Quantum Computing and Quantum RF Cryogenic Microwave Engineering Informational

What is the required attenuation at each temperature stage of a dilution refrigerator for qubit control lines?

The required attenuation at each temperature stage of a dilution refrigerator for qubit control lines is determined by the need to thermalize the microwave noise at each stage, reducing the thermal photon population to the equilibrium value at that temperature. Standard attenuation distribution for a 5-stage cryostat: Room temperature to 50K: cable loss only (0.5-2 dB), no discrete attenuator needed. 50K to 4K: typically no attenuator (rely on cable loss of 1-3 dB). At 4K plate: 20 dB attenuator. This reduces the effective noise temperature from ~300K to ~3K (the attenuator thermalizes at 4K, and the 20 dB kills 99% of the 300K noise). Still plate (800 mK): 6-10 dB attenuator. Thermalizes the 4K noise contribution. Cold plate (100 mK): 10 dB attenuator. Mixing chamber (20 mK): 20 dB attenuator. Total: approximately 60 dB from room temperature to qubit. The exact values depend on: (1) Available cooling power at each stage. (2) Number of control lines sharing each stage. (3) Average signal power delivered to the lines. (4) Cable type (stainless steel vs NbTi, affecting cable loss). (5) Target residual photon number at the qubit (<0.01 for high-coherence qubits). More attenuation at warmer stages (where cooling power is abundant) reduces the burden on colder stages. The 4K stage is the most cost-effective location for attenuation: 1.5W of cooling power can trivially absorb the heat from dozens of 20 dB attenuators.
Category: Quantum Computing and Quantum RF
Updated: April 2026
Product Tie-In: Cryogenic Components, Attenuators, Circulators, Cables

Stage-by-Stage Attenuation Requirements

The attenuation distribution is a fundamental design choice in quantum computing cryostats that directly affects qubit coherence, control signal fidelity, and thermal budget. Different research groups use slightly different distributions based on their refrigerator capabilities and qubit requirements.

ParameterOption AOption BOption C
PerformanceHighMediumLow
CostHighLowMedium
ComplexityHighLowMedium
BandwidthNarrowWideModerate
Typical UseLab/militaryConsumerIndustrial

Technical Considerations

At each stage, the attenuator serves two functions: (1) Kill the noise from warmer stages. (2) Thermalize the noise to the local temperature. After a perfect attenuator with value A (dB) at temperature T_stage, the effective noise temperature of the signal is: T_eff = T_input × 10^(-A/10) + T_stage × (1 - 10^(-A/10)). For A = 20 dB at T_stage = 4K with T_input = 300K: T_eff = 300 × 0.01 + 4 × 0.99 = 3 + 3.96 = 6.96K. The output noise is dominated by the stage temperature, which is the goal. If A = 10 dB instead: T_eff = 300 × 0.1 + 4 × 0.9 = 30 + 3.6 = 33.6K. The 300K noise is incompletely suppressed. Rule of thumb: the attenuation at each stage should be large enough that the first term (input noise × attenuation factor) is less than or equal to the stage temperature. For 300K input to 4K: need 300 × 10^(-A/10) ≤ 4, so A ≥ 10*log10(75) = 18.8 dB, confirming the standard 20 dB at 4K.

Performance Analysis

Google Sycamore (53 qubits): 20 dB at 4K, 20 dB at mixing chamber, 0-10 dB at intermediate stages. Total ~50-60 dB. IBM Quantum: 20 dB at 4K, 10 dB at cold plate, 20 dB at MC. Total ~50-60 dB. Academic labs (typical): 20 dB at 4K, 6 dB at still, 10 dB at cold plate, 20 dB at MC. Total ~56-60 dB. Some groups add extra attenuation at room temperature (10-20 dB in a fixed attenuator between the DAC output and the cryostat input) to reduce the signal level entering the cryostat, easing the thermal load at all stages. The room-temperature attenuator contributes no thermal benefit (it is at 300K) but reduces the absolute power dissipated at cryogenic stages.

  • Performance verification: confirm specifications against the application requirements before finalizing the design
  • Environmental factors: temperature range, humidity, and vibration affect long-term reliability and parameter drift
  • Cost vs. performance: evaluate whether the application demands premium components or standard commercial grades
  1. Interface compatibility: verify impedance, connector type, and mechanical form factor match the system architecture

Design Guidelines

The signal power at each stage determines the heat dissipated: Typical qubit control pulse at room temperature: -20 to 0 dBm (10 μW to 1 mW), very short duty cycle (0.01-1%). After 20 dB at 4K: -40 to -20 dBm at 4K. Heat dissipated in the 4K attenuator: 9.9 μW to 0.99 mW average (with duty cycle, this is negligible compared to 1.5W cooling). After another 40 dB to MC: -80 to -60 dBm at qubit. Heat at MC attenuator: 0.99 nW to 99 nW. With duty cycle of 0.1%: 1-100 pW average. This is 1000× below the MC cooling power of 15 μW, confirming the thermal feasibility. The qubit itself requires approximately -60 to -40 dBm at its input for typical gate operations (pi pulse power), so the total attenuation must be calibrated to deliver the correct signal amplitude at the qubit while maintaining the noise thermalization.

Common Questions

Frequently Asked Questions

Can I put all the attenuation at one stage?

No. If all 60 dB is at the 4K stage with no attenuation below, the cable from 4K to MC acts as a waveguide for 4K thermal noise, delivering ~16 photons at 5 GHz directly to the qubit. The cable does not attenuate the noise sufficiently (cable loss from 4K to MC is only 1-3 dB). Distributing attenuation ensures that noise is thermalized at each stage, reducing the photon population step by step from 1250 (300K) to 17 (4K) to 0.4 (100 mK) to 0.08 (20 mK). Each attenuator acts as a thermal reset point.

Does the readout line also need attenuation?

No. The readout output line carries the weak qubit signal that must be amplified, not attenuated. Placing attenuators on the output line would reduce the already tiny readout signal below the noise floor of the amplifier chain. Instead, the output line uses circulators/isolators for backward isolation (protecting the qubit from amplifier noise) and low-loss cables to preserve the signal. The output chain: qubit → circulator(s) → quantum-limited amplifier (JPA/TWPA at MC) → low-loss cable to 4K → HEMT LNA at 4K → cable to room temperature → room-temperature receiver.

What if my attenuator does not thermalize properly?

A poorly thermalized attenuator does not reset the noise temperature to its stage temperature. If a 20 dB attenuator at the MC stage is thermally floating at 200 mK instead of 20 mK (due to poor contact or insufficient thermal conductance), the output noise temperature is T_eff = T_in × 0.01 + 0.2 × 0.99 ≈ 0.2K instead of 0.02K. The residual photon number at 5 GHz increases from 0.083 to 0.83, a 10× degradation that significantly affects qubit coherence. Symptoms: unexpectedly low T1, elevated qubit excited-state population, and temperature-dependent performance changes when cycling the cryostat. Fix: ensure gold-plated copper mounting bracket with indium foil gasket, torque SMA connectors to specification, and verify thermometer reading at the attenuator body matches the stage temperature.

Need expert RF components?

Request a Quote

RF Essentials supplies precision components for noise-critical, high-linearity, and impedance-matched systems.

Get in Touch