Reliability Engineering

Bathtub Curve

Q: What are the three phases of the bathtub curve?

Phase 1 is infant mortality (decreasing failure rate), lasting from hours to a few hundred hours. Failures come from manufacturing defects: solder voids, contamination, wire bond defects, and die attach issues. Burn-in at elevated temperature (125 to 175 degrees Celsius) and voltage stress screens these out before shipping. Phase 2 is the useful life region (constant failure rate, lambda), lasting years to decades. Failures are random and unpredictable: cosmic ray upsets, ESD events, random contamination activation. The constant failure rate lambda is expressed in FIT (failures in 10^9 device-hours). MTBF = 1/lambda. Phase 3 is wearout (increasing failure rate), where cumulative degradation mechanisms dominate: electromigration in interconnects (current density dependent), hot carrier injection in MOSFETs, gate dielectric breakdown, and intermetallic growth in wire bonds. The Weibull distribution models this phase with shape parameter beta greater than 1.

Q: How are RF component reliability ratings determined?

RF semiconductor manufacturers use accelerated life testing: devices are stressed at elevated junction temperatures (200 to 300 degrees Celsius for GaN, 250 to 350 degrees for GaAs) and extrapolated to operating temperature using the Arrhenius model. The activation energy E_a determines how much temperature acceleration is achieved: E_a = 1.5 to 2.0 eV for GaN gate degradation, 1.0 to 1.5 eV for GaAs HBT degradation. Acceleration factor AF = exp(E_a/k times (1/T_use minus 1/T_stress)). A 1000-hour test at 300 degrees Celsius with E_a = 1.7 eV is equivalent to approximately 10 million hours at 150 degrees Celsius junction temperature. Typical results: GaN HEMT MTTF greater than 10^6 hours at 200 degrees Celsius channel temperature, GaAs HBT MTTF greater than 10^7 hours at 150 degrees Celsius junction temperature.

Q: What is the difference between MTBF and MTTF?

MTTF (Mean Time To Failure) applies to non-repairable items: it is the average time before first failure. MTTF = 1/lambda during the constant failure rate phase. MTBF (Mean Time Between Failures) applies to repairable systems: it includes both the operating time and repair time in the calculation, representing the average time between successive failures of a system that is restored to operation after each failure. For individual RF components (transistors, capacitors, ICs), MTTF is the correct metric since they are replaced, not repaired. For RF systems (base stations, radar units), MTBF is appropriate. A base station with MTBF of 50,000 hours and MTTR (Mean Time To Repair) of 4 hours has an availability of MTBF/(MTBF+MTTR) = 99.992%.

/bath-tub kerv/

A plot of component failure rate (λ) versus operational time that shows three distinct phases: infant mortality (decreasing rate, screened by burn-in), useful life (constant rate, characterized by FIT or MTBF), and wearout (increasing rate, driven by electromigration, hot carriers, or oxide breakdown). For RF semiconductors: GaAs HBT MTTF >10⁷ hours at 150°C T_j; GaN HEMT MTTF >10⁶ hours at 200°C channel temperature. Arrhenius acceleration enables extrapolation from 1000-hour stress tests.

Model: Weibull / Exponential

Unit: FIT (10⁹ hrs)

Burn-in: 125-175°C

Understanding the Bathtub Curve

The bathtub curve is the foundational model of hardware reliability. Its shape arises from the superposition of two competing mechanisms: early-life defect-driven failures that decrease over time as weak units are eliminated, and late-life degradation-driven failures that increase as cumulative damage accumulates. Between these two regions, the failure rate is approximately constant, and failures are considered random. This constant-rate region is where reliability specifications like MTBF and FIT rates apply.

For RF components, the infant mortality phase is addressed through manufacturing screening. Burn-in subjects devices to elevated temperature (125-175°C) and electrical stress for 48-168 hours to precipitate latent defects before shipment. High-reliability programs (MIL-STD-883, JEDEC) mandate specific burn-in profiles. The useful life phase is characterized through accelerated life testing using the Arrhenius model, which relates failure rate to junction temperature through an activation energy that depends on the dominant degradation mechanism.

Reliability Equations

Constant failure rate (useful life):
R(t) = e^−λt
MTTF = 1/λ
λ in FIT = failures per 10⁹ device-hours

Arrhenius acceleration:
AF = exp[(E_a/k) × (1/T_use − 1/T_stress)]
E_a = activation energy (eV)
k = 8.617 × 10⁻⁵ eV/K

Weibull (wearout phase):
f(t) = (β/η)(t/η)^β−1 × e^−(t/η)β
β < 1: infant mortality
β = 1: random (exponential)
β > 1: wearout

RF Component Reliability Comparison

Technology	MTTF (hours)	E_a (eV)	Max T_j (°C)	Dominant Failure
Si CMOS	>10⁸	0.7	125	HCI, NBTI, EM
SiGe HBT	>10⁷	1.0-1.2	150	Electromigration
GaAs HBT	>10⁷	1.2-1.6	175	Contact degradation
GaN HEMT	>10⁶	1.5-2.0	225	Gate degradation
InP HBT	>10⁶	1.0-1.4	150	Contact/EM

Common Questions

Frequently Asked Questions

What are the three phases of the bathtub curve?

Infant mortality (decreasing λ): manufacturing defects precipitated by burn-in at 125-175°C for 48-168 hours. Useful life (constant λ): random failures at a FIT rate; MTBF = 1/λ. Wearout (increasing λ): cumulative degradation from electromigration, hot carrier injection, or oxide breakdown; modeled by Weibull distribution with shape parameter β > 1.

How are RF component reliability ratings determined?

Accelerated life testing at 200-350°C, extrapolated using Arrhenius: AF = exp[(E_a/k)(1/T_use − 1/T_stress)]. GaN gate degradation E_a = 1.5-2.0 eV; 1000 hours at 300°C equates to ~10⁷ hours at 150°C. Results: GaN HEMT MTTF >10⁶ hours at 200°C channel temperature; GaAs HBT MTTF >10⁷ hours at 150°C junction temperature.

What is the difference between MTBF and MTTF?

MTTF applies to non-repairable items (individual components); it is the average time to first failure. MTBF applies to repairable systems (base stations, radar); it is the average time between successive failures. System availability = MTBF / (MTBF + MTTR). A base station with MTBF 50,000 hours and 4-hour MTTR achieves 99.992% availability.

Related Terms

← Battery Satellite Bayes Detection →