betaFeedback

WorksPhysics and Astronomy

Heavy fermion phase diagram in magic-angle twisted trilayer graphene

Abstract

The interplay between localized magnetic moments and itinerant electrons gives rise to exotic quantum states in condensed matter systems. Here, we demonstrate an electrically tunable heavy fermion phase diagram in magic-angle twisted trilayer graphene, achieved by controlling the Kondo hybridization between localized flat-band electrons and itinerant Dirac electrons via a displacement field. Our results reveal a continuous quantum phase transition from an antiferromagnetic semimetal to a paramagnetic heavy fermion metal. At quantum critical point, we observe effective mass divergence and Fermi surface reconstruction. This highly tunable platform offers unprecedented control over heavy fermion physics, establishing moire heterostructures as a versatile arena for exploring correlated quantum phases-including potential unconventional superconductivity-in two-dimensional limit.

Two-dimensional (2D) moiré superlattices with flat bands have emerged as a versatile platform for exploring strongly correlated matters [1]. The strong on-site interactions among localized electrons can give rise to a plethora of quantum phases, including unconventional superconductivity [2, 3], correlated insulators [4]. Beyond the established Hubbard physics, moiré materials can be engineered to simulate Kondo lattice physics by coupling localized moments (ff-electrons) with a sea of itinerant conduction electrons (cc-electrons) [5, 6, 7]. In magic-angle twisted bilayer graphene (MATBG), hybridization between ff-electrons centered at AA-stacking regions and plane-wave-like itinerant cc-electrons can produce topological flat bands near the Fermi level EfE_{\mathrm{f}} [8, 9, 10, 11, 12, 13, 14]. Its counterpart, magic-angle twisted trilayer graphene (MATTG), hosts both flat bands and a set of Dirac bands. These additional Dirac bands not only help stabilize superconducting states [15, 16, 17], but also provide extra cc-electrons and an unprecedented form of in-situ tunability. This enables the emulation of the full heavy-fermion phase diagram, going beyond what is possible in MATBG and semiconducting transition-metal dichalcogenide heterobilayers [18, 19, 20, 21].

In conventional rare-earth-based Kondo lattices, the ground state can evolve into a heavy Fermi liquid via the Kondo effect or develop magnetic order mediated the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction. In MATTG, the competing interplay can be continuously tuned via a perpendicular displacement field. Here, we report the experimental observation of a continuous crossover from a heavy Fermi liquid to an antiferromagnetic semimetal, along with direct evidence of Fermi surface reconstruction at filling factor ν=n/n0=3\nu= n/n_0 = 3 (here n0n_0 is the charge density corresponding to one electron per moiré unit cell). By mapping the generalized Doniach phase diagram in a single dual-gated device, our work may illuminate the mechanism behind unconventional superconductivity and reveal novel emergent phenomena beyond those found in bulk Kondo lattice materials.

The high-quality dual-graphite-gated MATTG devices, featuring three adjacent graphene layers twisted in an alternating sequence as illustrated in Figure 1(a), are prepared by the dry transfer method. The vertical mirror symmetry structure gives rise to a unique band structure at zero displacement field D=0D = 0 V nm−1^{-1}, comprising MATBG-like flat bands and Dirac bands originated from the additional monolayer graphene Figure 1(b). Upon applying a large D=0.9D = 0.9 V nm−1^{-1}, the flat bands hybridize with the Dirac bands, leading to the coupling of ff-electrons with cc-electrons and the formation of a Kondo lattice Figure 1(c), right]. To experimentally probe this Kondo phase, we have measured two MATTG devices with twist angles of 1.52∘1.52^\circ (device D1) and 1.45∘1.45^\circ (device D2). Despite the slight variation in twist angle, the electronic properties of both devices show high consistency (see Supplemental Material Fig. S1 [22]). Following established transport characterization [15, 17], we first identify the electronic phases in MATTG. Figure 1(d) displays a typical longitudinal resistance RxxR_{xx} mapping as a function of filling factor ν\nu and displacement field DD at T=1.8T = 1.8 K, where ν\nu and DD are controlled by top/bottom gate voltages. The correlated insulating resistance peaks at integer filling ν=1,±2,3\nu= 1, \pm2, 3 and charge density wave peaks at fractional fillings become more pronounced as the magnitude of ∣D∣\lvert D\rvert increases and eventually evolve into low-resistance metallic states beyond a critical displacement field.

Image not extracted

Figure 1. (a) Schematic of the dual-gated MATTG device. (b), (c) Tight-binding band structures for MATTG at zero (b) and finite displacement field (c). Right panel, schematic of localized moments (black arrow) interacting with itinerant electrons (purple arrow) at D=0D = 0 V nm−1^{-1} and D=0.9D = 0.9 V nm−1^{-1}. (d) Longitudinal resistance RxxR_{xx} as a function of ν\nu and DD at T=1.8T = 1.8 K. (e), (f) Landau fan diagrams of RxxR_{xx} (e) and RxyR_{xy} (f) at D=0D = 0 V nm−1^{-1}. (g) Quantum Hall states at D=0D = 0 V nm−1^{-1} and ν=3\nu= 3. Gray dashed lines mark quantized values corresponding to the sequence −2-2, −6-6, −10-10, …\ldots.

The coexistence of ff- and cc-electrons is further evident in the Landau fan diagram. As shown in Figure 1(e) and (f), the decoupled flat bands and Dirac bands at zero DD give rise to two distinct sets of Landau fans: dense MATBG-like fans originating from integer and fractional fillings at high magnetic field and extra quantum oscillations emanating from the charge neutrality point (CNP) at low field. By measuring the quantum Hall states [Fig. 1(g)], we can confirm that the additional quantum oscillations are in fact from Dirac bands. From the temperature dependence of Shubnikov-de Haas (SdH) oscillations at low magnetic field, we extract an effective mass of m∗∼0.005m0m^* \sim0.005m_0 (m0m_0 is the free electron mass), consistent with massless Dirac fermions in monolayer graphene (Supplemental Material Fig. S7 [22]) [23, 24].

Overall, the hallmarks of gate-tunable correlated insulators at low temperature are highly reproducible (see Supplemental Material Fig. S1 [22]) and consistent with prior experimental results [15, 17, 25, 26]. Intriguingly, we also observe counterintuitive resistance behaviors at high temperatures. Figure 2(a) presents the temperature-dependent longitudinal resistance (RxxR_{xx}-TT) curve at D=0.9D = 0.9 V nm−1^{-1} and ν=3\nu= 3. In the high temperature regime T>54T > 54 K, the RxxR_{xx}-TT curve exhibits a logarithmic temperature dependence, a hallmark signature of spin-flip scattering via Kondo effect [27]. At an intermediate temperature T≈54T \approx54 K, the resistance reaches a maximum, signaling the onset of coherence, where localized moments start to be screened by itinerant cc-electrons. Below this temperature, the resistance displays typical metallic behavior and transition into a Fermi liquid regime described by Rxx=R0+AT2R_{xx} = R_0 + AT^2 at T<10T < 10 K, where R0R_0 is the residual resistance, and A0.5A^{0.5} is linearly proportional to the quasiparticle effective mass m∗m^* according to Kadowaki-Woods scaling [28, 29]. For comparison, we also measured the RxxR_{xx}-TT curve obtained at D=0.9D = 0.9 V nm−1^{-1} and ν=5.5\nu= 5.5 [Fig. 2(c)], where only the Dirac bands and higher-energy dispersive bands are involved. As expected, the RxxR_{xx}-TT curve exhibits pure Fermi liquid behavior with an extracted coefficient AA to be 0.020.02 Ω\Omega/K2^2. The corresponding m∗m^* was determined by the temperature dependence of SdH oscillations, yielding a small value of 0.067m00.067m_0 (see Supplemental Material Fig. S8 [22]). Accordingly, we can estimate that the effective mass for ν=3\nu= 3, D=0.9D = 0.9 V nm−1^{-1} is enhanced by 33 times, yielding m∗∼2m0m^* \sim2m_0—consistent with previous results in graphene moiré superlattices [4, 15]. The standard RxxR_{xx}-TT curve and dramatic enhancement of effective mass strongly indicate the emergence of a heavy fermion state for ν=3\nu= 3 at high displacement field. We also compare the temperature dependent Hall resistance RxyR_{xy}-TT at ν=3\nu= 3 and ν=5.5\nu= 5.5 for D=0.9D = 0.9 V nm−1^{-1}. While RxyR_{xy}-TT curve for ν=5.5\nu= 5.5 is nearly temperature-independent [as shown in Fig. 2(d)], the corresponding curve for ν=3\nu= 3 exhibits characteristic heavy fermion behavior: RxyR_{xy} reaches a maximum at 77 K, changes sign at 45 K, passes through a minimum at 13 K, and then increases gradually with the expansion of Fermi surface at low temperature [30, 31].

Image not extracted

Figure 2. (a) RxxR_{xx} versus temperature at D=0.9 V nm−1D = 0.9\ \mathrm{V}\ \mathrm{nm}^{-1} and ν=3\nu= 3. The blue line represents the ln⁡T\ln T fit at high temperature regime, while the red line corresponds the T2T^2 fit at low temperature regime. (b) RxyR_{xy}-TT curves at ν=3\nu= 3 for D=0.9 V nm−1D = 0.9\ \mathrm{V}\ \mathrm{nm}^{-1}. (c),(d) RxxR_{xx}-TT (c) and RxyR_{xy}-TT (d) curves obtained beyond the flat bands at D=0.9 V nm−1D = 0.9\ \mathrm{V}\ \mathrm{nm}^{-1} and ν=5.5\nu= 5.5. Inset: The linear dependence of σAHE\sigma_{\mathrm{AHE}} versus σxx\sigma_{xx} for D=0 V nm−1D = 0\ \mathrm{V}\ \mathrm{nm}^{-1} and ν=3\nu= 3. (e) RxxR_{xx}-TT curves at ν=3\nu= 3 for various displacement fields. (f) RxxR_{xx}-TT curves at ν=3\nu= 3 below 20 K with small displacement field ranging from D=0 V nm−1D = 0\ \mathrm{V}\ \mathrm{nm}^{-1} to D=0.3 V nm−1D = 0.3\ \mathrm{V}\ \mathrm{nm}^{-1}. (g) Magnetic-field-dependent RxyR_{xy} for D=0 V nm−1D = 0\ \mathrm{V}\ \mathrm{nm}^{-1} and ν=3\nu= 3 at various temperatures. (h) Temperature dependence of the inverse magnetic susceptibility for D=0 V nm−1D = 0\ \mathrm{V}\ \mathrm{nm}^{-1} and ν=3\nu= 3.

Having established the presence of heavy fermion, we now investigate the evolution of Kondo hybridization with displacement field. Figure 2(e) and 2(f) display RxxR_{xx}-TT curves at ν=3\nu= 3 with displacement fields ranging from 0 V nm−10\ \mathrm{V}\ \mathrm{nm}^{-1} to 1.1 V nm−11.1\ \mathrm{V}\ \mathrm{nm}^{-1}. As shown in Figure 2(f), the RxxR_{xx}-TT curve displays semimetallic behavior with an abrupt resistance drop at low temperature, which we attribute to the suppression of spin-disorder scattering [32]. By analyzing the magnetic field dependence of both RxxR_{xx} and RxyR_{xy} [Figure 3(c) and Figure 2(e), respectively], we observe clear signatures of spin-flop transition magnetoresistance (MR) [32, 33, 34, 35] and a giant anomalous Hall effects (AHE) [35]. From the AHE measurements in Figure 2(g), we can extract the magnetic susceptibility as:

Image not extracted

Figure 3. (a) DD-tunable heavy fermion phase diagram. The Kondo screening temperature T∗T^* and Néel temperature TNT_{\mathrm{N}} are determined from dRxx/dTdR_{xx}/dT. The onset of heavy Fermi liquid, THFLT_{\mathrm{HFL}}, is identified when the RxxR_{xx}-TT curve deviates from the Fermi-liquid T2T^2 dependence [22]. (b) The fitted AA coefficient of the T2T^2 term (red circles) and effective mass m∗m^* (blue diamonds) extracted from SdH oscillations at low-temperature regime as a function of DD. Inset: Schematic illustration of the large Fermi surface (right) corresponding to the Kondo singlet ground state and the small Fermi surface (left) associated with the incompletely screened semimetal. (c), (d) Magnetic-field-dependent RxxR_{xx} (c) and RxyR_{xy} (d) for discrete DD. The black and red arrows in (c) indicate the occurrence of spin-flop at 1.8 T and 5.1 T.

χ≈∂Rxy∂H∣H→0.\chi\approx\left.\frac{\partial R_{xy}}{\partial H}\right|_{H \to0}.

As shown in Figure 2(h), the temperature dependence of 1/χ1/\chi follows the Curie-Weiss law:

χ=CT−θC.\chi= \frac{C}{T-\theta_C}.

The linear fit yields a negative Curie-Weiss temperature, θC=−115 K\theta_C = -115\ \mathrm{K}, indicating predominant antiferromagnetic interactions [36]. The Néel temperature shown in Figure 2(h) (TN≈15 KT_N \approx15\ \mathrm{K}), identified from the deviation from linearity in 1/χ1/\chi, aligns well with the TN=14.2 KT_N = 14.2\ \mathrm{K} determined from the minimum in the derivative dRxx/dTdR_{xx}/dT-TT at same D=0 V nm−1D = 0\ \mathrm{V}\ \mathrm{nm}^{-1} (Fig. S12). The yielded ∣θC∣|\theta_C| much larger than TNT_N indicates the competition between the RKKY interaction and the Kondo effect, which frequently occurs in heavy fermion systems [37, 38]. These features provide strong evidence for the emergence of an antiferromagnetic order at low displacement fields. The persistence of AHE up to 100 K is unlikely to originate from an intrinsic large Berry curvature [39, 40]. The linear relationship between σAHE\sigma_{\mathrm{AHE}} versus σxx\sigma_{xx}, as shown in the inset of Figure 2(d), suggests an extrinsic skew-scattering origin, where the cc-electrons are asymmetrically scattered by localized moments [41].

By normalizing the RxxR_{xx}-TT curve, we present the first 2D heavy fermion phase diagram in MATBG as shown in Figure 3(a), featuring an antiferromagnetic semimetal and a paramagnetic heavy fermion state. Starting at D=0 V nm−1D = 0\ \mathrm{V}\ \mathrm{nm}^{-1}, the antiferromagnetic order is progressively suppressed by the displacement field, with the Néel temperature TNT_N shifts toward zero. Beyond D=0.4 V nm−1D = 0.4\ \mathrm{V}\ \mathrm{nm}^{-1}, a new characteristic temperature T∗T^*, signaling the onset of Kondo screening, rises proportionally with the displacement field, as does the heavy Fermi liquid temperature THFLT_{\mathrm{HFL}}. This tendency is consistently reproduced in device D2 (see Supplemental Material Fig. S11 [22]). Intriguingly, as shown in Figure 3(b), the fitted coefficients AA for discrete displacement fields DD exhibit a divergent behavior upon approaching the critical displacement field D≈0.5 V nm−1D \approx0.5\ \mathrm{V\,nm}^{-1}.

The continuous evolution from a semimetal to a heavy fermion metal, accompanied by the divergence of the AA coefficients and effective mass m∗m^* (see Supplemental Material Fig. S7 [22]), suggests the presence of a quantum critical point (QCP) at the zero-temperature limit [42]. In this scenario, the Fermi surface transitions from a small to a large configuration, as illustrated in the inset of Figure 3(b). In the low-displacement-field AFM semimetal phase, static Kondo screening is absent. This means the ground state is not a Kondo singlet, and fully developed Kondo resonances do not form. Consequently, only itinerant Dirac electrons contribute to the Fermi volume, resulting in a small Fermi surface. In contrast, in the high-displacement-field heavy fermion liquid phase, the Kondo singlet ground state gives rise to pronounced Kondo resonances, which must be incorporated into the Fermi volume. This leads to a correspondingly large Fermi surface.

To unveil this crossover, we measured the magnetic-field dependence of RxxR_{xx} and RxyR_{xy} across various value of DD at T=2T = 2 K [Figure 3(c) and Figure 3(d)]. In the weak-coupling regime (D<0.3 V nm−1D < 0.3\ \mathrm{V\,nm}^{-1}), the MR exhibits prominent SdH oscillations at low magnetic fields (μ0H<1\mu_0H < 1 T). These oscillations are suppressed as the displacement field approaches D≈0.3 V nm−1D \approx0.3\ \mathrm{V\,nm}^{-1}, consistent with the observed Landau fan behavior (Supplemental Material Fig. S3 [22]). Owing to the incoherent Kondo scattering, localized moments do not contribute to the Fermi-surface formation, ultimately leading to magnetic order below TNT_{\mathrm{N}}. As DD increases, the localized moments hybridize with cc-electrons through the Kondo effect, forming composite fermions with a heavily renormalized effective mass. The resulting mass enhancement suppresses SdH oscillations. The composite fermions incorporate into the Fermi volume, driving Fermi surface expansion. The sign reversal of the low-field Hall coefficient RHR_{\mathrm{H}} further corroborates this transition, signaling a change in the dominant charge carriers from cc-electrons to positively charged Kondo singlets.

Although the band structure and Fermi surface evolution in a single-gate device can be directly imaged with a quantum twisting microscope [43], our dual-gated configuration necessitated a different approach. We therefore employed well-established indirect transport measurements to probe the Fermi surface changes comprehensively. Figure 4(f) shows the fast Fourier transform of Rxx(1/μ0H)R_{xx}(1/\mu_0H) of two representative displacement field, D=0 V nm−1D = 0\ \mathrm{V\,nm}^{-1} and D=0.77 V nm−1D = 0.77\ \mathrm{V\,nm}^{-1}, respectively. The oscillation frequencies reveal that the Fermi surface expands significantly as DD increases [44]. Additionally, the effective mass undergoes a tenfold enhancement (Supplemental Material Fig. S10 [22]).

Image not extracted

Figure 4. (a) Hall carrier density nHn_{\mathrm{H}} and longitudinal resistance RxxR_{xx} as a function of DD at ν=3\nu= 3. (b) Schematic band structures near the critical displacement field. (c) RxxR_{xx} as a function of DD and μ0H\mu_{0}H, taken at ν=3\nu= 3 and T=0.25T = 0.25 K. The gray lines denote MIOs emanating from critical displacement field Dc1=0.31D_{\mathrm{c1}} = 0.31 V nm−1^{-1}, red dashed line marks a representative heavy fermion oscillation at D=0.77D = 0.77 V nm−1^{-1}. (d) Fast Fourier transform (FFT) of Landau fan, calculated from the data in (c). (e) FFT spectra of the SdH oscillation for D=0D = 0 V nm−1^{-1} and D=0.77D = 0.77 V nm−1^{-1}. Inset: MR under in-plane and out-of-plane magnetic field at ν=3\nu= 3 and D=0.77D = 0.77 V nm−1^{-1} at 0.25 K.

Furthermore, we measure the SdH oscillations by continuously scanning displacement at fixed magnetic fields [45, 46, 47, 48]. Figure 4(c) displays RxxR_{xx} at ν=3\nu= 3 as a function of μ0H\mu_0H and DD, measured in device D2 at T=0.25T = 0.25 K. In the (DD-μ0H\mu_0H-RxxR_{xx}) phase space at low displacement field (∣D∣<0.2 V nm−1|D| < 0.2\ \mathrm{V\,nm}^{-1}) and magnetic field (μ0H<2\mu_0H < 2 T), periodic low-resistance states (blue regions separated by white regions) emerge and are gradually suppressed as DD increased. Analysis of the SdH oscillations confirms that this periodicity arises from Dirac band cc-electrons (Supplemental Material Fig. S10 [22]). At higher magnetic field (μ0H>2\mu_0H > 2 T), a dense set of resistance oscillations emerge, which are originated from the scattering of charge carrier between minivalleys of flat and Dirac bands [46, 49]. Their trajectories, along with oscillations originating from higher DD side, converge at Dc1=0.31 V nm−1D_{\mathrm{c1}} = 0.31\ \mathrm{V\,nm}^{-1} in the zero magnetic field limit [see the corresponding schematic in Figure 4(d)], signifying energy equilibrium between minivalleys at this critical displacement field. Similar magneto interminivalley oscillations (MIOs) can be observed at ν=2\nu= 2 (Supplemental Material Fig. S13 [22]). At the vicinity of Dc2=0.43 V nm−1D_{\mathrm{c2}} = 0.43\ \mathrm{V\,nm}^{-1}, we also observe a pronounced resistance peak and a divergence in Hall carrier density nHn_{\mathrm{H}}, accompanied by a sign reversal Figure 4(a), which are associated with Lifshitz transition in the Fermi surface topology.

Crucially, by tracking the SdH frequency as a function of DD, we observe a clear Fermi surface reconstruction. In the weak-coupling regime, the SdH frequency shifts from 1.1 T to zero gradually. Near Dc2=0.43D_{\mathrm{c2}} = 0.43 V nm−1^{-1}, however, the fermiology remains unclear due to the absence of well-defined SdH oscillations. The emergence of non-Fermi liquid behavior and nonlinear II-VV transport characteristics (Supplemental Material Fig. S11 [22]) suggests the possible absence of a well-defined Fermi surface in this regime, likely due to strong quantum fluctuations. Beyond the Dc2=0.43D_{\mathrm{c2}} = 0.43 V nm−1^{-1}, a new frequency emerges at 11.5 T, reflecting a dramatic reconstruction of the Fermi surface driven by the strong Kondo effect between ff- and cc-electrons. The observed Fermi surface expansion, mass renormalization and divergence of the AA coefficients provide strong evidence of a QCP in the MATTG heavy fermion system [42, 44].

To better understand the fermiology evolution across different displacement field regimes, a schematic illustration is provided in Figure 4(b). In MATTG, the displacement field modifies the band structure in two primary ways: (i) hybridizing the flat and Dirac bands at K/K′K/K^{\prime} points, and (ii) shifting the Dirac bands energetically [50, 51]. At ν=3\nu= 3, the system remains semimetallic in the weak coupling regime (D<Dc1D < D_{\mathrm{c1}}) due to the presence of Dirac bands. The Dirac bands shifted upward with displacement field increasing. At critical field Dc1=0.31D_{\mathrm{c1}} = 0.31 V nm−1^{-1}, the Fermi level EfE_{\mathrm{f}} aligns with the Dirac point, as evidenced by the convergence of MIOs in the zero-field limit Figure 4(d). Upon further increasing displacement field beyond Dc2=0.43D_{\mathrm{c2}} = 0.43 V nm−1^{-1}, the flat bands and Dirac bands undergo strong hybridization, leading to formation of new quasiparticle bands with ultra-flattened dispersions near EfE_{\mathrm{f}}. This reconstruction of the electronic structure results in significant Fermi surface expansion, marking the onset of heavy Fermi liquid behavior at high displacement fields.

To summarize the evolution of versatile quantum states with displacement fields, Figure 5 maps out a fully gate-tunable heavy fermion phase diagram in MATTG extracted from our experimental results. Crucially, in-situ electric field tuning provides unprecedented resolution to access the QCP, revealing abrupt Fermi surface reconstruction. Our findings establish a novel 2D Doniach phase diagram, distinct from conventional pressure-, doping-, or field-tuned heavy fermion systems. Similar phase diagrams, including heavy fermion superconductivity [52, 53], are expected in other 2D moiré systems with coexisting flat and dispersive bands.

Image not extracted

Figure 5. Phase diagram for heavy fermions in MATTG at ν=3\nu= 3. The phase boundaries are schematized according to three characteristic temperatures: T∗T^{*}, TNT_{\mathrm{N}}, and THFLT_{\mathrm{HFL}}.

We thank Dr. Chao Zhang from the Instrumentation and Service Center for Physical Sciences (ISCPS) at Westlake University for technical support in data acquisition. This work was funded by National Natural Science Foundation of China (Grant No. 12550402, Grant No. 12274354, Grant No. 12574203, Grants No. 12474136 and No. 12174429), the National Key R&D Program of China (Grant No. 2022YFA1402203), the Zhejiang Provincial Natural Science Foundation of China (Grant No. LR24A040003, XHD23A2001), Hangzhou Natural Science Foundation for Key Program (Grant No. 2025SZRJJ1093) and Westlake Education Foundation at Westlake University. K.W. and T.T. acknowledge support from the JSPS KAKENHI (Grant Numbers 21H05233 and 23H02052) and World Premier International Research Center Initiative (WPI), MEXT, Japan. Z.Z. acknowledges support from the European Union’s Horizon 2020 research and innovation programme under the Marie-Sklodowska Curie grant agreement No 101034431.

Data Availability

The data shown in the main figures and other findings that support this study are available from the corresponding authors upon reasonable request.

End Matter

In this work, we also demonstrate that the observed isospin Pomeranchuk effect arises from the Kondo screening of localized moments by itinerant cc-electrons, resulting in a heavy Fermi liquid ground state at low temperatures. Figure 6(a) and (b) show RxxR_{xx} transfer curves at temperatures ranging from 2 K to 150 K for two representative displacement field D=0.1D = 0.1 V nm−1^{-1} and D=0.9D = 0.9 V nm−1^{-1}, respectively. Additional temperature-dependent transfer curves for various displacement fields are provided in Supplemental Material Fig. S4 to Supplemental Material Fig. S6 [22]. At a small displacement field (D=0.1D = 0.1 V nm−1^{-1}), resistance peaks emerge at ν=1,2,3\nu= 1,2,3 upon cooling, attributable to the breaking of spin/valley degeneracy due to Coulomb interactions among localized moments [54, 55, 56]. In contrast, at ν=3\nu= 3 and larger displacement field D=0.9D = 0.9 V nm−1^{-1}, we observe a completely opposite transport behavior: resistance peaks associated with correlated phase develop with increasing temperature and merge into a broad resistance peak around ν=0\nu= 0 at temperatures above 54 K. This suggests that localized moments become detectable at higher temperature. The unusual behavior is identical to the recently confirmed isospin Pomeranchuk effect in MATBG [57, 58], rhombohedral trilayer graphene superlattices [59], and trilayer-MoTe2_2/WSe2_2 moiré superlattices [60]. The Pomeranchuk effect, originally observed in the isotope 3^3He, describes an entropy-driven liquid-to-solid transition upon heating, where the nuclear spin of atoms in the paramagnetic solid phase are disordered, leading to higher entropy compared to the liquid phase. In 2D moiré superlattices, the localized moments exhibit an analogous effect: a low-entropy Fermi liquid phase emerges at low temperatures, while a high-entropy, broadly isospin-polarized insulating phase dominates at high temperatures. In general, systems tend to maximize entropy at high temperatures by sampling all possible configurations. In heavy fermion materials, this is achieved by gaining free energy through the liberation of local moments, effectively rendering them free [61].

Image not extracted

Figure 6. Longitudinal resistance RxxR_{xx} versus ν\nu measured at a small displacement field D=0.1 V nm−1D=0.1\ \mathrm{V\,nm^{-1}} (a) and a large displacement field D=0.9 V nm−1D=0.9\ \mathrm{V\,nm^{-1}} (b) across successive temperatures.

As shown in Figure 7(a), we measured the (ν,D)(\nu, D) phase diagram of D2 at a temperature of T=0.25T = 0.25 K. Compared to the measurement at T=2T = 2 K (see Supplemental Material Fig. S1 [22]), two prominent superconducting regions with zero resistance emerge at ν=−2−δ\nu= -2 - \delta and ν=2+δ\nu=2+\delta (0<δ<10<\delta<1), which is consistent with previous reports. Hall measurements indicate that superconductivity is closely linked to flavor polarization at integer filling and to van Hove singularity at non-integer filling.

Image not extracted

Figure 7. (a) RxxR_{xx} as a function of ν\nu and DD at T=0.25 KT=0.25\ \mathrm{K}. (b) Anti-symmetrized Hall carrier density nHn_{\mathrm{H}} as a function of ν\nu at several DD values, corresponding to the horizontal arrows marked in (a). (c) RxxR_{xx} as a function of ν\nu at D=0.36 V nm−1D=0.36\ \mathrm{V\,nm^{-1}}, measured at temperatures ranging from 0.25 K0.25\ \mathrm{K} to 4 K4\ \mathrm{K}. (d), (e) Critical current as a function of out-of-plane (d) and in-plane (e) magnetic field at D=0.5 V nm−1D=0.5\ \mathrm{V\,nm^{-1}} and ν=2.3\nu=2.3 [marked by the orange dot in (a)]. (f) Temperature-dependent differential resistance dV/dI\mathrm{d}V/\mathrm{d}I at D=0.5 V nm−1D=0.5\ \mathrm{V\,nm^{-1}} and ν=2.3\nu=2.3. (g) dV/dI\mathrm{d}V/\mathrm{d}I versus d.c. current on a log-log scale. The dashed line corresponds to a V∝Idc3V\propto I_{\mathrm{dc}}^{3} power law dependence, indicating TBKT≈1.3 KT_{\mathrm{BKT}}\approx1.3\ \mathrm{K}.

Superconductivity is closely linked to flavor polarization at integer filling and to van Hove singularity at non-integer filling. For instance, at D=0.36 V nm−1D=0.36\ \mathrm{V\,nm^{-1}}, the nHn_{\mathrm{H}} increases linearly with filling factor until a charge carrier reset occurs at ν=1\nu=1. Following this flavor symmetry-breaking, nHn_{\mathrm{H}} exhibits divergent behavior at ν=1.8\nu=1.8, which corresponds to VHS. Superconductivity emerges after the correlated insulator gap at integer filling ν=2\nu=2 but vanishes at integer filling ν=3\nu=3 upon charge carrier resetting. Under higher displacement field, the superconductivity at ν=2+δ\nu=2+\delta is suppressed at the VHS. We also observe signature of superconductivity at ν=2−δ\nu=2-\delta, which is similarly bounded by the VHS. One possible explanation for the incipient superconductivity is that pairing might occur at temperatures higher than those at which the zero-resistance state is detected. In conventional electron-phonon weak-coupling Bardeen-Cooper-Schrieffer theory, the density of state (DOS) peaks near a VHS, typically enhancing superconductivity. Together with prior work, our observations suggest an unconventional origin of the superconductivity in MATTG. In addition, we find a large in-plane upper critical magnetic field for the superconducting phase at optimal ν\nu and DD, which exceeds the Pauli limit. This finding may stimulate further investigations into the paring mechanism of graphene-based superconductors, such as spin-triplet paring.

References

  1. [1]J. G. Checkelsky, B. A. Bernevig, P. Coleman, Q. M. Si, and S. Paschen, Flat bands, strange metals and the Kondo effect, Nat. Rev. Mater. 9, 509 (2024).
  2. [2]Y. Cao, V. Fatemi, S. Fang, K. Watanabe, T. Taniguchi, E. Kaxiras, and P. Jarillo-Herrero, Unconventional superconductivity in magic-angle graphene superlattices, Nature 556, 43 (2018).DOI
  3. [3]X. Lu, P. Stepanov, W. Yang, M. Xie, M. A. Aamir, I. Das, C. Urgell, K. Watanabe, T. Taniguchi, G. Zhang, A. Bachtold, A. H. MacDonald, and D. K. Efetov, Superconductors, orbital magnets and correlated states in magic-angle bilayer graphene, Nature 574, 653 (2019).
  4. [4]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).
  5. [5]Y.-Z. Chou and S. Das Sarma, Kondo Lattice Model in Magic-Angle Twisted Bilayer Graphene, Phys. Rev. Lett. 131, 026501 (2023).
  6. [6]A. Kumar, N. C. Hu, A. H. MacDonald, and A. C. Potter, Gate-tunable heavy fermion quantum criticality in a moiré Kondo lattice, Phys. Rev. B 106, 041116 (2022).
  7. [7]D. Guercı, J. Wang, J. Zang, J. Cano, J. H. Pixley, and A. Millis, Chiral Kondo lattice in doped MoTe₂/WSe₂ bilayers, Sci. Adv. 9, eade7701 (2023).
  8. [8]A. Ghosh, S. Chakraborty, R. Dutta, A. Agarwala, K. Watanabe, T. Taniguchi, S. Banerjee, N. Trivedi, S. Mukerjee, and A. Das, Thermopower probes of emergent local moments in magic-angle twisted bilayer graphene, Nat. Phys. 21, 732 (2025).
  9. [9]R. L. Merino, D. Călugăru, H. Hu, J. Díez-Mérida, A. Díez-Carlón, T. Taniguchi, K. Watanabe, P. Seifert, B. A. Bernevig, and D. K. Efetov, Interplay between light and heavy electron bands in magic-angle twisted bilayer graphene, Nat. Phys. 21, 1078 (2025).
  10. [10]A. T. Pierce, Y. Xie, J. M. Park, Z. Cai, K. Watanabe, T. Taniguchi, P. Jarillo-Herrero, and A. Yacoby, Tunable interplay between light and heavy electrons in twisted trilayer graphene, Nat. Phys. 21, 1237 (2025).
  11. [11]Z. D. Song and B. A. Bernevig, Magic-Angle Twisted Bilayer Graphene as a Topological Heavy Fermion Problem, Phys. Rev. Lett. 129, 047601 (2022).
  12. [12]G.-D. Zhou, Y.-J. Wang, N. Tong, and Z.-D. Song, Kondo phase in twisted bilayer graphene, Phys. Rev. B 109, 045419 (2024).DOI
  13. [13]H. Hu, B. A. Bernevig, and A. M. Tsvelik, Kondo Lattice Model for Magic-Angle Twisted-Bilayer Graphene: Hund’s Rule, Local-Moment Fluctuations, and Low-Energy Effective Theory, Phys. Rev. Lett. 131, 026502 (2023).
  14. [14]L. L. H. Lau and P. Coleman, Topological Mixed Valence Model for Twisted Bilayer Graphene, Phys. Rev. X 15, 041015 (2025).
  15. [15]J. M. Park, Y. Cao, K. Watanabe, T. Taniguchi, and P. Jarillo-Herrero, Tunable strongly coupled superconductivity in magic-angle twisted trilayer graphene, Nature 590, 249 (2021).
  16. [16]J.-X. Lin, P. Siriviboon, H. D. Scammell, S. Liu, D. Rhodes, K. Watanabe, T. Taniguchi, J. Hone, M. S. Scheurer, and J. L. A. Li, Zero-field superconducting diode effect in small-twist-angle trilayer graphene, Nat. Phys. 18, 1221 (2022).
  17. [17]Z. Hao, A. M. Zimmerman, P. Ledwith, E. Khalaf, D. H. Najafabadi, K. Watanabe, T. Taniguchi, A. Vishwanath, and P. Kim, Electric field-tunable superconductivity in alternating-twist magic-angle trilayer graphene, Science 371, 1133 (2021).
  18. [18]W. Zhao, B. Shen, Z. Tao, Z. Han, K. Kang, K. Watanabe, T. Taniguchi, K. F. Mak, and J. Shan, Gate-tunable heavy fermions in a moiré Kondo lattice, Nature 616, 61 (2023).
  19. [19]W. J. Zhao, B. W. Shen, Z. Tao, S. Kim, P. Knüppel, Z. D. Han, Y. C. Zhang, K. Watanabe, T. Taniguchi, D. Chowdhury, J. Shan, and K. F. Mak, Emergence of ferromagnetism at the onset of moiré Kondo breakdown, Nat. Phys. 20, 1772 (2024).
  20. [20]A. Ramires and J. L. Lado, Emulating Heavy Fermions in Twisted Trilayer Graphene, Phys. Rev. Lett. 127, 026401 (2021).
  21. [21]J. B. Yu, M. Xie, B. A. Bernevig, and S. Das Sarma, Magic-angle twisted symmetric trilayer graphene as a topological heavy-fermion problem, Phys. Rev. B 108, 035129 (2023).
  22. [22]See Supplemental Material at http://link.aps.org/supplemental/10.1103/zzks-vk12 for details on the device fabrications, the band structure calculations, and additional data analysis, which includes Refs. [62–72].
  23. [23]K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, M. I. Katsnelson, I. V. Grigorieva, S. V. Dubonos, and A. A. Firsov, Two-dimensional gas of massless Dirac fermions in graphene, Nature 438, 197 (2005).
  24. [24]Y. Zhang, Y. W. Tan, H. L. Stormer, and P. Kim, Experimental observation of the quantum Hall effect and Berry’s phase in graphene, Nature 438, 201 (2005).
  25. [25]C. Shen, P. J. Ledwith, K. Watanabe, T. Taniguchi, E. Khalaf, A. Vishwanath, and D. K. Efetov, Dirac spectroscopy of strongly correlated phases in twisted trilayer graphene, Nat. Mater. 22, 316 (2023).DOI
  26. [26]X. X. Liu, N. J. Zhang, K. Watanabe, T. Taniguchi, and J. I. A. Li, Isospin order in superconducting magic-angle twisted trilayer graphene, Nat. Phys. 18, 522 (2022).DOI
  27. [27]J.-H. Chen, L. Li, W. G. Cullen, E. D. Williams, and M. S. Fuhrer, Tunable Kondo effect in graphene with defects, Nat. Phys. 7, 535 (2011).
  28. [28]A. C. Jacko, J. O. Fjaerestad, and B. J. Powell, A unified explanation of the Kadowaki–Woods ratio in strongly correlated metals, Nat. Phys. 5, 422 (2009).DOI
  29. [29]K. Kadowaki and S. B. Woods, Universal Relationship of the Resistivity and Specific-Heat in Heavy-Fermion Compounds, Solid State Commun. 58, 507 (1986).DOI
  30. [30]S. Paschen, T. Luhmann, S. Wirth, O. Trovarelli, C. Geibel, and F. Steglich, Anomalous hall effect in YbRh₂Si₂, Physica B 359, 44 (2005).
  31. [31]A. Fert and P. M. Levy, Theory of the Hall effect in heavy-fermion compounds, Phys. Rev. B 36, 1907 (1987).pubmed.ncbi.nlm.nih.gov/9943035
  32. [32]Y. Luo, F. Ronning, N. Wakeham, X. Lu, T. Park, Z. A. Xu, and J. D. Thompson, Pressure-tuned quantum criticality in the antiferromagnetic Kondo semimetal CeNi₂−δAs₂, Proc. Natl. Acad. Sci. U.S.A. 112, 13520 (2015).DOI
  33. [33]K. Torres, J. Y. Park, V. A. Posey, M. E. Ziebel, C. E. Casaday, K. J. Anderton, D. Cui, B. Tang, T. Taniguchi, K. Watanabe, A. N. Pasupathy, X. Roy, and P. Kim, Glassy relaxation dynamics in the two-dimensional heavy fermion antiferromagnet CeSi, Nano Lett. 25, 6848 (2025).DOI
  34. [34]V. A. Posey, S. Turkel, M. Rezaee, A. Devarakonda, A. K. Kundu, C. S. Ong, M. Thinel, D. G. Chica, R. A. Vitalone, R. Jing, S. Xu, D. R. Needell, E. Meirzadeh, M. L. Feuer, A. Jindal, X. Cui, T. Valla, P. Thunstrom, T. Yilmaz, E. Vescovo, D. Graf, X. Zhu, A. Scheie, A. F. May, O. Eriksson, D. N. Basov, C. R. Dean, A. Rubio, P. Kim, M. E. Ziebel, A. J. Millis, A. N. Pasupathy, and X. Roy, Two-dimensional heavy fermions in the van der Waals metal CeSi, Nature 625, 483 (2024).DOI
  35. [35]Y. Guo, J. Pack, J. C. Swann, L. Holtzman, M. Cothrine, K. Watanabe, T. Taniguchi, D. G. Mandrus, K. Barmak, J. Hone, A. J. Millis, A. Pasupathy, and C. R. Dean, Superconductivity in 5.0° twisted bilayer WSe₂, Nature 637, 839 (2025).DOI
  36. [36]E. Anderson, F.-R. Fan, J. Cai, W. Holtzmann, T. Taniguchi, K. Watanabe, D. Xiao, W. Yao, and X. Xu, Programming correlated magnetic states with gate-controlled moiré geometry, Science 381, 325 (2023).
  37. [37]O. Trovarelli, C. Geibel, S. Mederle, C. Langhammer, F. M. Grosche, P. Gegenwart, M. Lang, G. Sparn, and F. Steglich, YbRh₂Si₂: pronounced non-Fermi-liquid effects above a low-lying magnetic phase transition, Phys. Rev. Lett. 85, 626 (2000).
  38. [38]P. Gegenwart, J. Custers, C. Geibel, K. Neumaier, T. Tayama, K. Tenya, O. Trovarelli, and F. Steglich, Magnetic-field induced quantum critical point in YbRh₂Si₂, Phys. Rev. Lett. 89, 056402 (2002).pubmed.ncbi.nlm.nih.gov/12144454
  39. [39]H. Polshyn, J. Zhu, M. A. Kumar, Y. Zhang, F. Yang, C. L. Tshirhart, M. Serlin, K. Watanabe, T. Taniguchi, A. H. MacDonald, and A. F. Young, Electrical switching of magnetic order in an orbital Chern insulator, Nature 588, 66 (2020).
  40. [40]D. Xiao, W. Yao, and Q. Niu, Valley-contrasting physics in graphene: magnetic moment and topological transport, Phys. Rev. Lett. 99, 236809 (2007).
  41. [41]S. Nair, S. Wirth, S. Friedemann, F. Steglich, Q. Si, and A. J. Schofield, Hall effect in heavy fermion metals, Adv. Phys. 61, 583 (2012).
  42. [42]Q. Si and F. Steglich, Heavy fermions and quantum phase transitions, Science 329, 1161 (2010).DOI
  43. [43]J. Xiao, A. Inbar, J. Birkbeck, N. Gershon, Y. Zamir, Y. Viturri, T. Taniguchi, K. Watanabe, E. Berg, and S. Ilani, Imaging the flat bands of magic-angle graphene reshaped by interactions, Nature 653, 68 (2026).DOI
  44. [44]H. Shishido, R. Settai, H. Harima, and Y. Ōnuki, A drastic change of the Fermi surface at a critical pressure in CeRhIn₅: dHvA study under pressure, J. Phys. Soc. Jpn. 74, 1103 (2005).DOI
  45. [45]I. Y. Phinney, A. Zimmerman, Z. Hao, P. J. Ledwith, T. Taniguchi, K. Watanabe, A. Vishwanath, and P. Kim, Modulation of superconductivity across a Lifshitz transition in alternating-angle twisted quadrilayer graphene, Phys. Rev. Lett. 136, 086501 (2026).arxiv.org/abs/2502.16700
  46. [46]P. Tomic, P. Rickhaus, A. Garcia-Ruiz, G. Zheng, E. Portoles, V. Fal’ko, K. Watanabe, T. Taniguchi, K. Ensslin, T. Ihn, and F. K. de Vries, Scattering between Mini-valleys in Twisted Double Bilayer Graphene, Phys. Rev. Lett. 128, 057702 (2022).
  47. [47]H. Zhou, L. Holleis, Y. Saito, L. Cohen, W. Huynh, C. L. Patterson, F. Yang, T. Taniguchi, K. Watanabe, and A. F. Young, Infospin magnetism and spin-polarized superconductivity in Bernal bilayer graphene, Science 375, 774 (2022).DOI
  48. [48]H. Zhou, T. Xie, A. Ghazaryan, T. Holder, J. R. Ehrets, E. M. Spanton, T. Taniguchi, K. Watanabe, E. Berg, M. Serbyn, and A. F. Young, Half- and quarter-metals in rhombohedral trilayer graphene, Nature 598, 429 (2021).DOI
  49. [49]I. Y. Phinney, D. A. Bandurin, C. Collignon, I. A. Dmitriev, T. Taniguchi, K. Watanabe, and P. Jarillo-Herrero, Strong Interminivalley Scattering in Twisted Bilayer Graphene Revealed by High-Temperature Magneto-Oscillations, Phys. Rev. Lett. 127, 056802 (2021).DOI
  50. [50]S. Carr, C. Li, Z. Zhu, E. Kaxiras, S. Sachdev, and A. Kruchkov, Ultralheav y and Ultrarelativistic Dirac Quasiparticles in Sandwiched Graphenes, Nano Lett. 20, 3030 (2020).
  51. [51]E. Khalaf, A. J. Kruchkov, G. Tarnopolsky, and A. Vishwanath, Magic angle hierarchy in twisted graphene multilayers, Phys. Rev. B 100, 085109 (2019).DOI
  52. [52]Y. Cao, J. M. Park, K. Watanabe, T. Taniguchi, and P. Jarillo-Herrero, Pauli-limit violation and re-entrant superconductivity in moire graphene, Nature 595, 526 (2021).DOI
  53. [53]S. Ran, C. Eckberg, Q. P. Ding, Y. Furukawa, T. Metz, S. R. Saha, I. L. Liu, M. Zic, H. Kim, J. Paglione, and N. P. Butch, Nearly ferromagnetic spin-triplet superconductivity, Science 365, 684 (2019).DOI
  54. [54]J. M. Park, Y. Cao, K. Watanabe, T. Taniguchi, and P. Jarillo-Herrero, Flavour Hund’s coupling, Chern gaps and charge diffusivity in moire graphene, Nature 592, 43 (2021).DOI
  55. [55]D. Wong, K. P. Nuckolls, M. Oh, B. Lian, Y. Xie, S. Jeon, K. Watanabe, T. Taniguchi, B. A. Bernevig, and A. Yazdani, Cascade of electronic transitions in magic-angle twisted bilayer graphene, Nature 582, 198 (2020).
  56. [56]U. Zondiner, A. Rozen, D. Rodan-Legrain, Y. Cao, R. Queiroz, T. Taniguchi, K. Watanabe, Y. Oreg, F. von Oppen, A. Stern, E. Berg, P. Jarillo-Herrero, and S. Ilani, Cascade of phase transitions and Dirac revivals in magic-angle graphene, Nature 582, 203 (2020).
  57. [57]A. Rozen, J. M. Park, U. Zondiner, Y. Cao, D. Rodan-Legrain, T. Taniguchi, K. Watanabe, Y. Oreg, A. Stern, E. Berg, P. Jarillo-Herrero, and S. Ilani, Entropic evidence for a Pomeranchuk effect in magic-angle graphene, Nature 592, 214 (2021).DOI
  58. [58]Y. Saito, F. Yang, J. Ge, X. Liu, T. Taniguchi, K. Watanabe, J. I. A. Li, E. Berg, and A. F. Young, Isospin Pomeranchuk effect in twisted bilayer graphene, Nature 592, 220 (2021).DOI
  59. [59]X. Han, Y. Zou, Q. Liu, Z. Wang, R. Niu, Z. Qu, Z. Li, C. Han, K. Watanabe, T. Taniguchi, B. Dong, Z. Song, J. Mao, Z. Han, Z. G. Cheng, Z. Gan, and J. Lu, Suppression of symmetry-breaking correlated insulators in a rhombohedral trilayer graphene superlattice, Nat. Commun. 15, 9765 (2024).DOI
  60. [60]M. J. Zhang, X. Zhao, K. Watanabe, T. Taniguchi, Z. Zhu, F. C. Wu, Y. Q. Li, and Y. Xu, Tuning Quantum Phase Transitions at Half Filling in 3L-MoTe2/WSe2 Moire Superlattices, Phys. Rev. X 12, 041015 (2022).DOI
  61. [61]P. Gegenwart, Q. Si, and F. Steglich, Quantum criticality in heavy-fermion metals, Nat. Phys. 4, 186 (2008).DOI

Paper details

vlnc:W_bvpg569w
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents