Print temporary

General Discussion
1

Reassessing Pump Penetration, Energy Storage, and Parasitic Limits in the BSF Molten-Salt Laser Concept

Executive summary

The BSF concept (as described on the forum) is: a spherical reactor filled with molten FLiBe doped with a rare-earth gain ion; the salt is optically pumped shortly before acoustic compression of a DT bubble; the bubble’s flash seeds laser amplification in the excited salt; and reflected light returns to the bubble to drive ignition.

Your central critique of the earlier “linear Beer–Lambert” pump-penetration model is directionally right: if the pump field is predominantly radial and the power crossing spherical shells is approximately conserved (minus absorption), then the local pump flux density can scale roughly like (\propto (R^2/r^2),\exp[-\alpha(R-r)]), i.e., geometric convergence competes with exponential attenuation.

Using the BSF forum’s own “~1000” fiber-bundle estimate and “~1 inch diameter” estimate for each bundle, the purely geometric “coverage” crossover where the sum of beam areas equals the area of a spherical shell occurs at (r\sim 0.2) m. That means a large “unpumped core” is not an inevitable consequence of periphery injection; even with collimated radial injection, central regions can be illuminated by many more sources than peripheral regions.

That said, the bigger feasibility limiter is still not “capacity” (total Nd ions available), but (i) the effective upper-state lifetime at ~800 K (how much energy can be stored per unit absorbed pump power), and (ii) unwanted optical feedback (parasitic lasing / gain clamping / ASE) in a many-path cavity. The forum itself uses an illustrative (\tau\sim 50\ \mu\text{s}) assumption for hot Nd:FLiBe, which—if true—drives pump-power requirements into the hundreds of gigawatts for multi-megajoule storage.

Switching the pump from 808 nm to ~869 nm plausibly helps in two ways: it modestly improves quantum efficiency (~7.6% on wavelength ratio alone), and it reduces Nd absorption relative to the strong ~808 nm band (deeper penetration / less surface-skewed pumping). However, neither change alone resolves parasitic feedback limits; reducing wall reflectivity from ~98% (gold-like) to “nickel-like” reflectivity (often ~75–78% for bulk nickel at normal incidence, depending on wavelength) increases the threshold for oscillation, but the resulting gain-clamping stored energy is still orders of magnitude below 5 MJ under Nd:YAG-like cross-sections.

Finally, regarding the fiber/plug claim: the BSF patent summary emphasizes fused-silica “plugs” and shock accommodation as a mechanical challenge. Laser-damage thresholds for fiber facets depend strongly on pulse duration and surface quality; some references cite fiber-bundle thresholds in the few MW/mm² range, while others cite very high best-case bulk thresholds for pristine silica for nanosecond pulses. With ~1000 one‑inch apertures distributing tens of gigawatts, the average intensity per aperture can be far below a few MW/mm², but millisecond-pulse thermal limits and molten-salt interface engineering remain first-order risks.

What the BSF forum says about the concept and assumptions

The forum’s “molten salt laser medium” description frames BSF as a spherical reactor filled with molten salt that simultaneously acts as coolant/blanket and laser gain medium (rare-earth doped), storing pump energy in excited ions.

The “sonoluminescent trigger” description adds the timing/sequence: shortly before the converging acoustic shock reaches the bubble, the salt is pumped to become laser-active; then the bubble’s flash seeds amplification; the amplified light is intended to return (after reflecting from the wall) and deposit energy into the fuel.

The “2009 abstract” post reiterates the same core functional chain and explicitly states that the bubble’s bright radiation seeds “laser cascades that return, greatly amplified, from the sphere’s polished innards.”

Two additional details from the patent-summary post matter directly to your requested recalculation:

  • Pump delivery hardware concept: on the order of ~1000 fiber-optic cables/bundles, each “estimated to be about one inch in diameter,” terminating in durable silica plugs set into drilled shafts and mechanically decoupled (springs / shock absorbing pistons).
  • Spherical ray geometry / multi-pass intuition: the post notes that (in a reflective sphere) ray paths are confined to a plane; rays that pass close to center tend to revisit similar locations after two reflections; and simulations found unexpectedly high “two-reflection” reabsorption.

These points collectively support your premise that “linear, single-pass attenuation” is not the whole story: there is both geometric convergence (more of the periphery contributes to smaller radii) and multi-pass persistence (some fraction of rays revisit the center region).

Corrected pump transport in a spherical cavity

A simple radial-flux model that captures your convergence point

If the TaC grid (per your stated assumption) strongly suppresses non-radial propagation, the pump field can be approximated as predominantly radial (inward and outward). Under the idealization of spherical symmetry, define (P(r)) as the total inward pump power crossing a spherical surface of radius (r). With an absorption coefficient (\alpha) (m(^{-1})) for the pump wavelength, the Beer–Lambert differential form gives:
[ \frac{dP}{dr}=-\alpha P \quad\Rightarrow\quad P(r)=P(R),e^{-\alpha (R-r)}. ]

The flux density (power per area) at radius (r) is:
[ F(r)=\frac{P(r)}{4\pi r^2}=\frac{P(R)}{4\pi R^2}\left(\frac{R^2}{r^2}\right)e^{-\alpha(R-r)}. ]

So the key ratio is:
[ \boxed{\frac{F(r)}{F(R)}=\left(\frac{R^2}{r^2}\right)e^{-\alpha(R-r)}} ]

This is exactly your “Beer–Lambert attenuation and spherical convergence” correction: the exponential term decreases with depth, but the (R^2/r^2) term increases strongly as you approach the center.

A geometric “overlap radius” from the forum’s own fiber numbers

A practical “reality check” is to estimate whether inward rays can plausibly “cover” inner spherical shells.

From the patent-summary post: ~1000 fiber bundles, ~1 inch diameter each.
Approximate one bundle area: [ A_\text{bundle}=\pi(0.0127\ \text{m})^2 \approx 5.07\times 10^{-4}\ \text{m}^2. ] Total bundle area: (N A \approx 1000\times 5.07\times 10^{-4}=0.507\ \text{m}^2.)

Set this equal to the shell area (4\pi r^2) to find the radius where the sum of bundle “beam footprints” equals the shell area: [ 4\pi r^2 \approx 0.507\ \Rightarrow\ r \approx 0.20\ \text{m}. ]

Interpretation: if beams remain roughly collimated and radial, then inside (r\sim 0.2) m the “beam area budget” is enough that many rays overlap and the assumption of a broadly illuminated inner region is no longer obviously wrong. This directly weakens the notion that periphery-only injection necessarily implies a meter-scale unpumped core.

Workflow diagram for the corrected “pump → store → trigger → deposit” logic

mermaid

Copy

flowchart TD
  A[Pump pulse injected via many fiber bundles at periphery] --> B[Predominantly radial propagation enforced by collimation grid]
  B --> C[Pump flux at radius r: scales ~ (R^2/r^2)*exp(-alpha*(R-r))]
  C --> D[Spatially varying absorbed pump power density]
  D --> E[Upper-state population builds for duration tp; limited by tau_eff at ~800 K]
  E --> F[Bubble compression flash provides seed photons]
  F --> G[Seed amplified; energy extracted from stored inversion]
  G --> H[Optical feedback & multi-pass determine deposition back into bubble]
  E --> I[Risk: parasitic lasing/ASE clamps gain before seed arrives]
  B --> I
  H --> I

The flowchart highlights the real “knife-edge”: the same geometry that helps inward flux and multi-pass also increases the number of closed (or quasi-closed) feedback paths that can trigger unwanted oscillation during the pumping period.

Recalculated dopant capacity and a best-guess NdF₃ range for 869 nm pumping

This section recalculates (i) “capacity” (how many Nd ions are available to store energy), (ii) pump penetration constraints with spherical convergence, and (iii) how 869 nm vs 808 nm changes the feasible dopant window.

Step one: capacity—how much Nd is needed in principle to store 5 MJ?

The forum’s laser-efficiency thread provides a baseline set of FLiBe properties and the 2.5 m radius sphere volume: (V\approx 65.4\ \text{m}^3), (\rho\approx 2.09\ \text{g/cc}), and molecular weight (\approx 98.9\ \text{g/mol}), giving a total FLiBe inventory (\approx 1.38\times 10^6) moles.

Photon energy at (\lambda_L\approx 1055) nm: [ E_\gamma = \frac{hc}{\lambda_L}\approx 1.88\times 10^{-19}\ \text{J}. ]

If the Nd upper state stored energy were fully extractable (best case), the number of excited ions needed for 5 MJ is: [ N_2 = \frac{5\times 10^6}{1.88\times 10^{-19}}\approx 2.7\times 10^{25}. ]

Convert that to moles of excited Nd: [ n_2=\frac{N_2}{N_A}\approx \frac{2.7\times 10^{25}}{6.022\times 10^{23}}\approx 45\ \text{mol}. ]

Relative to total FLiBe moles: [ x_\text{Nd,min}=\frac{45}{1.38\times 10^6}\approx 3.2\times 10^{-5}\approx 32\ \text{ppm (molar)}. ]

That ~32 ppm is only the theoretical minimum (100% excitation of Nd and 100% extraction). More realistic scaling is:

[ x_\text{Nd}\approx \frac{32\ \text{ppm}}{f_\text{inv}, f_\text{ext}}, ]

where (f_\text{inv}) is the achieved excitation fraction and (f_\text{ext}) is the fraction of stored energy extracted into the useful pulse.

Representative cases (2.5 m radius sphere):

  • (f_\text{inv}=1.0,\ f_\text{ext}=0.5 \Rightarrow x\approx 64) ppm
  • (f_\text{inv}=0.5,\ f_\text{ext}=0.5 \Rightarrow x\approx 128) ppm
  • (f_\text{inv}=0.25,\ f_\text{ext}=0.5 \Rightarrow x\approx 256) ppm
  • (f_\text{inv}=0.1,\ f_\text{ext}=0.5 \Rightarrow x\approx 640) ppm

So “capacity” alone suggests an NdF₃ dopant scale on the order of (10^1)–(10^3) ppm, not ~1 mol% (=10,000 ppm). The forum’s own worked example at 0.5 mol% already implies huge ion inventories relative to tens-of-moles needs.

Scaling note for 2.0 m radius: volume scales as (R^3). A 2.0 m radius sphere has ~0.512× the volume of a 2.5 m sphere, so required ppm roughly doubles for the same energy target.

Step two: pump absorption strength at 808 vs 869 (easy-to-get proxy data)

Two easily accessible references constrain how different 808 vs 869 pumping can be for Nd:

  • A Northrop Grumman Synoptics chart for 1% Nd:YAG shows an absorption coefficient peak near ~808 nm around ~10 cm(^{-1}), and a separate notable feature near ~869 nm around a few cm(^{-1}) (order ~3–4 cm(^{-1})).
  • RP Photonics provides Nd:YAG reference values: absorption cross-section at 808 nm (\sim 7.7\times 10^{-20}\ \text{cm}^2), emission cross-section at 1064 nm (\sim 28\times 10^{-20}\ \text{cm}^2), and fluorescence lifetime (\sim 230\ \mu\text{s}).

These don’t prove Nd:FLiBe will match Nd:YAG, but they justify a first-order ratio: 869 nm absorption is plausibly ~3× weaker than 808 nm absorption under diode-like pumping bandwidths.

Step three: quantum efficiency advantage of 869 → 1055 nm vs 808 → 1055 nm

Using only wavelength ratio (an upper-limit proxy for “quantum defect efficiency”):

[ \eta_q \approx \frac{\lambda_p}{\lambda_L}. ]

So:

  • 808 → 1055: (\eta_q\approx 0.766)
  • 869 → 1055: (\eta_q\approx 0.824)

That is a ~7.6% improvement in the idealized photon-energy ratio sense.
(And RP Photonics explicitly describes 869 nm pumping as exciting the upper laser manifold more directly, reducing the quantum defect relative to pumping into higher manifolds around 808 nm. )

Step four: penetration with spherical convergence—turning ppm into an absorption coefficient constraint

Assume the “overlap radius” (r_*\sim 0.2) m is the smallest radius at which it’s reasonable to talk about an approximately “filled” inward flux field (from the many periphery bundles).

Using the earlier ratio: [ \frac{F(r_)}{F(R)}=\left(\frac{R^2}{r_^2}\right)e^{-\alpha(R-r_*)}. ]

With (R=2.5) m and (r_=0.2) m, the geometric factor is: [ \frac{R^2}{r_^2}=\frac{2.5^2}{0.2^2}=156.25. ]

If you want the inward pump flux at (r_*) to be at least ~10% of what it is at the wall (a conservative “don’t starve the core” criterion), solve: [ 156.25,e^{-\alpha(2.3)} \gtrsim 0.1 \Rightarrow \alpha \lesssim 3.2\ \text{m}^{-1}. ]

If you want it ~50%, then (\alpha \lesssim 2.5\ \text{m}^{-1}).

Now map (\alpha) to ppm using the Nd:YAG 869 nm absorption coefficient magnitude as a proxy: the chart suggests order 3–4 cm(^{-1}) for ~1% Nd near 869 nm, i.e. order (300–400\ \text{m}^{-1}) per 1% Nd.
Scaling linearly, to make (\alpha\sim 2.5–3.2\ \text{m}^{-1}), you’d want: [ x_\text{Nd}\sim \frac{2.5–3.2}{300–400}\times 1% \approx 0.006–0.011% \approx 60–110\ \text{ppm}. ]

That overlaps nicely with the “capacity mid-cases” (~64–256 ppm), suggesting a plausible sweet spot: tens to a few hundred ppm for a 2.5 m radius sphere if 869 nm pumping is used and you want meaningful core pumping without multi-radius injection.

Summary table of what changes when you adopt 869 nm pumping and lower-reflectivity walls

Design knob Why it helps What it hurts Quantitative anchor
Pump at ~869 nm vs 808 nm Higher wavelength ratio (\lambda_p/\lambda_L) → smaller quantum defect; and typically weaker absorption → deeper penetration Weaker absorption can require longer effective pump path / more photons to reach target inversion; stimulated emission proximity can limit achievable upper-state population (general in-band-pumping caveat) RP Photonics notes 869 nm pumping targets the upper manifold and reduces quantum defect. Nd:YAG chart shows lower absorption near 869 than the strong ~808 band.
Lower wall reflectivity (gold-like ~98% → nickel-like ~75% or your assumed 60%) Raises oscillation threshold and reduces cavity photon lifetime → less chance of early self-oscillation Reduces multi-pass retention (both pump trapping and “return-to-bubble” light recycling) Bulk optical constants imply Au reflectivity near ~0.98 and Ni nearer ~0.75–0.78 around 0.8–1.1 µm (normal incidence).
Multi-radius pump injection (shells or internal feedthroughs) Cuts effective path length and relaxes dopant/absorption constraints; can shape inversion distribution More penetrations/structures in the wall; higher mechanical/thermal/chemistry complexity Patent summary already anticipates many silica “plug” penetrations and stress accommodation as a design constraint.

The “hidden dominant variable”: upper-state lifetime at ~800 K

The BSF forum’s laser-efficiency analysis explicitly highlights that a short lifetime (example assumption (\tau\approx 50\ \mu\text{s})) drives pump power requirements to extreme values.

For a pump pulse of duration (t_p) comparable to or longer than (\tau), the stored energy scales roughly with absorbed pump power times (\tau) (because excited ions decay while pumping continues). Under that regime, to store (E_\text{store}) you need: [ P_\text{abs}\sim \frac{E_\text{store}}{\eta_q,\tau}. ]

If (E_\text{store}=5) MJ, (\eta_q\approx 0.824) (869→1055), and (\tau=50\ \mu\text{s}), then: [ P_\text{abs}\sim \frac{5\times 10^6}{0.824\times 50\times 10^{-6}} \approx 1.2\times 10^{11}\ \text{W} \ (120\ \text{GW}). ]

So your stated 20 GW-class pump becomes plausible only if (\tau) is closer to a few (10^{-4}) s: [ \tau \sim \frac{5\ \text{MJ}}{0.824\times 20\ \text{GW}} \approx 0.30\ \text{ms}. ]

As an existence proof for (\sim 0.2) ms-class lifetimes in a neodymium system, Nd:YAG is tabulated at ~230 μs at room temperature. But the forum’s skepticism is specifically that hot molten-salt quenching may reduce (\tau) far below that.

Implication: your “869 nm + lower reflectivity + better geometry” options are worth exploring, but the research priority remains: measure or defensibly bound (\tau_\text{eff}) for Nd:FLiBe near 800 K under realistic chemistry/purity.

Parasitic oscillations, cavity feedback, and why wall reflectivity matters (and isn’t enough)

Why “stored energy ≪ 5 MJ” can still happen in a huge volume

In high-gain systems, unwanted closed optical paths can lase (parasitic lasing) and clamp gain, draining energy before the intended extraction event. This risk is explicitly highest during the pumping period in pulsed high-gain systems.

The steady-state threshold condition for a simple two-mirror cavity is commonly written: [ R_1 R_2, e^{2g_\text{th}l},e^{-2\alpha l}=1, ] or rearranged, [ g_\text{th}=\alpha_0-\frac{1}{2l}\ln(R_1R_2). ]

A “diameter-cavity” estimate for a 2.5 m radius sphere

Take the simplest closed radial path: between two opposite wall tiles (cavity length (l\approx 2R=5) m), with two reflections per round trip.

Assume:

  • mirror reflectivity per bounce (R_m),
  • (R_1R_2\approx R_m^2),
  • distributed loss at 1055 nm is small compared to mirror loss (optimistic).

Then: [ g_\text{th}\approx \frac{1}{2l}\ln\left(\frac{1}{R_m^2}\right). ]

Using Nd:YAG-like emission cross-section as a proxy to connect gain coefficient to stored inversion: [ g \approx \sigma_e N_2, ] with (\sigma_e(1064)\approx 28\times 10^{-20}\ \text{cm}^2=2.8\times 10^{-23}\ \text{m}^2.)

Compute a rough “gain-clamping” stored energy (E_\text{clamp}\sim N_2 h\nu V) at threshold:

  • If (R_m=0.98) (gold-like), (g_\text{th}\approx 0.0040\ \text{m}^{-1}) → (E_\text{clamp}\sim 1.8) kJ for a 65 m³ sphere.
  • If (R_m\approx 0.75) (bulk-Ni-like), (g_\text{th}\approx 0.058\ \text{m}^{-1}) → (E_\text{clamp}\sim 25) kJ.
  • If (R_m=0.60) (your proposal), (g_\text{th}\approx 0.102\ \text{m}^{-1}) → (E_\text{clamp}\sim 45) kJ.

These results are order-of-magnitude (and optimistic) because real geometries often contain shorter unintended loops, scattering paths, and wavelength-dependent losses; shorter cavities lower the threshold further. The qualitative conclusion is robust: reducing reflectivity helps, but does not by itself make MJ storage “automatic.”

Why this is consistent with real high-energy laser engineering practice

Large-aperture Nd:glass amplifiers face practical limits from amplified spontaneous emission (ASE) that depumps slabs and limits gain; even at far smaller scales than a multi-meter sphere, ASE can impose “practical size limits.” This is one reason high-energy systems incorporate deliberate parasitic/ASE suppression measures (e.g., absorbing edge cladding in disk/slab architectures).

Design improvements that directly target the parasitic/feedback problem

These are the options that most directly address the “kJ clamp vs MJ target” gap:

  • Decouple “pump trapping” optics from “lasing feedback” optics. A sphere that is very reflective at pump wavelengths but much less reflective at the lasing wavelength can absorb pump efficiently without creating a high-Q lasing cavity. In conventional systems this is done with dichroics and cavity engineering; in BSF it would require wavelength-selective wall behavior at ~800–900 nm vs ~1050–1070 nm (materials/coatings at ~800 K in molten salt are nontrivial).
  • Introduce an intensity-dependent loss element (passive hold-off). This is the conceptual role played by saturable absorbers in pulsed lasers: they suppress low-intensity oscillation and permit high-intensity extraction. Whether any chemically compatible absorber exists in molten FLiBe at ~1055 nm is unknown, but without some “hold-off” mechanism, suppressing premature oscillation in a multi-path cavity is structurally difficult.
  • Geometry that limits closed paths during pumping. If the TaC grid truly eliminates most non-radial propagation, the dominant closed paths become diameter-like. Any additional structure that breaks those (e.g., slight de-collimation + absorbing baffles, or intentionally roughened/diffuse walls during pumping) can raise thresholds—at the cost of reducing the intended “return-to-bubble” recycling.

Net: your “replace 98% gold with ~60% nickel” direction is consistent with “raise threshold,” but the math indicates it is unlikely to move the stored-energy ceiling from kJ to MJ without one of the above deeper changes.

Fiber-optic and fused-silica “plug” limits for the proposed pump pulse

What the forum’s patent-summary actually claims about the pump feedthroughs

The patent-summary post anticipates:

  • ~1000 fiber bundles,
  • ~1 inch diameter estimate,
  • fused cable ends forming “durable silica plugs,”
  • mechanical stress and shock accommodation as central engineering issues.

So the forum itself does not treat “fibers into molten salt” as a trivial detail; it flags it as a primary mechanical survivability challenge.

Damage-threshold reality check (what is easy to cite)

Two easy-to-access references show why quoting a single “LIDT number” is tricky:

  • NKT Photonics notes that fiber-facet damage thresholds depend critically on surface quality and pulse length; it cites a best-case 8 ns damage irradiance for pure silica around 4.1 kW/μm² (while emphasizing large variability and that facets are more fragile than bulk).
  • A Google Patents document states typical fiber-bundle damage thresholds “up to 2 MW/mm²” and “up to 5 MW/mm²” in laboratory arrays, again with strong dependence on processing/polishing/coatings.

Those values are not directly transferrable to 0.5 ms pulses in a molten-salt environment, but they’re enough to evaluate your order-of-magnitude claim: “20 GW peak should be fine.”

Area/intensity arithmetic for your proposed pump pulse

If total pump power is (P=20) GW and you distribute it among ~1000 one-inch bundles:

Total optical area: [ A\approx 1000\times \pi(12.7\ \text{mm})^2 \approx 5.07\times 10^5\ \text{mm}^2. ]

Average intensity: [ I\approx \frac{20\times 10^9\ \text{W}}{5.07\times 10^5\ \text{mm}^2}\approx 3.9\times 10^4\ \text{W/mm}^2 = 0.039\ \text{MW/mm}^2. ]

That is well below “a few MW/mm²” style thresholds (if they applied), by ~50× or more.

So the strongest technical objection is not “categorically impossible intensity,” but rather:

  • the ms-pulse thermal regime (not well captured by ns-pulse LIDT scalings),
  • contamination/defects at the facet,
  • bubble formation at the interface,
  • mechanical shock, and
  • chemical compatibility of silica plugs and any sealing/brazing scheme in hot fluoride salt.

Bubble blackbody seeding versus cascade triggering requirements

The patent-summary post’s blackbody discussion makes two explicit claims:

  • blackbody radiation from the hot coolant (at the laser frequency) is far below lasing threshold and not expected to trigger amplification;
  • blackbody from the compressed bubble can be hot enough to produce significant radiation at the laser frequency, “enough to trigger a laser amplification cascade.”

From an optics standpoint, there are two distinct regimes:

  • Above-threshold cavity: if net gain exceeds loss on any closed path, spontaneous emission (or scattered pump light) is enough to build a pulse; you do not need an unusually strong bubble seed. This is exactly why parasitic lasing is such a problem in high-gain pulsed systems.
  • Below-threshold amplifier: if you successfully suppress closed-path feedback, then you rely on the bubble flash as an injected signal, and the extraction depends on whether the seed fluence reaches (or exceeds) the gain medium saturation fluence scale: [ F_\text{sat}=\frac{h\nu}{\sigma_\text{em}}. ]

Using Nd:YAG-like (\sigma_\text{em}(1064)\approx 2.8\times 10^{-19}\ \text{cm}^2), (F_\text{sat}) is on the order of: [ F_\text{sat}\sim \frac{1.88\times 10^{-19}\ \text{J}}{2.8\times 10^{-19}\ \text{cm}^2}\approx 0.67\ \text{J/cm}^2. ]

Whether a DT bubble’s radiative flash can deliver an effective seed fluence of that order to the relevant optical modes/paths depends on information not specified in the prompt:

  • bubble radius at peak emission,
  • emission duration,
  • spectral content specifically near 1050–1070 nm,
  • how strongly the surrounding medium absorbs/scatters that band before amplification,
  • how much of the seed is in the “radially allowed” set of directions (given the grid filtering assumption).

The patent-summary claims this works in principle, but the engineering implication is: the stronger your parasitic-suppression strategy is (keeping the system sub-threshold), the more demanding the bubble seed requirement becomes, and the seed requirement should be quantified in the same “saturation fluence / stored energy extraction” language used in high-energy amplifiers.

Sources consulted

BSF forum sources (discourse.group), prioritized first:

  • “About the Liquid (molten salt) laser medium category.”
  • “About the Sonoluminescent trigger for lasing category.”
  • “Patent App. (2009 abstract)” (including the fiber count/size estimates and blackbody discussion).
  • “Comparitive Analysis of Laser Efficiency” (for the illustrative (\tau\sim 50\ \mu\text{s}) hot-lifetime assumption and pump-module discussion).

Additional high-quality / primary references:

  • RP Photonics on in-band pumping concepts (incl. 869 nm for Nd systems) and on parasitic lasing (gain clamping in high-gain pulsed contexts).
  • RP Photonics Nd:YAG property table (reference values for absorption/emission cross sections and lifetime).
  • Northrop Grumman Synoptics Nd:YAG absorption coefficient chart (808 vs ~869 nm absorption magnitude at 1% doping).
  • NKT Photonics note “Damage threshold of fiber facets” (pulse-length dependence; facet sensitivity; example irradiance value for ns pulses).
  • Fiber bundle damage-threshold statement via Google Patents.
  • Lawrence Livermore National Laboratory / National Ignition Facility amplifier design report noting ASE-imposed practical aperture limits (context for large-gain-media scaling).
  • Optical constants for bulk Ni and Au at 0.8–1.1 µm via refractiveindex.info dataset text exports (used to compute approximate normal-incidence reflectivity).