NASA Electronic Parts and Packaging (NEPP) Program Parametric Failures in COTS Capacitors Alexander Teverovsky*, Michael Sampson ** *ASRC AS&D, Inc. work performed for NASA GSFC Code 562 ** NEPP program Co-manager, NASA alexander.a.teverovsky@nasa.gov
List of Acronyms AF acceleration factor HTS high temperature storage BME base metal electrode NT new technology COTS commercial off the shelf PME precious metal electrode DCL direct current leakage PTC polymer tantalum capacitors ESR equivalent series resistance S&Q screening and qualification HALT highly accelerated life testing TTF Time to failure WTC wet tantalum capacitors 2
Outline General comments on insertion of COTS in hi-rel systems. Rating-related failures. Degradation-related failures. Catastrophic and parametric failures in tantalum capacitors and MLCCs. ESR in polymer tantalum capacitors. Degradation of leakage currents in tantalum capacitors. Conclusion. 3
Two Approaches for COTS Insertion 1. Reliability of COTS is inferior to MIL parts, and to qualify for space one need to run extensive testing per the existing requirements. The major concern is cost and time rather than technical issues. 2. COTS are NT devices and need analysis of new degradation processes and failure mechanisms. Existing procedures for S&Q have to be evaluated and adjusted. New mechanisms might require new testing techniques. o o o COTS as NT approach requires understanding of new degradation mechanisms, specific reliability issues, and development of adequate S&Q procedures. The consistency of COTS quality still remains a problem. HTS does not affect MnO2 caps, but causes degradation of PTC. Cracks in packages affect degradation of ESR in PTCs, but can be considered mostly as cosmetic defects for MnO2 caps. Weibull grading testing works with MnO2 caps, but does not with PTC. 4
ESR_avr, Ohm Rating-related Parametric Failures Failures during environmental testing might be due to the marketing pressure that forces manufacturers to squeeze performance of COTS components thus leaving insufficient margin between the rated and actual characteristics. Examples: Ripple currents. Temperature stability in wets. Leakage currents in wets. ESR in chip tantalum and polymer capacitors. 10 1 0.1 all MnO2 capacitors commercial trend LIMIT 0.01 0.01 0.1 1 10 ESR_limit, Ohm 5
current, A current, A Catastrophic and Parametric Failures in MnO2 Tantalum Capacitors Variations of leakage currents with time for two lots of 6.8 mf 35 V capacitors from the same Mfr. during 100hr HALT 1.E-03 6.8uF 35V Type M at 85C 70V 1.E-03 6.8uF 35V Type C at 85C 70V 1.E-04 1.E-04 1.E-05 1.E-05 1.E-06 100 1000 10000 100000 time, sec 1.E-06 100 1000 10000 100000 time, sec Type I: A sharp increase of DCL indicating breakdown. Type II: gradual increase of DCL resulting in parametric failures. Type I (catastrophic failures) is more often observed for MIL and Type II (parametric failures) for commercial capacitors. 6
current, A current, A Failures in BME and PME Capacitors 0.33mF 50V BME MLCCs with cracks 1.E-04 BME 0.33uF 50V 1210 with cracks at 125C 100V 0.33mF 50V PME MLCCs with cracks 1.E-04 CDR 1825 with cracks at125c 200V 1.E-05 1.E-05 1.E-06 1.E-06 Mfr.C Mfr.A 1.E-07 1.E+2 1.E+3 1.E+4 1.E+5 Cracking in BMEs does not affect IR measured at 125 C but facilitates degradation and parametric failures. Degradation in PMEs with cracks results often in instantaneous failures. Contrary to PMEs, degradation in BMEs occurs gradually and energy generated at the defect can be balanced by heat dissipation. Leaky BME capacitors often degrade, but do not fail catastrophically, whereas PME capacitors with prevailing avalanche-like breakdown might not degrade, but fail short circuit. time, sec 1.E-07 1.E-08 Mfr.C Mfr.V 1.E-09 1.E+2 1.E+3 1.E+4 1.E+5 time, sec 7
ESR, mohm c u m u la t iv e p r o b a b ilit y, % ESR Degradation in Polymer Ta Caps Approximations: ESR ESR exp t 1.E+5 1.E+4 1.E+3 1.E+2 1.E+1 0 E a 0.72 ev, which is close to results for 10 V capacitors. Simulations allow for the end-of-life predictions. More complex models might be necessary to accommodate for degradation inception times. exp E a kt Polymer Ta 10uF 25V at HTS 150hr 300hr 990hr = 5000hr 100C 125C 150C 175C 0 1000 2000 3000 4000 5000 time, hr 0 99 90 50 10 5 175C P( ) Polymer Ta 10uF 25V at HTS 150C 125C 100C 1 exp use level 55C 1 10 100 1000 10000 100000 1000000 tau, hr 8
A History Case Background Commercial 6.8 mf 25 V Ta caps that were screened and qualified to MIL-PRF-55365. Parts successfully tested in voltage regulator units. Operating conditions: 9 V at 45 C. Problem Per the project request, six capacitors have been subjected to life testing at 125 C and 16.7 V. Two parametric failures were observed. The risk of failure during the mission should be assessed. A model for leakage currents degradation should be developed to evaluate the probability of parametric failures. 9
current, A current, A rate, A/sec Degradation of Leakage Currents Linear approximation of I-t characteristics 1.E-05 6.8uF 25V at 125C 25V 1.E-02 6.8uF 35V at 85C 1.E-07 6.8uF 35V at 85C 8.E-06 1.E-03 1.36E-08 A/sec 1.E-08 6.E-06 1.E-04 1.4E-09 A/sec 1.E-09 4.E-06 2.E-06 0.E+00 0.E+0 2.E+4 4.E+4 6.E+4 8.E+4 1.E+5 1.E+5 time, sec 1.E-05 1.27E-10 A/sec 1.E-06 7.68E-12 A/sec 56V 63V 70V 77V 1.E-07 100 1000 10000 100000 time, sec 1.E-10 y = 2E-20e 12.415x 1.E-11 1.E-12 1.25 1.5 1.75 2 2.25 2.5 V/VR Linear approximation is applicable for initial stages of degradation. Degradation rate a = f(t, V). Degradation rate increases with voltage exponentially, B 12. 10
Technique, Cont d Capacitors: 6.8 mf 25 V and 6.8 mf 35 V. Monitored HALT: o Temperature: 85 ºC to 145 ºC in 20 ºC increments; o Voltage: 15 V to 35 V in 10 V increments; o step duration 30 hr. Degradation rate, a(t, V), was calculates for each sample. Distributions of a at different T and V were approximated with a general log-linear model: a 0 exp a 1 ( a) 1 exp ( T, V ) a exp T Acceleration constant B and the activation energy: B = a 2 VR, E a = -a 1 /k. F a Time to failure was calculated as 2 TTF I Acceleration factors for a AF AF V T crit a Vtest expb 1 VR E a 1 1 exp k T1 T2 I 0 V test 11
c u m u l a t i v e p r o b a b i l i t y, % c u m u la t iv e p r o b a b ilit y, % Distributions of Degradation Rates for 35V and 25V Capacitors 99 90 50 10 5 35 V capacitors at different temperatures and voltages Degradation rates for 6.8uF 35V capacitors wo 125/49 General Log-Linear Weibull 358 49 F=20 S=0 Stress Level Points Stress Level Line 358 56 F=19 S=0 Stress Level Points Stress Level Line 358 63 F=19 S=0 Stress Level Points Stress Level Line 358 70 F=19 S=0 Stress Level Points Stress Level Line 358 77 F=18 S=0 Stress Level Points Stress Level Line 398 42 F=20 S=0 Stress Level Points Stress Level Line 328 10 Use Level Line 99 90 50 10 5 25 V capacitors at 25 V and different temperatures Degradation rate, 6.8uF 25V at 25V 85C 105C 125C 145C 1 1.E-20 1.E-18 1.E-16 1.E-14 1.E-12 1.E-10 1.E-8 1.E-7 rate, A/sec 1 1.E-16 Unimodal distributions for 35 V capacitors. Bimodal distributions for 25 V capacitors. 1.E-15 1.E-14 1.E-13 1.E-12 1.E-11 1.E-10 1.E-9 rate, A/sec Slow- and fast-degrading subgroups in 25 V capacitors have been analyzed separately. 1.E-8 12
current@25v, 85C, A Parameters of the Model and Mechanism AF V Vtest expb 1 VR AF T exp E k a 1 T1 1 T 2 Variations of leakage currents caused by HALT and annealing 1.E-03 6.8uF 25V capacitors 6.8 mf 25V 6.8 mf 35V 1.E-04 Low-rate subgroup High-rate subgroup All data B 9.25 5.45 10.3 1.E-05 1.E-06 1.E-07 1.E-08 E a, ev 1.66 1.42 1.65 1.E-09 init HALT 125C 35V bake 20hr 175C HALT 145C 25V bake 10hr 175C bake 30hr 175C bake 60hr 175C Similar constants for the slow-degrading subgroup of 25 V and 35 V capacitors, B avr = 9.8 ±0.5 and E a_avr = 1.65 ev. E a = E migr + E leak. At E leak ~ 0.5 ev, E migr ~ 1.1 ev, which is typical for oxygen vacancies (E a, E migr and E leak are activation energies of the degradation rate, migration of V ++ o and of leakage current respectively) DCL degradation is reversible that is in agreement with the model. 13
c u m u la t iv e p r o b a b ilit y, % Distributions of Times to Failure 99 Simulation of TTFs for 6.8uF 25V capacitors Distributions of experimental and calculated times to failure at life test and use conditions 90 50 10 calculation 125C, 16.6V experimental data use conditions 55C, 10V 5 1 1.E+2 1.E+3 1.E+4 1.E+5 1.E+6 1.E+7 1.E+8 1.E+9 time, hr Calculated and experimental data are close thus validating the model. Model allows for conservative estimations. The slope of distributions, 2, indicates wear-out failures. The probability of failure at use conditions is negligibly small. 14
Conclusion Two reasons for parametric failures in COTS: (i) due to insufficient margin in specified parameters and (ii) due to degradation processes. Degradation is typically caused by wear-out processes and can be modeled relatively easily. Determining physical mechanisms of degradation is more challenging, but is important to justify the models. 15