Crystal-field effects in the formation of Wigner-molecule supercrystals in moiré TMD superlattices

Constantine Yannouleas [email protected]    Uzi Landman [email protected] School of Physics, Georgia Institute of Technology, Atlanta, Georgia 30332-0430
(27 May 2024)
Abstract

For moiré bilayer TMD superlattices, full-configuration-interaction (FCI) calculations are presented that take into account both the intra-moiré-quantum-dot (MQD) charge-carrier Coulombic interactions, as well as the crystal-field effect from the surrounding moiré pockets (inter-moiré-QD interactions). Such FCI calculations enable an effective computational embedding strategy and allow for a complete interpretation of the counterintuitive experimental observations reported recently in the context of moiré TMD superlattices at integer fillings ν=2𝜈2\nu=2italic_ν = 2 and 4. Two novel states of matter are reported: (i) a genuinely quantum-mechanical supercrystal of sliding Wigner molecules (WMs) for unstrained moiré TMD materials (when the crystal field is commensurate with the trilobal symmetry of the confining potential in each embedded MQD) and (ii) a supercrystal of pinned Wigner molecules when strain is involved and the crystal field is incommensurate with the trilobal symmetry of the confining potential in each embedded MQD. The case of ν=3𝜈3\nu=3italic_ν = 3 is an exception, in that both unstrained and strained cases produce a supercrystal of pinned WMs, which is due to the congruence of intrinsic (that of the WM) and external (that of the confining potential of the MQD) C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT point-group symmetries. Furthermore, it is shown that the unrestricted Hartree-Fock approach fails to describe the supercrystal of sliding WMs in the unstrained case, providing a qualitative agreement only in the case of a supercrystal of pinned WMs.

I Introduction

Earlier studies have revealed a novel fundamental-physics aspect in artificial twodimensional (2D) nanosystems in the regime of strong interelectronic correlations, namely formation of quantum Wigner molecules (WMs), originally described theoretically [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26] in 2D semiconductor (parabolic, elliptic, and double-well) quantum dors (QDs), as well as in trapped ultracold atoms, and subsequently observed experimentally in GaAs QDs [27, 28, 29, 30], Si/SiGe QDs [31], planar germanium QDs [32] and 1D carbon-nanotubes [33].

Remarkably, most recent work [34, 35, 36] has extended the WM portfolio to the newly emerging and highly regarded (due to the potential for fundamental-physics discoveries and for advancing quantum-device applications) field of transition-metal dichalcogenide (TMD) moiré materials and superlattices. Specifically, TMD moiré superlattices are highly pursued for solving the question of scalability in quantum computer architectures.

Refer to caption
Figure 1: FCI ground-state (with total spin S=0𝑆0S=0italic_S = 0) charge densities for N=2𝑁2N=2italic_N = 2 holes in the embedded central MQD (red triangle in the schematics) taking into consideration the crystal-field effect generated by the charge carriers in the surrounding six moiré pockets (see blue hexagon, with radius aMsubscript𝑎𝑀a_{M}italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT, in the schematics). (a) and (c) κ=5𝜅5\kappa=5italic_κ = 5 in agreement with the case of a TMD moiré superlattice in hBN environment. (b) and (d) were calculated, for purpose of comparison, with κ=2𝜅2\kappa=2italic_κ = 2, corresponding to a stronger Coulomb repulsion. In (a) and (b), all six surrounding MQDs (see schematics) were populated with Q=Ne𝑄𝑁𝑒Q=Neitalic_Q = italic_N italic_e charge carriers. In (c) and (d), only one of the surrounding MQDs (the one to the right, see schematics) has been populated with Q=Ne𝑄𝑁𝑒Q=Neitalic_Q = italic_N italic_e charge carriers. The charge-carrier distributions in each surrounding MQD have been represented by three point charges Q/3𝑄3Q/3italic_Q / 3 placed at the apices of an equilateral triangle (see schematics), with radius aM/6subscript𝑎𝑀6a_{M}/6italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT / 6. Remaining parameters: effective mass m=0.90mesuperscript𝑚0.90subscript𝑚𝑒m^{*}=0.90m_{e}italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = 0.90 italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT, moiré lattice constant aM=9.8subscript𝑎𝑀9.8a_{M}=9.8italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT = 9.8 nm, depth of a moiré pocket v0=10.3subscript𝑣010.3v_{0}=10.3italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 10.3 eV, and ϕ=20italic-ϕsuperscript20\phi=20^{\circ}italic_ϕ = 20 start_POSTSUPERSCRIPT ∘ end_POSTSUPERSCRIPT. Charge densities are in units of 1/nm2.

Building on the demonstrated formation [34] of WMs in the isolated moiré pockets [most often referred to as moiré quantum dots (MQDs)] at integer fillings, we investigate here the effects arising from embedding such single MQDs in the moiré superlattice structure, which is the actual system that was addressed experimentally [36]. Specifically, this paper adresses the key question whether the experimental observations in the unstrained-lattice case do reflect formation of a collective WM supercrystal, i.e., a novel, more complex insulating phase of matter beyond the well-known generalized Wigner crystal, or the trivial case of a collection of independent MQDs. The present investigations (see below) support formation of a WM supercrystal 111We note that the experimental results for the unstrained case in Ref. [36] were necessarily interpreted with formation of sliding WMs in isolated MQDs (for N=2𝑁2N=2italic_N = 2 confined charge carriers, see Refs. [34, 36]; for N=4𝑁4N=4italic_N = 4 confined charge carriers, see Ref. [34]), given the fact that the mean-field Hartree-Fock approach (that allows calculations for a large number of charge carriers distributed over several moiré pockets) is unreliable [35, 36] for describing sliding WMs. having sliding WMs as building blocks for ν=2𝜈2\nu=2italic_ν = 2 or ν=4𝜈4\nu=4italic_ν = 4, with the filling ν=3𝜈3\nu=3italic_ν = 3 being an exception having a pinned WM as a building block.

Specifically, this paper focuses on the effects on WM formation in a given MQD resulting from the crystal-field generated by the charge carriers (of same nature) in the surrounding MQDs (moiré pockets) located at the apices of a regular hexagon in the TMD moiré superlattice. This approach is inspired by the well-known crystal-field theory [38, 39] in molecular physics, and is applicable here because the moiré lattice constant aMsubscript𝑎𝑀a_{M}italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT is much larger than the extent of the Wigner molecule within a single MQD. Two different computational methodologies will be used in this endeavor, namely: (i) the full configuration interaction (FCI) [40, 8, 9, 41, 42, 14, 17, 43, 21, 22, 26, 26, 44, 45] and (ii) the spin-and-space unrestricted Hartree-Fock (sS-UHF) [1, 46, 5, 14, 47].

II Many-body Hamiltonian (including crystal field from surrounding moiré pockets)

Following earlier established literature [48, 49, 50, 34, 36], we approximate the potential of the 2D TMD moiré superlattice that confines the extra charge carriers as

V(𝐫)=2v0i=13cos(𝐆i𝐫+ϕ),𝑉𝐫2subscript𝑣0superscriptsubscript𝑖13subscript𝐆𝑖𝐫italic-ϕ\displaystyle V({\bf r})=-2v_{0}\sum_{i=1}^{3}\cos({\bf G}_{i}\cdot{\bf r}+% \phi),italic_V ( bold_r ) = - 2 italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT ∑ start_POSTSUBSCRIPT italic_i = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT roman_cos ( bold_G start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ⋅ bold_r + italic_ϕ ) , (1)

where 𝐆i=[(4π/3aM)(sin(2πi/3),cos(2πi/3))]subscript𝐆𝑖delimited-[]4𝜋3subscript𝑎𝑀2𝜋𝑖32𝜋𝑖3{\bf G}_{i}=[(4\pi/\sqrt{3}a_{M})(\sin(2\pi i/3),\cos(2\pi i/3))]bold_G start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT = [ ( 4 italic_π / square-root start_ARG 3 end_ARG italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT ) ( roman_sin ( 2 italic_π italic_i / 3 ) , roman_cos ( 2 italic_π italic_i / 3 ) ) ] are the moiré reciprocal lattice vectors. The materials specific parameters of V(𝐫)𝑉𝐫V({\bf r})italic_V ( bold_r ) are v0subscript𝑣0v_{0}italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT (which can also be experimentally controlled through voltage biasing), the moiré lattice constant aMsubscript𝑎𝑀a_{M}italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT, and the angle ϕitalic-ϕ\phiitalic_ϕ. aMsubscript𝑎𝑀a_{M}italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT is typically of the order of 10 nm, which is much larger than the lattice constant of the monolayer TMD material (typically a few Å).

The parameter ϕitalic-ϕ\phiitalic_ϕ controls the strength of the trilobal C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT anisotropy in each MQD potential pocket. This trilobal anisotropy can be seen by expanding V(𝐫)𝑉𝐫V({\bf r})italic_V ( bold_r ) in Eq. (1) in powers of r𝑟ritalic_r, and defining an approximate confining potential, VMQD(𝐫)subscript𝑉MQD𝐫V_{\rm MQD}({\bf r})italic_V start_POSTSUBSCRIPT roman_MQD end_POSTSUBSCRIPT ( bold_r ), for a single MQD as follows:

VMQD(𝐫)V(𝐫)+6v0cos(ϕ)mω02r2/2+𝒞sin(3θ)r3.subscript𝑉MQD𝐫𝑉𝐫6subscript𝑣0italic-ϕsuperscript𝑚superscriptsubscript𝜔02superscript𝑟22𝒞3𝜃superscript𝑟3\displaystyle\begin{split}V_{\rm MQD}({\bf r})\equiv V({\bf r})+6v_{0}\cos(% \phi)\approx m^{*}\omega_{0}^{2}r^{2}/2+{\cal C}\sin(3\theta)r^{3}.\end{split}start_ROW start_CELL italic_V start_POSTSUBSCRIPT roman_MQD end_POSTSUBSCRIPT ( bold_r ) ≡ italic_V ( bold_r ) + 6 italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT roman_cos ( italic_ϕ ) ≈ italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT italic_ω start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_r start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT / 2 + caligraphic_C roman_sin ( 3 italic_θ ) italic_r start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT . end_CELL end_ROW (2)

with mω02=16π2v0cos(ϕ)/aM2superscript𝑚superscriptsubscript𝜔0216superscript𝜋2subscript𝑣0italic-ϕsuperscriptsubscript𝑎𝑀2m^{*}\omega_{0}^{2}=16\pi^{2}v_{0}\cos(\phi)/a_{M}^{2}italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT italic_ω start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT = 16 italic_π start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT roman_cos ( italic_ϕ ) / italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT, and 𝒞=16π3v0sin(ϕ)/(33aM3)𝒞16superscript𝜋3subscript𝑣0italic-ϕ33superscriptsubscript𝑎𝑀3{\cal C}=16\pi^{3}v_{0}\sin(\phi)/(3\sqrt{3}a_{M}^{3})caligraphic_C = 16 italic_π start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT roman_sin ( italic_ϕ ) / ( 3 square-root start_ARG 3 end_ARG italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT ); msuperscript𝑚m^{*}italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT is the effective mass and the expansion of V(𝐫)𝑉𝐫V({\bf r})italic_V ( bold_r ) can be restricted to the terms up to r3superscript𝑟3r^{3}italic_r start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT. (r,θ)𝑟𝜃(r,\theta)( italic_r , italic_θ ) are the polar coordinates of the position vector 𝐫𝐫{\bf r}bold_r.

The effective many-body Hamiltonian associated with a given MQD embedded in the moiré superlattice is given by

HMB=HMQD+HCF,subscript𝐻MBsubscript𝐻MQDsubscript𝐻CF\displaystyle H_{\rm MB}=H_{\rm MQD}+H_{\rm CF},italic_H start_POSTSUBSCRIPT roman_MB end_POSTSUBSCRIPT = italic_H start_POSTSUBSCRIPT roman_MQD end_POSTSUBSCRIPT + italic_H start_POSTSUBSCRIPT roman_CF end_POSTSUBSCRIPT , (3)

where

HMQD=i=1N{𝐩i22m+VMQD(𝐫i)}+i<jNe2κ|𝐫i𝐫j|,subscript𝐻MQDsuperscriptsubscript𝑖1𝑁superscriptsubscript𝐩𝑖22superscript𝑚subscript𝑉MQDsubscript𝐫𝑖superscriptsubscript𝑖𝑗𝑁superscript𝑒2𝜅subscript𝐫𝑖subscript𝐫𝑗\displaystyle H_{\rm MQD}=\sum_{i=1}^{N}\left\{\frac{{\bf p}_{i}^{2}}{2m^{*}}+% V_{\rm MQD}({\bf r}_{i})\right\}+\sum_{i<j}^{N}\frac{e^{2}}{\kappa|{\bf r}_{i}% -{\bf r}_{j}|},italic_H start_POSTSUBSCRIPT roman_MQD end_POSTSUBSCRIPT = ∑ start_POSTSUBSCRIPT italic_i = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_N end_POSTSUPERSCRIPT { divide start_ARG bold_p start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_ARG start_ARG 2 italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT end_ARG + italic_V start_POSTSUBSCRIPT roman_MQD end_POSTSUBSCRIPT ( bold_r start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) } + ∑ start_POSTSUBSCRIPT italic_i < italic_j end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_N end_POSTSUPERSCRIPT divide start_ARG italic_e start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_ARG start_ARG italic_κ | bold_r start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT - bold_r start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT | end_ARG , (4)

with κ𝜅\kappaitalic_κ being the dielectric constant, and

HCF=i=1Nj=13eQ3κ|𝐫i𝐑j|subscript𝐻CFsuperscriptsubscript𝑖1𝑁superscriptsubscript𝑗13𝑒𝑄3𝜅subscript𝐫𝑖subscript𝐑𝑗\displaystyle H_{\rm CF}=\sum_{i=1}^{N}\sum_{j=1}^{3{\cal M}}\frac{eQ}{3\kappa% |{\bf r}_{i}-{\bf R}_{j}|}italic_H start_POSTSUBSCRIPT roman_CF end_POSTSUBSCRIPT = ∑ start_POSTSUBSCRIPT italic_i = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_N end_POSTSUPERSCRIPT ∑ start_POSTSUBSCRIPT italic_j = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 3 caligraphic_M end_POSTSUPERSCRIPT divide start_ARG italic_e italic_Q end_ARG start_ARG 3 italic_κ | bold_r start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT - bold_R start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT | end_ARG (5)

is the crystal-field potential from the surrounding moiré pockets, {\cal M}caligraphic_M denoting the number of surrounding moiré pockets populated with charge carriers, and the 𝐑jsubscript𝐑𝑗{\bf R}_{j}bold_R start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT’s being the positions of the point charges that mimick the charge carrier distributions. In each one of the surrounding moiré pockets, we use three point charges at the apices of an equilateral triangle with a total charge Q=Ne𝑄𝑁𝑒Q=Neitalic_Q = italic_N italic_e, thus respecting the C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT trilobal symmetry; see solid dots in the schematics in the figures.

A brief outline of the FCI and sS-UHF methodologies, used to solve the corresponding many-body Schrödinger equation, is presented in the Appendices. Before proceeding with the computational results, we point out that the arrangement having the equilateral three-point-charge distributions in all the six surrounding moiré pockets [see schematic in Fig. 1(a)] displays an overall C6subscript𝐶6C_{6}italic_C start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT point-group symmetry, which is commensurate with the trilobal C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry associated with the potential confinement [Eq. (2)] of a single MQD. Below, we will investigate the effect of both such a commensurate crystal field, as well as the effect of an incommensurate crystal field (IcCF) generated by placing the three-point-charge distribution in only one of the six surrounding moiré pockets [see schematic in Fig. 1(c)].

Refer to caption
Figure 2: sS-UHF ground-state (with total-spin projection Sz=0subscript𝑆𝑧0S_{z}=0italic_S start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0) charge densities for N=2𝑁2N=2italic_N = 2 holes in the central MQD (red triangle in the schematics) taking into consideration the crystal-field effect generated by the charge carriers in the surrounding six moiré pockets (see blue hexagon, with radius aMsubscript𝑎𝑀a_{M}italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT, in the schematics). (a) and (c) κ=5𝜅5\kappa=5italic_κ = 5 in agreement with the case of a TMD moiré superlattice in hBN environment. (b) and (d) were calculated, for purpose of comparison, with κ=2𝜅2\kappa=2italic_κ = 2, corresponding to a stronger Coulomb repulsion. In (a) and (b), all six surrounding MQDs (see schematics) were populated with Q=Ne𝑄𝑁𝑒Q=Neitalic_Q = italic_N italic_e charge carriers. In (c) and (d), only one of the surrounding MQDs (the one to the right, see schematics) has been populated with Q=Ne𝑄𝑁𝑒Q=Neitalic_Q = italic_N italic_e charge carriers. The charge-carrier distributions in each surrounding MQD have been represented by three point charges Q/3𝑄3Q/3italic_Q / 3 placed at the apices of an equilateral triangle (see schematics), with radius aM/6subscript𝑎𝑀6a_{M}/6italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT / 6. Remaining parameters: effective mass m=0.90mesuperscript𝑚0.90subscript𝑚𝑒m^{*}=0.90m_{e}italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = 0.90 italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT, moiré lattice constant aM=9.8subscript𝑎𝑀9.8a_{M}=9.8italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT = 9.8 nm, depth of a moiré pocket v0=10.3subscript𝑣010.3v_{0}=10.3italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 10.3 eV, and ϕ=20italic-ϕsuperscript20\phi=20^{\circ}italic_ϕ = 20 start_POSTSUPERSCRIPT ∘ end_POSTSUPERSCRIPT. Charge densities are in units of 1/nm2.

III Results for filling factor ν=2𝜈2\nu=2italic_ν = 2

The FCI and sS-UHF charge densities (CDs) for a moiré TMD (e.g., WS2 [36]) superlattice at integer filling factor ν=2𝜈2\nu=2italic_ν = 2 calculated using the many-body Hamiltonian (…) are displayed in Fig. 1 and Fig. 2, respectively. The configuration of the equilateral three-point-charge distributions that mimick the crystal-field effect from the surrounding six moiré pockets is illustrated in the schematics drawn in the upper right corner of each frame; see figure captions for details and parameters. The total charge associated with each equilateral three-point-charge distribution is Q=2e𝑄2𝑒Q=2eitalic_Q = 2 italic_e. (Note that Q/e𝑄𝑒Q/eitalic_Q / italic_e can be taken different than ν𝜈\nuitalic_ν in the case of a noninteger filling.)

The FCI calculations [see Figs. 1(a) and 1(b)] demonstrate that the full crystal-field effect (all six surrounding moiré pockets are contributing) maintains the charge densities that are associated with sliding WMs. Namely, the CDs preserve the C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry of the confining single-moiré-pocket potential, as was the case with the isolated MQDs [34]. Note that even a stronger Coulomb repulsion (κ=2𝜅2\kappa=2italic_κ = 2 instead of κ=5𝜅5\kappa=5italic_κ = 5, which corresponds to the experimental setup) does not destroy the the C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry; see Fig.  1(b). This symmetry-preserving behavior is counterintuitive for a crystalline material in that the CDs necessarily exhibit three modest humps instead of the naive expectation of a symmetry-breaking double hump associated with two Wigner-crystal-type [51, 52, 53, 54] localized charge carriers (one hump per charge carrier). Note further that this counterintuitive behavior is in remarkable agreement with the recent experimental findings [36] for the unstrained case at filling ν=2𝜈2\nu=2italic_ν = 2.

Refer to caption
Figure 3: Charge densities for N=4𝑁4N=4italic_N = 4 holes in the central MQD (red triangle in the schematics) taking into consideration the crystal-field effect generated by the charge carriers in the surrounding six moiré pockets (see blue hexagon, with radius aMsubscript𝑎𝑀a_{M}italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT, in the schematics). (a) and (b) FCI ground-states (with total spin S=1𝑆1S=1italic_S = 1 and total-spin projection Sz=0subscript𝑆𝑧0S_{z}=0italic_S start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0). (c) and (d) sS-UHF ground-states (with total spin projection Sz=0subscript𝑆𝑧0S_{z}=0italic_S start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 0). Dielectric constant κ=5𝜅5\kappa=5italic_κ = 5 in agreement with the case of a TMD moiré superlattice in an hBN environment. In (a) and (c), all six surrounding MQDs (see schematics) were populated with Q=Ne𝑄𝑁𝑒Q=Neitalic_Q = italic_N italic_e charge carriers. In (b) and (d), only one of the surrounding MQDs (the one to the right, see schematics) has been populated with Q=Ne𝑄𝑁𝑒Q=Neitalic_Q = italic_N italic_e charge carriers. The charge-carrier distributions in each surrounding MQD have been represented by three point charges Q/3𝑄3Q/3italic_Q / 3 placed at the apices of an equilateral triangle (see schematics), with radius aM/6subscript𝑎𝑀6a_{M}/6italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT / 6. Remaining parameters: effective mass m=0.90mesuperscript𝑚0.90subscript𝑚𝑒m^{*}=0.90m_{e}italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = 0.90 italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT, moiré lattice constant aM=9.8subscript𝑎𝑀9.8a_{M}=9.8italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT = 9.8 nm, depth of a moiré pocket v0=10.3subscript𝑣010.3v_{0}=10.3italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 10.3 eV, and ϕ=20italic-ϕsuperscript20\phi=20^{\circ}italic_ϕ = 20 start_POSTSUPERSCRIPT ∘ end_POSTSUPERSCRIPT. Charge densities are in units of 1/nm2.
Refer to caption
Figure 4: Charge densities for N=3𝑁3N=3italic_N = 3 holes in the central MQD (red triangle in the schematics) taking into consideration the crystal-field effect generated by the charge carriers in the surrounding six moiré pockets (see blue hexagon, with radius aMsubscript𝑎𝑀a_{M}italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT, in the schematics). (a) FCI ground-states (with total spin S=1/2𝑆12S=1/2italic_S = 1 / 2 and total-spin projection Sz=1/2subscript𝑆𝑧12S_{z}=1/2italic_S start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 1 / 2). (b) sS-UHF ground-states (with total spin projection Sz=1/2subscript𝑆𝑧12S_{z}=1/2italic_S start_POSTSUBSCRIPT italic_z end_POSTSUBSCRIPT = 1 / 2). Dielectric constant κ=5𝜅5\kappa=5italic_κ = 5 in agreement with the case of a TMD moiré superlattice in an hBN environment. In (a) and (b), all six surrounding MQDs (see schematics) were populated with Q=Ne𝑄𝑁𝑒Q=Neitalic_Q = italic_N italic_e charge carriers. The charge-carrier distributions in each surrounding MQD have been represented by three point charges Q/3𝑄3Q/3italic_Q / 3 placed at the apices of an equilateral triangle (see schematics), with radius aM/6subscript𝑎𝑀6a_{M}/6italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT / 6. Remaining parameters: effective mass m=0.90mesuperscript𝑚0.90subscript𝑚𝑒m^{*}=0.90m_{e}italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT = 0.90 italic_m start_POSTSUBSCRIPT italic_e end_POSTSUBSCRIPT, moiré lattice constant aM=9.8subscript𝑎𝑀9.8a_{M}=9.8italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT = 9.8 nm, depth of a moiré pocket v0=10.3subscript𝑣010.3v_{0}=10.3italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT = 10.3 eV, and ϕ=20italic-ϕsuperscript20\phi=20^{\circ}italic_ϕ = 20 start_POSTSUPERSCRIPT ∘ end_POSTSUPERSCRIPT. Charge densities are in units of 1/nm2.

The CDs in Figs. 2(a) and 2(b) demonstrate that the sS-UHF is unable to properly describe the unstrained TMD moiré supercrystal at filling ν=2𝜈2\nu=2italic_ν = 2. Indeed, these sS-UHF CDs exhibit a pinned WM with two well-localized humps in accordance with the traditional Wigner-crystal-type [51, 52, 53, 54] expectation (one hump per charge carrier). We stress again that these sS-UHF CDs do break the C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry of the single-moiré-pocket confinement and disagree with the experimental observations [36].

Heretofore, we used a crystal field with C6subscript𝐶6C_{6}italic_C start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT symmetry (associated with that of the six surrounding moiré pockets), which is commensurate with the C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry of the confinement of the embedded MQD. A natural question arising at this point concerns the effect of a crystal field that has a symmetry non-commensurate with the C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry of the single MQD. Experimentally, such a non-commensurability was generated [36] by deforming the moiré lattice through straining. Here, we generate a non-commensurability situation by maintaining the equilateral three-point-charge configuration only in a single moiré pocket [the one on the right; see schematics in Figs. 1(c,d) and 2(c,d)].

From the FCI CDs in Figs. 1(c,d), it is apparent that the IcCF produces an azimuthally pinned WM exhibiting two antipodal humps. These pinned-WM CDs do break the C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry of the single MQD. In addition, the extent of localization for each charge carrier becomes more pronounced for stronger Coulomb repulsion; compare the case of κ=5𝜅5\kappa=5italic_κ = 5 [Fig. 1(c)] to κ=2𝜅2\kappa=2italic_κ = 2 [Fig. 1(d)]. Concerning the corresponding sS-UHF CDs, it is obvious that, in qualitative agreement with the FCI result, the IcCF CDs result also in pinned two-humped WMs, exhibiting, nevertheless, a stronger charge-carrier localization; contrast Fig. 1(c) to Fig. 2(c) and Fig. 1(d) to Fig. 2(d). We mention here that our IcCF FCI CDs are in agreement with the recent experimental observations [36] of two-humped pinned WMs per moiré pocket in the case of strained moiré superlattices at a filling ν=2𝜈2\nu=2italic_ν = 2.

IV Results for filling factor ν=4𝜈4\nu=4italic_ν = 4

The ability of the crystal-field with the commensurate C6subscript𝐶6C_{6}italic_C start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT symmetry to preserve, and even enhance, the formation of a quantum sliding WM in the central MQD is revealed in an even more spectacular way in the case of the filling factor ν=4𝜈4\nu=4italic_ν = 4, where Ref. [36] has reported within each MQD the observation of CDs with 3 humps (C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry) for the unstrained lattice, but with 4 humps (broken symmetry) for the strained lattice. Our FCI CDs that take into account the crystal-field effect are in agreement with the experiment; see the 3-hump, sliding-WM CD in Fig. 3(a) (commensurate crystal field) and the 4-hump, distorted-pinne-WM CD in Fig. 3(b) (IcCF). We further note that the sS-UHF fails to describe the sliding WM, yielding instead a broken-C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT-symmetry CD (pinned WM) with 4 well defined humps even for the case of the commensurate crystal field; see Fig. 3(c). In the case of an IcCF, the sS-UHF agrees qualitatively with the FCI CD as it describes a deformed, 4-hump, pinned WM; compare Fig. 3(b) and Fig. 3(d), the charge-carrier localization being more pronounced in the sS-UHF CD.

V Results for filling factor ν=3𝜈3\nu=3italic_ν = 3

Compared to the cases of ν=2𝜈2\nu=2italic_ν = 2 and ν=4𝜈4\nu=4italic_ν = 4, the case of ν=3𝜈3\nu=3italic_ν = 3 is exceptional, a fact that was reported already in Ref. [34] where formation of WMs was investigated for the case of an isolated MQD in the absence of a crystal field from the surrounding moiré pockets. Specifically, the 3-hump pinned WM in the central MQD has a C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT symmetry which coincides with that of the confining potential. This congruence of intrinsic and external point-group symmetries prohibits the formation of a sliding WM and yields a strongly pinned WM. The commensurate crystal field does not alter this CD distribution, a fact that is reflected in the FCI calculation [see Fig. 4(a)]. In the ν=3𝜈3\nu=3italic_ν = 3 case, the sS-UHF result is in qualitatively aggreement with the FCI one [see Fig. 4(b)], with the charge-carrier localization, however, being more pronounced in the sS-UHF case. Again, it is remarkable that our FCI results (including the crystal-field effect) for ν=3𝜈3\nu=3italic_ν = 3 are in excellent agreement with the experiment [36].

VI Summary

Taking into consideration (via a computational embedding scheme) both the intra-moiré-QD charge-carrier interactions in the embedded MQD, as well as the crystal-field effect from the surrounding moiré pockets (inter-moiré-QD interactions), and using the full-configuration-interaction methodology, we offered a complete interpretation of the counterintuitive experimental observations reported in Ref. [36] in the context of moiré TMD superlattices at integer fillings ν=2𝜈2\nu=2italic_ν = 2 and 4. In particular, our results demonstrate that these experimental observations reflect the interplay between two novel states of matter; namely, (i) a supercrystal of sliding Wigner molecules 222When the QD confinement possesses circular symmetry, the sliding WMs were earlier referred to as rotating WMs or rotating electron molecules [14, 8]. for unstrained moiré TMD materials (when the crystal field is commensurate with the trilobal symmetry of the confining potential in each MQD) and (ii) a supercrystal of pinned Wigner molecules when strain is involved and the crystal field is non-commensurate with the trilobal symmetry of the confining potential in each MQD. The case of ν=3𝜈3\nu=3italic_ν = 3 is an exception, in that both unstrained and strained cases produce a supercrystal of pinned WMs; this is due to the congruence of intrinsic (that of the WM) and external (that of the confining potential of the MQD) C3subscript𝐶3C_{3}italic_C start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT point-group symmetries.

Furthermore, we demonstrated that the UHF approach (invoked in Refs. [36, 56]) utterly fails to describe the genuinely quantum-mechanical supercrystal of sliding WMs in the unstrained case, providing a qualitative agreement only in the case of a supercrystal of pinned WMs.

The insights gained in this work through systematic comparisons, employing the UHF method with exact diagonalization (FCI) calculations as benchmarks, provide impetus and pathways for the development of effective strategies for advancing economically feasible and accurate computational methodologies for treating strongly-correlated many-body states in materials, including the electron (hole) Wigner-molecular TMD superlattice systems studied here.

VII ackowledgements

This work has been supported by a grant from the Air Force Office of Scientific Research (AFOSR) under Award No. FA9550-21-1-0198. Calculations were carried out at the GATECH Center for Computational Materials Science.

Appendix A THE CONFIGURATION INTERACTION METHOD

The full configuration interaction (FCI) methodology has a long history, starting in quantum chemistry; see Refs. [40, 45]. The method was adapted to two dimensional problems and found extensive applications in the fields of semiconductor quantum dots [8, 9, 42, 16, 14, 43, 26, 44] and of the fractional quantum Hall effect [10, 22].

Our 2D FCI is described in our earlier publications. The reader will find a comprehensive exposition in Appendix B of Ref. [26], where the method was applied to GaAs double-quantum-dot quantum computer qubits. We specify that, in the application to moiré DQDs, we keep similar space orbitals, φj(x,y)subscript𝜑𝑗𝑥𝑦\varphi_{j}(x,y)italic_φ start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT ( italic_x , italic_y ), j=1,2,,K𝑗12𝐾j=1,2,\ldots,Kitalic_j = 1 , 2 , … , italic_K, that are employed in the building of the single-particle basis of spin-orbitals used to construct the Slater determinants ΨIsubscriptΨ𝐼\Psi_{I}roman_Ψ start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT, which span the many-body Hilbert space [see Eq. (B4) in Ref. [26]; the index I𝐼Iitalic_I counts the Slater determinants]. The orbitals φj(x,y)subscript𝜑𝑗𝑥𝑦\varphi_{j}(x,y)italic_φ start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT ( italic_x , italic_y ) are determined as solutions (in Cartesian coordinates) of the auxiliary Hamiltonian

Haux=𝐩22m+12mω02(x2+y2),subscript𝐻auxsuperscript𝐩22superscript𝑚12superscript𝑚superscriptsubscript𝜔02superscript𝑥2superscript𝑦2\displaystyle H_{\rm aux}=\frac{{\bf p}^{2}}{2m^{*}}+\frac{1}{2}m^{*}\omega_{0% }^{2}(x^{2}+y^{2}),italic_H start_POSTSUBSCRIPT roman_aux end_POSTSUBSCRIPT = divide start_ARG bold_p start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT end_ARG start_ARG 2 italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT end_ARG + divide start_ARG 1 end_ARG start_ARG 2 end_ARG italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT italic_ω start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ( italic_x start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + italic_y start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) , (6)

where mω02=16π2v0cos(ϕ)/aM2superscript𝑚superscriptsubscript𝜔0216superscript𝜋2subscript𝑣0italic-ϕsuperscriptsubscript𝑎𝑀2m^{*}\omega_{0}^{2}=16\pi^{2}v_{0}\cos(\phi)/a_{M}^{2}italic_m start_POSTSUPERSCRIPT ∗ end_POSTSUPERSCRIPT italic_ω start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT = 16 italic_π start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_v start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT roman_cos ( italic_ϕ ) / italic_a start_POSTSUBSCRIPT italic_M end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT, i.e., only the isotropic parabolic (harmonic) contribution of the MQD confinement VMQD(𝐫)subscript𝑉MQD𝐫V_{\rm MQD}({\bf r})italic_V start_POSTSUBSCRIPT roman_MQD end_POSTSUBSCRIPT ( bold_r ) [see Eqs. (2) and (4) in the main text] is included.

The matrix elements φi(x,y)|sin(3θ)r3|φj(x,y)quantum-operator-productsubscript𝜑𝑖𝑥𝑦3𝜃superscript𝑟3subscript𝜑𝑗𝑥𝑦\langle\varphi_{i}(x,y)|\sin(3\theta)r^{3}|\varphi_{j}(x,y)\rangle⟨ italic_φ start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ( italic_x , italic_y ) | roman_sin ( 3 italic_θ ) italic_r start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT | italic_φ start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT ( italic_x , italic_y ) ⟩ of the anisotropic term in the MQD confinement are calculated analytically using the algebraic language MATHEMATICA [57] and the Hermite-to-Laguerre (Cartesian-to-polar) transformations listed in Ref. [58], whereas the matrix elements φi(x,y)|HCF|φj(x,y)quantum-operator-productsubscript𝜑𝑖𝑥𝑦subscript𝐻CFsubscript𝜑𝑗𝑥𝑦\langle\varphi_{i}(x,y)|H_{\rm CF}|\varphi_{j}(x,y)\rangle⟨ italic_φ start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ( italic_x , italic_y ) | italic_H start_POSTSUBSCRIPT roman_CF end_POSTSUBSCRIPT | italic_φ start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT ( italic_x , italic_y ) ⟩ of the crystal field are calculated numerically.

Following Ref. [26], we use a sparse-matrix eigensolver based on Implicitly Restarted Arnoldi methods to diagonalize the many-body Hamiltonian in Eq. (3) of the main text. We stress that the FCI conserves both the total spin and the total-spin projection.

Appendix B THE SPIN-AND-SPACE UNRESTRICTED HARTREE-FOCK

Early on in the context of 2D materials, the spin-and-space unrestricted Hartree-Fock (sS-UHF) was employed in Ref. [1] to describe formation of Wigner molecules at the mean-field level. This methodology employs the Pople-Nesbet equations [45, 14]. The sS-UHF WMs are self-consistent solutions of the Pople-Nesbet equations that are obtained by relaxing both the total-spin and space symmetry requirements. For a detailed description of the Pople-Nesbet equations in the context of three-dimensional natural atoms and molecules, see Ch. 3.8 in Ref. [45]. For a detailed description of the Pople-Nesbet equations in the context of two-dimensional artificial atoms and semiconductor quantum dots, see Sec. 2.1 of Ref. [14]. We note that the Pople-Nesbet equations conserve the total-spin projection, but not the total spin. Convergence of the self-consistent iterations was achieved in all cases by mixing the input and output charge densities at each iteration step. The convergence criterion was set to a difference of 1012superscript101210^{-12}10 start_POSTSUPERSCRIPT - 12 end_POSTSUPERSCRIPT meV between the input and output total UHF energies at the same iteration step.
    

Appendix C CHARGE DENSITIES FROM FCI AND UHF WAVE FUNCTIONS

The FCI single-particle density (charge density) is the expectation value of a one-body operator

ρ(𝐫)=ΦFCI|i=1Nδ(𝐫𝐫i)|ΦFCI,𝜌𝐫quantum-operator-productsuperscriptΦFCIsuperscriptsubscript𝑖1𝑁𝛿𝐫subscript𝐫𝑖superscriptΦFCI\rho({\bf r})=\langle\Phi^{\rm FCI}|\sum_{i=1}^{N}\delta({\bf r}-{\bf r}_{i})|% \Phi^{\rm FCI}\rangle,italic_ρ ( bold_r ) = ⟨ roman_Φ start_POSTSUPERSCRIPT roman_FCI end_POSTSUPERSCRIPT | ∑ start_POSTSUBSCRIPT italic_i = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_N end_POSTSUPERSCRIPT italic_δ ( bold_r - bold_r start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) | roman_Φ start_POSTSUPERSCRIPT roman_FCI end_POSTSUPERSCRIPT ⟩ , (7)

where ΦFCIsuperscriptΦFCI\Phi^{\rm FCI}roman_Φ start_POSTSUPERSCRIPT roman_FCI end_POSTSUPERSCRIPT denotes the many-body (multi-determinantal) FCI wave function, namely,

ΦFCI(𝐫1,,𝐫N)=ICIΨI(𝐫1,,𝐫N),superscriptΦFCIsubscript𝐫1subscript𝐫𝑁subscript𝐼subscript𝐶𝐼subscriptΨ𝐼subscript𝐫1subscript𝐫𝑁\Phi^{\rm FCI}({\bf r}_{1},\ldots,{\bf r}_{N})=\sum_{I}C_{I}\Psi_{I}({\bf r}_{% 1},\ldots,{\bf r}_{N}),roman_Φ start_POSTSUPERSCRIPT roman_FCI end_POSTSUPERSCRIPT ( bold_r start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , … , bold_r start_POSTSUBSCRIPT italic_N end_POSTSUBSCRIPT ) = ∑ start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT italic_C start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT roman_Ψ start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ( bold_r start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , … , bold_r start_POSTSUBSCRIPT italic_N end_POSTSUBSCRIPT ) , (8)

with ΨI(𝐫)subscriptΨ𝐼𝐫\Psi_{I}({\bf r)}roman_Ψ start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT ( bold_r ) denoting the Slater determinants that span the many-body Hilbert space. The expansion coefficients CIsubscript𝐶𝐼C_{I}italic_C start_POSTSUBSCRIPT italic_I end_POSTSUBSCRIPT are a byproduct of the exact diagonalization of the many-body Hamiltonian HMBsubscript𝐻MBH_{\rm MB}italic_H start_POSTSUBSCRIPT roman_MB end_POSTSUBSCRIPT.

For the sS-UHF case, one substitutes ΦFCIsuperscriptΦFCI\Phi^{\rm FCI}roman_Φ start_POSTSUPERSCRIPT roman_FCI end_POSTSUPERSCRIPT in Eq. (7) with the single-determinant, ΨUHF(𝐫)superscriptΨUHF𝐫\Psi^{\rm UHF}({\bf r})roman_Ψ start_POSTSUPERSCRIPT roman_UHF end_POSTSUPERSCRIPT ( bold_r ), solution of the Pople-Nesbet equations. ΨUHF(𝐫)superscriptΨUHF𝐫\Psi^{\rm UHF}({\bf r})roman_Ψ start_POSTSUPERSCRIPT roman_UHF end_POSTSUPERSCRIPT ( bold_r ) is built out from the UHF spin-orbitals whose space part has the form:

uiα=μ=1K𝒞μiαφμ,i=1,,K,formulae-sequencesubscriptsuperscript𝑢𝛼𝑖superscriptsubscript𝜇1𝐾subscriptsuperscript𝒞𝛼𝜇𝑖subscript𝜑𝜇𝑖1𝐾\displaystyle u^{\alpha}_{i}=\sum_{\mu=1}^{K}{\cal C}^{\alpha}_{\mu i}\varphi_% {\mu},\;\;\;i=1,\ldots,K,italic_u start_POSTSUPERSCRIPT italic_α end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT = ∑ start_POSTSUBSCRIPT italic_μ = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_K end_POSTSUPERSCRIPT caligraphic_C start_POSTSUPERSCRIPT italic_α end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_μ italic_i end_POSTSUBSCRIPT italic_φ start_POSTSUBSCRIPT italic_μ end_POSTSUBSCRIPT , italic_i = 1 , … , italic_K , (9)

and

uiβ=μ=1K𝒞μiβφμ,i=1,,K,formulae-sequencesubscriptsuperscript𝑢𝛽𝑖superscriptsubscript𝜇1𝐾subscriptsuperscript𝒞𝛽𝜇𝑖subscript𝜑𝜇𝑖1𝐾\displaystyle u^{\beta}_{i}=\sum_{\mu=1}^{K}{\cal C}^{\beta}_{\mu i}\varphi_{% \mu},\;\;\;i=1,\ldots,K,italic_u start_POSTSUPERSCRIPT italic_β end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT = ∑ start_POSTSUBSCRIPT italic_μ = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_K end_POSTSUPERSCRIPT caligraphic_C start_POSTSUPERSCRIPT italic_β end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_μ italic_i end_POSTSUBSCRIPT italic_φ start_POSTSUBSCRIPT italic_μ end_POSTSUBSCRIPT , italic_i = 1 , … , italic_K , (10)

where the expansion coefficients 𝒞μiαsubscriptsuperscript𝒞𝛼𝜇𝑖{\cal C}^{\alpha}_{\mu i}caligraphic_C start_POSTSUPERSCRIPT italic_α end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_μ italic_i end_POSTSUBSCRIPT and 𝒞μiβsubscriptsuperscript𝒞𝛽𝜇𝑖{\cal C}^{\beta}_{\mu i}caligraphic_C start_POSTSUPERSCRIPT italic_β end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_μ italic_i end_POSTSUBSCRIPT are solutions of the Pople-Nesbet equations, with α𝛼\alphaitalic_α and β𝛽\betaitalic_β denoting up and down spins, respectively.

The UHF expression for the charge density simplifies to:

ρUHF(𝐫)=i=1Nα|uiα(𝐫)|2+j=1Nβ|ujβ(𝐫)|2,superscript𝜌UHF𝐫superscriptsubscript𝑖1superscript𝑁𝛼superscriptsubscriptsuperscript𝑢𝛼𝑖𝐫2superscriptsubscript𝑗1superscript𝑁𝛽superscriptsubscriptsuperscript𝑢𝛽𝑗𝐫2\displaystyle\rho^{\rm UHF}({\bf r})=\sum_{i=1}^{N^{\alpha}}|u^{\alpha}_{i}({% \bf r})|^{2}+\sum_{j=1}^{N^{\beta}}|u^{\beta}_{j}({\bf r})|^{2},italic_ρ start_POSTSUPERSCRIPT roman_UHF end_POSTSUPERSCRIPT ( bold_r ) = ∑ start_POSTSUBSCRIPT italic_i = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_N start_POSTSUPERSCRIPT italic_α end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT | italic_u start_POSTSUPERSCRIPT italic_α end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ( bold_r ) | start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT + ∑ start_POSTSUBSCRIPT italic_j = 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT italic_N start_POSTSUPERSCRIPT italic_β end_POSTSUPERSCRIPT end_POSTSUPERSCRIPT | italic_u start_POSTSUPERSCRIPT italic_β end_POSTSUPERSCRIPT start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT ( bold_r ) | start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT , (11)

with Nαsuperscript𝑁𝛼{N^{\alpha}}italic_N start_POSTSUPERSCRIPT italic_α end_POSTSUPERSCRIPT, Nβsuperscript𝑁𝛽{N^{\beta}}italic_N start_POSTSUPERSCRIPT italic_β end_POSTSUPERSCRIPT being the number of lowest-energy occupied spin-up and spin-down orbitals, respectively.

References

  • Yannouleas and Landman [1999] C. Yannouleas and U. Landman, Spontaneous Symmetry Breaking in Single and Molecular Quantum Dots, Phys. Rev. Lett. 82, 5325–5328 (1999).
  • Egger et al. [1999] R. Egger, W. Häusler, C. H. Mak, and H. Grabert, Crossover from Fermi liquid to Wigner molecule behavior in quantum dots, Phys. Rev. Lett. 82, 3320–3323 (1999).
  • Yannouleas and Landman [2000a] C. Yannouleas and U. Landman, Collective and Independent-Particle Motion in Two-Electron Artificial Atoms, Phys. Rev. Lett. 85, 1726–1729 (2000a).
  • Filinov et al. [2001] A. V. Filinov, M. Bonitz, and Y. E. Lozovik, Wigner Crystallization in Mesoscopic 2D Electron Systems, Phys. Rev. Lett. 86, 3851–3854 (2001).
  • Yannouleas and Landman [2002a] C. Yannouleas and U. Landman, Strongly correlated wavefunctions for artificial atoms and molecules, Journal of Physics: Condensed Matter 14, L591–L598 (2002a).
  • Mikhailov [2002] S. A. Mikhailov, Two ground-state modifications of quantum-dot beryllium, Phys. Rev. B 66, 153313 (2002).
  • Harju et al. [2002] A. Harju, S. Siljamäki, and R. M. Nieminen, Wigner molecules in quantum dots: A quantum monte carlo study, Phys. Rev. B 65, 075309 (2002).
  • Yannouleas and Landman [2003] C. Yannouleas and U. Landman, Two-dimensional quantum dots in high magnetic fields: Rotating-electron-molecule versus composite-fermion approach, Phys. Rev. B 68, 035326 (2003).
  • Tavernier et al. [2003] M. B. Tavernier, E. Anisimovas, F. M. Peeters, B. Szafran, J. Adamowski, and S. Bednarek, Four-electron quantum dot in a magnetic field, Phys. Rev. B 68, 205305 (2003).
  • Yannouleas and Landman [2004] C. Yannouleas and U. Landman, Structural properties of electrons in quantum dots in high magnetic fields: Crystalline character of cusp states and excitation spectra, Phys. Rev. B 70, 235319 (2004).
  • Szafran et al. [2004] B. Szafran, F. M. Peeters, S. Bednarek, and J. Adamowski, Anisotropic quantum dots: Correspondence between quantum and classical Wigner molecules, parity symmetry, and broken-symmetry states, Phys. Rev. B 69, 125344 (2004).
  • Romanovsky et al. [2006] I. Romanovsky, C. Yannouleas, L. O. Baksmaty, and U. Landman, Bosonic molecules in rotating traps, Phys. Rev. Lett. 97, 090401 (2006).
  • Li et al. [2006] Y. Li, C. Yannouleas, and U. Landman, From a few to many electrons in quantum dots under strong magnetic fields: Properties of rotating electron molecules with multiple rings, Phys. Rev. B 73, 075301 (2006).
  • Yannouleas and Landman [2007] C. Yannouleas and U. Landman, Symmetry breaking and quantum correlations in finite systems: Studies of quantum dots and ultracold Bose gases and related nuclear and chemical methods, Reports on Progress in Physics 70, 2067–2148 (2007).
  • Dai et al. [2007] Z. Dai, J.-L. Zhu, N. Yang, and Y. Wang, Spin-dependent rotating Wigner molecules in quantum dots, Phys. Rev. B 76, 085308 (2007).
  • Li et al. [2007] Y. Li, C. Yannouleas, and U. Landman, Three-electron anisotropic quantum dots in variable magnetic fields: Exact results for excitation spectra, spin structures, and entanglement, Phys. Rev. B 76, 245310 (2007).
  • Baksmaty et al. [2007] L. O. Baksmaty, C. Yannouleas, and U. Landman, Rapidly rotating boson molecules with long- or short-range repulsion: An exact diagonalization study, Phys. Rev. A 75, 023620 (2007).
  • Ghosal et al. [2007] A. Ghosal, A. D. Güçlü, C. J. Umrigar, D. Ullmo, and H. U. Baranger, Incipient Wigner localization in circular quantum dots, Phys. Rev. B 76, 085341 (2007).
  • Yang et al. [2008] N. Yang, J.-L. Zhu, and Z. Dai, Rotating Wigner molecules and spin-related behaviors in quantum rings, Journal of Physics: Condensed Matter 20, 295202 (2008).
  • Romanovsky et al. [2009] I. Romanovsky, C. Yannouleas, and U. Landman, Edge states in graphene quantum dots: Fractional quantum Hall effect analogies and differences at zero magnetic field, Phys. Rev. B 79, 075311 (2009).
  • Brandt et al. [2015] B. B. Brandt, C. Yannouleas, and U. Landman, Double-well ultracold-fermions computational microscopy: Wave-function anatomy of attractive-pairing and Wigner-molecule entanglement and natural orbitals, Nano Letters 15, 7105–7111 (2015).
  • Yannouleas and Landman [2021] C. Yannouleas and U. Landman, Exact closed-form analytic wave functions in two dimensions: Contact-interacting fermionic spinful ultracold atoms in a rapidly rotating trap, Phys. Rev. Research 3, L032028 (2021).
  • Ercan et al. [2021] H. E. Ercan, S. N. Coppersmith, and M. Friesen, Strong electron-electron interactions in Si/SiGe quantum dots, Phys. Rev. B 104, 235302 (2021).
  • Abadillo-Uriel et al. [2021] J. C. Abadillo-Uriel, B. Martinez, M. Filippone, and Y.-M. Niquet, Two-body Wigner molecularization in asymmetric quantum dot spin qubits, Phys. Rev. B 104, 195305 (2021).
  • Yannouleas and Landman [2022a] C. Yannouleas and U. Landman, Wigner molecules and hybrid qubits, J. Phys.: Condens. Matter (Letter) 34, 21LT01 (2022a).
  • Yannouleas and Landman [2022b] C. Yannouleas and U. Landman, Molecular formations and spectra due to electron correlations in three-electron hybrid double-well qubits, Phys. Rev. B 105, 205302 (2022b).
  • Ellenberger et al. [2006] C. Ellenberger, T. Ihn, C. Yannouleas, U. Landman, K. Ensslin, D. Driscoll, and A. C. Gossard, Excitation Spectrum of Two Correlated Electrons in a Lateral Quantum Dot with Negligible Zeeman Splitting, Phys. Rev. Lett. 96, 126806 (2006).
  • Kalliakos et al. [2008] S. Kalliakos, M. Rontani, V. Pellegrini, C. P. García, A. Pinczuk, G. Goldoni, E. Molinari, L. N. Pfeiffer, and K. W. West, A molecular state of correlated electrons in a quantum dot, Nature Physics 4, 467–471 (2008).
  • Jang et al. [2021] W. Jang, M.-K. Cho, H. Jang, J. Kim, J. Park, G. Kim, B. Kang, H. Jung, V. Umansky, and D. Kim, Single-Shot Readout of a Driven Hybrid Qubit in a GaAs Double Quantum Dot, Nano Letters 21, 4999–5005 (2021).
  • Jang et al. [2023] W. Jang, J. Kim, J. Park, G. Kim, M.-K. Cho, H. Jang, S. Sim, B. Kang, H. Jung, V. Umansky, and D. Kim, Wigner-molecularization-enabled dynamic nuclear polarization, Nature Communications 14, 2948 (2023).
  • Corrigan et al. [2021] J. Corrigan, J. P. Dodson, H. E. Ercan, J. C. Abadillo-Uriel, B. Thorgrimsson, T. J. Knapp, N. Holman, T. McJunkin, S. F. Neyens, E. R. MacQuarrie, R. H. Foote, L. F. Edge, M. Friesen, S. N. Coppersmith, and M. A. Eriksson, Coherent control and spectroscopy of a semiconductor quantum dot Wigner molecule, Phys. Rev. Lett. 127, 127701 (2021).
  • Palma et al. [2023] F. D. Palma, F. Oppliger, W. Jang, S. Bosco, M. Janík, S. Calcaterra, G. Katsaros, G. Isella, D. Loss, and P. Scarlino, Strong hole-photon coupling in planar ge: probing the charge degree and wigner molecule states (2023), arXiv:2310.20661 [quant-ph] .
  • Pecker et al. [2013] S. Pecker, F. Kuemmeth, A. Secchi, M. Rontani, D. C. Ralph, P. L. McEuen, and S. Ilani, Observation and spectroscopy of a two-electron Wigner molecule in an ultraclean carbon nanotube, Nature Physics 9, 576–581 (2013).
  • Yannouleas and Landman [2023] C. Yannouleas and U. Landman, Quantum Wigner molecules in moiré materials, Phys. Rev. B 108, L121411 (2023).
  • Yannouleas and Landman [2024] C. Yannouleas and U. Landman, Wigner-molecule supercrystal in transition metal dichalcogenide moiré superlattices: Lessons from the bottom-up approach, Phys. Rev. B 109, L121302 (2024).
  • Li et al. [2023] H. Li, Z. Xiang, A. P. Reddy, T. Devakul, R. Sailus, R. Banerjee, T. Taniguchi, K. Watanabe, S. Tongay, A. Zettl, L. Fu, M. F. Crommie, and F. Wang, Wigner molecular crystals from multi-electron moiré artificial atoms (2023), arXiv:2312.07607 [cond-mat.mes-hall] .
  • Note [1] We note that the experimental results for the unstrained case in Ref. [36] were necessarily interpreted with formation of sliding WMs in isolated MQDs (for N=2𝑁2N=2italic_N = 2 confined charge carriers, see Refs. [34, 36]; for N=4𝑁4N=4italic_N = 4 confined charge carriers, see Ref. [34]), given the fact that the mean-field Hartree-Fock approach (that allows calculations for a large number of charge carriers distributed over several moiré pockets) is unreliable [35, 36] for describing sliding WMs.
  • Bethe [1929] H. Bethe, Termaufspaltung in Kristallen, Annalen der Physik 395, 133–208 (1929).
  • Van Vleck [1932] J. H. Van Vleck, Theory of the variations in paramagnetic anisotropy among different salts of the iron group, Phys. Rev. 41, 208–215 (1932).
  • Shavitt [1998] I. Shavitt, The history and evolution of configuration interaction, Molecular Physics 94, 3–17 (1998).
  • Yannouleas and Landman [2006] C. Yannouleas and U. Landman, Electron and boson clusters in confined geometries: Symmetry breaking in quantum dots and harmonic traps, Proceedings of the National Academy of Sciences 103, 10600–10605 (2006).
  • Rontani et al. [2006] M. Rontani, C. Cavazzoni, D. Bellucci, and G. Goldoni, Full configuration interaction approach to the few-electron problem in artificial atoms, The Journal of Chemical Physics 124, 124102 (2006).
  • Li et al. [2009] Y. Li, C. Yannouleas, and U. Landman, Artificial quantum-dot Helium molecules: Electronic spectra, spin structures, and Heisenberg clusters, Phys. Rev. B 80, 045326 (2009).
  • Yannouleas and Landman [2022c] C. Yannouleas and U. Landman, Valleytronic full configuration-interaction approach: Application to the excitation spectra of si double-dot qubits, Phys. Rev. B 106, 195306 (2022c).
  • Szabo and Ostlund [1989] A. Szabo and N. S. Ostlund, Modern Quantum Chemistry (McGraw-Hill, New York, 1989) For the Slater-Condon rules, see Chap. 4.
  • Yannouleas and Landman [2002b] C. Yannouleas and U. Landman, Magnetic-field manipulation of chemical bonding in artificial molecules, International Journal of Quantum Chemistry 90, 699–708 (2002b).
  • Yannouleas and Landman [2000b] C. Yannouleas and U. Landman, Formation and control of electron molecules in artificial atoms: Impurity and magnetic-field effects, Phys. Rev. B 61, 15895–15904 (2000b).
  • Wu et al. [2018] F. Wu, T. Lovorn, E. Tutuc, and A. H. MacDonald, Hubbard model physics in transition metal dichalcogenide moiré bands, Phys. Rev. Lett. 121, 026402 (2018).
  • Angeli and MacDonald [2021] M. Angeli and A. H. MacDonald, ΓΓ\Gammaroman_Γ-valley transition metal dichalcogenide moiré bands, Proceedings of the National Academy of Sciences 118, e2021826118 (2021).
  • Zhang et al. [2020] Y. Zhang, N. F. Q. Yuan, and L. Fu, Moiré quantum chemistry: Charge transfer in transition metal dichalcogenide superlattices, Phys. Rev. B 102, 201115 (2020).
  • Wigner [1934] E. Wigner, On the interaction of electrons in metals, Phys. Rev. 46, 1002–1011 (1934).
  • Wigner [1938] E. Wigner, Effects of the electron interaction on the energy levels of electrons in metals, Trans. Faraday Soc. 34, 678–685 (1938).
  • Li et al. [2021] H. Li, S. Li, E. C. Regan, D. Wang, W. Zhao, S. Kahn, K. Yumigeta, M. Blei, T. Taniguchi, K. Watanabe, S. Tongay, A. Zettl, M. F. Crommie, and F. Wang, Imaging two-dimensional generalized Wigner crystals, Nature 597, 650–654 (2021).
  • Tsui et al. [2024] Y.-C. Tsui, M. He, Y. Hu, E. Lake, T. Wang, K. Watanabe, T. Taniguchi, M. P. Zaletel, and A. Yazdani, Direct observation of a magnetic-field-induced wigner crystal, Nature 628, 287–292 (2024).
  • Note [2] When the QD confinement possesses circular symmetry, the sliding WMs were earlier referred to as rotating WMs or rotating electron molecules [14, 8].
  • Reddy et al. [2023] A. P. Reddy, T. Devakul, and L. Fu, Artificial Atoms, Wigner Molecules, and an Emergent Kagome Lattice in Semiconductor Moiré Superlattices, Phys. Rev. Lett. 131, 246501 (2023).
  • [57] Wolfram Research, Inc., Mathematica, Version 13.2, Champaign, IL, 2022.
  • Kimel and Elias [1993] I. Kimel and L. Elias, Relations between Hermite and Laguerre Gaussian modes, IEEE Journal of Quantum Electronics 29, 2562–2567 (1993).