An Overview of Electronic Component Reliability Michael Choi

1
An Overview of Electronic Component Reliability

• Michael Choi
• Cornell B.S. E.E. 99
• Cornell M.Eng E.E. 00

2
Abstract

• In order to release an electronic component into
  the market, a series of evaluations exerting
  electrical and environmental stress must be
  performed in order to prove out component
  reliability. To carry out these evaluations
  without gating product release, elevated stress
  conditions are used in shorter durations and
  compared against lifetime usage of the component.

3
Outline

• Quality vs Reliability
• Bathtub Curve
• Common Failure Mechanisms
• Burn-in
• Survey of Common Reliability Evaluations
• Usage Models
• Summary

4
Quality vs Reliability

• Quality is a measure of overall product health at
  the time of release (i.e. percentage of products
  that are functional at shipment)
• Reliability is a measure of overall product
  health over the expected lifetime of the product.

5
(No Transcript)
6
The Bathtub Curve

• Infant Mortality
• Decreasing failure rate due to early life fails
  that were not caught at production test.
• Normal Life
• Constant failure rate since early life fails have
  been weeded out. These are random failures.
• Wearout
• This is the end of life for electronic
  components. The failure rate increases because
  operation is beyond what the component was
  designed to do.

7
Common Failure Mechanisms

• Silicon
• Gate oxide pinhole
• Metal-to-metal short
• Package
• Cu-line migration
• Delamination
• Silicon-Package interaction
• Die cracking

8
Burn-in

• What is burn-in?
• Burn-in is a method to stress silicon to emulate
  the type of stress it would undergo in its normal
  life in a shorter span.
• Why do burn-in?
• For any product, an assessment of the reliability
  over the intended lifetime of product is
  necessary to protect the customer.
• How is burn-in done?
• Utilizing the built-in logic test functionality,
  the silicon operates at a higher voltage and
  temperature than normal to induce a faster
  degradation

9
Temperature Acceleration

• One common way to understand how time-to-fail
  varies with temperature is the empirically based
  Arrhenius equation. It takes the form
• The acceleration factor between operation at two
  different temperatures is

10
Voltage Acceleration

• Voltage acceleration can also follow an
  exponential dependency
• Typically, voltage acceleration is much more
  effective at stressing silicon devices than
  temperature acceleration.

11
Temperature/Voltage Stress Use

• Temperature and voltage stress accelerate
  different failure mechanisms in varying ways
  depending on the process technology. Each
  technology will be affected differently depending
  on
• Susceptibility to different types of failure
  mechanisms
• Design rules for the process technology

12
Burn-in Equipment

• Burn-in is one of the main tools for assessing
  product reliability. It requires equipment to
  run the evaluation
• Burn-in socket holds the product and connects it
  to the burn-in board.
• Burn-in board board which holds an array of
  sockets and routes signals to each socket.
• Burn-in oven provides the vector patterns to
  stress logic in the device and regulates
  temperature.

13
Reliability Evaluation

• With the increasing complexity of devices, there
  are an increasing number of failure mechanisms
  and evaluations to assess them. Here are some of
  the common ones
• Silicon
• Life stress test (burn-in)
• Electrostatic Discharge (ESD)
• Package
• Temperature Cycling
• Highly Accelerated Stress Test (HAST)

14
ESD

• Electrostatic discharge occurs when charge flows
  from one place to another and an equilibrium is
  reached.
• Two objects with different potentials come into
  contact and current flows between them.
• This potential difference can be in the 1000s of
  volts and cause amps of current to flow in
  nanoseconds.
• Unprotected CMOS devices will be blown-out with
  less than 100V of potential difference.
• Thus, electronic components require ESD circuit
  protection to handle these potentials on the
  input/output pins.

15
ESD continued

• Sources of ESD can come from anywhere
• Human touch
• Manufacturing process
• Carpet
• Thus, different ESD tests are done to cover
  different sources
• Human body model (HBM)
• Machine model (MM)
• Charged device model (CDM)

16
Temperature Cycling

• Temperature cycling swings a packaged component
  from sub-zero to high temperatures to stress the
  package.
• The temperature cycling evaluation typically
  takes a sample of functionally good components
  and cycles them. These parts are then tested
  afterwards to check for functionality.
• Typical failures
• Package delamination
• Die cracking

17
HAST

• HAST exposes a packaged component to thermal and
  humidity stress.
• The components sit in a chamber which creates
  conditions of high temperature and a high
  relative humidity.
• Sometimes a biased HAST is done to exert voltage
  stress on the package metal traces.
• Typical failure modes
• Delamination
• Die-cracking
• Metal line migration (biased)

18
Usage Models

• Many evaluations exist for different applications
  and different types.
• Each of the aforementioned evaluations (ESD,
  HAST, Temperature Cycling) has different versions
  to fit different usage models.
• Therefore, for a given electronic component, a
  usage model must be developed to figure out what
  stresses should be used to emulate a true life
  test.

19
Usage Models continued

• Usage models vary greatly from component to
  component. Here are some cases
• Desktop computer
• Not mobile, operates inside.
• Notebook computer
• Mobile, operates in a variety of temperature,
  humid conditions. Might get dropped.
• Cellphone
• Mobile, operates in a variety of temperature,
  humid conditions. Will get dropped.
• In addition, each of these devices has a
  different expected usage life.
• Thus, understanding the usage of the electronic
  component is essential to ensuring proper
  reliability for the consumer.

20
Summary

• Quality and reliability are major concerns for
  any electronic component used today.
• Through reliability stress, we can understand
  what makes our products weak and address those
  issues before they affect customers.
• In order to properly stress electronic
  components, we must first understand how they are
  used.