Drying ICs any advice

We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. How do we dry ? what are the alternatives - Nitrogen was a favorite medium in the past - any good ?

reply »

| We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | How do we dry ? | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? If you are buying new IC's fresh out of the dry pack you will not need to bake. But since that is not always the case the chip manufacturer should have a recommendation for a bake profile. I have never baked PLCC84, have never had to, consider sending the package out to a lab to determine the failure mode. If it was indeed a moisture caused failure there should be evidence of a crack, albeit maybe small. We use a N2 tank to store our BGAs after baking (if required). N2 alone will not Remove the moisture from the component. But I have a question, Is there a danger of moisture intrusion into a component during a wash cycle. We run some moisture sensitive devices on double sided reflow boards. And we will wash the first side before flipping the board to run the other. Any input?

reply »

| We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | How do we dry ? | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? Alan, I don't know if you can dry out components very easily with Nitrogen. I think baking the components is probably a better solution. You may incur some oxidation on the leads which will affect solderability but the activity of your flux system may be sufficient to handle it. If you are no-clean, I would recommend staying away from baking. There are some active no-cleans out there but as a general rule of thumb, it's not a good idea. Storage is your best friend. Check out there on the used market for a dessicant cabinet. We picked one up with a dual purge system (Terra-Universal) for right around $2K. That's dirt cheap and well worth it. Nitrogen consumption is only around 1 5028 cu. ft. cylinder every 4 weeks. Not too bad for a cost (58.00 per cylinder) but well worth it when it comes to the process improvements. How old were those parts? Popcorning shouldn't happen on parts like that unless they've been sitting out forever. Bottom line. Bake a couple of parts if you have the facilities(125 C for 12-24 hrs). See how well they solder and test. If all goes well, you know what to do. Otherwise, prevention is the best medicine. Regards, Justin Medernach

reply »

| | We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | | How do we dry ? | | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? | If you are buying new IC's fresh out of the dry pack you will not need to bake. But since that is not always the case the chip manufacturer should have a recommendation for a bake profile. | I have never baked PLCC84, have never had to, consider sending the package out to a lab to determine the failure mode. If it was indeed a moisture caused failure there should be evidence of a crack, albeit maybe small. | We use a N2 tank to store our BGAs after baking (if required). N2 alone will not Remove the moisture from the component. | But I have a question, Is there a danger of moisture intrusion into a component during a wash cycle. We run some moisture sensitive devices on double sided reflow boards. And we will wash the first side before flipping the board to run the other. | Any input? Mike, I ran the process you speak of for a while. I wondered the same thing myself but I never saw anything that would lead me to believe this was a dangerous process. Doesn't make much sense, does it? I suspect that the part isn't in the washing environment long enough to incurr a significant amount of moisture. There is also the drying phase at the end of the aqueous cleaner. This may be enough to counteract the "wet" portion of the wash cycle and make the entire process inert. Maybe somebody will help us out... Regards, Justin

reply »

| | | We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | | | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | | | How do we dry ? | | | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? | | If you are buying new IC's fresh out of the dry pack you will not need to bake. But since that is not always the case the chip manufacturer should have a recommendation for a bake profile. | | I have never baked PLCC84, have never had to, consider sending the package out to a lab to determine the failure mode. If it was indeed a moisture caused failure there should be evidence of a crack, albeit maybe small. | | We use a N2 tank to store our BGAs after baking (if required). N2 alone will not Remove the moisture from the component. | | But I have a question, Is there a danger of moisture intrusion into a component during a wash cycle. We run some moisture sensitive devices on double sided reflow boards. And we will wash the first side before flipping the board to run the other. | | Any input? | Mike, | I ran the process you speak of for a while. I wondered the same thing myself but I never saw anything that would lead me to believe this was a dangerous process. Doesn't make much sense, does it? I suspect that the part isn't in the washing environment long enough to incurr a significant amount of moisture. There is also the drying phase at the end of the aqueous cleaner. This may be enough to counteract the "wet" portion of the wash cycle and make the entire process inert. Maybe somebody will help us out... | Regards, | Justin I agree it's currious. What's more printed circuit boards, which are hydroscopic, don't seem to absorb much moisture during washing between reflow cycles either. TTYL Dave F

reply »

| We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | How do we dry ? | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? I'll attempt to answer the questions that have come up on popcorning. First, moisture delamination has always been in surface mount devices. The reason we notice it more today is because the die sizes are increasing in relation to the plastic size. The greater the die size, the easier it is to trap moisture between the flag and the bottom of the die. You'll find more delamination problems with the thin plastic packages such as TQFP's and BGA's. Nitrogen cabinets came remove moisture, but at a very slow rate. A fully saturated package would have to be stored for months to dry it out. If the package has been out of dry pack beyond its specified limits, our recommendation is to bake at 125� C for 24 hrs. You should have no trouble soldering with no-clean fluxes, provided that you use a reflow profile that is optimized to the type of flux you use. Running long preheat times with no-clean fluxes can be detrimental to solderability, due to the fluxes drying out prior to reflow. In order to ruin solderability by baking, you would have to bake until the Cu/Sn intermetallic of the lead grew to one half of the original plating thickness. An average PLCC should have < 50 microinches of Cu/Sn intermetallic when you receive it from your supplier. You should expect a Sn/Pb plating thickness of at least 300 microinches. Baking at 125� C, grows the Cu/Sn intermetallic at a rate of approximately 10 microinches every 24 hrs. At 10 days of baking you would begin to lose solderability. The oxides which are formed on the surface of the Sn/Pb plating float to the surface of the solder during reflow. Finally, plastic SMD's do absorb some small amount of moisture during cleaning, but it's not a problem unless the board and SMD's are going in to another solder process. Plastic packages such as PLCC's have to experience temperatures greater than 150� C for popcorn to occur. The entrapped moisture than turns to steam. Once the steam pressure exceeds the adhesion strenght of the epoxy used to adhere the die to the flag area, you get delamination. Two area's to pay attention to are double sided reflow, and rework. If your building a double sided reflow board, and you reflow packages, on the first pass which are good for 48 hrs.(example) out of dry pack. Then you wait more than 48 hrs to reflow the second side, good chance you'll pop the first pass parts. When you remove a plastic package at rework, especially if it's for analysis, you should bake the entire board before rework

reply »

| | | | We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | | | | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | | | | How do we dry ? | | | | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? | | | If you are buying new IC's fresh out of the dry pack you will not need to bake. But since that is not always the case the chip manufacturer should have a recommendation for a bake profile. | | | I have never baked PLCC84, have never had to, consider sending the package out to a lab to determine the failure mode. If it was indeed a moisture caused failure there should be evidence of a crack, albeit maybe small. | | | We use a N2 tank to store our BGAs after baking (if required). N2 alone will not Remove the moisture from the component. | | | But I have a question, Is there a danger of moisture intrusion into a component during a wash cycle. We run some moisture sensitive devices on double sided reflow boards. And we will wash the first side before flipping the board to run the other. | | | Any input? | | Mike, | | I ran the process you speak of for a while. I wondered the same thing myself but I never saw anything that would lead me to believe this was a dangerous process. Doesn't make much sense, does it? I suspect that the part isn't in the washing environment long enough to incurr a significant amount of moisture. There is also the drying phase at the end of the aqueous cleaner. This may be enough to counteract the "wet" portion of the wash cycle and make the entire process inert. Maybe somebody will help us out... | | Regards, | | Justin | I agree it's currious. What's more printed circuit boards, which are hydroscopic, don't seem to absorb much moisture during washing between reflow cycles either. | TTYL | Dave F Dave, Dave, I was able to attend a PCB technology course offered by a group called DMG Engineering (led by a former VP of Manufacturing of Hadco. It was their opinion that one should never have to bake a PCB for moisture removal purposes. Keeping in mind the guy is looking at things from a PCB shop perspective, and not an assemblers point of view. I work with a medium sized Electronic Assembly CM, and we have not baked a board in 3 years. As far as components go, I have read some interesting material on the subject. One book in particular "Ball Grid Array Technology" by John Lau. The book shows some moisture absorption charts, and also some bakeout curves. He makes note of a safe zone, a delamination zone, and a popcorn zone. At any rate the safe zone is situated at about .15% to .20% moisture weight vs component weight. The bakeout chart featured 3 sets of components, one at .475% moisture, one at .425% moisture, and one at .2% moisture. At the 4 hour mark of baking at 125C, all 3 sets of components were below .05% moisture, in fact the the set that absorbed moisture at 23C & 55% humidity for 150 hours to reach .2% moisture content, was at 0% moisture in 4 hours. We preach proper handling and drypacking of components, and if bake is necessary, bake in a nitrogen enviromnent to reduce oxidation. But the 25 hour bakeout at 125C is not necessary. This is especially importand if one is baking in an oxygen rich environment. The sample set exposed to 85C at 85% humidity took 50 hours to absorb .475% moisture, but the moisture was baked out to .04% at 4 hours, yet took an additional 21 hours to remove the last small percentage of moisture. Very interesting info, and keep in mind the focus was on PBGA's, but I feel the principals apply to plastic molded components. Keep in mind the percentage absorbtion rate per package weight would be greater with the greatest amount of surface area per weight such as TSOP and SSOP components versus SOJ and large SOIC components Obviously if you are dealing with tape and reeled components and cannot bake at 125C, then a longer bake at a lower temp is necessary. Good Luck Steve Abrahamson

reply »

Hi All! I've been reading this conversation which prompted me to break out the IPC-SM-786A from the IPC manuals from our parent company in Maryland sent us (we're just starting up in California). It's the Procedures for Characterizing and Handling of Moisture/Reflow Sensitive IC's. It's pretty good, actually... As far as how moisture is absorbed by IC's, it happens over time starting the moment they leave the cure oven after molding the package at the factory. Different molding compounds will absorb moisture at different rates, and another variable is the percentage of plastic vs. the die paddle size. The die paddle is a part of the lead frame material in which the silicon die is mounted to. The larger the die paddle, the thinner the plastic usually is, which makes a part more likely to crack. It was interesting to read that moisture isn't the only mechanism that causes package cracking. Some of it can be caused just by the mismatch of CTE's between the three main components of an IC; the molding compound, the die itself, and the lead frame. The CTE mismatch between the lead frame and mold compound is less for copper than it is for alloy-42, but the CTE mismatch between the die and the lead frame is greater with copper than with alloy-42. So there is a trade off when deciding what kind of lead frame material to use, affecting how susceptable a part is to cracking. One other variable is the lead frame parameters. They've done Finite Element Analysis and found that cracks initiate at the sharp corners of the lead frame, so the more rounded and smooth the geometry of the lead frame is, the less likely it is to crack. Different types of lead frame surfaces which promote the adhesion of the molding compound can reduce the likelyhood of package cracking. I know, all that is well and good and nice to know, but doesn't help much when you got the part and you got to reflow it, no matter what it's made of, or how it was designed. There was mention in conversation earlier of the standard 24-hr., 125 C bake, the other bake in the 786A is a low temperature bake that can be done in the parts shipping packaging...of course you need to remove the bubble wrap and stuff, but it's done at 40 C for 192-hrs. in a less than 5% relative humidity environment...larger parts may take longer. One thing, if you bake reeled parts like that, be aware that the peel force is going to increase quite a bit with the top film. If your equipment doesn't get bothered by it, good. Other things mentioned in the 786A to reduce cracking were: 1. Don't exceed 220 C during reflow. 2. The more cycles a part sees at reflow temperature the more likely it is to crack. So a component assembled on the first side of a double side reflow board that is reworked will stand a good chance of cracking. 2. During rework don't exceed 200 C. 3. If during double-sided reflow using moisture sensitive devices there is a delay in assembling the second side, you should bake the board prior to assembly. Same goes for rework, if a board sits for a while before it gets a part reworked, it will absorb enough moisture to require baking before rework.

reply »

| | We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | | How do we dry ? | | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? | If you are buying new IC's fresh out of the dry pack you will not need to bake. But since that is not always the case the chip manufacturer should have a recommendation for a bake profile. | I have never baked PLCC84, have never had to, consider sending the package out to a lab to determine the failure mode. If it was indeed a moisture caused failure there should be evidence of a crack, albeit maybe small. | We use a N2 tank to store our BGAs after baking (if required). N2 alone will not Remove the moisture from the component. | But I have a question, Is there a danger of moisture intrusion into a component during a wash cycle. We run some moisture sensitive devices on double sided reflow boards. And we will wash the first side before flipping the board to run the other. | Any input?

I've read your question and some very insightful answers to it. All is good. All is not so good when suppliers do not always perform adequate component testing for requirements such as moisture. Part of the testing process must include residual gas analysis (RGA). This analysis determines moisture, and other deleterious fluid contamination, left within a device after introduction to whatever element is being tested for. Supplier certifications (certs) should be traceable to the testing done whether by lot, run, or whatever. Also, all qualfied suppliers of all device types gladly provide customers with testing done and results of said testing. As for PCB's, baking is done less because suppliers are getting better - as are laminate, core, and prepreg suppliers. However, even though epoxy PCB material is only .2% hygroscopic, this moisture is capable of blowing out plated holes when they do not meet specified requirements for plating thickness, ductility, and aspect ratio. For this reason, and because not all suppliers are proven capable, it often is wise to prebake before assembly. Even some device types require this attention when the above requirements are not met and testing evidence is not available. Earl Moon

reply »

| | | We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | | | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | | | How do we dry ? | | | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? | | If you are buying new IC's fresh out of the dry pack you will not need to bake. But since that is not always the case the chip manufacturer should have a recommendation for a bake profile. | | I have never baked PLCC84, have never had to, consider sending the package out to a lab to determine the failure mode. If it was indeed a moisture caused failure there should be evidence of a crack, albeit maybe small. | | We use a N2 tank to store our BGAs after baking (if required). N2 alone will not Remove the moisture from the component. | | But I have a question, Is there a danger of moisture intrusion into a component during a wash cycle. We run some moisture sensitive devices on double sided reflow boards. And we will wash the first side before flipping the board to run the other. | | Any input? | | I've read your question and some very insightful answers to it. All is good. All is not so good when suppliers do not always perform adequate component testing for requirements such as moisture. Part of the testing process must include residual gas analysis (RGA). This analysis determines moisture, and other deleterious fluid contamination, left within a device after introduction to whatever element is being tested for. | Supplier certifications (certs) should be traceable to the testing done whether by lot, run, or whatever. Also, all qualfied suppliers of all device types gladly provide customers with testing done and results of said testing. | As for PCB's, baking is done less because suppliers are getting better - as are laminate, core, and prepreg suppliers. However, even though epoxy PCB material is only .2% hygroscopic, this moisture is capable of blowing out plated holes when they do not meet specified requirements for plating thickness, ductility, and aspect ratio. For this reason, and because not all suppliers are proven capable, it often is wise to prebake before assembly. Even some device types require this attention when the above requirements are not met and testing evidence is not available. | Earl Moon

Also, with qualified suppliers, certs are traceable to testing and offer a means of recourse when failure, as described, do occur. Thank You, Earl Moon

reply »

| We had some failures on an IC package that the manufacturer has put down to moisture inside the package, leading to popcorning. | This seems a bit odd, the package is a PLCC 84 which is quite thick. I have only seen this problem on thin packages. | How do we dry ? | what are the alternatives - Nitrogen was a favorite medium in the past - any good ? I recommend baking the ICs in a vacuum oven (prior to assembly). This procedure will rapidly remove moisture, and does not require a high temperature. Pete Sorenson

reply »