So I have this bunch of very small InGaAs photodiodes (60-micron diameter) that are wire bonded to a ceramic carrier with 100 microinches of gold over 75 microinches of nickel.
Problem is, the entire pad dissolves in 60/40 solder. Good thing I had spares. :(
How do you solder to these things? Is 2% silver solder good enough?
Thanks
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
On Thu, 30 Jul 2015 13:41:28 -0400, Phil Hobbs Gave us:
60/40 solder is typically what is used for plumbing. 63/37 has been the industry standard (before the RoHS debacle), and still is. With yours, you could have a corrosive, acidic flux in there too.I would use a very tiny tip and a very tiny amount of 63/37, and apply the droplet with any wire at all with an application lasting less than a second, then using a fine stranded SPC wire make another short cycle heating reflow with the wire on top (heat through the wire with the fine point iron) and as it falls into place, halt the heat sources connection, again in a task lasting a mere second at most.
If delamination is still taking place, I would register a complaint to the device maker as well as the design for this type of 'pad' is not supposed to delam that easily, much less dissolve.
On Thu, 30 Jul 2015 13:41:28 -0400, Phil Hobbs Gave us:
The lead free RoHS solders require even higher temperatures to reflow, and would very likely make your particular current problem worse, not due to the alloy, but due to the poorly manufactured pads and higher temps delivered to them.
Can you use silver bearing conductive epoxy?
On Thu, 30 Jul 2015 14:31:53 -0400, "Tom Miller" Gave us:
That would also work.
The chip bonding industry as well as many mil applications use this:
On Thu, 30 Jul 2015 14:30:16 -0400, DecadentLinuxUserNumeroUno Gave us:
offset thermal shock down, which may be a contributing factor.
May have to. I've asked the mfg but they're a small outfit and it's the summer, so I may not hear back very soon.
Usually the gold just dissolves in the solder, leaving the actual joint as solder-nickel.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
On Thu, 30 Jul 2015 14:41:58 -0400, Phil Hobbs Gave us:
That is OK too, as long as the wetting (bonding) is full and complete.
Well, I got some Sn62 Pb36 Ag2 from eBay, but for the near term, four-nines indium seems to have worked OK.
It's a drag destroying a $65 photodiode in such a stupid way.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
That's embarrassing. :^)
I soldered copper and thin gold wires together, by heating the copper and letting the solder "sneak up" on the gold. So maybe tin a piece of wire and then try heating mostly from the wire end... limiting the amount of solder the gold will see??? (Whole thing on a hot plate??) How many do you have to play with?
Or some other "goofy" low temperature solder... indium?
George H.
On Thu, 30 Jul 2015 15:08:28 -0400, Phil Hobbs Gave us:
OMG!
Yeah. I recall seeing a workable method in a book on soldering.
But I would first ask the photodiode maker for advice.
Indium will likely work long term, if I recall that book correctly.
Joe Gwinn
For soldering something like this you will need a special solder. The most common one is In50/Pb50. Don't even think of using something containing either tin or silver. In/Pb alloys are _SPECIALLY_ designed for soldering everything gold plated.
BTW, contrary to popular belief all those nice gold plated pads and lids are _NOT_ better for soldering with regular solder. Gold dissolves in that solder in a blink of an eye and forms brittle intermetallics that are prone to cracking. If you _HAVE_ to solder something cold plated use In/Pb solder.
-- ****************************************************************** * KSI@home KOI8 Net < > The impossible we do immediately. *
But it isn't gold plated, it's 2.5 microns gold over 2 microns nickel. Usually the gold dissolves but the nickel doesn't, so you wind up with a nice solder joint to the nickel.
I used 99.99% indium, because I had that in stock. Looks okay.
The intermetallics are no big problem if the solder volume is high enough. These pads are pretty small.
There's still a bit of pad left on the dead photodiode, so I'll experiment when the SN62 36Pb 2Ag comes in. That's supposed to be good for soldering noble metals.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
It is :) That 2.5 microns of gold _IS_ the plating... It might be not _ELECTRO_ plating but it doesn't matter.
It gets brittle where you hit gold with your solder. So no matter how big is the volume it cracks at the pad and you get a big blob of solder barely touching the pad. Then it oxidizes (sulfurizes, whatever) and contact gets lost. You end up with a nice looking solder joint without electrical contact with the pad. It is very difficult to spot even under a microscope.
That is _NOT_ . It is better that regular Sn63/Pb37 for soldering _SILVER_ but it is a total disaster on gold. Nickel is not very easy to solder if you expect gold to dissolve so you would be soldering to nickel.
As a matter of fact one needs many different solders for different purposes. I do have something like 20 different alloys for different purposes :) Some of them I make myself e.g. almost unobtanium Cd70/Sn30 that is irreplaceable for low thermal EMF joints.
-- ****************************************************************** * KSI@home KOI8 Net < > The impossible we do immediately. *
All of our PC boards are now ENIG, electroless nickel and immersion gold over copper. The gold is something silly like 5 or 10 uinches, so it dissolves in solder at a 1000:1 or so ratio. It solders amazingly... a little accidental solder splash is permanent. We haven't seen any adhesion problems; a pad will rip off the board before the metals separate.
Phil's problem is probably that he has a ceramic substrate, and not copper. On a PCB, both the gold and the nickel can dissolve, and then you still have a nice solder-copper joint.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement
ENIG is thinner than this gold film, but you're right, you need some reasonable amount of gold in the solder to make the nasty brittle intermetallics that Sergey's talking about (which are very real if there's enough gold floating around).
It was the nickel dissolving that surprised me.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
Exactly. When you solder to _COPPER_ with ENiG it is just a thin cover over copper (immersion gold is literally several atoms thick) it is totally different thing. That is if that ENiG is fresh. If you let it stay for a month or so before soldering you will get all kind of issues. I had BGA balls hanging in the air barely touching the pads without electrical contact on such boards and even bigger parts like 1206 sometimes had solder balled without zero pad wetting. And once the pad failed to wet it is extremely difficult to solder -- there is no plating on it any more and it is oxidized beyond solderability. The only way around it with bigger parts is removing the part and then _MECHANICALLY_ removing the oxide (or whatever that gunk is) layer. That is why I use _LEADED_ HASL whenever possible.
In case of ceramic it is a russian roulette to solder it with wrong solder. Regular solder just dissolves that pad in a blink of an eye and there is nothing left to solder to at all.
BTW, depending on the metal (pure copper is better, bronze/brass is a total disaster) some of plumbing lead-free solders can easily etch up to 1/10th of an inch into it. Literally. I've seen that many times while helping friends and neighbors with failed plumbing of various kinds.
-- ****************************************************************** * KSI@home KOI8 Net < > The impossible we do immediately. *
I assume (TM) (c) that if solder wets something, it will dissolve it. That's why soldering iron tips eventually erode. Given that assumption, solder dissolves ENIG all the way down to the copper. The nickel is sacrificial and just keeps the copper from diffusing into the gold during storage.
Who needs metallurgists when you can make this stuff up?
-- John Larkin Highland Technology, Inc picosecond timing precision measurement
John, what is "(TM) (c)"?
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.