Quality and Reliability of Quality and Reliability of

Semiconductor DevicesSemiconductor Devices

Daniel R. LoughmillerDaniel R. Loughmiller

Product Engineering ManagerProduct Engineering Manager--

Network and Communications GroupNetwork and Communications Group

Micron Technology, Inc.Micron Technology, Inc.

What are Quality and Reliability and What are Quality and Reliability and

how are they measured?how are they measured?

Quality is typically defined as the “time zero” Quality is typically defined as the “time zero”

failure rate and is measured in DPM (Defects failure rate and is measured in DPM (Defects

Per Million) or PPM (Parts Per Million).Per Million) or PPM (Parts Per Million).

Reliability typically refers to the “intrinsic Reliability typically refers to the “intrinsic

failure rate” over time and is measured in failure rate” over time and is measured in

FITS (Failures In Time per billion device FITS (Failures In Time per billion device

hours).hours).

The The BathTub BathTub CurveCurve

BathTub Curve

Log Time

Failure Rate

What factors contribute to

What factors contribute to

semiconductor device failures?semiconductor device failures?

Packaging related failures:Packaging related failures:

Thermal/mechanical stress.Thermal/mechanical stress.

Moisture induced corrosion.Moisture induced corrosion.

Alpha radiation.Alpha radiation.

Silicon related failures:Silicon related failures:

Sensitive to thermal and voltage stress.Sensitive to thermal and voltage stress.

Particle contamination.Particle contamination.

Silicon lattice defects.Silicon lattice defects.

Thin film (oxide) integrity.Thin film (oxide) integrity.

Examples of Package Related

Examples of Package Related

Failures Failures –– Lifted Ball BondsLifted Ball Bonds

Examples of Package Related

Examples of Package Related

Failures Failures –– Lifted Ball BondsLifted Ball Bonds

Example of Silicon Related

Example of Silicon Related

FailureFailure--Particle ContaminationParticle Contamination

Example of Silicon Related

Example of Silicon Related

FailureFailure--Silicon Crystal DefectSilicon Crystal Defect

How do semiconductor How do semiconductor

manufacturers provide quality manufacturers provide quality

and reliability?and reliability?

The testing of semiconductor devices is The testing of semiconductor devices is

aimed at detecting defective units before aimed at detecting defective units before

shipment to customer.shipment to customer.

BURNBURN--IN (high voltage/high temperature IN (high voltage/high temperature

stress) is used to accelerate the failure of stress) is used to accelerate the failure of

defective components.defective components.

The effectiveness of these processes is The effectiveness of these processes is

monitored through accelerated testing.monitored through accelerated testing.

How are reliability failure rates

How are reliability failure rates

calculated?calculated?

It is very impractical to determine failure rates It is very impractical to determine failure rates

using “real time” data.using “real time” data.

Predicted failure rates must be validated before Predicted failure rates must be validated before

product release which would impact time to product release which would impact time to

market.market.

Semiconductor devices are subjected to Semiconductor devices are subjected to

“accelerated” stress conditions.“accelerated” stress conditions.

Statistical models are applied to accelerated data Statistical models are applied to accelerated data

to calculate predicted failure rates.to calculate predicted failure rates.

FIT Rate ComputationFIT Rate Computation

Where:Where:

FIT Rate = FIT Rate = Failures per billion device hours

Pn Pn = = Poisson Statistic at 60% confidence (can be referenced in

statistical tables based on sample size/failure rate).

Device Hours = Device Hours = Sample size X test hours at accelerated

conditions. Typically must have over a million device hours for

statistical significance.

AF = AF = Acceleration factor between stress environment and typical

operating conditions.

910E

AF

s

DeviceHour

Pn

FITRate

Voltage Acceleration FactorVoltage Acceleration Factor

Where:

V

s

=Accelerated Stress Voltage

V

o

=Typical Operating Voltage

= Experimentally determined constant based on dielectric

integrity data.

For a device that typically operates at 2.7V and is stressed at

3.7V: AFAF

VV

= 20.09= 20.09

)(

os

VV

V

eAF

Thermal Acceleration FactorThermal Acceleration Factor

Where:

T

s

=Accelerated Stress Temperature in Kelvins

T

o

=Typical Operating Temperature in Kelvins

E

a

= Activation Energy in eV. Activation energy for typical oxide

defects has been experimentally determined to be 0.3eV.

k = Boltzmann’s Const (8.617E-5 eV/

o

K)

For a device which typically operates at 50’C and is stressed at

125’C: AFAF

T T

= 7.623= 7.623

)

11

(

so

a

TTk

E

T

eAF

Overall Acceleration FactorOverall Acceleration Factor

To determine the overall acceleration factor To determine the overall acceleration factor

used in the FIT Rate calculation, the temperature used in the FIT Rate calculation, the temperature

and voltage stress factors are multiplied and voltage stress factors are multiplied

together.together.

AFAF

TT

= AF= AF

VV

X AFX AF

TT

= 20.09 X 7.623 = 153= 20.09 X 7.623 = 153

FIT Rate Calculation ExampleFIT Rate Calculation Example

A device typically operates at 2.7V and 50’C. 18,386

devices are stressed for 168 hours at 3.7V and 125’C.

There are 10 failures. What is the FIT rate?

Pn = Based on the number of failures in the sample size,

Pn is found to be 11.51 at 60% confidence.

Device Hours = 18,386 X 168 = 3.09 X 10

6

AF=153

FIT Rate=11.51(10FIT Rate=11.51(10

99

)/(3.09 X 10)/(3.09 X 10

66

)(153) = 24 FITS)(153) = 24 FITS

Or one could expect 24 failures every billion hours.

What would the MTBF be???

Thin Film (Dielectric) IntegrityThin Film (Dielectric) Integrity

TDDB (Time Dependent Dielectric Breakdown) is a TDDB (Time Dependent Dielectric Breakdown) is a

principal failure mechanism in semiconductor devices.principal failure mechanism in semiconductor devices.

1.1.Defect related breakdown (caught at BURNDefect related breakdown (caught at BURN--IN).IN).

Asperities at Asperities at SiSi--SiO2 interfaceSiO2 interface

Crystal defects/pinholes.Crystal defects/pinholes.

2.2.Intrinsic breakdown (Intrinsic breakdown (wearoutwearout).).

Hot electrons/charge trapping causes eventual Hot electrons/charge trapping causes eventual

rupture of dielectric.rupture of dielectric.

How are these modeled?How are these modeled?

TDDB failures exhibit a lognormal distribution that can TDDB failures exhibit a lognormal distribution that can

be approximated by:be approximated by:

)

ln

(

2

1

2

1

)(

t

e

t

tf

Where:Where:

sigma = standard deviation of the samplesigma = standard deviation of the sample

mumu = mean of the distribution= mean of the distribution

TDDB is a strong function of electric field stress and a TDDB is a strong function of electric field stress and a

weak function of thermal stress with an Activation weak function of thermal stress with an Activation

Energy of 0.3eV.Energy of 0.3eV.

Other accelerated tests typically Other accelerated tests typically

performed to measure device performed to measure device

reliability:reliability:

THB (Temp/Humidity/Bias) THB (Temp/Humidity/Bias) –– 85’C/85% RH/3.6V85’C/85% RH/3.6V

Autoclave Autoclave –– 121’C/100% RH/3.6V121’C/100% RH/3.6V

Temp Cycle Temp Cycle -- --40’C to +85C/1008 cycles40’C to +85C/1008 cycles

Thermal Shock Thermal Shock -- --55’C to + 125’C55’C to + 125’C

High Temp Storage High Temp Storage -- +150’C for 1008 hours+150’C for 1008 hours

ESD ESD –– Based on MIL STD 3015Based on MIL STD 3015

Latchup Latchup

Electromigration Electromigration StudiesStudies

Questions?Questions?

## Comments 0

Log in to post a comment