Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

SMTnet
»
Electronics Forum
»
Substrate Au/Ni thickness

Substrate Au/Ni thickness

Subscribe to Thread

I have done some simple DOE and found that substrate with t... - Feb 15, 2001 by

What do you mean by "But after performing temperature cycle ... - Feb 15, 2001 by davef

Expanding on the limit of the portion of gold acceptable in ... - Feb 19, 2001 by davef

Thanks Dave for your helpful response. I found that the thi... - Feb 20, 2001 by

Let�s be basic. The two most common causes of BGA opens tha... - Feb 20, 2001 by davef

Nick

#5241

Substrate Au/Ni thickness | 15 February, 2001

I have done some simple DOE and found that substrate with thicker gold plating thickness.let say..0.85um(Ni~9um) exhibit a higher ball shear strength than a thinner one->0.5um(Ni~6.5um) after reflow(data taken in 48hrs).But after performing temperature cycle test ,the former aging rate(from the shear strength ) seems faster than the latter.Why!?Someone call this the 'precipitation hardness',but what actually is it! Any answer would be appreciated.OH..by the way the solder ball i am using is eutectic 62/36/Ag02.

This message was posted via the Electronics Forum @

reply »

davef

#5243

Substrate Au/Ni thickness | 15 February, 2001

What do you mean by "But after performing temperature cycle test, the former aging rate (from the shear strength ) seems faster than the latter."?

Comments are:

I can't recall having heard the term "precipitation hardness" [not that that means anything].

Repeating an old refrain, shear testing is a waste of time and money. Solder joint strength in pull/shear varies with lead geometry, solder volume, lead metal/metallization, and the way the test is done. It is very difficult to shear solder consistantly. Shear tests are [in my opinion] senseless. The shear stress you measure depends more on the shear rate and on the point where the force is applied than on the grain structure. When shearing a component, you not only apply shear stress, but also roll the component over. This means that both shear and normal stress are present, again making the results very dependent on where you apply the load.

Your nickel over-coat is thicker than necessary. 50+ u" is sufficient. The down side of overly thick nickel is cost.

Your gold top-coat is too thick and will cause reliability problems by embrittling your solder. A 3 to 8 u" gold thickness is sufficient. Regardless of the thickness, gold dissolves in to the solder and reacts with the tin in the solder to form intermetallics [IM].

Generally, people try to limit the amount of gold available by limiting the thickness of the gold over-coat. A rule of thumb is to limit the portion of gold to LT3% [I want to say] by weight.

So obviously, as the gold that's available to react with the tin in your solder connection increases, the brittle characteristics of the connction increase. And you are just accelerating the IM formation during your elevated temperature aging process.

I would expect the solder connections on boards with the thicker gold plating to be more brittle than on the board with the thinner plating.

reply »

davef

#5277

Substrate Au/Ni thickness | 19 February, 2001

Expanding on the limit of the portion of gold acceptable in a solder connection [mentioned in an earlier post]:

The embrittlement culprit, AuSn4, is 29 weight percent gold. So even if one had 100% tin as the solder alloy and 10% gold were dissolved into the solder, then about a third of the solder joint would be rocks (i.e., intermetallic compound) and the remainder would be soft solder. Do not count on the rocks to bond to the solderable surface. So, most recommendations are in the 3 to 5 weight per cent gold range for the maximum allowable percent.

The maximum ratio of gold finish thickness to solder alloy thickness can be calculated, to help prevent a problem from excessive gold-tin (i.e., AuSn4) compound. For the equation and its derivation, refer: "The Use of Capillary Action Measurements for Solderability Improvement," Soldering and Surface Mount Technology, No. 9, Oct., 1991, M Wolverton, Wela Publications, Ayr, Scotland, Table 5.

reply »

Nick L

#5279

Substrate Au/Ni thickness | 20 February, 2001

Thanks Dave for your helpful response. I found that the thickness of gold & nickel that you mention differ in a great deal with ours : 3 ~ 8 u"(micro inches)---> 0.076 ~ 0.2 um(micron) but our substrate gold thickness is between 0.3~1.2um. Is there any related to the plating process of substrate! And how can they control this thickness?(May be I should ask substrate vendor their controlling limit).

If this is the case ,the % in weight of gold would far exceed the amount u mention .Would there be a higher risk of open ball after a period of time when the bga packge is mount on to another pcb.Because recently we have customer complaining of solder joint separating form the BGA and solder ball interface while stay firm with the pc board.

This message was posted via the Electronics Forum @

reply »

davef

#5290

Substrate Au/Ni thickness | 20 February, 2001

Let�s be basic. The two most common causes of BGA opens that we talk about here are: 1 Gold thickness and 2 Black pad. Given the thickness of gold that you talking about, it�s reasonable to consider thickness as the first alternative to investigate.

Two other recent threads on SMTnet along similar lines are:

Date: November 07, 2000 09:49 PM Author: Dave F Subject: Re: BGA problem: open after reflow

Date: February 09, 2001 08:05 PM Author: Doug Benefiel Subject: PCB Gold Plating

Depending on the process, immersion gold self-limits at LT 12 u" [Some people say 8u", others say 10"u. And they�re all probably correct. Bottom line � if it�s GT 12 u", it�s not solely immersion. (And therefore will not be called immersion.)] Assemblers that either gold wire bond to gold or have specific gold finger insertion requirements require gold thickness GT 12 u". For these applications, fabricators increase gold thickness over this limit for immersion gold with additional processing. * Electroless gold deposits to 15 to 25 u". * Electroplate gold can�t be any less than 30 u".

Your 12 to 50 u" sounds like a wire bond spec.

Um, customer complaints. That�s bad. So, what you�ve seen is probably just the old tip of the �burg. Tsh, tsh.

Yes, there would be a higher risk of open solder balls after a period of time after the BGA package is mounted on a PCB. There�s a boatload of possible explanations. Examples are: * Intermetallic compounds increase over time, dependent on the materials, their concentrations, temperature, etc. * All solder connections become weaker through the course of time. * Greater exposure to environmental stress.

reply »