3.1 Purpose of Reliability Testing

Reliability testing is for maintaining and improving the quality as well as for verifying whether a semiconductor device withstands the stresses in a customer’s environment, from manufacturing to transportation, or the stresses in customer’s equipment once marketed.

• Descriptions of Reliability Testing

There are three main types of reliability testing: tests in a design and development phase, a pre-production phase, and a mass production onset phase. These tests are conducted to confirm that no problems are found in each phase.

• Methods of Reliability Testing

Reliability testing should be performed under various methods that are standardized by the following: the Japanese Industrial Standards (JIS), the Japan Electronics and Information Technology Industries Association (JEITA), the United States Military Specifications and Standards (MIL), and the International Electrotechnical Commission (IEC).
We conduct the reliability testing complied with the JEITA standards.

3.2 Methods of Reliability Testing

1) New Process Development Phase

With advances in technology, integrated circuit (IC) products have recently been tending to become more miniaturized and sophisticated. This is why narrowing down the problems in all internal elements only by evaluating an actual product is considered inefficient and uneconomical. Therefore, such IC products will undergo the following tests: commonly known wafer-level life tests (HCI, TDDB, BTI, EM, SM) to ensure their expected life periods; individual element-level quality tests on the components essential to an IC, such as active elements (transistors), passive elements (diodes, capacitors, resistors), and traces to evaluate adequately and grasp the quality levels. In particular, during a new process development focusing on miniaturization, a dedicated TEG (Test Element Group) for reliability testing is used to verify if there is any fundamental product reliability issue.

2) Individual Product Development Phases

Each product development phase includes the two principal product prototyping runs.

• Post-design prototyping: To verify whether the characteristics and reliability defined in the planning phase are met
• Pre-production: To verify the points listed below through prototyping using a mass production line:
  

  － Confirm if the measures taken against the problems found in a post-design prototyping run are effective
  － Confirm if there is any potential problem in mass production

We basically perform reliability evaluations at each prototyping run.

3) Periodic Reliability Evaluation

We implement periodic reliability evaluations on our products to check if their reliability levels, which were defined in development phases, still maintain even after mass production. For such evaluations, we usually pick up a sample from those which are already mass produced. With considerations in the combination of several factors (e.g., wafer process, assembly process, and production site), we decide which product type to be picked up and evaluated.

3.3 Standards Related to Reliability Testing

3.4 Derating and Acceleration Models

The reliability of semiconductor devices greatly varies depending on derating factors applied even when they are used within the absolute maximum ratings or operating ranges individually specified. Therefore, when you design your equipment, be sure to perform appropriate derating in which adequate safety can be ensured. We conduct our reliability verifications under the operating environments that are generally assumed as per the quality grades defined in JEITA EDR-4708B/EDR-4711A.

The following subsections describe the three derating methods commonly employed to estimate market life periods from reliability test results: the temperature acceleration model, temperature difference acceleration model, and humidity acceleration model.
Electrical stress factors can be applied to self-heating temperature acceleration and temperature difference acceleration models. However, if your equipment requires abrupt voltage or current application, or has other potential stress factors, please contact us.

Temperature Acceleration Model (Arrhenius Model)

The Arrhenius model, proposed by the Swedish scientist Arrhenius, is a model to predict a chemical reaction speed at a given temperature. This is the most often used model in estimating the life periods of semiconductor devices.

Where:
L = Life
A = Constant value
Ea = Activation energy (eV)
k = Boltzmann constant of 8.6173×10-5 (eV/K)
T = Absolute temperature (K)

Let L1 be the life at the derating temperature T1, L2 be the implementation time at the reliability test temperature T2, then obtain the acceleration factor α by:

Below is an example of the temperature derating curve calculated from our own experimental values based on the definitions above.

Example Life of Intermetallic Compound Formation (Kirkendall Void) between Au-wire and Al-electrode
(Derived from Experimental Results of High-temperature Storage at 150 °C, 160 °C, and 175 °C)

Temperature Acceleration Model (Eyring Model)

The temperature difference acceleration model uses the Eyring model, proposed by the American theoretical chemist Eyring, to obtain the temperature difference repeatedly applied and the number of life cycles, as expressed in the equation below:

Where:
L = Life
A = Constant value
ΔT = Temperature difference
n = Temperature difference factor

Let L1 be the number of life cycles with the derating temperature difference ΔT1, L2 be the number of implementation cycles with the reliability test temperature difference ΔT2, then obtain the temperature difference acceleration factor αΔT by:

Below is an example of the temperature difference derating curve calculated from our own experimental values based on the definitions above.

Example Life of Chip Junction (Solder Deterioration)
(Derived from Experimental Results of Power Cycles at Δ70 °C, Δ90 °C, and Δ100 °C)

Humidity Acceleration Models

There have been a variety of proposed humidity acceleration models, but in this section, we explain the absolute vapor pressure model and the relative humidity model, respectively.

● Absolute Vapor Pressure Model / Vapor Acceleration Factor

This model correlates the time L, the period until a certain cumulative failure rate, with the vapor pressure VP, as expressed by:

Where:
A = Constant value
n = 2 (reference value)

From the equation above, the vapor pressure acceleration factor αVP can be defined as:

Where:
VP1 = Vapor pressure in the derated state
L1 = Life in the derated state
VP2 = Vapor pressure in the reliability test state
L2 = Life in the reliability test state

The saturated vapor pressure e (T) at a given temperature T can be obtained approximately from the Tetens equation:

● Relative Humidity Model / Humidity Acceleration Factor

This model correlates the time L, the period until a certain cumulative failure rate, with the relative humidity RH (%) and the temperature T (°C), as expressed by:

Where:
A = Constant value
Ea = Activation energy (eV)
k = Boltzmann constant of 8.6173×10-5 (eV/K)

From the equation above, the humidity acceleration factor αH can be defined as:

Where:
RH1 = Relative humidity in the normal state
L1 = Life in the normal state
RH2 = Relative humidity in the acceleration state
L2 = Life in the acceleration state

Below is an example of the correlation curves which were derived from the absolute vapor pressure and relative humidity models applied to our pressure cooker test (121 °C, 100%, 100 h), based on the definitions above.