Dirac buffer layers reshape topological terahertz laser cavities for power and robustness

Topological lasers have long promised a kind of optical immortality: light modes that refuse to scatter backward, no matter how badly the underlying crystal is fabricated. Now a systematic simulation study has shown how to push that promise from a laboratory curiosity toward a practical, electrically pumped terahertz laser. The work, focused on valley Hall photonic crystal cavities designed for the ~3.4 THz quantum cascade laser platform, introduces and rigorously evaluates a Dirac-buffer-layer architecture that simultaneously flattens the lasing mode’s dispersion, enlarges its footprint across the gain medium, and shields it from the wavelength-scale defects that plague real devices. The results point to a specific design, a five-layer Dirac buffer with triangular air holes, as the optimal candidate for high-performance topological terahertz lasers.

The starting point is a well-known bottleneck. Valley Hall photonic crystals break spatial inversion symmetry, opening a photonic bandgap and hosting robust edge states at the boundary between two topologically distinct domains. Electrically pumped topological quantum cascade lasers exploiting these states have already been demonstrated, marking a critical milestone for on-chip integration. But the bulk-boundary correspondence principle dictates that most topological lasing modes remain evanescently confined to the interface between the two domains. In a typical terahertz quantum cascade laser waveguide roughly twenty lattice constants wide, that tight confinement leaves approximately ninety percent of the electrically pumped gain medium optically idle, directly limiting output power and wall-plug efficiency.

The edge-state dispersion compounds the problem. Conventional valley Hall boundary states are approximately linear, with group velocities on the order of vg/c ~ 0.1. Because the photonic density of states scales inversely with group velocity, this steep slope limits the light-matter interaction strength and raises the lasing threshold. Worse still, the electric-field intensity peaks of conventional valley Hall modes sit precisely at the domain wall, where structural perturbations can induce inter-valley scattering, standing-wave formation, and mode localization, threatening fabrication yield in practical devices.

The Dirac-buffer-layer concept, introduced recently under the name Dirac-scaling, offers an elegant escape. In this ABA’-type heterostructure, a gapless Dirac-point-degenerate photonic crystal lattice is inserted as a buffer between two gapped valley Hall domains carrying opposite valley Chern numbers. The central Dirac domain possesses C6 rotational symmetry and supports Dirac point degeneracy without a bandgap, so there is no evanescent decay requirement in this region. The topological mode extends across the entire Dirac buffer as a propagating wave, transforming from a conventional boundary state into a quasi-extended bulk mode. A perturbation at any given location now intercepts a smaller fraction of the total mode energy, diluting the impact of localized structural defects.

The new study goes beyond previous demonstrations by systematically mapping the design space with finite-element simulations in COMSOL Multiphysics. The researchers analyzed five primary cavity configurations, spanning Dirac layer numbers DDirac = 1, 3, and 5 with circular holes, and DDirac = 5 with triangular and square holes, complemented by band-structure calculations extending to DDirac = 2, 4, 6, and 8 for all three geometries. The passive cavity was modeled as a lossless dielectric with refractive index 3.6, with gold-coated top and bottom boundaries treated as perfect electric conductors enforcing the transverse-magnetic polarization required by the intersubband selection rule of the quantum cascade laser.

The headline result concerns group velocity engineering. When a Dirac buffer is introduced, the edge-state wavefunction acquires an increasing admixture of the Dirac material’s intrinsic eigenstates, which possess characteristically low velocity near the Dirac point. The effective group velocity of the hybridized mode becomes a weighted average of the fast valley Hall boundary component and the slow Dirac-bulk component. As the Dirac layer number increases from 2 to 8, the weight of the slow component grows, progressively pulling the group velocity downward. For triangular-hole Dirac lattices with DDirac = 8, the calculations reveal a near-flat-band region with |vg|/c < 0.02 between -2/3π < kxa < 2/3π, and |vg|/c < 0.005 at the band minimum, all within the quantum cascade laser gain bandwidth of 3.23-3.68 THz. In stark contrast, the edge-state dispersion for circular-hole Dirac lattices remains almost unchanged regardless of layer number, underscoring the critical importance of lattice geometry.

This flat-band slow-light effect carries direct threshold benefits. Achieving a group index ng = c/vg > 50 across the working wavevector range represents more than a four-fold enhancement in the photonic density of states compared with a conventional valley Hall waveguide, where ng ≈ 12.5 at vg/c ≈ 0.08. Because the threshold gain scales as gth ∝ 1/(QΓ), the density-of-states enhancement translates into reduced threshold carrier density, lower operating current, and reduced Joule heating, factors that are particularly critical for terahertz quantum cascade lasers, where thermal management remains one of the primary obstacles to continuous-wave and room-temperature operation. Crucially, the flat-band region naturally overlaps with the spectral region where gain is available, without requiring post-fabrication tuning of the lattice constant or operating point.

The Dirac buffer also acts as a transverse mode filter. Only the edge state whose in-plane wavevector and field symmetry match the Dirac continuum can hybridize efficiently and form a slow-light mode with a high quality factor; mismatched transverse modes radiate into the surrounding gapped domains and exhibit much lower Q factors. The full cavity simulations show a dramatic modal transformation: the conventional DDirac = 0 cavity has a densely populated bandgap with no mode exceeding Q > 10^5, while introducing a Dirac buffer raises the count of high-Q modes and pushes the maximum Q factor to 6.76 × 10^5 for the DDirac = 5 triangular-hole design. The triangular geometry consistently outperforms circular and square holes, achieving both the highest Q factors and the sparsest mode distribution, likely because its lower rotational symmetry introduces additional scattering channels that more effectively couple boundary states to the slow Dirac modes while suppressing parasitic bulk defect modes. A sparse mode spectrum reduces the risk of mode competition and facilitates single-mode operation.

Power scaling follows from the expanded mode profile. In the DDirac = 5 configurations, the electric field fills the entire Dirac buffer region, forming a broad, quasi-uniform intensity plateau spanning the full waveguide width, so gain material across the whole pumped region contributes to the lasing mode. This directly enables the >10× power enhancement observed experimentally in previous work, and unlike simply widening a trivial waveguide, which would introduce competing higher-order transverse modes, the Dirac-buffer design suppresses those modes through its filtering mechanism. Steady-state current-density simulations add an electrical dimension: the near-insulating etched holes force the injected current to spread laterally, producing a more uniform current profile that alleviates current crowding, space-charge screening, and Joule-heating concentration, all major limiting factors for maximum achievable output power in narrow-channel devices.

Defect robustness was quantified by deliberately introducing three wavelength-scale air defects at random positions in each cavity. For the conventional DDirac = 0 cavity, the defects caused a moderate Q reduction of about 6.5 percent but also a large frequency shift of ~44 GHz and a strongly distorted field pattern, showing that Q alone does not fully characterize robustness. The DDirac = 5 triangular-hole cavity, by comparison, retained 94.8 percent of its pristine Q factor with a frequency shift of only 4.3 GHz and negligible field distortion. The geometry- and layer-number dependence was pronounced, with circular designs at DDirac = 1 and 3 showing larger Q reductions and frequency shifts of 58.5 and 56.2 GHz respectively. Collectively, the findings establish the DDirac = 5 triangular-hole design as the optimal candidate for practical terahertz quantum cascade laser cavity engineering, combining effective mode filtering, flat-band-enabled low-threshold operation, expanded-mode-area high-power capability, and the highest defect tolerance among all configurations tested. The design rules, the authors note, should also transfer to Dirac-buffer-layer topological cavities at other wavelengths, offering a viable pathway toward practical high-performance topological lasers under electrical pumping.

Subject of Research: Dirac-buffer-layer valley hall photonic crystal cavities for group velocity engineering and power scaling with terahertz quantum cascade lasers

Article Title: Dirac-buffer-layer valley hall photonic crystal cavities for group velocity engineering and power scaling with terahertz quantum cascade lasers

Article References: Sun, Z., Yao, S., Yao, X., Feng, J., Zhong, C., Ding, S., Deng, S., Li, N., Li, H., & Lu, J. (2026). Dirac-buffer-layer valley hall photonic crystal cavities for group velocity engineering and power scaling with terahertz quantum cascade lasers. Results in Optics, Article 101194. https://doi.org/10.1016/j.rio.2026.101194

Image Credits: AI Generated

DOI: 10.1016/j.rio.2026.101194

Keywords: topological photonics, terahertz lasers, quantum cascade lasers, valley Hall photonic crystals, Dirac buffer layer, slow light, photonic crystal cavities, group velocity engineering, Dirac-buffer-layer, valley, hall, photonic