Self-evolving photonic crystals for ultrafast photonics

The study addresses how to achieve ultrafast control of optical phenomena in nanophotonic materials (photonic crystals, metamaterials) without bulky or slow external actuation. Prior dynamic control has used optical pulse irradiation or electrical pulse application, but high-power optical pumping is bulky and electrical pulsing is difficult on sub-nanosecond timescales, with challenges worsening for larger devices and higher optical powers. Alternatives such as phase-change materials or MEMS suffer slow responses. The authors propose a self-evolving photonic crystal in which the photonic band structure’s spatial profile dynamically changes via stimulated-emission-induced refractive index modulation, eliminating the need for external ultrafast stimuli. This enables spontaneous short-pulse generation with high peak power under constant current injection.

Previous works demonstrated dynamic tuning in nanophotonic structures via: (i) optical pulse irradiation requiring external high-power lasers; (ii) electrical pulse application, difficult at <ns timescales; and (iii) temperature-sensitive phase-change materials or MEMS, which are slow. Conventional Q-switched semiconductor lasers achieve pulsing but typically require multi-section electrodes or saturable absorbers. These limitations motivate a method that scales to large devices and high optical powers without external ultrafast control or slow mechanisms.

Concept and device design: A graded double-lattice photonic crystal is integrated near the active layer within a p–n junction of a photonic-crystal surface-emitting laser (PCSEL). The lattice constant is monotonically increased along the u-axis, creating a spatial gradation of the band-edge frequency. Double-lattice holes are shifted ~a/4 along x and y to form anti-symmetric (A, C) and symmetric (B, D) band-edge modes; the lasing mode (A) is selected for low radiation loss. Before lasing, the mode-gap from the gradation localizes the resonance at the periphery; during lasing, stimulated-emission-induced refractive index changes dynamically flatten the band-edge profile, allowing the mode to propagate across the injection region, amplifying as it accesses carrier-rich regions, then quenching when carriers deplete, repeating cyclically.

Numerical simulation: Time-dependent 3D coupled-wave theory is used. The photonic crystal comprises paired elliptical and circular holes; parameters are tuned for moderate radiation constant of the lasing mode (~10 cm^-1) and large threshold margin to higher-order modes (Δα≈9 cm^-1). A 1-mm-diameter current injection region is modeled. Band-edge frequency gradation is implemented via lattice constant gradients α₁ (u-axis), α₂ (v-axis), and β (inside/outside mismatch). α₁ sets the mode-gap magnitude; β controls in-plane loss outside the injection area; α₂ compensates carrier-induced index variation along v to narrow divergence. Simulations compare uniform PCSEL (α₁=α₂=β=0 nm) versus graded devices (e.g., α₁=α₂=0.22 nm, β=0.11 nm) at I=20 A. Temporal output, carrier density, band-edge frequency, and photon density distributions are computed to visualize pulse formation and propagation. Robustness over currents and gradient parameters, and tolerance to random frequency fluctuations, are assessed. Thermal index effects are treated as quasi-static (microsecond scale) and compensable via gradient design.

Fabrication: A GaAs-based 1-mm-diameter graded PCSEL is fabricated with the same process as conventional PCSELs (details in Methods). Two devices are made: monoaxial gradation (α₁=0.22 nm, α₂=0, β=0.11 nm) and biaxial gradation (α₁=α₂=0.22 nm, β=0.11 nm). Although the per-period lattice change Δα < 0.5 nm is below SEM resolution, cumulative positional shifts of lattice points across many periods reach tens to hundreds of nanometers, enabling accurate fabrication of the designed gradation via integrated Δα over m,n periods: Δx(m,n)=∑{m'=1}^m Δα(m',n), Δy(m,n)=∑{n'=1}^n Δα(m,n'). The average lattice constant is 274 nm (lasing wavelength ~936 nm).

Experimental characterization: Emission is coupled into a streak camera with the slit aligned parallel to the u-axis to capture spatiotemporal evolution. Spatial integration yields temporal output power waveforms. Current is swept, demonstrating pulsation behavior, pulse widths, peak and average powers, repetition frequency, and far-field divergence (θ₁/₂).

• Simulation: Uniform PCSEL under 20 A exhibits steady CW output after relaxation oscillations. Graded PCSEL (α₁=α₂=0.22 nm, β=0.11 nm) exhibits intermittent short-pulse trains with peak powers >100 W and pulse widths ≈30 ps under constant current. Simulated carrier, band-edge, and photon distributions confirm self-evolution: initial edge-localized mode transitions to center as index changes flatten the band-edge gradient, enabling propagation and amplification.
• Experiment: Biaxially graded 1-mm-diameter device (α₁=α₂=0.22 nm, β=0.11 nm) shows periodic pulse trains on the streak camera at I≈20.9 A, with pulses initiating from one edge and propagating along +u, matching simulations. Stable self-pulsation is observed across 8–20 A. Temporal output (e.g., at 20.6 A) shows pulse widths <30 ps. Maximum measured peak power exceeds 80 W, about 4× higher than the previous record among Q-switched semiconductor lasers without amplifiers. The far-field shows a narrow divergence θ₁/₂≈0.2°, with slight elongation along y due to incomplete compensation along v.
• Trends with current: Increasing injection current reduces pulse width and increases repetition frequency, attributed to faster carrier accumulation and faster band-edge evolution; consistent with simulations.
• Pulse compression potential: Calculations show nearly linear wavelength chirp during each pulse; for α₁=α₂=0.22 nm (β=0.11 nm, I=20 A), second-order dispersion compensation can compress pulses to 6.4 ps with peak power 588 W (near Fourier limit of 3.4 ps for Δλ=0.38 nm). Larger device diameter and higher gradients could enable >1 kW peak power after compression.

The work demonstrates that engineered carrier–photon interactions within a graded photonic crystal can self-evolve the photonic band structure on picosecond timescales, transitioning between high-loss and low-loss states without external ultrafast stimuli. This mechanism generates high-peak-power, sub-30-ps pulses under constant current injection, addressing key limitations of prior approaches (bulky optical pumping, slow electrical or thermal modulation). The observed spatiotemporal propagation from the edge across the injection region validates the designed mode-gap-mediated localization and its dynamic compensation via stimulated-emission-induced refractive index change. The substantial intrinsic wavelength chirp enables straightforward external dispersion compensation to further shorten pulses and boost peak power. The approach simplifies device architecture (no multi-section electrodes or saturable absorbers) while maintaining excellent beam quality. These results open pathways to scalable, high-power ultrafast semiconductor sources and broaden applications in ultrafast photonics.

The authors introduce and validate self-evolving photonic crystals that dynamically reshape their band-edge profiles via stimulated-emission-induced refractive index changes under uniform DC current. They realize spontaneous short-pulse lasing with peak power >80 W and pulse widths <30 ps from a 1-mm-diameter GaAs-based PCSEL, with narrow beam divergence. Simulations and experiments elucidate the self-evolution dynamics and robustness across operating conditions. Leveraging intrinsic chirp with dispersion compensation could compress pulses to ~6.4 ps and hundreds of watts peak power, with prospects for >1 kW by scaling device size and gradients. Future directions include optimizing band-edge gradation to mitigate higher-order dispersion, refining lateral index compensation for symmetric beams, exploring RF superimposition for repetition-rate control, and exploiting thermal/current engineering to realize or tune gradations. The concept promises compact, high-peak-power ultrafast sources for a wide range of scientific and industrial applications.

• Experimental peak power (~80 W) is lower than simulated (>100 W), likely due to differences in radiation constant and current injection uniformity, affecting slope efficiency.
• The designed sub-nanometer lattice-constant gradation cannot be directly resolved by SEM; only cumulative positional shifts are observable.
• Slight beam asymmetry (elongation along y) indicates incomplete compensation of carrier-induced refractive index distribution along the v-axis in the fabricated device.
• Thermal refractive index changes, while slow and considered static, may require careful compensation via gradient design.
• Results are demonstrated on specific device configurations and parameters; performance depends on gradient values (α₁, α₂, β) and fabrication precision.