Soldering to Gold

We have some SMT components with gold plated leads. They're actually LCC led packages. A good solder alloy to use would be Indium / Lead to reduce tin dissolution into the gold and avoid the resultant brittle intermetallic compound. However, a recommendation has also been made to use an alternative Sn62 Pb36 Ag2. The introduction of silver is said to reduce the leaching of tin. This alloy is practically a "drop in" and it is less expensive than an indium alternative. Has anyone had experience with soldering to gold leads with this silver alloy? Thanks, DaveJ

reply »

| We have some SMT components with gold plated leads. They're actually LCC led packages. A good solder alloy to use would be Indium / Lead to reduce tin dissolution into the gold and avoid the resultant brittle intermetallic compound. However, a recommendation has also been made to use an alternative Sn62 Pb36 Ag2. The introduction of silver is said to reduce the leaching of tin. This alloy is practically a "drop in" and it is less expensive than an indium alternative. | Has anyone had experience with soldering to gold leads with this silver alloy? | Thanks, | DaveJ | In the past, soldering or using LCCC's or any other electroplated gold solder termination area devices, I and most of the military types always remove the gold in a solder pot. Sounds crude, but it's the only way to really assure no deleterious effects.

Earl Moon

reply »

| | We have some SMT components with gold plated leads. They're actually LCC led packages. A good solder alloy to use would be Indium / Lead to reduce tin dissolution into the gold and avoid the resultant brittle intermetallic compound. However, a recommendation has also been made to use an alternative Sn62 Pb36 Ag2. The introduction of silver is said to reduce the leaching of tin. This alloy is practically a "drop in" and it is less expensive than an indium alternative. | | Has anyone had experience with soldering to gold leads with this silver alloy? | | Thanks, | | DaveJ | | | In the past, soldering or using LCCC's or any other electroplated gold solder termination area devices, I and most of the military types always remove the gold in a solder pot. Sounds crude, but it's the only way to really assure no deleterious effects. | | Earl Moon | Dave: Earl's correct. Get that gold into your solder pot, rather than remaining in your solder pads. Don't forget that the gold will remain in the solder pot and could affect other uses of the pot.

The Sn62 Pb36 Ag2 story is an old wives tale. It may be a nice theory with the different sized atomic spacing and all that, but there is no documentation that it works.

Good luck

Dave F

reply »

| We have some SMT components with gold plated leads. They're actually LCC led packages. A good solder alloy to use would be Indium / Lead to reduce tin dissolution into the gold and avoid the resultant brittle intermetallic compound. However, a recommendation has also been made to use an alternative Sn62 Pb36 Ag2. The introduction of silver is said to reduce the leaching of tin. This alloy is practically a "drop in" and it is less expensive than an indium alternative. | Has anyone had experience with soldering to gold leads with this silver alloy? | Thanks, | DaveJ | Dave

I like the Ag2 alloy, but not for that. It's a tad tougher than 63/37 without losing much on the ductility, and the joints are slightly more brilliant (i.e. a doubtful one stands out slightly more). The dissolution rate for Au in 63/37 is c. 5 �m/sec at about 230�C. In 62/36/2, it will be about 4.5 �m/sec and you'll get the same intermetallics. Pretinning is the answer.

BTW, I hand-soldered tons (well, kgs) of 0.8 and 1.0 mm solid gold wire in sensors which my co. used to make, onto copper wire using 62/36/2. The results were horribly mat. The whole lot was cast into an epoxy mass: never once had a fracture in service and some of these were in service for 15 years under fairly nasty climatic conditions. What I'm suggesting here is that cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics from causing cracking if you don't pre-tin but, for Crissake, do your qualification tests first.

Wave soldering presents no problems because the �g of Au per device gets lost in that 300 kg mass of molten solder, as it is pretinned and soldered in the one operation. Reflow is hell, because your �g stays in the mg of paste you laid down.

My tuppence worth

Brian

reply »

| | We have some SMT components with gold plated leads. They're actually LCC led packages. A good solder alloy to use would be Indium / Lead to reduce tin dissolution into the gold and avoid the resultant brittle intermetallic compound. However, a recommendation has also been made to use an alternative Sn62 Pb36 Ag2. The introduction of silver is said to reduce the leaching of tin. This alloy is practically a "drop in" and it is less expensive than an indium alternative. | | Has anyone had experience with soldering to gold leads with this silver alloy? | | Thanks, | | DaveJ | | | Dave | | I like the Ag2 alloy, but not for that. It's a tad tougher than 63/37 without losing much on the ductility, and the joints are slightly more brilliant (i.e. a doubtful one stands out slightly more). The dissolution rate for Au in 63/37 is c. 5 �m/sec at about 230�C. In 62/36/2, it will be about 4.5 �m/sec and you'll get the same intermetallics. Pretinning is the answer. | | BTW, I hand-soldered tons (well, kgs) of 0.8 and 1.0 mm solid gold wire in sensors which my co. used to make, onto copper wire using 62/36/2. The results were horribly mat. The whole lot was cast into an epoxy mass: never once had a fracture in service and some of these were in service for 15 years under fairly nasty climatic conditions. What I'm suggesting here is that cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics from causing cracking if you don't pre-tin but, for Crissake, do your qualification tests first. | | Wave soldering presents no problems because the �g of Au per device gets lost in that 300 kg mass of molten solder, as it is pretinned and soldered in the one operation. Reflow is hell, because your �g stays in the mg of paste you laid down. | | My tuppence worth | | | Brian | Brian,

Though I agree with the big pot theory, I do question the solder medium you suggest and using wave soldering to attach leadless ceramic chip carriers or any other type leaded device with gold plating.

As you are aware, in the US military world, we required gold plating on many device types, especially LCCC's only to prevent solder termination oxidation. It has always been required, again in the military world, manufacturers remove gold before assembly. This is done using pure/controlled (constantly monitored contaminate levels not allowing them to exceed specified requirements) solder pots, or derivations thereof, to remove gold and apply tin/lead solder coatings in its place.

Also, I've never placed and soldered LCCC's using wave soldering. With these device types, very precise SMT solder pad configurations are required so solder paste deposition is made possible for reflow (started out in vapor phase type equipment). Additionally, as you are acutely aware, solder pad configuration is key to excessively mis-matched TCE's between ceramic and the substrate. Also, they are required even with constraining or tailoring core materials as CIC or Kevlar.

If all doesn't come together correctly, solder joint reliability suffers. Reasons include the obvious as gold embrittlement, as you very well know, and thermal mismatching.

Kind regards,

Earl Moon

reply »

| | | We have some SMT components with gold plated leads. They're actually LCC led packages. A good solder alloy to use would be Indium / Lead to reduce tin dissolution into the gold and avoid the resultant brittle intermetallic compound. However, a recommendation has also been made to use an alternative Sn62 Pb36 Ag2. The introduction of silver is said to reduce the leaching of tin. This alloy is practically a "drop in" and it is less expensive than an indium alternative. | | | Has anyone had experience with soldering to gold leads with this silver alloy? | | | Thanks, | | | DaveJ | | | | | Dave | | | | I like the Ag2 alloy, but not for that. It's a tad tougher than 63/37 without losing much on the ductility, and the joints are slightly more brilliant (i.e. a doubtful one stands out slightly more). The dissolution rate for Au in 63/37 is c. 5 �m/sec at about 230�C. In 62/36/2, it will be about 4.5 �m/sec and you'll get the same intermetallics. Pretinning is the answer. | | | | BTW, I hand-soldered tons (well, kgs) of 0.8 and 1.0 mm solid gold wire in sensors which my co. used to make, onto copper wire using 62/36/2. The results were horribly mat. The whole lot was cast into an epoxy mass: never once had a fracture in service and some of these were in service for 15 years under fairly nasty climatic conditions. What I'm suggesting here is that cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics from causing cracking if you don't pre-tin but, for Crissake, do your qualification tests first. | | | | Wave soldering presents no problems because the �g of Au per device gets lost in that 300 kg mass of molten solder, as it is pretinned and soldered in the one operation. Reflow is hell, because your �g stays in the mg of paste you laid down. | | | | My tuppence worth | | | | | | Brian | | | Brian, | | Though I agree with the big pot theory, I do question the solder medium you suggest and using wave soldering to attach leadless ceramic chip carriers or any other type leaded device with gold plating. | | As you are aware, in the US military world, we required gold plating on many device types, especially LCCC's only to prevent solder termination oxidation. It has always been required, again in the military world, manufacturers remove gold before assembly. This is done using pure/controlled (constantly monitored contaminate levels not allowing them to exceed specified requirements) solder pots, or derivations thereof, to remove gold and apply tin/lead solder coatings in its place. | | Also, I've never placed and soldered LCCC's using wave soldering. With these device types, very precise SMT solder pad configurations are required so solder paste deposition is made possible for reflow (started out in vapor phase type equipment). Additionally, as you are acutely aware, solder pad configuration is key to excessively mis-matched TCE's between ceramic and the substrate. Also, they are required even with constraining or tailoring core materials as CIC or Kevlar. | | If all doesn't come together correctly, solder joint reliability suffers. Reasons include the obvious as gold embrittlement, as you very well know, and thermal mismatching. | | Kind regards, | | Earl Moon | Sorry, Earl, I didn't make myself clear. I agree, wave soldering LCCCs and suchlike is not on the books: I was trying to state a generality. I remember, 20 years ago, I wave-soldered some preformed TO-18 package leads (gold plated Kovar) and we found next to no gold on microsections and wet analyses. However, we started getting returns from the tropics about 6-9 months later: the customers claimed fatal faults. The forming, nearer the device than the solder capillary rose, had fissured the gold plating and the kovar suffered from stress corrosion under the gold. In some cases, the transistors held on and were electrically connected just by the gold plating! A good tap and all the transistors fell to the ground! Cure: pretinning the transistors before forming right up to within 1-2 mm from the g-to-m seals. Moral of the story: never take ANYTHING for granted, not even what we tell you on this forum :-)

Brian

reply »

| | | | We have some SMT components with gold plated leads. They're actually LCC led packages. A good solder alloy to use would be Indium / Lead to reduce tin dissolution into the gold and avoid the resultant brittle intermetallic compound. However, a recommendation has also been made to use an alternative Sn62 Pb36 Ag2. The introduction of silver is said to reduce the leaching of tin. This alloy is practically a "drop in" and it is less expensive than an indium alternative. | | | | Has anyone had experience with soldering to gold leads with this silver alloy? | | | | Thanks, | | | | DaveJ | | | | | | | Dave | | | | | | I like the Ag2 alloy, but not for that. It's a tad tougher than 63/37 without losing much on the ductility, and the joints are slightly more brilliant (i.e. a doubtful one stands out slightly more). The dissolution rate for Au in 63/37 is c. 5 �m/sec at about 230�C. In 62/36/2, it will be about 4.5 �m/sec and you'll get the same intermetallics. Pretinning is the answer. | | | | | | BTW, I hand-soldered tons (well, kgs) of 0.8 and 1.0 mm solid gold wire in sensors which my co. used to make, onto copper wire using 62/36/2. The results were horribly mat. The whole lot was cast into an epoxy mass: never once had a fracture in service and some of these were in service for 15 years under fairly nasty climatic conditions. What I'm suggesting here is that cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics from causing cracking if you don't pre-tin but, for Crissake, do your qualification tests first. | | | | | | Wave soldering presents no problems because the �g of Au per device gets lost in that 300 kg mass of molten solder, as it is pretinned and soldered in the one operation. Reflow is hell, because your �g stays in the mg of paste you laid down. | | | | | | My tuppence worth | | | | | | | | | Brian | | | | | Brian, | | | | Though I agree with the big pot theory, I do question the solder medium you suggest and using wave soldering to attach leadless ceramic chip carriers or any other type leaded device with gold plating. | | | | As you are aware, in the US military world, we required gold plating on many device types, especially LCCC's only to prevent solder termination oxidation. It has always been required, again in the military world, manufacturers remove gold before assembly. This is done using pure/controlled (constantly monitored contaminate levels not allowing them to exceed specified requirements) solder pots, or derivations thereof, to remove gold and apply tin/lead solder coatings in its place. | | | | Also, I've never placed and soldered LCCC's using wave soldering. With these device types, very precise SMT solder pad configurations are required so solder paste deposition is made possible for reflow (started out in vapor phase type equipment). Additionally, as you are acutely aware, solder pad configuration is key to excessively mis-matched TCE's between ceramic and the substrate. Also, they are required even with constraining or tailoring core materials as CIC or Kevlar. | | | | If all doesn't come together correctly, solder joint reliability suffers. Reasons include the obvious as gold embrittlement, as you very well know, and thermal mismatching. | | | | Kind regards, | | | | Earl Moon | | | Sorry, Earl, I didn't make myself clear. I agree, wave soldering LCCCs and suchlike is not on the books: I was trying to state a generality. I remember, 20 years ago, I wave-soldered some preformed TO-18 package leads (gold plated Kovar) and we found next to no gold on microsections and wet analyses. However, we started getting returns from the tropics about 6-9 months later: the customers claimed fatal faults. The forming, nearer the device than the solder capillary rose, had fissured the gold plating and the kovar suffered from stress corrosion under the gold. In some cases, the transistors held on and were electrically connected just by the gold plating! A good tap and all the transistors fell to the ground! Cure: pretinning the transistors before forming right up to within 1-2 mm from the g-to-m seals. Moral of the story: never take ANYTHING for granted, not even what we tell you on this forum :-) | | Brian | Thank you for the clarification and story. Very informative and strikes some similar memories.

Earl

reply »

| | We have some SMT components with gold plated leads. They're actually LCC led packages. A good solder alloy to use would be Indium / Lead to reduce tin dissolution into the gold and avoid the resultant brittle intermetallic compound. However, a recommendation has also been made to use an alternative Sn62 Pb36 Ag2. The introduction of silver is said to reduce the leaching of tin. This alloy is practically a "drop in" and it is less expensive than an indium alternative. | | Has anyone had experience with soldering to gold leads with this silver alloy? | | Thanks, | | DaveJ | | | Dave | | I like the Ag2 alloy, but not for that. It's a tad tougher than 63/37 without losing much on the ductility, and the joints are slightly more brilliant (i.e. a doubtful one stands out slightly more). The dissolution rate for Au in 63/37 is c. 5 �m/sec at about 230�C. In 62/36/2, it will be about 4.5 �m/sec and you'll get the same intermetallics. Pretinning is the answer. | | BTW, I hand-soldered tons (well, kgs) of 0.8 and 1.0 mm solid gold wire in sensors which my co. used to make, onto copper wire using 62/36/2. The results were horribly mat. The whole lot was cast into an epoxy mass: never once had a fracture in service and some of these were in service for 15 years under fairly nasty climatic conditions. What I'm suggesting here is that cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics from causing cracking if you don't pre-tin but, for Crissake, do your qualification tests first. | | Wave soldering presents no problems because the �g of Au per device gets lost in that 300 kg mass of molten solder, as it is pretinned and soldered in the one operation. Reflow is hell, because your �g stays in the mg of paste you laid down. | | My tuppence worth | | | Brian | Brian:

I like Ag2 also. I prefer it over both 63/37 and 60/40, if the price is not a factor. Yes, gold rips through eutectic and near-eutectic solders like habanero slathered shiitake burritos through ...

I bought that, but I�m not buying ...

Intermetallics. I�m not buying "cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics." First, the intermetallics were formed "long" before the connection was cleaned or coated. They are formed during the soldering process and continue to form as long as the connection is warmer than 0�K. You don�t prevent intermetalics in soldering, you try to minimize them. And mechanical support has nothing to with it!!!

Wave Solder Pot Exemption. I don�t buy the �It�s OK to put gold in the solder pot, because it�s so big, you�ll never notice it.� (My interpretation) (So, is this analogous? ... It�s OK to dump nuclear wastes into the ocean, because those Siberian Sea plankton will never make it down to the Great Northwest and get eaten by my salmon. But it digress.) I assume (probably a mistake) adding gold to my solder pot is acceptable, because I�ll notice the gold reaching dangerous levels in my monthly solder pot analysis. While that�s probably true, but:

� What about marginal soldering situations that occur before changing-out the solder pot? Because, (1) I�m not sure of the safety margin of gold concentration that�s acceptable before I should clean the pot (versus when I must clean the pot), (2) things that happen in that month between when the pot was acceptable and the next analysis when it becomes unacceptable, and (3) trust me, I will put-off changing the solder in the pot as long I can, because ... � I just never have enjoyed changing-out the solder in wave solder pots. It�s too expensive in terms of danger to employees and lost time and materials.

Now changing-out the solder in a bench top solder pot, that�s something I can handle.

Taking a different tack on the "Wave Solder Pot Exemption:" Or is it OK to add gold to the solder pot, because the AuSn4 floats and ends-up getting tossed with the dross? (This was not the implication of your statement.)

My2�

Dave F

reply »

Dave, I'll respond by paragraph: | | | Brian: | | I like Ag2 also. I prefer it over both 63/37 and 60/40, if the price is not a factor. Yes, gold rips through eutectic and near-eutectic solders like habanero slathered shiitake burritos through ... | | I bought that, but I�m not buying ... | | Intermetallics. I�m not buying "cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics." First, the intermetallics were formed "long" before the connection was cleaned or coated. They are formed during the soldering process and continue to form as long as the connection is warmer than 0�K. You don�t prevent intermetalics in soldering, you try to minimize them. And mechanical support has nothing to with it!!!

In my case, which I cited, my joints were filled with intermetallics, but they never seemed to crack. I assume this is because they were supported mechanically by the epoxy resin. In other words, the mechanical strength is organic and the electrical contact was maintained, despite the gold-rich tin. I therefore hypothesised, but did not advocate, that conformal coating MAY have a similar effect. In no way did I wish to suggest that the intermetallics would miraculously disappear by cleaning and coating, which I infer is what you inferred. If you like, this speculation is rather like the proven method of using NON-conductive underfill adhesives for ensuring the electrical connection of BGAs, in that the resins provide the mechanical support that maintain the contact. And I did put a big caveat at the end of the paragraph. | | Wave Solder Pot Exemption. I don�t buy the �It�s OK to put gold in the solder pot, because it�s so big, you�ll never notice it.� (My interpretation) (So, is this analogous? ... It�s OK to dump nuclear wastes into the ocean, because those Siberian Sea plankton will never make it down to the Great Northwest and get eaten by my salmon. But it digress.) I assume (probably a mistake) adding gold to my solder pot is acceptable, because I�ll notice the gold reaching dangerous levels in my monthly solder pot analysis. While that�s probably true, but: | | � What about marginal soldering situations that occur before changing-out the solder pot? Because, (1) I�m not sure of the safety margin of gold concentration that�s acceptable before I should clean the pot (versus when I must clean the pot), (2) things that happen in that month between when the pot was acceptable and the next analysis when it becomes unacceptable, and (3) trust me, I will put-off changing the solder in the pot as long I can, because ... | � I just never have enjoyed changing-out the solder in wave solder pots. It�s too expensive in terms of danger to employees and lost time and materials.

If you have a big pot (say, 250 kg or more), gold levels will never rise to a dangerous level from just gold plated leads (if the PCBs are gold plated, as well, that may be a horse of a different colour). After a few weeks, the gold content will reach an asymptote which will be maintained at � the same level because the solder replenishment due to solder use, dross removal, etc. will provide enough tin to compensate for new gold being added. In the good old days when all semiconductors were in TO-18 and TO-5 cans, with gold plated leads (and some other components), I ran a 63/37 wave for three years before changing the solder (and that was not because of gold, but zinc because one supplier of mechanical components had, unbeknownst to us, changed from tinned bronze to tinned brass). That machine was soldering about 200 TO-x components per minute, day in and day out on an 8 hour/4.5 day basis. I would agree, that if the pot contains only a few kg of solder, like some jet wave machines, the gold-content asymptote will be much higher, possibly reaching dangerous levels quite rapidly.

In any case, there is no cut and defined threshold where gold suddenly becomes dangerous. It is a gradual process, which explains why different standards permit different levels. If you saw, on your monthly analysis, that you were approaching the arbitrary value that your standard (or you) decided was enough, your solder joints aren't suddenly going to start cracking because there was an extra 0.001% during the next month. The keyword is arbitrary and the standards all have a considerable safety margin built in (at least a factor of 5). | | Now changing-out the solder in a bench top solder pot, that�s something I can handle.

When? Do you do analyses? If so, your sample will half-empty the pot and the replenishment will reduce the level, anyway, so that the sample is meaningless. After n leads have passed through it? Who does the counting? Every day, week or month? Back to the arbitrary. | | Taking a different tack on the "Wave Solder Pot Exemption:" Or is it OK to add gold to the solder pot, because the AuSn4 floats and ends-up getting tossed with the dross? (This was not the implication of your statement.)

As you say, I did not imply this, but I don't believe that skimming solder pots is a very clever way of controlling impurity levels. However, the inadvertent removal of gold with the dross will enter into the asymptote equation, simply because the volume removed will be replaced by fresh solder. Again, dredging my memory back to the early days of wave soldering (c. 1954/5), when the whole lot was an art, rather than a science, Rolf Strauss (the inventor) suggested we keep copper levels down by cooling the bath to c. 190�C (i.e. below tin/copper intermetallic MPs) and removing the floating sludge, topping up the level with pure tin. He also provided a can of what I believe was sawdust laced with a mixture of chemicals which he told us to try mixing into the pot and then skimming off. I'm sure that did a lot more harm than good :-( | | My2� | | Dave F | In summary, Dave, despite what the pundits would have us believe, a lot of soldering "technology" is still a black art and not a science -- and I put the word technology in inverted commas because it is closer to a technique than a true technology. I say this because it is exactly 50 years ago to the month that, as a callow student, I took an optional undergraduate semester course on Soldering Technology: little did I know, at the time, that it would shape my career.

Keep smiling

Brian

reply »

| | | Dave, I'll respond by paragraph: | | | | | Brian: | | | | I like Ag2 also. I prefer it over both 63/37 and 60/40, if the price is not a factor. Yes, gold rips through eutectic and near-eutectic solders like habanero slathered shiitake burritos through ... | | | | I bought that, but I�m not buying ... | | | | Intermetallics. I�m not buying "cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics." First, the intermetallics were formed "long" before the connection was cleaned or coated. They are formed during the soldering process and continue to form as long as the connection is warmer than 0�K. You don�t prevent intermetalics in soldering, you try to minimize them. And mechanical support has nothing to with it!!! | | In my case, which I cited, my joints were filled with intermetallics, but they never seemed to crack. I assume this is because they were supported mechanically by the epoxy resin. In other words, the mechanical strength is organic and the electrical contact was maintained, despite the gold-rich tin. I therefore hypothesised, but did not advocate, that conformal coating MAY have a similar effect. In no way did I wish to suggest that the intermetallics would miraculously disappear by cleaning and coating, which I infer is what you inferred. If you like, this speculation is rather like the proven method of using NON-conductive underfill adhesives for ensuring the electrical connection of BGAs, in that the resins provide the mechanical support that maintain the contact. And I did put a big caveat at the end of the paragraph. | | | | Wave Solder Pot Exemption. I don�t buy the �It�s OK to put gold in the solder pot, because it�s so big, you�ll never notice it.� (My interpretation) (So, is this analogous? ... It�s OK to dump nuclear wastes into the ocean, because those Siberian Sea plankton will never make it down to the Great Northwest and get eaten by my salmon. But it digress.) I assume (probably a mistake) adding gold to my solder pot is acceptable, because I�ll notice the gold reaching dangerous levels in my monthly solder pot analysis. While that�s probably true, but: | | | | � What about marginal soldering situations that occur before changing-out the solder pot? Because, (1) I�m not sure of the safety margin of gold concentration that�s acceptable before I should clean the pot (versus when I must clean the pot), (2) things that happen in that month between when the pot was acceptable and the next analysis when it becomes unacceptable, and (3) trust me, I will put-off changing the solder in the pot as long I can, because ... | | � I just never have enjoyed changing-out the solder in wave solder pots. It�s too expensive in terms of danger to employees and lost time and materials. | | If you have a big pot (say, 250 kg or more), gold levels will never rise to a dangerous level from just gold plated leads (if the PCBs are gold plated, as well, that may be a horse of a different colour). After a few weeks, the gold content will reach an asymptote which will be maintained at � the same level because the solder replenishment due to solder use, dross removal, etc. will provide enough tin to compensate for new gold being added. In the good old days when all semiconductors were in TO-18 and TO-5 cans, with gold plated leads (and some other components), I ran a 63/37 wave for three years before changing the solder (and that was not because of gold, but zinc because one supplier of mechanical components had, unbeknownst to us, changed from tinned bronze to tinned brass). That machine was soldering about 200 TO-x components per minute, day in and day out on an 8 hour/4.5 day basis. I would agree, that if the pot contains only a few kg of solder, like some jet wave machines, the gold-content asymptote will be much higher, possibly reaching dangerous levels quite rapidly. | | In any case, there is no cut and defined threshold where gold suddenly becomes dangerous. It is a gradual process, which explains why different standards permit different levels. If you saw, on your monthly analysis, that you were approaching the arbitrary value that your standard (or you) decided was enough, your solder joints aren't suddenly going to start cracking because there was an extra 0.001% during the next month. The keyword is arbitrary and the standards all have a considerable safety margin built in (at least a factor of 5). | | | | Now changing-out the solder in a bench top solder pot, that�s something I can handle. | | When? Do you do analyses? If so, your sample will half-empty the pot and the replenishment will reduce the level, anyway, so that the sample is meaningless. After n leads have passed through it? Who does the counting? Every day, week or month? Back to the arbitrary. | | | | Taking a different tack on the "Wave Solder Pot Exemption:" Or is it OK to add gold to the solder pot, because the AuSn4 floats and ends-up getting tossed with the dross? (This was not the implication of your statement.) | | As you say, I did not imply this, but I don't believe that skimming solder pots is a very clever way of controlling impurity levels. However, the inadvertent removal of gold with the dross will enter into the asymptote equation, simply because the volume removed will be replaced by fresh solder. Again, dredging my memory back to the early days of wave soldering (c. 1954/5), when the whole lot was an art, rather than a science, Rolf Strauss (the inventor) suggested we keep copper levels down by cooling the bath to c. 190�C (i.e. below tin/copper intermetallic MPs) and removing the floating sludge, topping up the level with pure tin. He also provided a can of what I believe was sawdust laced with a mixture of chemicals which he told us to try mixing into the pot and then skimming off. I'm sure that did a lot more harm than good :-( | | | | My2� | | | | Dave F | | | In summary, Dave, despite what the pundits would have us believe, a lot of soldering "technology" is still a black art and not a science -- and I put the word technology in inverted commas because it is closer to a technique than a true technology. I say this because it is exactly 50 years ago to the month that, as a callow student, I took an optional undergraduate semester course on Soldering Technology: little did I know, at the time, that it would shape my career. | | Keep smiling | | Brian | Brian,

You're the best. The question is: Shape or deform? I think we all could ask that.

I do have a question. I was a staunch advocate of immersion gold over immersion nickel, over copper. I'm just not so convinced anymore. I'm seeing more "dark" pad phenomonon as well as phosphorous, not gold, embrittlement. Could you comment and discuss some of the factors invloved in both?

Also, being an old board fabricator, I know it's hard to control the phosphorous content in the immersion nickel, or any other chemical solution, so I continue being concerned about its affect on embrittlement - again, not just the minute gold amounts.

And on it goes, though a bit disjointed today (Bud day in Chippewa Falls, WI - most others partake of Linenkugel)

Earl Moon

reply »

| | | | | | Dave, I'll respond by paragraph: | | | | | | | Brian: | | | | | | I like Ag2 also. I prefer it over both 63/37 and 60/40, if the price is not a factor. Yes, gold rips through eutectic and near-eutectic solders like habanero slathered shiitake burritos through ... | | | | | | I bought that, but I�m not buying ... | | | | | | Intermetallics. I�m not buying "cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics." First, the intermetallics were formed "long" before the connection was cleaned or coated. They are formed during the soldering process and continue to form as long as the connection is warmer than 0�K. You don�t prevent intermetalics in soldering, you try to minimize them. And mechanical support has nothing to with it!!! | | | | In my case, which I cited, my joints were filled with intermetallics, but they never seemed to crack. I assume this is because they were supported mechanically by the epoxy resin. In other words, the mechanical strength is organic and the electrical contact was maintained, despite the gold-rich tin. I therefore hypothesised, but did not advocate, that conformal coating MAY have a similar effect. In no way did I wish to suggest that the intermetallics would miraculously disappear by cleaning and coating, which I infer is what you inferred. If you like, this speculation is rather like the proven method of using NON-conductive underfill adhesives for ensuring the electrical connection of BGAs, in that the resins provide the mechanical support that maintain the contact. And I did put a big caveat at the end of the paragraph. | | | | | | Wave Solder Pot Exemption. I don�t buy the �It�s OK to put gold in the solder pot, because it�s so big, you�ll never notice it.� (My interpretation) (So, is this analogous? ... It�s OK to dump nuclear wastes into the ocean, because those Siberian Sea plankton will never make it down to the Great Northwest and get eaten by my salmon. But it digress.) I assume (probably a mistake) adding gold to my solder pot is acceptable, because I�ll notice the gold reaching dangerous levels in my monthly solder pot analysis. While that�s probably true, but: | | | | | | � What about marginal soldering situations that occur before changing-out the solder pot? Because, (1) I�m not sure of the safety margin of gold concentration that�s acceptable before I should clean the pot (versus when I must clean the pot), (2) things that happen in that month between when the pot was acceptable and the next analysis when it becomes unacceptable, and (3) trust me, I will put-off changing the solder in the pot as long I can, because ... | | | � I just never have enjoyed changing-out the solder in wave solder pots. It�s too expensive in terms of danger to employees and lost time and materials. | | | | If you have a big pot (say, 250 kg or more), gold levels will never rise to a dangerous level from just gold plated leads (if the PCBs are gold plated, as well, that may be a horse of a different colour). After a few weeks, the gold content will reach an asymptote which will be maintained at � the same level because the solder replenishment due to solder use, dross removal, etc. will provide enough tin to compensate for new gold being added. In the good old days when all semiconductors were in TO-18 and TO-5 cans, with gold plated leads (and some other components), I ran a 63/37 wave for three years before changing the solder (and that was not because of gold, but zinc because one supplier of mechanical components had, unbeknownst to us, changed from tinned bronze to tinned brass). That machine was soldering about 200 TO-x components per minute, day in and day out on an 8 hour/4.5 day basis. I would agree, that if the pot contains only a few kg of solder, like some jet wave machines, the gold-content asymptote will be much higher, possibly reaching dangerous levels quite rapidly. | | | | In any case, there is no cut and defined threshold where gold suddenly becomes dangerous. It is a gradual process, which explains why different standards permit different levels. If you saw, on your monthly analysis, that you were approaching the arbitrary value that your standard (or you) decided was enough, your solder joints aren't suddenly going to start cracking because there was an extra 0.001% during the next month. The keyword is arbitrary and the standards all have a considerable safety margin built in (at least a factor of 5). | | | | | | Now changing-out the solder in a bench top solder pot, that�s something I can handle. | | | | When? Do you do analyses? If so, your sample will half-empty the pot and the replenishment will reduce the level, anyway, so that the sample is meaningless. After n leads have passed through it? Who does the counting? Every day, week or month? Back to the arbitrary. | | | | | | Taking a different tack on the "Wave Solder Pot Exemption:" Or is it OK to add gold to the solder pot, because the AuSn4 floats and ends-up getting tossed with the dross? (This was not the implication of your statement.) | | | | As you say, I did not imply this, but I don't believe that skimming solder pots is a very clever way of controlling impurity levels. However, the inadvertent removal of gold with the dross will enter into the asymptote equation, simply because the volume removed will be replaced by fresh solder. Again, dredging my memory back to the early days of wave soldering (c. 1954/5), when the whole lot was an art, rather than a science, Rolf Strauss (the inventor) suggested we keep copper levels down by cooling the bath to c. 190�C (i.e. below tin/copper intermetallic MPs) and removing the floating sludge, topping up the level with pure tin. He also provided a can of what I believe was sawdust laced with a mixture of chemicals which he told us to try mixing into the pot and then skimming off. I'm sure that did a lot more harm than good :-( | | | | | | My2� | | | | | | Dave F | | | | | In summary, Dave, despite what the pundits would have us believe, a lot of soldering "technology" is still a black art and not a science -- and I put the word technology in inverted commas because it is closer to a technique than a true technology. I say this because it is exactly 50 years ago to the month that, as a callow student, I took an optional undergraduate semester course on Soldering Technology: little did I know, at the time, that it would shape my career. | | | | Keep smiling | | | | Brian | | | Brian, | | You're the best. The question is: Shape or deform? I think we all could ask that. | | I do have a question. I was a staunch advocate of immersion gold over immersion nickel, over copper. I'm just not so convinced anymore. I'm seeing more "dark" pad phenomonon as well as phosphorous, not gold, embrittlement. Could you comment and discuss some of the factors invloved in both? | | Also, being an old board fabricator, I know it's hard to control the phosphorous content in the immersion nickel, or any other chemical solution, so I continue being concerned about its affect on embrittlement - again, not just the minute gold amounts. | | And on it goes, though a bit disjointed today (Bud day in Chippewa Falls, WI - most others partake of Linenkugel) | | Earl Moon | Earl

I'm not sure I'm really qualified to say much about immersion plating. I have done, many years ago, immersion gold on copper and immersion tin on copper and on electrolytic tin/lead (to cover the edges after etching, this being before immersion or IR reflowing was thought of). I'm darn sure that modern baths must be better. Certainly, immersion gold was, at that time, a no go as far as solderability is concerned. An edge dip test showed 50% non-wetting and 50% dewetting :-0

Anyway, my concerns: - nickel is always dicy, no matter what the type as it is hell to solder - if the gold is non-porous (hence not allowing the nickel to oxidise) then it is thick and expensive - as you say, phosphorus is hell for soldering (renders the solder mobility sluggish) and, if wave soldering, poisons the solder bath at <0.001% with no cure except to phone up your supplier and get him to ship n kg double fast.

That having been said, I repeat that I am not up to date on the details of modern immersion baths (I last made PCBs professionally in 1974, before starting up Protonique) so my fears may be unfounded.

BTW, for heavy duty edge connectors, I like 5 �m Ni + 2-3 �m hard Au on the fingers (electrolytic, of course).

You know the best finish we did in the 60s and 70s? Lonco Sealbrite. We gave a 1 year solderability guarantee on Sealbrite'd PCBs (6 months on all other finishes), including multilayers and never once had a return under guarantee (We had a few coming back after 3-4 years for re-treatment to restore solderability but, hell, do you want the earth?)

Soory can't help you more.

Brian

reply »

| | | | | | | | | Dave, I'll respond by paragraph: | | | | | | | | | Brian: | | | | | | | | I like Ag2 also. I prefer it over both 63/37 and 60/40, if the price is not a factor. Yes, gold rips through eutectic and near-eutectic solders like habanero slathered shiitake burritos through ... | | | | | | | | I bought that, but I�m not buying ... | | | | | | | | Intermetallics. I�m not buying "cleaning + conformal coating MAY provide sufficient mechanical support to prevent them 'orrible intermetallics." First, the intermetallics were formed "long" before the connection was cleaned or coated. They are formed during the soldering process and continue to form as long as the connection is warmer than 0�K. You don�t prevent intermetalics in soldering, you try to minimize them. And mechanical support has nothing to with it!!! | | | | | | In my case, which I cited, my joints were filled with intermetallics, but they never seemed to crack. I assume this is because they were supported mechanically by the epoxy resin. In other words, the mechanical strength is organic and the electrical contact was maintained, despite the gold-rich tin. I therefore hypothesised, but did not advocate, that conformal coating MAY have a similar effect. In no way did I wish to suggest that the intermetallics would miraculously disappear by cleaning and coating, which I infer is what you inferred. If you like, this speculation is rather like the proven method of using NON-conductive underfill adhesives for ensuring the electrical connection of BGAs, in that the resins provide the mechanical support that maintain the contact. And I did put a big caveat at the end of the paragraph. | | | | | | | | Wave Solder Pot Exemption. I don�t buy the �It�s OK to put gold in the solder pot, because it�s so big, you�ll never notice it.� (My interpretation) (So, is this analogous? ... It�s OK to dump nuclear wastes into the ocean, because those Siberian Sea plankton will never make it down to the Great Northwest and get eaten by my salmon. But it digress.) I assume (probably a mistake) adding gold to my solder pot is acceptable, because I�ll notice the gold reaching dangerous levels in my monthly solder pot analysis. While that�s probably true, but: | | | | | | | | � What about marginal soldering situations that occur before changing-out the solder pot? Because, (1) I�m not sure of the safety margin of gold concentration that�s acceptable before I should clean the pot (versus when I must clean the pot), (2) things that happen in that month between when the pot was acceptable and the next analysis when it becomes unacceptable, and (3) trust me, I will put-off changing the solder in the pot as long I can, because ... | | | | � I just never have enjoyed changing-out the solder in wave solder pots. It�s too expensive in terms of danger to employees and lost time and materials. | | | | | | If you have a big pot (say, 250 kg or more), gold levels will never rise to a dangerous level from just gold plated leads (if the PCBs are gold plated, as well, that may be a horse of a different colour). After a few weeks, the gold content will reach an asymptote which will be maintained at � the same level because the solder replenishment due to solder use, dross removal, etc. will provide enough tin to compensate for new gold being added. In the good old days when all semiconductors were in TO-18 and TO-5 cans, with gold plated leads (and some other components), I ran a 63/37 wave for three years before changing the solder (and that was not because of gold, but zinc because one supplier of mechanical components had, unbeknownst to us, changed from tinned bronze to tinned brass). That machine was soldering about 200 TO-x components per minute, day in and day out on an 8 hour/4.5 day basis. I would agree, that if the pot contains only a few kg of solder, like some jet wave machines, the gold-content asymptote will be much higher, possibly reaching dangerous levels quite rapidly. | | | | | | In any case, there is no cut and defined threshold where gold suddenly becomes dangerous. It is a gradual process, which explains why different standards permit different levels. If you saw, on your monthly analysis, that you were approaching the arbitrary value that your standard (or you) decided was enough, your solder joints aren't suddenly going to start cracking because there was an extra 0.001% during the next month. The keyword is arbitrary and the standards all have a considerable safety margin built in (at least a factor of 5). | | | | | | | | Now changing-out the solder in a bench top solder pot, that�s something I can handle. | | | | | | When? Do you do analyses? If so, your sample will half-empty the pot and the replenishment will reduce the level, anyway, so that the sample is meaningless. After n leads have passed through it? Who does the counting? Every day, week or month? Back to the arbitrary. | | | | | | | | Taking a different tack on the "Wave Solder Pot Exemption:" Or is it OK to add gold to the solder pot, because the AuSn4 floats and ends-up getting tossed with the dross? (This was not the implication of your statement.) | | | | | | As you say, I did not imply this, but I don't believe that skimming solder pots is a very clever way of controlling impurity levels. However, the inadvertent removal of gold with the dross will enter into the asymptote equation, simply because the volume removed will be replaced by fresh solder. Again, dredging my memory back to the early days of wave soldering (c. 1954/5), when the whole lot was an art, rather than a science, Rolf Strauss (the inventor) suggested we keep copper levels down by cooling the bath to c. 190�C (i.e. below tin/copper intermetallic MPs) and removing the floating sludge, topping up the level with pure tin. He also provided a can of what I believe was sawdust laced with a mixture of chemicals which he told us to try mixing into the pot and then skimming off. I'm sure that did a lot more harm than good :-( | | | | | | | | My2� | | | | | | | | Dave F | | | | | | | In summary, Dave, despite what the pundits would have us believe, a lot of soldering "technology" is still a black art and not a science -- and I put the word technology in inverted commas because it is closer to a technique than a true technology. I say this because it is exactly 50 years ago to the month that, as a callow student, I took an optional undergraduate semester course on Soldering Technology: little did I know, at the time, that it would shape my career. | | | | | | Keep smiling | | | | | | Brian | | | | | Brian, | | | | You're the best. The question is: Shape or deform? I think we all could ask that. | | | | I do have a question. I was a staunch advocate of immersion gold over immersion nickel, over copper. I'm just not so convinced anymore. I'm seeing more "dark" pad phenomonon as well as phosphorous, not gold, embrittlement. Could you comment and discuss some of the factors invloved in both? | | | | Also, being an old board fabricator, I know it's hard to control the phosphorous content in the immersion nickel, or any other chemical solution, so I continue being concerned about its affect on embrittlement - again, not just the minute gold amounts. | | | | And on it goes, though a bit disjointed today (Bud day in Chippewa Falls, WI - most others partake of Linenkugel) | | | | Earl Moon | | | Earl | | I'm not sure I'm really qualified to say much about immersion plating. I have done, many years ago, immersion gold on copper and immersion tin on copper and on electrolytic tin/lead (to cover the edges after etching, this being before immersion or IR reflowing was thought of). I'm darn sure that modern baths must be better. Certainly, immersion gold was, at that time, a no go as far as solderability is concerned. An edge dip test showed 50% non-wetting and 50% dewetting :-0 | | Anyway, my concerns: | - nickel is always dicy, no matter what the type as it is hell to solder | - if the gold is non-porous (hence not allowing the nickel to oxidise) then it is thick and expensive | - as you say, phosphorus is hell for soldering (renders the solder mobility sluggish) and, if wave soldering, poisons the solder bath at <0.001% with no cure except to phone up your supplier and get him to ship n kg double fast. | | That having been said, I repeat that I am not up to date on the details of modern immersion baths (I last made PCBs professionally in 1974, before starting up Protonique) so my fears may be unfounded. | | BTW, for heavy duty edge connectors, I like 5 �m Ni + 2-3 �m hard Au on the fingers (electrolytic, of course). | | You know the best finish we did in the 60s and 70s? Lonco Sealbrite. We gave a 1 year solderability guarantee on Sealbrite'd PCBs (6 months on all other finishes), including multilayers and never once had a return under guarantee (We had a few coming back after 3-4 years for re-treatment to restore solderability but, hell, do you want the earth?) | | Soory can't help you more. | | Brian | Brian,

You done good. I too came through the Sealbrite school. I also went through the H-P school of electroplated gold over electroplated nickel over copper. Fun times?

I'm just concerned about so many fine "rookies" and professionals professing the virtues of electroless gold over nickel over copper. The phosphorous thing continues as great material for debate, as well it should but for boron, etc.

I see so many folks decrying phosphorous as the lone culprit in the "dark" pad controversy. It's not that simple, and too many dismiss it as being another can of worms when it is that plus so much more.

Thanks again,

Earl

reply »