Replenishable Plasma Mirrors (PMs)

• Replenishable plasma mirrors are laser-driven reflective interfaces formed by ionizing a target to create an overdense plasma boundary that renews after each shot.
• They operate in sub-relativistic to relativistic regimes, enhancing beam contrast and enabling self-aligned focusing via radiation pressure-induced curvature.
• Implementations span tape-drive, liquid-crystal, and integrated photonic platforms, with design challenges focused on pre-plasma control, optical flatness, and high-repetition reliability.

Replenishable plasma mirrors (PMs) are transient, laser-driven reflective interfaces that are renewed between shots so that they can operate where conventional optics would be damaged or destroyed. In the standard implementation, a high-contrast ultrashort pulse ionizes a solid or liquid surface, producing an overdense plasma that specularly reflects the intense core of the pulse while transmitting or suppressing lower-intensity prepulses and pedestal; in staged laser-plasma accelerators (LPAs), this enables coupling optics to be placed close to focus without exposing solid mirrors to fluences far beyond coating limits (Tsai et al., 2016, Vazquez et al., 22 Sep 2025). A broader usage also includes intrinsically renewable free-carrier plasma mirrors in integrated photonic structures, where the reflecting boundary is regenerated by carrier recombination rather than by presenting a fresh material surface (Gaafar et al., 2017). Across these realizations, replenishment is the central engineering feature: the optical element is consumed, reformed, or reconstituted at the experiment’s repetition rate.

1. Definition, operating regimes, and distinction from conventional optics

A plasma mirror is formed when an intense pulse ionizes a material interface and creates an overdense plasma boundary near the critical density surface. For optical frequencies, reflection is governed by the condition that the electron density exceeds the critical density,

$n_c = \frac{\epsilon_0 m_e \omega^2}{e^2},$

with overdense plasma reflecting the main pulse while low-intensity prepulses and wings can transmit. Standard PMs are commonly operated at sub-relativistic intensities, approximately $10^{15}$–$10^{17}\,\mathrm{W/cm^2}$, primarily for temporal-contrast cleaning and spatial-profile improvement (Tsai et al., 2016).

In the relativistic regime, $I_0 \approx 10^{18}$–$10^{19}\,\mathrm{W/cm^2}$, the electron quiver motion becomes relativistic and the PM no longer acts as a passive specular switch. The normalized vector potential is written as

$a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$

or, at $\lambda = 0.8\,\mu\mathrm{m}$,

$a_0 \approx 0.68 \sqrt{I_0/10^{18}},$

so that $a_0 \approx 0.68$ at $10^{18}\,\mathrm{W/cm^2}$ and $10^{15}$0 at $10^{15}$1 (Tsai et al., 2016). In this regime, relativistic transparency and radiation pressure shift and deform the reflecting surface, converting a nominally flat PM into a dynamic concave optic.

Replenishability is required because the interaction region is destroyed every shot. In staged LPAs, this is not optional: the coupling optic must sit close to the focus of the post-acceleration beam, where state-of-the-art systems routinely reach intensities above $10^{15}$2, and up to $10^{15}$3 has been demonstrated, well beyond the damage threshold of any solid-state optic (Vazquez et al., 22 Sep 2025). Traditional tape-drive PMs can operate at $10^{15}$4, but the reported trade-offs include $10^{15}$5–$10^{15}$6 thickness, wavefront roughness, debris, and deleterious electron scattering (Vazquez et al., 22 Sep 2025).

A distinct replenishable class is the free-carrier PM in a silicon slow-light photonic crystal waveguide, where the reflective boundary is a two-photon-absorption-induced carrier front that is recreated optically and then cleared by recombination. In that system, the “mirror” is not a consumable surface but a moving refractive-index front confined within a solid-state waveguide, which makes replenishment intrinsic to the carrier dynamics (Gaafar et al., 2017).

2. Relativistic plasma mirrors as self-aligning concave optics

The defining optical feature of relativistic PMs is light-driven curvature. Two effects act simultaneously. First, radiation pressure dents the overdense layer. For near-normal reflection from a near-perfect reflector,

$10^{15}$7

which gives $10^{15}$8 at $10^{15}$9 and $10^{17}\,\mathrm{W/cm^2}$0 at $10^{17}\,\mathrm{W/cm^2}$1 (Tsai et al., 2016). Second, relativistic electron motion lowers the effective plasma frequency, shifting the reflecting surface inward according to $10^{17}\,\mathrm{W/cm^2}$2. At $10^{17}\,\mathrm{W/cm^2}$3, $10^{17}\,\mathrm{W/cm^2}$4 at $10^{17}\,\mathrm{W/cm^2}$5 and $10^{17}\,\mathrm{W/cm^2}$6 at $10^{17}\,\mathrm{W/cm^2}$7 (Tsai et al., 2016).

Near-normal incidence is crucial because it makes the dent symmetric about the optical axis. The reflected beam is therefore self-aligned to the incoming wavefront: the mirror “makes its own curvature” that points the reflection back toward the laser axis (Tsai et al., 2016). If the dent is approximately parabolic, $10^{17}\,\mathrm{W/cm^2}$8, then $10^{17}\,\mathrm{W/cm^2}$9 and $I_0 \approx 10^{18}$0. Using the observed $I_0 \approx 10^{18}$1 and $I_0 \approx 10^{18}$2–$I_0 \approx 10^{18}$3, the implied dent depth is of order $I_0 \approx 10^{18}$4–$I_0 \approx 10^{18}$5, consistent with the radiation-pressure and relativistic-transparency picture (Tsai et al., 2016).

Under $I_0 \approx 10^{18}$6, $I_0 \approx 10^{18}$7–$I_0 \approx 10^{18}$8, near-normally incident ($I_0 \approx 10^{18}$9), high-contrast pulses with $10^{19}\,\mathrm{W/cm^2}$0–$10^{19}\,\mathrm{W/cm^2}$1, relativistic PMs were found to be highly reflective, with angle-integrated, time-integrated reflectivity in an $10^{19}\,\mathrm{W/cm^2}$2 cone of approximately $10^{19}\,\mathrm{W/cm^2}$3 up to $10^{19}\,\mathrm{W/cm^2}$4, gradually decreasing to $10^{19}\,\mathrm{W/cm^2}$5 at $10^{19}\,\mathrm{W/cm^2}$6 (Tsai et al., 2016). A significant fraction of the reflected light is focused close to the surface: near-field imaging showed focusing at $10^{19}\,\mathrm{W/cm^2}$7 for $10^{19}\,\mathrm{W/cm^2}$8, and simulations found focal distances as small as $10^{19}\,\mathrm{W/cm^2}$9 with peak on-axis intensity in the focal region up to $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$0, collectively supporting gains up to $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$1 (Tsai et al., 2016).

The pre-plasma scale length $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$2 acts as the main control parameter. In EPOCH simulations with $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$3–$a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$4, reflectivity fell with increasing $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$5, while focal length decreased monotonically: representative values were $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$6 at $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$7 and $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$8 at $a_0 \approx 0.85 \sqrt{I_{18}\lambda_{\mu m}^2},$9 (Tsai et al., 2016). This establishes an adjustable-focus trade-off: shorter $\lambda = 0.8\,\mu\mathrm{m}$0 yields longer focal length and higher reflectivity, while longer $\lambda = 0.8\,\mu\mathrm{m}$1 yields tighter focusing but lower reflectivity.

A persistent misconception is that the measured reflectivity of a relativistic PM necessarily collapses at high intensity. The data show that narrow-angle, space-resolved measurements can exhibit a steep apparent drop because focused or refocused light leaves the collection cone; calorimetric wide-cone measurements better represent net absorption (Tsai et al., 2016). A second misconception is that far-field divergence determines the near-field focus. At relativistic intensity, spherical aberration of the light-driven concave surface decouples these observables: the near-surface focal region can contain the highest intensity even when other portions of the beam diverge strongly (Tsai et al., 2016).

3. Replenishable target formats and demonstrated implementations

Replenishable PMs have been implemented with rotating disks, tape targets, liquid jets or sheets, free-standing liquid-crystal films, and integrated free-carrier fronts. The choice of platform is governed by refresh rate, optical flatness, debris generation, thickness, and compatibility with transmitted particle beams.

Platform	Reported characteristics	Reported constraints
Tape-drive PMs	$\lambda = 0.8\,\mu\mathrm{m}$2; $\lambda = 0.8\,\mu\mathrm{m}$3–$\lambda = 0.8\,\mu\mathrm{m}$4 thick; reflectance up to $\lambda = 0.8\,\mu\mathrm{m}$5 at $\lambda = 0.8\,\mu\mathrm{m}$6; intrinsic wavefront RMS $\lambda = 0.8\,\mu\mathrm{m}$7	Large debris production; significant emittance growth in transmitted beams; surface flatness can be inconsistent (Vazquez et al., 22 Sep 2025)
8CB liquid-crystal windmill films	Central optically flat region $\lambda = 0.8\,\mu\mathrm{m}$8–$\lambda = 0.8\,\mu\mathrm{m}$9; average wavefront RMS $a_0 \approx 0.68 \sqrt{I_0/10^{18}},$0–$a_0 \approx 0.68 \sqrt{I_0/10^{18}},$1; angular fluctuation $a_0 \approx 0.68 \sqrt{I_0/10^{18}},$2–$a_0 \approx 0.68 \sqrt{I_0/10^{18}},$3; reliable up to $a_0 \approx 0.68 \sqrt{I_0/10^{18}},$4, $a_0 \approx 0.68 \sqrt{I_0/10^{18}},$5 success at $a_0 \approx 0.68 \sqrt{I_0/10^{18}},$6	Performance depends on smectic-phase temperature control, conditioning, wipe speed, and wiper variability (Vazquez et al., 22 Sep 2025)
Renewable 8CB double PM with SDI	Film thickness $a_0 \approx 0.68 \sqrt{I_0/10^{18}},$7–$a_0 \approx 0.68 \sqrt{I_0/10^{18}},$8; total throughput $a_0 \approx 0.68 \sqrt{I_0/10^{18}},$9; contrast enhancement $a_0 \approx 0.68$0–$a_0 \approx 0.68$1 orders of magnitude; pointing accuracy $a_0 \approx 0.68$2	Demonstrated at several shots per minute; precise per-shot alignment and film-thickness stability required (Czapla et al., 13 Feb 2025)
Silicon slow-light free-carrier PM	$a_0 \approx 0.68$3 forward reflection for $a_0 \approx 0.68$4 at $a_0 \approx 0.68$5	Reflection is governed by dispersion engineering rather than metallic plasma behavior; FCA and finite interaction length limit efficiency (Gaafar et al., 2017)

Among material-surface implementations, smectic 8CB liquid crystal has emerged as a particularly important replenishable medium. Its phase behavior is strongly temperature dependent: crystalline below $a_0 \approx 0.68$6, smectic from $a_0 \approx 0.68$7–$a_0 \approx 0.68$8, nematic from $a_0 \approx 0.68$9–$10^{18}\,\mathrm{W/cm^2}$0, and isotropic above $10^{18}\,\mathrm{W/cm^2}$1. The smectic phase is essential for repeatable film formation and optical flatness (Vazquez et al., 22 Sep 2025). Thickness tunability from $10^{18}\,\mathrm{W/cm^2}$2 to tens of $10^{18}\,\mathrm{W/cm^2}$3 is reported, and prior work demonstrated $10^{18}\,\mathrm{W/cm^2}$4 LC PMs that added only $10^{18}\,\mathrm{W/cm^2}$5 geometric emittance on a $10^{18}\,\mathrm{W/cm^2}$6, $10^{18}\,\mathrm{W/cm^2}$7 electron beam, negligible compared to tape-induced scattering (Vazquez et al., 22 Sep 2025).

Two mechanical formation schemes are described in detail. The “windmill” device uses a vacuum-compatible stepper motor driving a 12-arm rotor; each arm carries an $10^{18}\,\mathrm{W/cm^2}$8-square lens-tissue wiper that dips into an 8CB reservoir and sweeps across a $10^{18}\,\mathrm{W/cm^2}$9-diameter aperture in a polished aluminum plate, drawing a free-standing film (Vazquez et al., 22 Sep 2025). The Spinning Disk Inserter (SDI) is a rotary implementation in which a wheel with $10^{15}$00 apertures sweeps past a fixed absorbent wiper saturated with 8CB; in the BELLA double-PM system, a new film was formed at the same location before every shot, and the same wiper also cleaned the surface (Czapla et al., 13 Feb 2025).

The optical quality of LC PMs is unusually high. In the windmill study, a typical single-film example showed $10^{15}$01 RMS and $10^{15}$02 peak-to-valley wavefront error after subtraction of a flat-gold reference and removal of piston, tip, tilt, and focus, while averages over 50 consecutive films yielded $10^{15}$03–$10^{15}$04 RMS depending on operating condition (Vazquez et al., 22 Sep 2025). In the BELLA renewable double-PM system, the RMS wavefront after the DPM was $10^{15}$05, versus $10^{15}$06 for the reference beam, and the near-field reflected mode was reported to be very similar to the reference (Czapla et al., 13 Feb 2025).

4. Diagnostics, models, and time-dependent optical response

The optical behavior of replenishable PMs is strongly multidimensional and time dependent. A major advance was the simultaneous, shot-by-shot measurement of wavefront, spectrum, and temporal profile of the reflected pulse from a high-intensity PM, using a QuadriWave Lateral Shearing Interferometer, a spectrometer, and GRENOUILLE (Rakeeb et al., 23 Jun 2025). In that study, the reflected wavefront encoded the instantaneous three-dimensional PM shape via the optical path difference, and the measured surface depression was of the order of a few hundred nanometers at relativistic intensities, in quantitative agreement with 3D-PIC simulations (Rakeeb et al., 23 Jun 2025).

The same measurements showed that PM reflection is accompanied by substantial spectral and temporal restructuring. At $10^{15}$07, strong spectral broadening and partial absorption around $10^{15}$08 were observed at high intensity; at $10^{15}$09 ultrahigh contrast, the bandwidth increased monotonically with intensity and reached $10^{15}$10 the initial width at the highest intensity (Rakeeb et al., 23 Jun 2025). These observations imply spatio-temporal couplings arising from surface motion, transient density gradients, and polarization-dependent electron motion. The paper explicitly notes that the QWLSI measurement is time integrated over the reflected pulse per shot and therefore does not resolve intra-pulse attosecond dynamics (Rakeeb et al., 23 Jun 2025).

Pump-probe reflectivity measurements with a $10^{15}$11 probe established that PM focusing properties evolve on a picosecond timescale. Reflectivity begins rising at $10^{15}$12 to $10^{15}$13, consistent with the prepulse intensity reaching $10^{15}$14–$10^{15}$15; at high $10^{15}$16, a transient central “hole” appears near $10^{15}$17 because the dented surface focuses the reflected probe out of the collection cone rather than because true reflectivity vanishes; after $10^{15}$18–$10^{15}$19, central reflectivity partially recovers before later declining due to expansion, absorption, and scattering from surface ripples (Tsai et al., 2016). This time dependence is operationally important for any application that depends on the reflected fundamental.

For renewable double PMs, throughput modeling has been made explicit. The BELLA system decomposed the single-PM reflectivity as

$10^{15}$20

with the expansion-loss term written

$10^{15}$21

The model uses the measured temporal profile $10^{15}$22, a spatial Gaussian mode, pulse energy, incidence geometry, and an ionization threshold of $10^{15}$23 for 8CB; surface expansion is estimated from the interval between threshold crossing and pulse peak using an ion sound speed of $10^{15}$24 (Czapla et al., 13 Feb 2025). The paper states that this is the first model that accurately predicts the peak reflectivity of a plasma mirror when given the laser temporal profile (Czapla et al., 13 Feb 2025).

Model fidelity also depends on dimensionality. For relativistic focusing, 2D PIC underestimates peak focused field and intensity by $10^{15}$25 and $10^{15}$26, respectively, so 3D simulations are required to capture true focusing behavior and energy distribution (Tsai et al., 2016). This is consistent with the broader diagnostic result that PM response is intrinsically three-dimensional in space, spectrum, and time (Rakeeb et al., 23 Jun 2025).

5. Applications from contrast cleaning to staged acceleration and extreme-field optics

The most immediate application of replenishable PMs is high-contrast beam cleaning under conditions where upstream losses and optic damage are unacceptable. The BELLA renewable double PM demonstrated two to three orders of magnitude contrast enhancement on picosecond timescales, suppression of a $10^{15}$27 pre-pulse to below background, and total throughput exceeding $10^{15}$28 with $10^{15}$29 pulses; $10^{15}$30 was measured with $10^{15}$31 pulses without re-optimizing PM locations (Czapla et al., 13 Feb 2025). This directly addresses the use case of petawatt-class lasers operating at increasing repetition rate.

In staged LPAs, replenishable PMs serve as inter-stage coupling optics. The required mirror must sit near the post-acceleration focus, redirect a second high-intensity pulse onto the transported electron-beam axis, and survive neither by damage tolerance nor by cooling, but by renewal (Vazquez et al., 22 Sep 2025). The measured wavefront RMS of $10^{15}$32–$10^{15}$33 for 8CB windmill films is substantially below the BELLA PW intrinsic $10^{15}$34 RMS, which implies that LC PMs do not appreciably degrade the coupled beam wavefront (Vazquez et al., 22 Sep 2025). Thickness tunability into the $10^{15}$35 range further supports transmission of the electron beam with minimal emittance growth (Vazquez et al., 22 Sep 2025).

Relativistic PMs can also recycle and refocus the spent drive pulse. Under $10^{15}$36, $10^{15}$37, near-normal operation at $10^{15}$38–$10^{15}$39, they can reflect $10^{15}$40–$10^{15}$41 of the energy while focusing a significant fraction to $10^{15}$42–$10^{15}$43 from the surface, with peak gains up to $10^{15}$44 (Tsai et al., 2016). The source text states that this could support self-aligned retro-reflection and focusing for laser-plasma accelerator staging or Compton sources, potentially boosting backscatter brightness into the nonlinear Compton regime if geometry allows near-normal retro-reflection (Tsai et al., 2016).

At still higher intensity, curved relativistic PMs enter the relativistic oscillating mirror (ROM) regime and can focus Doppler-generated harmonics. Large-scale 3D-PIC simulations for a $10^{15}$45, $10^{15}$46, $10^{15}$47 laser at $10^{15}$48 and $10^{15}$49 found a combined gain of $10^{15}$50, reaching intensities near $10^{15}$51, with focused harmonic intensities in the $10^{15}$52–$10^{15}$53 range (Vincenti, 2018). The optimum in that study occurred at $10^{15}$54, where radiation-pressure-induced curvature and harmonic efficiency were jointly favorable (Vincenti, 2018). This places replenishable PM technology within the broader program of extreme-field and attosecond science.

Integrated photonics provides a different application domain. In a silicon slow-light photonic crystal waveguide, a two-photon-absorption-generated free-carrier front moving at $10^{15}$55 produced forward reflection of a co-propagating probe with $10^{15}$56 efficiency for the interacting CW portion, a maximal blue shift of $10^{15}$57, and operation at telecom wavelengths using a $10^{15}$58, $10^{15}$59 pump pulse (Gaafar et al., 2017). Reflection there is enforced by dispersion engineering and an indirect intraband transition rather than by a metallic plasma boundary, so the carrier density requirement is reduced to $10^{15}$60 (Gaafar et al., 2017).

A further theoretical extension treats the relativistically oscillating PM as a “spacetime mirror.” In that picture, the laser-driven interface $10^{15}$61 acts as a time-varying boundary that can exhibit a superluminal apparent boundary velocity, time reflection and refraction, and quantum light generation with pair creation (Pan et al., 2024). The proposal uses an $10^{15}$62, $10^{15}$63, $10^{15}$64 drive on a coated $10^{15}$65 target with $10^{15}$66 cavity length, and reports simulated photon-number growth from $10^{15}$67 at $10^{15}$68 to $10^{15}$69 at $10^{15}$70 for selected modes (Pan et al., 2024). This is not yet an established experimental application, but it shows how replenishable PMs intersect time-varying-media and strong-field quantum-optics research.

6. Limitations, misconceptions, and current design criteria

The main practical limitation is control of pre-plasma scale length $10^{15}$71. For relativistic PM performance at the fundamental, prepulses should not exceed $10^{15}$72 up to $10^{15}$73 before the main peak; in the cited experiments, this kept $10^{15}$74 in the micron range, with $10^{15}$75–$10^{15}$76 at $10^{15}$77–$10^{15}$78 and $10^{15}$79 at $10^{15}$80 (Tsai et al., 2016). In the high-intensity diagnostics study, high contrast of $10^{15}$81 on the picosecond scale was likewise used to suppress undesired pre-plasma (Rakeeb et al., 23 Jun 2025). This makes contrast management a first-order systems problem rather than a secondary diagnostic detail.

A related misconception is that PM reflectivity is a single scalar property. In practice, measured reflectivity depends on collection angle, time window, and whether the observable is the reflected fundamental, harmonics, or a probe beam (Tsai et al., 2016, Rakeeb et al., 23 Jun 2025). Another misconception is that a replenishable PM is automatically suitable for high repetition rate once a fresh surface can be presented. The 8CB windmill study showed that reliability remained high up to $10^{15}$82, but at $10^{15}$83 the success probability dropped to $10^{15}$84, attributed to thermal and flow constraints, wiper-LC dynamics, and conditioning effects (Vazquez et al., 22 Sep 2025). The BELLA renewable DPM operated at several shots per minute, and while the SDI alone demonstrated film formation up to $10^{15}$85, the full system still required $10^{15}$86–$10^{15}$87 per shot because wiping and alignment dominated the cycle (Czapla et al., 13 Feb 2025).

For liquid-crystal PMs, temperature control is decisive. 8CB must be kept in the smectic phase, $10^{15}$88–$10^{15}$89, and the windmill study used a thermoelectric recirculating chiller because motor heating could otherwise drive the LC out of the usable regime (Vazquez et al., 22 Sep 2025). A conditioning protocol of $10^{15}$90–$10^{15}$91 slow rotation after LC application was reported as critical for stable thickness and angular pointing (Vazquez et al., 22 Sep 2025). Pointing jitter correlated with wipe direction, wiper variability, LC dosing, and edge accumulations at the aperture; a motorized gimbal plus centroid-based feedback was identified as a route to keep LC PM pointing within the $10^{15}$92 staging requirement (Vazquez et al., 22 Sep 2025).

High-repetition deployment also requires separate treatment of near-field and far-field observables. Because spherical aberration decouples near-field focus from far-field divergence, applications that want the focused energy near the PM must place optics within $10^{15}$93–$10^{15}$94 of the surface or adopt compact pick-off geometries, whereas applications requiring high reflectivity into downstream optics should use a wide acceptance cone such as $10^{15}$95 or larger (Tsai et al., 2016). This suggests that PM beamline design should not treat “reflectivity” and “focusing” as interchangeable figures of merit.

Open problems remain well defined in the source literature. For LC PMs, these include high-intensity reflectivity characterization versus $10^{15}$96, incidence angle, and contrast; in-situ thickness metrology across the $10^{15}$97–$10^{15}$98 flat region; long-term automation and maintenance; and credible scaling to robust $10^{15}$99 and beyond (Vazquez et al., 22 Sep 2025). For relativistic PM physics, the unresolved engineering issues include systematic mapping of reflectivity versus $10^{17}\,\mathrm{W/cm^2}$00, $10^{17}\,\mathrm{W/cm^2}$01, polarization, and material; time-resolved wavefront metrology of sub-picosecond surface motion; and debris or contamination studies for high-average-power operation (Rakeeb et al., 23 Jun 2025). For extreme-intensity ROM applications, pre-structured PMs that increase curvature independently of $10^{17}\,\mathrm{W/cm^2}$02 were proposed as a route toward gains $10^{17}\,\mathrm{W/cm^2}$03, but preserving harmonic quality under such conditions remains technologically nontrivial (Vincenti, 2018).

Taken together, the literature defines replenishable PMs not as a single device class but as a family of renewable plasma-based optics spanning consumable solid and liquid interfaces, ultrathin liquid-crystal films, and carrier fronts in integrated photonics. Their shared premise is that the reflecting boundary is transient, high-damage-threshold, and re-established at the operating cadence of the experiment; their shared challenge is that optical quality, reflectivity, phase, curvature, and timing are all set by nonlinear plasma dynamics that must be controlled rather than merely tolerated.