Quantized heat flow in graphene quantum Hall phases: Probing the topological order

In single-layer graphene, carbon atoms are arranged in a honeycomb lattice structure as shown in  1 (a). To describe its lattice vectors, one can think of it as a triangular lattice with two sublattice basis A and B. The unit lattice vectors can be written as

where $a=1.42$ Å is the carbon-carbon bond length. The corresponding reciprocal lattice vectors are given by

The first Brillouin zone (BZ) of the graphene lattice is shown in Fig. 1(b). There are six corners in the first BZ, out of which only two points, known as ‘Dirac points’, $\mathbf{K}$ and $\mathbf{K}^{{}^{\prime}}$ are non-equivalent. The positions of $\mathbf{K}$ and $\mathbf{K}^{{}^{\prime}}$ points in momentum space are given by

SLG’s low energy band structure can be calculated using the tight-binding approach. For the scope of this review, we account for only the first three nearest neighbour sites, which are

The tight binding Hamiltonian accounting only the first three NN can be written in the basis of the wave-function amplitudes on the two sublattices $\Phi_{A}$ and $\Phi_{B}$, as follow;

where $\gamma_{0}=2.8$ eV is the nearest neighbour hopping between site $A\leftrightarrow B$. and

The resulting eigenvalues of the Hamiltonian(7) have two eigenvalues, which become identically zero at $\mathbf{K}$ and $\mathbf{K}^{{}^{\prime}}$ points in the Brillouin zone. In most cases, only the low-energy bands around these points ($\mathbf{K}$ and $\mathbf{K}^{{}^{\prime}}$), known as Dirac points and Valley in the Brillouin zone, are important. The low-energy Hamiltonian around these Dirac points can be written as

where $v_{F}=3\gamma_{0}a/2\hbar\simeq 1\times 10^{6}$ m/s and $\xi=\pm 1$ for $\mathbf{K/K^{{}^{\prime}}}$ point. The eigenvalues of the Hamiltonian(11) around $\mathbf{K}$ point is given by;

The resulting low-energy band dispersion of graphene has a four-fold degeneracy, 2 for the spin flavour and 2 for the Valley flavour. In the presence of a magnetic field perpendicular to the graphene plane $\mathbf{B}=B\hat{\mathbf{z}}$, the band dispersion gets modified; Following the minimal coupling approach of the vector potential $\mathbf{A}=(-By,0,0)$; the Hamiltonian(11) around $\mathbf{K}$ point can be written in terms of the defined operator $\pi\rightarrow(p_{x}-eBy+ip_{y}=-i\hbar\partial_{x}-eBy+\hbar\partial_{y})$ and $\pi^{\dagger}\rightarrow(p_{x}-eBy-ip_{y}=-i\hbar\partial_{x}-eBy-\hbar\partial_{y})$. The resulting Hamiltonian in the magnetic field near the Dirac point $\mathbf{K}$ is given by

Operator $\pi$ and $\pi^{\dagger}$ act as a creation and annihilation operator respectively, such that

Here, $\psi_{N}(x,y)$ is the scalar orbital Landau level wavefunction;

where $l_{B}=\sqrt{\hbar/eB}$ is the magnetic length scale, $N$ is the Landau level index and $H_{N}$ is $N^{th}$ Hermite polynomial. The resulting eigenvalues are given by,

The eigenstates at $\mathbf{K}$ point are given by

whereas near $\mathbf{K^{\prime}}$ point, eigenstates are

It should be noted that for all Landau level $N\geq 1$, the electronic density is shared by both sublattice site $A$ and $B$. However, for $N=0$ state, near $\mathbf{K/K^{\prime}}$ valley, electronic density only resides on the sublattice $A/B$. The resulting Landau level energy spectrum and the density of states are plotted in Fig. 2(b) and (c), respectively. The presence of the zero energy Landau level (ZLL) and four-fold spin valley degeneracy of SLG, the Hall conductivity $\sigma_{xy}$ exhibits an unconventional sequence of the quantization described by[18, 19, 20, 21, 22, 23, 24, 25]

It should be noted that the Landau level’s wavefunction in SLG differs significantly from the conventional \chGaAs/AlGaAs. First, here, for SLG, the wavefunction of the $N^{th}$ Landau level has the orbital component of the $N^{th}$ and $N^{th}-1$ conventional Landau level, except ZLL. This difference can lead to different values of Haldane pseudopotentials for the LLs with different orbital structures and hence affects the sequence of the FQH states in different LLS. The second major difference, which is more pronounced in ZLL is the different sublattice structure of wavefunctions near degenerate $\mathbf{K}$ and $\mathbf{K^{\prime}}$. As a consequence, it affects the valley and or spin symmetry-breaking of ZLL and the ground state of $\nu=0$ insulating state.

Along with the unconventional sequence of the QH plateaus due to the $\pi$ Berry’s phase, the spin and valley degeneracy makes graphene a wonderful system for observing the rich physics associated with the multicomponent quantum Hall phases[26, 27, 23, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38]. In SLG, the relevant energy scales are the Landau quantization energy $\epsilon_{N}$ (cyclotron energy) and the long-range Coulomb interaction $E_{C}=\frac{e^{2}}{\epsilon l_{B}}$. At low magnetic fields, $\epsilon_{N}$ and $E_{C}$ do not depend on the spin or valley degrees of freedom. So the combined spin-valley flavour degeneracy can be described in terms of a single approximate $SU(4)$ isospin symmetry which corresponds to the invariance of long-range electron-electron $(E_{C})$ interaction under a rotation within the fourfold spin/valley internal space. However, the long-range Coulomb interaction $(E_{C})$ can drive the system through a ferromagnetic instability, in which the order parameter corresponds to a finite polarization in a specific direction within the $SU(4)$ isospin space [26, 27, 23, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37]. At the integer fillings within a partially filled quartet Landau levels, this order parameter is predicted to lead to a finite gap for charged excitations and a robust quantum Hall effect for integers outside the sequence described in Eqn. 25. The exact $SU(4)$ polarization for a given sample depends on the interplay between the anisotropies arising from the $SU(4)$ symmetry-breaking term like Zeeman splitting, lattice scale interactions and even the disorder (that might not break the $SU(4)$ symmetries). Fig. 3(c) shows the plot of transverse conductance as a function of the back gate voltage at $B=9.8$ T and $40$ mK of bath temperature for a single layer graphene device encapsulated between the two hexagonal boron nitride substrate (hBN). In addition to the conventional plateaus at $2\frac{e^{2}}{h},6\frac{e^{2}}{h},10\frac{e^{2}}{h}$, well-developed symmetry broken plateaus of $N=0$ and $N=1$ LLs are also seen.

Similar to the IQH case (Fig. 4(a)), the charge density for the particle-like FQH states $(\nu<1/2)$ drops monotonically from its bulk values to zero as one approaches the edge from the bulk[52], as shown in Fig. 4(b, c) for $\nu=1/3,2/5$. It leads to the emergence of the one-dimensional downstream edge modes near the physical boundary of the sample[52], as shown in Fig. 4(e, f) for $\nu=1/3,2/5$. The applied magnetic field’s direction dictates the edge modes’ downstream chirality[53, 54]. However, the charge density profile does not decrease monotonically for the hole-conjugate FQH states $(1/2<\nu<1)$. For example, at $\nu=2/3$, the filling factor increases from $2/3$ to $1$ and then drops back to zero[52]. In other words, $2/3$ liquids can be considered an FQH state of $\nu=1/3$ holes in the IQH state at $\nu=1$ electron liquid. As a consequence, two counter-propagating edge modes of charge $e$ with conductance $e^{2}/h$ and charge $-e/3$ with conductance $e^{2}/3h$ emerge at the sample boundary[52, 55]. Fig. 5(a, b) shows the schematics of charge density profiles and edge states of the $\nu=2/3$ FQH state. However, in the presence of the disorder-dominated random inter-edge tunnelling between, Kane-Fisher-Polchinski[56] found a decoupled charge mode of conductance $2e^{2}/3h$ and a neutral mode, which does not carry any charge but can carry the energy. The schematics of such normalized edge modes are shown in Fig. 5(c, d).

Since the neutral modes do not couple with the charged excitations of the external probes, it becomes challenging to detect them via the conventional electrical conductance measurement. Although several proposals were made to see the neutral modes which involve the measurement of the tunnelling exponent in constrictions[57], looking for resonance in long constriction[58], and looking for the heating effects on shot noise[59, 60], the first experimental proof was demonstrated by Bid et al.;[61] for filling factor $\nu=2/3,3/5,$ and $5/2$ in \chGaAs/AlGaAs sample. This experiment involved an upstream quantum point contact (QPC) constriction from an energized ohmic contact. An excited neutral mode emanates from the ‘hot spot’ at the source, propagates upstream along the edge, and impinges on a partially pinched QPC. This leads to observed current fluctuations. Recently, upstream neutral modes for $\nu=2/3,3/5$ mode have been also observed in bilayer graphene devices[62].

Quantized electrical and thermal Hall conductance in quantum Hall states have been known for a long time back theoretically[63, 64] since their first set of experimental observations. Although electrical Hall conductance has been widely used to understand the topological order of a quantum Hall state, it is insufficient in the hierarchical fractional quantum Hall states, where the edge structure is complicated, and transport may occur via the downstream and upstream modes. The electrical Hall conductance only reveals the number and the conductance of the downstream charged chiral edge modes. Still, it is independent of the edge modes’ total number, chirality, and character. By contrast, the quantized thermal Hall conductance is not only sensitive to the downstream charged modes, it can also detect the other upstream modes, including the chargeless neutral modes, which are not detectable in electrical Hall conductance measurement[63, 64, 65].

Similar to the quantization of electrical conductance, the quantized thermal conductance of a single ballistic channel only depends on the fundamental constants of nature and is given by $\kappa_{0}T=\frac{\pi^{2}k_{B}^{2}}{3h}T$, where $k_{B}$ and $h$ are the Boltzmann’s and Planck’s constant, respectively. If multiple downstream edge modes exist, such as IQH and particle-like FQH states, quantized thermal conductance becomes $G_{Q}=N_{d}\kappa_{0}T$, where $N_{d}$ is the total number of downstream edge modes. However, the situation becomes complex in the inter-mode tunnelling between counter-propagating edge modes at certain fillings, such as for $2/3$ and $3/5$, where some chiral edge modes propagate in the upstream direction. In 1997, Kane and Fisher derived the quantized thermal Hall conductance for the abelian fractional quantum Hall states for an ideal impurity-free edge[64]. For the FQH states with counter-propagating edge modes, their analysis assumes a substantially longer edge propagation length than the thermal equilibration length. They found that quantized thermal Hall conductance at any filling of abelian FQH state is given by[64] $G_{Q}=(N_{d}-N_{u})\kappa_{0}T$, where $N_{d}$ and $N_{u}$ are the numbers of the downstream and upstream edge modes, respectively. However, if the edge propagation length is smaller than the thermal equilibration length $(l_{H})$, thermal Hall conductance takes the value of $G_{Q}=(N_{d}+N_{u})\kappa_{0}T$. Since quantized $G_{Q}$ depends on the $N_{d}$ and $N_{u}$, it can also be used as a powerful experimental tool to detect the upstream neutral modes and hence the exact topological order of complex FQH states, which were not possible in electrical conductance measurements. Furthermore, in contrast to quantized electrical conductance, quantized thermal conductance remains independent of the statistics of the carriers. It remains the same for the fermions, bosons, anyons [64, 66] except in the case of the Majorana mode, where the quantized thermal conductance becomes half of its quantum limit,[67, 68, 69, 70, 71]. It makes the thermal conductance measurement a powerful technique which can be used to distinguish non-abelian order of even-denominator FQH states. Usually, the non-abelian order of FQH state, such as $5/2$ is believed to host the Majorana modes in its edge structure[72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83]. Hence the observation of the half-quantized thermal conductance will be the smoking-gun evidence of the Majorana mode. Experimentally, the half quantum of the thermal conductance has been measured at $5/2$ FQH state in GaAs/AlGaAs based two-dimensional electron gas[84].

Different from electrical conductance measurement, quantized thermal conductance measurement is complicated and tricky. It is evident from the fact that although the quantization of the electrical and thermal conductance of the IQH and FQH states has been known for decades, the first experimental measurement of quantized thermal conductance of quantum Hall states was reported 33 years after the first experimental demonstration of the IQH states in the pioneering work of Jezouin et al. [85] in GaAs/AlGaAs based two-dimensional electron gas. The basic principle of the experimental set-up is shown in Fig. 6(a). In this measurement scheme, a micrometer-sized metallic floating reservoir is connected to the cold electronic reservoir via an adjustable number $n$ of the ballistic QH edge channels and a phonon cold bath at temperature $T_{0}$. The electron temperature of the floating metallic reservoir is heated to a temperature $T_{M}$ by injecting a known joule heating power $J_{Q}$ to it. The temperature of the floating reservoir is determined by a heat balance equation given by;

Here, the first and the second terms in this heat balance equation correspond to the electronic contribution to heat flow via $N$ ballistic edge channel and heat loss due to electron-phonon coupling from the floating reservoir to the phonon cold bath. The central result of Jezouin et al. is shown in Fig. 6(b), where $\alpha_{n}$ measured electronic thermal conductance normalized by $0.5\kappa_{0}$ is plotted as a function of the $n$ number of the ballistic edge channel. The unit slope of the experimental fit data corresponds to the theoretically expected quantized limit on the thermal conductance. Later, a similar measurement was extended to the FQH regime by Banerjee et al. [86, 84] in the same system. Banerjee et al. [86] demonstrated the universality of the quantized thermal conductance for the anyonic heat flow. They further reported the observation of a half-integer of the quantized thermal conductance for even-denominator FQH state $5/2$ [84]. The measured thermal conductance value of $2.5\kappa_{0}T$ was interpreted as the particle–hole Pfaffian (PH-Pfaffian) topological order of $5/2$ state. As mentioned above, the experiments by Jezouin et al. and Banerjee et al. were performed in \chGaAs/AlGaAs based two-dimensional electron gas.

Although The quantization of the electrical Hall conductance in graphene, a van der Waals material was reported in 2005 by A. Geim and Philip Kim group separately[87, 88], the first experiment of quantized thermal conductance in QH states in graphene was reported in 2019[89]. In this experiment, graphene was encapsulated between two hexagonal boron nitride substrates and was gated by either the \chSiO_2/Si gate or the graphite back gate. In addition to having the Hall probe metallic contacts, a micron-size metallic floating contact was connected in the middle of the graphene flake via one-dimensional edge contact. It is important to emphasize that below the metallic floating reservoir, the graphene region was entirely etched, and after applying the sufficient magnetic field and the gate voltages, the QH edge channels directly terminate and leave the floating reservoir from one side of the graphene chamber to the other side. To heat the floating reservoir, a DC current $I_{S}$ was injected at contact ‘S’ as shown in Fig.  7. This setup helps to create a hot reservoir at milli-Kelvin temperature. The cold-grounded (CG) metal contacts, as shown in Fig.  7, serve as a cold reservoir. Now, one has two reservoirs of different temperatures. To extract the thermal conductance, one needs to know the thermal current flowing between the two reservoirs via the electronic channel and the electron temperatures of the reservoir.

The thermal current flowing between the two reservoirs can be easily estimated for integer and particle-like fractional quantum Hall states, where only downstream edge mode exists. In this scenario, all dissipated power near the floating contact is used entirely to heat the metallic floating contact. Hence, the thermal current flowing between two reservoirs will be exactly the same as the total dissipated power near the floating contact. A DC current $I_{S}$ injected at the source contact $S$ (Fig. 7) flows from the source contact $S$ to the floating reservoir along the chiral QH edge channels.

The outgoing current from the floating reservoir splits into two equal parts, each propagating along the outgoing chiral edge from the floating reservoir to the cold grounds. The floating reservoir reaches a new equilibrium potential $V_{M}=\frac{I_{S}}{2\nu G_{0}}=\frac{V_{S}}{2}$ ( where $G_{0}=e^{2}/h$) with the filling factor $\nu$ of graphene determined by the $V_{BG}$, whereas the potential of the source contact is $V_{S}=\frac{I_{S}}{\nu G_{0}}$. Thus, the power input to the floating reservoir is $P_{in}=\frac{1}{2}(I_{S}V_{S})=\frac{I_{S}^{2}}{2\nu G_{0}}$, where the pre-factor of $1/2$ results because half power dissipates at the back of source contact (7). Similarly, the outgoing power from the floating reservoir is $P_{out}=\frac{1}{2}(2\times\frac{I_{S}}{2}V_{M})=\frac{I_{S}^{2}}{4\nu G_{0}}$. Thus, the resultant injected power dissipation in the floating reservoir due to joule heating is $J_{Q}=P_{in}-P_{out}=\frac{I_{S}^{2}}{4\nu G_{0}}$.

However, the situation may be complicated in hole-conjugate fractional quantum Hall states like $\nu=2/3$, which also supports the upstream and downstream modes. The downstream and upstream modes can have different chemical potentials near the metallic floating contact and the source contact. Due to the charge tunnelling between the counter-propagating modes, the edge channel attains the equilibrium chemical potential for the propagation length larger than the charge equilibration length ($l_{eq}^{c}$). In this scenario, the total dissipated joule power also remains the same. Still, some portion is dissipated in the tunnelling regions outside the ohmic reservoirs, and the rest is dissipated in the floating metallic contact. In this case, one can not equate the total dissipated power to the thermal current $J_{Q}$. However, suppose one assumes that all heat from the charge equilibration region returns to the nearby ohmic reservoirs. In that case, one can safely use the formula derived for the integer and particle-like fractional quantum Hall states. Such an assumption is justified if the charge equilibration length is much smaller than the reservoir size. Fortunately, the charge equilibration length in a graphene-based system is expected to be quite small; hence, we can safely equate the total dissipated power near the floating contact to the thermal current $J_{Q}$. Here we would also like to define the characteristic length scales like charge and thermal equilibration length, which will be extensively used in this review.

Charge equilibration length ($l_{eq}^{c}$): Charge equilibration length ($l_{eq}^{c}$) is defined as the minimum propagation length required to achieve the complete voltage equilibration between the counter-propagating edge modes. In other words, all counter-propagating edge modes should attain the same electrostatic voltage after a propagation length of $l_{eq}^{c}$.

Thermal equilibration length ($l_{eq}^{H}$): Thermal equilibration length ($l_{eq}^{H}$) is defined as the minimum propagation length required to achieve the same temperature between the counter-propagating edge modes. In other words, all counter-propagating edge modes should attain the same temperature profile after a propagation length of $l_{eq}^{H}$.

The resulting increase in the electron temperature $T_{M}$ of the floating contact was determined by measuring the excess thermal noise at contact ‘T’. To avoid the contribution from the 1/f noise, a resonant L//C tank circuit was used to shift the operating frequencies close to the 0.7 MHz regimes. The circuit consists of an inductor made of superconducting NbTi wire, and the capacitance developed along the coaxial line connecting the sample to the cryogenic amplifier. A ceramic capacitance of 10 nF was also introduced between the sample and inductor to block the DC current along the measurement line and isolate the DC voltage at the cryogenic amplifier gate port from the sample bias voltage. The output voltage from the cryogenic amplifier was further amplified using a room-temperature voltage amplifier. After the second amplification stage, the resulting signal was measured using a spectrum analyzer. The measured excess noise at contact ‘T’ is related to the electron temperature $(T_{M})$ of the floating contact via the relation,

where $\nu$ is the bulk filling of the QH state. Furthermore, it is worth emphasising the various conditions that need to be fulfilled to justify the use of noise thermometry for electron temperature $(T_{M})$ determination.

Quasi-equilibrated electronic distribution function: The metallic floating contact will act as a hot reservoir only if it has a well-defined quasi-equilibrated electronic distribution characterized by temperature $T_{M}$. For the realization of such a quasi-equilibrium regime, the hot electrons must dwell in metallic floating contact sufficiently longer than the electron-electron interaction (thermalization) time scale. The dwell time can be calculated as; $t_{dwell}=\frac{D_{E}\Omega h}{N}$[85, 90], where $D_{E}$ is the electronic density of states per unit volume per unit energy, $\Omega$ is the volume of micron-size floating contact, $h$ is the Planck’s constant and $N$ is the number of channels leaving the floating contact. In typical graphene devices used for thermal conductance measurement, the floating contact has a volume  $\Omega\approx 0.5\mu$m³, mostly made of gold. Using the typical density of states for gold $D_{E}\simeq 1.14\times 10^{47}J^{-1}m^{-3}$, estimated dwell time was found to be $t_{dwell}\simeq\frac{40\mu s}{N}$, which is much larger than the typical electron-electron interaction time of the order of 10 ns range for gold at a temperature down to few milli-kelvin[91]. This firmly establishes the quasi-equilibrium hypothesis that the electron’s energy distribution in the micron size ohmic contact is a hot Fermi distribution function characterized by a temperature $T_{M}$.

High-quality ohmic contact between the metallic floating reservoir and graphene channel: As mentioned, since the electron temperature determination relies on excess thermal noise measurement, other unwanted noise sources should be avoided. One such factor is that the metallic floating reservoir should make good quality ohmic contact with graphene to minimise the current reflection, which otherwise will generate unwanted Shot noise. Thanks to the development of a high-quality ohmic one-dimensional edge contact technique for graphene, the minimal reflection coefficient was obtained, which was negligible for all devices used for the thermal conductance measurements in literature.

We will first discuss the results reported for the Integer quantum Hall states in graphene. Fig. 8(a) shows the quantum Hall response of the device as a function of back-gate voltage at 9.8 T of the magnetic field at $\sim$ 40 mK of the bath temperature. In addition to the signature QH plateau of SLG at $2\frac{e^{2}}{h},6\frac{e^{2}}{h},10\frac{e^{2}}{h}$, QH plateaus at $1\frac{e^{2}}{h},4\frac{e^{2}}{h},5\frac{e^{2}}{h}$, are also visible, which are associated with the degeneracy lifting of zeroth and first Landau levels of the single layer graphene.

To measure the thermal conductance, the central floating contact was heated to a temperature $T_{M}$ due to the Joule dissipation $J_{Q}$ near the floating contact. The plot of $T_{M}$ against the $J_{Q}$ is shown in Fig. 8(b). To extract the electronic contribution to the thermal conductance of the QH states, $J_{Q}$ can be plotted against the $T_{M}^{2}-T_{0}^{2}$ and the slope of the linear fit provides the value of the quantized thermal conductance $G_{Q}$ provided the electron-phonon cooling contribution $(J_{Q}^{e-ph})$ is negligible. Experimentally, It was found that $J_{Q}^{e-ph}$ is negligible for the typical device geometry till $\sim T_{M}=100$ mK[89]. However, to completely rule out any electron-phonon cooling ($J_{Q}^{e-ph}$) contribution, dissipated joule power between the two QH filling factors was subtracted at constant $T_{M}$ under the assumption that the $J_{Q}^{e-ph}$ does not depend on the number of edge channels leaving the floating contact. The resulting plot between $\lambda=\Delta J_{Q}/0.5\kappa_{0}$ (where $\Delta J_{Q}=J_{Q}(\nu_{i},T_{M})-J_{Q}(\nu_{j},T_{M})=0.5\Delta N\kappa_{0}(T_{M}^{2}-T_{0}^{2})$ ) vs $T_{M}^{2}-T_{0}^{2}$ is shown in Fig. 8(c). From the slope of the curve $(m)$, thermal conductance per ballistic channel was calculated as $G_{Q}=(m/\Delta N)\kappa_{0}T$ and found to be $\approx 1\kappa_{0}T$ as predicted theoretically. This plot is the first experimental demonstration of the quantized heat flow for integer quantum Hall states in graphene since its first mechanical exfoliation isolation. This experiment also emphasized that the universality of the quantized heat flow is the same for intrinsic QH states ($\nu=2,6$) and symmetry-broken quantum Hall states ($\nu=1$) in graphene devices, as expected.

The four-fold unique spin valley degeneracy of single-layer graphene distinguishes it from the conventional \chGaAs/AlGaAs system. In the regime of fractional quantum Hall states, the presence of a large magnetic field can give rise to different possibilities of the symmetry breaking of the internal four-fold degeneracy, such as the complete lifting of the spin and valley degeneracy, hence the emergent FQH states will be either spin or valley polarised. Another possibility includes the degeneracy lifting of only one flavour, either spin or valley and hence, the emergent FQH phases preserve the approximate SU(2) symmetry in the remaining flavour sector. In addition to these two possibilities, there might be a scenario where none of the spin and valley flavours degeneracy is broken. Consequently, the emergent FQH phases have mixed spin-valley flavours. The earlier transport experiments and the local compressibility measurements suggest that while the FQH states with filling $0<\nu<1$ follow the standard sequence of non-interacting single component two flux composite fermion model, only the even numerator FQH states emerge for the filling $1<\nu<2$, suggesting the possibility of two-component composite fermion model, with an approximately SU(2) symmetric states for the interacting electrons. This complexity or richness of FQH states in graphene raised concern over the universality of the quantized heat flow for FQH states in graphene.

The earlier experimental result of the heat flow measurement for the particle-like FQH state $(\nu=4/3)$[89] in an hBN encapsulated graphite back-gated single-layer graphene device is shown in Fig. 9. Fig. 9(a) shows the QH plateau of $\nu=4/3$ FQH state between the integer plateau of $\nu=1$ and $2$. Fig. 9(b) shows the plot of $J_{Q}$ (solid circles) as a function of $T^{2}_{M}-T^{2}_{0}$ for $\nu$ = 1, $\frac{4}{3}$, and 2 over the temperature window where the curve is linear, implying the dominance of the electronic contribution to the heat flow. The solid lines in Fig. 9(b) represent the linear fits (in $0.5\kappa_{0}$) and give the values of 2.04, 4.16 and 4.04, which corresponds to $G_{Q}=1.02,2.08$ and $2.02\kappa_{0}T$ for $\nu$ = 1, $\frac{4}{3}$, and 2, respectively. Conventionally, $\nu$ = $\frac{4}{3}$, is thought of as $1+1/3$. As a result, one would expect two downstream charge modes, one integer and one fractional (inner $\nu$ = $\frac{1}{3}$ with effective charge, $e^{*}=\frac{e}{3}$). If the universality of quantized heat flow is preserved, the thermal conductance of $\nu=\frac{4}{3}$ should be the same as $\nu$ = 2 having two integer downstream charge modes. This is indeed the same as shown in Fig. 9(b). For $\nu=4/3$, $G_{Q}$ was found to be $(2.02\pm 0.02)\kappa_{0}T$. This measurement established the universality of the quantized thermal conductance in graphene for both integer and fractional QH edges.