Mercury receives ~9,100 W/m² of solar flux — roughly 6.7× what Earth's surface gets at noon on the equator. A solar panel that yields 200 W on your roof would deliver about 1,340 W there. Tempting. Let's see what breaks.

The thermal problem dominates everything. Mercury's dayside surface hits 700 K (427 °C). Silicon photovoltaics lose roughly 0.4%/°C above 25 °C; extrapolating naively, a Si panel at 700 K would have negative efficiency — it'd be a heater. Real answer: silicon's bandgap collapses thermally and the p-n junction stops rectifying around 200 °C. You need wide-bandgap cells: GaInP/GaAs/Ge multijunctions with active cooling, or thermophotovoltaics tuned for the spectrum.

So we cheat: build at the poles. Mercury's axial tilt is only 2°, so polar crater rims see near-continuous sunlight at grazing incidence (~250 K ambient — manageable) while crater floors stay in permanent shadow at 100 K. Perfect radiator sink right next door.

Back-of-envelope: a 1 GW polar array.

• Grazing solar flux at the rim: ~9,100 × sin(5°) ≈ 790 W/m² (about Earth-equivalent — the irony)
• Multijunction efficiency at 250 K: ~38%
• Required collector area: 1×10⁹ / (790 × 0.38) ≈ 3.3 km²
• Mass at 2 kg/m² thin-film mounted on lightweight truss: ~6,600 tonnes

For comparison, the ISS took 42 launches to assemble 420 tonnes. You'd need roughly 16× that, but to Mercury — a Δv budget of ~13 km/s from LEO using gravity assists, versus 4 km/s to the Moon. Mercury is harder to reach than Pluto because you're falling into the Sun's gravity well and need to brake hard. The MESSENGER probe took 6.5 years and used six planetary flybys to shed velocity.

Now the real question: how do you get the power home?

Wires won't work — 77 million km of superconductor isn't happening. The two options:

• Microwave beam: A 1 GW beam at 10 GHz needs a transmitter aperture of roughly D ≈ 1.22 × λ × R / d_rx. For a 10 km receiver on Earth and R = 1×10¹¹ m, transmitter diameter ≈ 370 km. Unbuildable.
• Laser at 1 μm: Same equation gives transmitter diameter ≈ 12 m. Suddenly tractable. But atmospheric absorption + pointing stability at 0.0000007 arcseconds is its own nightmare.

The honest use case isn't powering Earth. It's powering Mercury itself — running a robotic mining operation refining the planet's iron-nickel crust (Mercury is 70% metal by mass) and slinging refined material out via electromagnetic mass driver. Mercury's escape velocity is 4.25 km/s, low enough that a 100 km railgun pulling 9 g could launch payloads on Hohmann transfers to Earth or the asteroid belt.

The structural curveball: Mercury has no atmosphere, but its day is 176 Earth days long, and the surface undergoes 600 K thermal cycling. Aluminum truss with α=23 μm/m/K expanding across 600 K means 1.4% length change — a 100 m beam shifts 1.4 m every Mercury day. You'd need invar (α=1.2) or compliant joints, not rigid welds. Anything bolted will fatigue-crack within years.