Tara Tubbs & Philip Procter
September 1994
Abstract
Adhesion of the mold compound to its substrates has received increasing attention over the past few years. In these studies, the relationship between delamination induced by moisture exposure, as measured by Scanning Acoustic Microscopy (SAM), and device performance, as measured by: corrosion, ball lift and moisture absorption, was assessed. With the Scanning Acoustic Microscopy techniques, we are now able to measure the amount of separation between the mold compound and the die and leadframe surfaces. This development has in turn generated an immense quantity of work on the relationships between compound adhesion and properties such as popcorn resistance, corrosion resistance, etc. (Section 1.2). In many cases, however, secondary modes of failure are present which are overlooked in an attempt to assign a direct cause and effect relationship between two variables (Section 1). This is of particular concern today due to the increasing standards being developed based upon acoustic microscopy. For instance, the current revision of the proposed IPC codes for dry-packing state a percentage delamination allowable before added testing and/or dry-packing is indicated. Our work, however, indicates that delamination may occur without a measurable effect on semiconductor device reliability (Section 3). Some compounds may in fact contain additives that enhance device reliability while at the same time promote delamination (Section 2.3). It further illustrates that highly accelerated testing can induce secondary failure modes that may not be encountered under normal device operating conditions (Section 2).
1. Study A
1.1 Overview of Study A
Study A involved a screening design measuring moisture absorption, pad corrosion, ball lift, die surface and lead surface delamination on six current high performance, low stress mold compounds (A–F). The test vehicle was a 28 lead copper small outline integrated circuit (SOIC) with a bipolar chip. Each of these compounds was subjected to 85 °C and 85% relative humidity for 168 hours followed by 3 vapor phase cycles at 215 °C/FC70 and 30 PSI (gauge) moisture exposure. Another set of parts were submitted to PCT at 15 PSI (gauge) with only 85°C and 85% relative humidity precondition. Parts were removed periodically for testing. A flow diagram for these parts Is given in Figure 1.
Traditional logic theorizes that delamination at the chip and lead surface provides a pathway for moisture and other contaminants to reach the ball bonds which in turn leads to ball lift and pad corrosion failures during PCT testing. While this makes sense intuitively, the direct cause and effect relationship between delamination and chip failure has not been fully proven. In this study, six significantly different molding compounds were used to establish what correlation, if any, there was between delamination and moisture absorption, ball lift or pad corrosion.
1.2 Test Results for Study A
1.2.1 Moisture Absorption
As can be seen in Figure 2, System C had the greatest weight gain at 15 PSI followed by Systems F and A. All the other systems had equivalent moisture absorption at 15 PSI. Similar results were obtained from the parts at 30 PSI PCT (Figure 3).
at 15 psi PCT
at 30 psi PCT
1.2.2 Corrosion Testing
Each bonding pad was evaluated by a trained operator for corrosion and given a corrosion rating code as follows: 0 for no corrosion, 1 for slight corrosion, 2 for moderate corrosion and 3 for severe corrosion. Then the total corrosion rating was calculated as the summation of the number of pads at each corrosion level (none, slight, moderate, severe) multiplied by the corrosion rating code for that level. At 15 PSI, System F had the worst corrosion rating. System E also had a high corrosion rating under these test conditions (Figure 4). Analogous results were obtained under the 30 PSI conditions (Figure 5).
1.2.3 Ball Bond Pull Strength Testing
There was virtually no ball bond strength degradation at 15 PSI PCT until the 800 hour readpoint (Figure 6). Only System F showed significant degradation of ball bond strength at this point. At 30 PSI PCT, both Systems A and C had significant ball bond strength degradation, as well as System F (Figure 7).
1.2.4 Delamination Measured by Scanning Acoustic Microscopy
Lead and die surface delamination was measured using Scanning Acoustic Microscopy for the parts from the 30 PSI pathway. Delamination for the die surface is given based upon the proportion of the area of the die which was delaminated (i.e. if 1/2 of the die was delaminated, then the surface delamination rating = 0.5). The surface delamination rating for the leads is given by the summation of the fraction that each lead is delaminated, divided by the total number of leads. As can be seen from Figures 8 and 9, Systems A and D had high chip surface delamination. At the lead surface, Systems E and F had very high lead delamination. In addition, Systems A and D had some lead delamination.
1.3 Discussion
Based on these results, there is no correlation between cumulative pad corrosion and chip surface delamination. There may be a relationship between ball lift and chip adhesion, however, it is not a direct correlation.
In general, parts which exhibited lead delamination did poorly on ball lift testing and cumulative pad corrosion, but again, it is not a direct cause-and-effect relation ship based on this data. Further, those parts which exhibited the greatest moisture absorption had the greatest degradation of bond pull strength. However, ball bond degradation is known to be affected by many other factors including formulation additives (1). There was not a consistent tie between moisture absorption and cumulative pad corrosion. If moisture absorption were directly correlatable to delamination (2), System C should have the greatest delamination, followed by Systems F and A. Clearly, there were other modes of delamination failure generated, therefore, delamination cannot be expected, in all cases, to be a predictor of moisture performance.
In conclusion, Compound F exhibited high moisture absorption, ball bond degradation, corrosion and lead delamination but had minimal chip delamination. This was the most severe case, and is likely related to compound formulation and processing effects. However, all of the other compounds clearly had mitigating factors affecting their performance in specific areas. Secondary routes to lead and chip delaminations such as formulation additives may explain the lack of corrosion on these devices. This will be explored further in Study B.
Another possible cause of delamination is lead frame condition. For instance, FTIR analysis of solvents used to wash unused leadframes showed carbonyl bend and hydroxyl stretch indicative of the presence of esters, probably the residues of stamping oils used to prepare the leadframes. The presence of these oils may cause delamination to show up through scanning Acoustic Microscopy. But, they may also protect the leadframes from the oxidation and the device from the corrosive effects of moisture from sensitive areas of the device.
2. Study B
2.1 Overview of Study B
In this matrix, a fractional factorial design was used to evaluate the effects of compound formulation on moisture absorption, delamination and corrosion, as well as the correlation of these responses. System D (from Study A) was chosen as the control for this experiment because, although it exhibited lead and chip delamination, it had good real world performance including: low moisture absorption, high bond pull strength and moderate levels of corrosion.
This system was designated Compound A for this matrix and its formulation was used as the basis for the other six modifications (Compounds A-G). There were seven compounding parameters analyzed in the design. The test vehicle was again the 28 lead copper SOIC. The die was of the same dimensions as the original but had an updated circuitry. The matrix was designed as shown in Table 1. The test sequence is shown in Figure 10.
2.2 Test Results for Study B
2.2.1 Moisture Absorption
The moisture absorption curves generated at 15 PSI PCT indicate that Compound D had the greatest weight gain followed by Compounds E and F. The least weight gain was exhibited by Compounds C and G (Figure 11). At the 30 PSI PCT conditions, Compound E had the greatest moisture absorption followed by the rest of the compounds in a fairly tight group with the exception of Compound G. Its moisture absorption was considerably lower than that of the rest at 30 PSI PCT (Figure 12).
2.2.2 Corrosion Testing
There was virtually no corrosion on these parts after 800 hours of PCT at either 15 PSI or 30 PSI. This is attributed to the fact that a different chip was used than that of the previous study. It apparently has better corrosion resistance than the chip that was originally used.
2.2.3 Delamination Measured by Scanning Acoustic Microscopy
Scanning Acoustic Microscopy showed no die surface delamination at any of the readpoints at either 15 PSI or the 30 PSI conditions. After 250 hours of PCT at 15 PSI, the cumulative lead delamination of Compound G was significantly higher than the other six samples (Figure 13).
Compound C represented the other end of the spectrum with virtually no lead delamination after 250 hours of 15 PSI PCT. At the 30 PSI condition, Compound G still had quite high lead delamination and Compound C was still the lowest of the seven samples. But, a startling increase in lead delamination was seen in Compounds D and F (Figure 14). While most of the compounds’ lead delamination had doubled between the 15 PSI and the 30 PSI conditions, Compounds D and F exhibited a seven to nine fold increase. Clearly there was a different mode of failure occurring at the 30 PSI PCT condition, in these two systems, than that occurring at the 15 PSI PCT. This dramatic difference in level of lead delamination can be seen in the comparison of C-mode scanning acoustic microscope (C-SAM) results in Figure 15.
2.2.4 Live Device Testing
Live Device Testing was done on a 14 pfn DIP package, copper leadframes using a MC34002 die. The testing was done in a highly accelerated stress test (HAST) chamber at 85% relative humidity, 158 °C, and using a 5 V bias. The results of this testing are given in Table 2 (testing was discontinued after 2342 hours). Compounds B, D, F, and G are clearly superior with regards to this live device testing. While Compound B performed at the middle of the series in delamination testing, Compounds D, F and G had the highest lead delamination, traditionally considered detrimental to live device performance.
(MC34002 on Cu at 158 °C, 85% RH, 5 VDC Bias)
2.3 Discussion
Compounds D, F and G all contain formulation additive A. There is a strong correlation between this additive and the high level of lead delamination exhibited at 30 PSI PCT. Despite the low moisture absorption exhibited by Compound G at 30 PSI, its lead delamination is quite high due to the presence of additive A. Clearly, moisture absorption, a bulk property, cannot be relied upon to predict the adhesion performance of a compound.
With regards to device performance, it is clear that the presence of additive A in the formulation mitigates any negative side effects due to high moisture absorption or lead delamination. Additive B also appears to aid live device performance as can be seen in Compounds B and E.
Delamination is a questionable rejection criteria to use without supporting compound-specific data to confirm that such delamination is strongly correlated to device reliability.
3. Conclusion
From the evidence presented, it can be concluded that, at least in some cases, different failure modes occur at 15 PSI than at 30 PSI. In fact, these modes may be more prevalent than originally thought because, due to time and resource constraints, compounds which exhibit negative characteristics under highly accelerated testing are often rejected before complete evaluation of the compound has occurred. Under normal device operating conditions, there may be yet another set of failure modes in operation. An investigation of this hypothesis could have serious implications for all accelerated testing.
Secondly, delamination alone is not a reliable predictor of device performance. From the data presented, delamination may occur in the presence of certain formulation additives without catastrophically affecting device reliability. This is not to say that delamination is advantageous, or that it does not affect performance. However, delamination is a questionable rejection criteria to use without supporting compound-specific data to confirm that such delamination is strongly correlated to device reliability.
4. References
- A.A. Gallo, “Effect of Mold Compound Components on Moisture Induced Degradation of Gold-Aluminum Bonds in Epoxy Encapsulated Devices”, Proc. IEEE International Reliability Physics Symposium, 1990, p. 244
- S. Golwalkar, et al, “Moisture Sensitivity of Thin Small Outline Packages”, AOC 41st ECTC, 1991,p. 745
5. Acknowledgment
The contributions to this work of the following colleagues are gratefully acknowledged: John Gapa, sample preparation, Carl Snyder, bond pull and corrosion tests. Discussions with Anthony Gallo regarding ball bond degradation were most helpful.
