Peculiarities Of High-Speed Dynamics Of Two-Photon Absorption In Si Nanowire Waveguides

Technical Summary: Peculiarities Of High-Speed Dynamics Of Two-Photon Absorption In Si Nanowire Waveguides

Problem Statement
The integration of dense CMOS electronics with high-speed photonic components has enabled on-chip data transmission approaching 10 Tbit/s. However, as optical power and bit rates increase, two-photon absorption (TPA) in silicon nanowire waveguides becomes a critical limiting factor. TPA induces nonlinear optical loss that can substantially exceed linear loss, degrading the bit error ratio (BER) and restricting both achievable data rates and the maximum usable length of silicon waveguides.

A fundamental challenge in mitigating TPA is the unclear physical origin of the "virtual midgap level" invoked in conventional models. Standard quantum mechanics describes TPA as a process where an electron transitions from the valence band to a virtual state (violating energy conservation momentarily) before absorbing a second photon to reach the conduction band. However, this model fails to explain several experimental inconsistencies:

1. Decoupling of Linear and Nonlinear Loss: Linear and nonlinear losses do not correlate as expected; for instance, TE modes exhibit lower linear loss but higher nonlinear loss than TM modes.
2. Inconsistency in Rates: Conventional models assume that nonlinear photon absorption, electron excitation, and free-carrier generation occur at identical rates. Previous studies have not systematically compared these rates across the full dynamical pathway.
3. Nature of the Virtual State: It remains uncertain whether the virtual level is a real quantum state with an extremely short lifetime, a field-dressed state created by the optical pulse, or if the process occurs via a mechanism (such as interband tunneling) that does not require an intermediate state at all.

Methodology
To resolve these discrepancies, the authors employed three independent high-speed measurement techniques to trace the complete dynamical pathway of photon-electron interactions in standard silicon nanowire waveguides (430 nm × 220 nm cross-section, 5 mm length). The study utilized 13 ps optical pulses at 1553 nm.

The experimental approach evaluated three distinct stages of the TPA process:

1. Nonlinear Photon Absorption: Measured by detecting the intensity loss of a single optical pulse passing through the waveguide. A specialized setup with two computer-controlled variable attenuators ensured that the total linear loss remained constant while varying the pulse power inside the waveguide, allowing for precise detection of small intensity changes.
2. Electron Excitation (TPA Transitions): Measured using a pump-probe technique. A continuous-wave (CW) probe beam (1559 nm) was co-propagated with the pump pulse. The TPA-induced loss of the CW beam was monitored, which corresponds to the rate of electron transitions from the valence to the conduction band.
3. Free-Carrier Generation: Measured by analyzing the residual optical loss of the CW beam after the pump pulse had passed. This residual loss, attributed to free-carrier absorption (FCA), was used to calculate the density of free carriers generated by the TPA process.

The authors calculated the "photon loss rate" for each stage, normalizing the data to the number of photons in the input pulse to allow for direct comparison of the conversion efficiencies across the three stages.

Key Results
The measurements revealed significant discrepancies between the three stages of the TPA pathway, contradicting the conventional model where these processes occur at identical rates:

• Discrepancy in Absorption vs. Excitation: The measured rate of nonlinear photon absorption was more than twice the rate required to account for the measured TPA transitions. This indicates that a substantial portion of absorbed photons does not result in successful electron transitions to the conduction band, suggesting the presence of additional absorption pathways or nontrivial TPA dynamics where electrons return to the valence band without completing the transition.
• Discrepancy in Excitation vs. Free-Carrier Generation: The number of TPA transitions substantially exceeded the measured free-carrier density. This implies that long-lifetime free carriers represent only a small fraction of the TPA-excited electrons. The majority of excited electrons recombine rapidly back to the valence band on a timescale shorter than 13 ps.
• Distinct Saturation Behaviors: The three stages exhibited saturation at different photon densities. Nonlinear photon absorption saturated at approximately 200 million photons per pulse, TPA transitions saturated at roughly 50 million photons per pulse, and free-carrier generation saturated at 20 million photons per pulse.
• Virtual Level Origins: The authors identify three physically plausible origins for the virtual midgap level: (1) ordinary mid-gap quantum states with extremely short lifetimes; (2) field-dressed dynamic states created by the combined oscillating electric field of the photons; or (3) a mechanism involving interband tunneling that bypasses the need for a virtual state. However, the study concludes that further investigation is required to determine the true nature of this level.

Significance and Claims
The paper claims to provide the first systematic investigation of the complete dynamical pathway of TPA in silicon, directly comparing conversion efficiencies across all three stages of the process. The primary significance of these findings lies in:

1. Challenging the Conventional Model: The results demonstrate that TPA in silicon is more complex than described by standard models, which assume a direct, one-to-one correspondence between absorbed photons, excited electrons, and generated free carriers.
2. Insight into Physical Mechanisms: The identification of rapid recombination pathways and the mismatch between absorption and excitation rates suggests the existence of multiple competing pathways for this optical transition.
3. Strategic Implications: Understanding these dynamics is essential for developing strategies to minimize unwanted nonlinear losses in high-speed silicon photonic circuits. Furthermore, if the properties of the virtual midgap level can be controlled (e.g., via interface states or electric fields), TPA could potentially be exploited for high-speed optical switching and signal processing on timescales faster than 100 fs, transforming a limiting effect into a functional advantage.

The authors maintain a modest stance regarding the specific nature of the virtual level, acknowledging that while they have identified plausible origins, the exact physical mechanism remains an open question requiring further study.