Semiconductor Power Semiconductor Assembly

I caught up with Alan Rae after a recent IWLPC committee meeting, where he jokingly asked me to, “Stop asking important questions” - LOL! He was kind enough to give me a few moments of his time to share his wit and wisdom, and answer some technology questions that, yes, I thought were kind of important…

[Andy Mackie] You’re increasingly being seen as “Dr Nano” by the electronics industry – how did you arrive as the focus of so much of this technology?

[Alan Rae] At the start of my career I was in the structural ceramics business. In the days of “ceramic fever” in the 1980’s the mantra was sub-micron and monosize (monodisperse) for lower temperature processing and better properties. It worked. Then at TAM Ceramics we made “sub-micron” barium titanate and other ceramic materials but we didn’t call it nano then. When I was at Cookson Electronics in the early 2000’s we started to see nanotechnology emerging from the woodwork with people saying the same about nanomaterials for the electronics industry. Then I joined NanoDynamics in 2004 and realized the scope and potential, ranging from semiconductors to touch screens to printable electronics, to LED lighting, to solar power, to materials such as nano solders, dielectrics, conductors…the list is growing but the leitmotiv is the same – small, monosize, tightly-controlled. 

[Andy Mackie] OK, so Nanotechnology has been a buzzword for quite a while – is there a clear definition yet, and what current uses are there for nanotechnologies that may not be immediately obvious?

[Alan Rae] Well, the definition has been really tough to derive – ISO TC 229 “Nanotechnologies” came up with a definition that one dimension of a particle, needle or plate should be less than 100nm but it’s really tough to define…should all particles be less than 100 nm? 50%? Any? And should it be exactly 100nm? There are a lot of opinions. The Woodrow Wilson Institute lists over 800 consumer products containing nanomaterials on the market now – industrially the products range from semiconductors, to fillers in packaging materials and underfills, to antimicrobial and self-cleaning coatings for phones. Solar panels, especially thin film ones, depend on nanomaterials in their manufacture.

[Andy Mackie] What is in the pipeline for nanotech electronics and semiconductor interconnect materials? I know that nanosolders are starting to gain ground in some areas – what else is upcoming?

[Alan Rae] Much of the work in nano metals is being done by universities and small companies – for example my small company is working with Purdue and the Air Force to develop a novel solder technology – but commercialization will come by partnering with established companies like Indium Corporation, who have the distribution and technical support so that customers will be comfortable with a new material. Cost and reliability are king. Indium is already in the reactive nano foil business; there are existing and near-term applications for silver, silver-coated copper, alumina coated boron nitride and their combinations in adhesives, shielding materials and thermal interface materials.

[Andy Mackie] Several years ago, quantum dots were being promulgated for tunable band-gap detectors and quantum computers. How close are quantum dots to seeing real uses, and what else is on the horizon?

[Alan Rae] Quantum dots are unique and have great potential in medical imaging and as frequency shifters for LEDs. The markets haven’t developed yet because of the cost and because some of the best dots are cadmium (toxic metal) based. I’m working with a group at University of Buffalo which has a silicon quantum dot process that looks like a promising alternative. Quantum dots will have their time…but not just yet. In terms of new developments – they range from core shell and modulated structures for thermoelectric to replacing indium tin oxide with carbon nanotubes or graphene. The US National Nanotechnology Initiative tracked $1.6 billion in Government spending (check out www.nano.gov) in the last year at Universities and small businesses and NSF has set up centers of excellence at Cornell and other great universities that are really working hard to translate science into technology so we can make practical products.

=======

Alan, many thanks for your time, and for sharing your insights with us.

Cheers!  Andy

Just before the New Year, I had the opportunity to sit down with my old friend and colleague Dr Conrad Sorenson. Conrad is an expert in the fusion of electronics technology, marketing, and results-oriented business strategy – a discipline he refers to as Technical Scouting, and he specializes in sorting out and quantifying trends in electronic materials.

[Andy Mackie] The USPTO already has a search engine on its site – why is it sometimes ineffective at helping me find the patents I need?

[Conrad Sorenson] The USPTO system is a giant Oracle database.  We interact with the database by matching keywords against its contents. Unfortunately, because there can a thousand ways to describe an innovation, using the English language to retrieve them can be quite difficult. Ambiguous results can be improved by cross-filtering using CCL codes and sub-codes (aka the US Patent Classification System, a rather old-fashioned listing of technologies which can be used to limit results to specific fields of use) or restricting inquiries to specific fields such as abstracts, inventors, assignees, etc. Unfortunately, because the database must serve the public, the USPTO limits the length of queries to the patent database, which impedes our ability to extract relevant information.

[Andy Mackie] How does your patent search technology work?

[Conrad Sorenson] I have automated both the initial basic search criteria and the process of downloading the results to a local database. Because filtering is done offline, my strategy has evolved to use the USPTO selection criteria broadly, to minimize the amount of pertinent intellectual property overlooked by the initial filtering process. After pertinent patent data is downloaded locally, my patent scanner confers with proprietary databases to automatically segment results into markets, technologies, and supply chains. 

 Another important feature of the scanner is that it not only looks into the databases of granted patents but it also simultaneously explores published documents in the patent application database. Patent applications are fascinating. They can reveal knowledge about the internal development investments which could only otherwise be gained under non-disclosure. According to research I performed a few years ago on Atomic Layer Deposition, patents are published within about 9 months ± 6 months from the filing date. By analyzing changes in the number of patent filings in over time, you can learn whether a technology approach is being emphasized, how it ranks with other technology development efforts, and by closely examining CCL codes, learn if the technology has shifted directions – for example from ALD process for depositing high dielectric constant materials toward diffusion barrier coatings. The downside to including patent application data is that results can often be obscured – only market share leaders clearly identify themselves as the assignee in patent applications. My scanner sniffs this information out by examining each inventor’s recent assignment information, double-checking to ensure if they are in the same general geographical location, and then determines a likely patent assignee.

      As you can tell from the above description, my development efforts continue to be around the management of an abundance of data, to winnow it to the knowledge we seek. When restrictive filters are removed in the search for patents, we have to develop new skills for managing this abundance of data. Because such techniques can yield thousands of results, even after market, technology, and supply chain segmentation, I am applying artificial intelligence concepts to the search for knowledge. I am having excellent success using a neural network to patent search results to distil highly relevant patents from a pool of mixed relevance and non-relevant IP.

[Andy Mackie] Can you explain “neural networks” in simple terms?

[Conrad Sorenson] Neural networks are method of teaching a computer to “think” much like a human. They emulate the functioning of the brain through the use of weighted linkages between synapses or “nodes” (see figure). 


The style of network I am using (a back-propagation neural network) actually learns a task by comparing results against a training set, then adjusting weighted linkages between synapses until error rates drop below a threshold. I “train” the network by providing it with an exemplary set of patents classified by relevance (High, Medium, Low, and Not Relevant). The patent scanner manipulates node weights until it is able to reproduce my exemplary results, then applies this to un-reviewed patents. It automatically sorts through an almost unlimited number of patents (more than one per second) and comes up with results that are very interesting! In this manner, I am able to break through the limitations imposed by a keyword searching methodology, by duplicating and applying my judgment of patent relevance to the process. 

[Andy Mackie] How do you find the “key phrase” that differentiates the patents I am truly looking for, from the “chaff”?

[Conrad Sorenson] That’s my secret: I don’t! I unfetter the USPTO keyword search process, and then apply the techniques outlined above to the results. For example, I recently did a search of patents for “semiconductor vacuum chamber seasoning” for a client. It turns out that restricting the USPTO search to only semiconductor applications culled out some very useful data. I got the best results using the keyword search terms ‘“chamber” AND “seasoning”’. As expected, the results included more references to Kentucky Fried Chicken than IBM! By segmenting results to electronics markets, I quickly achieved the results which were useful for my client.

[Andy Mackie] As a technology marketing guy, you look at patents in a different way than intellectual property specialists or R&D professionals. Can you give us some insights from your experiences?

[Conrad Sorenson] I do not look at patents from a traditional standpoint of intellectual property. I consider the USPTO to be a rich source of competitive intelligence. This is analogous to the use of signals intelligence to obtain important information. In World War II, after a change in codes, Allied code breakers would go for long stretches of time without being able to read our adversary’s coded messages. During these dry spells, they could still obtain highly-valuable information by determining where messages were transmitted, when they were sent, and who responded to them. Similarly, what companies are doing – or not doing – and how much they are spending to do so can be determined by information published in the USPTO database.

Combined with more than a decade of experience looking at online patent information; I can also infer technology maturity from supply chain segmentation. Because a technology progresses in a predictable way, early IP has a supply chain which emphasizes Universities, National Laboratories, transitioning to large industrial laboratories. Materials companies then emerge, followed by OEMs and End Users. In a well-developed market such as semiconductor manufacturing, there is a ratio of 1:10:100 between materials suppliers, OEMs, and End Users. If this ratio is skewed, it indicates that there is a competitive challenge in the outlier.

[Andy Mackie] I really appreciate your sharing this with me and our Blog readers. If you’d like to learn more, Conrad can be reached through his website , phone +1(716) 639-0721, or email him at Conrad.Sorenson@EnablingMaterials.com

Cheers! Andy

Forming gas is a complicated topic, so I will provide some preliminary background in this section, then get into the soldering part next time.

Don't you mean "formic"?
 

Forming gas is a mixture of hydrogen (H2) and an inert gas (usually nitrogen, N2) that is used to reduce oxides on metal surfaces to water. Please don’t confuse this with formic acid (HCO2H), which I hope to touch on in another posting later this year.

Safety

The reason for the dilution of hydrogen by the inert gas is to keep the hydrogen below 5.7% (by volume), as this is the point above which the hydrogen can spontaneously combust. Gas companies such as Linde and Air Products consider forming gas at less than this level to be an inert mixture, so the fittings used for gas cylinder attach are the standard CGA580 type used for nitrogen, argon, helium and so on. Depending on the gas supplier, they may allow a maximum of either 5.0% or 4.0% hydrogen, to ensure they are within safety margins.

All this notwithstanding, 100% hydrogen furnaces are used around the world in a variety of different processes, and I have also seen soldering processes around the world where 10% and even 20% hydrogen/nitrogen forming gas is in use. I am not saying that >5% H2/N2 can not be safely used, but you have to be careful when using it.

 
Gas Supply

There are three ways of supplying gas for forming gas-based soldering processes:

1/ Mixing hydrogen and nitrogen in a special panel. Sometimes this may also incorporate a catalytic reactor that reacts ppm traces of oxygen, with hydrogen to form water: the water  is then removed by adsorption. This process makes a very "clean" forming gas that will have optimal reducing properties. Usually, the nitrogen source is from vaporised cryogenic N2, and the hydrogen is from a cylinder or "tube"-based sources.

2/ Cylinder supply. A single cylinder, or a manifolded bank of cylinders may be used to provide the gas as a mixture. Usually, this is used as-received without being cleaned up.

3/ Ammonia cracking. Basically, NH3 -> 3N2 + H2. This is feasible, but results in a fixed 3:1 ratio of N2 to H2, and is never used (to my knowledge) in soldering. It is also massively inefficient in terms of costs and power usage to make the ammonia, plus the ammonia usually has a much higher moisture content than a nitrogen plus hydrogen gas mixture.

What does it do in soldering? I’ll get into that next time: I'll be talking thermodynamics and kinetics, and there WILL be a test.

Cheers! Andy

Andy Mackie: What role does gold play in protecting surfaces in SMT and semiconductor assembly processes?

Lenora Toscano: Gold does not form an oxide; it protects the nickel from oxidation or passivation. A clean nickel surface has very high solderability for most solder types, but its oxide is very difficult to remove with standard flux types. Also, gold dissolves almost instantaneously into most solders during assembly, thus promoting superior wettability.

Andy Mackie: What standards exist on the thickness of gold for different electronics and semiconductor assembly applications?

Lenora Toscano: The main application of ENIG (electroless nickel/immersion gold) coating is in chip-on-board (COB) technology, the typical thickness of the immersion gold layer on the HDI substrate being 3-5 micro-inches.

Edge connectors typically require the use of hard gold. Acid gold deposits are used for compliance with MIL-STD-275, which states that gold shall be in accordance with MIL-G-45204, Type II, Class 1. The thickness shall be 50-100 micro-inches, typical thickness is 30-50 micro-inches on 150micro-inches nickel.

On the other hand, for solderable surfaces, typical thickness is 5-15 micro-inches on 150micro-inches nickel.

For wire bonding, in general, gold plating of a minimum of 30 micro-inches on 200 micro-inches nickel works well. Soft gold is generally preferred. Soft gold processes are also used for boards designed for semiconductor chip (die) attachment. These qualities comply with Type I and III of MIL-G-45204.

Andy Mackie:  What are the differences between gold layers deposited by immersion gold and electroplated gold processes?

Lenora Toscano: There are five main differences:

1. The coating thickness is different. Immersion gold is a displacement reaction, gold displaces the nickel on the surface, and is self-limiting as the nickel surface is coated with the immersion gold. Common baths cannot produce thicknesses of much more than 10 micro-inches, while with electroplated gold the thickness depends on current and time. The higher current or longer the plating time the thicker the gold coating.
2. The structure of the gold deposit layers is different. Electroplated gold is denser that the naturally porous immersion deposit.
3. The hardness is usually different. Electroplated gold often has other metals introduced into the plating that make the deposit harder.
4. Porosity is different. Immersion deposits have more porosity that electroplated deposits; it is the nature of the plating system.
5. Deposition composition (purity) varies with additives in the bath. Immersion gold baths contain gold as the only plated metal, while electroplating systems may introduce small amounts of other metals.

Andy Mackie: How thick does gold have to be to fully protect the underlying surface, and what are the trade-offs as customers attempt to reduce their gold costs?

Lenora Toscano: Per IPC-4552 ENIG specification, 1.97 micro-inches is the recommended minimum at +/-4 sigma from the mean, with 3 – 5 micro-inches being typical.

The immersion gold deposit is porous by definition. It does offer very good protection to the underlying nickel, but over time the porosity of the deposit results in the passivation of the nickel surface and the wetting forces will be reduced. Of course, this process should take years to occur, but if the gold coating is too thin (below the minimum requirement), it will occur sooner and affect the solderability. 

Andy Mackie: What advantage does gold have over silver or other metals?

Lenora Toscano: Again, gold has good tarnish resistance and solderability after storage because it does not form an oxide or hydroxides, so it is unaffected by temperature and storage conditions that might reduce the shelf-life of the other finishes. It meets requirements for lead-free (Pb-free) assembly while offering a coplanar surface that is both solderable and aluminum-wire and gold-wire bondable.

Gold has good electrical conductivity, and produces a contact surface with low electrical resistance. Electroplated gold is also an excellent etch resist.

Electroplated silver is not widely used in the printed circuit industry. Under certain conditions or electrical potential and humidity, silver will migrate along the surface of the deposit and through the body of insulation to produce low-resistance leakage paths. Alkaline cyanide baths for silver electroplating are highly toxic.

Immersion silver is susceptible to problems if not correctly stored and even packaged. Packaging materials that contain sulfur or allow exposure to air will result in tarnishing of the surface (sulfide, sulfate, and chloride formation). High levels of surface contamination can detrimentally affect solderability.

As a supplier of electronics materials, Indium Corporation is constantly faced with customer requests for “halogen-free” (HF) soldering fluxes and associated materials. This is an interesting trend, but we face several challenges here:

1/ What is “halogen-free”? We have not seen any consistent message from our customers on what they mean by a halogen-free flux. As a materials-supplier this is an absolute show-stopper.

Based on several conversations with interested parties, my understanding of the IPC status is as follows, and apologies for any misunderstandings to Tim Jensen (ICA) and Tom Newton (IPC). The IPC’s 4-33a Task Group, which was looking at a universal halogen-free material standard (J-STD-709), saw a failure of a second ballot on the standard, even when it got downgraded to a guideline. The 4-33a group faced numerous differences of opinion: on what materials should be included; what halogen levels are allowable; or even whether a single component could be considered a "homogeneous material” to be ground up and analyzed for halogens and so on. The task of defining HF will now reportedly be taken up by two separate groups from IPC and JEDEC.

Meanwhile, in March of this year, the Japanese organization JEITA quietly released their understanding (ET-7304) of what is meant by HF fluxes and solder pastes, using a 1000ppm halogen limit. This definition is clearly at odds with the IEC's definition of HF. That is, 900ppm by weight maximum of chlorine or bromine atoms, or a maximum of 1500ppm of both: the so-called “9-9-15” limit. .

2/ Which halogens? The strict definition of a Group VII element (halogen) is one of Fluorine (F), Chlorine (Cl), Bromine (Br), Iodine (I) and radioactive Astatine (At). From environmental reasons, chlorine and bromine in halogenated fire-retardant (HFR) materials that emit dioxins and similar compounds when heated should be eliminated. However, some customers are also throwing fluorine and iodine into this definition, too. This may be based on fears of electrical reliability, but from my perspective the customer is becoming defocused from the necessity of meeting environmental concerns.

3/ Does halogen = halogenated fire retardant? Not every halogen found in an electronic material is an indicator of a halogen-containing fire retardant!

Greenpeace is the main driver behind this, and I have, to date, been unable to get a response from them on how they will detect halogens on circuit boards. The fear from our customers, and our customers’ customers, is that an electronic device (iPhone / flat-screen TV or other) will be obtained by an environmental group; pulled apart; and X-ray fluorescence (XRF) used to detect halogens. The minimum sample size that can give a quantifiable result for halogen-free is reportedly 2grams: contrast this with the milligrams of material (residue) present from a no-clean flux in a cellphone, and you can see the issues in quantifying halogen levels based on flux residues. We can’t do it reliably.

4/ How can we confirm “zero” halogens? Contrast the JEITA standard with the requests from many customers for zero halogens / “no intentionally added” halogens / “elemental halogen-free” fluxes or some such. However, since many customers insist on third-party data-reporting, we are reliant on these analytical labs to reliably give us data. One of the challenges we face is when, for example, a lab reports “63ppm of chlorine”, based on a reported limit of detection (LOD) of 50ppm. Our customer is outraged: “You said it was halogen-free!”

Those of you familiar with the statistics of analytical chemistry will immediately see the two fallacies here: the first is that they have reported not the method detection limit (MDL), but the much-lower LOD. The MDL is a function of analytical equipment PLUS the errors in sample preparation and handling. The second fallacy is that you can not report 63ppm as a reliable, reproducible number, since the limit of quantitation (LOQ) – the limit at which you can actually give a figure for the concentration – is more than 3 x the MDL. The limit of analytical capability to reliably quantify the amount of halogen present is therefore around 150ppm or greater.

Instead of reporting “63ppm halogen”, a more accurate statement is: “In our single test, we found a small peak in our spectrum at the same elution time as a halide-ion. It may be a halogen, or it could be one of the millions of anionic organic species taht elute at the same retention time. The quantity found is well below the method detection limit, so we have no way of knowing if it is from contamination during the sample preparation, and we certainly can not tell how much is present.”

5/ What is a ‘homogeneous material”? Some customer standards require the level of halogen in a homogeneous material to be reported. We can probably safely say that a flux is a “homogeneous material”, but is a solder paste truly homogeneous? Both JEITA and Indium Corporation can agree that the flux-content needs to be the focus of the analysis, but an unscrupulous solder paste supplier may, for example, take the analysis of a 90%w/w metal solder paste, and report the results as “890ppm chlorine”, knowing that the level in the (10%) flux is 8,900ppm chlorine, essentially diluted by the 90% metal content.

Conclusion:

As a global electronics materials supplier, we at Indium Corporation can see three possible solutions to all these dilemmas:

a/ Adopt the JEITA specification – even though it goes against the 9-9-15 EIC recommendation. This allows us to be on a level footing with our Japanese competitors, but appears to put us at odds with the needs of some of the semiconductor assembly and electronics assembly industries.

b/ Adopt a three-tier specification based on the IPC/IEC recommendation – the Indium Corporation approach is given here (below).



Why three levels? Because our more discriminating customers are telling us that truly halogen-free fluxes are simply not as effective as those that contain small amounts of halogen. For those who are concerned about end product reliability, a “halogen-compliant” tier allows the best of both worlds.

c/ Report the atomic chlorine and bromine levels present in the flux component, and allow the customer to choose what they want, based on this.

If you are a user of Indium Corporation materials, or even a competitor of ours - what makes most sense here? Or is there a fourth or fifth way?

Chers! Andy

Had some very interesting conversations at the IWLPC show last week, as always: one discussion was with my good friend Jeff Schake of Dek. He knew I had done some work a few years ago on a system for preventing solder paste drying out on the stencil by maintaining a set solvent vapor pressure, or “%RS” (the solvent equivalent of a %RH) over the paste.

Almost as a byproduct to my engineering work, I realized that I would need a tool to differentiate between solder paste that was losing solvent, and one that was simply “thickening up” due to time-dependent rheological changes (TDRC). TDRC includes phenomena such as thixotropy and rheopexy. Thixotropy, or shear-thinning, will be familiar to everyone who has had to shake a glass bottle of tomato ketchup in order to get the darned stuff to come out of the bottle: the ketchup has a weak gel-like “structure” that can be broken down to a low viscosity structure by simply shaking it. The (retardation) time it takes to break the “structure” down can be modeled by a simple exponential function based on the rheological equivalent of a half-life, as can the (relaxation) time. "Relaxation" here means the rebuilding of the gel-structure with time of the paste when it simply sits with no shearing forces, and the rate at which it rebuilds is dependent only on the diffusion kinetics -  [just in case you read the title of this blog entry and had mental images of solder paste tanning itself in the sand on Palm Beach].

The no-clean solder paste studied certainly did change in viscosity with time, but the change was reversible, fitting the rheological model rather well (data shown below) but demonstrating that irreversible drying (solvent loss) was NOT the prime cause of the stencil clogging and other problems seen.


Contrast this with the water-wash paste (data shown below), which even after a couple of hours showed signs of irreversible performance degradation that could be prevented by maintaining a controlled atmosphere as a specific %RS over the paste surface during its stencil-life.




 

Most no-clean pastes these days require little or no kneading (stirring or printing on the stencil) before they are used, but as a rule of thumb:  the more kneading a paste requires before it is useful – the less time it can be left on the stencil between prints.

Mechanistic insights like this are also helping us to develop improved wafer-bumping and flip-chip solder pastes.

Cheers! Andy

Why 8 random things? Because that’s the number of random things I’ve got listed below, that’s why – do I look like Letterman?

OK, so I spent this week closing some business in California, and also 3 days at the very well-attended and immaculately-organized IWLPC show, which is starting to mutate into the WLP and MEMS show, in the same way that Semicon West budded into the Semicon West plus Solar Wild-West show in 2008 and 2009. I will say a big thank you to Lee Smith of Amkor and Dan Baldwin of Engent for their excellent and worthwhile training programs - they have a lot to teach us. Here’s some other stuff I learned: 

1. From Dr Rao Tummala’s Keynote Speech: Moore’s Law may be leading the smaller and smaller revolution, and Packaging is severly lagging behind (no surprise to anyone who’s been watching the ITRS for the last five years), but the real surprise is how the overall system that the silicon chip and the package are housed in is similarly dragging its feet. See 7. (below) for some insight into the "why" here.

2. If you are chairing a session, you really will need a good 30minutes start on the session, to ensure everything is in apple pie order. Corollary: Just because the projector is working, doesn’t mean it is connected to your laptop, and my apologies to Dr Raj Gupta of Volant Technologies for slowing him down. He did take the slight delay to starting his presentation in his stride, however, which shows what a gentleman he is!

3. WL-CSP for high-end, fine pitch processors with ULK dielectric layers may never get off the ground, and the competition is cheap, inorganic interposer technologies that have CTE’s close to silicon. Getting low pitch, high I/O conductive holes in glass, however, is no picnic. And would all those muttering about “Flip-Chip on CBGA?!” please keep their comments to themselves: what’s old is new again.

4. If you ask the same question of experts from 3 different marketing research companies, you will get three radically different answers, each within a certain range of “correct” depending on how they view the market. A summary of all of the answers will be very helpful in determining a course of action, and you will learn a lot about the extent to which each organization is blind-sided in their market view.

5. The phrases “killer app” and “smaller, cheaper, faster” have died the death. Yay! I don’t mourn the latter, but I really didn’t get a good feel for what actual DEVICE market need is driving WLP and 3D integration. Yes: Toshiba camera modules in 2007. 8 stack DRAM modules in 2008…but the sensor / logic / RF / NAND / DRAM / MEMS / bacon / lettuce / tomato stack up envisioned for 2013 may be way off base: beaten back by simple throughput, KGD and heat-dissipation issues.

Corollary: Speed of technology change is fine in and of itself, but what concerns me is the VELOCITY of change. As you’ll recall, velocity is a vector quantity, necessitating that you state the direction of motion. Going fast down a dark alley is never a good idea. Speaking of which...

6. DARPA does the freakiest things! Dr Alan Rae (one of the presenters at the Plenary MEMS session I chaired) shared one of the ways the US government is planning on spending our tax dollars. Radio-controlled flies. No Halloween joke. The basis of this project proposal is to put a chip into an insect (e.g. your regular Musca Domestica: the common house fly), and use the energy (motion/heat/electrochemical) generated by a baby fly… OK, a maggot (thank you, John Fowles)… to power the device, which will have nanowires embedded in the insect’s nervous system. These microwires will interface with the fly’s nerves as they grow, allowing you robotic control over the fly, so it goes wherever you want it to.

Over a glass of wine, Alan and I saw some issues here: most particularly as said fly will also have some degree of freewill. One can envision the vexed fly oscillating in some metastable state: torn between the radio-control chip telling it to continue spying on a terrorist cell, and its innate instincts telling it to commence chowing down in the terrorists’ outhouse. “HELP MEEEE!”

7. There is no attempt made yet to shrink the human finger or the human eyeball, so input and output devices will remain severe limitations. Witness a recent text message to my boss on “Fkuxes” this week. Ross Berntson responds immediately “Fkuxes is a new product line for us ;)…” Yeah, LOL! But you see the problem: Blackberry keys and human thumbs…

After chatting to my favorite futurist, Dr Ken Gilleo (pictured right with Leslee Johns) this week, I think we’re safe in assuming that, based on all the above, the next step by 2020 will be DIS. Device in Skull. Bluetooth for the central nervous system: it will look like telepathy. Nanowires will communicate with visual, vocal and aural nerves, and the whole thing could be placed in a sinus or some other pretty vacant bio-real-estate. You can use a microturbine to harvest the energy from breathing. Again, shunting (not merely tapping into, but SHUNTING) and switching the nerves from DIS to brain will present problems, but I’m sure we will see some biotech answers to this – binuclear Schwann cells or some such. An idea not to be sneezed at. Not that sneezing would be a good idea anyway, come to think of it. Which reminds me....

8. Extremely important: Do not make a funny comment to Alan Rae when he just took a bite of dessert. Particularly during the keynote speech. Sorry, Alan. Was funny, though, right?

Cheers! Andy

I just sat down to talk to Tommy Acchione (pronounced “akki-OWN”) Applications Engineer with Indium Corporation’s  new product line, Reactive NanoTechnologies’ (RNT) NanoFoil^®, about the technology, and its offerings into the semiconductor, power semiconductor assembly, LED and display assembly industries.

[ACM] First of all: welcome to Indium Corporation! Can you tell us, in just a few words, what the basis of the RNT Technology is?

[Tommy Acchione] NanoFoil^® technology is a thin metal sheet (“foil”) made up from alternating ultrathin layers of aluminum and nickel (Al and Ni). The reaction between these two metals is stoichiometrically very simple:

            Al+2Ni -> AlNi₂

And extremely exothermic (heat-generating). This reaction (see picture) is started by a very localized heat or other high-energy source, such as a 9V battery or even a laser beam. For a fraction of a second, the alternating thousands of sandwiched layers reach temperatures as high as 2000degC, and this isotropic heatwave radiates away from the initial hot-spot through the foil at speeds of about 5-8meters/second.

Just banging two lumps of Ni and Al together will never initiate a reaction this intense, as the two large pieces of metal act as very effective heat sinks, but by layering the metals together, the heat-generating reaction propagates by allowing the adjacent layers of Ni and Al to rapidly interdiffuse, so giving out more heat, causing the nearby layers of Ni and Al to interdiffuse and so on.

[ACM] How are these materials manufactured?

[Tommy Acchione] First, we pull a high vacuum, equivalent to those vacuums found in outer space, then we sequentially deposit the alternating layers by a sputtering process onto a specially-made metal block.

For a bonding material, a layer of a specialized brazing material is initially deposited onto the metal block, then the Al and Ni are put down, then a final capping layer of braze is deposited. The initial brazing layer both enhances subsequent bonding and also helps with easy removal from the surface of the metal block.

[ACM] I understand that the uses of these materials are expanding all the time. Can you give some examples that you can talk about?

Well, as you know we have about 30 patents on this technology and 35 outstanding patent applications, but I still have to be careful talking about newer applications, which are emerging all the time.

The biggest uses are in sputtering target manufacture (which is a little ironic, since that is how they are made!); Component mounting; and what we can call “reaction initiation”, or “energetics” - things requiring an instantaneous heat-source.

Sputtering Targets: For sputtering targets of non-refractory metals, standard indium or diffusion may be the preferred method. For most refractory metals and ceramics, solder wetting and CTE mismatches make bonding with standard processes difficult. NanoFoil^® allows for these materials to be bonded at room temperature, thus removing any CTE mismatches during bonding or subsequent cooling processes.

However, as targets get larger for flat panel displays (and we are seeing needs for up to 3m x .4m targets with higher generation depositing), indium starts to become too weak to take the weight of the indium-tin oxide (ITO or InTO) target itself, and only the strength of a NanoBond^® is sufficient to hold the target in place. Another key factor is that a manual bond of a large target to its backing plate starts to become simply physically unwieldy for an operator, as its size and weight increase. NanoFoil^® becomes the elegant and simple solution here.

Component Bonding: One major market that we are seeing is in component bonding. I can’t talk too much about this, but for high-brightness LED’s (HB-LED’s) and photovoltaic concentrators (CPVs) there is a growing demand for a high-temperature stable, thermally-conductive flux-less bonding material able to provide low junction temperatures over the lifetime of the device.

Energetics: Here we are talking about fuses and timed devices, with specially-shaped initiators that take advantage of the ignition properties and the reaction rate and energy produced by the NanoFoil^®.

[ACM] How can people contact you to discuss applications further?

[Tommy Acchione] Well, they can call me at Indium HQ (+1-800-4INDIUM or +1-315-853-4900, or simply email me at tacchione@indium.com .

[ACM] Tommy: very interesting! Many thanks for your time.

Steve Adamson of Asymtek

I had the chance to discuss fluid dispensing recently with my fellow-Brit, Steve Adamson of Asymtek.

After you've reflowed solder in contact with a flux, you're always left with a certain amount of flux residue. There are no clear industry guidelines on how you refer to the residue, and new terminology is emerging all the time. If you leave it up to me, here is what I recommend : 

1/ "No clean" flux residues:

- Standard Residue:  >40%
- Low Residue (LR): Between > 10% to 40%
- Ultralow Residue (ULR): Between >2% to 10%
- Near Zero Residue (NZR): Between 0 to 2%

Each % is given as the weight percent of flux residue after a real reflow process, and refers to the fraction of the raw flux, or flux component of a mixture (such as solder paste or metal-filled epoxy). Note that the exact amount of residue will vary with the reflow profile; the mass of flux or solder paste studied; and the rate of gas flow over the sample material, as well as secondary factors, such as the oxygen level in the reflow atmosphere.

Thermogravimetric analysis (TGA) is a pretty poor method for determining post-reflow residue levels. Results from the use of a platinum TGA sample cup with nitrogen flowing over it have been found in our testing to vary significantly with the mass of sample present, probably because the headspace in the cup acts as a "dead zone" for entrapment of vapor: TGA may therefore give artificially high % residue readings, compared to the results on a flat leadframe or other substrate.

From the viewpoint of a standard semcionductor assembl process, now consider the situation of a low-clearance direct chip attach "flip-chip" or package-on-package application, where the flux is essentially entrapped in a "cage" of I/O's, sandwiched between two flat diffusion barriers. As well as issues of flux residue, this also raises the question of how the electrical properties of the flux will be affected, if more of the solvent and other volatiles from the flux are trapped in the residue.

2/ "Water-soluble" (same principles apply for "Solvent cleanable") flux residues:

- Water-soluble: Residues can be truly dissolved in water to leave a transparent liquid: the color of the this rinse liquid is immaterial,
- Water-dispersible: Non-transparent rinse liquid with any hint of translucency or turbidity

I know that the differences here will be very dependent on rinse-water quality and temperature; chemistry of any cleaning agents; stage of bath-life and so on, but to my mind, if the rinsed liquid is not transparent, then the solids from the flux must be suspended as fine particulates. These particulates usually have refractive indices different from the bulk liquid: the result - turbidity. There may be a means of bath-life end-point determination by turbidity or dynamic light scattering (DLS) or a similar technique; possibly in combination with the standard refractive index measurement that is most commonly used.

In conclusion, note that ULR and NZR fluxes are showing increased usage in flip-chip applications, since these types of material interfere less with the curing of underfill polymers. NZR fluxes are becoming critical for copper-pillar bumping applications.

Just my thoughts - let me know what you think.

Cheers!   Andy

Got contacted last week by Tony Hilvers of the IPC (Association Connecting Electronics Industries). Tony tells me that the Canadian Government is considering banning some rosin and resin-based chemicals that may be of interest to flux formulators for both no-clean; solvent-clean; and even water-wash solder pastes and fluxes. The Canadians are at an early, investigative, stage here: allowing the various interested parties six months to respond.

My initial, knee-jerk reaction is as follows:

1/ While Tony and the IPC's rapid response is commendable, note that we in the electronics industry are not alone. Even a cursory Google search shows that the vast majority of these types of material are used in the following, multi-billion dollar, industries:

- Paper manufacturing

- Cosmetics

- Adhesives and glues

- Synthetic rubbers

- Coatings

- Printing inks and toners

We in the electronics industry are relatively small fry: combining our voice with that from these other industries, may give the Canadians pause for thought.

2/ If you're wondering why I'm so interested in this, it's simply because after the Pb-free switch in most of the Electronics Assembly industry, I am now seeing the Electronics Assembly and Outsourced Assembly and Test industries still in turmoil over the exact meaning of "halogen-free" solder fluxes. Industry sources are telling me that there is a strong movement to pull back from  the absolutist "zero tolerance for halogens of any kind" to a more rational call for a certain limit to them, based around the standard "9-9-15" halogen classification. The hard truth is that truly "no-intentionally-added" (NIA: that is, TOTALLY halogen-free) solder fluxes may, in some instances, simply not be as effective as those containing moderate amounts of halogens.. More on "halogen-free" in a couple of weeks.

3/ Eliminating rosins and rosin-derivatives, including materials that may be present in naturally-occurring rosin products, may not only have the beleagured Canadian timber industry up in arms, but will probably result in another protracted round of setting of allowable limits for "proscribed substances" some of which.... umm.... occur naturally.

All comments, corrections and clarifications gratefully received.

Cheers!  Andy

Dr Jennie Hwang

I recently had the opportunity to discuss several issues in Pb-free die-attach and other solder applications with Jennie Hwang PhD, DSc, and world-renowned consultant in solder and  electronics assembly processes. 

Cheers! Andy

Figure 1: 15% Voiding with air reflow

Figure 2: ~0% Voiding after vacuum reflow

Figure 3: Multiple voids

While at the Semicon West 2009 show in July, I had a chance to sit down with Herr Klaus Roemer of Pink GmbH. PINK is most famous in the die-attach and power module manufacturing world for their reflow ovens with vacuum, but are also known in the medical and aerospace industries for manufacturing extremely high precision, one-off, vacuum equipment for applications as diverse as particle-accelerators for ion bombardment, and large-volume chambers for helium leak-detection. I asked him some questions about Pink vacuum soldering technology.

Cheers! Andy

... and we're back now with the real situation of solder powder size and its effect on solder paste rheology.

Size distribution - Real solder powder is not monodisperse (single diameter), but has a spread of sizes (typically approximating to a log-normal distribution). Generally, the wider the distribution, the higher the maximum packing fraction. This can be easily understood by looking at the picture from the last post (below) and imagining tiny spheres fitting into the little interstices between the particles: they would have to be around 1/10 of the diameter of the larger spheres. Theoretically, you can get 0.99 or even higher packing fractions with a very specific multimodal distribution, but you never see this in real life.

Boundary layer - Every time a fluid flows over a surface, the part of that fluid closest to the solid surface does not move, relative to the surface. On a molecular level, individual molecules are diffusing in and out of this so-called "static boundary layer", but essentially, the fluid right next to the surface is completely immobile. The fluid just above this static layer is moving slowly, and the next layer out moves faster still, until the velocity is the same as that in the bulk fluid. "Ok" you say, "so what?" Well the fluid around the solder particle therefore forms a kind of shell that is almost like an extension of the particle into the fluid, and the thickness of the shell is not dependent on the particle size. Complicating things further is the fact that solder paste fluxes are plastic, not Newtonian, so the boundary layer goes out even further. To sum up: smaller solder particles have a "virtual shell" around them that means you need a lower metal weight percent to get the same rheology.

Non-sphericality - The sphere is an ideal solid, and any deviation from perfect roundness causes an increase in the "k" factor, which is 2.5 for spheres, and increases as the particle become sincreasingly deformed, more and more fluid being trapped either within or around the particle (see picture). Most powder these days is spherical, so it's not a big deal

Chemical reactions - Activators (see previous posts) are very good at removing oxides from metal surfaces. There is a myth that there is a magical temperature at which the activation (metal oxide plus activator) reaction occurs, but that's exactly what it is: the myth of the "activation temperature". How can I demosnstrate it's a myth? Simply because solder paste has to be stored in a refrigerator, or else it increases in viscosity with time through the so-called "concretion" reaction which is just the slow reaction of activators with metal oxides to form solid reaction products, just in the same way that water hydrates cement to swell the crystals and cause them to change shape and grow, interlocking together into a solid mass.

Air - No matter how hard you try, you will always have a little air mixed in with your solder paste.

A complicated answer to a simple question!

Oh, and by the way, once you've embarked on a study of solder paste rheology, there is is the little matter of "artefacts". A subject for another time....

Cheers!  Andy

A well-respected colleague asked this week if the metal loading of type '4' solder powder in a paste was lower than that for type '3' because the particles were smaller. The implication being that smaller and smaller particles will fill up more and more volume. He is correct in that typically, in solder paste made from type 4 solder powder, the metal loading is usually slightly lower than that for the same alloy in type '3', but not for the reasons he is thinking of. Bear in mind that (per the ANSI/IPC powder J-STD-006A, the type '4' is typically between 25 and 38 microns, while the type '3' is from 25 to 45microns in diameter, so the type '4' is just a smaller range of particle diameters.

There are two aspects of this: the geometry of solder particle packing... and some things we can lump together under the heading of "harsh realities".

Particle-Packing:

Just from a theoretical point of view (monodisperse powder diameter / perfectly spherical particles), the maximum volume that can be occupied by these particles is 0.740. This is the so-called "maximum packing fraction". You can prove this yourself by taking a triangular prism as a unit repeating cell and calculating the volume of fractional-spheres inside it - see the picture for the two ways that hexagonally-close-packed spheres can be arranged - the math is easier for the A-B-A packing. Note that this is [I}independent of the diameter of the particles[/I], which is counter-intuitive, but nonetheless correct. In other words, no matter how small the particles are, you should always be able to get them to achieve a maximum volume fraction of 0.74.

For randomly-arranged spheres, the maximum packing fraction is usually agreed to be somewhere between 0.67-0.69. Of course, at the maximum fraction, any solder paste would just be a solid mass, and the typical volume fraction of a solder paste is from 0.4 to 0.54 by volume of solder (from 75-93%w/w powder, depending on the alloy and the application).

So to achieve the same viscosity needed for the application, you need less volume of solder powder, and hence less % by weight, if the powder is more monodisperse.

Theory is nice, but we'll talk about the "harsh realities" of solder paste in the next post.

Cheers!  Andy

A colleague in California was on the phone last week: "We need a flux with greater activity!" He said. Choking down comments like "Maybe we can take it jogging next time you're in town?", I asked him to be more specific. "You know perfectly well what I mean!" He was right. I did. Sort-of... But I was also trying to force him to think what he was asking for. Too often people (even intelligent, well-educated people) ask for a flux with a higher/greater/stronger/better activity flux, without pausing to consider what effect they really want. If you're starting to lose the thread here, maybe it's best to begin at the beginning and consider what an activator does...

Activators are the chemicals that are added to solder fluxes to remove oxides from metal surfaces, and so allow them to join together to form a strong metallurgical bond. When I first started out as a solder paste formulation chemist, I thought like many people, that the more reactive the chemical was, the better activator it would be. Also, I thought that the key to this was the pKa, or dissociation constant: equating corrosivity with activity. For a chemical of formula AB that dissociates in solution:

   AB <> A + B

Its dissociation constant, pKa, is therefore:

   pKa = -log { [A].[B] / [AB] }

so the smaller the pKa, the greater the extent of dissociation. The data, however, did not make sense when I did the experiments. Low pKa chemicals often (but not always) did not work well; and in testing homologous series' of activators: some worked well, and many did not, but always without any apparent pattern. There were three things that quickly became evident:

1/ pKa is almost always given in aqueous solution
2/ I was not focusing on a single effect (we'll talk about what I mean by an "effect" in a minute)
3/ Certain chemicals can be added to activators to enhance their effect

With the exception of certain very corrosive/reactive fluxes, water is a poor choice for a solvent, so any analysis based on (for example) acids and bases is, at best, rather simplistic, so 1/ is inapplicable. Talking about 2/ and 3/... OK, so what is this "effect"? At the time, I was focused only on solder paste, and also only on the issue of solderballing; where individual solder particles partially coalesce, leaving some individual particles behind that do not join up with the main body of reflowed paste. When I took a solder paste that gave excellent solderballing data, and reflowed it into OSP-copper, the result was often very poor wetting onto the copper surface, so this was my fourth learning, and it is the key one in solder paste formulation:

4/ What effect are you looking for?

Activators that remove oxides from the surface of solder are, more often than not, not good at removing oxides from substrate surfaces, such as nickel or copper oxides and hydroxides. There was a fifth learning, too:

5/ Activators have to do two things (see figure), not one:

- Firstly: react with the surface metal oxides to form a metal-activator reaction product (MARP)
- Secondly: remove the MARP's from the metal surface

This leads me naturally to the activator koan: What good is done by replacing an insoluble metal oxide with an MARP, unless it moves away from the metal surface? Note that these MARP's can either be dissolved into the residue, or they can be made volatile: although this latter approach does raise the question of how toxic volatile organo-metallic compounds are, and I don't believe this interesting concept was ever reduced to practise. Maybe you know better?

The next time someone asks you to formulate a flux with "higher activity", smile cheerfully  and send them a link to this blog posting (and tell 'em Andy Mackie sent you!) and finally , ask them "What is it you want the flux to DO exactly?", or to be more specific:

           * Describe in detail the soldering defect the customer is seeing
           * What solder alloy is used?
           * Wetting onto what surface?
           * Under what reflow conditions (profile / atmosphere)?

This way, you will be better able to guide the formulation chemist towards resolving the issue, rather than trying to best-guess their approach to the problem.

Cheers!  Andy

...and we're back with more nitrogen and inert soldering myths.

Myth 6: "I turned on the nitrogen flow, so my oven is now inerted"
Fact: It takes time to purge an oven down (and yes, I have an equation to determine this). The final equlibrium oxygen level, as we have seen in the previous discussion, will be somewhere between that found in the incoming nitrogen gas and the 209,000ppm oxygen level found in air.

Myth 7: "If I increase the nitrogen flow rate, I'll get a lower oxygen level"
Fact: Yes you will, to a certain point. But every reflow oven has an oxygen level below which you will not be able to go, due to turbulent mixing (see Myth 3 in the previous post) and you will never be able to reach the same oxygen level as that seen in the incoming gas, as by definition, it is impossible to hermetically seal an in-line reflow oven.

Myth 8: "I have 26 identical reflow ovens using nitrogen gas, with the same inlet pressure, and I know the ppm oxygen level in one of the ovens, so the ppm will be the same in the others."
Fact: I had a customer make this same claim to me, then tell me that he had soldering defects on some ovens and not on others. Why? Because when we went in and actually measured the oxygen level, it was different in each one - and not a small difference either. I measured from 80ppm oxygen up to 1% (10,000ppm) simply because the oven curtains were trimmed to different lengths. A power semiconductor customer was carrying out reflow with forming gas and his process started having huge voids over time. Why? He was not doing sufficient preventative maintenance, and flux residues were choking or diverting the gas flow, allowing oxygen to contaminate the process. He cleaned the oven out, and the problem disappeared.

Myth 9: "I measured the oxygen level in my ovens once, so now I know exactly what it is."
Fact: Reflow equipment "air tightness" changes over time. Amongst the factors causing this change are:
 - Flux residues can clog diffusers and block exhausts:
 - Operators can trim curtains and adjust the oven exhaust balance.
 - The copper-filled silicone gasket materials can lose pliability and harden so they no longer seal correctly.
One contract manufacturer I visited was having some reflow problems, and it was only when he took the external exhaust system to pieces, and removed the twofoot long piece of rosin that was almost completely blocking his 4inch duct that he realized what was going on. Remember, flux "volatiles" only remain volatile if they are hot.

Myth 10: "I can balance the reflow exhausts easily by having the inlet and outlet vents tied together".
Fact: This is one for the engineers amongst you. The face velocity (gas flow rate in ft/min or cm/min) into the exhausts at both ends of the reflow oven has to be the same (see diagram). Not the volumetric gas flow rates. Not the vent pipe ID (inner diameter). The reason is simple: if you have the same face velocity into the ducts at each end, you exactly balance the Bernoulli effect across the oven, so that the pressure is the same. The minute you have a difference in the face velocity, you have a difference in the air pressure, which will drive air into the oven, at a rate that is a function of the difference of the squares of the face velocities.

One final note: Inert or "nitrogen" soldering really should be renamed "low oxygen level" soldering to get the emphasis on the control variable here: the ppm oxygen (O2). From an engineering perspective it IS a critical control variable, and ppm levels within reflow ovens should be measured on a continuous basis.

Remember: a variable that is not monitored can not be controlled: just good engineering practise.

Next time: forming gas.

Cheers!  Andy

In my twenty years in the electronics manufacturing industry, I have heard a lot of claims made about the use of nitrogen in inerted soldering processes: many of them completely wrong. In this discussion, I'll use the example of reflow in an enclosed oven, although many of these discussions may pertain to wavesoldering and even vacuum soldering.

Let's start with the real reason an assembly engineer uses nitrogen in inert soldering: because it is the cheapest gas available that does not react with hot metal surfaces. to form an unsolderable film. That's it. Period. People who use nitrogen for reflow are not using it because it has any wonderful properties, they are using it because it has low oxygen and moisture levels, and can purge (dilute) oxygen down to a low enough level to prevent or slow the oxidation of metal surfaces during heating.

To understand any process using inert (unreactive) gases, you have to understand the composition of air; the most abundant gas available to us. Air is around 78% nitrogen, 20.9% oxygen and 0.9%argon, with small amounts of other gases, carbon dioxide and so on, along with varying amounts of water vapor. Water vapor may go up to around 4%, and of course, at this level, it will dilute the other levels of gases by (96/100), just in case you think there as a problem with the math. The oxygen level (20.9%) equates to 209,000ppm (parts per miilion). The ppm unit is a much more useful measure when you are down at low percentage levels, for example 0.01% = 100ppm. It is also important to note that the fractional measure (ppm or %) correlates to the amount by volume and, from the ideal gas equation, also the molar percentage.

I'll cover half of the of the myths now, and half next time.

Myth 1: "Nitrogen removes oxides"
Fact: Nitrogen used at reflow temperatures has no fluxing (oxide-removing) properties whatsoever and does not chemically react with anything with anything at these temperatures. Nitrogen prevents or slows oxidation or (in the case of a flux-cleaned surface) re-oxidation simply because is is not an oxidizing gas. Forming gas (a mixture of hydrogen and nitrogen) is very different, and I will discuss this in a subsequent note.

Myth 2: "Nitrogen improves heat-transfer"
Fact: It has no practical thermal effect on the soldering process. Heat transfer in gases at the same pressure and temperature is governed by the molecular weight of the gases: nothing else. Since nitrogen has a molecular weight of 28, and oxygen almost the same at 32, the difference in heat transfer properties between air and nitrogen is minimal whether you are talking about laminar or turbulent flow.

Myth 3: "If I measure the oxygen level in my incoming nitrogen, then I know the level in the oven"
Fact: Even an apparently well-sealed inerted reflow oven is actually mixing ambient air with your nitrogen to some degree. It does this through simple diffusion (driven by difference in partial pressure of oxygen in air versus inside the oven) or by turbulent mixing of nitrogen with air near an opening. What happens in a real oven is shown in the illustration (above). Putting a low flow rate of nitrogen into the oven will have little or no effect (a), then putting more in will reduce the level, but you will see large variations (b), then finally you will reach a plateau (c) where you have obtained the minimum oxygen level possible, but turbulent mixing is still introducing oxygen from the outside air. As you increase the nitrogen flow rate, you are simply increasing the turbulence, and hence the rate of mixing

Myth 4: "Purer nitrogen will give me better results"
Fact: Standard, cryogenic quality nitrogen has around 2-5ppm of oxygen in it. Even purging a well-sealed oven will not get you down to exactly the same level as the incoming gas. You will see no difference if you are using a nitrogen source at 10ppm or 10ppt (parts per trillion) oxygen. As you can see from the illustration above (c), the effect of the highly pure gas is completely negated by the mixing with air.

Myth 5: "Nitrogen reduces all soldering defects"
Fact: Nitrogen can help with some wetting-related defects, and can often turn a so-so (marginally acceptable) soldering process into an acceptable one. Other Fact: It may not only increase wetting ("wicking") uncontrollably onto leadframes or other surfaces, but may also cause solderspatter and contribute to die tilt (power semiconductor assembly) or tombstoning (SMT).

More next time.

Cheers! Andy

Something that seems counter-intuitive to many customers, yet seems obvious to a rheologist is why the percent metal loading of solder paste varies so much. The most crucial control variables are:

- Alloy type

- Powder size

- Application / usage

I'll tackle the two last points in a later blog entry, but the first point is fairly simple to explain. Assuming that solder powder is formed of identically-sized, unreactive, perfect spheres, the viscosity of the solder paste will depend only on the VOLUME of solder present, and is therefore related to the alloy density. This finding goes straight back to one of Einstein's first papers* where he showed that, for dilute dispersions of identical spheres, the relative viscosity (the measured viscosity of the suspension divided by the viscosity of the carrier fluid), is 2.5 (the so-called 'k' factor or Einstein constant) times the volume fraction. The picture (right) shows the definition of the volume fraction.

As can be readily seen: the metal density of solders can vary from 6.5 to 14.5g/cm3, and changing the metal density therefore necessitates adjusting the metal weight percent accordingly. This also varies with the usage:

 1/ For printable paste: a 6%w/w spread is possible

 2/ For package-on-package paste: over 10%w/w spread in metal loading may be needed

For both 1/ and 2/, of course, the goal is to maintain the SAME volume fraction, so the rheology of the solder paste remains the same.

Simple algebra will allow you to derive an equation so you can plug in any density of alloy for a solder paste and calculate the required weight percent of that metal. I can email you the solution if you're stuck, just click on the "Contact Me" button (left).

Also note that changing the flux will not only change the density (flux densities can range from 0.85 - 1.05g/cm3), but will also change the rheological properties of the paste significantly. Quick plug: Ron Lasky was good enough to give me a chance to discuss solder paste rheology a few weeks back, and there will be more about this topic in the coming months.

Cheers!  Andy

* A. Einstein, "Concerning the motion of particles in quiescent liquids as required by the molecular- kinetic theory of heat," Ann. Phys., i_7549-560 (1905)