Thermal crossover from a Chern insulator to a fractional Chern insulator in pentalayer graphene

Abstract

By theoretically analyzing the recent temperature dependent transport data [Lu et al., arXiv:2408.10203] in pentalayer graphene, we establish that the experimentally observed transition from low-temperature quantum anomalous Hall effect to higher-temperature fractional quantum anomalous Hall effect is a crossover phenomenon arising from the competition between interaction and disorder energy scales, with the likely zero temperature ground state of the system being either a localized insulator or a Chern insulator with a quantized anomalous Hall effect. In particular, the intriguing suppression of FQAHE in favor of QAHE with decreasing temperature is explained as arising from the low-temperature localization of the carriers where disorder overcomes the interaction effects. We provide a detailed analysis of the data in support of the crossover scenario.

In a recent follow-up experiment JuEQAH2024 to their breakthrough discovery of quantum anomalous Hall (QAHE) and fractional quantum anomalous Hall (FQAHE) effects in pentalayer graphene (PLG) PentaGraphene2023 , Long Ju and collaborators at MIT have made a startling discovery that, with the lowering of temperature (T) from $>300$ mK to $<100$ mK, the original PLG FQAHE reported in Ref. PentaGraphene2023, is suppressed in favor of a QAHE phase over a large range of the filling factor ($\nu$) in some range of the applied displacement field ($D$). In addition, as T decreases, the strongly insulating (SI) phase, also observed in some low-$\nu$ corners of the $D-\nu$ space in the original paperPentaGraphene2023 , expands along with the QAHE phase. The purpose of the current note is to analyze the new low-T data in Ref. JuEQAH2024, and suggest some possible physical explanations for the puzzling phenomena reported therein. All our analyses of the experiment use the actual data underlying Ref. JuEQAH2024, , which were supplied to us by the authors of Ref. JuEQAH2024, .

We earlier analyzed XiePenta2024 the higher-T results reported in Ref. PentaGraphene2023, , which can be summarized as follows. We found a seemingly puzzling constant activation gap ($\Delta_{\nu}\negmedspace\sim\negmedspace 5$ K) for all the reported FQAHE fractions, $\nu=2/5,3/7,4/9,4/7,3/5,2/3$ along with a QAHE gap $\sim 10$ K for $\nu=1$. In addition, the FQAHE for all $\nu$ generically manifest a mysterious residual longitudinal resistance $R_{0}>10$ $\text{\,}\mathrm{k\SIUnitSymbolOhm}$ at low temperatures, which is unheard of in the quantum Hall literature, where the quantization of $R_{xy}$ is always associated with the vanishing of $R_{xx}$. By contrast, the QAHE at $\nu=1$ shows a much smaller value of $R_{0}\sim 0.1-0.5\,\textrm{$\text{\,}\mathrm{k\SIUnitSymbolOhm}$}$. There was an additional mystery in the original experiment PentaGraphene2023 , as shown in Fig. 1. The T-dependence of $R_{xx}$ for all fractions is essentially identical between $\nu=2/3$ and $\nu=2/5$ including at $\nu=1/2$, which is extremely puzzling. But $R_{xx}(T)$ at $\nu=1$ is qualitatively different as also shown in Fig. 1. We note that Ref. JuEQAH2024, basically manifests an SI phase for $\nu<2/5$ (e.g. $\nu=1/3$ does not manifest any FQAHE as it falls within the strongly insulating regime in both Refs. JuEQAH2024, ; PentaGraphene2023, ), so only the regime $\nu>1/3$ is relevant for our consideration.

The results in Fig. 1 clearly bring out the two essential features of the data in Ref. PentaGraphene2023, , namely, the constancy of the activation energy for all fractional $\nu$ values (as reflected in the almost parallel rise in the activated $R_{xx}$ for $0.3$ K$<T<1$ K) including $\nu=1/2$, and the existence of a large $R_{0}$ for all the fractions as reflected in $R_{xx}(T<0.3K)$ approaching a constant value $R_{0}\sim 10-15\,\textrm{$\text{\,}\mathrm{k\SIUnitSymbolOhm}$}$. By contrast, $\nu=1$ QAHE produces a very small $R_{0}$ in Fig. 1.

Figure 1: Temperature dependence of the longitudinal resistance $R_{xx}$ at integer and fractional filling factors. The data is taken from Device 1 reported in Ref. PentaGraphene2023, (Courtesy of L. Ju, et al.)

Another surprising feature of Fig. 1, which was not reported in Ref. XiePenta2024, (and was entirely missed in Ref. PentaGraphene2023, ), is that the data points to an activated temperature dependence of $R_{xx}$ at $\nu=1/2$, with a gap $\Delta_{1/2}\sim 5$ K (and $R_{0}\sim 12$ k$\Omega$), implying an apparent FQAHE at $\nu=1/2$. The corresponding Hall resistance $R_{xy}$, however, reflects no plateau at $\nu=1/2$ in Ref. PentaGraphene2023, . This apparent conflict remained a mystery for the data of Ref. PentaGraphene2023, and was therefore not mentioned in Ref. XiePenta2024, .

The new experiment JuEQAH2024 now sheds some light on this mystery by extending the temperature range down to $T\sim 40$ mK (the reliable $T$ values in Ref. PentaGraphene2023, were above $100-300$ mK). The new discovery of Ref. JuEQAH2024, is stunning: With decreasing $T$, the QAHE phase takes over three nearly connected $\nu-D$ regions, extending from $\nu=0.5$ to above $\nu=1$, and FQAHE states in these regions are suppressed with the $R_{xy}$ quantization now (at lower $T$) systematically becoming $h/e^{2}$ QAHE quantization! The corresponding $R_{xx}(T)$ shows a crossover behavior (in contrast to a clear activated behavior) until T is low enough to reach QAHE. In addition to these extended regions of QAHE, the SI regime also expands appreciably covering much of the $\nu<1/2$ region where both $R_{xx}$ and $R_{xy}$ become very large at low $T$. The experimental results of Ref. JuEQAH2024, are consistent with the system (at some $D$ values) showing an extended quantum anomalous Hall effect (EQAHE) over a large range of $\nu$ values with $R_{xy}$ being quantized at $h/e^{2}$ as in the regular QAHE. The EQAHE appears only as T decreases, since for $T\sim 0.3K$ the QAHE manifests only at $\nu\sim 1$, and the FQAHE is observed for the usual Jain fractions Jain1989 of $\nu=2/3,3/5$, etc, as in Ref. PentaGraphene2023, . In fact, the results of Ref. JuEQAH2024, seem to indicate that the $T=0$ phase is likely to be QAHE at all $\nu$ values except perhaps for $\nu<2/5$ which is SI. The universal large background $R_{0}$ for the FQAHE phase in Fig. 1 is indicating that the observed FQAHE is a high-T crossover phenomenon, and therefore, it can happen over a background parent phase with a large $R_{xx}$ at higher $T$. Lowering $T$ then leads to the true ground state, the EQAHE phase with $R_{xy}\sim h/e^{2}$, and a small residual $R_{xx}$.

Figure 2: Temperature dependence of the longitudinal resistance $R_{xx}$ for the same Device 1 as in Fig. 1 but at lowered temperatures extracted from Ref. JuEQAH2024, (Courtesy of L. Ju, et al.)

Before discussing the possible reason for this crossover from a higher-T FQAHE to a lower-T QAHE, we first show our analyses of the data of Ref. JuEQAH2024, , establishing that the crossover phenomenology described above is consistent with the data. In Fig. 2, we show our extracted $R_{xx}(T)$ from Ref. JuEQAH2024, for $\nu=1/2,3/5,2/3,1$ in the low-T regime $T\in(0.02\textrm{ K},0.3\textrm{ K})$. It is clear that $R_{xx}$ is showing a crossover from one phase ($T<0.1$ K) to a different phase ($T>0.3$ K) with no clear activation behavior. It is also manifestly obvious that the low-T ‘phase’ (up to $T\sim 0.1$ K) is the same for all three fractions whereas the high-T phase is different for different $\nu$ as expected for FQAHE footnote . The low-T phase is the EQAHE phase, which eventually succumbs to the FQAHE phase around $T\sim 0.3$ K. We mention that the corresponding $R_{xx}(T)$ for the QAHE around $\nu\sim 1$ manifests an almost T-independent $R_{xx}\sim 0$ up to $T\sim 0.3$ K consistent with the activation gap for the QAHE around $\nu\sim 1$ being larger than 10K as seen in Ref. PentaGraphene2023, . Note that the physics depends on the value of $D$, and Fig. 2 shows results for a specific D since this is what is provided in Ref. JuEQAH2024, . The low-T EQAHE to the high-T FQAHE crossover depends on the applied displacement field.

Figure 3: Temperature dependence of the Hall resistance $R_{xy}$ for Device 1 extracted from Ref. JuEQAH2024, . (Courtesy of L. Ju, et al.) The dashed lines mark the expected integer and fractional quantized values of $R_{xy}$.

To further emphasize the crossover, we show in Fig. 3 the explicit $R_{xy}(T)$ extracted from Ref. JuEQAH2024, corresponding to the $R_{xx}(T)$ shown in Fig. 2. While the $R_{xy}(T)$ at $\nu=1$ remains quantized within $0.1\%$ of $h/e^{2}$, the results at fractional fillings emphatically bring out the crossover phenomenon—a low-T QAHE for $\nu=2/3,3/5$ is becoming a higher-T FQHE for $T\sim 0.3$ K. Although at the same D value, the $R_{xy}(T)$ for $\nu=1/2$ does not approach any quantized value, it in fact approaches the value $2h/e^{2}$ expected for the composite Fermi liquid phase West1993 ; Read1993 at a slightly smaller $D$. We note that Fig. 3 (extracted by analyzing the data presented in Ref. JuEQAH2024, ) indicates that the experimental temperature range for the fractional fillings is dominated by crossover physics (as we emphasized above in the context of Fig. 2) with neither the QAHE nor the FQAHE well-established at high and low temperatures respectively since the quantization is hardly exact at any T footnote , except at $\nu=1$, where the system remains a well-quantized QAHE with a large activation gap and vanishing $R_{xx}$ throughout. This crossover behavior from low-T EQAHE to high-T FQAHE (in contrast to a phase transition) is further reinforced by our direct calculation of the activation gap for the FQHE observed in Ref. JuEQAH2024, for different $\nu$ values as presented below in our Figs. 4-6.

Figure 4: Thermal activation fitting of the longitudinal resistance for $\nu=2/3$. The two panels correspond to fitting with different temperature ranges. In the lower panel, data in the shaded region is excluded from the fitting.

Figure 5: Thermal activation fitting of the longitudinal resistance for $\nu=3/5$. The same convention is used as in Fig. 4.

The thermal activation fits to the data of Ref. JuEQAH2024, follow the same procedure as what we used in our earlier work XiePenta2024 for analyzing the reported high-T ($>0.3$ K) FQAHE of Ref. PentaGraphene2023, . We use $R_{xx}(T)=R_{0}+A\exp(-\Delta/2k_{B}T)$, where A is a constant $\sim h/e^{2}$, and obtain the best fit values of $R_{0}$ and $\Delta$ for different $\nu=1/2,3/5,2/3$ using the low-T data of Ref. JuEQAH2024, as shown in Figs. 4-6. These fits work over a very narrow range of T, consistent with the physics being dominated by crossover rather than by a phase (or even two phases) over the whole range of $T=10$ mK$-1$ K. The values of $\Delta$ for the observed FQAHE (at higher T in the data of Ref. JuEQAH2024, ) are $\sim 0.2$ K for all $\nu$ values in sharp contrast to the much larger $\Delta\sim 5$ K found for the FQAHE reported in Ref. PentaGraphene2023, . In addition, the extracted residual resistance ($R_{0}$) is much smaller in Ref. JuEQAH2024, , being $R_{0}\sim 0.2-1\textrm{$\text{\,}\mathrm{k\SIUnitSymbolOhm}$}$ in sharp contrast to $R_{0}>10$ $\text{\,}\mathrm{k\SIUnitSymbolOhm}$ found for the FQAHE in Ref. JuEQAH2024, . In addition, the high-T ($>0.3$ K) results in Fig. 3 are hardly quantized at the expected values of $R_{xy}=h/(\nu e^{2})$, with the deviation from the quantization being of the order of $10-15\%$, which is not inconsistent with having a background residual resistance $R_{0}\sim 0.2-1$ $\text{\,}\mathrm{k\SIUnitSymbolOhm}$footnote . These results decisively show that the observed FQAHE in both Refs. JuEQAH2024, ; PentaGraphene2023, are fragile intermediate crossover phases, and not the ground states. This also solves the puzzle of why the extracted gaps for the FQAHE at different $\nu$ were all the same ($\sim 5$ K) as well as why the background residual resistance was large for the FQAHE in Ref. PentaGraphene2023, as analyzed in Ref. XiePenta2024, . We emphasize that the QAHE at $\nu=1$ is, however, a real ground state topological phase and not a crossover, and it has a small $R_{0}$ and comparable activation gaps in both Refs. JuEQAH2024, ; PentaGraphene2023, although the available low-T data in Ref. JuEQAH2024, are not sufficient to precisely estimate a QAHE gap at $\nu=1$—all we can say is that the gap at $\nu=1$ is orders of magnitude larger than $0.1$ K ($>10$ K most likely) and the associated $R_{0}<100\textrm{$\text{\,}\mathrm{\SIUnitSymbolOhm}$}$.

Figure 6: Thermal activation fitting of the longitudinal resistance for $\nu=1/2$. The same convention is used as in Fig. 4.

We provide in Fig. 11 the low-T JuEQAH2024 EQAHE to high-T PentaGraphene2023 FQAHE crossover for fractional filling by showing in Figs. 7-10 our calculated T-dependence of $R_{xx}$ and $R_{xy}$ as well as $G_{xx}$ and $G_{xy}$, obtained by inverting the resistance matrix $R$, for $\nu=1,2/3,3/5,1/2$ at fixed $D$ (the same as in Fig. 2c and 2d in Ref. JuEQAH2024, ). These results in Figs. 7-10 reinforce the crossover nature of the quantum Hall phenomenology at fractional fillings, while also establishing that the $\nu=1$ QAHE is indeed a stable integer quantum Hall phase.

Figure 7: Temperature dependence of the resistances and conductances at integer filling factor $\nu=1$. Dashed lines mark the expected quantized values.

The results of Figs. 7-10 indicate that only $\nu=1$ is a true QAHE topological phase with a reasonable quantization of $G_{xy}=e^{2}/h$ and $R_{xy}=h/e^{2}$ over the whole range of $T\sim 0.01K-0.3$ K as well as having a rather small background resistance $R_{0}\sim 100\textrm{$\text{\,}\mathrm{\SIUnitSymbolOhm}$}$ appropriate for the quantum Hall quantization. All the fractional states manifest strong crossover behavior from a putative QAHE phase (with $G_{xy}$ varying from $e^{2}/h$ at very low $T$ to within $15\%$ of the exact value $\nu e^{2}/h$ at higher $T$) over this temperature range. We emphasize that the physics is strictly crossover physics at fractional fillings, and only the QAHE phase at $\nu=1$ is a real phase with a true ground state quantization.

Figure 8: Temperature dependence of the resistances and conductances at filling factor $\nu=2/3$. Dashed lines mark the expected quantized values.

Figure 9: Temperature dependence of the resistances and conductances at filling factor $\nu=3/5$. Dashed lines mark the expected quantized values.

To complete the phenomenology, we provide in Fig. 11 the low-T JuEQAH2024 EQAHE to high-T PentaGraphene2023 FQAHE crossover for fractional filling by showing our calculated $\nu$-dependence of $R_{xx}$ and $R_{xy}$ as well as $G_{xx}$ and $G_{xy}$ for fixed T (between $7$ mK base temperature to $380$ mK) at the same D value as in Figs. 7-10. Note that this analyses use data exclusively from Ref. JuEQAH2024, and no attempt is made to reconcile the raw data between Refs. JuEQAH2024, ; PentaGraphene2023, , particularly since the temperature calibration in Ref. PentaGraphene2023, most likely has errors footnote . (Note also that the very low-T temperature measurements in these data from Ref. JuEQAH2024, most likely refer only to the base temperature and not necessarily the electron temperature LongPrivate .) What is obvious from Fig. 11 is that at the lowest T, the system is almost entirely in the EQAHE phase from $\nu\sim 0.5$ to $1.0$, and for $\nu<0.4$, the system is a strong insulator (where $G_{xx}\sim 0$, but $R_{xx}$ is large). As T increases, strong crossover effects set in, but the underlying EQAHE phase survives up to 100 mK. For $T>100$ mK, signatures for FQAHE start developing, but they are neither robust nor definitive when one looks at the totality of the analyzed data over the whole temperature and filling ranges.

Figure 10: Temperature dependence of the resistances and conductances at filling factor $\nu=1/2$. Dashed lines mark the expected quantized values.

Having established that the observed phenomenology of Ref. JuEQAH2024, is a low-T to high-T crossover between QAHE and FQAHE originally reported in Ref. PentaGraphene2023, , we now address the possible cause underlying this crossover. We first mention that this crossover is surprising because the FQAHE is considered to be an interaction-induced fragile phase which typically manifests itself at low temperatures whereas here JuEQAH2024 ; PentaGraphene2023 , it is in fact disappearing at the lowest temperatures while existing at higher temperatures (or rather at intermediate temperatures, since FQAHE obviously disappears at very high temperatures again when $T>\Delta_{\nu}$). Although this phenomenology of an FQHE existing only in the intermediate temperatures, while disappearing at the lowest temperatures, is unusual, it is not unheard of. For example, it often happens in the regular high-field fractional quantum Hall effect (FQHE), but only for very low density samples where the low fractions in the lowest Landau level are involved in the physics ManfraP ; ShayeganP . For example, Figure 2 in Ref. Shayegan2022, gives an example of a $1/7$ FQHE state arising in a 2D GaAs electron system (with a low carrier density of $6\times 10^{10}$ cm^-2) only in an intermediate temperature range $\sim 90$ mK and vanishing at lower (and higher) temperatures. In addition, this FQHE at $1/7$ filling occurs on a very highly resistive ($\sim 10$ M$\Omega$!) $R_{xx}$ background, which is rather similar to the FQAHE in Ref. PentaGraphene2023, happening with a large $R_{0}$ ( $>10\textrm{$\text{\,}\mathrm{k\SIUnitSymbolOhm}$}$). Similarly, Figure 3a in Ref. Shayegan2023, gives an example of an intermediate temperature ( $\sim 125$ mK) FQHE at $3/13$ filling in a low density ($\sim 10^{11}$ cm^-2) 2D GaAs hole system, which disappears at both lower ($\sim 100$ mK) and higher ($\sim 150$ mK) temperatures, appearing only in the intermediate temperature range similar to the FQAHE phenomenology in PLG being discussed in the current work. This $3/13$ intermediate temperature FQHE in the 2D hole system also happens in a very large background $R_{xx}\sim 10$ M$\Omega$. Thus, the phenomenology of a higher-T FQHE disappearing at lower temperatures along with a large background resistance is not unheard of in the regular FQHE literature. What is new in the PLG FQAHE is that the low-T phase turns out to be an EQAHE phase for some $\nu$ values whereas at other $\nu$ values ($\nu<2/5$), the PLG manifests a highly resistive insulating phase similar to that in Refs. Shayegan2022, ; Shayegan2023, . In the regular high-field FQHE literature (e.g. 2D GaAs), such a thermal crossover from a high-T FQHE-like phase to a low-T QHE phase has never been reported to the best of our knowledge —there the crossover is from a high-T apparent FQH phase to a low-T SI phase Shayegan2022 ; Shayegan2023 . We cannot rule out the possibility that at still lower temperatures the PLG manifests only an SI phase for all filling as it does now (at perhaps 40mK electron temperature) for $\nu<2/5$. In regular QHE/FQHE literature, three distinct situations exist generically at the lowest temperatures for 2D low-density systems in the lowest Landau level: (1) no QHE or FQHE is observed with the lowest Landau level being entirely an SI phase; (2) the lowest Landau level is entirely a QHE phase with perhaps a transition to the SI phase at the lowest filling; (3) the lowest LL manifests FQHE at some fillings transitioning to the SI phase at sample dependent values of very low fillings. We emphasize that the temperature-induced QHE/FQHE dichotomies manifested in Refs. Shayegan2022, ; Shayegan2023, and in Refs. JuEQAH2024, ; PentaGraphene2023, are not ground state phases but temperature induced crossover physics as we emphasize in the current work. The ultimate $T=0$ phase in the PLG is likely to be an SI phase for all filling or a QAHE phase for all filling, but more lower-T experiments are needed to settle this issue.

Figure 11: Filling factor dependence of the resistances and conductances at fixed temperatures (a-b) $T=0.07$ K, (c-d) $T=0.15$ K and (e-f) $T=0.38$ K. Dashed lines mark the unit of resistance/conductance. Data is from Fig. 2c-d of Ref. JuEQAH2024, . (Courtesy of L. Ju, et al.)

What could be the reason for the observed thermal crossover? We speculate that the crossover arises from a competition between disorder and interaction. If the effective disorder is very strong (as it must be for $\nu<2/5$ in Ref. JuEQAH2024, ) the ground state is an SI. The fact that strong disorder suppresses FQHE Haldane1985 ; Haldane2003 ; DasSarma1985 as well as FQAHE DasSarma2012 , leading to a strongly insulating phase, is theoretically known in the literature, and is of course well-established experimentally. For lower effective disorder (as is happening for $\nu>2/5$), the ground state is QAHE (manifesting as EQAHE for $\nu<1$ and QAHE for $\nu\sim 1$). The reason the system crosses over to an FQAHE phase at slightly higher (but not very high) temperatures is most likely because some fraction of the localized electrons forming the $T=0$ SI phase are thermally activated into delocalized carriers, and as they start interacting with each other, they may form an intermediate FQAHE phase (perhaps as happening in Refs. Shayegan2022, ; Shayegan2023, in the high-field FQHE). Whether such a phenomenon happens or not is a delicate function of disorder, screening, interaction, and temperature. The phenomenology also depends sensitively on the filling factor and the displacement field since the screening properties of the carriers depend sensitively on these two, and what matters is the screened effective disorder instead of the bare disorder. The fact that the screened disorder in the quantum Hall systems depends sensitively on the filling factor has been well-known for a long time SDS1981 ; SDS1990 . It is obvious that the effective disorder must depend on the applied displacement field since the whole PLG bandstructure depends sensitively on the applied field. Thus, there are two separate energy scales in the problem associated with the effective disorder strength ($E_{dis}$) and the effective interaction strength ($E_{int}$), leading to the observed crossover phenomena. SI and/or QAHE dominate for $E_{int}<T<E_{dis}$, but there may be an optimal fine-tuned situation where $E_{dis}<T<E_{int}$ allowing a crossover from the low-T QAHE to the intermediate-T FQAHE. For even higher T, when $E_{int}<T$ (or more precisely the temperature exceeds the FQAHE excitation gap), the FQAHE itself disappears.

The existence of two distinct energy scales $E_{dis}$ and $E_{int}$, arising from two distinct terms in the Hamiltonian (i.e., disorder and interaction), is what may lead to the competition necessary for the crossover physics. The reason that this crossover is so much more prominent in PLG Chern insulators than in regular 2D high-field QHE/FQHE physics is most likely because PLG has three different knobs allowing for the fine-tuning of this competition and crossover: filling factor $\nu$, displacement field D, and temperature T. (The temperature is an important scale because the screening is typically strongly T-dependent.) By contrast, regular 2D high-field QHE/FQHE has only the filling factor as the tuning knob, making it highly unlikely that this competition between $E_{dis}$ and $E_{int}$ can be fine tuned except at very low densities where disorder effects are the strongest leading to very large $R_{xx}$ values. This conclusion about the importance of the displacement field in the QAHE/FQAHE physics of PLG is also directly corroborated by the fact that the tuning of the displacement field is crucial in the experimental observation of QAHE, FQAHE, and EQAHE in Refs. JuEQAH2024, ; PentaGraphene2023, . It is possible that if the disorder could be fine tuned at will (independent of the filling factor $\nu$) in regular 2D high-field Landau level systems, then the QHE to FQHE crossover could be observed at arbitrary filling factors in 2D systems too JainPrivate . The lack of such an additional tuning knob makes this crossover possible only at very low $\nu$ in 2D landau level systems, and thus in 2D systems the QHE to FQHE crossover as T increases has not been reported although the crossover from SI to FQHE has been occasionally seen at very low densitiesManfraP ; ShayeganP ; Shayegan2022 ; Shayegan2023 .

We note that our conclusion for the observed thermal QAHE to FQAHE crossover in PLG as arising from a competition between interaction and disorder is agnostic about whether the PLG QAHE ground state is a Wigner crystal or not (the so-called Hall crystal Halperin1989 ) as has been speculated recently in several Dong2024 ; Parker2024 ; Sheng2024 ; Dong2023 ; Zhang2023 but not all Xiao2024 ; FanZhang2024 ; Bernevig4 ; Bernevig3 ; Bernevig2 ; Liu2023 publications. The Hall crystal is essentially a Wigner crystal which is pinned by disorder and for all practical purposes, it is simply a localized insulator for our discussion. For the Wigner crystal possibility, the energy scale $E_{dis}$ could also include the possible melting temperature of the Wigner crystal Vu2022 , which also must depend sensitively on the filling factor $\nu$ and the displacement field D as discussed above. We emphasize that any actual calculation of the energy scales $E_{dis}$ and $E_{int}$ (and therefore, of the crossover temperature scale for the QAHE to FQAHE crossover) is essentially impossible at this point since the bare disorder itself is unknown in PLG and even calculating the energy scales for the pristine system (i.e. $E_{int}$ and $\Delta$) is a challenge and is highly controversial since the results depend crucially on the approximation scheme as emphasized in Ref. FanZhang2024, from a microscopic perspective and in Ref. XiePenta2024, from a phenomenological perspective. More experiments and considerable theoretical work are necessary for a better understanding of the quantitative aspects of the QAHE to FQAHE crossover observed in Ref. JuEQAH2024, , but the basic physics of the thermal crossover arising from a competition among interaction, disorder, and temperature is reasonably clear.

A very rough estimate of the crossover temperature scale from the EQAHE at lower-T to the FQAHE at higher-T reported in Ref. JuEQAH2024, can be obtained if we assume that the crossover is arising specifically from the thermal melting of an underlying Wigner crystal, i.e., if we accept the theoretical premise that the $T=0$ ground state of the PLG Chern insulator is an anomalous Hall crystal Halperin1989 with $R_{xy}=h/e^{2}$ for all $\nu$ and $R_{xx}=0$ (due to the crystal being pinned by disorder) as claimed in some of the recent theoretical worksDong2024 ; Parker2024 ; Sheng2024 ; Dong2023 . We note that the anomalous Hall crystal ground state produces an integer Hall resistance quantization with $R_{xy}=h/e^{2}$ throughout the whole range of $\nu$ from 0 to 1, and as such, the observation of the FQAHE in Refs. JuEQAH2024, ; PentaGraphene2023, at higher temperatures is inconsistent with a Hall crystal ground state, but is consistent with our crossover scenario if the crystal happens to melt around the crossover temperature scale ($100-300$ mK in Figs. 7-10 above) where the gradual transition from the EQAHE to FQAHE occurs.

There is no available estimate of the thermal melting temperature of an anomalous Hall crystal in the literature, but there are estimates for the thermal melting temperature of the 2D Wigner crystal Vu2022 ; Huang2024 ; Hwang2001 ; Anderson1979 ; Jain2013 . Using the available literature for the melting of the 2D Wigner crystal (and assuming that such estimates apply to the 2D Hall crystal in PLG), we find that any existing PLG-based anomalous Hall crystal would have a melting temperature of $50-500$ mK, depending on the precise carrier density. Interestingly, this estimate is roughly consistent with the observed EQAHE-FQAHE crossover scale $\sim 0.1-0.3$ K in Ref. JuEQAH2024, . This could very well be a coincidence since the melting temperature of the PLG anomalous Hall crystal is theoretically unknown. In addition, the pristine Hall crystal must be pinned by disorder, and hence in the end the crossover may have nothing to do with the melting of the Hall crystal and everything to do with the strength of the disorder. Also, the observed EQAHE ground state clearly depends both on the filling fraction $\nu$ and the applied displacement field $D$, which argue for the importance of the effective disorder playing a key role through screening. Unfortunately, nothing is known about the amount of disorder in the experimental PLG samples, so there is no clue on how to estimate the activation gap for the disorder-induced localization (or equivalently, pinning of the Hall crystal) in the sample.

Future temperature dependent QAHE/FQAHE experiments on additional PLG samples would be highly desirable in this context since the Hall crystal induced crossover phenomenology should be fairly universal for all samples (at similar carrier densities and displacement fields) whereas the disorder effects should vary randomly from sample to sample, thus producing different crossover behaviors in different samples. It may be useful here to mention the corresponding situation in the regular 2D high-field Landau level QHE/FQHE experiments, where the physics depends crucially on disorder, and there is no universal behavior for samples manifesting QHE or FQHE —everything depends on the sample quality (characterized by the sample mobility), and in general, higher mobility samples manifest both QHE and FQHE, and lower mobility samples manifest only QHE, and very poor quality samples do not manifest even QHE in the lowest Landau level. A classic example is the original discovery Klitzing1980 of QHE by von Klitzing in the 2D Si-SiO₂ MOSFETs, where the QHE manifested only in the higher Landau level and the lowest spin- and valley-split Landau level only manifested strong localization with $R_{xx}$ ($G_{xx}$) being very large (small). In fact, FQHE has never been observed in Si MOSFETs because of their high disorder level. Only the highest mobility 2D GaAs (and graphene) samples typically show many FQHE fractions in the lowest Landau level at high magnetic field, with the system eventually crossing over from the FQHE ground state to a SI ground state at very low filling fractions ($\sim 1/7$), which is sometimes interpreted as a transition to the Wigner crystal phase, but could very well be simply a strongly localized Anderson insulator arising from the effective disorder becoming very strong at low carrier densities Tsui1988 ; Cunningham1990 . It is interesting to note in this context that Refs. JuEQAH2024, ; PentaGraphene2023, also manifest a strongly localized insulating phase (for $\nu<2/5$) over a large regime of $D$ and $\nu$ values, indicating the importance of disorder in PLG QAHE/FQAHE physics.

Finally, while we have focused on the intriguing temperature-induced FQAHE to QAHE crossover phenomenon reported in Ref. JuEQAH2024, , we mention the interesting low-temperature, displacement-field-dependent feature observed in the same work (see Figs. 4(i)-4(k) therein). The displacement field induces a quantum phase transition between FQAHE and QAHE for several fractional $\nu$s at the lowest temperature ($T\sim 10$ mK). Such a direct quantum phase transition between FQHE and QHE has never been reported in the high-field 2D quantum Hall literature (e.g. GaAs), but is theoretically allowed here because the $D$ field modifies the band structure (and thus, the quantum metric), enabling the phase transition as shown in Figs. 4(i)-4(k) of Ref. JuEQAH2024, . This is thus akin to something like changing the effective mass (or perhaps the effective interaction) in the corresponding regular high-field QHE problem (which is not experimentally feasible since there is no available knob for continuously tuning the effective mass or interaction). On the other hand, changing $D$ also modifies the effective disorder, which can induce a phase transition between the SI phase and the FQAHE/EQAHE phase, as also shown in Figs. 4(i)-4(k) and Figs. 4(a)-4(b) of Ref. JuEQAH2024, at lower $D$ values. Such SI-FQHE transitions are often reported in regular high-field 2D QHE systems (as we have discussed above already) by varying $\nu$, and are interpreted as Wigner crystallization or Anderson localization. Very interestingly, in the zero-field PLG system, these transitions between WC/AI phases and QAHE/FQAHE phases can be induced by either tuning $\nu$ or $D$. The tunability by the displacement field allows for a much richer phenomenology in PLG than in the high-field 2D systems. We emphasize that although there is no temperature induced quantum phase transition in Ref. JuEQAH2024, , $D$ and $\nu$ induced quantum phase transitions at $T=0$ are indeed observed in Refs. JuEQAH2024, and PentaGraphene2023, .