Comparison of Mechanical Reliability of Three Underfill Materials for Flip Chip Bumped on High Tg PCB HDI for Automotive Applications
G. Fabbri, C. Sartori, G. Scarano, Magneti Marelli Electronic Systems S.p.A., V.le Carlo Emanuele II, 118, 10078 Venaria Reale (TO) ITALY, Phone: +390116879-256 / 799 / 277, Fax: +390116879199, e-mails: giancarlo.fabbri@venaria.marelli.it; claudio.sartori@venaria.marelli.it; giuseppina.scarano@venaria.marelli.it
Abstract
The importance of the presence of underfill in flip chip technology bumped on high Tg PCB HDI is checked. Three different underfills are compared in term of stress distribution at ambient and at high temperature, and in term of mechanical reliability after thermal aging by means of C-SAM, microsections, SEM and electrical resistance measurements. Thermal fatigue tests are performed in order to reproduce environmental conditions of use typical of automotive electronics. A different behaviour of the three underfills is found but no failures are detected.
Key words: Underfill, Flip Chip, HDI, Reliability
1. Introduction
The aim of this work is to study the importance of the presence of underfill encapsulant in flip chips assembled on high Tg (glass transition temperature) PCB HDI (Printed Circuit Board High Density Interconnection) and the mechanical reliability of underfill after environmental tests usually performed for automotive applications.
In the first part of this work, we examine the effect of voids in underfill on flip chip bumps after thermal fatigue tests, using C-SAM (Scanning Acoustic Microscope) images and metallographic microsections.
In the second part, three different underfills (named A, B and C) are studied. Camber measurements are made at ambient temperature and at 100°C in order to evaluate stress distribution dependence on temperature; flip chip electrical resistance measurements are recorded before, durin and after thermal stresses and finally filler size measurements and filler deposition analysis are performed at SEM (Scanning Electron Microscope).
Flip chips assembled for these tests are in a daisy chain 18x18 full matrix array pitch 0.5 mm, bump alloy is Sn62/Pb36/Ag2, the bump diameter is 250 µm and the gap between substrate and die is 140 µm.
The substrate is 2 (core) +2 (build-up top) +2 (build-up bottom) layers PCB made in HDI technology. Core layers are an epoxy resin with glass cloth, its thickness is about 550 µm, while build-up layers are epoxy with silica filler. Both materials have Tg = 160 ˜ 170 °C. Total thickness is about 1 mm. The choice of this kind of PCB is to investigate the possibility to assemble flip chip components on organic substrate for high temperature applications as a possible alternative for ceramic boards.
Three different thermal fatigue tests were performed on the samples. Test conditions and the number of flip chips subjected to each test are reported in table 1.
2. Analysis of voids in underfill materials
A suitable means to investigate the presence of voids in underfill is C-SAM acoustic microscopy that is a non-destructive inspection technique.1 In fig. 1, three flip chips with voids in underfill are shown: the lack of encapsulant is evident in wide regions, leaving tens of bumps uncovered. It is interesting to evaluate if these voids affect the reliability of the device. For this reason, the electrical resistance of the dies was recorded before and after thermal shocks. The first step of 83 thermal shocks already revealed the failures (open circuit) of the three dies. In order to verify the typology of the damage, microsections of the devices were performed. In fig. 2, a general view of voids is shown. In fig. 3, a zoom on a crack in solder bump is shown. In fig. 4, the delamination of the Under Bump Metallization layer is evident along with a crack in the silicon die. This analysis demonstrates how important the presence of an encapsulant between the bumps in CSP (Chip Scale Packaging) technology is for the reliability of a flip chip. In fact, underfill reduces the CTE (Coefficient of Thermal Expansion) mismatch between the organic substrate and the silicon die acts as a stress relief material for the solder joints.2 Therefore, the lack of underfill in the studied samples is the reason of formation of cracks in the bumps due to the stresses induced by thermal treatment.
3. Underfill analysis
The reliability of three encapsulant materials is studied in order to check their mechanical properies in terms of cam-ber (laser pro- filometer analysis), degradation after en-vironmental tests (electrical resistance and C-SAM) and filler content. The char- acteristics of the three underfills are reported in table 2.
Curing conditions are quite different among the three underfills. For example, the curing temperature and time of underfill C are great- er than he ones of the other two encapsulants. Therefore, flip chips underfilled with this resin could be more stressed during the assembly process.
3.1. Laser profilometer measurement
Camber measurements have been performed at ambient temperature and at 100°C on a flip chip for each kind of underfill by means of a laser profilometer. The camber is evaluated recording the height gap between the centre and the edges of the die. The values measured are reported in table 3. The profiles at ambient temperature for underfills A, B and C are shown in fig. 5, 7, 9 respectively, while in fig. 6, 8, 10 the measurements at 100 °C are reported.
Analysing fig. 5 to 10, camber profile differences are evident: at ambient temperature, the curvature is much more pronounced compared to the high temperature one. This is due to the softening of the resin approaching the glass transition temperature of the material. Data reported in table 3, show that the three underfills have a different behaviour at ambient temperature: underfill A has the lowest camber as we expected for its lowest CTE (table 2); on the other hand, at high temperature, the height gap is practically the same for the three encapsulants. It must be remarked that the height gaps measured are affected also by the deformation of the substrate, but it can be assumed that this influence is the same for every measure.
3.2. Environmental tests
The thermal fatigue tests (table 1) were performed in order to evaluate the behaviour and the structural modification of the encapsulant material after aging. Analysis is focused on flip chip electrical resistance values to evaluate how underfills affect this property, and on C-SAM images, to see eventual delaminations or cracking in the resin.
No failures (open circuit) were detected in the measure of the electrical resistance, but some modifications are evident. In fig. 11, 12 and 13 the relative variations of electrical resistance values before and after the three aging treatments (storage, slow thermal cycles and thermal shocks respectively) are shown in respect to the three underfills. The mean, maximum and minimum values are reported. The relative variations for underfill C are always smaller than the variations for underfills A and B. For slow cycles and thermal shocks, this difference is stronger than for storage. The mean variation values are reported in table 4.
From these data, it is evident that the relative variation is larger after storage than after the other two tests. Probably, it is more critical for the capability of an underfill to protect the joints remaining at high temperature for a long time than during thermal cycling. Finally, generally speaking, the decrease of electrical resistance values is not only due to the component modification after thermal aging, but it’s also due to the encapsulant properties.
After environmental tests, all the flip chips assembled have been examined with C-SAM: there is no evidence of voids, delamination and cracking in underfill. Therefore, after thermal aging, encapsulants don’t show any visual defects or any structural modification as observed on microsections examined by means of a metallographic microscope.
To complete the analysis, some device microsections were performed in order to observe the three underfills before and after thermal aging, for evaluating filler content, sedimentation and modification after environmental tests. The filler consists of silica spheres embedded in epoxy resin uniformly distributed in the gap between the sub- strate and the silicon die.3 SEM images of underfills A, B and C are shown in fig. 14, 15 and 16, respectively, after thermal aging. It’s evident from the different filler size: underfill A is characterised by silica spheres with diameters from 2 µm to more than 20 µm, for underfill B, the diameter of the spheres is more homogeneous (around 1 µm), while underfill C contains filler whose size is in the range 2-5 µm. No change in filler size or in sphere distribution is noticed after environmental tests.
4. Conclusions
The presence of underfill material is needed to assure good reliability of a flip chip bumped on high Tg PCB HDI. We have found a close correlation between the presence of voids in the encapsulant and the occurrence of failures (open circuits): in fact during the device life, if some voids are present, there is a high probability for crack propagation on solder bumps due to thermal-mechanical stresses.
Laser profilometer analysis revealed that mechanical stresses on flip chips are higher at ambient temperature than at temperatures approaching the Tg of the underfill resin. These stresses depend on the type of underfill.
No failures have been found after thermal fatigue tests, but a reduction of electrical resistance values is evident. Underfill maintains its capability to preserve solder joints after aging (no delaminations, no cracking), even if a different behaviour of electrical resistance value is verified depending on the environmental treatment and on the type of underfill. In particular, one underfill has shown greater stability on the relative variation of electrical resistance values than the other two regardless of thermal treatment. Moreover, for the three kinds of encapsulants, it is critical to remain at high temperatures for a long time than to perform thermal cycling.
The filler content is checked by means of SEM and metallographic microscopy: no changing of silica spheres is evident after environmental tests.
Finally, according to the results of analysis performed in this work, it’s possible to state that the assembly of flip chip devices on high Tg PCB in HDI technology for automotive applications is feasible.
Acknowledgments
The authors would like to thank O. Di Tollo (Centro Ricerche Fiat) for the useful technical support in performing SEM analysis and R. Brignolo for reviewing this work.
References
[1] J. E. Semmens, S. R. Martell, L. W. Kessler, IJMEP, volume 18, number 4, pg. 382, 1995
[2] J. H. Lau, C. Chang, IJMEP, volume 22, number 1, pg. 20, 1999
[3] M. B. Vincent, C. P. Wong, SSMT, volume 11, number 3, pg. 33, 1999
About the Authors
Giancarlo Fabbri received his degree in Material Science from the Universit di Torino in 1998. He worked on deposition and characterisation of glasses for telecommunications deposited by PCVD. He joined Magneti Marelli in 2000 and is presently working on lead free soldering and on encapsulant materials.
Claudio Sartori received his degree in Nuclear Engineering in 1992 and Ph.D. in Plasma Physics in 1997 from the Politecnico di Torino. He worked four years in the semiconductor field and joined Magneti Marelli in 2000 where he is presently working on thermal analysis and reliability on HDI substrates.
Giuse Scarano received her degree in Physics from the Universit di Torino in 1985. She worked on characterisation of semiconductor materials for optoelectronic applications. She has over 15 years of experience in hybrids, PCB, packaging and assembly. She is presently working on lead free soldering.
| 
