Cryogenic Cable
Balancing RF Loss Against Heat Leak
The defining problem of a cryogenic cable is that the two properties an engineer cares about, electrical conductivity and thermal conductivity, track each other through the Wiedemann-Franz law for ordinary metals. A wire that carries RF beautifully also carries heat beautifully, and in a dilution refrigerator the mixing-chamber stage may have only a few microwatts of cooling power at roughly 10 mK. Routing dozens of read-out and control lines down to that plate with solid copper coax would overwhelm the cryocooler instantly. The accepted solution is to break the Wiedemann-Franz link at the surface: plate a thin layer of silver or gold (which only matters for the RF skin depth, a fraction of a micron at 10 GHz) over a body of 304 stainless steel, whose bulk thermal conductivity collapses from about 15 W/m-K at room temperature toward 0.3 W/m-K near 4 K.
For the lowest-loss links, particularly the read-out lines that must preserve a weak signal coming up from a quantum device, designers switch to NbTi superconducting coax. Below its critical temperature of about 9.2 K the niobium-titanium center conductor and shield carry RF with vanishing resistive loss while conducting almost no heat, because in the superconducting state the electronic contribution to thermal conductivity is suppressed. The penalty is that NbTi cannot pass meaningful DC current and ceases to superconduct above its critical field and frequency, so it is reserved for the cold, low-power stages.
Thermal management does not stop at material choice. Each cable is clamped and heat-sunk to every temperature plate so the conducted heat is intercepted there rather than landing on the coldest stage, and fixed attenuators are added at the 4 K and still plates to absorb the room-temperature thermal-noise photons traveling down the input lines. This is why a cryogenic coax assembly is specified not just by frequency and impedance but by its per-stage thermal load.
The Governing Heat-Conduction Relations
Q ≈ (A / L) × ∫TcTh k(T) dT W
Skin-effect attenuation (lossy metal coax):
α ∝ Rs ≈ √(π f μ ρ) → loss ∝ √f
Wiedemann-Franz (why copper is avoided):
k / σ = L0 × T, L0 ≈ 2.44 × 10−8 W·Ω/K2
Where A = conductor cross-section, L = length between stages, k(T) = temperature-dependent thermal conductivity, Rs = surface resistance, f = frequency, ρ = resistivity, σ = electrical conductivity, T = temperature. Example: a 0.085 in stainless coax conducts ≈ 1 to 5 μW between the 300 K and 4 K stages.
Cryogenic Cable Material Comparison
| Conductor Type | Loss @ 10 GHz | Rel. Heat Leak | DC Current | Useful Range | Typical Stage |
|---|---|---|---|---|---|
| Solid copper (reference) | ~0.5 dB/m | Very high (1x) | Excellent | DC to 110 GHz | 300 K bench only |
| 304 stainless steel | 4 to 8 dB/m | Low (~0.02x) | Limited | DC to 40 GHz | 300 K → 4 K input |
| Silver-plated SS | 2 to 5 dB/m | Low to moderate | Moderate | DC to 50 GHz | 300 K → 4 K |
| Beryllium copper | 1 to 3 dB/m | Moderate | Good | DC to 40 GHz | 4 K stage runs |
| NbTi superconductor | < 0.1 dB/m (cold) | Very low | None (RF only) | To ~12 GHz, below 9.2 K | Cold read-out lines |
Frequently Asked Questions
Why are cryogenic coaxial cables made from stainless steel or NbTi instead of copper?
Copper conducts both electricity and heat extremely well (k ≈ 400 W/m-K), so solid copper coax would dump unacceptable heat onto each cold plate. Cryogenic cables use poor thermal conductors instead: 304 stainless steel (about 15 W/m-K at 300 K, near 0.3 W/m-K at 4 K), beryllium copper, or NbTi. A thin silver or gold plating restores RF skin-effect surface conductivity while the steel body keeps bulk heat conduction low. Below 9.2 K, a NbTi center conductor carries RF with near-zero loss and almost no heat.
How much heat does a single cryogenic coax conduct between temperature stages?
It follows Q = (A/L) × ∫k(T)dT between stage temperatures. A 0.085 in (2.2 mm) semi-rigid stainless coax leaks roughly 1 to 5 μW between the 4 K and still (~0.7 K) plates, and 0.1 to 1 μW between the still and mixing chamber. Because a refrigerator may carry dozens of lines, designers heat-sink and attenuate each cable at every plate to keep the total under the few-microwatt budget of the coldest stage.
What insertion loss should I expect from a stainless steel cryogenic cable at microwave frequencies?
Stainless coax is intentionally lossy: about 4 to 8 dB/m at 10 GHz, rising with √f to roughly 8 to 15 dB/m near 40 GHz. Silver plating reduces this by 30 to 50 percent, and loss drops slightly as the cable cools. For low-loss read-out, NbTi superconducting coax can reach under 0.1 dB/m once below its transition temperature, at the cost of carrying essentially no DC current.