Correlated states of a triangular net of coupled quantum wires: Implications for the

phase diagram of marginally twisted bilayer graphene

Chuan Chen,1, 2 A. H. Castro Neto,1, 3 and Vitor M. Pereira1, 3, ∗
1
Centre for Advanced 2D Materials and Graphene Research Centre,
National University of Singapore, Singapore 117546
2
Max-Planck Institute for the Physics of Complex Systems, 01187 Dresden, Germany
3
Department of Physics, National University of Singapore, Singapore 117542
(Dated: April 29, 2020)
We explore in detail the electronic phases of a system consisting of three non-colinear arrays
of coupled quantum wires, each rotated 120 degrees with respect to the next. A perturbative
renormalization-group analysis reveals that multiple correlated states can be stabilized: a s-wave or
arXiv:1907.11450v2 [cond-mat.str-el] 28 Apr 2020

d ± id superconductor, a charge density wave insulator, a two-dimensional Fermi liquid, and a 2D

Luttinger liquid (also known as smectic metal or sliding Luttinger liquid). The model provides an
effective description of electronic interactions in small-angle twisted bilayer graphene and we discuss
its implications in relation to the recent observation of correlated and superconducting ground states
near commensurate densities in magic-angle twisted samples, as well as the “strange metal” behavior
at finite temperatures as a natural outcome of the 2D Luttinger liquid phase.

I. INTRODUCTION tight-binding models for those bands22–25 enabled, on the

one hand, predictions of possible broken symmetries aris-
ing from weak-coupling mechanisms, such as Fermi sur-
The low-energy physics of interacting fermions in one face nesting or enhanced density of states24,26–29 ; on the
dimension (1D) is determined by collective spin and other hand, it revealed trilobed Wannier functions cen-
charge density excitations that define what is known tered at the AB/BA positions23–25 , which has in turn mo-
as Luttinger liquid (LL) behavior1–3 . Soon after high- tivated strong-coupling perspectives based on extended
temperature superconductivity was discovered in cuprate and non-conventional Hubbard-type interactions30–32 .
oxides4 , it was proposed that the charges added upon Electronic interactions are an undisputed factor given
doping a Mott insulator could end up distributed in that the ratio of the local Coulomb integral to bandwidth
stripes5–9 . This led Anderson et al. to suggest that con- is estimated in the range U/w ∼ 5–1023 . The extremely
fined fermionic excitations in such presumed LL arrays large Moiré unit cells involved ( ∼ 172 nm2 , about 100
(the stripes) could explain the non-Fermi-liquid nature times those of canonical Mott-insulators like cuprates)
of cuprates’ “normal” state10 . Since then, theoretical in- has also prompted the suggestion that the insulating
vestigation has assessed whether LL behavior can emerge phase can be a Wigner crystal, consistently with the ex-
in higher dimensions, especially in 2D10–15 , the natural tremely low densities, and the emergence of SC a result
route to that having been to study systems of coupled of its melting33,34 .
LLs in different guises. It is now known, for example, Evidence accumulated from recent experiments and
that, in an array of parallel LLs, marginal inter-wire theoretical work motivates a different perspective over
density-density and current-current interactions lead to the effective electronic model governing correlations in
strong transverse charge-density fluctuations at incom- MTBG, which we develop in this paper. We note, first,
mensurate wave vectors which can frustrate electron crys- that, by allowing more than the minimum two orbitals24
tallization and indeed stabilize a LL state, commonly per Moiré unit cell, Carr et al. have recently shown
designated “smectic metal” or “sliding Luttinger liquid” that the weight of the Bloch states belonging to the
state12–15 . However, previous work has been limited to flat-band sector is overwhelmingly distributed among
exploring consequences of couplings among either one or Wannier functions situated at the AA positions and at
two perpendicularly crossed arrays, without ever consid- the AB/BA domain boundaries 35 . Second, it is well
ering LLs interlinked in the form of a triangular net, pos- known that, in the presence of perpendicular electrical
sibly for lack of a realistic representative system. fields, AB-BA domain boundaries host protected helical
The recent discovery of strongly correlated physics in modes36–40 . MTBG accommodates well defined, intrin-
marginally twisted bilayer graphene (MTBG) near the sic, and periodically alternating AB/BA regions21 whose
magic angle θ ≈ 1.1◦16,17 set off a flurry of interest in network of boundaries was shown to likewise support
the origin of the observed insulating and superconduct- the propagation of such confined states41–43 . Moreover,
ing (SC) phases. (For the purposes of this work, MTBG since AB is favored against AA stacking, a considerable
refers to bilayers twisted by ∼ 1◦ or less, including the atomic relaxation within the Moiré unit cell maximizes
first magic angle.) It had been previously suggested that, the AB/BA regions, leaving sharply defined, atomic-scale
at magic angles, the quasi-flatness of the electronic bands domain boundaries44–49 . Crucially, there is now unequiv-
closest to the undoped Fermi level could promote elec- ocal spectroscopic44,50–53 and transport50,54–56 evidence
tronic instabilities18–21 . The development of effective of the reality of this network of 1D modes in MTBG,
 2

(a) (b) II. COUPLED-WIRE MODEL

A. Interconnected Luttinger liquid array

d
K2 Mirroring the experimental network of AB-BA bound-
aries, we consider three families of quantum wires, each
family consisting of an array of parallel wires depicted
1
2 by the same color in Fig. 1(a). Although each link in
3 MTBG supports two helical channels per spin along the
K1 forward and backward direction, we study the simplified
case of a single mode per wire, which should capture the
FIG. 1. (a) Schematic of the local net spanned by the three essential physics at play. Each independent wire is a LL
coupled arrays of quantum wires, labeled {1, 2, 3}. One array whose excitations are best described in the boson formal-
consists of a set of parallel, identically colored wires. The ism by a separated spin (s) and charge (c) Hamiltonian
superposition of the three arrays reproduces the net of AB- (see Appendix for the convention of bosonization)1–3 ,
BA domain boundaries in MTBG. (b) Domain of stability.  
The array’s effective Luttinger parameter, κ(k⊥ ), is positive X Z vα 1
in the shaded domain for all k ∈ (−π/d, π/d). The LL phase H= dx (∂x θα )2 + Kα (∂x φα )2 , (1)
α=c,s
2 Kα
exists only in the red-to-yellow region surrounding the upper
boundary. Dashed lines indicate the cuts chosen to generate
where θα (x) and φα (x) are the conventional phase-field
the phase diagrams in Fig. 2.
operators, Kc/s is the (inverse) charge/spin Luttinger
parameter (Kc ≷ 1 for repulsive/attractive interactions),
and vα defines the velocity of each excitation. Spin
√ back-


scattering adds the term 2gs /(2πα)2 dx cos 2 2πφs ,
R
including in single-gated devices. where α is the (length) cutoff of the theory, to Hamil-
tonian (1), leading to a sine-Gordon-like action and spin
There are additional hints that warrant a description couplings that flow according to the equations2
in terms of such “network of linked quantum wires” to de-
scribe the observed correlated behavior: the confinement dgs 1 dKs gs2
= 2(1 − )gs , = . (2)
of electrons to 1D naturally boosts correlations; a phase dl Ks dl 2π 2 vs2
diagram similar to that of magic-angle samples arises at As a result, when there is a spin gap, Ks → ∞.
other twist angles under pressure57 , in line with the ex- To describe each periodic array of parallel wires sep-
pectation that an inter-layer coupling enhanced by pres- arated by d as in Fig. 1, one must include the long-
sure would amplify the lattice relaxation, in turn defining wavelength (charge) density-density and current-current
sharp domain boundaries and the emergence of the wire interactions among wires in the fixed-point Hamiltonian,
network for more generic twists54–56 . as first noted by Emery et al.12 . It can then be shown
All the above aspects and observations call for an inves- that the charge part of the action reads, in Fourier space,
tigation of the implications of a coupled-wire description X v(k⊥ )k 2  |θc,q |2 
2
of the low-energy physics of MTBG. Wu et al. have re- S= + κ(k⊥ )|φc,q |
cently advanced arguments to justify the insulating and q
2 κ(k⊥ )
SC phases in such a scenario, but only considering cou- + iωn kφ∗c,q θc,q , (3)
pling at the wire intersections and spin isotropic interac-
tions within each wire58 . By generalizing to MTBG the where q ≡ (ωn , k, k⊥ ) and k/k⊥ is the momentum
approach developed to study sliding LL phases, we scru- along/perpendicular to the wires12–14 . The spin part,
tinize not only the competition among SC and charge- for θs and φs , is obtained by replacing [v(k⊥ ), κ(k⊥ )] →
density wave (CDW) states, but also the emergence of [vs , Ks ]. Direct comparison with Eq. (1) shows that an
Fermi liquid (FL) and sliding LL phases that, stabi- array of LLs effectively behaves as a LL, the net effect
lized by inter-wire interactions, might explain the exper- of the inter-wire coupling being a Luttinger parameter κ
imental progression of MTBG from an insulator to SC that is now a 2π/d-periodic function of k⊥ . This period-
to a metal as density deviates from commensurate fill- icity justifies a Fourier expansion,
ings. The SC order parameter acquires either s or d ± id κ(k⊥ ) = K0 [1 + K1 cos(k⊥ d) + K2 cos(2k⊥ d) + . . . ], (4)
symmetry, depending on the Josephson coupling at the
wire crossings. Moreover, it suggests that these corre- which we shall use below with K0,1,2 as free
lated phases could also happen in “marginally” (θ  1◦ ) parameters13–15 .
twisted bilayer graphene. Finally, we emphasize that, in- Each of the three LL arrays depicted in Fig. 1 is as-
dependently of its direct relevance to MTBG, this work signed a (superscript) label j ∈ {1, 2, 3}. Within each ar-
reports the first detailed study of coupled quantum wires ray j, we consider the single-electron hopping (t⊥ ) be-
and sliding LL phases in a triangular net geometry. tween nearest-neighboring wires, as well as inter-wire
 3

CDW (Vn ) and SC (Jn ) singlet interactions between n- Coupling (0, K− ) (K− , 1/2) (1/2, 1) (1, K+ ) (K+ , ∞)
th neighboring wires. These are described, respectively, intra-array (among parallel wires)
by the intra-array, inter -wire couplings
Vn (CDW) – – 3 3 3
Jn (SC) – – 3 3 3
X X j†
Hhj = t⊥ j
ψl,ν,σ ψl+1,ν,σ + H. c., (5a)
l,σ ν=±1
t⊥ (hop) – 3 3 3 –
j
X j† j j† j inter-array (at wire crossings)
Hc,n = Vn ψl,ν,σ ψl,−ν,σ ψl+n,−ν,σ 0 ψl+n,ν,σ 0 , (5b)
l,σ,σ 0 ,ν
V0 (CDW) – – – 3 3
j
X j† j† j j J0 (SC) – – – 3 3
Hsc,n = Jn ψl,µ,↑ ψl,−µ,↓ ψl+n,ν,↓ ψl+n,−ν,↑ + H. c., t (hop) – – – – –
l,µ,ν
(5c) TABLE I. Relevance of each coupling for the different ranges
i
of the spin Luttinger parameter (Ks ) specified in the first row.
where ψl,±1,σ
is the field operator for a right/left-moving The symbol 3 means that a coupling may be relevant while
electron of spin σ in the l-th wire of array i. The wires are “–” indicates
√ it is always irrelevant
√ within that interval of Ks .
also coupled at each intersection [white dots in Fig. 1(a)], K− ≡ 3 − 2 2 ' 0.17, K+ ≡ 3 + 2 2 ' 5.83.
requiring us to consider the additional inter -array hop-
ping, CDW and SC interactions:
X X i†
Hhi,j = t j
ψl,µ,σ ψm,ν,σ + H. c., (6a)
l,m,σ µ,ν
XXX i† j† j
Hci,j = V0 ψl,µ,σ i
ψl,µ,σ ψm,ν,σ 0 ψm,ν,σ 0 , (6b) second order. Furthermore, in order to have a stable the-
l,m σ,σ 0 µ,ν ory, κ(k⊥ ) must be positive for k⊥ ∈ (−π/d, π/d), which
X i† i† j j
i,j
Hsc = J0 ψl,µ,↑ ψl,−µ,↓ ψm,ν,↓ ψm,−ν,↑ + H. c.. (6c) constrains K0 > 0 and (K1 , K2 ) to the shaded domain
l,m,µ,ν shown in Fig. 1(b).

B. Renormalization group equations At this level of approximation, the RG equations (7)

are independent. The relevancy of the different couplings
can thus be immediately established and is summarized
Once all the couplings in Eqs. (5) and (6) are written
in Table I. Since κ is strictly positive, ∆C,0 + ∆S,0 ≥ 2
in terms of the bosonic fields, we proceed by developing
which, according to Eq. (7d), implies that the single-
a perturbative renormalization group (RG) analysis. To
electron hopping at the wire intersections (t) is, at most,
the lowest order, the flow equations for the hopping (t,
marginal if Ks = 1 and κ(k/d) = 1 for all k; it is otherwise
t⊥ ), CDW (Vn ), and SC (Jn ) coupling parameters read:
irrelevant in nearly the whole phase space. This justi-
fies considering t globally irrelevant and, accordingly, it
dVn 1 will be ignored in the subsequent analysis. Similarly, one
= (2 − δn,0 − − ∆C,n )Vn , (7a)
dl Ks can see that ∆C,1 + ∆S,1 ≥ 2 so that the
√ intra-array hop-
√
dJn 1 ping (t⊥ ) may be relevant when 3 − 2 2 ≤ Ks ≤ 3 + 2 2.
= (2 − δn,0 − − ∆S,n )Jn , (7b) The CDW and SC couplings are relevant only if Ks > 1/2
dl Ks
dt⊥

1 1 1
 in the intra-array case (V1,2 and J1,2 ), while the corre-
= 2 − (Ks + ) − (∆C,1 + ∆S,1 ) t⊥ , (7c) sponding inter-array couplings (V0 and J0 ) are relevant
dl 4 Ks 4 for Ks > 1.
 
dt 1 1 1
= 1 − (Ks + ) − (∆C,0 + ∆S,0 ) t, (7d)
dl 4 Ks 4
Up to this point, the spin Luttinger parameter Ks has
where, been considered free; Table I thus covers the most gen-
Z π eral scenario in relation to the possible magnetic phases.
dk 1 However, addition of the spin backscattering term men-
∆C,n ≡ [1 − (1 − δn,0 ) cos(nk)] , (8a)
−π 2π κ(k/d) tioned earlier to Eq. (1) makes Ks a running coupling,
π
governed by the flow Eqs. (2). The solution where
Z
dk
∆S,n ≡ [1 − (1 − δn,0 ) cos(nk)] κ(k/d). (8b) Ks → ∞ corresponds to a spin-gapped state, in which
−π 2π
case we find the single-electron hoppings t and t⊥ to
It is physically reasonable to expect the intra-array cou- be irrelevant (last column of Table I), in correspondence
plings to decay rapidly so, henceforth, we only con- with previous calculations for a single array of coupled
sider intra-array CDW and SC interactions up to second- quantum wires12 . In contrast, if Ks → 0 we have a spin
neighbors. As for κ(k⊥ ), in line with Vishwanath and gapless state and all the couplings considered here are
Carpentier13 , we truncate its Fourier expansion at the irrelevant — the system consists of decoupled LLs.
 4

(a) phase diagram, while the CDW eventually becomes the

LL
1.4
only relevant phase for large K0 . In the crossover re-
gion K0 ∼ 1, the domains of relevancy for the CDW and
s or
d ± id
SC SC orders overlap; in addition, the intra-array hopping
K1 CDW is relevant as well in this case (gray region enclosed by
1.3 the dot-dashed line) which, should the hopping become
dominant over the CDW and SC instabilities, implies the
1.2 existence of a 2D FL phase. This indicates that the tran-
1.1
0 sition between SC and CDW with increasing repulsion
0.5 1 5 10 50
(b)
K0 (increasing K0 ) can occur either directly or via an in-
1.0 tervening FL phase, depending on the magnitude of K1
s or SC
d ± id
(which is a measure of the nearest-neighbor inter-wire
0.8 coupling within an array). The precise outcome of this
0.6 phase competition, or coexistence, depends on how the
K1 running of one coupling constant affects the others, whose
0.4 CDW analysis requires a perturbative RG calculation beyond
first order, which is not in the scope of this paper.
0.2
When K2 is finite, in addition to a FL, one finds that
0.0 a LL phase is stabilized between the SC and CDW√re-
0.5 1 5 10 gion close to the parameter-space boundary at K1 → 2.
K0
This is marked by the green area in Fig. 2(a) or the or-
ange region in Fig. 1(b). Physically, the appearance of
FIG. 2. Phase diagram along the horizontal cuts marked in
a LL phase in this case arises from the fact that K2
Fig. 1 and for Ks = 2: (a) K2 = 0.5, (b) K2 = 0. The param-
eters K0 and K1 (axes) are the Fourier coefficients defined promotes interaction between next-nearest-neighboring
in Eq. (4). In the region above the solid-blue line, one of J1 wires within an array, which is detrimental to the sta-
or J2 is relevant (SC order). In the region below the solid- bility of the CDW. As first pointed out by Vishwanath
orange line, at least one of V1 and V2 is relevant (CDW order). and Carpentier13 , when K2 is included, near the (K1 , K2 )
The intra-array hopping (t⊥ ) is relevant in the gray domain space boundary, the minimum of the κ(k⊥ ) is located
bounded by the dash-dotted line, implying that the system at some incommensurate k⊥ and the value is close to
might be a Fermi liquid in this region. To the left of the blue- zero, which indicate strong fluctuations of a transverse
dashed line, the inter-array SC coupling is relevant, whereas incommensurate CDW order [the density correlation
the inter-array CDW coupling is relevant to the right of the hφ∗c,k,k⊥ φc,k,k⊥ i ∝ 1/κ(k⊥ )]. The incommensurate CDW
orange-dashed line. The green area indicates a regime where
fluctuations then destroy the crystallization so that all
all the couplings are irrelevant, corresponding to a 2D LL
state. The main difference between (a) and (b) is the absence
couplings are irrelevant within that region. In contrast,
of the LL phase in the latter. for K2 = 0, the most divergent transverse CDW is com-
mensurate as K1 → 1; i.e., 1/κ(k⊥ ) = 1/K0 [1 + cos(k⊥ d)]
diverges at k⊥ = π/d, so the next-nearest-neighbor intra-
array CDW coupling will crystallize the system and there
III. PHASE DIAGRAM AND ANALYSIS is no LL phase. Indeed, comparing Figs. 2(a) and 2(b),
we see that a LL phase is stabilized at the expense of
A. Instability tendencies the CDW phase in the repulsive region (K0 > 1), without
much impact on the region of SC stability. At large K0 ,
both intra-array (rightward off the solid orange line) and
While one may explore any range of Ks , we will now fo- inter-array (rightward off the dashed orange line) CDW
cus on Ks = 2. Table I shows that this falls in the regime couplings are relevant — the electrons crystallize and an
where all couplings but t are relevant and, therefore, it is insulator ensues.
representative of the physical scenarios involving phase If K0 < 0.5, the inter-array Josephson coupling is rele-
competition, as is the case of MTBG, either driven by vant. At each wire crossing, the phases of the three SC
inter- or intra-array interactions (or both). Figure 2 order parameters couple via
shows the phase diagram in two representative scenar-
ios, defined by different magnitudes of the second har-
X
∝ J0 cos(ϕir − ϕjr ). (9)
monic in the Fourier expansion (4). Although K0 is not i,j∈{1,2,3}
strictly the Luttinger (charge) parameter of an individual
wire, Eq. (3) implies it does represent the effective Lut- Assuming that the intra-array SC coupling promotes uni-
tinger parameter of an array behaving collectively as a form SC within each array, Eq. (9) indicates that the
LL12 . Therefore, K0 < 1 signals an effectively attractive global SC phase depends on the sign of J0 : if J0 < 0, the
regime while K0 > 1 describes repulsion. In this context, Josephson coupling favors s-wave SC with all ϕi equal;
one qualitatively understands the fact that the SC phase but, if J0 > 0, that coupling is frustrated and will re-
(blue region) dominates in the small-K0 portion of the sult in a 2π/3 difference between the phase of the SC
 5

order parameter of one array (j) with respect to the next IV. DISCUSSION
(j + 1). This originates a d ± id SC symmetry. A similar
conclusion has been drawn by Wu et al. who have fur- A. Landscape of correlated states
ther considered triplet pairing and discuss the additional
possibility of p ± ip symmetry58 . The propagation of interacting electrons along the
quantum channels provided by the well defined AB-BA
domain boundaries of small-angle MTBG [Fig. 1(a)] pro-
vides a natural low-energy picture for the emergence of
competing SC and insulating states. At generic densi-
B. Commensurability ties and moderate Luttinger parameter (|K0 | ∼ 1), we ob-
tain SC and possibly FL as the dominant phases, with
SC stabilized even for repulsive Coulomb interactions
In a conventional (i.e., single) LL problem, the prox- (K0 > 1), which is noteworthy; at commensurate densi-
imity to commensurate electron densities is described ties, the system is a charge-insulator. This holds both
by considering the Umklapp process within each wire2 when Ks < 1 and Ks > 1, particularly in the spin-gapped
which, in the notation of Eq. (1), has the form regime (Ks  1) where the only qualitative difference is
the possible loss of the FL phase at very high Ks — this
is significant for the model applicability to MTBG, where
gU
Z √
dx cos[2 2πφc + (4kF − G)x], (10) a magnetic field has been seen to destroy the insulating
(2πα)2 state16,57 .

where G is a vector of the reciprocal superlattice. The B. Anomalous metallic behavior

couplings gU and Kc flow according to Eq. (2), with the
replacements gs → gU , Ks → Kc 2 . In the present case, Most interestingly, we see that the interaction among
however, the effective Luttinger parameter κ is a func- parallel wires contributes to stabilize both “sliding” and
tion of the transverse momentum [cf. Eq. (3)] due to “crossed sliding” LL phases, thus extending previous
the marginal interactions between wires within each ar- findings for square arrays12–15 to this triangular geom-
ray; this complicates the flow equations in the charge etry as well. These phases are extremely interesting be-
sector. We proceed by assuming, as a first approxima- cause, on the one hand, they define a regime of metallic
tion, that the flow equations for gU and K0 behave anal- 2D transport underpinned entirely by Luttinger-liquid
ogously to those in Eq. (2), in which case we naturally behavior and interactions, with the consequence that
obtain distinct behavior at and away from half-filling: physical observables scale anomalously with tempera-
Our phase diagram in Fig. 2 indicates that, away from ture, size, and fields14 . On the other hand, and as a
half-filling, the system is a SC provided K0 is not too result, these regimes of 2D transport are entirely dif-
large; at (or near) half-filling, a large enough gU is able ferent from that of an effective circuit of independent
to drive the system to an insulating state even for very 1D wires. Perhaps most significantly for current exper-
small K0 . Such SC-to-insulator transition is a general iments is the fact that charge transport in these phases
feature of the competing instabilities in a LL with com- would have anisotropic fingerprints and an anomalous
mensurate density, because the Umklapp terms provide temperature dependence, thus being a natural candi-
a “condensation” energy gain that ultimately makes the date for the “strange metal” behavior reported in magic-
charge-gapped CDW state energetically favorable2,59,60 . angle MTBG right above the temperatures where the in-
This competition between SC and CDW insulating sulating and SC states disappear61 . Moreover, in this
states bears directly on the current experimental ob- picture, a symmetry breaking among the three equiva-
servations with magic-angle MTBG, which show the lent wire arrays would naturally impart both local and
ground-state to be either a FL at generic densities, global electronic “nematicity”, a feature that has recently
a SC near commensurate fillings, or an insulator at been inferred from high-resolution scanning tunneling
commensurability16,17 — the coupled LL scenario is con- microscopy (STM) experiments62–65 .
sistent with such observation. For quantitative compar-
isons in this regard, it is worth noting that the electronic
filling/density reported for the 2D experimental system C. The nature of the 1D wire net
needs to be converted to 1D electronic densities by tak-
ing into account that, in the coupled-wire picture, each As the presence of 1D modes traveling along the
Moiré unit cell contains three non-equivalent wires. For AB/BA domain boundaries is a decisive precondition for
example, increasing the electron density by one electron modeling MTBG in terms of the proposed network of
per Moiré unit cell away from charge neutrality corre- coupled Luttinger liquids, in the remainder we elaborate
sponds to adding 1/3 electrons per segment of each non- on their evidence so far as well as on means through which
equivalent wire within that unit cell. they can be ensured.
 6

It is now well established that MTBG undergoes con- it is know that, similarly to a graphene monolayer37,69–72 ,
siderable internal deformation within each Moiré unit cell the Moiré and relaxation induced by boron nitride sub-
so as to maximize extension of the energetically more fa- strates generates a spectral gap for certain crystallo-
vorable Bernal stacking at the expense of the AA-stacked graphic orientations in bilayer graphene as well71,73–75 .
regions. This energetic tendency is constrained by frus- Therefore, a bulk gap that guarantees and stabilizes the
tration at the interface between AB and BA regions and 1D modes can be engineered with appropriate substrate
results in bilayers with uniform Bernal stacking essen- conditions.
tially everywhere, except at sharp AB/BA domain walls Finally, we note that there is a strong pseudomag-
and AA vertices whose geometry is depicted in Fig. 1(a). netic field due to the lattice relaxation, with magnitudes
Numerical calculations reveal this relaxation effect very that might exceed 10 T45,66 . As a result of the trian-
clearly44–48 : it starts becoming prominent for twist an- gular shape of the AB and BA domains, that field is
gles below ∼ 2–3◦47,66,67 and is completely established quasi-uniform within the Bernal regions, but with op-
for θ . 1◦ , at which point the width of the domain bound- posite polarity — the polarity sharply switches precisely
aries saturates at ∼ 6–9 nm (becomes independent of along the domain walls. The combined effect of large
twist amount for smaller angles)44,47,51 . This threshold pseudomagnetic fields and abrupt polarity changes along
is confirmed experimentally55 and MTBG samples have the domain walls is likely to efficiently confine snake-type
been reported with domain walls extending up to the mi- chiral states76,77 . These would be chiral 1D modes of a
cron scale while retaining their atomic-scale width44 . different kind, which do not require a bulk gap76 .
The electronic modes localized at the AB/BA domain
boundaries of deliberately biased bilayer graphene are ex-
pected to behave as perfect 1D quantum wires, so long D. Conclusions
as intervalley scattering remains unimportant36 ; explicit
bandstructure calculations have recently shown this to Different experimental probes and theoretical work are
be indeed realized in relaxed MTBL35,49 . More impor- persuasive enough of the conclusion that the triangular
tantly, this has been confirmed by measurements that array of coupled quantum wires illustrated in Fig. 1(a)
probed electric transport along isolated AB/BA bound- is the natural starting point to describe transport and
aries extending over several microns, which revealed the correlated states in biased MTBG. The accuracy of this
expected conductance quantization at ≈ 4e2 /h and or- picture increases with larger bulk gaps and angles . 1◦ ,
ders of magnitude enhancement of the mean-free path which ensures the sharpest domain boundaries as well
associated with these modes, when compared with that as sufficiently long channels between the AA vertices for
elsewhere in the sample50,68 . Their confinement to the a valid LL description of each quantum wire (the inter-
domain walls has been confirmed directly by local STM vertex distance is 14 nm for θ = 1◦ ). The facts that such
and STS measurements51,52 as well as indirectly: (i) by modes have been equally seen in unbiased experiments
the observation of Aharanov-Bohm oscillations in mag- and that pseudomagnetic fields can themselves beget ad-
netotransport with spatial periods that correlate with ditional 1D states of a different nature, suggest the rele-
closed paths along adjacent domain walls54,56 ; (ii) by the vance of this description to unbiased devices as well. In-
saturation of resistance near h/(4e2 )55,56 and metallic deed, the phenomenology of the correlated states, which
temperature dependence despite increases in interlayer so far has been scrutinized only in the unbiased case, tal-
bias (i.e., resistivity saturation and metallic temperature lies with the phase diagram arising from the coupled LL
dependence with increasing bulk gap)56 ; (iii) and by the model, namely when it comes to: the types of correlated
local enhancement of infrared optical conductivity at the states involved and their competition, the sequence of
AB/BA interfaces, which is associated with the presence phase transitions with doping, the association of CDW
of the 1D modes53 . insulating states with commensurate fillings, the exis-
All of the above indicates that the coupled-wire model tence of non-Fermi liquid metallic states, and nematicity.
should provide an adequate description of biased MTBG, Seeing as the detailed mechanisms underpinning both
where the tunable bulk gap and topological character the insulating and SC states in MTBG remain an open
of the 1D states ensure the robustness of these elec- problem, it is of utmost interest to experimentally scru-
tronic modes, in the regime where the Fermi level re- tinize the evolution of the correlated phase diagram in
mains within the bulk gap. In the case of magic-angle MTBG (at the magic as well as smaller angles) as a func-
MTBG, it would be interesting to experimentally inves- tion of the bulk gap through interlayer bias. This would
tigate the fate of the correlated insulator and SC states place the system in the regime where our model most re-
under a finite interlayer bias. liably applies, while it would also assess its relevance to
Whether these domain-wall-bound modes survive and the strictly unbiased case.
remain influential at zero interlayer bias can depend on Note — Recently, a preprint emerged with a similar
the conditions of the substrate. For example: these con- formulation78 , but with more restricted applicability di-
fined modes have been seen directly by STM/STS against rectly to MTBG: it considers a wire net with C4 sym-
a gapped Bernal background of graphene bilayers de- metry rather than C6 , assumes persistent local SC order
posited on graphite, without any electrostatic bias51 ; and in puddle regions that encompass the wire intersections,
 7

and couplings are considered only at those intersections, ACKNOWLEDGMENTS

without intra-array interactions. In specific cases (pa-
rameter ranges) where the two models can be compared, CC thanks X. Y. Gu and J. N. Leaw for helpful discus-
the conclusions agree. sions. This work was supported by the National Research
Foundation of Singapore under its Medium-Sized Centre
Programme.

∗ 27
Correspondence to: vpereira@nus.edu.sg H. Isobe, N. F. Q. Yuan, and L. Fu, Phys. Rev. X 8,
1
F. D. M. Haldane, Journal of Physics C: Solid State Physics 041041 (2018).
28
14, 2585 (1981). Y.-Z. You and A. Vishwanath, npj Quantum Materials 4,
2
T. Giamarchi, Quantum Physics in One Dimension (Ox- 16 (2019).
29
ford University Press, 2003). E. Laksono, J. N. Leaw, A. Reaves, M. Singh, X. Wang,
3
E. Fradkin, Field Theories of Condensed Matter Physics S. Adam, and X. Gu, Solid State Commun. 282, 38
(Cambridge University Press, 2013). (2018).
4 30
J. G. Bednorz and K. A. Müller, Zeitschrift für Physik B C. Xu and L. Balents, Phys. Rev. Lett. 121, 087001 (2018).
31
Condensed Matter 64, 189 (1986). X. Y. Xu, K. T. Law, and P. A. Lee, Phys. Rev. B 98,
5
J. M. Tranquada, D. J. Buttrey, V. Sachan, and J. E. 121406 (2018).
32
Lorenzo, Phys. Rev. Lett. 73, 1003 (1994). X. Gu, C. Chen, J. N. Leaw, E. Laksono, V. M.
6
J. Zaanen and O. Gunnarsson, Phys. Rev. B 40, 7391 Pereira, G. Vignale, and S. Adam, arXiv e-
(1989). prints , arXiv:1902.00029 (2019), arXiv:1902.00029 [cond-
7
K. Machida, Phys. C Supercond. 158, 192 (1989). mat.supr-con].
8 33
M. Kato, K. Machida, H. Nakanishi, and M. Fujita, J. B. Padhi, C. Setty, and P. W. Phillips, Nano Lett. 18,
Phys. Soc. Japan 59, 1047 (1990). 6175 (2018).
9 34
V. J. Emery, S. A. Kivelson, and J. M. Tranquada, Proc. B. Padhi and P. W. Phillips, Phys. Rev. B 99, 205141
Natl. Acad. Sci. 96, 8814 (1999). (2019).
10 35
S. P. Strong, D. G. Clarke, and P. W. Anderson, Phys. S. Carr, S. Fang, Z. Zhu, and E. Kaxiras, Phys. Rev. Res.
Rev. Lett. 73, 1007 (1994). 1, 013001 (2019).
11 36
X. G. Wen, Phys. Rev. B 42, 6623 (1990). I. Martin, Y. M. Blanter, and A. F. Morpurgo, Phys. Rev.
12
V. J. Emery, E. Fradkin, S. A. Kivelson, and T. C. Luben- Lett. 100, 036804 (2008).
37
sky, Phys. Rev. Lett. 85, 2160 (2000). M. Kindermann, B. Uchoa, and D. L. Miller, Phys. Rev.
13
A. Vishwanath and D. Carpentier, Phys. Rev. Lett. 86, B 86, 115415 (2012).
38
676 (2001). F. Zhang, A. H. MacDonald, and E. J. Mele, Proc. Natl.
14
R. Mukhopadhyay, C. L. Kane, and T. C. Lubensky, Phys. Acad. Sci. 110, 10546 (2013).
39
Rev. B 63, 081103 (2001). A. Vaezi, Y. Liang, D. H. Ngai, L. Yang, and E.-A. Kim,
15
R. Mukhopadhyay, C. L. Kane, and T. C. Lubensky, Phys. Phys. Rev. X 3, 021018 (2013).
40
Rev. B 64, 045120 (2001). M. Koshino, Phys. Rev. B 88, 115409 (2013).
16 41
Y. Cao, V. Fatemi, A. Demir, S. Fang, S. L. Tomarken, P. San-Jose and E. Prada, Phys. Rev. B 88, 121408 (2013).
42
J. Y. Luo, J. D. Sanchez-Yamagishi, K. Watanabe, D. K. Efimkin and A. H. MacDonald, Phys. Rev. B 98,
T. Taniguchi, E. Kaxiras, R. C. Ashoori, and P. Jarillo- 035404 (2018).
43
Herrero, Nature 556, 80 (2018). H. K. Pal, S. Spitz, and M. Kindermann, Phys. Rev. Lett.
17
Y. Cao, V. Fatemi, S. Fang, K. Watanabe, T. Taniguchi, 123, 186402 (2019).
44
E. Kaxiras, and P. Jarillo-Herrero, Nature 556, 43 (2018). J. S. Alden, A. W. Tsen, P. Y. Huang, R. Hovden,
18
E. Suárez Morell, J. D. Correa, P. Vargas, M. Pacheco, L. Brown, J. Park, D. A. Muller, and P. L. McEuen, Proc.
and Z. Barticevic, Phys. Rev. B 82, 121407 (2010). Natl. Acad. Sci. 110, 11256 (2013).
19 45
G. Li, A. Luican, J. M. B. Lopes dos Santos, A. H. Castro N. N. T. Nam and M. Koshino, Phys. Rev. B 96, 075311
Neto, A. Reina, J. Kong, and E. Y. Andrei, Nat. Phys. 6, (2017).
46
109 (2010). M. Andelković, L. Covaci, and F. M. Peeters, Phys. Rev.
20
R. Bistritzer and A. H. MacDonald, Proc. Natl. Acad. Sci. Mater. 2, 034004 (2018).
47
108, 12233 (2011). F. Gargiulo and O. V. Yazyev, 2D Mater. 5, 015019 (2017).
21 48
J. M. B. Lopes dos Santos, N. M. R. Peres, and A. H. P. Lucignano, D. Alfè, V. Cataudella, D. Ninno, and
Castro Neto, Phys. Rev. B 86, 155449 (2012). G. Cantele, Phys. Rev. B 99, 195419 (2019).
22 49
J. M. B. Lopes dos Santos, N. M. R. Peres, and A. H. N. R. Walet and F. Guinea, 2D Mater. 7, 015023 (2019).
50
Castro Neto, Phys. Rev. Lett. 99, 256802 (2007). L. Ju, Z. Shi, N. Nair, Y. Lv, C. Jin, J. Velasco, C. Ojeda-
23
M. Koshino, N. F. Q. Yuan, T. Koretsune, M. Ochi, Aristizabal, H. A. Bechtel, M. C. Martin, A. Zettl, J. An-
K. Kuroki, and L. Fu, Phys. Rev. X 8, 031087 (2018). alytis, and F. Wang, Nature 520, 650 (2015).
24 51
H. C. Po, L. Zou, A. Vishwanath, and T. Senthil, Phys. L.-J. Yin, H. Jiang, J.-B. Qiao, and L. He, Nat. Commun.
Rev. X 8, 031089 (2018). 7, 11760 (2016).
25 52
J. Kang and O. Vafek, Phys. Rev. X 8, 031088 (2018). S. Huang, K. Kim, D. K. Efimkin, T. Lovorn, T. Taniguchi,
26
C.-C. Liu, L.-D. Zhang, W.-Q. Chen, and F. Yang, Phys. K. Watanabe, A. H. MacDonald, E. Tutuc, and B. J.
Rev. Lett. 121, 217001 (2018). LeRoy, Phys. Rev. Lett. 121, 037702 (2018).
 8

53 64
S. S. Sunku, G. X. Ni, B. Y. Jiang, H. Yoo, A. Stern- Y. Choi, J. Kemmer, Y. Peng, A. Thomson, H. Arora,
bach, A. S. McLeod, T. Stauber, L. Xiong, T. Taniguchi, R. Polski, Y. Zhang, H. Ren, J. Alicea, G. Refael, F. von
K. Watanabe, P. Kim, M. M. Fogler, and D. N. Basov, Oppen, K. Watanabe, T. Taniguchi, and S. Nadj-Perge,
Science 362, 1153 (2018). Nat. Phys. (2019), 10.1038/s41567-019-0606-5.
54 65
P. Rickhaus, J. Wallbank, S. Slizovskiy, R. Pisoni, Y. Xie, B. Lian, B. Jäck, X. Liu, C.-l. Chiu, K. Watanabe,
H. Overweg, Y. Lee, M. Eich, M.-H. Liu, K. Watanabe, T. Taniguchi, B. A. Bernevig, and A. Yazdani, Nature
T. Taniguchi, T. Ihn, and K. Ensslin, Nano Lett. 18, 6725 572, 101 (2019).
66
(2018). M. M. van Wijk, A. Schuring, M. I. Katsnelson, and A. Fa-
55
H. Yoo, R. Engelke, S. Carr, S. Fang, K. Zhang, solino, 2D Mater. 2, 034010 (2015).
67
P. Cazeaux, S. H. Sung, R. Hovden, A. W. Tsen, K. Zhang and E. B. Tadmor, J. Mech. Phys. Solids 112,
T. Taniguchi, K. Watanabe, G.-C. Yi, M. Kim, M. Luskin, 225 (2018).
68
E. B. Tadmor, E. Kaxiras, and P. Kim, Nat. Mater. 18, J. Li, K. Wang, K. J. McFaul, Z. Zern, Y. Ren, K. Watan-
448 (2019). abe, T. Taniguchi, Z. Qiao, and J. Zhu, Nat. Nanotechnol.
56
S. G. Xu, A. I. Berdyugin, P. Kumaravadivel, F. Guinea, 11, 1060 (2016).
69
R. K. Kumar, D. A. Bandurin, S. V. Morozov, W. Kuang, B. Hunt, J. D. Sanchez-Yamagishi, A. F. Young,
B. Tsim, S. Liu, J. H. Edgar, I. V. Grigorieva, V. I. Fal’ko, M. Yankowitz, B. J. LeRoy, K. Watanabe, T. Taniguchi,
M. Kim, and A. K. Geim, Nat. Commun. 10, 4008 (2019). P. Moon, M. Koshino, P. Jarillo-Herrero, and R. C.
57
M. Yankowitz, S. Chen, H. Polshyn, Y. Zhang, K. Watan- Ashoori, Science 340, 1427 (2013).
70
abe, T. Taniguchi, D. Graf, A. F. Young, and C. R. Dean, Z.-G. Chen, Z. Shi, W. Yang, X. Lu, Y. Lai, H. Yan,
Science 363, 1059 (2019). F. Wang, G. Zhang, and Z. Li, Nat. Commun. 5, 4461
58
X.-C. Wu, C.-M. Jian, and C. Xu, Phys. Rev. B 99, 161405 (2014).
71
(2019). P. Moon and M. Koshino, Phys. Rev. B 90, 155406 (2014).
59 72
T. Giamarchi, Physica B: Condensed Matter 230-232, J. Jung, A. M. DaSilva, A. H. MacDonald, and S. Adam,
975 (1997), proceedings of the International Conference Nat. Commun. 6, 6308 (2015).
73
on Strongly Correlated Electron Systems. M. Mucha-Kruczynski, J. R. Wallbank, and V. I. Fal’ko,
60
V. J. Emery, “Theory of the one-dimensional electron gas,” Phys. Rev. B 88 (2013), 10.1103/PhysRevB.88.205418.
74
in Highly Conducting One-Dimensional Solids, edited by C. R. Dean, L. Wang, P. Maher, C. Forsythe, F. Ghahari,
J. T. Devreese, R. P. Evrard, and V. E. van Doren Y. Gao, J. Katoch, M. Ishigami, P. Moon, M. Koshino,
(Springer US, Boston, MA, 1979) pp. 247–303. T. Taniguchi, K. Watanabe, K. L. Shepard, J. Hone, and
61
Y. Cao, D. Chowdhury, D. Rodan-Legrain, O. Rubies- P. Kim, Nature 497, 598 (2013).
75
Bigordà, K. Watanabe, T. Taniguchi, T. Senthil, and H. Kim, N. Leconte, B. L. Chittari, K. Watanabe,
P. Jarillo-Herrero, arXiv preprint (2019), 1901.03710. T. Taniguchi, A. H. MacDonald, J. Jung, and S. Jung,
62
Y. Jiang, X. Lai, K. Watanabe, T. Taniguchi, K. Haule, Nano Lett. 18, 7732 (2018).
76
J. Mao, and E. Y. Andrei, Nature (2019), 10.1038/s41586- G. W. Jones, D. A. Bahamon, A. H. Castro Neto, and
019-1460-4, 1904.10153. V. M. Pereira, Nano Lett. 17, 5304 (2017).
63 77
A. Kerelsky, L. J. McGilly, D. M. Kennes, L. Xian, J. E. Müller, Phys. Rev. Lett. 68, 385 (1992).
78
M. Yankowitz, S. Chen, K. Watanabe, T. Taniguchi, Y.-Z. Chou, Y.-P. Lin, S. Das Sarma, and R. M. Nandk-
J. Hone, C. Dean, A. Rubio, and A. N. Pasupathy, Nature ishore, Phys. Rev. B 100, 115128 (2019).
572, 95 (2019).

Appendix A: Bosonization conventions

According to the bosonization approach2,3 , the fermion field from the l-th wire of array j with spin σ and direction
µ (+1 for right and −1 for left) reads

1 √ j
j
ψl,µ,σ ∝√ eiµ2 πφl,µ,σ , (A1)
2πα

where α is the length cut-off. The bosonic fields are related by:

φjl,σ = φjl,+,σ + φjl,−,σ , (A2a)

j
θl,σ = −φjl,+,σ
+ φjl,−,σ , (A2b)
1
φjl,c/s = √ (φjl,↑ ± φjl,↓ ), (A2c)
2
j 1 j j
θl,c/s = √ (θl,↑ ± θl,↓ ). (A2d)
2
 9

The subscript c/s in Eq. (A2)(c,d) stands for the charge/spin sector. The bosonic fields are related to the electron
densities through:

1
√ ∂xj φjl,µ,σ = ρjl,µ,σ , (A3a)
π
1 1
√ ∂xj φjl,c/s = √ (ρjl,+,c/s + ρjl,−,c/s ), (A3b)
π 2
1 1
√ ∂xj θl,c/s = √ (−ρjl,+,c/s + ρjl,−,c/s ),
j
(A3c)
π 2

where xj is the variable along the direction of j-th array and ρc/s stands for the charge/spin density, i.e., ρc/s = ρ↑ ±ρ↓ .

Appendix B: Bosonized form of the couplings

Once we know the bosonized form of the fermion fields, it is straightforward to write all the couplings in terms of
the bosonic fiels defined in the previous section. For the intra-array couplings, we consider an array of parallel wires
along the x direction [see the schematic of array-1 in Fig. 1(a) as an example]: The Hamiltonian terms describing the
coupling between wires separated by a distance nd are:

X X Z j,† j j,† j
j
Hc,n = Vn dxψl,µ,σ ψl,−µ,σ ψl+n,−µ,σ 0 ψl+n,µ,σ 0

l µ,σ,σ 0
2Vn XZ h√ i √ √
= dx cos 2π(φjl,c − φjl+n,c ) cos( 2πφjl,s ) cos( 2πφjl+n,s ) (B1a)
(πα)2
l

XZ j,† j,† j j
j
Hsc,n = Jn dxψl,µ,↑ ψl,−µ,↓ ψl+n,ν,↓ ψl+n,−ν,↑ + H. c.
l,µ,ν
2Jn X
Z h√ i √ √
j j
= dx cos 2π(θ l,c − θ l+n,c ) cos( 2πφjl,s ) cos( 2πφjl+n,s ) (B1b)
(πα)2
l

XXZ
Hhj = t⊥ j,†
dxψl,µ,σ j
ψl+1,µ,σ + H. c.
l µ,σ
Z r  r 
2t⊥ X π j j π j
= dx cos (φl,c − θl,c − φjl+1,c + θl+1,c
j
) cos j
(φl,s − θl,s − φjl+1,s + θl+1,s
j
)
πα 2 2
l
r  r 
π j j j j π j j j j
× cos (φ + θl,c − φl+1,c − θl+1,c ) cos (φ + θl,s − φl+1,s − θl+1,s ) (B1c)
2 l,c 2 l,s
 10

Meanwhile, the inter-array couplings at a crossing point (e.g., the intersection of the l-th wire from array k and the
m-th from array j) can be written in the form:
k,j
X k† j† j
k
Hcdw = V0 ψl,µ,σ ψl,−µ,σ ψm,ν,σ 0 ψm,−ν,σ 0

µ,ν,σ,σ 0
4V0 √ √ √ √
= 2
cos( 2πφkl,c ) cos( 2πφkl,s ) cos( 2πφjm,c ) cos( 2πφjm,s ) (B2a)
(πα)

X k† k† j j
k,j
Hsc = J0 ψl,µ,↑ ψl,−µ,↓ ψm,ν,↓ ψm,−ν,↑ + H. c.
µ,ν
4J0 i√2π(θl,c
k j
−θm,c )
√ √
= 2
e cos( 2πφll,s ) cos( 2πφjm,s ) + H. c. (B2b)
(πα)

X
Hhk,j = t k†
ψl,µ,σ j
ψm,ν,σ + H. c.
µ,ν,σ

t X i√ π2 (θl,c
r  r 
k k
+σθl,s j
−θm,c j
−σθm,s ) π k π j
= e cos (φl,c + σφkl,s ) cos (φm,c + σφjm,s ) + H. c. (B2c)
πα σ 2 2

Having obtained the identities above, one can proceed with the perturbative RG calculation to explore the potential
instabilities of the system.