resolved local density of states (LDOS) to being barely visible. Owing to the chemical potential
shift, both peak features near the Fermi level are now
The orbital-resolved spectral functions can be interpreted as part of the UHB. The low-energy spectrum
obtained via analytic continuation of the imaginary- develops a splitting feature at a hole doping of ∼ 0.2,
time Green functions, which encode rich information which evolves into more and more complex structure e.g.
about quasiparticle excitations and gap features in at ∼ 0.6 doping. Notice that not only the single ZRS peak
corresponding orbitals. Fig. 4(a)–(c) display the impact at small dopings splits into many irregular peaks; but also
on LDOS caused by doping, ϵp and Udd , separately. In part of the O spectra shift toward a higher energy, which
Fig. 4(a), we display a fairly large density range from 0.4 hints as the breakdown of ZRS at heavliy hole doped
to 1.6, i.e., 0.6 for both electron and hole doping. The systems8 . At the electron doping side of ∼ 0.6 doping,
value of Udd = 6.0 and ϵp = 3.0 are set to be relevant for the spectra shows no anomalous feature which retains
cuprates and Fig. 4(d)-(e) further explore the influence high Cu-O hybridization. As one of our major findings,
of Upp . the possible ZRS breakdown is consistent with recent
At half-filling, the system is insufficent to open a experiment39 , which showed signatures of an additional
complete energy gap near Fermi level due to relatively O K-edge excitation above the Fermi level in extremely
low Udd and ϵp as well as our moderate simulated overdoped La2−x Srx CuO4 (up to x = 0.6).
temperature scale. The first noticeable feature is the Next we display the role of ϵp and Udd in modifying
broad blue O band near -4 eV. To the left and right of the the LDOS for a selected hole doping level of 0.2 in
 4

1.2 the density in the range of 0.6 ∼ 1.2 density successfully

(a) Udd = 8.0 reproduce the low energy feature illustrated by Moritz et
al.43 in the single-band Hubbard model, although the QP
0.9 in our LDOS is less coherent due to the high simulating
temperature. This may reflect the validity of the low-
energy physics of the single-band model at low doping.
0.6
nCu

The splitting feature motivates us to further check the

impact of the onsite repulsion Upp on O sites. Due to
0.3 p=2.0 the limitation of the sign problem, we perform this at
p=4.0 Udd = 6.0 and β = 10.0. When Upp is taken into account,
p=6.0 as illustrated in Fig. 4(d) and (e), it significantly pushes
0.00.0 0.2 0.4 0.6 0.8 1.0
the onset doping of low energy splitting to a higher level.
In Fig. 4(e), the low energy peak of Cu remains intact
(b) p = 6.0 at ⟨ntot⟩ = 1.43 when Upp = 2.0, though there is more
pronounced O spectral weight transfer to ∼ 1.0 eV as the
1.2 doping level increasing. In other words, the ZRS is more
robust with hole doping at finite Upp .
0.9 Moreover, increasing Upp appears to drive the UHB
nCu

closer to the Fermi level by comparing Fig. 4(a) with

0.6 (e). Regardless of the value of Upp , within our parameter
range, the UHB in the half-filled LDOS consistently
Udd=4.0
0.3 Udd=6.0
exhibits a double-peak structure. Such behavior may
Udd=8.0 indicate either additional excitations or a nontrivial
0.00.0 redistribution of spectral weight. On the electron-doped
0.2 0.4 0.6 0.8 side, there is no clear indication that the spectrum is
nO significantly affected by Upp . Since the remaining holes
primarily lie on the Cu orbitals in this situation, double
FIG. 3. The hole density on Cu orbital ⟨nCu ⟩ versus that occupancy on the O orbitals, which inherently have a
on O orbital ⟨nO ⟩ with ϵp or Udd being fixed. The dashed
lower density due to higher onsite energy, is unlikely
line denotes the half-filling ⟨ntot ⟩ = 1.0 case. The dotted line
indicates ⟨nCu ⟩ = ⟨nO ⟩, i.e. equal occupancies on Cu and O. to occur. The two low-energy peaks, linked to the
UHB, show contrasting behavior with increasing electron
doping: the right peak diminishes and the left peak
Fig. 4(b) and (c), respectively. In Fig. 4(b), Udd is fixed becomes more prominent.
at 6.0 while ϵp varies from 2.0 to 6.0; in Fig. 4(c), ϵp To identify more details on the spectral features
is fixed at 4.0 while Udd varies from 4.0 to 8.0. Both near the UHB at half-filling, Fig. 4(f) illustrates the
increasing ϵp and Udd effectively enlarge the distance temperature β = 1/T evolution corresponding to the
between the UHB and the broad O band, with the effect situation in Fig. 4(a). As the temperature cools down
of ϵp being significantly more pronounced which reflects (increasing β), the insulating gap gradually becomes
the CTI nature of our undoped model. Interestingly, clearer. The strong Cu-O hybridization reflected by the
either increasing ϵp or decreasing Udd can enhance the coincidence of their spectral peaks is maintained at each
splitting near the Fermi level. As shown in Fig. 3(a), β value. Notably, the anomalous features above ω = 0
large ϵp induces more deviated Cu-O hole distribution of Cu spectra emerges from β = 8.0 and persists up to
so that restricts the stability of ZRS and leads to the β = 15.0. Conversely, the O spectra never shows any
peak splitting in Fig. 4(b), although a strong ϵp enlarges anomalous behavior.
the effective repulsion on Cu to promote the localized
behavior. The same reasoning applies for small Udd by
combining Fig. 3(b) and Fig. 4(c). C. k-resolved spectra: pseudogap feature
In contrast to the 0.6 hole doping, the prominent Cu-O
hybridization is preserved at 0.2 doping in Fig. 4(b) and The momentum dependent single-particle spectral
(c), which is evidenced by the coincidence of the spectral function Aα (k, ω) can manifest more information than
peaks of Cu and O. This may point to a distinct origin of the local DOS. Prompted by the anomalous LDOS
the spectral splitting from the “ZRS breakdown” in the near the Fermi level at high doping levels, we further
heavily overdoped regime shown in Fig. 4(a). Extensive examine the structure of Aα (k, ω) in Fig. 5. Here
earlier theoretical investigations have established the we choose two representative hole doping levels, 0.05
appearance of a new quasiparticle peak (QP) in the and 0.6, and two typical charge transfer energy scales,
vicinity of the Fermi level induced by hole doping, which ϵp = 3.0, 6.0, for our analysis. The first notable
is commonly interpreted as a dynamical spectral weight feature is the more pronounced spectral weight near
transfer9,40–42 . Specifically, in Fig 4(a), our LDOS of the Fermi level along the nodal (N) than the anti-
 5

(a) Udd = 6.0 Cu

n (b) Udd = 6.0 (c) p = 4.0
p = 3.0 ntot = 1.2 p Udd
O 2.0 4.0
Upp = 0.0 ntot = 1.2
1.6
1.51
1.4 3.0 5.0
1.35
1.2
1.08 4.0 6.0
1.0
0.86
0.8 5.0 7.0
0.69
0.6
0.4 6.0 8.0
6 4 2 0 2 4 6 4 2 0 2 4 6 4 2 0 2 4
(d) Udd = 6.0 n (e) Udd = 6.0 n (f) Udd = 6.0
p = 3.0 p = 3.0 p = 3.0
Upp = 1.0 Upp = 2.0 15.0
1.6 1.57 Upp = 0.0
1.51 1.48 ntot = 1.0
1.46 1.43 12.0
1.41 1.39
1.32 1.29 10.0
1.23 1.2
1.14 1.12
1.0 1.0 8.0
0.95 0.94
0.86 0.84
0.68 0.67 5.0
6 4 2 0 2 4 6 4 2 0 2 4 6 4 2 0 2 4

FIG. 4. Local density of states (LDOS) as a function of (a) doping level, (b) ϵp , and (c) Udd . The varied parameter values
are indicated to the right of each panel. The spectra of O orbital is summed over the x- and y-direction. The orange triangles
denote the location of the UHB.

(a) Udd = 6.0 (b) Udd = 6.0

p = 3.0
Cu p = 6.0
ntot = 1.05 O ntot = 1.05

X X
M M

6 4 2 0 2 4 6 4 2 0 2 4
(c) Udd = 6.0 (d) Udd = 6.0
p = 3.0 p = 6.0
ntot = 1.6 ntot = 1.6

X X
M M

6 4 2 0 2 4 6 4 2 0 2 4

FIG. 5. The orbital-resolved spectral function Aα (k, ω) along the high-symmetry path Γ-M -X-Γ in the Brillouin zone. Following
the discussion of LDOS, Udd is kept constant at 6.0 as well. The spectral weight of O at Γ point is truncated for better clarity.
Owing to the momentum-space anisotropy of the spectral function, the O spectrum is obtained by summing the contributions
along the x- and y-directions.

nodal (AN) direction in Fig. 5(a), which is widely the underdoped regime of cuprates2,3,44 . With the
considered to be characteristic of the pseudogap in constraint of QMC sign problem, we are unable to reveal
 6

1.0
a more pronounced pseudogap feature by reducing the
temperature so that a decisive conclusion cannot be (a) Udd = 8.0
drawn at present. Nevertheless, for ϵp = 6.0 in Fig. 5(b), 0.8 p=2.0
the strong momentum differentiation for smaller ϵp = 3.0 p=4.0
largely weakens. Specifically, the low-energy peaks are 0.6 p=6.0
predominantly located around (π, π) and the spectra do

m2
not show any pseudogap feature. This difference implies
the weakening or absence of the pseudogap in infinite- 0.4
layer nickelates due to its large charge transfer energy,
which also results in lower superconducting Tc than 0.2
cuprates and the absence of long-range antiferromagnetic
magnetic order19 . Additionally, the peak broadening is 0.0
more evident than the ϵp = 3.0 case, indicating a larger 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
1.0
scattering arising from stronger correlation31 .
At the heavily overdoped regime, as illustrated in
(b) p = 4.0
Fig. 5(c) and (d), the low-energy peaks become much 0.8 Udd=4.0
more coherent and the UHB is hardly visible. Compared Udd=6.0
to the underdoped regime shown in Fig. 5(a-b), the 0.6 Udd=8.0
difference between panels (c) and (d) is generally less

m2
obvious, reflecting the minor role of large ϵp in the 0.4
heavily overdoped regime. Instead, one common feature
of ϵp = 3.0, 6.0 lies that both show stronger zero- 0.2
energy excitations along AN than the N direction. This
phenomenon has been widely reported across various
0.0
cuprates accompanied by the Lifshitz transition of the 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Fermi surface45–48 , though the doping level here is much ntotal
higher. Consistent with the previous research35 , the
strong Cu-O hybridization feature near the Fermi level FIG. 6. Orbital-resolved local moment ⟨m2 ⟩α versus ⟨ntot ⟩
with Udd or ϵp varies. Here α =Cu/O is distinguished by
persists up to a doping level of 0.6 when ϵp = 3.0.
solid/dashed line.
However, a notable suppression of hybridization can be
identified in Fig. 5(d). There is no multi-peak feature
near the Fermi level at any k-point, suggesting that
arising from the effectively larger Udd . The local moment
the multi-peak structure in LDOS of Fig. 4 originates
at both electron and hole doping sides show nearly linear
from momentum integration over distinct regions of the
dependence with strong slope asymmetry. At the hole
Brillouin zone. The additional Cu peaks near the Fermi
doping side, the decreasing ⟨m2 ⟩Cu with doping when ϵp
level in Fig. 4(a) at 0.6 hole doping primarily originates
is larger than 2.0 suggests that the magnetic correlations
around Γ. On the other hand, the extra O peak above
would also be weakened. Although ⟨m2 ⟩O obviously
the Fermi level emerges near M , hence leading to a
increases at the hole doping side, its amplitude is much
complicated peak structure near the Fermi level.
smaller compared to Cu and contributes less to the
magnetic response in the system, as clearly evidenced in
precious DQMC study35 . On the other hand, Fig. 6(b)
D. Magnetic properties indicates the less significant impact of Udd , especially at
the electron doping side. The onsite repulsion on the
From now on, we concentrate on two-particle Cu orbital effectively limits the double occupancy and
quantities such as the spin-spin correlation and the spin thereby enhances ⟨m2 ⟩Cu at the hole doping side.
structure factor, as both experimental and theoretical In our SU(2)-symmetric system under investigation,
studies49,50 have revealed Néel antiferromagnetic the complete spin rotational symmetry is maintained.
ordering near zero doping. Therefore, a careful Nonetheless, numerically z-component quantities show
examination of how Udd and ϵp influence these quantities less uncertainty so that we adopt the z-component of
is warranted. We first examine the orbital-resolved spin-spin correlation function
local moment, ⟨m2 ⟩α = ⟨(nα α 2
↑ − n↓ ) ⟩, which quantifies
the localized behavior of spins as a precondition for 1 X
nα α
nα α



Sα (l) = i↑ − ni↓ i+l,↑ − ni+l,↓ (4)
the emergence of magnetic order. Since ϵp adjusts the N i
distribution of the doped electron/hole directly, one
can readily anticipate its strong influence on the local and the static spin structure factor
moment. X
On the one hand, as illustrated in Fig. 6(a), increasing Sα (q) = eiq·l Sα (l) (5)
ϵp apparently enhances the local moment on Cu orbital l
 7

0.02
(a) Udd = 8.0 2.00 (a) Udd = 8.0 p=2.0
0.00
1.75 p=4.0
0.02 p=6.0
1.50
0.04

SCu(q)
1.25
SCu(l)

0.06
0.08 1.00
0.04
0.10 p=2.0 0.75
0.02
0.12 0.00 p=4.0 0.50
0.02 p=6.0
0.14 0.0 0.5 1.0 1.5 0.25
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 (0, 0) ( , ) ( , 0) (0, 0)
0.02
(b) p = 4.0 2.00 (b) p = 4.0 Udd=4.0
0.00 Udd=6.0
0.02 1.75 Udd=8.0
0.04 1.50
SCu(l)

SCu(q)
0.06
1.25
0.08 0.06
0.04 1.00
0.10 0.02 Udd=4.0
0.12 0.00 Udd=6.0 0.75
0.02 Udd=8.0 0.50
0.14 0.0 0.5 1.0 1.5
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75
ntotal (0, 0) ( , ) ( , 0) (0, 0)
1.00
FIG. 7. Nearest neighbor spin-spin correlation function (c) Udd = 8.0
SCu (1, 0) versus ⟨ntot ⟩ with Udd or ϵp varying. The inset 0.90 ntot = 1.4
shows the next-nearest neighbor SCu (1, 1). The dashed line
is an indicator of sign change. 0.80
SCu(q)

0.70
to study how the magnetism evolves with respect to
various parameters. The dominant spin-spin correlation
of our model is in Cu-Cu channel, especially the short
0.60 p=2.0
range part SCu (1, 0) and SCu (1, 1), where (1, 0) and (1, 1) p=4.0
denote the spatial separation.
0.50 p=6.0
Fig. 7(a) shows SCu (1, 0) and SCu (1, 1) versus total (0, 0) ( , ) ( , 0) (0, 0)
hole filling. The increasing slope difference between
each side of half-filling with increasing ϵp indicates FIG. 8. Cu-Cu spin structure factor SCu (q) with q along the
the intrinsic electron–hole asymmetry caused by charge high-symmetry path Γ-M -X-Γ in the first Brillouin zone for
transfer energy of our model. Another notable feature various values of Udd or ϵp at (a-b) half-filling or (c) 0.4 hole
is that the larger ϵp causes a smaller critical doping, at doping.
which SCu (1, 0) changes its sign to positive, suggesting
the emerged short-range ferromagnetic correlation51–54 .
The inset shows the next-nearest neighbor SCu (1, 1). from antiferromagnetic to ferromagnetic neighboring
Interestingly, in a wide doping range around half- correlations motivates us to further investigate the
filling, both SCu (1, 0) and SCu (1, 1) reach their maximum parameter dependence of the Cu-orbital’s spin structure
value when ϵp equals to 4.0. This feature implies the factor SCu (q). In Fig. 8(a), the antiferromagnetic
existence of an optimal charge transfer energy scale, ordering vector q = (π, π) dominates at half-filling as
which is reminiscent of an earlier study identifying an expected. Similar to the spin-spin correlation function,
optimal ϵp for the maximal superconducting Tc 32 . SCu (q) exhibits a maximal antiferromagnetic peak at
In addition, Fig. 7(b) indicates that varying Udd an intermediate value of ϵp . The existence of an
does not obviously affect the amplitude of the short optimal ϵp is supported by numerous computational
range magnetic correlation at all doping levels. Both methods32,55,56 . Away from the antiferromagnetic wave-
the peak of (1, 1) and the valley of (1, 0) show strong vector, the existence of optimal ϵp gradually disappears
antiferromagnetic order at half-filling. and the intensity exhibits the monotonic rise as ϵp
The sign change feature reflecting the transition increases, which indicates enhanced spin fluctuations and
 8

that the Cu orbitals are more localized. Again Fig. 8(b) IV. CONCLUSION
reveals the minor modification from Udd . In fact, this
insensitivity to Udd is naturally expected since the local In summary, by employing the large-scale DQMC
moment is robust against varying Udd in Fig. 6(b) as simulations, we have investigated the influence of Udd ,
well as the robustness of neighboring spin correlations in ϵp , Upp , as well as the electron/hole doping in the
Fig. 7(b). three-orbital Emery model on physical quantities such
When the hole doping level reaches 0.4 and the optimal as orbital occupancy, local and k-resolved spectral
ϵp behavior has disappeared in Fig. 7(a), Fig. 8(c) functions, as well as spin correlation functions.
presents the corresponding spin structure factor. Because We concentrate on the difference arising from the large
of the increasing local moment in Fig. 6(a), the overall ϵp that is believed to be relevant to infinite-layer nickelate
intensity is shifted up by enlarging ϵp . Meanwhile, the superconductors19,31 . The pseudogap features at small
AFM dominated peak gradually diminishes and the q charge transfer energy scale (relevant to cuprates) are
distribution becomes more evenly, with possible weak shown to diminish at larger ϵp , which implies the
incommensurate magnetic ordering. Consistent with weakening or absence of the pseudogap in the infinite-
the spin correlations, there is an enhancement of short- layer nickelates. In addition, the spectra of low doping
range ferromagnetic (FM) correlations with increasing ϵp . levels are basically consistent with the characteristic of
These phenomena at high hole doping may indicate an dynamical spectral weight transfer. However, signatures
important role of paramagnon57–59 in describing heavily of ZRS breakdown have been identified via the spectra
doped systems. in the heavily overdoped regime. This undoubtedly
challenges the applicability of the single-band Hubbard
Finally, we provide a brief interpretation for this model. The anomalies in the spectral function motivate
optimal ϵp behavior. Consistent with Cui et al.55 , our further investigation of magnetism. Around the
the local magnetism is naturally strengthened at high antiferromagnetic wave vector (π, π), an optimal ϵp ∼
ϵp because of effectively enlarged Udd . The low 4.0 which gives rise to the largest spin correlation and
AFM correlation at small ϵp is clear because of the structure factor near the half-filling is detected, which
insufficient local moment on Cu orbitals. Nonetheless, is closely related to the superexchange mechanism in
at large ϵp values, the superexchange given in four-order cuprates. At high doping levels, a higher ϵp results in
perturbation as6,19,36,60–62 a stronger short-range FM fluctuation.
All these findings above highlight the pivotal role of the
charge transfer energy in shaping both the spectral and
4t4pd
 
1 1 magnetic responses, and shed light on the applicability of
J= + (6) the three-orbital Emery model as a common framework
∆2 ∆ Udd
for capturing the intertwined phenomena in cuprates
and infinite-layer nickelates31 . In the CTI regime, a
decreases monotonically with increasing ϵp . At ϵp = negligible role of the onsite interactions on investigating
6.0, the remaining superexchange is only about 20% the magnetism is revealed, which may significantly
of the typical value in cuprates. The much weaker mitigate the sign problem in the future.
superexchange results in the diminished spin correlations
and structure factor in spite of a stronger local magnetic
moment. Hence, the competition between the effective V. ACKNOWLEDGMENTS
magnetic moment and the superexchange leads to the
maximum AFM correlation for a moderate ϵp . Since We would like to thank George A. Sawatzky, Xinmao
the unconventional SC is widely considered to be closely Yin, Xiongfang Liu, Rui Peng, and Wenxin Ding for
associated with AFM correlations63–65 , our results may illuminating discussions. We acknowledge the support by
offer some insights for adjusting the magnetism to National Natural Science Foundation of China (NSFC)
optimize the superconducting Tc . Grant No. 12174278.

∗ 3
jiangmi@[Link] Setsuko Tajima, Yuhta Itoh, Katsuya Mizutamari, Shigeki
1
J. G. Bednorz and K. A. Müller, “Possible high Tc Miyasaka, Masamichi Nakajima, Nae Sasaki, Shunpei
superconductivity in the Ba–La–Cu–O system,” Zeitschrift Yamaguchi, Kei-ichi Harada, and Takao Watanabe,
für Physik B Condensed Matter 64, 189–193 (1986). “Correlation between Tc and the pseudogap observed in
2
Wei Wu, Mathias S. Scheurer, Shubhayu Chatterjee, the optical spectra of high Tc superconducting cuprates,”
Subir Sachdev, Antoine Georges, and Michel Ferrero, Journal of the Physical Society of Japan 93, 103701 (2024).
4
“Pseudogap and Fermi-surface topology in the two- Boris Ponsioen, Sangwoo S. Chung, and Philippe Corboz,
dimensional Hubbard model,” Phys. Rev. X 8, 021048 “Superconducting stripes in the hole-doped three-band
(2018). Hubbard model,” Physical Review B 108, 205154 (2023).
 9

5
Peizhi Mai, Benjamin Cohen-Stead, Thomas A Maier, (R= La-Lu),” Physical Review B 102, 205130 (2020).
21
and Steven Johnston, “Fluctuating charge-density- Andreas Kreisel, Brian M Andersen, Astrid T Rømer,
wave correlations in the three-band Hubbard model,” Ilya M Eremin, and Frank Lechermann, “Superconducting
Proceedings of the National Academy of Sciences 121, instabilities in strongly correlated infinite-layer nickelates,”
e2408717121 (2024). Physical Review Letters 129, 077002 (2022).
6 22
F. C. Zhang and T. M. Rice, “Effective Hamiltonian for the Xiang Ding, Yu Fan, Xiaoxiao Wang, Chihao Li, Zhitong
superconducting Cu oxides,” Phys. Rev. B 37, 3759–3761 An, Jiahao Ye, Shenglin Tang, Minyinan Lei, Xingtian
(1988). Sun, Nan Guo, et al., “Cuprate-like electronic structures
7
J Zaanen, GA Sawatzky, and JW Allen, “Band gaps in infinite-layer nickelates with substantial hole dopings,”
and electronic structure of transition-metal compounds,” National Science Review 11, nwae194 (2024).
23
Physical Review Letters 55, 418 (1985). Frank Lechermann, “On the oxygen p states in
8
Gideok Kim, Ksenia S. Rabinovich, Alexander V. Boris, superconducting nickelates,” (2024), arXiv:2410.06891.
24
Alexander N. Yaresko, Y. Eren Suyolcu, Yu-Mi Wu, Yuhao Gu, Sichen Zhu, Xiaoxuan Wang, Jiangping Hu,
Peter A. van Aken, Georg Christiani, Gennady Logvenov, and Hanghui Chen, “A substantial hybridization between
and Bernhard Keimer, “Optical conductivity and correlated Ni-d orbital and itinerant electrons in infinite-
superconductivity in highly overdoped La2−x Cax CuO4 layer nickelates,” Communications Physics 3, 84 (2020).
25
thin films,” PNAS 118 (2021), 10.1073/pnas.2106170118. Hanghui Chen, Alexander Hampel, Jonathan Karp, Frank
9
C-C Chen, M Sentef, YF Kung, CJ Jia, R Thomale, Lechermann, and Andrew J Millis, “Dynamical mean
B Moritz, Arno P Kampf, and TP Devereaux, “Doping field studies of infinite layer nickelates: Physics results
evolution of the oxygen K-edge x-ray absorption spectra and methodological implications,” Frontiers in Physics 10,
of cuprate superconductors using a three-orbital Hubbard 835942 (2022).
26
model,” Physical Review B—Condensed Matter and Chihao Li, Yutong Chen, Xiang Ding, Yezhao Zhuang,
Materials Physics 87, 165144 (2013). Nan Guo, Zhihui Chen, Yu Fan, Jiahao Ye, Zhitong An,
10
Yangmu Li, A Sapkota, PM Lozano, Zengyi Du, Hui Suppanut Sangphet, et al., “Observation of electride-like s
Li, Zebin Wu, Asish K Kundu, RJ Koch, Lijun Wu, states coexisting with correlated d electrons in NdNiO2 ,”
BL Winn, et al., “Strongly overdoped La2−x Srx CuO4 (2025).
27
: Evidence for josephson-coupled grains of strongly Antoine Georges, Gabriel Kotliar, Werner Krauth, and
correlated superconductor,” Physical Review B 106, Marcelo J. Rozenberg, “Dynamical mean-field theory of
224515 (2022). strongly correlated fermion systems and the limit of infinite
11
VJ Emery, “Theory of high-Tc superconductivity in dimensions,” Rev. Mod. Phys. 68, 13–125 (1996).
28
oxides,” Physical Review Letters 58, 2794 (1987). Alexander Weiße and Holger Fehske,
12
R. T. Scalettar, D. J. Scalapino, R. L. Sugar, and S. R. “Exact diagonalization techniques,” in
White, “Antiferromagnetic, charge-transfer, and pairing Computational Many-Particle Physics, edited by Holger
correlations in the three-band Hubbard model,” Physical Fehske, Ralf Schneider, and Alexander Weiße (Springer,
Review B 44, 770–781 (1991). Berlin, Heidelberg, 2008) pp. 529–544.
13 29
Th Maier, MB Zölfl, Th Pruschke, and J Keller, HQ Lin and JE Hirsch, “Pairing in the two-dimensional
“Magnetic properties of the three-band Hubbard model,” hubbard model: An exact diagonalization study,” Physical
The European Physical Journal B-Condensed Matter and Review B 37, 7359 (1988).
30
Complex Systems 7, 377–383 (1999). Mark H. Fischer and Eun-Ah Kim, “Mean-field analysis
14
M. Guerrero, J. E. Gubernatis, and Shiwei Zhang, of intra-unit-cell order in the Emery model of the CuO2
“Quantum Monte Carlo study of hole binding and pairing plane,” Phys. Rev. B 84, 144502 (2011).
31
correlations in the three-band Hubbard model,” Physical Tianzong Mao and Mi Jiang, “Non-Fermi-liquid behavior
Review B 57, 11980–11988 (1998). of the scattering rate in the three-orbital Emery model,”
15
Luca de’ Medici, Xin Wang, Massimo Capone, and Physical Review B 109, 245102 (2024).
32
Andrew J. Millis, “Correlation strength, gaps, and particle- Peizhi Mai, Giovanni Balduzzi, Steven Johnston, and
hole asymmetry in high-Tc cuprates: A dynamical mean Thomas A Maier, “Pairing correlations in the cuprates:
field study of the three-band copper-oxide model,” Phys. A numerical study of the three-band Hubbard model,”
Rev. B 80, 054501 (2009). Physical Review B 103, 144514 (2021).
16 33
Hao Xu, Chia-Min Chung, Mingpu Qin, Ulrich Schollwöck, George Sawatzky and Robert Green, “The explicit role
Steven R. White, and Shiwei Zhang, “Coexistence of anion states in high-valence metal oxides,” Quantum
of superconductivity with partially filled stripes in the Materials: Experiments and Theory Modeling and
Hubbard model,” Science 384, eadh7691 (2024). Simulation 6, 1–36 (2016).
17 34
Hong-Chen Jiang, “Pair density wave in the doped OK Andersen, AI Liechtenstein, O Jepsen, and F Paulsen,
three-band Hubbard model on two-leg square cylinders,” “LDA energy bands, low-energy hamiltonians, t′ , t′′ ,
Physical Review B 107, 214504 (2023). t⊥ (k), and J⊥ ,” Journal of Physics and Chemistry of Solids
18
Danfeng Li, Kyuho Lee, Bai Yang Wang, Motoki Osada, 56, 1573–1591 (1995).
35
Samuel Crossley, Hye Ryoung Lee, Yi Cui, Yasuyuki Y. F. Kung, C.-C. Chen, Yao Wang, E. W. Huang,
Hikita, and Harold Y Hwang, “Superconductivity in an E. A. Nowadnick, B. Moritz, R. T. Scalettar, S. Johnston,
infinite-layer nickelate,” Nature 572, 624–627 (2019). and T. P. Devereaux, “Characterizing the three-orbital
19
Mi Jiang, Mona Berciu, and George A. Sawatzky, “Critical Hubbard model with determinant quantum Monte Carlo,”
nature of the Ni spin state in doped NdNiO2 ,” Phys. Rev. Phys. Rev. B 93, 155166 (2016).
36
Lett. 124, 207004 (2020). R. Blankenbecler, D. J. Scalapino, and R. L. Sugar,
20
Jesse Kapeghian and Antia S Botana, “Electronic “Monte Carlo calculations of coupled boson-fermion
structure and magnetism in infinite-layer nickelates RNiO2 systems. I,” Physical Review D 24, 2278–2286 (1981).
 10

37
Nailong Wu, The maximum entropy method, Vol. 32 communications 5, 3314 (2014).
52
(Springer Science & Business Media, 2012). Bayo Lau, Mona Berciu, and George A Sawatzky, “High-
38
NP Armitage, P Fournier, and RL Greene, “Progress spin polaron in lightly doped CuO2 planes,” Physical
and perspectives on electron-doped cuprates,” Reviews of Review Letters 106, 036401 (2011).
53
Modern Physics 82, 2421–2487 (2010). Bayo Lau, Mona Berciu, and George A Sawatzky,
39
“Private communication with Xinmao Yin,”. “Computational approach to a doped antiferromagnet:
40
G Sordi, GL Reaney, N Kowalski, P Sémon, and A- Correlations between two spin polarons in the lightly
MS Tremblay, “Ambipolar doping of a charge-transfer doped CuO2 plane,” Physical Review B—Condensed
insulator in the Emery model,” Physical Review B 111, Matter and Materials Physics 84, 165102 (2011).
54
045117 (2025). Hadi Ebrahimnejad, George A Sawatzky, and Mona
41
Cédric Weber, Kristjan Haule, and Gabriel Kotliar, Berciu, “The dynamics of a doped hole in a cuprate is
“Apical oxygens and correlation strength in electron- and not controlled by spin fluctuations,” Nature Physics 10,
hole-doped copper oxides,” Phys. Rev. B 82, 125107 951–955 (2014).
55
(2010). Zhi-Hao Cui, Chong Sun, Ushnish Ray, Bo-Xiao Zheng,
42
Xin Wang, Luca de’ Medici, and A. J. Millis, “Theory Qiming Sun, and Garnet Kin-Lic Chan, “Ground-
of oxygen K edge x-ray absorption spectra of cuprates,” state phase diagram of the three-band Hubbard model
Phys. Rev. B 81, 094522 (2010). from density matrix embedding theory,” Physical Review
43
B Moritz, F Schmitt, W Meevasana, S Johnston, Research 2, 043259 (2020).
56
EM Motoyama, M Greven, DH Lu, C Kim, RT Scalettar, Michal Zegrodnik, Andrzej Biborski, Maciej Fidrysiak,
ZX Shen, et al., “Effect of strong correlations on the and Józef Spalek, “Superconductivity in the three-band
high energy anomaly in hole-and electron-doped high- model of cuprates: Variational wave function study and
tc superconductors,” New Journal of Physics 11, 093020 relation to the single-band case,” Physical Review B 99,
(2009). 104511 (2019).
44 57
Fedor Šimkovic IV, Riccardo Rossi, Antoine Georges, and Lichen Wang, Guanhong He, Zichen Yang, Mirian
Michel Ferrero, “Origin and fate of the pseudogap in the Garcia-Fernandez, Abhishek Nag, Kejin Zhou, Matteo
doped Hubbard model,” Science 385, eade9194 (2024). Minola, Matthieu Le Tacon, Bernhard Keimer, Yingying
45
Yong Zhong, Zhuoyu Chen, Su-Di Chen, Ke-Jun Xu, Peng, et al., “Paramagnons and high-temperature
Makoto Hashimoto, Yu He, Shin-ichi Uchida, Donghui superconductivity in a model family of cuprates,” Nature
Lu, Sung-Kwan Mo, and Zhi-Xun Shen, “Differentiated Communications 13, 3163 (2022).
58
roles of Lifshitz transition on thermodynamics and Mathieu Le Tacon, G Ghiringhelli, J Chaloupka,
superconductivity in La2−x Srx CuO4 ,” Proceedings of the M Moretti Sala, V Hinkov, MW Haverkort, Matteo
National Academy of Sciences 119, e2204630119 (2022). Minola, M Bakr, KJ Zhou, S Blanco-Canosa, et al.,
46
Y-G Zhong, J-Y Guan, X Shi, J Zhao, Z-C Rao, C-Y Tang, “Intense paramagnon excitations in a large family of high-
H-J Liu, ZY Weng, ZQ Wang, GD Gu, et al., “Continuous temperature superconductors,” Nature Physics 7, 725–730
doping of a cuprate surface: Insights from in situ angle- (2011).
59
resolved photoemission,” Physical Review B 98, 140507 O. J. Lipscombe, S. M. Hayden, B. Vignolle, D. F.
(2018). McMorrow, and T. G. Perring, “Persistence of high-
47
Masafumi Horio, Kevin Hauser, Yasmine Sassa, Zarina frequency spin fluctuations in overdoped superconducting
Mingazheva, Denys Sutter, Kevin Kramer, Ashley Cook, La2−x Srx CuO4 (x = 0.22),” Phys. Rev. Lett. 99, 067002
Elisabetta Nocerino, Ola Kenji Forslund, Oscar Tjernberg, (2007).
60
et al., “Three-dimensional Fermi surface of overdoped La- Yabin Yu, Guanghan Cao, and Zhengkuan Jiao,
based cuprates,” Physical Review Letters 121, 077004 “Superexchange in the cuprates: a mean-field study,”
(2018). Physica C: Superconductivity 307, 137–144 (1998).
48 61
Ma Yong-Chang, Zhao Jie, An Yu-Kai, and Liu Jan Zaanen and Andrzej M. Oleś, “Canonical
Ji-Wen, “Dominance of antinodal quasiparticles on perturbation theory and the two-band model for high-Tc
the transport properties of heavily overdoped high-Tc superconductors,” Phys. Rev. B 37, 9423–9438 (1988).
62
cuprates: Infrared-reflectance spectra,” Chinese Physics Henk Eskes and John H. Jefferson, “Superexchange in the
Letters 26, 027401 (2009). cuprates,” Phys. Rev. B 48, 9788–9798 (1993).
49 63
A. I. Lichtenstein and M. I. Katsnelson, Hidekazu Mukuda, Sunao Shimizu, Shin-ichiro Tabata,
“Antiferromagnetism and d-wave superconductivity Keita Itohara, Yoshio Kitaoka, Parasharam M Shirage,
in cuprates: A cluster dynamical mean-field theory,” and Akira Iyo, “Superexchange interaction and
Phys. Rev. B 62, R9283–R9286 (2000). magnetic moment in antiferromagnetic high-Tc cuprate
50
Russell A Hart, Pedro M Duarte, Tsung-Lin Yang, superconductors,” Physica C: Superconductivity and its
Xinxing Liu, Thereza Paiva, Ehsan Khatami, Richard T applications 470, S7–S11 (2010).
64
Scalettar, Nandini Trivedi, David A Huse, and Randall G Alexandre Foley, Simon Verret, A-MS Tremblay, and
Hulet, “Observation of antiferromagnetic correlations in David Senechal, “Coexistence of superconductivity and
the Hubbard model with ultracold atoms,” Nature 519, antiferromagnetism in the Hubbard model for cuprates,”
211–214 (2015). Physical Review B 99, 184510 (2019).
51 65
CJ Jia, EA Nowadnick, K Wohlfeld, YF Kung, C-C Chen, NM Plakida, “Theory of antiferromagnetic pairing in
S Johnston, T Tohyama, B Moritz, and TP Devereaux, cuprate superconductors,” Low Temperature Physics 32,
“Persistent spin excitations in doped antiferromagnets 363–374 (2006).
revealed by resonant inelastic light scattering,” Nature