Type 1 ICDs; the causes of most Type 1 ICDs are not obvious. What to book for, and how to be certain of what you see.
[FIGURE 1 OMITTED]
Only unetched specimens should be used to inspect for ICDs. Only polished, unetched microsections should be used verify whether the defect is present. Proper procedure requires the sections be solder-floated (one to three times), then potted for inspection. If there is line of demarcation on the interconnect (again, without etching the specimen), it is reasonable to suspect an ICD. At this point, one can etch the specimen to bring out the grain structure and determine exactly which of the three types of ICDs is present. Etching the specimens prior to inspecting for ICDs will cause preferential etching of the electroless deposit due to the presence of palladium and the overall grain structure of the electroless copper. This is often referred to as pseudo smear.
Sometimes, a defect is seen such as the one shown in Figure 3. Is this a true ICD, caused by problems with the metalization system?
[FIGURE 3 OMITTED]
FIGURE 2 shows an etched microsection with the line of demarcation. However, this line was not present on the unetched specimen and is therefore not a Type 1 ICD. The line is due to preferential etching of the electroless copper deposit. There is no separation. There is another defect present, however: delamination.
[FIGURE 2 OMITTED]
One must be cognizant that not all defects seen at the interconnect are caused by plating issues or remaining drill smear. In FIGURE 3, close examination shows debris from drilling as the cause for the open circuit.
In Figure 3, debris is quite evident. In this case, the fabricator had two problems: 1) the vacuum on the drill machine had failed to remove drill chips, and 2) the company had switched to a poor-quality back-up material that was contaminating the holes. Fixing the vacuum and switching the back-up material to aluminum-clad composite solved the debris issues.
While the cause-and-effect is obvious in Figure 3, most Type 1 ICDs are not. Keep in mind that the weakest bond breaks first. Generally, Type 1 ICDs can be attributed to:
* "Invisible" drill smear not removed by the desmear process. Check parameters if using alkaline permanganate or plasma. Is the drilling process creating more smear than normal due to dull drill bits, faster in-feed rates, poor-quality entry and back-up material or drill-bit remains in hole for an excessive amount of time?
* Micro-etch in the electroless copper pre-plate line insufficiently "roughening" the copper surface to provide adequate anchoring sites for the copper deposit.
* Excessive dwell time in the catalyst or inadequate tin removal in the acceleration step. This leaves a "barrier" on the copper interconnect. Anything that weakens the bond between the copper deposit and interconnect can cause separation.
* Drag-in of cleaners into the various process tanks or poor rinsing due to insufficient time, rinse water quantity or quality.
* Electroless copper deposit exhibits inherent stress. This is usually caused by an excessively high deposition rate (the main cause of the stress is the amount of micro-inches of copper deposited per minute). The process should be operated within its designed parameters. Any change in the process that causes the electroless to be deposited at a higher-than-normal rate would be sufficient to cause Type 1 ICD.
A good rule to follow: if the electroless copper process is designed to deposit 40-50 microinches in 30 min. (as an example) and the actual deposit is somewhere in the 70-100 microinch range, then there is a problem. The copper is being deposited much too fast. This will "reorient" the grains of copper, leading to stress in the deposit. The only way to relieve the stress is for the deposit to pull away from the interconnect.
MICHAEL CARANO is vice president of sales and marketing at Electrochemicals (electrochemicals.com). He can be reached at email@example.com.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||The Plating Rack|
|Publication:||Printed Circuit Design & Manufacture|
|Date:||Aug 1, 2004|
|Previous Article:||Baking a pie: a review of a recent embedded passives conference shows the market smells good.|
|Next Article:||The history of embedded distributed capacitance; embedded distributed capacitance is much older than most think, with the first known U.S. patents...|