---

Advancing Microelectronics • Volume 29, No. 3 • May/June, 2002

---

Lead-Free Soldering Causes Reliability Risks for Systems with Harsh Environments

---

Hans Danielsson, MIKROELEKTRONIK KONSULT AB, Kumla, Sweden, hans.danielsson@mbox301.swipnet.se

ABSTRACT

The Japanese bans on solders containing Lead in mobile telecommunication products have caused the other telecommunication industries to go for Lead-Free solutions. The Lead-Free solder used by most telecommunication industries is SnAgCu-solder (95,8Sn3,5 Ag0,7Cu). This solder has a melting point of about 217°C. This indicates that the peak reflow temperature will increase to 250-260°C, compared to the peak reflow temperature of 215-230°C with eutectic PbSn-solder.

The increased solder reflow temperature will cause thicker intermetallics (SnCu and SnNi). This will have a negative effect on long term reliability, especially for harsh applications like under hood electronics in cars which will be shown.

The increased reflow temperature can give reliability problems for plastic packages used in harsh environments after soldering.

INTRODUCTION

The Japanese industries have banned Lead in mobile telecommunication products from January 1, 2002. Therefore Ericsson has decided that 80% of their NEW telecommunication products shall be Lead-Free from January 1, 2002, and Nokia has taken a similar standpoint. Most other telecommunication companies have made similar decisions.

The telecom market is the biggest user of semiconductor packages (see table 1).

From Table 1 (Reference 1) it can be seen:

• Telecom is by far the biggest market sector with the fastest increase in % per year
  
• The harsh market sector is Automotive with 5.8% of the market 2004 and Military/Space with 1.1% of the market 2004
  
• The benign environment market (Total market — harsh market) is using 93.1 % of all components in the world
  
• The automotive components market can be split into two groups the “Under hood” components and passenger compartment components. The “under hood” components take probably not more than 1-2 % of the total sale of the semiconductors in the world. The risk is then that the semiconductor manufacturers ignore the harsh market (Automotive and Military/Space) with its very high reliability demand
  

RELIABILITY OF CARS

In Reference 2 it is shown how many percent of cars sold in Sweden 1977, still were running on Swedish roads 1977, in spite of the salt (NaCl) sprayed on the roads in wintertime in order to melt the snow and ice. It is shown that more than 20-30% of the Volvo, Mercedes, Saab and BMW cars were in use 20 years after they were sold.

Let us define a vital system in a car as a system necessary for the car's safety or the car's ability to fulfil legal environmental regulations. Therefore from Reference 2 it can be concluded that vital mechanical systems in cars can last for more than 20 years. Consequently vital electronic systems in cars should have a life length of 15-20 years with high reliability.

If a car brakes down, the concern of the driver/owner is how much is the repair cost and how long time it takes to repair. In this stage it doesnít matter if it is a vital mechanical or vital electronic system that failed.

In Reference 2 it is shown that vital mechanical and electronic systems are allowed to have maximum 2040 ppm of failure (0,204%) per system, if there are 10 vital mechanical and 10 vital electronic systems in the car, and if the total functional probability (=reliability) of the car is 0,96.

If, however, the number of vital electronic systems is increased from 10 to 15 over a certain time period, the maximum allowed failures on ALL vital electronic systems have to be maximum 1360 ppm failures per vital electronic system (0,136%), if the reliability of the car is unchanged (functional probability=0,96). This is an important observation for car manufacturers, especially today, when the electronic systems in cars have a trend towards being more and more integrated with each other. It will be harder to say which electronic system is vital and which electronic system is not.

NEW QUALITY DEMANDS ON AUTOMOTIVE ELECTRONICS

The Swedish car manufacturers Volvo and Saab have now started to demand increased quality of the vital electronic systems they are buying. They demand a life length of 15-20 years and with a driven distance of 160.000 km (100.000 miles) the accumulated failure per vital electronic system shall fulfill the following (variations exist between different systems):

This is a very much increased quality demand compared to what is custom in the automotive industry, however it is needed.

EXTREME ENVIRONMENTS IN CARS

Delphi Delco (Reference 3) has given the following extreme environments in cars (Table 2):

Reference 3 also pointed out that Typical junction temperatures for ICs are 10-15°C higher than the baseplate temperature, and that power dervices can reach 25°C higher temperatures than the baseplate.

CONSEQUENCES OF TELECOM MARKET GOING TO LEAD-FREE PRODUCTS

If the telecommunication market sector is going to Lead-Free Products, which is highly probable, the semiconductor manufacturers will start supplying Lead-Free Components. This will in turn affect all other market. If the components are having Lead-Free bumps and platings, all other market sectors have to use such components.

One study (Reference 4) has shown that there are more than 400 plastic molding compounds in use in the semiconductor market to day.

The semiconductor industry is mainly interested in giving their ”big” customers the best value for the money and will therefore optimize their semiconductor plastic packages for the big users. This means that a semiconductor manufacturer can make big or small changes in the assembly of the plastic packages (molding materials, adhesive for mounting the chip, lead frame surface finish, solder resist, etc.), without giving their customers a notice, as long as these changes will not have negative effects for the ”big” customers. Therefore it is clear that it will be “very hard and very costly” for customers with harsh environments to make “Qualifications” in the same way as the military and space people (maybe?) can afford.

FAILURE MODES CAUSED BY THE INCREASED REFLOW TEMPERATURE

1. FAILING SOLDER JOINTS DUE TO TOO THICK INTERMETALLIC LAYERS

Chan et al (Reference 5) and Tu et al (Reference 6) are working in the same group on the intermetallics between Sn and Cu and Sn and Ni and its influence on reliability. They studied a µBGA-package with 46 bumps (bumps soldered to Cu) mounted on a PCB with 105µm Cu, 15µm Ni and flash Au in vibration with an extra weight of 56 gram glued on the µBGA, bending test (the µBGA mounted on a PCB 70 x 40 mm with a strain gauge mounted just under the the µBGA. One end of the PCB fixed, the other end bended).

Figure 1 shows that too low heating factor, i.e., too short time with the solder in melted condition, has low vibration and bending fatigue life. This is because of too short time in melted condition, the solder hardly had time for wetting the pads. At a heating factor of about 680, the vibration and bending fatigue life is longest. Higher heating factors will decrease the vibration and bending fatigue life. As explained in Reference 6, there is a direct correlation between the vibration fatigue life and the thickness of intermetallics (formed during the soldering process) between the solder and the pads on the components and PCBs.

The reflow time for maximum vibration and bending fatigue life in Figure 1 is 49 seconds above solder melting temperature (183°C) and peak temperature 203°C giving a Q-factor of about 680 (Reference 5). (Q=Soldering Heating Factor)

The use of Lead-Free Components will increase the peak reflow temperature from todayís 215-230°C to 250-260°C. For example NEMI recommends a Lead-Free Soldering profile (Reference 7) for (Sn95,8Ag3,5Cu0,7) with a solder MP=217°C: “Temperature maintained over 217°C=60-150 seconds with peak temperature 260°C (-5 + 0°C).”

With a lead-free solder Q°2x20x43/2+20x43= 1720°C-sec. to > 3000 depending on actual soldering profil (2x40x43/2+70x43=4730). If we study Figure 1 we see that the vibration fatigue life at this Q-level is about 15% of its maximum value (at Q=680). Note that this is the situation just after the package is soldered to the PCB!

If then the electronics is engine mounted, the operating temperature is often over 100°C for several thousand of hours during the life of the car, creating further increase in intermetallic thickness. In Reference 6 they measured the SnCu- and SnNi- intermetallic thickness after aging up to 36 days (864 hours) at 120°C. This thickness increased lineary with the square root of time, at least up to 36 days. When they vibrated the µBGA-packages at 9g rms at 30 Hz with an extra weight of 56 grams (Reference 6) they found:

As can be seen from table 3, there is a clear relation between the increased thickness of the intermetallics between Sn-Ni and Sn —Cu and increased percentage of failing of the solder joints.

2. CRACKS AND DELAMINATIONS IN IC PLASTIC IC PACKAGES

Intel together with University of Maryland has recently (Reference 7) presented a very interesting study. They tested PBGA-packages intended for PbSn-eutectic soldering (230°C) for a Lead-Free soldering profile with the peak temperature at 260°C.

They had packages soldered at 230°C as reference. Before soldering the packages were inspected with a Scanning Acoustic Microscope (SAM) to detect delaminations, then preconditioned for 120 hours at 60% RH. After soldering the packages were inspected in SAM again and submitted to unbiased HAST Test (130°C, 8 % RH) for up to 500 hours.

The result was that they could not find any delaminations in the SAM-inspection directly after soldering with the Lead-Free reflow profile (260°C peak temperature).

However after 200 hours HAST-test they found a lot of delaminations and cracks on packages going through the Lead-Free soldering profile. No delaminations were found on packages going through the PbSn soldering profile. The result can be explained in the following “visual” way:

This Figure shows that for this specific chip size and other specific details by Intel-tested package soldered at 230°C can stand the 200 hours HAST test. However if soldered at 260°C this specific package could not. Please note that the tested package could probably stand the telecommunications environment very well! How many of the 400 molding compositions on the market today can stand the telecommunications environment but NOT the automotive?

The result is very frightening for users with harsh environments, because there is no way to check the packages directly after a Lead-Free soldering process and see if any damage has occured. Only the field use will tell.

3. ELECTROMIGRATION AND SHORTING OF PCBs

If the reflow temperature is increased to 250-260°C, there are consequences for the cleanliness during manufacturing of the Printed Circuit Boards (PCBs).

In (Reference 8) Munson reports that they measured 6 batches of PCBs for cleanliness in incoming inspection. All batches met the military cleanliness criteria of having < 3,3µg/inch2 NaCl equivalent. When they took PCBs from each batch (with soldered components) and put them on a biased humidity test (85°C, 85% RH) all boards failed due to shorting (electro migration) in less than 168 hours.

If they instead cleaned the boards (after soldering the components) all boards passed the biased humidity test for more than 336 hours.

Then they measured the cleanliness on the bare PCBs before and after different soldering profiles with the following results:

The above shows that the higher temperatures in the soldering process, the more contaminations imbedded in the interior of the board, are diffusing through the solder masks. To counteract this failure mode, the PCB-manufacturers should be forced to clean the boards at critical manufacturing steps, and measure the cleanliness, before applying the solder mask to the PCBs. A protocol of the level of cleanliness should follow each manufacturing batch of PCBs. In a Lead-Free soldering process this is more important than ever.

POSSIBLE SOLUTIONS

A. SMD-electronics for passenger compartments.

A suitable way to prevent moisture-induced failures is to prevent the moisture to reach the interior of the plastic packages. This will eliminate or drastically reduce the cracking and delamination of plastic packages in harsh environments as well as prevent migration and shorting of the PCBs. A possible method is to use a ProofCap as a moisture barrier. The ProofCap system is a patented very low cost system containing metal sheets that are non-penetrable for moisture. See Figure 3.

It is possible to simulate the moisture diffusion into ProofCap based on results from accelerated tests and models for moisture absorption and diffusion in plastic materials (Reference 9). For example in the Middle East countries the maximum RH of 100% (Relative Humidity) occurs at night and the low RH of 35% during warm summer days.

Figure 4 shows a simulation for RH inside a typical ProofCap enclosure (green line) as function of ambient RH (red line) for a typical Middle-East country. The condition for this simulation is that internal heat dissipation is causing the temperature inside the enclosure to be 10°C over ambient temperature during 5% of the time. For example a car driven 500 hours per year, during 20 years, is in use 5.7% of the total time.

As can be seen the internal RH is hardly over 60% after moisture ingress in the enclosure, even after 20 years. If this RH is too high, it can be lowered by inserting a suitable moisture getter inside the enclosure.

B. Electronics for under hood applications.

This type of application has high operation temperatures, often high vibration levels and see large DTs. As has been shown in this report, the intermetallic thickness will grow too thick in this type of applications. The problem with the plastic packages in harsh environments (delamination and corrosion), together with the often high vibration level and high operating temperature in under hood applications, will make SMD-technology on PCBs useless as a packaging technology, because the reliability long-term demand cannot be met.

A suitable packaging technology for this kind of application is to use the bare chip technology. This is because then the MCM (Multi Chip Module) manufacturers can control the assembly and encapsulation process. Today, this technology is increasingly used in mobile phones, often as stacked 3-D-modules with bare chips. This in turn has interested the semiconductor industry to develop test and burn-in procedures on the wafer level (Reference 11). For example one semiconductor company has today more than 200 different chip types where they guarantee the same parameter values and the same quality of the chip as the equivalent plastic encapsulated circuits (Reference 10). This chip situation is similar for other semiconductor companies.

The environmental protection is by the well-known silicon gel. The interconnection between the chip and substrate can be either wire bonding or Flip Chip depending on the substrate material. Substrate material can be either silicon or ceramics or stainless steel with printed ceramic dielectric layers on the steel. Stainless steel substrates are a very robust technology that is supposed to have a great future in automotive electronics (Reference 12,13). Flip Chip technology is, however, very demanding and the conductor material on the substrate must be robust, have the “right” TCE and meet the long-term reliability demand. It must be noted that the bare chip technology is a technology demanding deep knowledge about materials and processes for harsh applications. A knowledge not found too often in the industry today.

CONCLUSION

The biggest and fastest growing market sector, the telecommunication sector, is going to use Lead-Free soldering materials. This will cause the peak reflow soldering temperature to increase to 250-260°C from todayís 215-230°C. The increased reflow soldering temperature will have negative influence on the long time reliability on the plastic packages.

There are over 400 molding compositions for encapsulation semiconductors on the market to day. Therefore it is not possible to select the “right” molding material for harsh applications.

It is shown in this report that at least some plastic packages in use today, will not stand a harsh and humid environment, after being soldered in a Lead-Free process (250-260°C). The soldering process will not make any detectable defects that can be screened out directly after soldering; the delamination will occur during field use in harsh environment.

The Lead-Free soldering process (250-260°C) will drive more contaminations from the interior of the PCBs to the outside of the boards compared to the PbSn soldering process (220-230°C). Therefore very stringent cleanliness demands have to be implemented in the PCB-manufacturing for automotive applications.

If the humidity could be prevented from being absorbed in the plastic packaged components, the delamination and later corrosion failure mode could be eliminated. Therefore, the ProofCap system is suggested as a solution for automotive electronics in the passenger compartment to allow plastic packages “of today” to be used and then also the normal SMT-technology.

The under-hood applications have high operation temperatures and large DTs and often high vibration levels. For such applications the Surface Mount Technology cannot stand the reliability demands if the life length with very high reliability should be longer than 5 years. For such applications a bare chip approach is suggested.

REFERENCES

1. R.Branchato,”Millennium Prediction: A Package for all Occasions,” Advanced Packaging, January 2000, pp. 30-34
   
2. H. Danielsson, “Reliability Risks with Lead-Free Soldering and Possible Solutions,” SAE-World Congress, March 4-7, 2002, Paper 2002-01-1048
   
3. M.Ray Farchild et al., “Emerging Substrate Technologies for Harsh-Environment Automotive Electronics Applications,” SAE-World Congress, March 4-7, 2002, Paper 2002-01-1052
   
4. John K. Hagge, “ROBOCOTS: A Program to Assure Robust Packaging of Commercial-Off-The-Self (COTS) Integrated Circuits,” IMAPS 2000 Proceeding, September 20-22,2000, Boston, USA, pp.139-147
   
5. Y.C. Chan et al., “Reliability Studies of µBGA Solder Joints-Effect of Ni-Sn Intermetallic Compound,” IEEE Transactions on ADVANCED PACKAGING, Vol 24, Nr 1, February 2001, pp.25-31
   
6. P.L.Tu et al., “Effect of Intermetallic Compounds on Vibration Fatigue of of µBGA Solder Joint,” IEEE Transactions on ADVANCED PACKAGING, Vol 24, No 2, May 2001, pp.197-205
   
7. L. Yang et al., “The Impact of Lead-Free Soldering on Electronic Packages,” Microelectronics International, Vol 18, No 3, September 2001, pp.20-26
   
8. N. Munson, “Identifying the variables in Fine Pitch Technology that affect Reliability,” Nepcon West Proceeding 1993, pp 1165-1173
   
9. Björneklett, ProofCap Simulation Program, Ericsson Components 1996, Revised 1977
   
10. National Semiconductors Chip Seminar, October 8, 2001, Baltimore,USA
    
11. R.Arnold et al., "Test Methods Used to Produce High Reliable Known Good Die (KGD) MCM 1998 Proceeding, April 15-17,1998, Denver, USA, pp.374-381
    
12. M.E.Ellis, "Thick Film on Stainless Steel Circuit Board,” IMAPS 1999 Proceeding, October 1999, Chicago USA, pp.106-111
    
13. M. Levitski et al., “A Cofireable Dielectric on Stainless Steel Thick Film Material System for Automotive Applications,” IMAPS 1999 Proceeding, October, Chicago, USA, pp. 112-117
    
    

For information on ProofCap see www.proofcap.se

| Table of Contents | Previous Page | Next Page |