1. Environmental hysteresis in voltage reference PEMs Alexander Teverovsky QSS Group, Inc./Goddard Operations Alexander.A.Teverovsky.1@gsfc.nasa.gov

2. CMSE Feb' 032 Purpose and Outline Purpose: To discuss effects of environmental stresses (temperature and humidity) on characteristics of precision voltage reference microcircuits encapsulated in plastic packages. Outline:  Introduction. Effect of MS in linear devices  Experimental  Effect of external mechanical stresses  Effect of moisture sorption/desorption  Effect of high temperature exposure  Effect of low temperature exposure  Conclusions

3. CMSE Feb' 033 Mechanical stresses in PEMs Cracks in dies Illustration of significance of mechanical stresses in PEMs   Packaging related mechanical stresses are characteristic features of PEMs.   These stresses are caused by cure shrinkage and CTE mismatch of MC and die assembly materials.   Typical compressive stresses caused by packaging are 50 MPa to 100 MPa.    pac <<  st (strength of Si).  int cracking Parametric shifts Packaging stresses

4. CMSE Feb' 034 Effect of mechanical stresses on reliability Parametric instability:  Mechanical stresses caused by packaging are sufficient to cause significant shift in Vout of voltage reference and in Vos of opamp PEMs.  Temperature and humidity might change stresses with time due to creeping, swelling, and shrinkage of molding compounds.  Variations in mechanical stresses due to changes in environmental conditions might cause instability in precision linear devices. Stress-related failure mechanisms reported in literature: For nMOSFETs tensile stresses might increase hot-carrier- induced degradation. Mechanical stresses might affect radiation sensitivity of bipolar devices.

5. CMSE Feb' 035 Why mechanical stresses cause changes in characteristics of linear devices? The effect of MS is due to changes in the electronic band structure of Si, which causes changes in E g and . Both effects are anisotropic. Piezo-resistance effect (caused by  variations): –Si resistors: has different sighs for n- and p- types, and is larger for p-type resistors than for n-types (dR/R up to ~2 - 3% at  ~ 100 MPa); –MOSFETS: affects Idsat; is larger in n-type with longer channels(>1 um). Piezo-junction effect – the change in the saturation current of BJT (caused by E g and  variations): –results in VBE changes up to 4 mV at 200 MPa, which is especially important for band-gap voltage references; –The piezo-junction coefficient decreases with increasing temperature, is larger for NPN than for PNP transistors, and is larger for compressive stress than for tensile stress.

6. CMSE Feb' 036 What level of the output stability is necessary? Voltage references are the biggest source of error in AD/DA systems Data sheet information:   Temperature drift;   Line regulation;   Load regulation;   Long-term stability:   AD780 ± 20 ppm/1khr;   LT1461 ± 60 ppm/1khr. System resolution requirements:

7. CMSE Feb' 037 Environmental factors affecting stability   Temperature: 1 to 3 ppm/ o C   Temperature hysteresis (TH): 75 to 130 ppm? Soldering reflow: 150 to 300 ppm?   Relaxation of the Vout deviations caused by TH: ?   Effect of temperature on TH: ?   Radiation effects [JPL’97]: Co60, 100 krad(Si): ~ 150 ppm p-200 MeV, 45 krad(Si) ~ 500 ppm   Humid conditions: ?   Vacuum conditions: ? How significant are Vout variations caused by T and/or RH excursions? How fast are these variations relaxing? How significant are Vout variations caused by T and/or RH excursions? How fast are these variations relaxing?

8. CMSE Feb' 038 Experimental: Precision 2.5 V Vref PEMs Average characteristics of materials and standard deviations (in brackets) Part (DC) LF CTE, ppm/ o C MC TG, o C MC CTE1, ppm/ o C MC CTE2, ppm/ o C AD780AR (0106) 17.5 (Cu) 171 (4.4) 15.1 (2.3) 89 (12) AD780BR (0127) 17.5 (Cu) 173 (5) 16.4 (1) 77 (25) LT1461AI (0113) 4.3 (alloy 42) 138 (3) 9.8 (0.5) 65 (4.4)

9. CMSE Feb' 039 Effect of temperature AD780 Vin = 5V AD780 Vin = 5V, Iout = 10 mA Temperature hysteresis Output drift with temperature Temperature drift is partially caused by mechanical stresses. Temperature hysteresis is due to creep of MC. Temperature drift is partially caused by mechanical stresses. Temperature hysteresis is due to creep of MC. failures

10. CMSE Feb' 0310 Effect of mechanical stresses Compressive stress causes negative drift. Gage factor = 1 to 3 Compressive stress causes negative drift. Gage factor = 1 to 3 Parts glued to an epoxy block

11. CMSE Feb' 0311 Moisture induced hysteresis Package thickness 2h = 1.5 mm D 85 = 4  10 -8 cm 2 /s  85 = h 2 /D  39 hr Initial Vout shift ~150-250 ppm is caused by moisture induced swelling. Output relaxation is due to moisture outdiffusion. Initial Vout shift ~150-250 ppm is caused by moisture induced swelling. Output relaxation is due to moisture outdiffusion.  Soaking at 85 o C/85% RH for 168 hrs Output relaxation during storing at 85 o C.

12. CMSE Feb' 0312 Output voltage moisture sorption isotherm A negative shift of ~150 to 300 ppm can be expected in vacuum due to moisture outdiffusion. Test conditions: Storing at 85 o C for 168 hrs at 20%, 40%, 60%, 80%, and 90% RH Measurements at RT.

13. CMSE Feb' 0313 Effect of HAST testing Swelling in MCs at 85% RH   The rate of parametric shifts exponentially increases with T, whereas dVout does not vary significantly.   Sorbton/desorption at RT causes output drift similar to HAST.   The rate of parametric shifts exponentially increases with T, whereas dVout does not vary significantly.   Sorbton/desorption at RT causes output drift similar to HAST. Moisture pressure increases >100 times, however Vout and swelling variations are <50%. Test conditions: soaking at 85% RH for 72 hr to 168 hr at temperatures of 75, 85, 110, 130, and 150 o C

14. CMSE Feb' 0314 Reversibility of dVout during HAST testing Three parts failed load testing due to damage in output transistors. Swelling-induced deviations are mostly reversible during testing at T = 85 o C. However, HAST at T = 150 o C causes Vout changes, which do not restore after baking. LT1461 AD780

15. CMSE Feb' 0315 Swelling of MCs due to HAST Test: moisture uptake and swelling after HAST were measured on several types of PEMs. Conditions: saturation with moisture at 85% RH and different temperatures. Irreversible increase in volume causes irreversible deviations in Vout. It also might cause delaminations and mechanical failures. Irreversible increase in volume causes irreversible deviations in Vout. It also might cause delaminations and mechanical failures. dM/M dV/V

16. CMSE Feb' 0316 Coefficient of moisture expansion  = (  L/L)/(  m/M). Typically varies from 0.1 to 0.5. Mechanical stress in die:   A  E  [(  MC -  Si )  T +   m ] At   ~ (7 to 13) ppm/ o C, and  m ~ ± 0.25%, moisture- induced stress variations are equivalent to temperature variations of  T ~ ± (20 to 180) o C. (Positive sign in  m corresponds to moisture sorption and negative - to moisture desorption) Variation of mechanical stresses in PEMs due to moisture induced swelling in MCs

17. CMSE Feb' 0317 Effect of exposure to high temperatures Room temperature relaxation after exposure to 250 o C 15 min exposure to T from 150 to 250 o C Negative deviation of Vout at T < 200 o C. Deviations up to 250 ppm at T > 200 o C. No relaxation at RT. An increase in Vout is due to moisture uptake. Negative deviation of Vout at T < 200 o C. Deviations up to 250 ppm at T > 200 o C. No relaxation at RT. An increase in Vout is due to moisture uptake. History case: 11 out of 30 Vref PEMs failed parametrically SMT simulation

18. CMSE Feb' 0318 Effect of exposure to low temperatures Effect of temperature cycling from 20 o C to -65 o C Experiment conditions: exposure to low temperatures for 15 min. Output deviations saturate at T < -65 o C; Most of the output deviations occur after the first low temperature exposure. Output deviations saturate at T < -65 o C; Most of the output deviations occur after the first low temperature exposure. AD780 LT1461

19. CMSE Feb' 0319 Relaxation of LT hysteresis Some relaxation at 75 o C after 15 min exposure to –65 o C No relaxation at 55 o C after 15 min exposure to –65 o C Partial relaxation occurs during a few minutes at 75 o C. No relaxation was observed during 6000 hours at 55 o C. Partial relaxation occurs during a few minutes at 75 o C. No relaxation was observed during 6000 hours at 55 o C.

20. CMSE Feb' 0320 LT relaxation at room temperature No relaxation during ~ 6500 hours at ~ 22 o C. Complete relaxation (overshooting) during 1 hr at 85 o C. No relaxation during ~ 6500 hours at ~ 22 o C. Complete relaxation (overshooting) during 1 hr at 85 o C. Snap effect. ( 1 hour at 85 o C) 15 min exposure to -65 o C.

21. CMSE Feb' 0321 Effect of moisture on LT hysteresis Test conditions: 15 min – 65 o C after bake and after moisturizing at different conditions. Results: - LT hysteresis in humid cond. ~ 170 ppm - LT hysteresis in dry cond. ~ 265 ppm Mechanism of LT hysteresis: At LT, high tensile stresses in MC result in creep and in decrease of compressive stresses in the die. Upon returning to RT the die experiences less compressive stresses. Moisture plastisizes MC thus decreasing tensile stresses at LT. Stress relaxation at T < Tg might indicate interfacial effects. Mechanism of LT hysteresis: At LT, high tensile stresses in MC result in creep and in decrease of compressive stresses in the die. Upon returning to RT the die experiences less compressive stresses. Moisture plastisizes MC thus decreasing tensile stresses at LT. Stress relaxation at T < Tg might indicate interfacial effects.

22. CMSE Feb' 0322 Conclusion  Environmental stresses result in parametric shifts large enough to cause instability in systems with resolution of more than 10 bits.  Moisture sorption causes positive shift of Vout. Increase in T at 85% RH from 85 o C to 150 o C increases dVout from ~250 ppm to ~ 400 ppm. Kinetics of Vout variations in humid/dry conditions can be calculated based on moisture diffusion characteristics of MCs.  A negative shift of ~150 ppm to 300 ppm can be expected in vacuum due to moisture outdiffusion.  Low temperature exposure (LTE) results in positive output deviations up to 250 ppm, which saturate at T < -65 o C. These deviations occur mostly after the first cycle of LTE.  LTE specifics: Snap effect. No relaxation during ~ 6500 hrs at RT and complete relaxation during 1 hr at 85 o C. Moisture effect. Moisture significantly decreases LTE effect.