Orbital Thermal Balance & Solder Fatigue

Every picosatellite in low Earth orbit exchanges heat solely by radiation against four thermal inputs: direct solar irradiance S₀ = 1,361 W/m² intercepted over the tumbling projected area A/4, sunlight reflected as planetary albedo (mean factor a = 0.30, peaking at the sub-solar point as max(cos φ, 0)), Earth's continuous outgoing longwave infrared flux q_IR = 237 W/m², and internal avionics dissipation Q_int, balanced against fourth-power grey-body emission ε·σ·A·T⁴ with σ = 5.670 × 10⁻⁸ W/(m²·K⁴).

Both planetary fluxes couple into the hull through the orientation-averaged spherical Earth view factor F(h) = ½·[1 − √(1 − (R_E / (R_E + h))²)], which evaluates to F = 0.305 at h = 550 km (95.5 min period, 37.2% eclipse fraction at zero beta angle). Because outgoing infrared depends on thermal emittance ε just like radiative rejection, Earth IR sets a cold-case floor while the ratio of solar absorptance to infrared emittance (α/ε) governs the orbital-mean temperature as (α/ε)^(1/4).

On a 250 g 1P hull, low-α/ε white paint drives the orbital average down near −53 °C and clear-anodized aluminium sits near −28 °C, whereas triple-junction solar cells, black paint, and black anodize (α ≈ ε ≈ 0.85) hold the mean inside +5 to +9 °C. Uncoated bare aluminium on structural rails or the deployment backplate (low ε, higher α/ε) shifts equilibrium temperatures sharply, and ultraviolet plus charged-particle exposure gradually raises solar absorptance α from beginning of life to end of life.

Linearizing fourth-power emission around a reference temperature T₀ = 290 K yields the radiative thermal time constant τ_th = m·c_p / (4·ε·σ·A_tot·T₀³), which scales directly with the areal mass ratio m / A_tot. At c_p = 900 J/(kg·K) and ε = 0.80, a 250 g 1P cube (150 cm² exterior area) exhibits τ_th ≈ 56 min, compared with ≈ 104 min for a 1.33 kg 1U CubeSat and ≈ 87 min for a 3U bus, while elongated 2P and 3P hulls remain near the 1P value.

Across a 35–36 min LEO umbral transit, a 1P spends 0.63·τ_th in shadow versus 0.34·τ_th for a 1U, completing 1 − exp(−t_ecl / τ_th) ≈ 46–47% of exponential relaxation toward its cold asymptote compared with only ≈ 29% on 1U and ≈ 33% on 3U. In full single-node integration at 550 km (α = ε = 0.85, c_p = 858 J/(kg·K), Q_int = 0.25 W), a 250 g 1P cycles between −10.8 °C and +19.5 °C (ΔT = 30.3 K, widening to 36.8 K at the 200 g SMOG-1 mass), whereas 1.33 kg and 2.00 kg 1U hulls with 1.0 W internal power swing 23.4 K and 15.8 K; the /lab/thermal 525 km baseline (0.24 kg, α = 0.62, ε = 0.78, Q_int = 0.32 W) spans −16.0 °C to +7.4 °C (ΔT = 23.4 K).

Fourth-power radiation also makes heating and cooling asymmetric: the 250 g 1P warms at up to ≈ 0.75 K/min at sunlit emergence but plunges at 1.2 K/min upon eclipse entry while the hull is still warm. Thermally isolated FR-4 solar panels (through-thickness k ≈ 0.3 W/(m·K) versus 155–167 W/(m·K) for Al 6061-T6 / 7075-T73) swing much wider than bulk nodes, reaching −26 °C to +15 °C in SMOG-1 finite-element models and up to 75 °C total excursion in Delfi-PQ flight telemetry.

At 15.1 orbits per day, a low-Earth-orbit spacecraft accumulates roughly 5,500 thermal cycles every year, turning moderate temperature swings into a low-cycle thermomechanical fatigue problem. Structural aluminium alloys (6061-T6 and 7075-T73) expand at 23.4–23.6 ppm/K, woven FR-4 PCB laminates expand at 14–17 ppm/K in-plane (and 50–70 ppm/K out-of-plane below glass transition, straining plated through-holes), whereas germanium triple-junction solar cells and ceramic QFN or BGA packages expand at only ≈ 6 ppm/K.

Across a rigid leadless solder joint of standoff height h_s and neutral-point-to-corner distance L_D, differential expansion Δα ≈ 10–11 ppm/K imposes cyclic shear strain range Δγ = (F_E · L_D · Δα · ΔT) / h_s each orbit. Under the IPC-SM-785 Engelmaier creep-fatigue model implemented in lib/physics/thermal.ts (eutectic Sn–Pb fatigue ductility ε'_f = 0.325 and temperature/frequency-dependent exponent c ≈ −0.442 − 6×10⁻⁴·T_sj + 0.0174·ln(1 + f) ≈ −0.55 to −0.60), cycles to 50% failure scale as ΔT^(1/c), so a 30% wider orbital ΔT cuts fatigue life nearly in half.

For a rigid germanium-on-FR-4 cell joint (Δα ≈ 11 ppm/K) experiencing ΔT = 30.3 K on a 250 g 1P, median fatigue life drops near N_f ≈ 5,100 cycles—under one orbital year, versus roughly 1.8 years on a 1.33 kg 1U CubeSat. In the /lab/thermal leadless package benchmark (L_D = 4 mm, h_s = 75 µm, Δα = 10 ppm/K, ΔT = 23.4 K), median life reaches N_f = 10,661 cycles (~1.9 years), whereas CTE-matched ceramic or metal-core substrates (Δα = 2 ppm/K) extend life beyond decades.

Wide 1P temperature swings propagate directly into adjacent subsystems documented in monograph SQ-R-002: Li-ion graphite anodes cross the 0 °C lithium-plating boundary during eclipse, sintered NdFeB permanent magnets lose ≈ 12% of remanence across 100 K (versus 2–3.5% for AlNiCo 5 and SmCo), copper PCB magnetorquer traces change resistance at +0.39%/K (producing >10% dipole drift per orbit under fixed bus voltage unless current-regulated), and unheated water micropropulsion lines freeze. Crucially, shaving structural mass off a 1P hull lowers thermal capacitance m·c_p and widens ΔT further.

Thermomechanical mitigation combines surface finish tailoring (balanced α/ε from solar cells and selective anodizing), conductive thermal straps linking sunlit and shaded panels, and compliant interconnect geometry: space-grade low-outgassing silicone adhesives (CVCM ≤ 0.1%, TML ≤ 1.0%) or stress-relieved silver/kovar solar-cell tabs replace rigid solder fillets, split solder pads shrink effective L_D, spacer beads raise standoff h_s, and leaded eutectic Sn63Pb37 solder or hot-solder-dipped terminations per ECSS-Q-ST-70-38C suppress vacuum pure-tin whisker shorts.

Before flight, ECSS-E-ST-10-03C thermal-vacuum cycling (such as Delfi-PQ's −20 °C to +50 °C qualification campaign) exposes infant-mortality joints, outgassing condensates, and cold-stalled mechanisms. Across the 32 CubeSat and PocketQube anomalies catalogued in monograph SQ-R-004, 71.4% of materials-initiated failures occurred on or before orbital day 60 (median onset day 1), and standard pre-flight TVAC cycling, vacuum deployment, and vibration tests would plausibly have caught 90.5% (19 of 21) prior to launch.