Temporal Beam Self-Cleaning in Nonlinear Optics

• Temporal beam self-cleaning is a nonlinear phenomenon that purges high-contrast noise and sidelobes from laser pulses, resulting in cleaner temporal and spatial profiles.
• The process is modeled using generalized nonlinear Schrödinger equations in systems like GRIN fibers, air–plasma channels, and χ² frequency conversion, demonstrating significant pulse compression and energy redistribution.
• Practical implementations improve imaging resolution and ultrafast pulse generation, though they require precise power thresholds and alignments to overcome inherent technical challenges.

Temporal beam self-cleaning refers to a class of nonlinear spatiotemporal phenomena in which the temporal structure of a laser beam becomes purged of high-contrast noise, sidelobes, or multi-mode beating—evolving, through nonlinear processes, into a cleaner, more coherent temporal profile. This process has been observed in a variety of nonlinear optical systems, including graded-index (GRIN) multimode fibers (MMFs), air–plasma channels, and χ² frequency-conversion processes such as second-harmonic generation (SHG). Temporal self-cleaning typically occurs in concert with spatial self-cleaning effects, resulting in beams that exhibit both high spatial quality and enhanced temporal coherence, with substantial implications for advanced photonic applications in microscopy, ultrafast science, and high-power laser engineering.

1. Theoretical Models of Temporal Beam Self-Cleaning

The theoretical formalism underlying temporal beam self-cleaning varies with the physical system but is universally rooted in the nonlinear Schrödinger equation (NLSE) framework. In GRIN MMFs, the field envelope $A(x, y, t; z)$ obeys a generalized (3+1)D NLSE with Kerr nonlinearity and a parabolic refractive index potential:

$$i\,\frac{\partial A}{\partial z} + \frac{\beta_2}{2}\,\frac{\partial^2 A}{\partial t^2} - \frac{1}{2\,k_0}\,\nabla_\perp^2 A + k_0\,n_2\,|A|^2\,A + V(r)\,A = 0,$$

where $\beta_2$ is the group-velocity dispersion, $n_2$ the Kerr coefficient, and $V(r)$ the GRIN potential, giving rise to self-imaging cycles. When the input peak power $P$ exceeds a threshold $P_{\mathrm{th}}$, intermodal four-wave mixing redistributes energy toward lower-order modes, inducing both spatial and temporal beam cleaning (Moussa et al., 2020).

In air–plasma channels, the field evolution of the envelope $\mathcal{E}(x, y, t, z)$ is governed by the coupled NLSE and plasma equations:

$$i \frac{\partial\mathcal{E}}{\partial z} + \frac{1}{2k_0} \nabla_\perp^2 \mathcal{E} + \frac{k^{(2)}}{2} \frac{\partial^2 \mathcal{E}}{\partial t^2} - k_0 n_2 |\mathcal{E}|^2 \mathcal{E} - k_0 \frac{\rho}{2\rho_{c0}} \mathcal{E} = 0,$$

with plasma density $\rho$ dictating the negative dispersion, and Kerr terms providing self-compression and bandwidth (Xu et al., 2023).

In χ² media undergoing SHG, coupled-wave equations for the fundamental $E_f(z, t)$ and second-harmonic $E_s(z, t)$ fields model energy transfer and selective depletion of high-peak temporal features, with preservation of modal phases across the frequency conversion:

$$\left(\frac{\partial}{\partial z} + \frac{1}{v_f}\frac{\partial}{\partial t}\right)E_{f}(z,t) = i\,d_{\rm eff}\,\frac{\omega_f^2}{2\,c^2k_f}E_f^*(z,t)E_s(z,t)e^{-i\Delta kz}$$

$$\left(\frac{\partial}{\partial z} + \frac{1}{v_s}\frac{\partial}{\partial t}\right)E_{s}(z,t) = i\,d_{\rm eff}\,\frac{\omega_s^2}{4\,c^2k_s}E_f^2(z,t)e^{+i\Delta kz}$$

(Chen et al., 17 Jan 2026).

2. Physical Mechanisms of Temporal Self-Cleaning

In nonlinear fiber systems, temporal self-cleaning emerges from the interplay of Kerr self-focusing, periodic self-imaging, and nonlinear mode coupling. As the multimode field propagates through the GRIN fiber, spatial focusing cycles intensify nonlinear interactions at each imaging plane. This mechanism preferentially channels energy into the fundamental or low-order modes, leading over multiple cycles to a temporally compressed, bell-shaped pulse profile. Temporal pulse shortening results from self-phase modulation at the self-imaging foci, further enhanced under anomalous dispersion where soliton-like pulse compression occurs. Experimentally, an input pulse of ~80 ps can shrink to 25–30 ps, a compression factor $C \simeq 3.2$ (Moussa et al., 2020).

In air–plasma channels, plasma engineering modulates the dispersion to enable spatiotemporal soliton formation. After an initial self-compression and pulse splitting, two phenomena lead to temporal beam cleaning of the soliton: ionization-induced blue shift causes the soliton to overtake and absorb leading sidelobes (due to their lower group velocity), while plasma lensing defocuses and eliminates trailing sidelobes. This process generates single-cycle, sidelobe-free pulses with duration as short as 3.7–17.3 fs, with sidelobe suppression exceeding 30 dB (Xu et al., 2023).

In SHG self-cleaning, the mechanism is fundamentally statistical: strong conversion efficiency ($\eta_{\rm SHG} \propto |E_f|^2$) results in preferential depletion of high-intensity temporal pulses (high-peak attenuation), while linear phase preservation across modes ensures temporal coherence. This process reduces temporal fluctuations and enhances autocorrelation background, transforming noisy multi-mode beating into a temporally stable profile (Chen et al., 17 Jan 2026).

3. Experimental and Quantitative Evidence

Empirical studies have rigorously characterized temporal beam self-cleaning. In GRIN MMF systems, the onset of self-cleaning occurs above $P_{\mathrm{th}} \sim 5–10$ kW (1064 nm, $$n_2 \approx 3 \times 10^{-20}\,\mathrm{m}^2/\mathrm{W}$$), where the beam's near-field transition from a speckled multimode structure ($M^2 \gg 2$) to bell-shaped ($M^2 \simeq 1.3–1.5$), and the temporal profile compresses by a factor of 3 (Moussa et al., 2020).

In air–plasma channel experiments and simulations, precise preformed electron densities ($\rho_p$) as a function of wavelength ($0.8–3.0\,\mu$m) are required for optimal soliton self-cleaning. Clean, sub-two-cycle pulses are generated for $\rho_p$ spanning $1.4 \times 10^{16}$ to $2.9 \times 10^{17}\,\mathrm{cm}^{-3}$, with effective suppression of all pre- and post-pulse sidelobes (Xu et al., 2023).

In the SHG approach, quantitative metrics confirm the effect:

Regime	Std (FW → Residual FW)	PV (FW → Residual FW)	Autocorrelation Background (FW → Residual FW)
SLM	0.0143 → 0.0089	0.1320 → 0.0848	0.9998 → 0.9999
DLM	0.1875 → 0.0814	0.6120 → 0.2831	0.93 → 0.99
MLM	0.6122 → 0.1890	5.6846 → 0.8847	0.72 → 0.96

These results demonstrate order-of-magnitude reduction in temporal noise and large increases in temporal coherence (background tending to unity) (Chen et al., 17 Jan 2026).

4. Applications in Imaging and Ultrafast Science

Temporal beam self-cleaning enables substantial advances in nonlinear microscopy, endoscopy, ultrafast pulse generation, and coherent beam delivery. In spatiotemporal beam self-cleaning within MMFs, improved spatial and temporal profiles result in increased brightness and higher peak power, directly enhancing the efficiency of multiphoton (two- and three-photon) fluorescence imaging. Lateral resolutions as low as 0.37 µm (two-photon, SB-SC) and 0.54 µm (three-photon, soliton regime) are demonstrated, with robust output under fiber bending ($>$98% image correlation) and efficient operation at low average powers (5 mW on sample) (Moussa et al., 2020).

In air–plasma channels, the generation of sub-two-cycle, sidelobe-free pulses with peak powers of tens of GW and single-cycle duration vastly expands the utility of ultrafast pulses in time-resolved spectroscopy, attosecond science, and strong-field physics, without the need for complex pulse-shaping or amplifier hardware (Xu et al., 2023).

In SHG-based temporal self-cleaning, the process can be seamlessly integrated into MOPA laser systems to suppress temporal noise ahead of amplification stages, improving nonlinear threshold margins (e.g., for SBS/SRS suppression) and boosting the signal fidelity in coherent spectroscopy. The coherent, temporally stabilized output is critical for precision sensing and low-noise measurement scenarios (Chen et al., 17 Jan 2026).

5. Practical Implementations, Challenges, and Limitations

Implementation of temporal beam self-cleaning mechanisms is sensitive to system parameters. In fibers, achieving the requisite peak power above $P_{\mathrm{th}}$ and matching dispersion regimes (e.g., anomalous vs. normal) are essential. Limitations include supercontinuum background in epifluorescent endoscopy configurations, mitigable by fiber-length and SC-band engineering (Moussa et al., 2020).

Air–plasma channel approaches require generation and maintenance of long, uniform plasma columns, typically via high-energy, picosecond pulses, and precise spatial overlap (alignment tolerances $<100\,\mu$m) between the ultrafast beam and plasma core. Excess or insufficient plasma density can degrade pulse duration or leave residual sidelobes, and environmental constraints restrict implementation to open or large geometries (Xu et al., 2023).

In χ² SHG systems, crystal properties (length, phase-matching bandwidth), fundamental power, and modal structure define the degree of temporal cleaning. The approach is highly compatible with fiber-laser architectures, requiring no feedback, active stabilization, or intracavity modifications. The effect persists across single-, dual-, and multimode longitudinal configurations (Chen et al., 17 Jan 2026).

6. Perspectives and Future Directions

Temporal beam self-cleaning continues to expand in both conceptual understanding and practical relevance. Immediate extensions include pursuit of stronger spatiotemporal compression in the femtosecond regime (for efficient higher-order soliton formation), lowering self-cleaning thresholds through input parameter control, and pursuing complete spatiotemporal beam purification by cascading spatial and temporal self-cleaning via multiple nonlinearities (Moussa et al., 2020, Chen et al., 17 Jan 2026).

The self-organized, environment-robust nature of these processes—bypassing the need for external wavefront correction or temporal gating—positions them as foundational tools in fiber-based super-resolution imaging, ultrafast pulse engineering, remote laser delivery, and quantum optics. Limitations due to technical practicalities (e.g., SNR losses, plasma-channel creation) motivate ongoing research in material, geometry, and process optimization.

Recent results in SHG open a new domain for passive, phase-preserving temporal stabilization at the frequency-conversion stage, with broad implications for coherent source engineering and seed-pulse preparation in high-power systems (Chen et al., 17 Jan 2026). The interplay between statistical depletion, phase preservation, and nonlinear waveform reshaping underlies the universality of temporal self-cleaning, and remains a subject of active investigation.