the significance of breakdown voltages for quality ... · 1.e+06 1.e+07 1.e+08 1.e+09 1.e+10 1.e+11...
TRANSCRIPT
The Significance of Breakdown Voltages for Quality Assurance of Low-Voltage BME Ceramic
Capacitors
NASA Electronic Parts and Packaging (NEPP) Program
Alexander TeverovskyAS&D Inc.
(work performed for GSFC Parts, Packaging, and Assembly Technologies Office, Code 562)
[email protected] (BME)
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
https://ntrs.nasa.gov/search.jsp?R=20140005413 2020-08-03T14:32:57+00:00Z
List of Acronyms and Symbols
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
2
AF acceleration factor Rd resistance associated with defectsBME base metal electrode Ri intrinsic resistance
C capacitance RSH resistance to soldering heatCVR constant voltage ramp S&Q screening and qualificationDCL direct current leakage T temperatureDF dissipation factor Tc Curie temperature
DWV dielectric withstanding voltage TS thermal shockHALT highly accelerated life testing TSD terminal solder dip
HV high voltage TTF time to failureIR insulation resistance VBR breakdown voltageLV low voltage VBR75 third quartile of VBR distribution
MF mechanical fracture VO++ charged oxygen vacancy
MLCC multilayer ceramic capacitor VR rated voltage
PME precious metal electrode X-sect cross-sectionr charge absorption resistance PDA percent defective allowable
Outline
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
3
What limits application of BME MLCCs in hi-relelectronics?
Drawbacks of the existing system of quality assurance for MLCCs.
Factors affecting VBR. Distributions of VBR and their
use for quality assurance. Modeling of life testing. Flash testing for screening. Recommendations. Summary.
Flashover
What Impedes Application of BME Capacitors in Hi-Rel Systems?
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
4
It is assumed that reliability is limited by high concentration of VO
++. The wear-out problem has been resolved
by using new materials and processes.(Life testing (1000 hr, 2VR, 125 °C) is equivalent to thousands of years at 65 °C and 0.5 VR).
Intrinsic wear-out failures caused by VO
++ do not limit applications. Failures are caused by defects.
1.E+02
1.E+03
1.E+04
1.E+05
1.E+06
1.E+07
1.E+08
2 4 6 8 10
time,
yea
rs
voltage acceleration constant, n
Time at 65C, 0.5VR that is equivalent to 1khr life testing
0.7eV 0.9eV1.1eV 1.3eV1.5eV 1.7eV1.9eV
Factors impeding applications: (1) lead-free termination finishing; (2) insufficient data on long-term reliability; (3) the presence of defects introduced either during manufacturing or assembly and handling processes.How effective the existing S&Q system for revealing defects in
MLCCs?
What is in the Toolbox for S&Q?
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
5
Specified characteristics: C, DF, and IR. DF and IR have only max limit (unknown margin). VR is assigned by manufacturers so that the parts
meet their performance and reliability requirements. DWV test assures that VBR exceeds 2.5VR. Environmental stresses during S&Q testing: HT/HV testing (voltage conditioning and life): Thermal testing (TS, RSH). Humidity testing; No measurements are required during the
environmental testing, and PDA has no limits.Improvements in the S&Q system can be achieved by: increasing the level of stress as necessary; setting PDA limits; monitoring DCL during environmental testing; using new, defect-sensitive, characteristics and tests.
C, DF, IR, DWV
∆T
C, DF, IR, DWV
The Effectiveness of IR Measurements
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
6
IR is presented by three resistors connected in parallel: absorption (r), intrinsic (Ri), and defects (Rd). Limits for military MLCCs: IR > 1011 Ω or 103 MΩ-µF at +25°C(×0.1 at +125°C).
1.E+06
1.E+07
1.E+08
1.E+09
1.E+10
1.E+11
1.E+12
1.E+06 1.E+07 1.E+08 1.E+09 1.E+10 1.E+11 1.E+12
IR_d
amag
ed, O
hm
IR_initial, Ohm2225 1uF 50V 1206 4.7uF 25V 1210 0.1uF 50V 1812 1uF 100V2220 1uF 50V 0805 3.3uF 6.3V 2220 10uF 50V 1210 0.1uF 50V1825 0.1uF 100V 1206 82nF 50V 1825 1uF 50V 2220 22uF 25V2223 22nF 50V 1206 0.47uF 16V 2220 100uF 6.3V 1825 0.47uF 50V
Room temperature
High absorption currents at RT, and large intrinsic leakage currents at HT mask the presence of defects.
The probability of revealing a defect decreases for capacitors with larger C and smaller VR.
1.E+05
1.E+06
1.E+07
1.E+08
1.E+09
1.E+10
1.E+11
1.E+05 1.E+06 1.E+07 1.E+08 1.E+09 1.E+10 1.E+11
IR_d
amag
ed, O
hm
IR_virgin, Ohm
M1206 22uF 6.3VC1206 4.7uF 25VM1206 10uF 16VA1206 0.47uF 25VV1825 0.47uF 50VA1825 0.47uF 50VC1825 0.47uF 50VA 1825 1uF 50VP 1825 1uF 50V
85C, 125C, 165C
[ ] [ ] 1111 ),(),( −−−− ++= di RVTRCtrIR
The Effectiveness of DWV Testing
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
7
27 lots of low-voltage MLCCs were tested in virgin condition, after mechanical fracture (MF) at the corner, after cross-sectioning (X-sect), and after thermal shock (TS).
The probability of failure during DWV testing for capacitors even with rough mechanical fractures was low.
The existing screening techniques (C, DF, IR, and DWV) are capable to determine only gross defects in the parts.
Gr.2 1812 Mfr.A 1uF 50V
breakdown voltage, V
cumu
lative
prob
abilit
y, %
60 2000100 10001
5
10
50
90
99
virgin
TS
X-sect
MF
DWV
Gr.6 2225 Mfr.C 1uF 50V
breakdown voltage, Vcu
mulat
ive pr
obab
ility,
%60 2000100 1000
1
5
10
50
90
99
virgin
TS
X-sect
MF
DWV
The Effectiveness of VBR Measurements
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
8
Local defects do not change C, DF and IR, but affect VBR.
If a capacitor passed DWV testing, should we care about VBR?(Designers’ and QA approaches)
defect
High values of VBR (similar to IR) provide assurance that the parts have no gross defects that might cause failures.Contrary to IR, VBR can be measured accurately.The consistency of the distributions indicates stability of the
manufacturing process and quality of the product.
VBR intrinsic VBR intrinsic
VBR defectVBRi < VBRd
Constant Voltage Ramp Technique
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
9
Factors affecting VBR: Ramp rate. Maximum current. Medium.
No significant variations in VBR at ramp rates from 25 V/s to 100 V/s.
Current limit in the range from 1 mA to 10 mA will not result in substantial variations of VBR.
CVR failures are due to thermal breakdown.
Testing in oil increases VBR by 10% to 50%.
400
420
440
460
480
500
520
540
0 50 100 150 200 250
VB
R, V
ramp rate, V/sec
MLCC 0805 1uF 10V
A B C
0.E+00
2.E-03
4.E-03
6.E-03
8.E-03
1.E-02
300 400 500 600
curr
ent,
A
voltage, V
MLCC 0805 1uF 10V CVR at 100 V/sec
Gr.BGr.AGr.C
MLCCs size 1825 0.47uF 50V
breakdown voltage, V
cumu
lative
prob
abilit
y, %
600 2100900 1200 1500 18001
5
10
50
99
BME_A, oil
BME_C, oil
BME A
BME_A, air
BME_C, air
PME_C, oil/airPME_V, oil/air
Distributions of VBR
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
10
In most cases distributions of VBR for BME capacitors are bimodal. The HV mode has tight distributions (STD/Mean ~4%) indicating
intrinsic breakdown. The presence of LV subgroup is due to defects.
The point of interception indicates proportion of defects in the lot.The spread of VBR towards low voltages indicates the
significance of these defects.
0.1uF 25V and 50V BME MLCCs
breakdown voltage, V
cumu
lative
prob
abilit
y, %
0 2500500 1000 1500 20001
5
10
50
99
0805 15um 25V M
0805 13um 25V A
0603 8um 50V C
1210 26um 50V C
BME capacitors with bimodal distributions
breakdown voltage, Vcu
mulat
ive pr
obab
ility,
%200 2200600 1000 1400 18001
5
10
50
99
1812 1uF 50V
0805 0.1uF 25V
0805 0.12uF 50V
1210 0.1uF 50V12 um, floating electrode
15 um
13 um
14 um
Distributions of VBR, Cont’d
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
11
Comparison of distributions for the same type of BME capacitors.
Analysis of distributions allows for assessment of the quality of capacitors and revealing improvements.Automotive industry and general purpose parts might be from
the same population and have similar defects.
MLCC size 0805 1uF 10V Mfr.A
breakdown voltage, V
cumu
lative
prob
abilit
y, %
100 600200 300 400 5001
5
10
50
99
2001, 6um
2013, 5um
BME MLCCs 1210 10uF 25V
breakdown voltage, V
cumu
lative
prob
abilit
y, %
250 500300 350 400 4500.1
0.51.0
5.010.0
50.0
99.9
0.1
Mfr.M, 5um
Mfr.T general, 4.5umMfr.T auto, 4.5um
Analysis of VBR Distributions for QA
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
12
Out of 27 types of BME capacitors 67% had bimodal distributions.
The proportion of capacitors with defects was from 2.6% to 80%.
The significance of the defects (the ratio of VBRmin and VBR75) varied from 0.3 to 0.95.
Assuming that acceptable defects would reduce breakdown voltage less than 2 times, 5 out of 27 lots should be rejected.
Analysis of distributions of VBR allows for the assessment of the quality of the lot.The lot is acceptable for space applications if VBR/VBR75 > 0.5.
BME_A 1825 1uF 50V
breakdown voltage, V
cumu
lative
prob
abilit
y, %
800 18001000 1200 1400 16001
5
10
50
99
VBR75
VBRmin
Rated and Breakdown Voltages
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
13
VBR exceeds VR more than 15 times.
Breakdown voltages on average increased with the rated voltage.
The spread of the data was large: at VR = 50 V, VBR varied from 17VR to 45VR.
Breakdown voltage is not a limiting factor in assigning VR to low-voltage capacitors.
In class II dielectrics VR is limited by the non-linearity of polarization that results in decrease of capacitance with voltage.
0
500
1000
1500
2000
2500
0 20 40 60
brea
kdow
n vo
ltage
, V
rated voltage, V
Low-voltage BME MLCCs
Effect of Soldering Stresses
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
14
3 cycles of TSD testing at Tsolder = 350 °C. Case size 1825, 0.47 µF, 50 V capacitors
were tested after manual soldering with a soldering iron set to 350 °C.
Stresses related to manual soldering can degrade VBR.VBR should be used to assess the
possibility of manual soldering.
BME 0805 0.12uF 50V Mfr.A
breakdown voltage, V
cumu
lative
prob
abilit
y, %
500 1500700 900 1100 13001
5
10
50
99
initialafter TSD_350
BME 0805 0.47uF 16V Mfr.Pa
breakdown voltage, V
cumu
lative
prob
abilit
y, %
500 800560 620 680 7401
5
10
50
99
initial
after TSD_350BME 1825 0.47uF 50V, Mfr.A
breakdown voltage, V
cumu
lative
prob
abilit
y, %
0 2500500 1000 1500 20001
5
10
50
99
manual soldering
TSD_350Cvirgin
intrinsicbreakdown
Simulation of Life Testing (2VR, 125°C)
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
15
Assumptions: intrinsic degradation for a defect-free part results in failures at TTF0 and the voltage AF follows Prokopowicz-Vaskas equation with n ~ 3 for PME and n from 4 to 9 for BME capacitors.
Similar defects are more likely to cause life test failures in BME compared to PME capacitors.
The greater the voltage acceleration constant n and the lowerVBR/VBR75 , the more probable infant mortality failures are.
n
i
dVBR
VBRTTFTTF
×= 0
VBR intrinsic VBR
defectV0+
+
TTFs were calculated based on VBR assuming for PME TTF0=50 khr, and for BME TTF0=10 khr.
BME
0805 0.1uF 25V, life test simulation
time, hrcu
mulat
ive pr
obab
ility,
%1.E-1 1.E+51 10 100 1000 100001
5
10
50
90
99
BME, n = 9
BME, n = 3PME, n = 3
β=0.3
β=0.9
PMEBME
β=3 to 9
2.E-09
2.E-08
2.E-07
1 10 100
curre
nt, A
time, hr
MLCCs 1210, 0.1uF, 50V
PME (7 damaged and 3 virgin)
BME (3 virgin)
BME (7 damaged)
BME_A 1825 1uF 50V
breakdown voltage, Vcu
mulat
ive p
roba
bility
, %0 2000400 800 1200 1600
1
5
10
50
99
initialafter HALT(160hr, 175C, 100V)
Effect of Defects in MLCCs on Reliability
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
16
BME and PME MLCCs with introduced defects were stressed by HALT.
Defects resulting in DCL degradation of BME capacitors might not affect PME capacitors.VBR for BME capacitors degrades after HALT.The effect is more significant for the LV subgroup of MLCCs.
Flash Testing
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
17
DWV testing can reveal only gross defects in low-voltage BME capacitors.
Testing at higher voltages would be more effective.
Time of exposure to HV should be reduced.
The third quartile is close to the mean value of VBR for the HV subgroup of capacitors.
Flash testing should be carried out at the same conditions as VBR measurements, except for it does not require oil as the medium and maximum applied voltage is limited to 0.5×VBR75_air.
BME 0805 0.1uF 25V
breakdown voltage, Vcu
mulat
ive pr
obab
ility,
%
0 1500300 600 900 12001
5
10
50
99
DWV Flash
75%
Flash
VBR_75%
Can Exposure to HV Cause Damage?
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
18
Different types of BME MLCCs were characterized before and after stressing with 1 to 5 sec pulses at V ~ 0.7VBR
No effect on leakage current and VBR. Remnant polarization can decrease C; however, this effect
decreases with time of storage and disappears after annealing at T > Tc (~ 120 °C).
No risks associated with HV exposure.
20
30
40
50
60
70
80
0 20 40 60 80 100
leak
age
curr
ent,
uA
cycle
BME Mfr.T 0603 0.47uF 10V cycling at 300V (VBR = 422V)
SN10 SN11
SN12 SN13
SN14 300V
300
600
900
1200
300 600 900 1200
VB
R_p
ost_
cycl
ing,
V
VBR_init, V
Effect of high-voltage cycles
Mfr.T 0603 0.47uF 10V after 100c 300VMfr.M 0.01uF 25V 0603 after 100c 500VMfr.M 0.1uF 25V 0805 after 100c 400V
0.5
0.6
0.7
0.8
0.9
1
1.1
init 1 min af ter HV
1.5hr af ter HV
70hr af ter HV
af ter 1hr at 175C
C/C
_bak
e
0805 0.1uF 25V Mfr.M 0603 0.01uF 25V Mfr.M0603 0.1uF 50V Mfr.C 0805 0.1uF 25V Mfr.A0805 1uF 10V Mfr.M
Recommendations
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
19
A lot should be characterized by a distribution of VBR based on 50 samples tested in oil (and in air for gr. C qualification testing) at maximum current of 10 mA and voltage ramp of 25 V/sec for C ≥ 1 µF and 100 V/sec for C < 1 µF. The values of VBR75_oil, VBR75_air, and VBRmin allow for the assessment of the consistency in quality, as a baseline to determine conditions for the flash testing, and to evaluate results of the RSH test.
Lots with VBRmin< 0.5×VBR75_oil have a higher risk of failures and should not be accepted for space applications.
Flash testing should be used as a screening procedure at the same conditions as VBR measurements, except for it does not require oil as the medium and maximum applied voltage is limited to 0.5×VBR75_air.
VBR measurements should be used as a lot acceptance procedure after RSH testing using 30 samples minimum. The acceptance criterion: VBRmin > 0.5×VBR75_oil or VBRmin > 0.5×VBR75_air for the parts soldered onto a test board.
Summary
Deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov originally presented by Alexander Teverovsky at the Capacitors and Resistors Technology Symposium (CARTS) conference, Santa Clara, California, April 1-3, 2014.
20
Reliability of BME MLCCs is limited by defects. The existing S&Q system can reveal only gross
defects. Analysis of VBR distributions allows for estimation
of the proportion and severity of the defects. Recommendations for using VBR measurements
for S&Q are suggested.