# Beam Fractionation — Spectral Allocation

> Sell each band to the customer who wants it. Stop charging anyone for the rest.

## Why it works

Solar output is broadband. Industrial demand is not. Broadband-only delivery makes customers pay for photons they reject and forces the swarm to sell at the lowest-margin band's price. **Fractionation splits the spectrum and prices each band separately.**

## Physical mechanism

| Tool | Use |
|---|---|
| Dichroic stack (multilayer dielectric film) | Bulk band split, >99% per-stage reflectivity, minimal mass overhead on standard mirror film |
| Volume Bragg grating | Sub-nm bandpass for ultra-selective applications |
| Prismatic disperser | Angular fan-out — multiple receivers at known angular positions, each a different band |

**Cascade:** solar flux → UV reflector → resonance reflector → visible reflector → NIR reflector → dump residual (long-wave IR).

## Three architectural locations

| Where | Trade-off |
|---|---|
| Source-side (mirror coating) | Most efficient; inflexible — band fixed by coating |
| Intermediate sorting tier | Flexible; higher per-installation cost |
| Conversion node (final) | Last-mile customer segmentation when upstream is broadband |

Mature swarm uses all three: stable mass-market bands sorted at source, specialty demand at intermediate, last-mile at node.

## Band-by-band market

| Band | Tier | Customers |
|---|---|---|
| Coherent narrowband UV | Premium | Photolithography, sterilization-at-scale |
| Soret-equivalent resonance | Premium | ATP-fed slime, photosynthetic optimization |
| Comm-frequency lines | Premium | Relay feeds, isolated comms |
| Coherent narrowband visible | Mid-premium | Metrology, precision fab |
| Broadband visible (illumination) | Commodity | Habitat illumination, general PV |
| Broadband NIR | Commodity | Thermal PV, mass-driver power |
| Mid-IR | Discount | Heat-cycle thermal |
| Far-IR / long-wave | Often stranded | Dump or radiate-to-space |

Premium bands trade multi-decade forward. Commodity bands trade near-spot. Stranded bands go to short-notice discount auctions or back to vacuum.

## ATP-resonance band — canonical example

Hyperscale ATP-fed Schleimfarmen contract for a 4–6 nm Soret-equivalent band. Near-quantum efficiency at receiver. **10 GW spectral grant ≈ 25–35 GW broadband-equivalent useful work.** Operator saves on radiator capacity. SMA saves on allocation. Spectral grant is more expensive per delivered watt, cheaper per useful watt. This is the structural reason hyperscale ATP-fed installations pencil out at all.

The ATP industry and spectral fractionation infrastructure **co-evolved**. Each justifies the other.

## Stranded dump

~8–15% of intercepted flux currently unallocated. Dumped = simply not reflected. The swarm's slack. Reducing it is a standing SMA strategic objective.

**Discount-band auctions** — short-notice contracts (hours to weeks) for soon-to-dump bands. Opportunistic buyers: temporary thermal applications, low-rate compute, deep-space relay heating. Spot pricing, no queue priority.

## Customer-side hardware

Narrowband receiver ≠ broadband + filter. Aperture, conversion stage, thermal management all band-specific. **Grants are sticky** — switching bands is a refit, not a setting change.

## SMA grant becomes multi-dimensional

```
Grant := receiver coords
       × total power
       × spectral profile (band × wattage)
       × coherence (broadband / narrow / coherent narrow)
       × modulation schedule
       × duration
```

Pricing reflects all dimensions. **Grant brokers** emerge as a recognized specialty class — independent intermediaries translating operator demand into SMA-acceptable grant specifications.

Secondary market: broadband grants resell freely; narrowband grants trade thin and volatile.

## Cone publication includes spectral profile

A cone carrying UV is a different hazard than a cone carrying NIR at the same intensity. Cone-edge skimming kits are now band-tuned. Senior flea pilots distinguish themselves by band-fluency.

→ `atmospheric-beam-safety.md`, `fleas.md`

## Coverage trajectory

| Year | Spectral coverage | State |
|---|---|---|
| ~3,240 (now) | 40–60% | Construction Spine mid-buildout |
| ~mid-3,500s | >80% (projected) | Late Construction Spine |
| Mature forward-canon | Full | "Directed microwave and laser beaming" era |

→ `dyson-swarm.md`, `solar-monetary-authority.md`, `pure-atp.md`