The beam architecture

Array diameter is set by the sail

A tougher membrane buys a smaller array.

The beam must stay smaller than the sail for the entire acceleration. A phased array of diameter D produces a diffraction-limited spot whose diameter at range L is approximately λL/D, and requiring that spot to fit inside a sail of diameter d gives a minimum array diameter of λL/d. The acceleration distance is the final velocity squared over twice the acceleration. So the array diameter is set, inversely, by how much acceleration the sail can survive.

Modeled, assuming a 1.06-micron beam, a 4-meter sail, perfect reflectivity and no pointing loss. Real systems carry a factor of three to six margin on power for spillage, reflectivity below unity and jitter, which is why Starshot specified 100 GW against an ideal of 15.

That table contains the single most important design coupling in the architecture and it is rarely stated this way. The structural limit of a 100-nanometer dielectric membrane under 10,000 g sets the capital cost of a multi-kilometer phased optical array, and the two problems belong to entirely different engineering disciplines that must be optimized together.

Note the direction the trade runs. Halving the acceleration triples the array diameter and cuts the required power by two thirds. Since array cost scales roughly with area and power cost scales with peak generation, there is a genuine optimum, and where it sits depends on which is cheaper in 2080: square kilometers of phased optics, or gigawatts of collected sunlight. In an economy with a thin-film industry at 0.39 AU, sunlight is cheaper, which pushes the design toward high acceleration and small arrays.

Where the power comes from

A few hundred tons of film.

Solar flux is 1,361 watts per square meter at 1 AU and scales as the inverse square. At 0.39 AU, Mercury’s semi-major axis, it is 8,948. At 0.2 AU it is 34,025.

Delivering 100 gigawatts at 20 percent collection-to-beam efficiency requires 5.59 × 10⁷ square meters of collector at 0.39 AU — 56 square kilometers, a square 7.5 kilometers on a side. At a thin-film areal density of 10 grams per square meter, that masses 559 tons. At 0.2 AU the same beam needs 15 square kilometers and 147 tons.

In an economy with a lunar mass driver moving millions of tons per year and orbital thin-film production already established for compute and power infrastructure, that part is a satellite program rather than a megaproject. Two corrections are owed to that sentence.

Two corrections from Mercury

The radiator, and what is actually shared.

The first correction: 559 tons is the collector alone, and in the obvious architecture the collector is not the system.

Sunlight is incoherent and the Sun is an extended source, so brightness conservation forbids collimating sunlight directly. The light must be converted to coherent light, conversion has losses, and in space losses become radiators. At 20 percent chain efficiency, delivering 100 gigawatts means collecting 500 and rejecting 400, most of it at the laser near 350 to 400 kelvin: on the order of 100 square kilometers of radiator and roughly 320,000 tons, six hundred times the film. Worse, that radiator is the same size at 1 AU, so the whole location advantage of being at 0.39 AU is worth about one percent of system mass and the array might as well be in Earth orbit.

The escape is the sorting rule’s own prediction: an architecture whose losses are rejected optically rather than thermalized. At 8 percent solar-to-laser, 100 gigawatts needs 137 square kilometers of collector at Mercury and a 4.4 square kilometer radiator — on the order of 15,000 tons, and now the full flux advantage applies, because the collector is the system. Solar-pumped laser efficiency, demonstrated in the low single digits and needing five to eight percent, is a named open item here as well as at Mercury.

The second correction is about what is shared with the inner-system array and what is not.

It is tempting to say that the Mercury constellation is the power plant of an interstellar launcher without any modification to its purpose. That is not right. A sail beam line wants a different wavelength agility, a different duty cycle, an acceleration tolerance five orders of magnitude away from a cargo tug’s, and optics matched to a four-meter target across a 22 percent Doppler band.

A correction on “statite”

1.53 grams per square meter, everywhere.

A statite is a sail that hovers by balancing solar radiation pressure against solar gravity. Radiation pressure on a perfect reflector at 1 AU is 9.08 × 10⁻⁶ newtons per square meter; solar gravitational acceleration at 1 AU is 5.93 × 10⁻³ meters per second squared. Both scale as the inverse square of heliocentric distance, so the balance condition is independent of where you are.

The critical areal density for a true statite is 1.53 grams per square meter, everywhere in the solar system. That is a factor of six below the 10 grams per square meter assumed above, and well below anything flown: IKAROS and LightSail 2 were both near 10.

So the 56-square-kilometer collector is an orbiting constellation, not a levitating one, and this site says so rather than using “statite” loosely. It matters less than it first appears. A sail with a lightness number below unity cannot hover, but it can hold non-Keplerian orbits — displaced above the ecliptic, or precessing off the Keplerian rate — which is most of what a beam array wants from its station-keeping. True statites become available if thin-film areal density falls below 1.53 grams per square meter, which is a manufacturing threshold rather than a physical one.

What has just been specified

A launcher, and the most capable weapon in the architecture.

The array derived above is between 4.8 and 47.6 kilometers across and delivers 100 gigawatts. It is worth stopping to say plainly what that object is.

At the small end, 4.8 kilometers, it holds structural steel at its melting point to about 6 AU and aluminum to about 21. At the large end, 47.6 kilometers, those become 60 and 212 AU — past Pluto, and most of the way to the gravitational-lens working distance. Mercury treats a 5-kilometer array as a strategic asset and as the reason a defense constituency for that destination exists at all. This is up to an order of magnitude more capable, and has so far been called a launcher.

The saving grace is the one Mercury derives, and it applies here with more force rather than less. A shooter cannot correct faster than one round-trip light time. The dodge margin d/s = aLD / 1.22λc² grows with range and aperture together, so the 47.6-kilometer array — the best-collimated instrument anywhere in the architecture — breaks even against a one-millimeter-per-second-squared target at 0.016 AU and is defeated by an ion thruster at every range where it has useful flux. Spreading the beam to guarantee a hit costs (d/s)² in intensity, which at these ranges is four to five orders of magnitude.

Two consequences carry forward.

The array’s weapon character is not an argument against building it. The same physics that makes it fearsome makes it nearly useless against anything that moves, and its lethal envelope covers fixed installations a launcher has no reason to shoot at. What it does mean is that the array will attract defense funding and defense oversight, which is how large arrays get paid for at all, and that an interstellar program will find itself institutionally downstream of a weapons program whether or not it wants to be.

The sail is the ideal target. It is a gram-class object with no thrust authority whatever, a deliberately enormous cross-section, and a published trajectory — fixed, unpowered and predictable at once, which is the entire hittable set. Everything that makes it accelerate makes it defenseless. The swarm below is the only mitigation available, and it was adopted for entirely unrelated reasons. That is luck, and it should be named as luck rather than as foresight.

The swarm

The one regime where attrition beats redundancy.

Because the vehicle is a gram, its marginal cost is the energy of the shot rather than the hardware. At 100 gigawatts for 600 seconds each launch consumes 6 × 10¹³ joules, or 16.7 gigawatt-hours. A collector array sized for continuous 100-gigawatt operation can fire as fast as it can retarget, which is limited by beam-line thermal recovery and orbital geometry rather than by energy. Even at one shot per day, a 15-year campaign launches about 5,000 craft.

This inverts the design philosophy of every other destination on this site. The no-rescue principle holds that because nothing can be rescued, everything must be dual-redundant, gracefully degrading and repairable in place. That governs every industrial destination. It does not govern here.

The redundancy is not internal to the vehicle. It is statistical and external. You do not engineer the craft to survive. You engineer the population to survive, and you accept a 90 percent loss rate because the marginal craft costs a day of sunlight.

This is the only regime in the architecture where attrition beats redundancy, and the reason is that it is the only regime where the vehicle is cheaper than the environment is hostile. A reader who has absorbed the high-availability doctrine that governs the Moon or Callisto will otherwise find these tolerances baffling.