Hole clustering and mutual interplay in three-band Hubbard model

Guang-Ming Zhang; Mi Jiang; Yi-feng Yang

REVIEW 5 major objections 6 minor 18 references

Hole clustering and mutual interplay in three-band Hubbard model

T0 review · 5 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Two doped holes in a three-band Hubbard model, with an intermediate local potential mimicking a Ca vacancy, arrange themselves into a $4\times4$ $d$-$p$ unit-cell cluster that the paper identifies as the building block of cuprate charge…

desk verdict A solid DQMC observation of two-hole clustering in a three-band model is stretched into a building-block narrative; the 4x4 scale is fitted, not predicted. read the letter →

arxiv 2506.00787 v1 pith:RDJHHDKX submitted 2025-06-01 cond-mat.str-el

Mi Jiang , Yi-feng Yang , Guang-Ming Zhang This is my paper

classification cond-mat.str-el PACS 71.10.Fd74.20.Mn

keywords three-bandHubbardmodelcupratesuperconductivity4x4supercellCavacancyholeclusteringdeterminantquantumMonteCarlochargeorderlocaldensityofstates

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that the $4\times4$ supercell seen in scanning tunnelling experiments on hole-doped cuprates is a genuine electronic building block, not a coincidental pattern. Using determinant quantum Monte Carlo on an $8\times8$ three-band $d$-$p$ Hubbard lattice with two doped holes, the authors add a local potential on the four oxygen sites nearest a Ca vacancy. At intermediate strength $V=0.75$, the hole density decays from the central oxygen ring outward and fills a $4\times4$ $d$-$p$ unit-cell region, with low-energy in-gap spectral weight appearing at the Fermi level near the vacancy. The paper argues that these two-hole clusters are the elementary units of the 1/8 charge order, that touching clusters form stripe-like density patterns, and that superconductivity emerges from percolation of such clusters. If this picture is right, the natural low-energy theory of underdoped cuprates starts from interacting $4\times4$ clusters rather than from a clean uniform model.

What carries the argument

The central object is the three-band $d$-$p$ Hubbard model in hole language on an $8\times8$ lattice with periodic boundary conditions. A local potential $-V$ acts only on the four nearest oxygen sites of a central plaquette (called set $A$), which is the paper's caricature of a Ca vacancy. Determinant quantum Monte Carlo gives approximation-free finite-temperature Green's functions, and maximum-entropy analytic continuation converts imaginary-time Green's functions into local densities of states. The controlling diagnostic is the hole-density difference $\Delta n_h = n_h(V)-n_h(0)$ with respect to the undoped, potential-free system; this difference isolates the spatial envelope of the two doped holes and yields the three $V$ regimes (homogeneous, intermediate, collapsed). Pair-cluster physics is studied by placing two identical potentials at separations $(4,0)$, $(3,0)$, and $(2,0)$ lattice spacings.

What would settle it

Recompute the same two-hole problem with a realistic Ca-vacancy potential, for instance a screened Coulomb potential acting on all Cu and O sites within a few angstroms rather than only on the four nearest oxygens, and test whether any potential strength produces a two-hole density envelope that decays on a $4\times4$ region. If no such potential reproduces the $4\times4$ scale, or if the envelope changes materially with lattice size or boundary conditions, the building-block claim fails. An experimental check: in STS maps at very low Ca-vacancy concentration, the two-hole puddle around an isolated vacancy should have a fixed $4a_0\times4a_0$ size that does not vary with the bias window.

Watch

Extended reading notes

Core claim

The paper reports that two doped holes near a Ca vacancy, modeled by a repulsive local potential on the four nearest oxygen sites of the central plaquette, spontaneously form a $4\times4$ $d$-$p$ unit-cell cluster. The hole density, measured relative to the undoped potential-free system, is homogeneous for $V=0$, collapses onto the innermost oxygen ring for $V\ge1$, and for intermediate $V=0.75$ decays through the O1, O2, and O3 rings to define the $4\times4$ envelope. The same calculation puts a Fermi-level peak in the local density of states on the O1 and Cu1 sites nearest the vacancy, which the authors associate with a singlet between holes on O1 oxygens and neighboring Cu1 spins, while O2 and O3 retain weaker in-gap weight. In two-cluster simulations with separations $(4,0)$, $(3,0)$, and $(2,0)$, the clusters are independent at $(4,0)$, begin to redistribute density at $(3,0)$, and at $(2,0)$ produce a stripe-like enhancement between the clusters while preserving the Fermi-level spectral peaks. The paper concludes that the $4\times4$ two-hole cluster is the building block of hole-doped cuprates, giving a natural geometric explanation of 1/8 charge order and a percolation route to phase-coherent superconductivity.

Load-bearing premise

The load-bearing premise is that a constant local potential $V=0.75$ placed only on the four oxygen sites nearest a Ca vacancy faithfully represents a real Ca vacancy, so that the computed $4\times4$ density envelope is a property of cuprates rather than an artifact of a hand-picked potential.

Editorial extensions

If this is right

The puzzling 1/8 doping anomaly acquires a geometric meaning: one two-hole cluster per $4\times4$ supercell, with a local repulsion between clusters, pins holes into a checkerboard charge order that can act as the parent state of superconductivity.
Underdoped cuprates are inhomogeneous at a fundamental scale, so descriptions of hole doping that start from a clean uniform model omit an intrinsic length scale of the problem.
When two clusters touch at separation $2a_0$, the hole density redistributes into a stripe-like pattern while the Fermi-level spectral weight survives, linking local nematic order to preformed pairing.
A crude effective model of bosons hopping between $4\times4$ supercells (an effective $t$-$V$ model) should reproduce 1/8 charge order and, away from 1/8 doping, produce a phase-coherent superconducting condensate through percolation of overlapping clusters.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Editorial inference: the $4\times4$ scale is not a generic consequence of two holes near a repulsive potential; the paper's own scan shows the cluster exists only for intermediate $V$. A decisive test is whether a realistic extended vacancy potential reproduces the same envelope without tuning the potential strength.
The paper stops at density and single-particle spectra. A direct calculation of two-hole pairing correlations on one cluster and between neighboring clusters would turn the 'preformed local Cooper pair' suggestion into a testable prediction.
The effective $t$-$V$ boson picture predicts a specific doping phase diagram—charge order pinned at 1/8, with superconductivity growing as extra holes create overlapping clusters—so STS maps across that doping range would test the percolation scenario.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick take: the paper reports a real, reproducible numerical result—two holes in a three-band Hubbard model with a local potential on the four O sites do produce a hole-density profile that decays over roughly a 4x4 region for intermediate V, and the two-cluster merging pattern at (2,0) separation is new. The authors are careful that DQMC is exact for the stated model, and they don't hide the severe simplifications. The soft spots are real but mostly about the interpretive leap, not the simulation.

What is actually new: Ref 13 was single-band, one hole, no cluster picture. Here it is two holes in a three-band d-p model, plus two interacting local potentials. The density redistribution when clusters touch at (2,0) is a genuinely new observation. The local DOS showing in-gap weight on O1/Cu1 is consistent with earlier impurity studies and is useful.

Where it gets shaky: the 4x4 building block is not a prediction. The potential is restricted to the four nearest O sites (admitted in the model section), and V=0.75 is picked after the fact because it gives the decay that matches the STS pattern. For V=0 or V>=1 the cluster looks different, so the intermediate regime is doing all the work. No error bars, no finite-size scaling, and the MaxEnt spectra should be treated as indicative. The 1/8 charge order and superconductivity discussion is explicitly a 'most crude' t-V model suggestion, so I don't hold that against them—it's labeled speculation. But the abstract's 'numerically support' and 'building block' language is stronger than the evidence warrants.

Also minor: the claim 'finite-size effects should not be significant' is asserted rather than shown; on an 8x8 lattice with a 4x4 cluster that's plausible, but a comparison of 6x6 versus 8x8 would have helped.

Bottom line: it's a competent DQMC study of a simplified model, worth publishing as a conditional with tightened claims and some robustness checks. It doesn't resolve the cuprate puzzle, but it gives a concrete, citable data point for the local-potential/cluster picture. I'd send it to referees; they'll ask for error bars and at least one alternative potential shape. Reading group maybe, mostly because the whole paper is a good example of how easy it is to make a fitted local potential look like an emergent length scale.

Referee Report

5 major / 6 minor

Summary. The paper studies two doped holes in a three-band Hubbard (d-p) model on an 8×8 lattice with an additional local potential −V placed on the four oxygen sites surrounding a central plaquette, intended to mimic a Ca vacancy in Ca2CuO2Cl2. Using finite-temperature determinant quantum Monte Carlo, the authors compute the hole density distribution and local spectral functions. They find that for an intermediate value V=0.75 the hole density decays from the inner O1 ring through O2 and O3 rings, leading them to identify an approximately 4×4 d-p unit-cell cluster as the natural building block of hole-doped cuprates. They then place two such local potentials at separations (4,0), (3,0), and (2,0) to study cluster interplay, report density redistribution and stripe-like features when clusters touch, and argue that this supports a percolation picture for 1/8 charge order and high-Tc superconductivity.

Significance. If the central claim holds, the paper would provide a concrete microscopic justification for the experimentally proposed 4×4 supercell as a building block in underdoped cuprates, linking local spectroscopic features to a specific model mechanism. The use of numerically exact DQMC on the three-band model is a strength, as the densities for the stated Hamiltonian are computed without approximation. The two-hole cluster-pair calculations go beyond previous one-hole studies and give a first look at how neighboring vacancies interact. However, the significance is conditional on the modeling of the Ca vacancy and on the robustness of the 4×4 identification; as presented, the conclusion rests on a post hoc choice of V and on a visual definition of the cluster extent.

major comments (5)

[Model and methodology, Eq. (1)] The mapping of a Ca vacancy to a local potential restricted to the four nearest oxygen sites is a strong modeling assumption that is not stress-tested. The text acknowledges this with 'For simplicity', but the central building-block conclusion depends on the resulting density decaying through O2 and O3. A realistic vacancy potential would act on Cu and O sites at multiple distances, and it is not shown whether adding a central Cu term or a longer-range tail preserves the 4×4 decay. The authors should report results for at least one alternative potential form (e.g., including a Cu-site term or a 1/r tail) to demonstrate that the 4×4 scale is robust rather than an artifact of the chosen restriction to set A.
[Hole density distribution, Fig. 2] The choice V=0.75 is post hoc: Fig. 2(a,b) show that for small V the density is nearly homogeneous and for V≥1.0 the holes collapse onto O1, so the O2/O3 decay that defines the 4×4 cluster exists only in a narrow intermediate window. To support the claim that the 4×4 building block is a robust physical feature, the authors should provide a quantitative criterion for the intermediate regime, show how the inferred cluster size varies across that window, and report Monte Carlo error bars on the density differences. Without error bars the non-monotonic dependence in Fig. 2(b), which is used to separate the three regimes, cannot be assessed.
[Hole density distribution, Fig. 2(c)] The identification of the 'approximately 4×4' supercell is made visually by drawing a red square. The paper should define a quantitative measure, for example the fraction of added hole density contained within a given distance from the central plaquette, the fitted decay length of the density on O rings, or a comparison of the density at the 4×4 boundary with the background fluctuation level. As written, the central claim that the cluster is 4×4 rather than, say, 3×3 or 5×5 is not quantitatively established.
[Local spectra, Fig. 3] The spectral statements rely on maximum-entropy analytic continuation, which carries uncontrolled systematic uncertainty. The claims about peaks at the Fermi level on O1 and Cu1 and about in-gap spectral weight should be qualified accordingly, and the authors should indicate how stable these features are with respect to the MaxEnt parameters or, at minimum, note the absence of error estimates. This point is secondary to the density-based cluster claim, but it matters because the spectra are used to connect the model to the STS in-gap features.
[Hole cluster pair and Discussion] The step from two-cluster density overlaps to the percolation picture and to 1/8 charge order is speculative. The paper states that two holes per 4×4 supercell 'is exactly 1/8 doping' and then argues for long-range 1/8 charge order, but no calculation at 1/8 doping on a larger lattice is presented, and the effective bosonic model is only sketched. These remarks should be clearly separated from the numerical results, or supported by additional simulations at lower density and larger system size.

minor comments (6)

[Hole cluster pair] There is a typo: 'we have demonstrated the capbility' should read 'capability'.
[Abstract] The phrase 'The model numerically support the role' should be 'supports', and the sentence structure of the abstract could be tightened.
[Fig. 2] The legend labels such as 'nh nh(d = 1, V = 0)' are difficult to parse; please define the subtracted quantity explicitly in the caption or in the text.
[References] Reference [10] is listed as 'to be published (2022)' and reference [12] is an arXiv preprint from 2022; the authors should update these if published versions are now available.
[Model and methodology] The phrase 'principally exact numerical technique' is unusual; 'numerically exact' or 'approximation-free' would be clearer.
[Eq. (3)] The integral kernel e^{-ωτ}/(e^{-βω}+1) should be checked against the standard fermionic Green's function convention, since a sign or factor in the denominator affects the MaxEnt input and hence the reported spectra; a brief statement of convention would help.

Circularity Check

1 steps flagged · score 4.0 of 10

The 4x4 'building block' conclusion is selected by tuning the local potential V to reproduce the STS pattern, rather than derived from an independently fixed vacancy potential.

fitted input called prediction [Model and methodology, Eq. (1); Hole density distribution, Fig. 2]
"For simplicity, we restrict this potential only to the four nearest surrounding oxygen sites, which are labeled as A in Eq. 1 and denoted as cross symbols in Fig. 1. ... Evidently, a moderate local potential can produce the density modulation such that it decays until O3 sites, namely within a 4×4 supercell around the central plaquette applied with local potential, to match with STS experimental findings."

The local potential is the sole source of inhomogeneity and its strength V is a free parameter; the paper scans V and then selects the intermediate value V=0.75 because the resulting density decay 'matches' the experimental 4x4 STS pattern. The later statement that the calculations 'confirm the formation of 4a0 x 4a0 basic plaquette' is therefore not an independent prediction: the target observable is the selection criterion for the fitted parameter. Because no separate determination of the Ca-vacancy potential is given, the building-block claim is an interpolation through the experimental pattern rather than a first-principles derivation. The 4x4 size itself is not hard-coded, and the two-cluster overlap is emergent, so the circularity is partial.

full rationale

The central numerical machinery (DQMC) and the three-band Hubbard parameters are standard and externally grounded, and no load-bearing self-citation or uniqueness theorem is invoked. The principal circularity is parameter fitting: V=0.75 is chosen in the intermediate regime specifically so that the two-hole density decays through O3 and reproduces the observed 4x4 pattern, after which the paper presents the same pattern as theoretical confirmation of the 4x4 building block. The two-cluster interplay and spectral results are genuine outputs of the calculation, and the 1/8 charge-order remark is dimensional counting that inherits the fitted cluster size. The authors' own limitation note (Discussion: cannot reach larger lattice sizes, higher doping, or lower temperature to study percolation and pairing) is a caveat on the superconducting claims rather than an additional circularity. This warrants a moderate partial-circularity score rather than a higher one, because the local potential does not encode the 4x4 length scale explicitly and the density profile at fixed V is a nontrivial correlated-electron result.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central results rest on a standard Hubbard model, a specific cuprate parameter set, a simplified local potential on four oxygen sites, and numerical assumptions (beta=10, 8x8, MaxEnt continuation). The only hand-tuned parameter that directly shapes the central claim is V, whose characteristic value 0.75 is selected to match the experimental cluster size.

free parameters (3)

Local potential V = V=0.75 (characteristic; scanned 0, 0.25, 0.5, 0.75, 1.0)
Chosen by hand to mimic the Ca vacancy; the 4x4 cluster appears only for intermediate V, so this choice is load-bearing.
Chemical potential mu = Tuned to yield two doped holes
mu is adjusted to fix the average hole count at two; its actual value is not reported, and the doping level is an input rather than an output.
Oxygen Hubbard repulsion U_pp = 0
Set to zero 'for simplicity'; the paper claims realistic nonzero U_pp does not change the conclusions, but no U_pp > 0 data is shown.

assumptions (5)

domain assumption The three-band d-p Hubbard model with U_pp=0 and the stated parameter set (U_dd=8.0, t_pd=1.3, t_pp=0.49, eps_p=3.24) is an adequate description of local cuprate physics.
Invoked in Model and methodology; the validity of this model for the STS-observed local states is assumed.
ad hoc to paper The Ca vacancy is represented by a local potential acting only on the four nearest oxygen sites, with all other sites unaffected.
Stated: 'we restrict this potential only to the four nearest surrounding oxygen sites' (Eq. 1, set A). This simplification is not tested against more realistic potentials.
domain assumption DQMC on an 8x8 lattice at beta=10 is sufficient to capture the local charge distribution, because finite-size effects are limited to a 4x4 region.
The paper asserts 'the finite-size effects should not be significant' without systematic scaling.
domain assumption Maximum entropy analytic continuation yields reliable local densities of states from imaginary-time Green's functions.
The local spectra in Fig. 3 are obtained via MaxEnt; the ill-posedness of this continuation is not discussed.
domain assumption The hole language and subtraction of the undoped V=0 system cleanly isolate the doped-hole response.
The density difference used throughout assumes that the undoped configuration is the correct reference and that the total added hole number is exactly two.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Hole clustering and mutual interplay in three-band Hubbard model." pith.science (2026). https://pith.science/paper/RDJHHDKX

@misc{pith2026250600787,
  author       = {Pith},
  title        = {Pith review of: Hole clustering and mutual interplay in three-band Hubbard model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RDJHHDKX}},
  note         = {Machine review of arXiv:2506.00787}
}

abstract

Recent scanning tunnelling spectroscopy (STS) experiments revealed remarkable role of a supercell consisting $4\times4$ CuO$_2$ unit cells in the emergence of local nematic state and preformed local Cooper pairs and phase coherent cuprate superconductivity. By employing the numerically exact determinant Quantum Monte Carlo simulations, we mimic the effects of experimental Ca vacancy by an external local potential to investigate the charge and spectral properties of the system hosting two doped holes. The model numerically support the role of the $4\times4$ supercell as the building block of hole doped cuprates via the hole density distribution and local spectra around the local potential. Our results might provide a theoretical support on the experimental observations and a platform for investigating local charge order and local Cooper pairs on the $4\times4$ supercell as the plausible route to understanding unconventional cuprate superconductivity.

Figures

Figures reproduced from arXiv: 2506.00787 by the authors.

**Figure 2.** FIG. 2: (a) Evolution of the charge density of two doped holes [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** FIG. 3: Local density of states (DOS) of inequivalent Cu and O sites [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 5.** Figure 5: FIG. 5: Local spectra for two clusters of [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

18 extracted references · 17 canonical work pages

[1]

F. C. Zhang and T. M. Rice, Phys. Rev. B37, 3759 (1988)

1988
[2]

Santoso, W

I. Santoso, W. Ku, T. Shirakawa, G. Neuber, X. Yin, M. Enoki, M. Fujita, R. Liang, T. Venkatesan, G.A. Sawatzky, A. Kotlov, S. Yunoki, M. Rübhausen, and A. Rusydi, Phys. Rev. B95, 165108 (2017)

work page 2017
[3]

Adolphs, S

C.P.J. Adolphs, S. Moser, G.A. Sawatzky, and M. Berciu, Phys. Rev. Lett.116, 087002 (2016)

work page 2016
[4]

B. Lau, M. Berciu, and G.A. Sawatzky, Phys. Rev. Lett.106, 036401 (2011)

work page 2011
[5]

Ebrahimnejad, G.A

H. Ebrahimnejad, G.A. Sawatzky, and M. Berciu, Nat. Phys. 10, 951 (2014)

work page 2014
[6]

Ebrahimnejad, G

H. Ebrahimnejad, G. A. Sawatzky and M. Berciu, J. Phys.: Cond. Mat.28, 105603 (2016)

work page 2016
[7]

Jiang, M

M. Jiang, M. Moeller, M. Berciu and G. A. Sawatzky, Phys. Rev. B 101, 035151 (2020)

work page 2020
[8]

Qin, C.-M

M. Qin, C.-M. Chung, H. Shi, E. Vitali, C. Hubig, U. Schollwöck, S. R. White, and S. Zhang, Phys. Rev. X 10, 031016 (2020)

work page 2020

Show all 18 references

[9]

H. Li, S. Ye, J. Zhao, C. Jin, and Y . Wang, Science Bulletin, 66, 1395 (2021)

2021
[10]

Ye, C.Zou, H

S. Ye, C.Zou, H. Yan, Y . Ji, M. Xu, Z. Dong, Y . Chen, X.J. Zhou, and Y . Wang, to be published (2022)

2022
[11]

Y . Li, A. Sapkota, P. M. Lozano, Z. Du, H. Li et al, arXiv: 2205.01702 (2022)

2022 arXiv
[12]

H. Li, H. Li, Z. Wang, S. Wan, H. Yang, H.-H. Wen, arXiv: 2207.00783 (2022)

2022 arXiv
[13]

He, S.-L

C.-P. He, S.-L. Yu, T. Xiang, and J.-X. Li, Chin. Phys. Lett.39, 057401 (2022)

2022
[14]

Emery, Phys

V .J. Emery, Phys. Rev. Lett.58, 2794 (1987)

1987
[15]

Kung, C.-C

Y .F. Kung, C.-C. Chen, Y . Wang, E. W. Huang, E. A. Nowadnick et al, Phys. Rev. B 93, 155166 (2016)

2016
[16]

Blankenbecler, D.J

R. Blankenbecler, D.J. Scalapino, and R.L. Sugar, Phys. Rev. D24, 2278 (1981)

1981
[17]

W. Hu, R. T. Scalettar, E. W. Huang, and B. Moritz, Phys. Rev. B 95, 235122 (2017)

2017
[18]

Gubernatis, M

J.E. Gubernatis, M. Jarrell, R.N. Silver, and D.S. Sivia, Phys. Rev. B 44, 6011 (1991)

1991

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

F. C. Zhang and T. M. Rice, Phys. Rev. B37, 3759 (1988)

1988

[2] [2]

Santoso, W

I. Santoso, W. Ku, T. Shirakawa, G. Neuber, X. Yin, M. Enoki, M. Fujita, R. Liang, T. Venkatesan, G.A. Sawatzky, A. Kotlov, S. Yunoki, M. Rübhausen, and A. Rusydi, Phys. Rev. B95, 165108 (2017)

work page 2017

[3] [3]

Adolphs, S

C.P.J. Adolphs, S. Moser, G.A. Sawatzky, and M. Berciu, Phys. Rev. Lett.116, 087002 (2016)

work page 2016

[4] [4]

B. Lau, M. Berciu, and G.A. Sawatzky, Phys. Rev. Lett.106, 036401 (2011)

work page 2011

[5] [5]

Ebrahimnejad, G.A

H. Ebrahimnejad, G.A. Sawatzky, and M. Berciu, Nat. Phys. 10, 951 (2014)

work page 2014

[6] [6]

Ebrahimnejad, G

H. Ebrahimnejad, G. A. Sawatzky and M. Berciu, J. Phys.: Cond. Mat.28, 105603 (2016)

work page 2016

[7] [7]

Jiang, M

M. Jiang, M. Moeller, M. Berciu and G. A. Sawatzky, Phys. Rev. B 101, 035151 (2020)

work page 2020

[8] [8]

Qin, C.-M

M. Qin, C.-M. Chung, H. Shi, E. Vitali, C. Hubig, U. Schollwöck, S. R. White, and S. Zhang, Phys. Rev. X 10, 031016 (2020)

work page 2020