Energy Participation Ratio (EPR) in Quantum Circuits

• Energy Participation Ratio (EPR) is a metric that quantifies the fraction of a mode’s electromagnetic energy distributed among circuit elements, crucial for understanding surface loss and nonlinearity.
• EPR computation utilizes a two-step finite-element simulation approach to extract detailed energy fractions in both dielectric and inductive components, enabling accurate parameter extraction.
• EPR underpins quantum Hamiltonian construction by facilitating prediction of eigenfrequencies, Kerr nonlinearities, and dissipation rates across various superconducting circuit architectures.

The Energy Participation Ratio (EPR) is a central concept for the theoretical description, simulation, and characterization of multi-mode, distributed, and nonlinear superconducting circuits. The EPR formalism provides a quantitative measure of how the electromagnetic energy of a circuit mode is distributed among its physical elements—both dissipative and nonlinear. Initially developed to address surface loss in 3D transmon qubits, the formalism has since become a cornerstone for black-box quantization, parameter extraction, and optimization of superconducting quantum devices across architectures including transmon, fluxonium, and multimode systems (Wang et al., 2015, Minev et al., 2020, Yu et al., 2023, Yilmaz et al., 2024).

1. Formal Definition of Energy Participation Ratio

In its general form, the EPR quantifies the fraction of a given mode's electromagnetic energy residing in an element or region of interest. Two variants are widely adopted:

• Dielectric/Electric-Energy Participation Ratio: For a surface or thin dielectric region $i$ (e.g., metal–substrate, substrate–air, or metal–air interface), the participation ratio is

$$p_i = \frac{\int_{V_i} \epsilon_i |E(\mathbf{r})|^2\, dV}{\int_{V_{\text{total}}} \epsilon(\mathbf{r}) |E(\mathbf{r})|^2\, dV}$$

where $V_i$ is the region volume, $\epsilon_i$ its permittivity, and $E(\mathbf{r})$ the electric field distribution (Wang et al., 2015).

• Inductive-Energy Participation Ratio (IEPR): For element $n$, mode $m$,

$$r_{mn} = \frac{\mathcal{E}^I_{mn}}{\mathcal{E}^I_m}$$

where $\mathcal{E}^I_{mn}$ is the average inductive energy stored in element $n$ when mode $m$ is excited, and $\mathcal{E}^I_m$ is the mode's total average inductive energy. In circuit-theory terms,

$$p_{mj} = \frac{\frac{1}{2} L_j I_{mj}^2}{\mathcal{E}_{\mathrm{ind}}^{(m)}}$$

where $I_{mj}$ is the peak current through inductor $L_j$ for mode $m$ (Minev et al., 2020, Yu et al., 2023, Yilmaz et al., 2024).

EPRs obey universal constraints: each $p_{mj}$ is in $[0,1]$; $\sum_m p_{mj} = 1$ for each element $j$; and $\sum_j p_{mj} \leq 1$ for each mode $m$ (Minev et al., 2020).

2. Methodologies for EPR Computation

Surface and Dielectric EPR:

Numerically evaluating EPRs for thin, lossy layers on complex 3D geometries is accomplished using a two-step finite-element procedure (Wang et al., 2015):

1. Global 3D Simulation: A coarse-mesh eigenmode solution models the full device, treating interfaces as 2D sheets and extracting the large-scale electric fields.
2. Local Fine Simulations: Targeted finely-meshed simulations capture nm-scale field variation near critical regions (metal edges, junction leads). Universal scaling functions $f_i(x,z)$ are constructed for perimeter/edge regions.
3. Recombination: The total participation $p_i$ is calculated by integrating contributions from "interior" and "perimeter" regions, combining results from both simulation levels.

Inductive EPR (IEPR) Workflow:

IEPR is computed in the context of black-box quantization and Hamiltonian extraction (Yu et al., 2023, Minev et al., 2020):

1. 3D Electromagnetic Eigenmode Simulation: Full-chip CAD models are solved (e.g., in ANSYS HFSS) with Josephson elements modeled as zero-thickness sheets with lumped inductance boundary conditions.
2. Energy Extraction: Post-processing integrates field energies to obtain $\mathcal{E}_m^I$, $V_{mn}$ (peak voltage across each element), $\mathcal{E}_{mn}^I$ (inductive energy in $n$).
3. Participation Calculation: The raw matrix $r_{mn}$ is formed and used (with sign conventions) to construct the unitary transformation $U$ connecting bare and normal mode bases.

3. Role of EPR in Quantum Hamiltonian Construction

EPR methodology underpins the quantization of distributed Josephson circuits:

• Linear Part: After finite-element eigenmode analysis, the Hamiltonian's quadratic sector is diagonal in "normal mode" operators $a_m^\dagger, a_m$ with frequencies $\omega_m$; the transformation to the original ("bare") basis is encoded in the EPR/IEPR matrix.
• Nonlinear Terms: Each Josephson element's potential is Taylor-expanded:

$$H_{\mathrm{nl}} = \sum_j \sum_{p=3}^\infty E_j c_{jp} \varphi_j^p$$

where zero-point amplitudes $\varphi_{mj}$ for each mode–element pair are fixed by the EPR:

$\varphi_{mj}^2 = p_{mj} \frac{\hbar\omega_m}{2E_j}$

• Effective Kerr Hamiltonian: In weakly anharmonic systems (e.g., transmons), the resulting effective (dispersive, quartic) Hamiltonian is given by

$$H_{\mathrm{eff}} = \sum_m (\omega_m - \Delta_m)\hat{n}_m - \frac{1}{2}\sum_m \alpha_m \hat{n}_m(\hat{n}_m-1) - \sum_{n<m} \chi_{mn} \hat{n}_m \hat{n}_n$$

with cross-Kerr nonlinearities $$\chi_{mn} = \sum_j \frac{\hbar\omega_m\omega_n}{4E_j} p_{mj} p_{nj}$$ (Minev et al., 2020, Yilmaz et al., 2024).

For circuits with strong anharmonicity (e.g., fluxonium), truncated expansions are inadequate; the full $-E_J\cos(\varphi_j-\varphi_{\mathrm{ext}})$ nonlinearity, parameterized by EPR-extracted amplitudes, must be diagonalized numerically in a truncated Fock space (Yilmaz et al., 2024).

4. Experimental Validation and Device Parameterization

Empirical studies have demonstrated the predictive power of EPR-based approaches:

• Surface Loss Studies: A linear relationship between the inverse relaxation time $1/T_1$ and the total surface EPR $$p_{\mathrm{surf}} = p_{\mathrm{MS}} + p_{\mathrm{SA}} + p_{\mathrm{MA}}$$ has been established for 3D transmon qubits:

$$\frac{1}{T_1} = \omega(2.6\times10^{-3})\,p_{\rm surf} + (3 \pm 1)\,\mathrm{ms}^{-1}$$

where the combined loss tangent, extracted from fit, lies near $2.6\times10^{-3}$ (Wang et al., 2015).

• Multi-mode Hamiltonians and Kerr Parameters: The EPR formalism allows efficient prediction of frequencies, nonlinearities (self- and cross-Kerr), and dissipation rates across multi-junction circuits. Experimental tests against a variety of geometries deliver agreement within a few percent for Kerr parameters and Hamiltonian eigenfrequencies over five orders of magnitude in energy (Minev et al., 2020).
• Highly Anharmonic Circuits: For fluxonium (Yilmaz et al., 2024), EPR-based predictions for both eigenfrequency and dispersive shift as a function of external flux accurately reproduce experimentally measured values. Improved fidelity is achieved over lumped-element approaches, particularly in regions of strong mode hybridization.

5. Loss, Dissipation, and Surface Participation

The EPR framework rigorously quantifies how loss in a superconducting circuit arises from dissipative mechanisms. Each lossy element $l$ (dielectric, metal, seam, port) is assigned a "loss EPR" $p_{ml}$ by integrating field energies over its region, yielding

$\frac{1}{Q_m} = \sum_l p_{ml} \frac{1}{Q_l}$

for the inverse quality factor of mode $m$ (Minev et al., 2020). In 3D transmons, dielectric surface loss—dominated by thin interfacial layers—currently sets the $T_1$ limit up to $Q \sim 10^7$. Reducing $p_i$ (by geometric engineering) and $\tan\delta_i$ (by improved fabrication) are both required for further progress (Wang et al., 2015).

A crucial subtlety is the "spatial discreteness" of surface-loss participation: in the immediate sub-micron region of the Josephson junction, the predicted $p_i$ may have no effect on $T_1$ if the volume is too small to contain a resonant two-level system (TLS) defect. Experimental fits exclude regions within $1\,\mu$m of the junction to avoid unphysical loss assignments—a consequence of the quantized, localized nature of dominant surface defects.

6. Extensions, Limitations, and Best Practices

Recent developments address the limitations of EPR methods for strongly nonlinear systems:

• For circuits with large phase excursions (e.g., fluxonium), the full Josephson cosine potential must be retained; EPR-extracted zero-point amplitudes are used directly in the Hamiltonian, which is then diagonalized numerically (Yilmaz et al., 2024).
• The inductive EPR (IEPR) generalizes the approach to distributed circuit elements beyond kinetic inductance and allows full recovery of the transformation between bare and normal representations, facilitating parameter extraction in both frames in a single eigenmode solve (Yu et al., 2023). IEPR further enables automated, scalable analysis for multi-qubit chips and distributed modes.
• Best practices include: always including capacitive nonlinearities if non-negligible, using sufficiently high Fock-space cutoffs for accurate diagonalization, and extracting unknown material parameters by fitting minimal experimental reference points.

Advances in EPR and IEPR allow for single 3D eigenmode simulations to deliver both the full quantum Hamiltonian and dissipation parameters for complex, high-coherence superconducting circuits (Yu et al., 2023, Yilmaz et al., 2024).

7. Representative Table: EPRs in Device Families

Empirically determined EPRs for surface participation in four distinct transmon families:

Design	$p_{\mathrm{MS}}$	$p_{\mathrm{SA}}$	$p_{\mathrm{MA}}$
A	$4 \times 10^{-5}$	$5 \times 10^{-5}$	$1 \times 10^{-6}$
B	$2 \times 10^{-5}$	$3 \times 10^{-5}$	$8 \times 10^{-7}$
C (1.5μm)	$5 \times 10^{-5}$	—	—
C (30μm)	$5 \times 10^{-6}$	—	—
D	$1 \times 10^{-5}$	$1.2 \times 10^{-5}$	$3 \times 10^{-7}$

These figures, obtained via the two-step simulation approach, illustrate order-of-magnitude reduction in EPR with increased device dimensions, directly mapping to lower dielectric loss (Wang et al., 2015).

---

The Energy Participation Ratio has evolved into a foundational metric for predicting and optimizing the behavior of superconducting quantum circuits, unifying classical simulation, quantum Hamiltonian construction, and experimental validation within a mathematically rigorous and computationally tractable framework (Wang et al., 2015, Minev et al., 2020, Yu et al., 2023, Yilmaz et al., 2024).