Perfect Coulomb Drag in Excitonic Bilayers

• Perfect Coulomb Drag is a transport regime where current in one layer induces an almost equal, oppositely directed current in the adjacent layer via excitonic interactions.
• Experimental studies in GaAs quantum Hall systems and TMD heterostructures demonstrate near-unity drag ratios under controlled conditions with negligible tunneling.
• Theoretical models using conductivity matrices and circuit analogies reveal that perfect drag arises when exciton transport dominates and residual charged conduction is minimized.

Perfect Coulomb drag denotes a transport regime in which a current driven in one electrically isolated conductor induces a current in a second conductor with the same magnitude and opposite sign, so that $I_{\rm drag}\approx -I_{\rm drive}$ and $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$. In the literature surveyed here, that regime is most closely associated with excitonic bilayers, where transport is carried by neutral interlayer electron–hole pairs and therefore appears electrically as counterflow in the two layers (Nandi et al., 2012). The broader review literature also uses the term for an ideal clean-limit Drude model of two capacitively coupled conductors, but the experimentally consequential meaning is the excitonic one: drag ceases to be a weak frictional correction and becomes the dominant low-energy transport channel (Narozhny et al., 2015).

1. Definition and conceptual scope

Ordinary Coulomb drag is the induction of a voltage, or in a closed passive circuit a current, in one conductor by a current flowing in another, with the two conductors coupled only by Coulomb interaction. In standard notation, the open-circuit drag response is written as $R_D=-V_2/I_1$, and in translationally invariant systems one often uses the drag resistivity $\rho_D=-\rho_{xx}^{(12)}$ (Narozhny et al., 2015). In ordinary weakly coupled Fermi liquids, screening strongly suppresses the effect, so the induced response is much smaller than the drive response (Nandi et al., 2012).

Perfect Coulomb drag is qualitatively different. In the excitonic formulation, a moving interlayer exciton consists of an electron current in one layer paired with a hole current in the other, so exciton motion necessarily generates equal counterflowing electrical currents. The GaAs bilayer quantum Hall work states this directly as “a transport current of electrons driven through one layer is accompanied by an equal one of holes in the other,” which in electrical-current language implies $I_2=-I_1$ and $|I_2|=|I_1|$ in the ideal limit (Nandi et al., 2012). The review literature distinguishes this longitudinal current-locking phenomenon from quantized Hall drag, even though both can occur in the same interlayer-coherent $\nu_T=1$ quantum Hall state (Narozhny et al., 2015).

A second conceptual distinction is equally important: perfect drag is not identical to dissipationless transport. The MoSe$_2$/WSe$_2$ excitonic-insulator studies explicitly report regimes with nearly unit drag efficiency while emphasizing that this does not by itself prove exciton superfluidity or zero counterflow resistance (Nguyen et al., 2023). The optical transport study sharpens that point by observing perfect Coulomb drag together with a finite exciton counterflow resistance $R_x$, and therefore interpreting the state as an exciton gas in an excitonic insulator rather than a superfluid condensate (Qi et al., 2023).

2. Quantum Hall bilayers and the original experimental realization

The canonical experimental realization of perfect Coulomb drag is the strongly correlated $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$0 bilayer quantum Hall state in a GaAs/AlGaAs electron bilayer. The device used two $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$1 GaAs quantum wells separated by a $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$2 $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$3As barrier, giving a center-to-center separation $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$4, and was patterned into a Corbino annulus so that transport between inner and outer rims necessarily probed the bulk rather than quantum Hall edge channels (Nandi et al., 2012). In the balanced case $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$5, when $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$6 at total filling factor $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$7, the bilayer enters the interlayer-coherent quantum Hall phase with quantized Hall resistance $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$8, a gap for charged excitations, and neutral excitonic transport across the bulk (Nandi et al., 2012).

Within that regime, the Corbino geometry is decisive. The inner and outer edges are disconnected, so an equal drag current cannot be attributed to conventional edge transport. The experimental logic was therefore to drive current through one layer, close the second layer through an external resistor, and test whether bulk transport enforces equal counterflow. At low bias and low temperature, the measured drive and drag currents were nearly equal, with a reported drag ratio $\left|I_{\rm drag}/I_{\rm drive}\right|\to 1$9 under the best conditions, and the authors described the low-bias response as “essentially perfect” (Nandi et al., 2012).

The same experiment also addressed the main trivial alternative, namely direct interlayer tunneling. By tilting the sample to $R_D=-V_2/I_1$0, the Josephson-like zero-bias tunneling anomaly of the $R_D=-V_2/I_1$1 state was “essentially obliterated.” In the drag configuration the measured interlayer voltage remained below about $R_D=-V_2/I_1$2 even when the intralayer drive voltage reached $R_D=-V_2/I_1$3, implying an estimated tunnel current $R_D=-V_2/I_1$4, negligible compared with the drag current (Nandi et al., 2012). This established that the nearly equal currents were not produced by charge tunneling between layers.

The review article places this result in a larger transport framework. In the interlayer-coherent $R_D=-V_2/I_1$5 state, the symmetric channel carries the quantum Hall response while the antisymmetric channel is superfluid-like, with $R_D=-V_2/I_1$6. In Hall-bar language this yields quantized Hall drag and vanishing Hall voltage in counterflow; in Corbino drag-current geometry it yields the stronger statement of “perfect” longitudinal drag, namely equal-magnitude opposite-sign currents in the two layers (Narozhny et al., 2015).

3. Zero-field excitonic insulators in transition-metal dichalcogenide bilayers

A second major setting for perfect Coulomb drag is the zero-magnetic-field excitonic insulator in Coulomb-coupled MoSe$R_D=-V_2/I_1$7/WSe$R_D=-V_2/I_1$8 double layers. In the electrical transport experiment, the device was a dual-gated MoSe$R_D=-V_2/I_1$9/hBN/WSe$\rho_D=-\rho_{xx}^{(12)}$0 heterostructure with type-II band alignment, a $\rho_D=-\rho_{xx}^{(12)}$1–$\rho_D=-\rho_{xx}^{(12)}$2 hBN barrier in the channel, thicker $\rho_D=-\rho_{xx}^{(12)}$3–$\rho_D=-\rho_{xx}^{(12)}$4 spacers in the contact regions, and separate Bi and Pt contacts to the electron and hole layers, respectively (Nguyen et al., 2023). Penetration and interlayer capacitance measurements identified a triangular region in $\rho_D=-\rho_{xx}^{(12)}$5 space that was charge insulating but exciton-compressible, with an inferred exciton binding energy of about $\rho_D=-\rho_{xx}^{(12)}$6, a reduced channel gap of about $\rho_D=-\rho_{xx}^{(12)}$7, and a Mott density $\rho_D=-\rho_{xx}^{(12)}$8 (Nguyen et al., 2023).

In the closed-circuit drag-counterflow geometry, a small in-plane AC bias applied to the Mo layer produced a drag current in the W layer with opposite sign, and the drag efficiency $\rho_D=-\rho_{xx}^{(12)}$9 remained above $I_2=-I_1$0 up to about $I_2=-I_1$1 at low pair density (Nguyen et al., 2023). At $I_2=-I_1$2, the drag ratio was essentially unity throughout the low-density excitonic-insulator region. The effect was sharply localized to equal electron and hole densities, and it collapsed abruptly once the pair density exceeded the Mott density, where the authors interpreted the system as undergoing Mott dissociation into an electron–hole plasma with only weak Fermi-liquid frictional drag $I_2=-I_1$3 (Nguyen et al., 2023).

The optical transport study examined a closely related MoSe$I_2=-I_1$4/hBN/WSe$I_2=-I_1$5 system by a different route. Because direct electrical transport in TMD bilayers is limited by large contact resistances, the authors drove transport capacitively between two device regions and used reflectance modulation to measure density oscillations in the electron and hole layers. They defined the drag ratio as $I_2=-I_1$6, and in the excitonic-insulator regime found that at $I_2=-I_1$7 the two responses had the same amplitude, the same phase, and the same cutoff frequency, which they identified as perfect Coulomb drag in dynamic transport (Qi et al., 2023). In that device, near-perfect drag persisted at low density with $I_2=-I_1$8 up to approximately $I_2=-I_1$9 (Qi et al., 2023).

These TMD results materially broaden the scope of perfect drag. Unlike the GaAs $|I_2|=|I_1|$0 experiments, they occur at zero magnetic field and in atomically thin, separately contacted electron–hole bilayers. At the same time, both studies delimit the phenomenon sharply: it is strongest at density balance, it weakens with density imbalance and increasing temperature, it disappears across the Mott transition, and it requires negligible interlayer tunneling (Nguyen et al., 2023).

4. Transport theory and equivalent-circuit descriptions

The theoretical description of perfect Coulomb drag is unusually transparent because the ideal limit can be stated directly at the circuit level. For the Corbino $|I_2|=|I_1|$1 bilayer, the transport picture of Su and MacDonald used by Nandi and collaborators assumes negligible bulk charge conductivity and dissipationless exciton transport. If $|I_2|=|I_1|$2 and $|I_2|=|I_1|$3 denote the total series resistances of the drive and drag circuits, including external resistors, 2DES-arm resistances, and contact resistances, then the ideal prediction is $|I_2|=|I_1|$4 (Nandi et al., 2012). Introducing a finite parallel-flow Corbino conductance $|I_2|=|I_1|$5 reduces the drag ratio to

$|I_2|=|I_1|$6

so imperfect drag is attributed primarily to residual charged bulk conduction rather than to failure of the exciton channel itself (Nandi et al., 2012).

The electrical TMD experiment formulates the same logic in a conductivity-matrix language. The layer currents obey

$|I_2|=|I_1|$7

where $|I_2|=|I_1|$8 is the exciton conductivity, $|I_2|=|I_1|$9 and $\nu_T=1$0 are free-carrier conductivities, and $\nu_T=1$1 is the ordinary frictional drag conductivity from unbound carriers (Nguyen et al., 2023). In the closed-circuit geometry with drag-layer contact resistance $\nu_T=1$2, the measured current ratio becomes

$\nu_T=1$3

so the drag ratio approaches unity whenever the drag resistance greatly exceeds the contact resistance (Nguyen et al., 2023). This is experimentally useful because it ties the current-locking criterion directly to a measurable open-circuit quantity.

The optical TMD work uses a closely related decomposition in terms of effective resistive channels. Its conductance matrix is

$\nu_T=1$4

with $\nu_T=1$5 and $\nu_T=1$6 the effective resistances of unpaired electrons and holes, and $\nu_T=1$7 the exciton counterflow resistance (Qi et al., 2023). In the ideal excitonic-insulator limit, $\nu_T=1$8 and $\nu_T=1$9, leaving only the counterflow channel and enforcing $_2$0. The significance of this formulation is that it separates perfect drag from superfluidity: perfect drag requires freezing out unpaired charged channels, whereas superfluidity would additionally require $_2$1 (Qi et al., 2023).

5. Contrast with strong but nonperfect drag

Perfect Coulomb drag is best understood by comparison with systems that exhibit large, optimized, or unusual drag without current locking. The following cases are representative.

Platform	Reported behavior	Relation to perfect drag
Coulomb-coupled double quantum dots (Sierra et al., 2019, Keller et al., 2016, Lim et al., 2016)	Fluctuation-driven or cotunneling-assisted drag; optimized near $_2$2; strong-correlation/Kondo physics suppresses low-bias drag	Explicitly not one-to-one current locking
Vertically integrated 1D wires (Laroche et al., 2010, Zheng et al., 2024)	Drag as large as $_2$3 of the drive voltage; simultaneous momentum-transfer and rectification components	Large drag signal, but not perfect drag
Ballistic or helical 1D wires (Dmitriev et al., 2012, Kainaris et al., 2016)	Forward scattering alone does not produce nonzero dc drag in ballistic wires; helical-edge drag usually vanishes as $_2$4 for $_2$5	Strong constraints against naive ideal drag
Clean double layers and double-layer graphene (Chen et al., 2015, Gorbachev et al., 2012)	Plasmon- and hydrodynamic-enhanced drag; giant anomalous drag near graphene double neutrality and large magnetodrag	Strong coupling precursor, not equal-current locking
Graphene/Josephson-array hybrid (Tao et al., 2020)	“Super” Coulomb drag with observed passive-to-active ratio $_2$6 and theoretical extrapolation $_2$7	Giant or super-unitary drag, not excitonic perfect drag

The quantum-dot literature is especially explicit on the distinction. The fluctuation-driven DQD theory shows that finite drag requires energy-dependent asymmetry in the drag-lead couplings, that the effect is inherently nonlinear with $_2$8 for proportional drive couplings, and that in the orbital-Kondo regime low-bias drag is suppressed rather than stabilized (Sierra et al., 2019). The large-$_2$9 three-state model of engineered drag currents goes further in isolating a purely cotunneling-enabled drag mechanism, but still reports only $_2$0, far from perfect conversion (Lim et al., 2016).

The 1D wire literature makes a different point. Closely spaced vertically integrated GaAs/AlGaAs quantum wires can show unusually strong drag, including values as large as $_2$1 of the drive voltage and both positive and negative drag regions (Laroche et al., 2010). However, later measurements on similarly spaced wires demonstrate that the observed signal can contain both current-reversal-odd momentum-transfer drag and current-reversal-even rectification drag, so a large signal alone does not establish ideal frictional current locking (Zheng et al., 2024). In the ballistic-wire kinetic theory, pairwise forward scattering alone gives $_2$2, and nonzero dc drag requires chirality-changing processes; in that sense, the theory directly rules out a naive perfect-drag scenario generated solely by forward scattering (Dmitriev et al., 2012).

Double-layer graphene reaches a different strong-coupling regime. With hBN spacers down to $_2$3, the two Dirac liquids enter the limit $_2$4, drag saturates instead of following weak-coupling $_2$5 expectations, the strongest zero-field drag occurs at double neutrality, and in magnetic field the magnetodrag can exceed $_2$6 at $_2$7 (Gorbachev et al., 2012). This suggests very strong interlayer correlations, and the authors explicitly frame it as a precursor to possible excitonic physics, but it is not a demonstration of equal current locking.

6. Misconceptions, diagnostics, and significance

A common misconception is that any unusually large drag is “perfect” drag. The literature does not support that identification. Perfect drag requires a specific transport signature: equal-magnitude opposite-sign currents, or an experimentally equivalent near-unity drag ratio in a geometry that excludes trivial alternatives. This is why the Corbino annulus was crucial in the GaAs experiment and why the TMD studies stress separate electron and hole contacts, equal-density tuning, and negligible interlayer tunneling (Nandi et al., 2012).

A second misconception is that perfect drag proves superfluidity. The zero-field TMD literature explicitly rejects that inference. One paper states that perfect drag demonstrates dominance of exciton transport but does not by itself prove dissipationless transport or exciton superfluidity, and the optical study reports finite counterflow resistance $_2$8 even when the drag ratio is essentially unity (Nguyen et al., 2023). This indicates that current locking can arise from transport by bound excitons even when the exciton fluid remains resistive (Qi et al., 2023).

A third misconception is that perfect drag is merely an extreme limit of ordinary screened Fermi-liquid friction. The contrast cases show otherwise. In ordinary drag, the induced response is typically much smaller than the drive response because screening suppresses interlayer coupling (Nandi et al., 2012). In perfect drag, the low-energy transport mode itself is changed: the relevant degrees of freedom are neutral excitons or an interlayer-coherent condensate, and electrical transport appears as counterflow. The review article therefore links the strongest form of perfect drag to collective interlayer coherence, especially the $_2$9 exciton-condensate regime with quantized Hall drag and perfect longitudinal drag in Corbino geometry (Narozhny et al., 2015).

The most reliable diagnostics follow from those distinctions. Suppression of direct tunneling is essential. Bulk-sensitive geometries are preferred over edge-dominated ones. Near-unity drag must be confined to the correlated phase and degrade in the expected way with temperature, density imbalance, increasing $R_x$0, or Mott dissociation. In 1D systems, current-reversal symmetry decomposition is additionally necessary because rectification can mimic strong drag without representing momentum-locked or excitonic transport (Zheng et al., 2024).

In this sense, perfect Coulomb drag is significant not because it is merely a large transresistive signal, but because it identifies the elementary transport channel of the bilayer. In the quantum Hall and excitonic-insulator realizations, the evidence points to a regime in which low-energy transport is excitonic and counterflowing rather than independent electron motion in the two layers (Nandi et al., 2012). That transport signature, rather than absolute magnitude alone, is what has made perfect Coulomb drag a central diagnostic of interlayer coherence in contemporary condensed-matter physics.