Quantum Computing and Quantum RF Cryogenic Microwave Engineering Informational

How do I calculate the thermal budget for microwave components at the mixing chamber stage?

The mixing chamber (MC) of a dilution refrigerator provides limited cooling power, typically 10-20 microwatts at 20 mK and 200-400 microwatts at 100 mK (scaling approximately as T^2). Every microwave component at this stage contributes heat load that must fit within this budget. The thermal budget calculation sums all heat sources: (1) Thermal conduction through coaxial cables: each cable from the cold plate to MC conducts heat determined by its thermal conductivity integral. Stainless steel semi-rigid cable (UT-085-SS): ~0.5 microwatts per cable for a 20 cm length between 100 mK and 20 mK. NbTi superconducting cable: ~0.05 microwatts (10× lower due to superconducting outer conductor). (2) Microwave signal dissipation: a 20 dB attenuator at MC dissipating -50 dBm (10 nW) absorbs 9.9 nW. With 10 qubit lines, total: ~100 nW. (3) Active components (circulators, isolators): residual magnetic heating from ferrite elements, typically 0.02-0.1 microwatts each. (4) Passive heat leaks: connector thermal paths, support structures, wiring for DC bias lines. Total MC thermal budget allocation: cables (40-50%), attenuators/filters (20-30%), active components (10-15%), contingency (10-20%). For a 50-qubit system with 150+ microwave lines, the total heat load at MC can reach 5-15 microwatts, consuming a large fraction of the available cooling power and requiring careful thermal design.

Category: Quantum Computing and Quantum RF

Updated: April 2026

Product Tie-In: Cryogenic Components, Attenuators, Circulators, Cables

Mixing Chamber Thermal Budget

The mixing chamber thermal budget is one of the most critical constraints in scaling quantum computers. As qubit count increases, the number of microwave lines grows proportionally (2-3 lines per qubit for control, readout, and flux bias), and the total heat load can exceed the dilution refrigerator's cooling capacity, raising the base temperature and degrading qubit performance.

Heat Load Sources

Cable thermal conduction: the dominant heat source. Each coaxial cable conducts heat proportional to its thermal conductivity integral between temperature stages. For a cable of cross-sectional area A, length L, between temperatures T_cold and T_hot: Q = (A/L) × integral(k(T)dT from T_cold to T_hot). Common cable materials and their thermal load per cable (20 cm length, 100 mK to 20 mK): Stainless steel UT-085-SS: ~0.3-0.5 μW. CuNi (copper-nickel): ~0.2-0.4 μW. NbTi superconducting: ~0.03-0.08 μW. For 200 cables (typical for a 50-100 qubit system): stainless steel: 60-100 μW (exceeds MC cooling capacity). NbTi: 6-16 μW (feasible). This is why large quantum computers use NbTi superconducting cables for the MC-to-cold plate section despite their higher cost ($200-500 per cable vs $20-50 for stainless steel).

Budget Allocation Strategy

A disciplined thermal budget allocates the available cooling power across all heat sources with margin: Total available at 20 mK: 15 μW (typical for BlueFors LD-400 or similar large DR). Allocation: Coaxial cables (NbTi): 6 μW (40%). DC wiring (filtered phosphor bronze or manganin): 2 μW (13%). Attenuators and filters: 1.5 μW (10%). Circulators/isolators: 1 μW (7%). Support structure and radiation: 1 μW (7%). Contingency: 3.5 μW (23%). This budget supports approximately 200 NbTi cables. For 1000+ qubit systems, additional measures are needed: thermalization schemes that pre-cool cables at intermediate stages, multiplexing to reduce cable count, and higher-capacity dilution refrigerators (50-100 μW at 20 mK from systems like BlueFors XLDsl).

Measurement and Verification

Verify the thermal budget experimentally by: (1) Measuring the base temperature as a function of applied heat load (using a resistive heater on the MC). This gives the actual cooling power curve. (2) Monitoring MC temperature as cables and components are installed incrementally, tracking the temperature rise per added component. (3) Using calibrated cryogenic thermometers (RuOx or Cernox) at each stage to verify temperatures match the thermal model. Common discrepancies: radiation leaks through gaps in radiation shields (can add 1-10 μW at MC), thermalization failures (attenuator or filter not in good thermal contact, remaining warmer than the stage), and eddy current heating from time-varying magnetic fields from neighboring experiments or building vibrations.

Thermal Budget Equations

Cable Heat Load: Q = (A/L) × ∫k(T)dT
MC Cooling Power: Q_cool ≈ 15μW × (T/20mK)²
Thermal Photon Power: P_noise = kT × Δf × n_th
Max Cables: N_cable ≤ Q_cool_budget / Q_per_cable

Common Questions

Frequently Asked Questions

What is the cooling power of a typical dilution refrigerator?

Modern commercial dilution refrigerators provide approximately 10-25 μW at 20 mK (base temperature), 200-500 μW at 100 mK, and 10-20 mW at 700 mK (still stage). Larger systems (BlueFors XLDsl, Oxford Instruments Proteox): up to 50-100 μW at 20 mK, 1 mW at 100 mK. Cooling power scales approximately as T^2 near base temperature. The 4K stage (pulse tube cooler) provides 1-2 W of cooling power, which is comparatively abundant. All heat loads should be intercepted at the warmest possible stage to minimize impact on the MC.

How many qubit lines can a dilution refrigerator support?

With NbTi cables and careful thermal engineering: a standard DR (15 μW at 20 mK) supports 100-200 coaxial lines. A large DR (50 μW at 20 mK) supports 300-600 lines. IBM Quantum System Two uses multiple custom DRs to support 1,000+ qubit systems. Google Sycamore (53 qubits) uses approximately 150 coaxial lines in a single DR. Reducing the number of lines through frequency multiplexing (multiple qubits per readout line) and cryo-CMOS control electronics (moving DACs/ADCs to the 4K stage to reduce cable count) are active research areas to break through the cabling bottleneck.

What happens if the thermal budget is exceeded?

Exceeding the MC cooling power raises the base temperature. A 15 μW DR with 20 μW total heat load will stabilize at a higher temperature, approximately 25-30 mK instead of 20 mK. The consequences: thermal photon population increases (n_th at 5 GHz: 0.001 at 20 mK vs 0.003 at 30 mK), quasiparticle density increases exponentially, qubit T1 may degrade by 20-50%, and readout fidelity decreases. In severe cases (2-3× cooling power exceeded), the MC temperature rises to 50-100 mK, rendering superconducting qubits inoperable. Always design with 20-30% thermal margin to accommodate unexpected heat loads and system aging.

Keep Reading