Destructive and Non-Destructive Testing
Two tests are industry standards for establishing wire bond strength and the acceptability of part lots based on that bond strength. They are:
- MIL-STD-883, Method 2011.7 Bond Strength (Destructive Bond Pull Test)
- MIL-STD-883, Method 2023.5 Nondestructive Bond Pull
- Link to these Test Methods
The military standard for high reliability microcircuits, MIL-PRF-38535, "General Specification For Integrated Circuits (Microcircuits) Manufacturing", guides the manufacturers to perform Destructive Bond Pull on twenty-two bonds from a minimum of 4 units from each production lot and to accept that lot only if there are zero failures. It also suggests that the Non-destructive test be performed on all wires in units intended to be sold as “Class S” or space grade product.
The Destructive test (Method 2011.7) applies a grams-force pull to the bond wires installed in the package. This force is increased until the break is detected. Different instructions are given for different types of bonds: single bond, double bond, flip chip, and beam lead. A table is given in the test method, which specifies minimum bond strength required (pass/fail criteria). A graph is also provided for this purpose. The values are specific to the type of bond and the composition and diameter of the wire. The values range from 1.5 g-f for standard single or double wire bonds made out of 0.7 mil aluminum wire to 30 g-f for beam lead bonds. These values are for parts that have not been sealed yet. The requirement decreases by between 17% and 40% for parts that have been de-lidded to do the test. Failures must be recorded by how the bond broke. The categories for the different types of breaks are listed in the test method.
A special note exists at the end of this test method which allows RF/microwave devices which contain extremely flat loops (to reduce electrical interference) to use a different pull strength than shown in the table. The reduction in requirement is calculated by multiplying the value shown for normal wire bonds by the sine of the “greatest calculated wire loop angle”.
Harmon1/ treats this consideration as well with calculations for understanding how much force is extended to each side of a two sided bond for various positions of the pulling hook. For cases where the die and the substrate bonds are at equal height (? = 0 in Figure X below), or very flat, the pulling force is applied, by using a hook, approximately in the center of the wire in a direction normal to the die or substrate. Subsequently, this pulling results in an equal force on each end. If the bonds are not equal height, the bond pull shall be performed where the bend (loop) is formed between the two bonds. The proper wire pull value can be calculated by using the following equation:
V1=New pull value
V2=Pull strength from table in MIL-STD-883 for the size of wire tested
q=Greatest Calculated Wire Loop Angle
|| Flat wire bond approach from bond pad at package shelf (top, wide end) down to the bond pad on the die. |
The Non-destructive test is intended for “Class S” parts or parts intended for use by the high reliability space flight community. In this test, the bonds are pulled, though not to destruction. Since this test is used as a “100% screen” any failures are unacceptable. Based on the program requirements, rework and retest may be acceptable. The force of the pull is 80% of the pre-seal minimum bond strength shown in the 2011 test method. A table is given in the test method for commonly used wire sizes and their test pull forces. These range from 1.2 g-f for 0.7 mil aluminum wire to 12.0 g-f for 3 mil gold wire. Again, the reduced pull force for “flat” RF/microwave bonds applies.
The Non-destructive test is not recommended for a package having 84 or more external terminations and with nominal bonding wire pitch at the package post of less than or equal to12 mils. Alternative approaches are listed in this test method for looking at wire bond quality. These include review of manufacture quality records and raw material control, a thermal mechanical analysis of the package and the bonds over time and with temperature cycling, and a 100% visual inspection of all bonds. When the package has gold plated posts, and the alternate method must be used, a sample is required to be submitted to a bake procedure (45 bonds, no failures) where the unit, post bonding, is exposed to 300°C for 1 hour in air or an inert atmosphere and then the bonds are pulled to destruction. This regimen is intended to look for contaminated plating. We recommend here that the temperature be increased to 360°C if the part can tolerate it, and the atmosphere be inert and non-contaminating. An appendix of this test method gives detailed instructions about how to do a wire bonding capability study on the manufacturer’s process.