Reliability Stress Testing With a Modified Daisy Chain Test Die

An integral part of the development of IC packaging technology is the ability to stress test the package at the earliest practical phase. Testing, however, inevitably requires a package assembled with a die. This is particularly necessary for flip-chip technology with organic substrates because of the multiplicity of organic compounds used and the consequential unique thermo-mechanical strains developed with die attached.
 
A common logistical problem is that, in order to be able to develop the device in a timely manner, package development must be accomplished simultaneously or, in many instances, prior to the active die development.

LSI Logic has taken the approach of designing a modified daisy chain (MDC) die that can be used for all phases of accelerated reliability stress testing. We have successfully been using MDCs for our development, qualification and reliability monitor program for the past six years.

A die test vehicle is needed that can be easily designed and built, and it must be able to:

• Provide full stress and testing capability of the package.
• Test all die-package interactive stress.
• Test all package metallurgical, mechanical and chemical stresses.
• Be tested on standard automated test equipment (ATE).
• Be easy to perform failure analysis.
• Be adaptable for second-level testing.

Package reliability testing includes the standard JEDEC tests, as described in JESD 47, and some custom testing: bias/humidity/temperature, high accelerated stress test (HAST), unbiased HAST, temperature cycling, high-temperature storage life, component temperature cycling, second-level temperature cycling and temperature shock. The MDC design must also be able to support acceleration of all the known failure mechanisms associated with a fully assembled package and detect any associated degradation mechanisms.

The purpose of the daisy chain chip is to ensure that packaging and assembly can be efficiently stressed and electrically tested. Failure sites must be easily identified. To accomplish this, the design must consider the following criteria:

• Allow opposite bias to the maximum pairs of adjacent traces.
• Have electrical contact to all the V_ss and V_dd planes.
• Provide traces that can detect die cracking.
• Monitor bond integrity and effect to underlying vias and metal.
• Stress effects to die redistribution layers for flip-chip.
• Monitor die-induced strain distribution effects.
• Monitor general thermo-mechanical stress effects.
• Monitor die attach and underfill adhesion problems.

1. The 313 BGA wire bond modified daisy chain (MDC) die.

The daisy chain die is designed to capture all of the thermo-mechanical failures attributed to accelerated stress testing. To this endeavor, we make sure that the MDC has the same metallurgy and passivation as the appropriate LSI active die. If necessary, we add the appropriate bond vias and redistribution layers. The assembly of the MDC die package is done in the exact manner as the operational chip with the same adhesives and at the same site.

The design may also contain loops that transverse each quadrant of the device and/or entire width of the device to detect die cracking. The die designs for 313 EPBGA wire-bonded package and the 1517 PBGA for flip-chip are typical.

Figures 1 and 2 show a bond-pad-design MDC with a two-layer metal to replicate the bond-pad via and ensure that the die undergoes all of the same wire-bonding stresses produced in die bonding. The die-crack monitoring traces are also shown.

2. Shown here is a magnified view of upper-left quadrant of the 313 BGA MDC.

The 313 daisy chain die is designed with two metal layers consisting of metal 2 (M2) and metal 3 (M3), representing one of our silicon metal architecture. The size is 9 × 9 mm. There are 256 bond pads with two pads chained in pairs, making a total of 128 chains.

The chains vary in length from a short chain loop that connects alternating pads for bias testing to the longest chain loop that crosses the center of the die. The long chains serve as die-crack detectors. These are strategically placed on the center and corner of the die covering all four quadrants.

Figure 3 shows a flip-chip MDC with 1517 balls and 512 daisy chains. The die size is 12.6 × 12.6 mm with a 10 mil pad pitch. All the chain loops have the same dimensions, and the die contains only one metal layer.

Accelerated stress environmental parameters with the MDC are exactly the same as those used with an active die.

3. Shown here is a cutaway view of the left quadrant of the 1517 flip-chip daisy chain die.

Because of the simple direct path of the circuit, the leakage or loop resistance can be continuously monitored during the stress cycle. Also, since the MDC has no current flowing, there is no Kelvin self-heating or self-drying, thereby ensuring that every part of the package/assembly is subjected to the intended humidity conditions. Because of this, the device is subjected to the same stress, relative humidity and temperature as unbiased HAST during the biased HAST test. This allows an evaluation of electrochemical vs. galvanic or direct chemical degradation, because both biased and unbiased HAST are tested at the same temperature.

For the above reasons, the MDC is a superior test vehicle for corrosion/open failures, especially for electrochemical corrosion.

It is also an effective structure for performing unbiased HAST or Autoclave.

Temperature cycling — The MDC design allows standard thermal cycling testing. All of the stress points are tested and can be isolated. In addition, the MDC design is conducive for performing second-level testing and/or continuous resistance monitoring during the temperature cycle test. Monitoring can be performed on all the pins.

The MDC has the ability to monitor:

• Die attach or underfill degradation.
• Die crack in multiple directions.
• Substrate degradation (traces and vias).
• Second-level degradation.

Another advantage is that, for wire-bonded devices, an available MDC die with a different bond footprint than that designed for a specific package can be used. Choosing only selective bond pads and altering wire bond directional angles accomplish this, allowing determination of the thermo-mechanical stress effects of different die sizes without designing a new die.

High-temperature storage life — The predominant failure mechanisms associated with high-temperature storage life stress test are metal diffusion and the formation of intermetallics. Again, since all of the intrametal interface joints are performed in the exact manner as in the live die, all of the interface metallurgy, like the underbump metallization (UBM)/bump and gold wire/aluminum pad, are effectively tested. Fault isolation is also easier to determine than with a live die.

The MDC allows testing to be easily performed with a wide range of test methodology ranging from a hand-held multimeter to a Kelvin meter to an automated resistance tester or with a full-blown ATE tester. Testing basically involves a measurement of resistance changes.

Parametric changes in the substrate or die connections can be monitored as a function of degradation over time rather than a threshold failure as is the case of an active die test. The MDC allows the option to do real-time degradation monitoring as performed in second-level TC testing. This will allow a more effective prediction of package field operating life.

The MDC greatly simplifies the failure analysis process. One of the main problems in using an active die for package reliability evaluation is determining if the failure is in the die or the package. This is particularly true for flip-chip, where the active surface is face down and embedded in an underfill epoxy. Because the MDC has a minimal amount of metallization layers, we are able to inspect the area between the die and substrate by using an infrared reflectance microscopy technique.¹

Wire bond open and short sites are easily identified by first determining the loop pin pairs that are failing, then exposing the bond wires by decapsulation or delidding (depending on package construction), severing the wires and performing resistance measurements between:

• The pin and connected bond wire.
• Pin to pin.
• Wire bond to wire bond wire segments still attached to the die.

The MDC vehicle is an excellent vehicle to evaluate the reliability of different package designs efficiently, with minimal cost and a minimal time requirement. It is routinely used to evaluate package development and process changes, as well as final qualification evaluation. We use the MDC vehicle for our reliability monitoring program and, because of the simple electrical circuitry, it can easily be electrically tested at various package assembly sites.