First-Principles Wannier Representation of Proximity Effects

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "First-Principles Wannier Representation of Proximity Effects".

Mira: Here's the gist of the paper. This work introduces a dynamical proximity operator derived from density functional theory to represent proximity effects in layered heterostructures, moving beyond static fitting methods to capture energy dependence, momentum transfer, and spatial structure.

Kai: Overcoming Limitations of Static Projections Proximity effects in layered heterostructures are usually represented by static parameters fitted to first-principles bands, which discards the energy dependence of the virtual hybridization, the momentum transfer, and the spatial structure <ref:2607.25690#pg2> The construction separates direct matrix elements from virtual hybridization through all remaining states <ref:2607.25690#pg2>.

Mira: This dynamical approach is superior because it reproduces the spectrum exactly within a fixed low-energy target Wannier subspace, overcoming the limitations of static projections <ref:2607.25690#pg2>.

Kai: The Downfolding Scheme The method evaluates the proximity coupling as an operator in a fixed low-energy Wannier subspace, without fitting to bands or re-wannierizing the heterostructure <ref:2607.25690#pg2>.

Title and authors: Kai: So we're looking at this paper, "First-Principles Wannier Representation of Proximity Effects," which is trying to move away from those old static fitting methods for layered heterostructures. It sounds like they are building something more dynamic.

Mira: Exactly. Instead of just picking some numbers that fit the bands in a static way, they're deriving this dynamical proximity operator, V(k, k'; omega), directly from density functional theory by downfolding the Kohn-Sham Hamiltonian onto a fixed low-energy target Wannier subspace.

Lev: So it’s not just about fitting parameters; it’s about calculating this operator across frequency and momentum space within that specific subspace, which is interesting because that immediately raises questions about how much overhead we need to manage on actual hardware.

Kai: Right, so the core idea is that they separate the direct matrix elements from virtual hybridization by looking at all remaining states, which they say overcomes the limitations of static projections <ref:2607.25690#pg2>.

Mira: They emphasize that this dynamical approach reproduces the spectrum exactly within a fixed low-energy target Wannier subspace, and they do this without needing to re-wannierize the entire heterostructure <ref:2607.25690#pg2>.

Lev: That's a big deal for implementation, because if we were to try and run this on a real quantum processor, we’d need to make sure that the downfolding process is tractable without introducing too much noise or computational complexity in the control layer.

Kai: So they describe this downfolding scheme where they split the direct projection from virtual hybridization using all remaining states, which gives them an energy-dependent effective Hamiltonian V(z) where z is a complex frequency <ref:2607.25690#pg2>.

Mira: And what’s key there is that since the P subspace is invariant under H0, the coupling between the P and Q sectors comes entirely from V, with the first term being direct projection and the second being renormalization through virtual hybridization with Q <ref:2607.25690#pg2>.

Lev: If we think about running this on a real system, that structure means we are dealing with coupling between distinct sectors that need careful handling when mapping it onto physical gates or couplings in our error-correction codes.

Kai: The paper then gives these downfolded matrix elements as V mn(k one k two; i omega) = w m k one V w n k two + X k three w m k one V Q psi n k two (k three i omega) <ref:2607.25690#pg2>.

Mira: That formula clearly shows the retention of the full momentum, spatial structure, and energy dependence of the coupling through that term involving X k three <ref:2607.25690#pg2>.

Title and authors: Lev: Retaining that full structure means we aren't losing information about intervalley or momentum transfer just because we are restricting ourselves to a low-energy subspace; it keeps the physics intact, which is what we need when trying to understand transport mechanisms.

Kai: They demonstrate this on three graphene-based heterostructures, isolating different aspects of proximity effects <ref:2607.25690#pg2>.

Mira: On graphene on hBN/Co(one), they show that the virtual hybridization generates more than ninety-nine percent of the proximity exchange, and it has a resonant frequency dependence set by the Co d states <ref:2607.25690#pg2>.

Lev: Ninety-nine percent is high, but when you’re trying to model things for real hardware where fidelity matters, that percentage tells you how much of your effective Hamiltonian comes from those virtual processes versus the direct ones.

Kai: Then there's graphene on PtSe2, which resolves a sublattice-selective intervalley coupling associated with a sqrt three times sqrt three charge modulation <ref:2607.25690#pg2>.

Mira: And for graphene on WSe2, they reveal a bond-resolved Rashba coupling of zero point two four meV against below one microvolt for the direct projection alone <ref:2607.25690#pg2>.

Lev: A zero point two four meV is relatively small compared to some of the larger effects we see in magnetic systems, but resolving that bond-resolved coupling spatially is important because it tells us how local structure dictates the spin-orbit interaction.

Kai: In graphene on hBN/Co(one), they find that the exchange splitting of order ten meV originates from the hybridization of Co d orbitals with graphene pz states and varies with the number of hBN layers and their stacking order <ref:2607.25690#pg3>.

Mira: They also found that in that system, more than ninety-nine percent of the proximity exchange arises from virtual hybridization with Co and hBN bands in Q <ref:2607.25690#pg3>. Spin-flip elements, which carry the proximity spin-orbit coupling, stay below zero point one five meV, two orders of magnitude beneath the exchange splitting, and they're absent from the direct projection <ref:2607.25690#pg3>.

Lev: So for a real system simulation, that means if you were looking for spin-flip effects in a qubit coupling model based on this, those elements are likely negligible compared to the main exchange term unless you’re pushing very high fidelity requirements.

Kai: On graphene on PtSe2, the intervalley coupling is spin conserving and fully sublattice polarized at the Fermi level, with its only sizable elements being diagonal on the A sublattice <ref:2607.25690#pg3>.

Mira: They note that this coupling reduces to a Kekulé proximity-induced Hamiltonian H K = lambda K tau x s zero one/two (sigma zero + sigma z) where lambda K is four point seven eight meV <ref:2607.25690#pg3>.

Title and authors: Lev: A coupling defined by a specific Hamiltonian like that one is much easier to analyze when mapping it onto an error model, because you can see exactly how the structure translates into the effective qubit interaction terms.

Kai: For graphene on WSe2, they resolve spin-dependent hopping over graphene bond by bond <ref:2607.25690#pg3>.

Mira: The resulting map in Figure three(c) spans spin-flip amplitudes from zero point one to zero point two nine meV across the supercell <ref:2607.25690#pg3>.

Lev: Spanning that range means you’ve got a whole spectrum of coupling strengths depending on where you measure in the lattice, which is crucial information for designing local control pulses.

Kai: And the frequency dependence in Figure three(d) is resonant at the WSe2 band edges, where graphene pz states hybridize with TMDC orbitals <ref:2607.25690#pg3>.

Mira: This means that when you look at the spectrum, you see those sharp resonances where the coupling is strongest, which directly relates to how the energy of your driving fields interacts with the system <ref:2607.25690#pg3>.

Lev: So what this gives us for a model is that we can generate these effective couplings without fitting anything, and those parameters—like that zero point two four meV Rashba coupling—are derived directly from the band structure, which is much more trustworthy than arbitrary inputs.

Kai: So to wrap up on "First-Principles Wannier Representation of Proximity Effects," this method gives us a dynamical proximity operator V(k, k'; omega) that retains direct matrix elements and virtual hybridization through eliminated orbitals <ref:2607.25690#pg1>.

Mira: It allows us to generate model Hamiltonians for graphene on PtSe2 and graphene on WSe2 that reproduce the downfolded bands with no adjustable parameters <ref:2607.25690#pg3>.

Lev: And that means we can take those derived couplings and feed them into T-matrix transport or spin-relaxation theory, which is how we actually predict what happens in a device, without needing any arbitrary fitting of the input parameters <ref:2607.25690#pg1>.

Kai: So for the listener who might be driving or cooking, this paper means we can get these effective couplings of proximitized materials from first-principles calculations rather than fitting them <ref:2607.25690#pg1>.

Mira: It’s a route from first-principles heterostructure calculations to effective low-energy models where the output is a dynamical proximity operator V(k, k'; omega) acting in the target Wannier subspace <ref:2607.25690#pg1>.

Lev: It supplies those effective couplings of proximitized materials rather than fitting them for T-matrix transport and spin-relaxation theory and correlated phase modeling <ref:2607.25690#pg1>.

Kai: We’ll keep an eye on this one, as it really helps us bridge the gap between the first-principles calculations we do and building something that actually runs on hardware.

The paper's summary: Kai: So, this paper is basically saying we need to stop using those static parameters for proximity effects in layered materials because they miss how things change with energy and momentum <ref:2607.25690#pg2>.

Mira: Right, they are building a dynamical proximity operator that comes straight from density functional theory by downfolding the Kohn-Sham Hamiltonian onto a fixed low-energy Wannier subspace <ref:2607.25690#pg1>.

Lev: I mean, it’s trying to reproduce the spectrum exactly within that fixed subspace, which is a huge technical challenge when you’re trying to model things for real hardware <ref:2607.25690#pg1>.

Kai: They show how they separate the direct matrix elements from the virtual hybridization using all remaining states <ref:2607.25690#pg1>.

Mira: That separation lets them keep the full momentum, spatial structure, and energy dependence of the coupling in this downfolded Hamiltonian V(z) where z is a complex frequency <ref:2607.25690#pg1>.

Lev: So they’re not just doing some simple fitting; they’re keeping the physics of intervalley scattering and momentum transfer intact, which means we don't lose information about how electrons move through the layers <ref:2607.25690#pg1>.

Kai: They show this on graphene on hBN/Co(one), where they find that the virtual hybridization of Co d states creates more than ninety-nine percent of the proximity exchange and it has a frequency dependence tied to those Co d resonances <ref:2607.25690#pg3>.

Mira: And what’s interesting is that while most of the effect comes from that virtual hybridization, the spin-flip elements which carry the proximity spin-orbit coupling stay very small, below zero point one five meV <ref:2607.25690#pg3>.

Lev: That means if you’re modeling something like a qubit interaction, those tiny spin-flip terms are way smaller than the main exchange splitting, so they might be negligible for your error model unless you’re pushing extreme precision <ref:2607.25690#pg3>.

Kai: They also look at graphene on PtSe2 and find a sublattice-selective intervalley coupling that is spin conserving and fully sublattice polarized at the Fermi level <ref:2607.25690#pg3>.

Mira: That specific coupling reduces to a Hamiltonian with a lambda K of four point seven eight meV, which means we can generate the model Hamiltonians for those systems with no adjustable parameters <ref:2607.25690#pg3>.

Lev: No adjustable parameters is what we want when you’re trying to build an effective Hamiltonian for a system because it means the physics is derived from first principles, not just tuned to fit a few data points <ref:2607.25690#pg3>.

Kai: And on graphene on WSe2, they resolve the spin-dependent hopping over graphene bond by bond and show that the real space structure of this coupling maps onto spin-flip amplitudes ranging from zero point one to zero point two nine meV across the supercell <ref:2607.25690#pg3>.

Mira: They also pinpoint that the frequency dependence in their results is resonant at those WSe2 band edges where graphene pz states hybridize with TMDC orbitals <ref:2607.25690#pg3>.

Lev: So for someone who only listens to the show, this means we have a way to take these complex, first-principles results and turn them into effective couplings for transport or spin relaxation theory using T-matrix methods <ref:2607.25690#pg1>.

Kai: This method gives us a dynamical proximity operator V(k, k prime; omega) that retains direct matrix elements and virtual hybridization through eliminated orbitals <ref:2607.25690#pg1>.

Mira: The real output is the ability to generate model Hamiltonians for graphene on PtSe2 and WSe2 that reproduce the downfolded bands with no adjustable parameters <ref:2607.25690#pg3>.

Lev: And those derived couplings are exactly what you need to feed into T-matrix transport or correlated phase modeling without needing any arbitrary fitting of input parameters <ref:2607.25690#pg1>.

Kai: This is a route from first-principles heterostructure calculations to effective low-energy models where the output is this dynamical proximity operator acting in the target Wannier subspace <ref:2607.25690#pg1>.

Mira: It supplies those effective couplings of proximitized materials rather than fitting them for T-matrix transport and spin-relaxation theory and correlated phase modeling <ref:2607.25690#pg1>.

Lev: It really means we can get these effective couplings of proximitized materials from first-principles calculations rather than fitting them <ref:2607.25690#pg1>.

Kai: So for the listener who might be driving or cooking, this paper means we can get these effective couplings of proximitized materials from first-principles calculations rather than fitting them <ref:2607.25690#pg1>.

The paper's improvements: Tom: So we’re looking at how these authors suggest they can take this dynamical operator and actually use it to build something practical <ref:2607.25690#pg1>.

Kai: They argue that what they did, deriving this V operator from DFT, is a great starting point but you still need a way to connect it to the actual hardware we’re building with superconducting qubits <ref:2607.25690#pg1>.

Mira: Right, they suggest that the next step is projecting this operator onto a low-energy model Hamiltonian, which means taking that full momentum and frequency structure and mapping it down to something manageable <ref:2607.25690#pg3>.

Lev: That’s where the overhead gets real, because if you try to simulate all that spectral detail directly on a real chip, the computational cost explodes unless you simplify it drastically <ref:2607.25690#pg1>.

Kai: They emphasize that by using this Wannier downfolding scheme, they can generate model Hamiltonians for graphene on PtSe2 and WSe2 that reproduce the bands with no adjustable parameters at all <ref:2607.25690#pg3>.

Mira: So the improvement is generating parameter-free effective Hamiltonians that directly map to band structures, which is a huge deal for testing theories without needing arbitrary fitting <ref:2607.25690#pg1>.

Lev: If you can generate that Hamiltonian directly from DFT, then when we feed it into T-matrix transport or spin relaxation theory, we’re using physics derived from first principles instead of just tuning parameters to match some noisy measurement <ref:2607.25690#pg1>.

Kai: They also point out that this operator is the input for things like T-matrix transport and correlated phase modeling which is exactly what we need to see how these materials behave in a real device <ref:2607.25690#pg1>.

Mira: The implication there is that this isn't just a calculation exercise; it’s providing the effective couplings for things like T-matrix transport and spin relaxation theory and correlated phase modeling <ref:2607.25690#pg1>.

Lev: So, for us in error correction, this means we can use these derived couplings to design better error models without having to guess those interaction strengths <ref:2607.25690#pg1>.

Kai: The limitation they flag is that while the method retains a lot of information—the momentum, the real space texture, the energy dependence—it stops short of describing things outside that fixed low-energy target subspace <ref:2607.25690#pg1>.

Mira: That’s fair; it's designed to be an effective model for a specific regime rather than a complete picture of every possible interaction in the system <ref:2607.25690#pg3>.

Lev: So, they’re building a tool that bridges the gap between first-principles calculations and actual device simulation by providing these high-fidelity effective couplings for T-matrix transport and spin relaxation theory <ref:2607.25690#pg1>.

Kai: It's a route from first-principles heterostructure calculations to effective low-energy models where the output is this dynamical proximity operator V(k, k prime; omega) acting in the target Wannier subspace <ref:2607.25690#pg1>.

Conclusion: Kai: So to wrap up, this paper on "First-Principles Wannier Representation of Proximity Effects" introduces a dynamical proximity operator <ref:2607.25690#pg1>.

Mira: It’s basically taking those first-principles calculations and turning them into an effective low-energy model where the output is this operator V(k, k prime; omega) acting in the target Wannier subspace <ref:2607.25690#pg1>.

Lev: And that operator retains direct matrix elements and virtual hybridization through eliminated orbitals <ref:2607.25690#pg1>, which is crucial for figuring out how much overhead we’d need to manage on real hardware <ref:2607.25690#pg1>.

Kai: They showed they can generate model Hamiltonians for graphene on PtSe2 and WSe2 that reproduce the bands with no adjustable parameters at all <ref:2607.25690#pg3>.

Mira: And those derived couplings are exactly what you need to feed into T-matrix transport or spin relaxation theory without needing any arbitrary fitting <ref:2607.25690#pg1>.

Lev: So, for us in error correction, this means we can use these derived couplings to design better error models without having to guess those interaction strengths <ref:2607.25690#pg1>.

Kai: This is a route from first-principles heterostructure calculations to effective low-energy models where the output is this dynamical proximity operator V(k, k prime; omega) acting in the target Wannier subspace <ref:2607.25690#pg1>.

Mira: It supplies those effective couplings of proximitized materials rather than fitting them for T-matrix transport and spin relaxation theory and correlated phase modeling <ref:2607.25690#pg1>.

Lev: It really means we can get these effective couplings of proximitized materials from first-principles calculations rather than fitting them <ref:2607.25690#pg1>.

Kai: We’ll keep an eye on this one, as it really helps us bridge the gap between the first-principles calculations we do and building something that actually runs on hardware <ref:2607.25690#pg1>.