betaFeedback

WorksPhysics and Astronomy

Super Moiré Domain Tessellations, Sliding Ferroelectricity, and Reconfigurable Quantum Dot Arrays in Twisted Trilayer Hexagonal Boron Nitride

Abstract

At very small twist angles, bilayer moiré systems exhibit characteristic stacking domain patterns, where the moiré length scale is determined solely by the twist angle. In contrast, the additional stacking and twisting degrees of freedom in twisted trilayer systems give rise to richer and more intricate domain tessellations. In twisted trilayer hexagonal boron nitride (TTBN), the interplay between polar and nonpolar domains and their domain walls is shown to result in unconventional responses to external electric fields, including electric-field tunability of the moiré-of-moiré or super moiré pattern–features absent in bilayer counterparts. We demonstrate that at the vertices of super moiré domains, TTBN can support arrays of quantum dots hosting localized quantum harmonic oscillator (QHO) states with diverse spatial symmetries. Futhermore, we show that the shape of the array and the spacing between the localized QHO states can be dynamically reconfigured by electric fields, enabling facile switching between fully isolated and strongly coupled regimes. The local potentials for the quantum dot state are predicted to be sufficiently deep to support a series of QHO states with nonzero angular momentum. This tunability enables control over the transport of quantum dot states and their interdot coupling, facillitating long-range quantum state transfer. Combined with the feasibility of large-scale fabrication of homogeneous twisted trilayer materials, these properties position TTBN as a promising platform for a wide range of quantum technologies.

Introduction

Substantial progress has been achieved in the development of quantum hardware designed to tackle computationally intractable problems that lie beyond the reach of classical computing architectures [1, 2, 3, 4, 5, 6, 7]. Multiple platforms have emerged as viable candidates, each with distinct advantages [6]. The cornerstone for these implementations lies in precise controls over individual quantum states as well as their mutual interactions [6]. This requires the reliable preparation of well-defined quantum objects such as atoms [2, 3], ions [4, 5], superconducting circuits [7], spin defects [8] and quantum dots (QDs) [9, 10, 11]. Simultaneously, it also necessitates advanced engineerings for scalable and on-demand development of many quantum objects with homogeneous quality controls [6].

Among these candidates, neutral atom and ion arrays have demonstrated exceptional quantum control capabilities over a large number of individual quantum units [2, 3, 4, 5]. In addition, superconducting circuits have already shown significant advancement [7]. Considering well-established nanofabrication technology of semiconductor industry, solid-state implementations, particularly QDs, also hold considerable promise [9, 10, 11]. Despite impressive progress in quantum information applications using QD arrays [9, 10, 11], significant challenges still remain in achieving their large-scale arrangements with uniform electronic properties [6] and their controlled on-demand reconfigurations [12, 13] — capabilities that are more readily available in atomic physics platforms [3].

Concurrently, intensive research efforts have focused on engineering quantum states through the controlled stacking of two-dimensional (2D) materials with twist angles between adjacent layers [14, 15]. These configurations generate large-scale periodic superlattices, known as moiré lattices [14, 15], that extend beyond conventional atomic periodicity. In twisted bilayer systems, these structures exhibit various notable quantum phenomena through interactions between quasiparticles or excitations modulated by moiré periodic potentials [15, 16, 17]. Furthermore, the versatility afforded by combining various characteristic 2D materials creates valuable opportunities for realizing unique quantum devices [18].

For quantum phenomena observed in moiré systems, the electron localizations governed by superlattice potentials have attracted particular interest due to their potential to realize strongly correlated physics as well as various quantum device applications [14, 15, 16, 17]. In twisted bilayer systems, localized states arise at the centers or vertices of superlattices from moiré potentials, forming ordered arrays of QDs [19, 20, 21, 22, 23, 24, 25, 26, 27]. A major limitation of bilayer systems is the inherent rigidity of the arrays, as their positions and symmetries are determined by the twist angle. Consequently, these configurations remain unaffected by noninvasive external perturbations, such as electric fields, rendering them analogous to conventional QD devices. While recent progress in in-situ mechanical rotation techniques [28, 29] enables partial control over the moiré supercell size, this approach could only facilitate the collective displacement of the entire dots in the array. So, achieving flexible and reconfigurable local QD patterns as demonstrated in the array of Rydberg atoms [3] remains a challenge, prompting the exploration of alternative approaches to unlock new functionalities.

Advances in twisted trilayer systems could provide a chance to overcome this limitation. Recent studies on their moiré-of-moiré or super moiré lattices demonstrate various domain structures with energetically disparate stacking orders, sharply distinguished from the bilayer ones having the energetically equivalent stackings for all domains [30, 31, 32, 33]. Notably, twisted trilayer graphene (TTG) with very small twist angles are shown to realize distinct domain lattices, for which the driving force is the total energy difference between different stackings [33]. So, if one can control the energy for each domain, their local configurations will alter, presenting a new opportunity to control their electronic properties [33].

In this work, we theoretically demonstrate that twisted trilayer hexagonal boron nitride (TTBN) forms well-ordered QD arrays with diverse arrangements and novel reconfigurability based on its ferroelectricity. For twist angles near 0.1 degrees, TTBNs are shown to generate intricate super moiré domain lattices, including triangular, kagome, and hexagram patterns. The localized quantum states at vertices of these lattices exhibit deep confinement, with binding energies of a few hundred millielectron volts, and display discrete energy eigenstates analogous to those of quantum harmonic oscillators (QHOs). Consequently, the uniform QD states could be possible across the whole array provided the superstructure maintains proper orderings [30, 31, 32, 33].

Furthermore, multilayer hexagonal boron nitride (h-BN) can exhibit out-of-plane ferroelectricity, arising purely from broken inversion symmetry induced by the stacking configuration, rather than conventional ionic displacements [34, 35]. In bilayer hexagonal boron nitride (BBN), this polarization is directly coupled with external electric fields and results in an ultrafast and reversible change of polarizations through sliding motions of each layer, offering a new platform for robust and switchable electronic functionality [36, 37]. In TTBN, the response is no longer trivial and, depending on the type of tessellation, dynamic repositioning of the QD states at domain vertices is possible. This enables seamless local transitions between isolated, capacitively interacting, and tunnel-coupled QDs [38], presenting a promising route toward various forms of quantum hybridization. Our findings suggest that twisted van der Waals (vdW) materials systems hold promise for quantum device applications, including information processing.

Materials and Methods

Stacking orders and polarizations

The h-BN has a honeycomb structure with boron (B) and nitrogen (N) atoms occupying each sublattice. Here, for multilayer h-BN, we focus on parallel stack configurations only, where the second layer is positioned directly above the first layer without rotation. Among the various parallel stackings accessible through in-plane sliding, the

AB and BA stackings are most favorable and energetically equivalent [39, 40]. In the AB (BA) stacking, boron (nitrogen) atoms in the top layer are positioned directly above nitrogen (boron) atoms in the bottom layer, with the remaining atoms in each layer aligned with vacant hexagon centers as shown in Figure 1 (a). We note that BA stacking can be assigned as AC stacking also. The AB and BA stackings possess opposite out-of-plane polarizations as denoted in Figure 1 (a) [39, 40]. For the twisted BBN (TTBN), a moiré pattern emerges where triangular regions of AB and BA stackings with opposite polarization are arranged in an alternating configuration [34, 35].

Stacking orders of h-BN bilayers and trilayers

Figure 1. Stacking orders of h-BN bilayers and trilayers. (a) Ball and stick models of bilayer h-BN with AB and BA stackings and their polarizations. Trilayer h-BN structures and their polarizations. (b) The ABA and ACA stacking orders are nonpolar and (c) ABC and ACB stackings exhibit opposite polarizations.

In contrast to TTBN, its trilayer counterpart exhibits four distinct stackings. Among these, the nonpolar ABA (ACA) stacking configurations have AB (AC) stacking for the pair of the first and second layers, and BA (CA) stacking for the pair of the second and third layers, respectively. As a result, the polarization between the first (bottom) and second (middle) layers cancels the other between the second and third (top) layers, as illustrated in Figure 1 (b). The remaining two polar ABC (ACB) stacking are defined by two successive AB (AC) and BC (CB) stackings, thus almost doubling the total polarization. The resulting polarizations in ABC and ACB are opposite to each other, as shown in Figure 1 (c).

Computational methods for lattice relaxations and electronic structures

We developed and employed a newly designed interatomic potential method [33, 41] to obtain fully relaxed TTBN structures containing millions of atoms in a commensurate super cell. Note that an incommensurate case can be represented by a nearby commensurate condition without loss of generality [33]. Our method achieved excellent agreement with ab initio calculations while remaining computationally efficient for large-scale systems [41]. This method integrates harmonic intralayer and long-ranged interlayer potentials, under the assumption that the intralayer potential remains unaffected by the interlayer atomic registry. Due to the large in-plane elastic constants of h-BN, the in-plane strain is found to remain below 0.5 %, justifying the neglect of anharmonic effects. The intralayer potential was constructed by fitting to phonon dispersions of monolayer h-BN obtained from density functional perturbation theory (DFPT) [42], showing excellent agreement with the reference data (see Appendix A for detailed information).

To accurately describe interlayer interactions, we extended the Kolmogorov-Crespi (KC) potential by incorporating long-range interactions across all three layers [33, 43]. While the KC potential provides an accurate description of bilayer interactions, it fails to distinguish between Bernal and rhombohedral h-BN trilayer stackings. This interaction was also found to be crucial for small-angle TTG [33]. To resolve this shortcoming, an additional Gaussian-type second-nearest-neighbor interaction between the first and third h-BN layers is introduced, enabling our potential model to reproduce the energy differences between ABA and ABC stackings in close agreement with ab initio benchmarks (see Appendix A for detailed parameters of the modified KC potential and the additional long-range potential).

Furthermore, we incorporate the out-of-plane electric polarization of PzP_z arising between two h-BN layers, which governs the material response to external electric fields. To quantitatively capture disparate PzP_z under various conditions, we introduced a pairwise dipole model having both screened and Gaussian-overlap functional forms, and parameterized it against ab initio calculations. This model accurately reproduces the dependence of PzP_z on various stacking arrangements and interlayer spacings, including intermediate configurations between AA and AB stackings, and successfully captures the polarization patterns at polar domain boundaries [25] of TBBN under applied electric fields (see details in Appendix A).

To compute the electronic structure, we developed an efficient and accurate method based on a tight-binding (TB) approximation. The reference electronic structures of few-layer h-BN were calculated using self-consistent extended Hubbard-corrected density functional theory (DFT+UU+VV) [44, 45, 46]. This method reproduces the band gap of monolayer h-BN with high accuracy (5.95 eV), comparable to the GWGW approximation [47], and provides on-site (UU) and intersite (VV) Hubbard parameters, which were self-consistently determined for monolayer h-BN and verified to be applicable to bilayer and trilayer systems. Thus, our calculations are free of band gap errors in typical DFT calculations [48]. Using these results, we obtained Wannier functions constructed from low-energy states using the Wannier90 code [49]. Details for ab initio calculations are in Appendix B.

The TB model describes the trilayer h-BN using pzp_z orbitals at each atom as basis states, with intralayer

hopping parameters up to seventh-nearest neighbors fitted to the DFT+UU+VV Wannier Hamiltonian. Interlayer hopping up to next-nearest-layer is parametrized using distance and direction dependent functions, reproducing band splittings and detailed band-edge features shown in ab initio calculations for trilayers. Additionally, our TB model incorporates on-site energy corrections reflecting local dipole moments arising at the two interfaces depending on the local stacking profile. Effects of external electric fields were simulated by adding a position-dependent potential with a fitted dielectric constant against corresponding first-principles calculations. Detailed parameters are described in Appendix B.

Results

Unconventional super moiré lattices and polarization domains

Among the various trilayer stacking orders without twist shown in Figure 1, the nonpolar stacking configurations (ABA and ACA) exhibit a slightly lower total energy, less than 0.5 meV per atom, compared to the polar stackings (ABC and ACB) [50]. Due to the competing stacking orders, TTBN with small twist angles can exhibit a variety of super moiré lattices, as illustrated in Figure 2. Similar to prior studies on TTG [33], as the twist angles decrease to near 0.1∘0.1^\circ, the energy gain from increasing the area of the lowest-energy stacking configuration can surpass the energy cost associated with the growing boundaries between domains. In contrast to marginally TTG, however, TTBN structures not only feature domain lattices with varying stackings but also possess an associated network of polarization domains, which may not precisely follow the pattern of the stacking domains as will be discussed below.

Super moiré domain lattices and associated polarization maps

Figure 2. Super moiré domain lattices and associated polarization maps. (a) Twist angles for alternate stacking TTBN with θ12=−θ\theta_{12}=-\theta and θ23=2θ\theta_{23}=2\theta, and (b) its relaxed structure for a case of θ=0.082∘\theta=0.082^\circ showing various stacking domains (four different colors) in the left panel and its associated polarization distribution in the right panel. Red and blue colored polarizations follow directions in Fig. 1. The black scale bar is 100 nm. (c) and (d), Twist angles, relaxed structures and polarization distributions for helical stacking with θ12=θ23=θ=0.082∘\theta_{12}=\theta_{23}=\theta=0.082^\circ. (e) and (f), Similar displays for a twisted monolayer-bilayer with θ=0.082∘\theta=0.082^\circ.

Considering various domain shapes, we will focus on three representative ones having characteristic sets of twist angles. The first case is the alternate stacking where the twist angle (θ12\theta_{12}) between the first and second layer has an opposite direction with respect to the twist angle (θ23\theta_{23}) between second and third layer as shown in Figure 2 (a). For a specific case with θ23=−2θ12\theta_{23}=-2\theta_{12}, the relaxed super moiré lattice shows a tessellation of triangular domains with different stackings, as shown in Figure 2 (b), which consists of hexagonal arrangements of alternating triangular domains of ABA and ACA stackings surrounded by triangular domains of ABC and ACB stackings. Because of the nonpolar nature of ABA and ACA orders, the associated polarization domain shows a kagome-shaped distribution of alternating out-of-plane directions as displayed in Figure 2 (b).

Second, we will consider an opposite twist geometry to the first one, where two twist angles have the same direction. In the case of helical stacking with θ12=θ23\theta_{12}=\theta_{23} shown in Figure 2 (c), the relaxed structure shows a network of corner-shared hexagrams as shown in Figure 2 (d), being similar to the TTG case [33]. The hexagonal area of ACA stacking and rhombic area of ABA stacking are nonpolar so that the resulting polarization distribution displays an alternating set of triangular domain clusters with opposite polarization directions as shown in Figure 2 (d) and that all domain clusters with opposite polarizations are connected through their vertices.

The third configuration is the twisted monolayer-bilayer, a bilayer (θ23=0\theta_{23}=0) on top of a monolayer with a twist by θ12>0\theta_{12}>0 as shown in Figure 2 (e). This twist geometry displays two distinguishable domain tessellations. Figure 2 (f) shows one of the two domains, the kagome-like twisted monolayer-bilayer, which has undergone a spontaneous symmetry breaking at a smaller twist angle [33]. Here, being similar to the case in TTG [33], the hexagonal Bernal stacked region is divided into ABA and ACA stackings by a nematic domain boundary, which is surrounded by alternating rhombohedral stacked ABC and ACB triangular domains, thus forming the kagome-like domain tessellation. As will be discussed later, this is a metastable structure that could be stabilized at very small twist angle. In a moderately small twist angle (θ>0.1∘\theta>0.1^\circ), the twisted monolayer-bilayer prefers the

simpler triangular domain network composed of alternating polar and nonpolar regions, as illustrated in Figure 3, which yields a net polarization in the system. The related unique sliding ferroelectricity in this geometry will be discussed below.

Polarization domains of twisted monolayer-bilayer system

Figure 3. Polarization domains of twisted monolayer-bilayer system with moderate twist angles (θ12>0.1∘\theta_{12}>0.1^\circ and θ23=0\theta_{23}=0). (a) BA-stacked and (c) AB-stacked bilayer on top of monolayer with twist angle θ12\theta_{12}. The twist geometry in (c) can be obtained from a uniform sliding of the top layer in (a) by Δx=a\Delta x=a (aa is the B–N bond length). The corresponding polarization domains are drawn in (b) and (d), respectively. The red, blue and white correspond to upward, downward and zero polarizations, respectively. In the left, middle, and right panels, cross sectional, pair-wise layer-resolved and total polarizations are displayed. See the text for a detailed description.

Sliding ferroelectrics in twisted monolayer-bilayer moiré systems

When a Bernal-stacked BBN is placed on top of a monolayer with a moderate twist angle, a triangular moiré domain forms, similar to twisted bilayer systems. In these types of stackings, the net polarization density is the sum of two interlayer polarizations: one from the twisted monolayer–bilayer interface and the other from the Bernal-stacked bilayer itself, as shown in Figure 3.

Since a lateral translation of one layer by Δx=a\Delta x=a (aa is a nearest-neighbor bond length between B and N atoms) can reverse the polarization direction as shown in Figure 1 (a), even the same moiré patterns may display the opposite polar direction. In the left panels of (b) and (d) in Figure 3, the red and blue arrows denote schematic polarization between the adjacent layers. For BA (AB)-stacked BBN for the top and middle layers, it has a uniform polarization domain (also denoted as uniform red and blue hexagons in the middle panels) while the alternating red and blue triangular domains denote polarization map between the middle and bottom layers with a twist angle of θ12\theta_{12}. In the right panels, by adding up two polarization domains, the final patterns show an interesting switching behavior: the sliding top layer by aa causes the previously polarized domains to be unpolarized and vice versa, as shown in Figure 3 (c) and (Figure 3d).

Due to the polarization reversal by the translation, we expect that an external electric field could potentially switch between these two configurations. However, unlike a previous demonstration of sliding ferroelectric behavior in bilayer studies [34, 35], the TTBN has competing Bernal and rhombohedral stacking orders so that the locally stable moiré domain other than a simple triangular one may be possible.

To quantify the switching mechanism, we calculated the energy barrier between the two polarization states as the top layer slides. Figure 4 (a) shows the barrier profiles for the twisted monolayer-bilayer with θ12=0.082∘\theta_{12} = 0.082^\circ, 0.165∘0.165^\circ and 0.275∘0.275^\circ, compared with bilayer sliding (dashed line). Notably, unlike the maximum energy at Δx=a/2\Delta x = a/2 for sliding BBN, the full atomic relaxation significantly lowers the energy barrier in trilayer cases compared to rigid sliding, resulting in a broad plateau with a shallow minimum at Δx=12a\Delta x = \frac{1}{2}a, as is clearly shown for the case with θ12=0.082∘\theta_{12} = 0.082^\circ. The corresponding energy landscape is schematically illustrated in Figure 4 (b).

Sliding ferroelectricity and energy landscape of twisted monolayer-bilayer hBN

Figure 4. Sliding ferroelectricity and energy landscape of twisted monolayer-bilayer hBN. (a) Energy landscapes as a function of the translation Δx\Delta x (in units of B–N bond length aa) of the top layer in twisted monolayer–bilayer h-BN. Three twist angles are considered. The dashed line represents the sliding energy landscape of the bilayer BN system. For comparison, the energies for each twist angle are vertically shifted such that the zero energy of each colored curve corresponds to its respective intercept with the energy axis. (b) Schematic energy landscape from (a) showing a plateau around Δx=a/2\Delta x = a/2. (c) Polarization switching during the translation and zero net polarization occurs at Δx=a/2\Delta x = a/2. (d) The energy difference ΔE\Delta E between the triangular type (Δx=0\Delta x = 0) and kagome-like domains (Δx=a/2\Delta x = a/2). (e) Layer-resolved interlayer polarization densities as a function of top-layer sliding by Δx\Delta x. P12P_{12} (P23P_{23}) indicates the interlayer polarization density between the bottom and middle (middle and top) layers. Top panels are variation of total polarizations as a function of translation (Δx\Delta x) of the top layer.

Figure 4 (e) displays the layer-by-layer polarization densities along the sliding path, revealing that most of the change arises from P23P_{23} (polarization between the middle and top layers), while P12P_{12} remains relatively constant. When atoms are fixed during sliding, the energy profile should resemble that of bilayer sliding. With relaxation, however, the system can lower its energy by expanding low-energy stacking domains at the cost of in-plane strain. Because stacking energy scales with area (∝L2\propto L^2), while strain energy scales with length (∝L\propto L), at small twist angles, all areas tend to be occupied by pure AB and AC domains. The area ratios of the AB and AC domains in P23P_{23} are approximately Δxa\frac{\Delta x}{a} and 1−Δxa1-\frac{\Delta x}{a}, respectively. At Δx=12a\Delta x=\frac{1}{2}a, P23P_{23} is partitioned into equally spaced AB and AC stripes as shown in Figure 4(e). Deviations from this point introduce energy costs due to weak interwall interactions and result in a very shallow minimum around Δx=12a\Delta x=\frac{1}{2}a. If the two walls are too close (Δx=0.2a\Delta x=0.2a or Δx=0.8a\Delta x=0.8a), the stripes break into small ellipses to reduce the interwall energy, manifesting as small humps in the energy landscape in Figure 4(b).

The net polarization density along the sliding path is shown in Figure 4(c). At Δx=12a\Delta x=\frac{1}{2}a, a kagome-like domain pattern emerges as a local minimum energy configuration and maximizes nonpolar domains. As shown in Figure 4(d), the energy difference ΔE\Delta E between the triangular type and the kagome-like domains decreases with decreasing twist angle, due to the increasing importance of the stacking energy relative to the energy of the domain wall. This suggests that at sufficiently small twist angles (≪0.1∘\ll0.1^\circ), the kagome domain structure could be the true ground state of the twisted monolayer-bilayer BN.

Electric field control of super moiré domains

At small twist angles, the local domain lattice within each moiré supercell forms to maximize the lowest-energy stacking configuration. As a result, domain tessellation strongly depends on the relative energies of competing stackings and can vary if these energies change [33]. Marginally twisted BBNs follow the same principle, with the added feature of polarization domains. In TBBN [25, 26], the most favorable stackings of AB and BA (or equivalently AC) are energetically degenerate. But if we apply the out-of-plane electric field EE, the domains with the parallel polarization to EE become energetically favorable, and the domain walls become warping to increase the area of the domains at the expense of additional energy cost from the increased length of the domain walls. Figure 5 (a) shows the polarization density map of TBBN and their evolutions in the external electric fields for three different twist angles. We note that the side length of the triangular moiré domain is inversely proportional to the twist angle θ\theta. We also notice that the vertices of the triangular domain are all fixed for a given twist angle irrespective of the strength of EE.

Polarization densities and electric susceptibilities of TBBNs under electric fields

Figure 5. (a) Out-of-plane polarization densities of TBBNs with twist angles of θ=0.110∘\theta= 0.110^\circ, 0.165∘0.165^\circ and 0.329∘0.329^\circ with increasing electric fields (EE). The scale bar (white line) is 100 nm. The field direction is (anti)parallel to the polarization direction of BA (AB) stacking (red and blue). (b) Schematic geometry of domain wall under the external electric field. LL represents the length of the triangular domain wall and Δs\Delta s is the increased area of the triangular domain by a warped boundary under EE. The warped boundary (thick black line) is described by y(x)y(x) along the xx-axis denoted by a dotted black line. The warped boundary has an arc shape that can be defined as a segment of a circle with radius RR, subtended by a central angle α\alpha. (c) Net polarization densities for the three cases as a function of EE. (d) Electric susceptibilities χ\chi of the three cases as a function of EE. (e) The computed electric susceptibility of χ\chi as a function of LL for the fully relaxed TBBNs under electric fields using our interatomic potential method.

Figure 5 (c) shows the net polarization density of PP per the moiré cell with increasing EE. As θ\theta decreases, PP increases for a given EE. The electric susceptibilities χ=εr−1\chi= \varepsilon_r - 1 (εr\varepsilon_r: relative permittivity) are plotted in Figure 5 (d). If we neglect the interaction between the walls, χ\chi has a simple relation with the energy of the domain wall. Let us assume that EE makes the energy difference δE\delta_E between the AB and AC domains. For δE≠0\delta_E \ne0, the domain wall becomes warped, and the energy FF of the system using the geometry shown in Figure 5(b) can be written as

F=∫0Ldx EW(y′(x))2+1+∫0Ldx δE y(x),(1)F = \int_0^L dx\, E_W \sqrt{(y'(x))^2 + 1} + \int_0^L dx\, \delta_E\, y(x), \tag*{(1)}

where EWE_W is the energy of the domain wall per unit length. By applying the Euler-Lagrange equation, we obtain the following.

ddx(y′(y′)2+1)=δEEW.(2)\frac{d}{dx}\left(\frac{y'}{\sqrt{(y')^2+1}}\right)=\frac{\delta_E}{E_W}. \tag*{(2)}

Because the left-hand side in Eq. (2) is the curvature of the domain wall, the shape of the domain wall is the arc of a circle with a radius R=EWδER = \frac{E_W}{\delta_E}. The increased area of ΔS\Delta S can be approximated for small EE as

ΔS=R2α2−RL2cos⁡(α2)≃L316R\Delta S = \frac{R^2\alpha}{2}-\frac{RL}{2}\cos\left(\frac{\alpha}{2}\right)\simeq\frac{L^3}{16R}

in the case of R≫LR \gg L. Then, the increased PP upon warping can be written as

P=43pΔSL2≃34δEEWLp=34Lp2EWEP=\frac{4\sqrt{3}p\Delta S}{L^2}\simeq\frac{\sqrt{3}}{4}\frac{\delta_E}{E_W}Lp=\frac{\sqrt{3}}{4}\frac{Lp^2}{E_W}E

where pp is the polarization density per unit area and δE=2pE\delta_E = 2pE. Then, the electric susceptibility can be written as

χ=Pε0E=CχLEW,(3)\chi=\frac{P}{\varepsilon_0 E}=C_\chi\frac{L}{E_W}, \tag*{(3)}

where Cχ=0.0024672C_\chi= 0.0024672 and the units of LL and EWE_W are Å and meV/Å, respectively. We confirm that the linear relationship between χ\chi and LL in Eq. (3) is numerically verified for E=0.08E = 0.08 V/Å using our interatomic potential method as shown in Figure 5 (e) and the estimated energy of the domain wall is 56.5 meV/Å.

Under external electric fields, the moiré domains in TBBN have been shown to deform, with their boundaries becoming distorted, consistent with previous experimental and theoretical studies [25, 51]. However, it is important to note that both the supercell lattice and the positions of the domain vertices are determined solely by the twist angle. Reversing the direction of the applied electric field does not alter these structural features; it merely switches the polarization [34, 35]. Consequently, precise spatial control over the domain symmetries and the associated local quantum states at the domain vertices of twisted bilayer systems are hardly realizable.

Unlike the twisted bilayer cases discussed so far, the trilayer systems show markedly different behaviors under external electric fields. In TTBN under a homogeneous perpendicular electric field, the local domain walls distort to increase the domain area where its polarization is parallel to the field, as was demonstrated in TBBN under the field. In contrast to TBBNs, all the other domains surrounding the expanded domains start to contract with the field and all the other boundaries become shortened as shown in Figure 6. In the case of alternate stacking displayed in Figure 2 (b), the area of equilateral triangular domains corresponding to ACB stackings decreases significantly with increasing field and eventually enters a saturation regime at 0.22 V/Å, as depicted in Figure 6 (a) and (Figure 6c). Its fully saturated side length is approximately 10 nm at E>0.3E > 0.3 V/Å. This saturation in domain size arises from interactions between domain boundaries and vertices, which limit further shrinkage. We note that this highly nonlinear variation of domain area with electric field follows a single, well-defined scaling rule, unifying all calculated results across varying twist angles, as discussed in Appendix § 3.8. We also find that the critical field required to reach this saturation remains well below the dielectric breakdown threshold of trilayer h-BN, larger than 0.8 V/Å [48].

Field-induced evolution of domain structures of TTBNs

Figure 6. The field-induced (E<0E < 0) evolution of domain structures of TTBNs with (a) alternate and (b) helical stacking configurations, corresponding to Figure 2 (b) and (Figure 2d), respectively. The scale bar is 100 nm. (c) and (d) The changes in the distances between boundary vertices that host quantum dot states. The dot positions are indicated in the leftmost panels of (a) and (b) by black dots. In (c), the distance measures the side of equilateral triangles formed by three dots, whereas in (d) the distances vary, as labeled by the two dot indices shown in (b).

In the case of helical stacking, illustrated in Figure 2 (d), a similar trend is observed: The isosceles triangular ACB stacking domains contract rapidly with increasing field and eventually saturate at a shortest side length of approximately 5 nm, as shown in Figure 6 (b) and (Figure 6d). The polarization distributions shown in Figure 12 in Appendix § 3.0.3 confirm this trend, demonstrating that the regions with polarization aligned in the direction of the applied field grow in area, consistent with energetic favorability under the external field.

Variation of polarization distributions of TTBN under electric fields

Figure 12. Variation of polarization distributions of TTBN under electric fields. (a) Changes of polarized domains of alternate stacking TTBN in Figure 6 (a) under a perpendicular electric field of which direction is parallel to the polarization of ABC (blue) stacking. (b) Same figures for helical stacking TTBN in Figure 6 (b).

The metastable kagome-shaped domain pattern of the twisted monolayer-bilayer TTBN exhibits minimal domain wall distortion under applied external electric fields as shown in Figure 13. Energetically favorable domains expand primarily through domain wall motion, and the system undergoes a structural transition into a simple triangular domain configuration even under a weak external field of 0.04 V/Å, as elaborated in Appendix § 3.0.3. Moreover, reversing the direction of the electric field alters only the polarization direction within each triangular domain, without shifting the QD-hosting vertex position itself. Thus, for the twisted monolayer-bilayer case, the electric-field-induced switching introduces a new sliding ferroelectric transition between kagoma-shaped to triangular-shaped polarization domain lattices.

Switching between kagome-shaped and triangular domains under electric fields

Figure 13. Switching between kagome-shaped and triangular domains under electric fields. (a) Schematic description demonstrating the evolution of energy landscapes for θ12=0.082∘\theta_{12} = 0.082^\circ when a small electric field EE is applied. (b) Layer-resolved polarization densities (P12P_{12} for the first and second layer and P23P_{23} for the second and third one) as a function of applied electric field of EE. Top panels are total polarization maps. (c) The critical fields for switching between a kagome domain lattice and a triangular domain lattice. As the moiré length LL decreases, the critical field also decrease and finally the kagome domain becomes unstable for L<500L < 500 Å.

We note that the striking difference between TTBN and TTBN arises from the presence of competing stacking domains within the supermoiré periodicity of trilayer systems. In TTBN, multiple domains with distinct stacking orders coexist within a single superlattice unit cell defined by two twist angles. As a result, the domain boundaries are not rigidly fixed within the supercell, allowing external electric fields to modify the local arrangement of the supermoiré domain tessellation and the positions of the associated vertices.

Realization of quantum dot arrays

As we mentioned earlier, several twisted bilayer systems including TTBN can host arrays of localized states at the center or vertices of their moiré lattice potentials [19, 20, 21, 22, 23, 24, 25, 26, 27]. For marginal TTBNs, we also expect similar but distinct localized states. In Figure 7 (a), total density of states (TDOS) of the alternate stacking case in Figure 2 (b) is drawn. The local density of states (LDOS) for two distinguishable vertices [1 and 2 in Figure 7 (b), which will be referred to as Dot 1 and Dot 2] are also displayed. Below the energy of the conduction band edge (ECB_{\rm CB}) and above the valence band edge (EVB_{\rm VB}), several discrete energy levels from localized states are identified. Here we define ECB_{\rm CB} and EVB_{\rm VB} as the corresponding band extrema obtained at the centers of triangular domains.

Localized quantum harmonic oscillator states in TTBN

Figure 7. Localized quantum harmonic oscillator states in TTBN. (a) Electronic structure of alternate stacking geometry with θ12=−0.082∘\theta_{12}=-0.082^\circ and θ23=0.164∘\theta_{23}=0.164^\circ shown in Figure 2 (b). The total density of states (denoted as “Total”) and local density of states (LDOS) for rectangular sites 1 and 2 in (b) (denoted as “Dot 1” and “Dot2”) are shown for the conduction bands and valence bands (top and bottom panels). The energy level is a relative value to the conduction and valence band edges (ECBE_{\mathrm{CB}} and EVBE_{\mathrm{VB}}) of the bulk trilayer stackings at the center of triangular domains. (b) The fully relaxed interlayer spacing between the bottom and top layers drawn for the same region as Figure 2 (b). (c) and (d) The enlarged interlayer spacings of Dot 1 and Dot 2 in (b) are shown, respectively. (e) and (f) The corresponding potential wells of conduction band energy. (g) and (h) The spatial distributions of the lowest eigenstates ∣ψ∣2|\psi|^2 at Dot 1 and the second lowest state at Dot 2, respectively. The colormap range is normalized to the maximum value of each panel. The dashed lines are visual guides for the polar domain boundaries. (i) The enlarged spatial densities of the first three lowest states for Dot 1, of which energies are indicated by small black triangles in (a). The upper panels are from our TB calculations, denoted as TB and the lower ones from 2D QHO potentials denoted as QHO with corresponding quantum numbers of nxn_x and nyn_y. (j) Same as (i) for Dot 2 with the six lowest states of which energies are also marked by the triangles in (a).

We found that the origin of confinements for these discrete states is the spatial variation of band energy around vertices depending on their stacking orders. In Figure 7 (b), the spatial variation of interlayer distances between top and bottom layers are drawn. Near Dot 1, its enlarged view in Figure 7 (c) shows threefold rotational symmetric variations circling around the isotropic maximum area. Near Dot 2, the corresponding distance map shows twofold symmetry around the elongated maximum region as shown in Figure 7 (d). It turns out that each area for interlayer distance maximum corresponds to unstable AAA stacking and mixture of AA and SP stackings for Dots 1 and 2, respectively. The symmetry differences in Figure 7 (c) and (Figure 7d) arise from the surrounding stackings. Dot 1 is encircled by nonpolar ACA and ABA stackings with three-fold symmetry, while Dot 2 is surrounded by non-polar ACA and ABA stacking together with polar ACB and ABC stackings that break this symmetry as shown in Figure 7 (b). Then, we plot the energy of conduction band minima from energy bands for corresponding local stackings in Figure 7 (e) and (Figure 7f). It is noticeable that both energy maps follow the distance profiles in an opposite way, and show deep isotropic and anisotropic potentials for Dot 1 and Dot 2, respectively. These smooth potentials prompt us to consider two-dimensional (2D) QHO potential, Vdot(x,y)=12mxωx2x2+12myωy2y2V_{\mathrm{dot}}(x,y)=\frac{1}{2}m_x\omega_x^2x^2+\frac{1}{2}m_y\omega_y^2y^2, where mxm_x and mym_y are effective masses and ωx\omega_x and ωy\omega_y are angular frequencies. The four parameters are fit to generate the low energy landscapes of each dot. The resulting parameters for the QHO potentials for Dot 1 and Dot 2 are summarized in Table I. We note that for Dot 1, the fitted parameters satisfy mx=mym_x=m_y and ωx=ωy\omega_x=\omega_y, while for Dot 2, anisotropy is necessary. In Figure 7 (g) and (h), we present the wavefunctions of the lowest eigenstate localized at Dot 1 and the second-lowest one at Dot 2, respectively, to illustrate their spatial distributions. The eigenstate associated with Dot 1 is localized at the centers of the nonpolar hexagons within the kagome-shaped polar domain lattice, whereas the state associated with Dot 2 is at the vertices of these hexagons. To further investigate the characteristics of the localized states, we plot the wavefunctions of the three lowest eigenstates for Dot 1 and the six lowest eigenstates for Dot 2 in Figure 7 (i) and (j), respectively. Simultaneously, we compute the corresponding eigenstates of the effective 2D Schrödinger equation with QHO potential using parameters listed in Table I, and we compare these with the results obtained from full-scale TB simulations involving multimillion atoms. The 2D QHO model yields eigenenergies of EDot1=ℏω(nx+ny+1)E_{\mathrm{Dot1}}=\hbar\omega(n_x+n_y+1) for Dot 1 and

EDot2=ℏωx(nx+12)+ℏωy(ny+12)E_{\mathrm{Dot2}}=\hbar\omega_x\left(n_x+\frac{1}{2}\right)+\hbar\omega_y\left(n_y+\frac{1}{2}\right)

for Dot 2 where nx,ny=0,1,2…n_x,n_y=0,1,2\ldots. The eigenstates from our TB calculations closely match those of the 2D quantum harmonic oscillator (QHO) in both spatial wavefunction profiles and energy level structures, characterized by quantum numbers of nxn_x and nyn_y. Detailed energy comparisons with the 2D QHO model are summarized in Table II. Minor deviations appear in the second and third eigenstates of Dot 1, and the fourth and fifth of Dot 2, due to boundary-induced perturbations.

For the helical stacking configuration [Figure 2 (d)] and the twisted monolayer–bilayer system [Figure 2 (f)], electronic structures (TDOS and LDOS), spatial maps of interlayer distances, corresponding potential landscapes, and 2D QHO type eigenstate wavefunctions are shown in Figure 15 and Figure 16 in Appendix § 3.0.5, respectively. Associated parameters for these QHO potentials and energies of QD states are also summarized in Tables V and VI in Appendix E. Furthermore, the QD states and their energy level spacings are robust under a moderate inhomogeneity such as twist angle and strain disorders (see Appendix § 3.9). These results confirm that TTBN with varying stacking geometries can host robust arrays of localized 2D QHO states.

QD states in the helical TTBN and their electrical reconfiguration

Figure 15. QD states in the helical TTBN (θ12=θ23=θ=0.082∘\theta_{12}=\theta_{23}=\theta=0.082^\circ) and their electrical reconfiguration. (a) and (b) The total density of states (Total) and local density of states (LDOS) of conduction bands for (a) ∣E∣=0|E|=0 and (d) ∣E∣=0.1|E|=0.1 V/Å, where the projections are for sites 1, 1′, 2, and 2′ in (c) and (d), respectively. The energy level is a relative value to the conduction edges ECBE_{\mathrm{CB}} of the bulk trilayer stackings. (c) and (d) The fully relaxed interlayer spacing between the bottom and top layers drawn for the same region as Figure 2(d) and the first panel of Figure 6(b), respectively. (e) The density distribution of the lowest QD states at sites 2 and 2′ shown for the same region as (c), of which energies are indicated by small black triangles in (a) Dot 2 LDOS. (f) Similar data but under the electric field ∣E∣=0.1|E|=0.1 V/Å, for the same region as (d) and energy peaks indicated in (b) Dot 2 LDOS. (g) The enlarged interlayer spacing around the site 1 in (c). (h) The potential well of the conduction band energy landscape at QD site 1 shown in (g). (i) The enlarged density distributions of the first three lowest QD site 1 states, of which energies are indicated in (a) Dot 1 LDOS. For each panel, corresponding quantum numbers of isotropic 2D QHO states nx+nyn_x+n_y are displayed.

QD states in the twisted monolayer-bilayer TTBN and their electrical reconfiguration

Figure 16. QD states in the twisted monolayer-bilayer TTBN (θ=0.082∘\theta= 0.082^\circ) and their electrical reconfiguration. (a) and (b) The total density of states (Total) and local density of states (LDOS) of conduction bands for (a) ∣E∣=0|E| = 0 and (d) ∣E∣=0.03|E| = 0.03 V/Å, where the projections are for sites 1, 2, and 3 indicated in (c) and corresponding sites in (d), respectively. The energy level is a relative value to the conduction band edges ECBE_{\mathrm{CB}} of the bulk trilayer stackings. (c) and (d) The fully relaxed interlayer spacing between the bottom and top layers drawn for the same region as Figure 2 (f) and equivalent area under the field ∣E∣=0.03|E| = 0.03 V/Å, respectively. (e) The enlarged density distribution of the lowest QD states at sites 2 and 3, the energies of which are indicated by small black triangles in (a) Dot 2 LDOS. (f) For the square area marked in (d), Gaussian-weighted spatial distributions of eigenstates under the electric field ∣E∣=0.03|E| = 0.03 V/Å corresponding to range 1 (R1) and range 2 (R2) indicated by shadings in (b). The shadings show the ±σ\pm\sigma (broadening factor, 1 meV) range around the Gaussian-weight center. (g) The enlarged interlayer spacing around the site 1 in (c). (h) The potential well of the conduction band energy landscape at QD site 1 shown in (g). (i) The enlarged density distributions of the first six lowest QD states at site 1, of which energies are indicated in (a) Dot 1 LDOS. For each panel, corresponding quantum numbers of 2D QHO states (nx,ny)(n_x,n_y) are displayed.

mxm_xmym_yℏωx\hbar\omega_xℏωy\hbar\omega_yE0E_0
Dot 1 (CB)0.4510.451103103−471
Dot 1 (VB)−0.616−0.6166666222
Dot 2 (CB)0.4430.5235492−339
Dot 2 (VB)−0.590−0.6912965169

Table 1. The parameters of 2D QHO potentials for Dot 1 and Dot 2 in Fig. 7. CB (VB) stands for QHO states below (above) the conduction (valence) band edge. The unit for mxm_x and mym_y is the electron mass mem_e. E0E_0 is the zero level of the harmonic potential with respect to the bulk reference level. The unit of ℏωx\hbar\omega_x, ℏωy\hbar\omega_y and E0E_0 is meV.

In addition to the regular arrays of deep QD states discussed so far, we note the distinct one-dimensional (1D) features in the DOS between the discrete QD levels and the bulk band edges, as shown in Figure 7 (a). These 1D-like DOS features are observed in both the alternate and helical stacking configurations and are attributed to boundary states localized at the interfaces between distinct stacking domains.

To elucidate the nature of these boundary states, the real-space distributions of the characteristic states are obtained for all three stacking geometries (see Appendix § 3.0.4 and Figure 14 for details). For the alternated stacking, 1D boundary states are found both between two nonpolar domains and at the interfaces between nonpolar and polar domains. In the helical stacking case, the three lowest-energy boundary states encircle the hexagonal domains of Bernal stacking inside the hexagrams, forming a regular array of closed loops. In contrast to the well-defined threefold rotational symmetry exhibited by the boundary states in the alternate and helical stacking cases, the boundary modes in the twisted monolayer-bilayer configuration [Figure 14 (f)] exhibit nematic order, breaking the rotational symmetry. Importantly, unlike the boundary states in TTG, which hybridize with bulk metallic states [33], the 1D boundary states in TTBNs are energetically well separated from the bulk states.

Domain boundary states of TTBNs

Figure 14. Domain boundary states of TTBNs. Shown are the total density of states (DOS; upper panels) and boundary-projected local density of states (LDOS; lower panels) for (a) alternate stacking, (b) helical stacking, and (c) twisted mono–bilayer TTBNs. In (a) and (c), “NP–NP” and “NP–P” denote LDOS at boundaries between nonpolar domains, and between nonpolar and polar domains, respectively. In (b), “3–6” and “3–4” refer to boundaries between triangular and hexagonal, and triangular and tetragonal domains. Panels (d)–(f) display Gaussian-weighted spatial distributions of eigenstates for the corresponding stackings, centered at the LDOS peak energies. Each set of four panels is arranged in order of increasing energy.

(nx,ny)(n_x,n_y)(0,0)(0,0)(1,0)(1,0)(0,1)(0,1)(2,0)(2,0)(1,1)(1,1)(0,2)(0,2)
Dot 1 (CB) TB−378−284−284−204−204−204
Dot 1 (CB) QHO−368−265−265−162−162−162
Dot 1 (VB) TB163111111---
Dot 1 (VB) QHO1558989---
Dot 2 (CB) TB−270−219−190−173−150−123
Dot 2 (CB) QHO−266−212−174−158−120−81
Dot 2 (VB) TB125102----
Dot 2 (VB) QHO12193----

Table 2. Energies of QD states with quantum numbers (nx,ny)(n_x,n_y) using TB approximation and 2D QHO models. Relative values to the conduction band (CB) edge E−ECBE-E_{\mathrm{CB}} and to the valence band (VB) edge E−EVBE-E_{\mathrm{VB}} in meV.

Here, we highlight both the similarities and differences in the nature of quantum dot (QD) states in TTBN by comparison with other moiré systems. In twisted bilayer graphene (TBG), the electronic density becomes strongly localized near AA-stacking regions, primarily due to moiré-scale interference of interlayer tunneling processes rather than confinement by a real-space potential well. In this case, the relevant energy scale is set by the separation between the flat bands and the remote bands, typically on the order of 30–60 meV, with a corresponding AA-centered localization length of 5–10 nm [16, 52]. In contrast, in marginally twisted TMDC heterobilayers, localization arises from hydrostatic strain “hot spots” that form at the vertices of the reconstructed domain network. These strain fields generate confinement potentials of approximately 100 meV, leading to QD bound states with characteristic sizes of 2 nm [24, 53]. For the TTBN structures studied here, the localization mechanism is instead governed by stacking-dependent electrostatic potentials that emerge at the vertices of the super-moiré domain network. Our calculations show that these potentials give rise to discrete bound QD states with substantially deeper confinement energies, ranging from 200 to 400 meV, and characteristic localization lengths of about 1–2 nm. Overall, this comparison demonstrates that although vertex-localized states are a common feature across several moiré platforms, their microscopic origin and confinement strength differ significantly among TBG, TMDC heterobilayers, and TTBN.

Reconfigurable quantum dot arrays

Having demonstrated that local domain shapes undergo significant changes under applied electric fields, we now explore the implications for the electronic structure of the localized states and their mutual interactions. In Figure 8 (a), we show the local density of states (LDOS) for valence-band states at Dot 1 and Dot 2 in the alternate stacking configuration previously discussed in Figure 7, as a function of the electric field. As the field increases, the confinement potential of each dot becomes shallower, and the energies of the lowest hole states approach the valence band edge of the bulk. Concurrently, the energy difference between the lowest hole states in Dot 1 and Dot 2 decreases and becomes nearly degenerate at a field strength of 0.3 V/Å. This behavior is accompanied by a field-induced reduction in the band gap due to the Stark effect [48], which causes higher-energy localized hole states and boundary states to hybridize with bulk states, leaving only the lowest-energy localized hole states well isolated as demonstrated in Figure 8 (a). As shown in Figure 6 (a) and (Figure 6c), the interdot distance among the three Dot 2 sites—located at the vertices of the equilateral ACB-stacking triangles—decreases notably with increasing field. As a result, the spatial distribution of the localized hole states reveals a clustering effect, where three localized states move closer to each other to form equilateral triangles centered at the mass center of a larger triangle composed of Dot 1 sites as shown in Figure 8(b). Simultaneously, interaction between localized states at Dot 1 and those at Dot 2 will be completely quenched with increasing fields.

Spatial and energetic variations of quantum dot eigenstates with electric fields

Figure 8. Spatial and energetic variations of quantum dot eigenstates with electric fields. (a) Variations of the local density of states (LDOS) at Dot 1 and Dot 2 of the alternate stacking case shown in Figure 7(b) in the presence of the external electric field. From the left to the right panels, the electric field increases from 0.1 to 0.4 V/A˚\mathring{\mathrm{A}} and corresponding relaxed structures are shown in Figure 6(a). The valence band extrema from the center of domains for each field are set to zero. (b) Integrated spatial distribution of states above the bulk states energy of 0.0 eV under the electric field ∣E∣=0.4|E| = 0.4 V/A˚\mathring{\mathrm{A}}. The modulus of three-fold degenerate wavefunctions for three Dot 2’s forming equilateral triangles [their energies are denoted by red arrows in (a)], ∣ψi∣2|\psi_i|^2 (i=1,2,3i = 1, 2, 3) with the applied electric field, (c) ∣E∣=0.2|E| = 0.2 V/A˚\mathring{\mathrm{A}}, (d) 0.3 V/A˚\mathring{\mathrm{A}}, and (e) 0.4 V/A˚\mathring{\mathrm{A}}. The colormap range is normalized to the maximum value in each panel.

As the interdot spacing between Dot 2 sites decreases, their localized wavefunctions begin to overlap, leading to quantum hybridization. At each Dot 2 site, the lowest energy localized hole states, indicated by the red arrow in the middle panel of Figure 8(a), are spin- and valley-degenerate. For each dot site, we can assign the ground QHO state of ϕi\phi_i with nx=ny=0n_x = n_y = 0 for three sites (i=1,2,3i = 1, 2, 3), respectively. Up to E=0.2E = 0.2 V/A˚\mathring{\mathrm{A}}, the states at the vertices of the triangle are well-localized, so that each dot wavefunction can be written as ψD2i=ϕi\psi_{\mathrm{D2}}^i = \phi_i (i=1,2,3i = 1, 2, 3) as shown in Figure 8(c). So, in this regime, we expect that the coupling between the triangular QDs can be well described by the capacitance model [38, 54]. However, when the field increases above 0.3 V/A˚\mathring{\mathrm{A}}, the wavefunctions starts to hybridize, forming linear combinations of the localized states at each site, i.e., ψD2i=c1iϕ1+c2iϕ2+c3iϕ3\psi_{\mathrm{D2}}^i = c_1^i\phi_1 + c_2^i\phi_2 + c_3^i\phi_3 where coefficients of cjic_j^i (i,j=1,2,3i, j = 1, 2, 3) are set by detailed interactions between QDs. We found that the valley and spin-degenerate ground states are slightly split by about 0.1 meV while maintaining their coupling through tunnelings. For example, at the upper-left vertex of the triangle shown in the left panel of Figure 8(d), the mixing ratios are ∣c21/c11∣2=0.24|c_2^1/c_1^1|^2 = 0.24 and ∣c31/c11∣2=0.12|c_3^1/c_1^1|^2 = 0.12. Similar mixing behavior is observed at the other two Dot 2 sites in the middle and right panels. As shown in Figure 8(e), with a further increase of the field, the hybridization becomes stronger as can be inferred from increasing mixing ratio of ∣c21/c11∣2=0.33|c_2^1/c_1^1|^2 = 0.33, and ∣c31/c11∣2=0.22|c_3^1/c_1^1|^2 = 0.22 for the same QDs. This phenomenon is turned on at the critical field E=0.22E = 0.22 V/Å, where the corresponding ratio is finite but marginal (<0.01< 0.01).

Discussions and conclusions

All the calculations hitherto confirm that the external electric field enables control over the spatial arrangement of QDs, resulting in clustering within certain regions while causing others to move apart. Under the field, the isolated localized states without the field begin to hybridize across neighboring QDs. When they form clusters, the systems we present exhibit other intriguing local features. The equilateral triangular configuration of three QDs shown in Figure 8 suggests potential applications in charge- and spin-based interactions, including spin qubits [55], due to several fascinating aspects such as spin or charge frustrations, topological properties and Nagaoka ferromagnetic molecular states, to name a few [38, 56, 57, 58]. To achieve these conditions, one must address the unavoidable fluctuations arising from the nuclear spins. Unlike silicon and germanium, which possess zero nuclear spins, B and N atoms cannot completely nullify their nuclear spins [47]. However, recent advances in isotope engineering [59] and dynamical decoupling techniques [60] have demonstrated significant improvements in the coherence of spin defects in bulk h-BN materials. So, if properly prepared, the aforementioned intriguing charge- and spin-related physics can be realized in TTBNs with added features such as dynamic repositionings of QDs as discussed so far.

We can envisage further a possible long-ranged information shuttling in solid state systems. For the alternate stacking geometry, if the applied electric field is first turned on along the polarization direction of ABC stacking, the domain area of ABC stacking increases while all the other domains shrink as already shown in Figure 6 (a). Then, if the field is reversed in direction, the enlarged ABC stacking area will return to its original shape and subsequently contracts, while the contracted ACB stacking area begins to expand. This can be expected well considering the upside down configuration of Figure 6 (a). During the process, field-induced coupled states in a triangular QD cluster first move apart as the field is reduced to zero. Then, upon reversing the field direction, these states reassemble into new triangular clusters composed of QDs from different initial clusters, facilitating long-ranged transport of quantum states encoded in a coupled QD state across the array (see details in Figure 9 as was recently demonstrated in a Rydberg atoms array [3].

Schematic redrawing of localized quantum harmonic oscillator states and reconfigured quantum dot clusters

Figure 9. (a) Schematic redrawing of the localized quantum harmonic oscillator (QHO) states shown in Figure 7 associated with individual sites i=1,…,9i = 1,\ldots,9 without electric field. (b) With an electric field (E<0E < 0 along polarization of ABC stacking), the reconfigured QD states form clusters denoted as c1, c2 and c3. Gray arrows indicate movement direction of each QD and empty dotted circles indicate their original positions. (c) With the opposite directional field (E>0E > 0), QDs form different clusters and one of them is denoted as c4.

Specifically, as illustrated in Figure 9, the localized quantum harmonic oscillator (QHO) states, φi\varphi_i, can be associated with individual sites i=1,…,9i = 1,\ldots,9 without electric field as denoted in Figure 9 (a). Upon application of an external electric field along the polarization direction of ABC stacking, the QD states corresponding to sites 1, 2 and 3 undergo spatial reorganization, forming a cluster of c1 denoted by red in Figure 9 (b). Concurrently, the states at sites 4, 5, 6 and 7, 8, 9 aggregate into distinct clusters, c2 and c3, respectively. Each of these cluster states, φci\varphi_{c_i} (i=1,2,3i = 1,2,3), can realize either capacitive coupled or tunnel-coupled states. If the field is then adiabatically reduced to zero, the cluster states are now redistributed at sites 1, 6, and 8 in Figure 9 (a), respectively. Reversing the field to E>0E > 0 in Figure 9 (c) induces a second reconfiguration, in which these three states converge to form a new coupled state at the cluster c4. Backtracking the transformations from the right to left panels (changing electric field from positive to negative directions again), we can then return to the original situation. If coherence is maintained within local clusterings and regroupings, it becomes possible to observe the encoded coupled states that connect distant regions, spanning more than a few hundred nanometers, despite each individual dot moving back and forth locally.

We remark that the electric reconfiguration of QDs and their potential for the delivery of quantum coupling is not limited to the alternated stack TTBN. The helical stacking (hexagram domain pattern) and the twisted monolayer–bilayer stacking (kagome) TTBNs also host QD states at lower symmetric vertices that can realize the electric reconfigurations with different features and domain network topologies. For example, in the helical stacking TTBN, a honeycomb array of QDs exhibits alternating pairings of QDs, an initial hexagonal unit into three pairs, under an electric field as shown in Figure 15 in Appendix E. In the kagome TTBN, alternating pairings in a linear array of QDs that also approach the 1D boundary states are expected, as shown in Figure 16 in Appendix E.

Lastly, as highlighted in previous studies on semiconducting twisted bilayer systems [15, 24], we also consider the significant potential that QD arrays offer for photonic applications, including single-photon emission (SPE) [61, 62, 63]. h-BN is well established for such applications and demonstrates efficient deep ultraviolet emission and room-temperature SPE capabilities [61, 64]. Given the excellent optical properties associated with defects in monolayer and bulk h-BN, our proposed system having arrays of localized QDs could play an important role in the realization of movable and programmable SPE arrays. Many of the fundamental characteristics in h-BN and its twisted forms such as stacking geometries, domain polarization, and sliding ferroelectricity are also prevalent in transition metal dichalcogenides (TMDCs) [14, 15]. So, we anticipate that certain theoretical predictions derived from our study may also be applicable to marginally twisted trilayer TMDCs.

In summary, we have shown that TTBN exhibits structurally and electronically tunable moiré patterns that arise from the interplay of multiple stacking and twisting degrees of freedom. These features distinguish TTBN from bilayer systems, where the domain structure is solely determined by a single twist angle. In TTBN, we identify regular arrays of localized quantum dot states at the vertices of super moiré domains, which can host well-defined localized quantum states with various spatial symmetries and nonzero angular momentum. The spatial arrangement and coupling strength of these states can be modulated by external electric fields, allowing for transitions between isolated and interacting regimes. This electric-field tunability, together with the well-defined domain structure and compatibility with large-scale fabrication, suggests that TTBN provides a versatile platform for investigating controllable quantum states and designing moiré-based quantum electronic devices.

Acknowledgments

We gratefully acknowledge fruitful discussions with Mikito Koshino, Yun-Pil Shim and Gil Young Cho. K.Y. was supported by Korea Institute for Advanced Study (KIAS) individual grants (CG092501 and CG092502). C.P. was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT and MOE) (No. RS-2025-16063688), and the Basic Science Research Program (No. RS-2024-00349573) through NRF, funded by the Ministry of Science and ICT of the Republic of Korea. Y-W.S. was supported by the KIAS individual Grant No. (CG031509). Computations were supported by Center for Advanced Computation of KIAS.

Data availability

All data is available in the manuscript.

Atomic Structure Optimizations

We employed a newly developed interatomic potential method [33, 41] to compute atomic forces and obtain fully relaxed TTBN structures. This method is specifically designed to accurately replicate ab initio calculation results for large-scale systems, and effectively captures the delicate structural phase transitions in various materials. The harmonic intralayer potential is extracted from the phonon dispersion of monolayer boron nitride, which is obtained from our density functional theory (DFT) calculation with a generalized gradient approximation (GGA) [65] using density functional perturbation theory (DFPT) [42]. In Figure 10 (a), phonon dispersions obtained by DFPT (dots) and our interatomic potential (lines), respectively, are compared.

Interatomic potential of stacked hBN systems

Figure 10. Interatomic potential of stacked hBN systems. (a) Phonon dispersions of monolayer h-BN from first-principles calculations based on density function perturbation theory (dots) and our interatomic potential (lines). (b) Sliding energetics of parallel-stacked bilayer h-BN calculated from DFT-SCAN-rVV10 (dots) and our interatomic potential methods (lines) as a function of relative positions between two layers. Red (blue) denotes configurations in which the interlayer distance hh is fixed at 3.3 Å (optimized individually) for each sliding configuration. (c) PP calculated using DFT-SCAN-rVV10 (dots) and our fitting function, Eq. (5) (lines) with slidings. (d) Variations of interlayer distance hh for AB and 1/2(AA+AB) stacking (see the text for definition). Dots from the first-principles and lines from our method.

For interlayer interactions, we incorporate additional pairwise forces by modifying the Kolmogorov-Crespi (KC) potential [33, 43] to account for long-range interlayer forces across all three layers [33]. The original KC potential is given by the following form [43] [66]:

Vij=e−λ(rij−z0)[C+2f(rij,∥)]−A(rijz0)−6,(4)V_{ij}=e^{-\lambda(r_{ij}-z_0)}[C+2f(r_{ij,\parallel})]-A\left(\frac{r_{ij}}{z_0}\right)^{-6}, \tag*{(4)}

where rijr_{ij} and rij,∥r_{ij,\parallel} are the distance and in-plane distances between two atoms ii and jj, respectively, and f(r)=e−(r/δ)2∑n=02C2n(r/δ)2nf(r)=e^{-(r/\delta)^2}\sum_{n=0}^{2}C_{2n}(r/\delta)^{2n}. The potential is defined for all pairs of atomic species, and each pairwise potential is constructed with eight parameters as presented in Table III. The parameters are fitted to reproduce the sliding energetics of bilayer BN calculated from DFT using SCAN+rVV10 for exchange-correlation and van der Waals functionals [67]. For sliding configurations from AA to AB and to AC parallel stackings, the interlayer distance hh is either fixed to 3.3 Å (red) or optimized (blue) as shown in Figure 10 (b). Although the KC potential is quite accurate for bilayer systems [43], it cannot distinguish the energy of Bernal and rhombohedral stacking which is proved to be crucial in the small-angle (<0.1∘<0.1^\circ) TTG [33]. We also confirm this in multilayer h-BN. To correct this error, we added second nearest neighboring interlayer interactions to the original KC potential as was done for TTG [33]. A simple Gaussian-type potential E2nn=−Ve−3r∥2E_{\mathrm{2nn}}=-Ve^{-3r_\parallel^2} where r∥r_\parallel is the layer-projected distance between the atoms in the first and third layers and V=−0.4V=-0.4 meV is used to make the energy difference between the ABA and ABC stacking h-BN to be −0.42-0.42 meV/atom, which is similar to the reported ab initio calculations [50, 68].

z0z_0 (Å)C0C_0 (meV)C2C_2 (meV)C4C_4 (meV)CC (meV)δ\delta (Å)λ\lambda (Å−1^{-1})AA (meV)
B–B3.445115.116912.19524.94472.03580.91624.421912.8737
B–N3.002414.864712.21374.94742.25051.05026.569213.4167
N–N3.026015.245512.23184.93332.50010.42824.948412.0878

Table III. Parameters for interlayer Kolmogorov-Crespi potential of multilayer boron nitrides.

Between the two layers of h-BN, there can be finite out-of-plane electric polarization of PzP_z, which determines the response to the external electric field. For example, for AC (or BA) stacking, Pz>0P_z>0 develops. For 180∘180^\circ-rotated bilayer, by symmetry, Pz=0P_z=0. PzP_z between two layers are described by our pairwise dipole model,

Pz(rBN)=P0zre−λ(r−z0)e−kr∥2,(5)P_z(\mathbf{r}_{\mathrm{BN}})=P_0\frac{z}{r}e^{-\lambda(r-z_0)}e^{-kr_\parallel^2}, \tag*{(5)}

where rBN\mathbf{r}_{\mathrm{BN}} is a vector from a B atom to a N atom between two neighboring layers, and rr, r∥r_\parallel and zz are ∣rBN∣|\mathbf{r}_{\mathrm{BN}}|, in-plane and zz-component of rBN\mathbf{r}_{\mathrm{BN}}, respectively. P0P_0, λ\lambda, z0z_0 and kk are fitting parameters. We found that the screening-like term e−λ(r−z0)e^{-\lambda(r-z_0)} alone cannot accurately reproduce the reference calculation results. So, the Gaussian overlap-type fitting function e−kr∥2e^{-kr_\parallel^2} is additionally introduced. The total polarization of the system is P=∑{rBN}Pz(rBN)P=\sum_{\{\mathbf{r}_{\mathrm{BN}}\}}P_z(\mathbf{r}_{\mathrm{BN}}), where {rBN}\{\mathbf{r}_{\mathrm{BN}}\} is the set of interlayer B-N vectors. For our SCAN+rVV10 calculations for BBN, h=3.3h=3.3 Å and

Pz=4.50×10−3 eA˚/AunitP_z=4.50\times10^{-3}\ \mathrm{e\AA}/A_{\mathrm{unit}} where Aunit=5.43A_{\mathrm{unit}}=5.43 Å2^2 is the unit area. The four parameters are fitted to reproduce PzP_z of sliding configurations in Figure 10 (c) and additionally the hh-dependence of PzP_z for AB stacking and halfway between AA and AB stacking denoted as 1/2(AA+AB)1/2(\mathrm{AA}+\mathrm{AB}). The parameters P0=1.61×10−2 eA˚P_0=1.61\times10^{-2}\ \mathrm{e\AA}, λ=1.94\lambda=1.94 Å−1^{-1}, z0=3.30z_0=3.30 Å, and k=2.17×10−1k=2.17\times10^{-1} Å−2^{-2} reproduce ab initio results for both configurations as shown in Figure 10 (c) and (d).

Electronic Structure Calculation Methods

To calculate electronic structures of TTBNs with multimillion atoms, we developed an efficient calculation method with tight-binding (TB) approximations. Independent DFT calculations were performed as references of the electronic structures of the few-layer h-BN. We adopted the self-consistent extended Hubbard corrected DFT method (DFT+U+V+U+V) [44, 45, 46, 69] implemented in our in-house version [70] of the Quantum ESPRESSO suite [71, 72]. DFT+U+V+U+V method can efficiently reproduce the electronic band gap in an accuracy comparable to the GWGW approximations without serious computational resources [44, 45, 46]. We calculated the self-consistent on-site (UU) and intersite (VV) Hubbard interactions for the monolayer h-BN and used them for the bilayer and trilayer systems. We checked the self-consistently determined UU and VV values in the different bi- and trilayer systems and confirmed no significant deviation from the monolayer case. Obtained interactions are UBp=0.5112U_{\mathrm{Bp}}=0.5112 eV and UNp=4.9670U_{\mathrm{Np}}=4.9670 eV, where the subscripts refer to the atom and orbital species, e.g., Bpp indicate pp orbitals of B atom, and intersite parameters are VBs−Ns=1.0198V_{\mathrm{Bs-Ns}}=1.0198 eV, VBs−Np=1.9254V_{\mathrm{Bs-Np}}=1.9254 eV, VBp−Ns=1.4700V_{\mathrm{Bp-Ns}}=1.4700 eV, and VBp−Np=2.9767V_{\mathrm{Bp-Np}}=2.9767 eV. The energy cutoffs are 100 and 1000 Ry for wavefunction and charge density, respectively. The kk-points grid is Γ\Gamma-centered 24×24×124\times24\times1. The lattice parameter is 2.504 Å, the interlayer spacing c0=3.253c_0=3.253 Å and the zz-direction length of the unit cell including the vacuum 30 Å. We also considered the dipole correction to the electrostatic potential. Finally, we obtained Wannier function constructed from low energy states by using Wannier90 code [49]. We adopted pzp_z orbitals for each B and N atom as initial projectors.

We constructed a TB model to investigate the electronic structure of the trilayer h-BN, which adopted the pzp_z orbital at each atom as the basis. The hopping parameters are parametrized by fitting to the Wannier Hamiltonian obtained from the DFT+UU+VV calculations. The on-site orbital levels are given as 2.625 eV for B and −3.495 eV for N. Since the deformation of each layer is smooth and moderately small, intralayer bond lengths also only slightly deviate from their unstrained references. So, for the intralayer hopping parameters, the Wannier Hamiltonian components of the corresponding bonds are adopted similarly to a previous study [25, 73]. Specifically, we considered only the intralayer bonds within the cutoff of 5.5 Å, which correspond to up to the fourth nearest B-N bond and the third nearest of each B-B and N-N bond. To incorporate the strain effect, the intralayer hopping parameters vary linearly with respect to the bond length change from the reference structure, i.e., t(r)=t0+s(r−d0)t(r) = t_0 + s(r - d_0), where rr is the bond length, t0t_0 and d0d_0 are the reference hopping parameter and bond length of the unstrained system, respectively, and ss is the gradient of the hopping parameter. The gradient is determined from the Wannier Hamiltonians of the ±2% strained systems. The detailed values of d0d_0, t0t_0, and ss for each type of bond are listed in Table IV.

Bond typed0d_0 (Å)t0t_0 (eV)ss (eV/Å)Vσ0V^0_{\sigma} (eV)βσ\beta_{\sigma}Vπ0V^0_{\pi} (eV)kπk_{\pi} (Å−1^{-1})dπd_{\pi} (Å)
B–N 1st1.4457−2.69095.10670.32334.4827−0.49531.55334.3920
B–B 1st2.5040−0.24491.57940.40864.3051−1.22461.52544.0241
N–N 1st2.5040−0.2138−0.62850.38023.2378−0.41452.24344.0699
B–N 2nd2.89140.03210.1510See text for 2nd (NNL interaction)
B–B 2nd4.3371−0.11700.0307
N–N 2nd4.3371−0.05830.1026
B–N 3rd3.82490.2500−0.2743
B–B 3rd5.0080−0.1019−0.0084
N–N 3rd5.0080−0.07980.1464
B–N 4th5.21250.0309−0.0393

Table IV. Parameters used in our tight binding models. “B-N 1st” indicates the nearest neighboring B and N atoms on intralayer or B and N atoms between the interfacing two layers (interlayer). Others follow the same rule.

For our TB model to reproduce all these reference data, the interlayer hopping is given by Slater-Koster type parametrization with decaying functions of distance, t(r)=n2Vσ(r)+(1−n2)Vπ(r)t(r) = n^2 V_{\sigma}(r) + (1 - n^2)V_{\pi}(r), where n=z/∣r∣n = z/|r|, Vσ(r)=Vσ0exp⁡[−βσ(r−c0)/c0]V_{\sigma}(r) = V^0_{\sigma}\exp[-\beta_{\sigma}(r - c_0)/c_0], and Vπ(r)=Vπ0[1+tanh⁡(−kπ(r−dπ))]/2V_{\pi}(r) = V^0_{\pi}[1 + \tanh(-k_{\pi}(r - d_{\pi}))]/2. The fitting references were Wannier Hamiltonian of AA, AB, and slightly displaced stackings from AB stacked bilayers. The cutoff distance of the interlayer hopping is 6 Å. The detailed parameters are listed in Table IV.

To reproduce the degeneracy splitting in abab initio bands of the trilayer BN, we also considered next-neighboring layer (NNL) interactions. The NNL hopping parameters are given as tNNL(r)=t0[1+tanh⁡(−γ(r−r0))]/2t^{\mathrm{NNL}}(r) = t_0[1 + \tanh(-\gamma(r - r_0))]/2, where t0=0.02t_0 = 0.02, −0.03, and −0.01 eV for B-N, B-B, and N-N hoppings, respectively, γ=5.96\gamma= 5.96 Å−1^{-1} and r0=6.753r_0 = 6.753 Å. The cutoff distance of the NNL hopping is 7 Å.

In addition, we considered the on-site level correction that mimics the potential energy gradient induced by the local dipole moment. To calculate the local gradient, we first determined the atomic portion of the dipole moment PiP_i of the atom ii for the interfaces between adjacent layers. The preliminary atomic portion PipreP_i^{\mathrm{pre}} is determined by dividing the dipole moment given by Eq. (5) in half and assigning it to the two atoms determining rBN\mathbf{r}_{\mathrm{BN}}, and adding it up for every B-N pair. Then, PiP_i is given as Pi=Pipre/2+∑j∈NNPjpre/6P_i = P_i^{\mathrm{pre}}/2 + \sum_{j\in\mathrm{NN}} P_j^{\mathrm{pre}}/6, where NN means the three nearest neighboring atoms. In this way, we can determine the smoothly distributed atomic portion of the dipole moments preserving their total moment. The on-site correction term is given as ΔEB/N=∓cB/NPi\Delta E_{\mathrm{B/N}} = \mp c_{\mathrm{B/N}}P_i for B and N atoms and cB=82c_{\mathrm{B}} = 82 V/Å, and cN=40c_{\mathrm{N}} = 40 V/Å, where the signs are for the upper and lower side of the interface, respectively. The resulting TB band structures are shown in Figure 11. The applied electric field is also simulated as an on-site energy level shift depending on the zz-coordinate by ΔU(z)=zEz/ϵr\Delta U(z) = zE_z/\epsilon_r with the fitted dielectric constant ϵr=1/0.413\epsilon_r = 1/0.413 to reproduce the DFT results.

Band structures near the CBM and VBM for monolayer and trilayer h-BN

Figure 11. Band structures near the CBM and VBM. (a) Monolayer h-BN band structures from DFT, Wannier Hamiltonian, and TB model calculations. Trilayer band structures from DFT and TB model calculations for (b) ABC, (c) ABA, and (d) ACA stackings.

The moiré supercell contains about three million atoms, which is equivalent to the dimension of the Hamiltonian matrix. Since the full-diagonalization is impractical, we only consider the low-energy eigenstates by using the ARPACK eigenvalue-solver which adopts the implicitly restarted Lanczos method [74, 75]. The linear equation solver implemented in the PETSc library is utilized for the shift-invert mode of the Lanczos scheme [76].

Area Variation of Polarized Domains under Electric fields

In this appendix, we discuss the field-induced variations in an area of polarized domains. In alternate and helical stackings shown in Figure 6 (a) and (b) respectively, with the external electric field parallel to the polarization direction of ABC stacking (blue in Figure 12), the total energy of the ABC stacking decreases and then maximize their area while the nonpolar (white) and opposite polarization (red) area decrease simultaneously.

In the twisted monolayer-bilayer system, a finite net polarization density modifies the potential energy landscape [illustrated in Figure 4 (a) and (b)] under an external electric field, EE as −P(Δx)⋅E−χE2-P(\Delta x) \cdot E - \chi E^2. The second term is negligible for small fields E<0.1E < 0.1 V/Å. Figure 13(a) shows the altered energy profile for a twist angle θ12=0.082∘\theta_{12} = 0.082^\circ under a weak applied field. Due to the shallow energy landscape around the kagome domain configuration, increasing EE eventually destabilizes this local minimum at a critical field strength of approximately E=0.04E = 0.04 V/Å. Figure 13 (b) illustrates how the polarization domains in each layer evolve as a function of the electric field. The dominant variation occurs in P23P_{23}, indicating that structural relaxation primarily involves the displacement of one-dimensional dislocation domain walls.

The critical fields at which the kagome domain becomes unstable are plotted in Figure 13 (c) with an uncertainty of 0.0025 V/Å. As the moiré length LL decreases, the energy barrier is reduced as shown in Figure 4 (a), and the critical field vanishes for L<500L < 500 Å, indicating that the kagome domain becomes unstable.

One-dimensional Domain Boundary States

We present a detailed analysis of domain boundary states in various TTBN stacking geometries by comparing the total density of states (DOS) and the boundary-projected local density of states (LDOS) for the three stacking configurations shown in Figure 2. In the case of alternate and twisted mono–bilayer stackings shown in Figure 14 (a) and (Figure 14c), the boundary-projected LDOS is further resolved according to the polarization nature of the adjacent domains. For the boundaries between two nonpolar domains, the well-defined 1D boundary DOS appears. Likewise, the LDOS between polar and nonpolar domains also supports 1D states. For the helical stacking geometry illustrated in Figure 14 (b), domain boundaries are labeled according to their adjacent domain

types—specifically triangular, hexagonal, and tetragonal domains—highlighting the geometric complexity of corner-shared hexagram domain tessellations unique to this configuration.

To further elucidate the spatial localization and energy characteristics of these domain boundary states, we calculate Gaussian-weighted integrals of the eigenstate distributions centered at representative boundary LDOS peak energies. These are shown in panels Figure 14(d)–(Figure 14f) for alternate, helical, and twisted mono–bilayer stackings, respectively. The Gaussian weight function is centered at the LDOS peaks identified in panels Figure 14(a)–(Figure 14c), with a smearing width of σ=5\sigma=5 meV; the corresponding ±σ\pm\sigma energy windows are indicated by gray-shaded regions in the LDOS plots. For clarity and systematic comparison, the four subpanels in each of Figure 14(d)–(Figure 14f) are arranged in ascending order of energy, allowing for an understanding of how the electronic states evolve across the energy spectrum in relation to their localization at domain boundaries. We note that all the boundary states are well localized at the interfaces between domains except a boundary state with an energy around 0 meV for the helical stacking in Figure 14(b). This state is a mixture of extended states on hexagonal domains and localized boundary states as shown in the fourth panel of Figure 14(e).

Quantum Dot States in Helical Stacking and Twisted monolayer-bilayer Systems

We investigate the relation between structural and electronic characteristics of QD states in TTBN with both helical and monolayer-bilayer configurations, as well as their electric reconfigurations. For the helical TTBN, Figure 15(c) shows the spatial variation of the fully relaxed interlayer spacing between the bottom and top layers without an applied field for the same region as Figure 2(d). Therein, localized increases in the interlayer distance imply QD sites labeled as 1, 1′, 2, and 2′. The enlarged view of site 1 is in Figure 15(g). The corresponding local conduction band edge profile, derived from TB models parametrized for multilayer h-BN, is presented in Figure 15(h) exhibiting pronounced minima for electrons.

The hole-counterpart from the valence band also manifests a well-defined maxima profile (not shown).

The electronic structures and charge density distributions of the confined states are obtained from direct TB calculations for multimillion atoms. For the helical case, calculated LDOS projected at each QD site shown in Figure 15 (a) exhibits well-isolated peaks representing QD states.

mxm_xmym_yℏωx\hbar\omega_xℏωy\hbar\omega_yE0E_0
H-CB0.4960.4965555−215
H-VB−0.662−0.6623232148
M-CB0.4420.5255082−330
M-VB−0.586−0.6932660164

Table V. The parameters of 2D QHO potentials for the helical stacking (θ12=θ23=θ=0.082∘\theta_{12} = \theta_{23} = \theta= 0.082^\circ) and twisted monolayer-bilayer TTBN (θ=0.082∘\theta= 0.082^\circ). The parameters and their unit follow the convention in Table I. H-CB (H-VB) indicates QHO potential parameters for potential wells near the conduction (valence) band edge of helical stacking while M-CB (M-VB) is that of twisted monolayer-bilayer stacking.

The first three lowest-energy conduction band states at site 1 are shown in Figure 15 (i). These states exhibit circular symmetry and nodal structures consistent with the isotropic 2D QHO states. Parameters charac-

(nx,ny)(n_x,n_y)(0,0)(0,0)(1,0)(1,0)(0,1)(0,1)(2,0)(2,0)(1,1)(1,1)(0,2)(0,2)
H-CB TB−164−112−112−63−63−63
H-CB QHO−161−107−107−52−52−52
H-VB TB1178888---
H-VB QHO1158383---
M-CB TB−271−220−192−174−151−126
M-CB QHO−264−215−182−165−133−100
M-VB TB12410273---
M-VB QHO1219662---

Table VI. Comparisons of QD state energies using TB approximation and 2D QHO models for the helical stacking and twisted monolayer-bilayer TTBNs. Relative values to the conduction band edge E−ECBE - E_{\mathrm{CB}} and to the valence band edge E−EVBE - E_{\mathrm{VB}} in meV. H-CB(VB) and M-CB(VB) follow the convention in Table V.

characterizing the QHO potentials and the corresponding QD energy levels are provided in Tables V and VI.

Additional well-defined QD states appear at even lower-symmetric vertices, which are labeled as 2 and 2′ in Figure 15(c) and indicated by corresponding LDOS peaks in Figure 15(a). These QDs form a honeycomb array [Figure 15(e)], with threefold rotational equivalence around the 1′ site. Under an out-of-plane field of ∣E∣=0.1|E|=0.1 V/Å, the QD states at moved 2 and 2′ sites [Figure 15(d)] remain localized but move closer by ∼15 nm [Figure 15(f)], demonstrating their electric reconfigurability. It is noteworthy that the LDOS of sites 1 and 1′, which were degenerate at zero field, also split under the applied field, further verifying field-tunable coupling between QDs.

For the twisted monolayer–bilayer TTBN, as shown in Figure 16(a), we also analyze the electronic structure, in which the LDOS peaks imply the well localized QD states. The QD sites are indicated in Figure 16(c) depicting the fully relaxed interlayer spacing in the same spatial region of Figure 2(f). A zoomed-in view of the site 1 in Figure 16(g) also reveals a distinct local maximum in interlayer spacing, again suggesting strong local confinement potential.

The conduction band energy landscape at this site [Figure 16(h)] also shows similar features to the helical case, with deeper potential wells for electrons and holes (not shown) indicated by darker regions in the respective maps. The associated localized electronic states include the six lowest-energy states in the conduction band as shown in Figure 16(i). The wavefunction profiles of these states closely follow the expected nodal patterns of anisotropic 2D QHO eigenfunctions. The full set of QHO potential parameters and the energy spectrum of the QD states are also listed in Tables V and VI for comparison.

In addition, the lower-symmetry vertex sites 2 and 3 labeled in Figure 16(c) also host spatially well-isolated QD states, as indicated by the LDOS peaks in Figure 16(a). Their density distributions are shown in Figure 16(e). Interestingly, the energy levels of Dot 2 and Dot 3 states fall within the energy range of the 1D boundary states that interconnect the Dot 1 sites [Figure 14(c) and (f)], even though these states are spatially well separated from the boundary states. Under an electric field ∣E∣=0.03|E|=0.03 V/Å, the three QD sites move closer to each other by ∼20 nm as shown in Figure 16(d). Furthermore, Dot 2 and Dot 3 states approach the 1D boundary “quantum-wire” states as shown in Figure 16(f), where the QD and 1D states appear together in the Gaussian-weighted density distributions. This represents a distinctive aspect of electric reconfiguration in the kagome-like pattern. As described in Appendix § 3.0.3, applying a stronger electric field than 0.03 V/Å induces a transition to a triangular pattern, rather than further reducing the interdot distance. For a similar reason, the QD motion in the kagome-like pattern is prominently more susceptible to the applied electric field.

A Scaling Rule of Inder-dot Distances in Alternated stacking Systems

In this appendix, we derive a single scaling relation between the electric-field and twist-angle dependence of interdot distance in the alternated stack TTBN, which provides practical information to estimate the critical electric field ECE_C to achieve the interdot coupling for different moiré periods. Figure 17(c) shows the electric-field dependence of the interdot distances ll of the highest Dot 2 states of nx=ny=0n_x=n_y=0 in the valence band of fully relaxed atomic structures of alternated stack (θ12,θ23)=(−θ,2θ)(\theta_{12},\theta_{23})=(-\theta,2\theta) with different θ\theta. It shows that the critical electric field of ECE_C required to saturate interdot spacing to lCl_C increases as the twist angle increases. The saturation can be well understood once we consider the hardcore effects on the vertex region where the elastic compression energy increases sharply as interdot distances decreases. Due to the local hardcore effects at the vertex, the lCl_C does not depend on the twist angle when the twist angle is small.

Universal scaling rule for interdot distance in alternating stack

Figure 17. Universal scaling rule for interdot distance in alternating stack. Domain pattern of alternating stacking TTBN of (θ12,θ23)=(−θ,2θ)(\theta_{12},\theta_{23})=(-\theta,2\theta) (a) without and (b) with the out-of-plane electric field. Blue (red) denotes the rhombohedral stack region whose polarization is (anti)parallel to the field, and gray denotes the Bernal stack. (c) Fully relaxed atomistic simulations on electric-field dependence of the interdot distances l(E)l(E) of the Dot 2 states of nx=ny=0n_x=n_y=0 in the valence band for different twist angle θ\theta. The filled circles represent the calculated data, and dashed lines are the corresponding fitting curves. (d) The same data as in (c), rescaled using the parameters in the analytic formula, A=aE/θA=aE/\theta and x=l(E)/l0x=l(E)/l_0, with fitting parameter a=0.203a=0.203 degrees·Å/V. “no vertex core” curve represents xmin⁡no−core=1−A/(1−A2)x_{\min}^{\mathrm{no-core}}=1-A/(1-A^2).

To elucidate these numerical simulation results phenomenologically, we formulated the total energy change (Δε)(\Delta\varepsilon) under an applied field (E)(E) in terms of the normalized interdot spacing x(θ,E)=l(θ,E)/l0x(\theta,E)=l(\theta,E)/l_0 where $l_0\equiv l(θ,E=0)=1.23/θl(\theta, E = 0) = 1.23/\theta is an angle-dependent interdot spacing in the absence of EE (its unit is Å/rad). l0l_0 without EE and ll with EE are denoted in Figure 17(a) and (Figure 17b), respectively. By using the domain geometries without and with EE [Figure 17(a) and (Figure 17b)], Δε\Delta\varepsilon upon the applied EE can be written as,

Δε=12εWl01+13(x−1)2−43δSl02(1−x)−6δS2εWl03(1−x)2+fc(x),(6)\Delta\varepsilon= 12\varepsilon_W l_0\sqrt{1+\frac{1}{3}(x-1)^2-4\sqrt{3}\delta_S l_0^2(1-x)}-\frac{6\delta_S^2}{\varepsilon_W}l_0^3(1-x)^2+f_c(x), \tag*{(6)}

where εW\varepsilon_W denotes the domain-wall energy per unit length, and δS\delta_S the stacking energy difference per unit area between Bernal (nonpolar) and rhombohedral (polar) domains. We note δS∼±PE\delta_S \sim\pm PE for the red and blue regions in Figure 17 where PP is the absolute magnitude of the layer polarization of the rhombohedral domain. The first term represents the elastic energy cost arising from the increased total length of domain walls with increasing field and the second term describes the energy gain from the field-induced expansion or shrinkage of the rhombohedral domains. The third term proportional to l03l_0^3 accounts for the warping of domain walls, and the last term of fc(x)f_c(x) introduces a hardcore vertex energy that diverges when ll approaches lCl_C. Here, we simply assume fc(x)=εc/(x−xC)f_c(x)=\varepsilon_c/(x-x_C) where xC=lC/l0x_C=l_C/l_0 and εc\varepsilon_c is a constant.

Equation (F1) can be recast in dimensionless form as

Δε2εWl0≈(1−a2E2θ2)(1−x)2−2aEθ(1−x)+bθx−xC,(7)\frac{\Delta\varepsilon}{2\varepsilon_W l_0}\approx\left(1-\frac{a^2E^2}{\theta^2}\right)(1-x)^2-2\frac{aE}{\theta}(1-x)+\frac{b\theta}{x-x_C}, \tag*{(7)}

where aa and bb are independent of twist angle and field strength. The equilibrium interdot spacing is obtained by minimizing Eq. (7) with respect to xx.

Comparison against the simulated data l(θ,E)l(\theta,E) is performed without the hardcore vertex effect term bθ/(x−xC)b\theta/(x-x_C) for further simplicity. Then, Eq. (7) becomes a much simplified form of (1−A2)(1−x)2−2A(1−x)(1-A^2)(1-x)^2-2A(1-x), where A≡aE/θA\equiv aE/\theta, thus directly giving a minimization condition xmin⁡no−core(A)=1−A/(1−A2)x_{\min}^{\mathrm{no-core}}(A)=1-A/(1-A^2). We fitted the parameter aa against the simulated l(θ,E)l(\theta,E) only for angles smaller than 0.20.2 degrees and for EE smaller than the saturation regime.

The single parameter a=0.203a=0.203 degrees·Å/V reproduces all the numerical data with different twist angles accurately, as shown by the dashed lines in Figure 17(c). Furthermore, when the interdot distances and electric fields are rescaled in terms of the variables xx and AA, all data before the saturation collapse onto a single scaling curve by xmin⁡no−corex_{\min}^{\mathrm{no-core}} [Figure 17(d)]. These results establish a single, well-defined scaling rule for the variation of domain area with electric field that unifies all calculated results across varying twist angles. With this, for all considered angles, the critical ACA_C for saturating xCx_C is almost constant for all angles so that EC=ACθ/a∼θE_C=A_C\theta/a\sim\theta, explaining our calculations very well.

Inhomogeneity effect

Inhomogeneity of the sample is often thought to hinder the reproducibility of phenomena in moiré systems. The inhomogeneity may come from two different materials parameters, namely twist angle disorder and local variations of strains. However, as demonstrated in the previous study on TTG [33], the formation of stacking domain patterns is robust over a finite range of twist-angle disorder and incommensuration. Even under these conditions, the trilayer lattice tends to self-organize into locally commensurate domain structures [33], which would generate almost identical QD states at each vertex. In particular, when we examine the sequence of twist angles used in Figure 17(c), the QD states at E=0E=0 in the alternate stack persist and the corresponding energy levels and spacing vary by only a few meV, except for shallow bound states. Moreover, to check the effect of strain on the QD energy levels, we consider a 0.5%0.5\% uniaxial tensile strain along either the armchair or zigzag direction for the fully relaxed θ=0.08∘\theta=0.08^\circ alternated stack structure without performing additional relaxation. All QD states remain, with a uniform shift in their absolute energies, while the level spacings stay nearly unchanged. The overall band gap decreases by about 4040 meV, and the threefold degeneracy of the equivalent Dot 2 sites is lifted by only a few meV. Together, these results confirm that the QD energy levels and couplings are robust against small twist-angle disorder and weak strain, ensuring the experimental feasibility and reproducibility of the proposed field-controlled QD arrays. Furthermore, experimental advances in reducing moiré sample inhomogeneity [77, 78] will strengthen the feasibility of implementing our proposal.

References

  1. [1]T. D. Ladd, F. Jelezko, R. Laflamme, Y. Nakamura, C. Monroe, and J. L. O’Brien, Quantum computers, Nature 464, 45 (2010).DOI
  2. [2]S. Ebadi, T. T. Wang, H. Levine, A. Keesling, G. Semeghini, A. Omran, D. Bluvstein, R. Samajdar, H. Pichler, W. W. Ho, S. Choi, S. Sachdev, M. Greiner, V. Vuletić, and M. D. Lukin, Quantum phases of matter on a 256-atom programmable quantum simulator, Nature 595, 227 (2021).
  3. [3]D. Bluvstein, H. Levine, G. Semeghini, T. T. Wang, S. Ebadi, M. Kalinowski, A. Keesling, N. Maskara, H. Pichler, M. Greiner, V. Vuletić, and M. D. Lukin, A quantum processor based on coherent transport of entangled atom arrays, Nature 604, 451 (2022).DOI
  4. [4]C. D. Bruzewicz, J. Chiaverini, R. McConnell, and J. M. Sage, Trapped-ion quantum computing: Progress and challenges, Appl. Phys. Rev. 6, 021314 (2019).DOI
  5. [5]C. Monroe, W. C. Campbell, L.-M. Duan, Z.-X. Gong, A. V. Gorshkov, P. W. Hess, R. Islam, K. Kim, N. M. Linke, G. Pagano, P. Richerme, C. Senko, and N. Y. Yao, Programmable quantum simulations of spin systems with trapped ions, Rev. Mod. Phys. 93, 025001 (2021).DOI
  6. [6]N. P. de Leon, K. M. Itoh, D. Kim, K. K. Mehta, T. E. Northrup, H. Paik, B. S. Palmer, N. Samarth, S. Sangtawesin, and D. W. Steuerman, Materials challenges and opportunities for quantum computing hardware, Science 372, eabb2823 (2021).DOI
  7. [7]R. Acharya, D. A. Abanin, L. Aghababaie-Beni, I. Aleiner, others, G. Q. Ai, and Collaborators, Quantum error correction below the surface code threshold, Nature 638, 920 (2025).DOI
  8. [8]S. Pezzagna and J. Meijer, Quantum computer based on color centers in diamond, Appl. Phys. Rev. 8, 011308 (2021).DOI
  9. [9]S. G. J. Philips, M. T. Mađzik, S. V. Amitonov, S. L. de Snoo, M. Russ, N. Kalhor, C. Volk, W. I. L. Lawrie, D. Brousse, L. Tryputen, B. P. Wuetz, A. Sammak, M. Veldhorst, G. Scappucci, and L. M. K. Vandersypen, Universal control of a six-qubit quantum processor in silicon, Nature 609, 919 (2022).DOI
  10. [10]G. Burkard, T. D. Ladd, A. Pan, J. M. Nichol, and J. R. Petta, Semiconductor spin qubits, Rev. Mod. Phys. 95, 025003 (2023).DOI
  11. [11]F. Borsoi, N. W. Hendrickx, V. John, M. Meyer, S. Motz, F. van Riggelen, A. Sammak, S. L. de Snoo, G. Scappucci, and M. Veldhorst, Shared control of a 16 semiconductor quantum dot crossbar array, Nat. Nanotechnol. 19, 21 (2024).DOI
  12. [12]Y. Pan, J. Yang, S. C. Erwin, K. Kanisawa, and S. Fölsch, Reconfigurable quantum-dot molecules created by atom manipulation, Phys. Rev. Lett. 115, 076803 (2015).pubmed.ncbi.nlm.nih.gov/26317740
  13. [13]L. Yan and P. Liljeroth, Engineered electronic states in atomically precise artificial lattices and graphene nanoribbons, Adv. Phys.: X 4, 1651672 (2019).
  14. [14]E. Y. Andrei, D. K. Efetov, P. Jarillo-Herrero, A. H. MacDonald, K. F. Mak, T. Senthil, E. Tutuc, A. Yazdani, and A. F. Young, The marvels of moiré materials, Nat. Rev. Mater. 6, 201 (2021).DOI
  15. [15]D. M. Kennes, M. Claassen, L. Xian, A. Georges, A. J. Millis, J. Hone, C. R. Dean, D. N. Basov, A. N. Pasupathy, and A. Rubio, Moiré heterostructures as a condensed-matter quantum simulator, Nat. Phys. 17, 155 (2021).
  16. [16]R. Bistritzer and A. H. MacDonald, Moiré bands in twisted double-layer graphene, Proc. Natl. Acad. Sci. U.S.A. 108, 12233 (2011).DOI
  17. [17]Y. Cao, V. Fatemi, A. Demir, S. Fang, S. L. Tomarken, J. Y. Luo, J. D. Sanchez-Yamagishi, K. Watanabe, T. Taniguchi, E. Kaxiras, R. C. Ashoori, and P. Jarillo-Herrero, Correlated insulator behaviour at half-filling in magic-angle graphene superlattices, Nature 556, 80 (2018).
  18. [18]A. R.-P. Montblanch, M. Barbone, I. Aharonovich, M. Atatüre, and A. C. Ferrari, Layered materials as a platform for quantum technologies, Nat. Nanotechnol. 18, 555 (2023).DOI
  19. [19]G. Trambly De Laissardière, D. Mayou, and L. Magaud, Localization of Dirac Electrons in Rotated Graphene Bilayers, Nano Lett. 10, 804 (2010).DOI
  20. [20]J. M. B. Lopes dos Santos, N. M. R. Peres, and A. H. Castro Neto, Continuum model of the twisted graphene bilayer, Phys. Rev. B 86, 155449 (2012).
  21. [21]K. Wang, K. De Greve, L. A. Jauregui, A. Sushko, A. High, Y. Zhou, G. Scri, T. Taniguchi, K. Watanabe, M. D. Lukin, H. Park, and P. Kim, Electrical control of charged carriers and excitons in atomically thin materials, Nat. Nanotechnol. 13, 128 (2018).DOI
  22. [22]M. H. Naik, S. Kundu, I. Maity, and M. Jain, Origin and evolution of ultraflat bands in twisted bilayer transition metal dichalcogenides: Realization of triangular quantum dots, Phys. Rev. B 102, 075413 (2020).DOI
  23. [23]X.-J. Zhao, Y. Yang, D.-B. Zhang, and S.-H. Wei, Formation of block flat bands in polar twisted bilayers without magic angles, Phys. Rev. Lett. 124, 086401 (2020).DOI
  24. [24]V. V. Enaldiev, F. Ferreira, J. G. McHugh, and V. I. Fal'ko, Self-organized quantum dots in marginally twisted MoSe2/WSe2 and MoS2/WS2 bilayers, npj 2D Mater. Appl. 6, 74 (2022).DOI
  25. [25]F. Li, D. Lee, N. Leconte, S. Javvaji, Y. D. Kim, and J. Jung, Moiré flat bands and antiferroelectric domains in lattice relaxed twisted bilayer hexagonal boron nitride under perpendicular electric fields, Phys. Rev. B 110, 155419 (2024).DOI
  26. [26]L. Xian, D. M. Kennes, N. Tancong-Dejean, M. Altarelli, and A. Rubio, Multiflat bands and strong correlations in twisted bilayer boron nitride: Doping-induced correlated insulator and superconductor, Nano Lett. 19, 4934 (2019).DOI
  27. [27]N. Nakatsuji, T. Kawakami, H. Tateishi, K. Kato, and M. Koshino, Moiré band engineering in twisted trilayer WSe2 (2025), 2504.20449.DOI
  28. [28]R. Ribeiro-Palau, C. Zhang, K. Watanabe, T. Taniguchi, J. Hone, and C. R. Dean, Twistable electronics with dynamically rotatable heterostructures, Science 361, 690 (2018).
  29. [29]H. Tang, Y. Wang, X. Ni, K. Watanabe, T. Taniguchi, P. Jarillo-Herrero, S. Fan, E. Mazur, A. Yacoby, and Y. Cao, On-chip multi-degree-of-freedom control of two-dimensional materials, Nature 632, 1038 (2024).DOI
  30. [30]N. Nakatsuji, T. Kawakami, and M. Koshino, Multiscale lattice relaxation in general twisted trilayer graphenes, Phys. Rev. X 13, 041007 (2023).DOI
  31. [31]I. M. Craig, M. Van Winkle, C. Groschner, K. Zhang, N. Dowlatshahi, Z. Zhu, T. Taniguchi, K. Watanabe, S. M. Griffin, and D. K. Bediako, Local atomic stacking and symmetry in twisted graphene trilayers, Nat. Mater. 23, 323 (2024).DOI
  32. [32]M. Van Winkle, N. Dowlatshahi, N. Khaloo, M. Iyer, I. M. Craig, R. Dhall, T. Taniguchi, K. Watanabe, and D. K. Bediako, Engineering interfacial polarization switching in van der waals multilayers, Nat. Nanotechnol. 19, 751 (2024).DOI
  33. [33]D. Park, C. Park, K. Yananose, E. Ko, B. Kim, R. Engelke, X. Zhang, K. Davydov, M. Green, H.-M. Kim, S. H. Park, J. H. Lee, S.-G. Kim, H. Kim, K. Watanabe, T. Taniguchi, S. M. Yang, K. Wang, P. Kim, Y.-W. Son, and H. Yoo, Unconventional domain tessellations in moiré-of-moiré lattices, Nature 641, 896 (2025).DOI
  34. [34]M. Vizner Stern, Y. Waschitz, W. Cao, I. Nevo, K. Watanabe, T. Taniguchi, E. Sela, M. Urbach, O. Hod, and M. Ben Shalom, Interfacial ferroelectricity by van der Waals sliding, Science 372, 1462 (2021).
  35. [35]K. Yasuda, X. Wang, K. Watanabe, T. Taniguchi, and P. Jarillo-Herrero, Stacking-engineered ferroelectricity in bilayer boron nitride, Science 372, 1458 (2021).
  36. [36]K. Yasuda, E. Zalys-Geller, X. Wang, D. Bennett, S. S. Cheema, K. Watanabe, T. Taniguchi, E. Kaxiras, P. Jarillo-Herrero, and R. Ashoori, Ultrafast high-endurance memory based on sliding ferroelectrics, Science 385, 53 (2024).DOI
  37. [37]R. Bian, R. He, E. Pan, Z. Li, G. Cao, P. Meng, J. Chen, Q. Liu, Z. Zhong, W. Li, and F. Liu, Developing fatigue-resistant ferroelectrics using interlayer sliding switching, Science 385, 57 (2024).DOI
  38. [38]C.-Y. Hsieh, Y.-P. Shim, M. Korkusinski, and P. Hawrylak, Physics of lateral triple quantum-dot molecules with controlled electron numbers, Rep. Prog. Phys. 75, 114501 (2012).DOI
  39. [39]G. Constantinescu, A. Kuc, and T. Heine, Stacking in bulk and bilayer hexagonal boron nitride, Phys. Rev. Lett. 111, 036104 (2013).pubmed.ncbi.nlm.nih.gov/23909342
  40. [40]J. H. Warner, M. H. Rümmeli, A. Bachmatiuk, and B. Büchner, Atomic resolution imaging and topography of boron nitride sheets produced by chemical exfoliation, ACS Nano 4, 1299 (2010).DOI
  41. [41]C. Park and Y.-W. Son, Condensation of preformed charge density waves in kagome metals, Nat. Commun. 14, 7309 (2023).DOI
  42. [42]S. Baroni, S. de Gironcoli, A. Dal Corso, and P. Giannozzi, Phonons and related crystal properties from density-functional perturbation theory, Rev. Mod. Phys. 73, 515 (2001).DOI
  43. [43]A. N. Kolmogorov and V. H. Crespi, Registry-dependent interlayer potential for graphitic systems, Phys. Rev. B 71, 235415 (2005).DOI
  44. [44]S.-H. Lee and Y.-W. Son, First-principles approach with a pseudohybrid density functional for extended Hubbard interactions, Phys. Rev. Res. 2, 043410 (2020).DOI
  45. [45]N. Tancogne-Dejean and A. Rubio, Parameter-free hybridlike functional based on an extended Hubbard model: DFT + U + V, Phys. Rev. B 102, 155117 (2020).
  46. [46]I. Timrov, F. Aquilante, L. Binci, M. Cococcioni, and N. Marzari, Pulay forces in density-functional theory with extended Hubbard functionals: From nonorthogonalized to orthogonalized manifolds, Phys. Rev. B 102, 235159 (2020).DOI
  47. [47]W. Liu, N.-J. Guo, S. Yu, Y. Meng, Z.-P. Li, Y.-Z. Yang, Z.-A. Wang, X.-D. Zeng, L.-K. Xie, Q. Li, J.-F. Wang, J.-S. Xu, Y.-T. Wang, J.-S. Tang, C.-F. Li, and G.-C. Guo, Spin-active defects in hexagonal boron nitride, Mater. Quantum Technol. 2, 032002 (2022).DOI
  48. [48]M. Otani and S. Okada, Gate-controlled carrier injection into hexagonal boron nitride, Phys. Rev. B 83, 073405 (2011).DOI
  49. [49]G. Pizzi, V. Vitale, R. Arita, S. Blügel, F. Freimuth, G. Géranton, M. Gibertini, D. Gresch, C. Johnson, T. Koretsune, J. Ibañez-Azpiroz, H. Lee, J.-M. Lihm, D. Marchand, A. Marrazzo, Y. Mokrousov, J. I. Mustafa, Y. Nohara, Y. Nomura, L. Paulatto, S. Poncé, T. Ponweiser, J. Qiao, F. Thöle, S. S. Tsirkin, M. Wierzbowska, N. Marzari, D. Vanderbilt, I. Souza, A. A. Mostofi, and J. R. Yates, Wannier90 as a community code: new features and applications, J. Phys.: Condens. Matter 32, 165902 (2020).
  50. [50]Electronic states and modulation doping of hexagonal boron
  51. [51]A. Weston, E. G. Castanon, V. Enaldiev, F. Ferreira, S. Bhattacharjee, S. Xu, H. Corte-León, Z. Wu, N. Clark, A. Summerfield, T. Hashimoto, Y. Gao, W. Wang, M. Hamer, H. Read, L. Fumagalli, A. V. Kretinin, S. J. Haigh, O. Kazakova, A. K. Geim, V. I. Fal’ko, and R. Gorbachev, Interfacial ferroelectricity in marginally twisted 2d semiconductors, Nature Nanotechnol. 17, 390 (2022).DOI
  52. [52]M. Koshino, N. F. Q. Yuan, T. Koretsune, M. Ochi, K. Kuroki, and L. Fu, Maximally localized wannier orbitals and the extended hubbard model for twisted bilayer graphene, Phys. Rev. X 8, 031087 (2018).
  53. [53]M. H. Naik and M. Jain, Ultraflatbands and shear solitons in moiré patterns of twisted bilayer transition metal dichalcogenides, Phys. Rev. Lett. 121, 266401 (2018).DOI
  54. [54]S.-Y. Lee and Y. Chung, Charge stability of a triple quantum dot with a finite tunnel coupling, Phys. Rev. B 87, 045302 (2013).DOI
  55. [55]D. Loss and D. P. DiVincenzo, Quantum computation with quantum dots, Phys. Rev. A 57, 120 (1998).DOI
  56. [56]M. Seo, H. K. Choi, S.-Y. Lee, N. Kim, Y. Chung, H.-S. Sim, V. Umansky, and D. Mahalu, Charge frustration in a triangular triple quantum dot, Phys. Rev. Lett. 110, 046803 (2013).pubmed.ncbi.nlm.nih.gov/25166188
  57. [57]A. Hamo, A. Benyamini, I. Shapir, I. Kivrich, J. Weissman, K. Kaasbjerg, Y. Oreg, F. von Oppen, and S. Ilani, Electron attraction mediated by coulomb repulsion, Nature 535, 395 (2016).DOI
  58. [58]J. P. Dehollain, U. Mukhopadhyay, V. P. Michal, Y. Wang, B. Wunsch, C. Reichl, W. Wegscheider, M. S. Rudner, E. Demler, and L. M. K. Vandersypen, Nagaoka ferromagnetism observed in a quantum dot plaquette, Nature 579, 528 (2020).DOI
  59. [59]R. Gong, X. Du, E. Janzen, V. Liu, Z. Liu, G. He, B. Ye, T. Li, N. Y. Yao, J. H. Edgar, E. A. Henriksen, and C. Zu, Isotope engineering for spin defects in van der waals materials, Nat. Commun. 15, 104 (2024).DOI
  60. [60]R. Rizzatto, M. Schalk, S. Mohr, J. C. Hermann, J. P. Leibold, F. Bruckmaier, G. Salviati, C. Qian, P. Ji, G. V. Astakhov, U. Kentsch, M. Helm, A. V. Stier, J. J. Finley, and D. B. Bucher, Extending the coherence of spin defects in hBN enables advanced qubit control and quantum sensing, Nat. Commun. 14, 5089 (2023).
  61. [61]T. T. Tran, K. Bray, M. J. Ford, M. Toth, and I. Aharonovich, Quantum emission from hexagonal boron nitride monolayers, Nat. Nanotechnol. 11, 37 (2016).
  62. [62]J. Große, M. von Helversen, A. Koulas-Simos, M. Hermann, and S. Reitzenstein, Development of site-controlled quantum dot arrays acting as scalable sources of indistinguishable photons, APL Photonics 5, 096107 (2020).
  63. [63]C. Su, F. Zhang, S. Kahn, B. Shevitski, J. Jiang, C. Dai, A. Ungar, J.-H. Park, K. Watanabe, T. Taniguchi, J. Kong, Z. Tang, W. Zhang, F. Wang, M. Crommie, S. G. Louie, S. Aloni, and A. Zettl, Tuning colour centres at a twisted hexagonal boron nitride interface, Nat. Mater. 21, 896 (2022).
  64. [64]J. D. Caldwell, I. Aharonovich, G. Cassabois, J. H. Edgar, B. Gil, and D. N. Basov, Photonics with hexagonal boron nitride, Nat. Rev. Mater. 4, 552 (2019).DOI
  65. [65]J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized gradient approximation made simple, Phys. Rev. Lett. 77, 3865 (1996). M. H. Naik, I. Maity, P. K. Maiti, and M. Jain, Kolmogorov–Crespi potential for multilayer transition-metal dichalcogenides: Capturing structural transformations in moiré superlattices, J. Phys. Chem. C 123, 9770 (2019).
  66. [67]H. Peng, Z.-H. Yang, J. P. Perdew, and J. Sun, Versatile van der waals density functional based on a meta-generalized gradient approximation, Phys. Rev. X 6, 041005 (2016).DOI
  67. [68]E. A. Cortés, J. M. Florez, and E. S. Morell, Ferroelectric response to interlayer shifting and rotations in trilayer hexagonal boron nitride, J. Phys. Chem. Solids 173, 111086 (2023).DOI
  68. [69]W. Yang, S.-H. Jhi, S.-H. Lee, and Y.-W. Son, Ab initio study of lattice dynamics of group IV semiconductors using pseudohybrid functionals for extended Hubbard interactions, Phys. Rev. B 104, 104313 (2021).
  69. [70]https://github.com/KITAS-CMT/DFT-U-V.
  70. [71]P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni, D. Ceresoli, G. L. Chiarotti, M. Cococcioni, I. Dabo, A. Dal Corso, S. de Gironcoli, S. Fabris, G. Fratesi, R. Gebauer, U. Gerstmann, C. Gougoussis, A. Kokalj, M. Lazzeri, L. Martin-Samos, N. Marzari, F. Mauri, R. Mazzarello, S. Paolini, A. Pasquarello, L. Paulatto, C. Sbraccia, S. Scandolo, G. Sclauzero, A. P. Seit­sonen, A. Smogunov, P. Umari, and R. M. Wentzcovitch, QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials, J. Phys.: Condens. Matter 21, 395502 (2009).DOI
  71. [72]P. Giannozzi, O. Andreussi, T. Brumme, O. Bunau, M. Buongiorno Nardelli, M. Calandra, R. Car, C. Cavazzoni, D. Ceresoli, M. Cococcioni, N. Colonna, I. Carnimeo, A. Dal Corso, S. De Gironcoli, P. Delugas, R. A. DiStasio, A. Ferretti, A. Floris, G. Fratesi, G. Fugallo, R. Gebauer, U. Gerstmann, F. Giustino, T. Gorni, J. Jia, M. Kawamura, H.-Y. Ko, A. Kokalj, E. Küçükbenli, M. Lazzeri, M. Marsili, N. Marzari, F. Mauri, N. L. Nguyen, H.-V. Nguyen, A. Otero-de-la-Roza, L. Paulatto, S. Poncé, D. Rocca, R. Sabatini, B. Santra, M. Schliph, A. P. Seit­sonen, A. Smogunov, I. Timrov, T. Thonhauser, P. Umari, N. Vast, X. Wu, and S. Baroni, Advanced capabilities for materials modeling with Quantum ESPRESSO, J. Phys.: Condens. Matter 29, 465901 (2017).
  72. [73]S. Javvaji, F. Li, and J. Jung, Ab initio tight-binding models for mono- and bilayer hexagonal boron nitride (h-BN), Phys. Rev. Materials 9, 024004 (2025).arxiv.org/abs/2404.13041
  73. [74]R. B. Lehoucq, D. C. Sorensen, and C. Yang, ARPACK users’ guide: solution of large-scale eigenvalue problems with implicitly restarted Arnoldi methods (Society for Industrial and Applied Mathematics, Philadelphia, 1998).DOI
  74. [75]Opencollab, Arpack-ng, https://github.com/opencollab/arpack-ng, [Accessed: 24-July-2025].
  75. [76]S. Balay, W. D. Gropp, L. C. McInnes, and B. F. Smith, Efficient Management of Parallelism in Object-Oriented Numerical Software Libraries, in Modern Software Tools for Scientific Computing, edited by E. Arge, A. M. Bruaset, and H. P. Langtangen (Birkhäuser, Boston, MA, 1997) pp. 163–202.DOI
  76. [77]M. Kapfer, B. S. Jessen, M. E. Eisele, M. Fu, D. R. Danielsen, T. P. Darlington, S. L. Moore, N. R. Finney, A. Marchese, V. Hsieh, P. Majchrzak, Z. Jiang, D. Biswas, P. Dudin, J. Avila, K. Watanabe, T. Taniguchi, S. Ulstrup, P. Bøggild, P. J. Schuck, D. N. Basov, J. Hone, and C. R. Dean, Programming twist angle and strain profiles in 2D materials, Science 381, 677 (2023).DOI
  77. [78]B. Wu, H. Zheng, S. Li, C.-T. Wang, J. Ding, J. He, Z. Liu, J.-T. Wang, and Y. Liu, Enhanced homogeneity of moiré superlattices in double-bilayer WSe₂ homostructure, ACS Appl. Mater. Interfaces 15, 48475 (2023).DOI

Paper details

vlnc:W_k5y8txpe
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents