MesoGlue will not replace solder
Paul Rako - September 23, 2016
There has been a minor media frenzy over some very good scientific work at Northeastern
University. Researchers discovered that you can put well-spaced silver nanowires on two surfaces
you want to bond; one set of nanowires has an indium coating, and the other set has a gallium
coating. The nanowires have to be well-spaced since you shear the two surfaces together a bit to get
the wires to stand on end. If the wires are well-spaced, there is room for them to mesh together and
form a metallic glue, called MesoGlue (Figure 1).
Figure 1 MesoGlue is comprised of silver nanowires. One set is coated with indium, the other with
gallium. When they touch the indium and gallium form a liquid, that later solidifies as they further
intermingle (Source: Northeastern University).
As every Russian schoolchild knows*, the eutectic point of an indium and gallium alloy is 15°C
(Figure 2). This is below room temperature. The delightful thing about this discovery is that the
indium-coated silver nanowires will form a eutectic alloy when they touch the gallium-coated silver
nanowires. Now you have a room-temperature conductive metallic interface. Better yet, the indium
and gallium will continue to diffuse until the alloy moves off its eutectic point, and then it becomes a
solid at room temperature.
Figure 2 The indium-gallium phase diagram shows the eutectic point is below room temperature,
only 15°C (Source: Harvard University).
The non-technical media are ranting that this will replace soldering. This is what happens when PR
departments and media treat a scientific breakthrough as an engineering breakthrough. It starts as
a slick glossy abstract (PDF) by the researchers, no doubt with an eye out to department budgets
and commercialization. Then the Northeastern University press office gets a hold of it and it might
replace solder. A materials trade-paper (23M PDF) picks up on the press office release, and there is
a bit more amplification on how this will change the world. Pretty soon the mainstream media pick
up the story and its breathless headlines: “Will This Fancy Metallic Glue Kill Soldering?” I love the
comment to one fan-boy article, “This seems like a game changer.” No, not really.
As a writer, I note that stories go from scientific jargon laced with passive voice to click-bait titles
asking a rhetorical question. All are no-nos in good technical writing. Fran Granville, my copy editor
when I worked at EDN full-time, taught me to spot these technical writing shortcomings and to avoid
them in my own work.
Engineering is science intersected with economics. That is why engineering is harder than science,
and that is why it pays better. Engineering is also a continuum of solutions. We already have
conductive silver-filled epoxy, and conductive nickel-filled epoxy, and I assume the boffins have
whipped up conductive carbon nano-tube epoxies. Scientific papers don’t provide the context for the
application. Engineers need to see a comparison of the costs and times and rework procedures, and
reliability, and a dozen other things, including toxicity to assembly workers and harm to the
environment.
I applaud this discovery, and readily admit it is some brilliant science. But the engineer in me asks
the universal first engineering question: “What does this cost?” The silver is expensive, and indium
and gallium are expensive. You make the well-spaced silver nanowires using glancing angle
deposition (GLAD). That is a variant of oblique angle deposition (OAD). Both techniques appeared
around the year 2000. Since this physical vapor deposition is a vacuum process, the engineers
amongst us cringe. It takes a long time to pump down a chamber. If there is plastic in the chamber,
now you have volatility that increases the pump-down time and pollutes the chamber. Time is money.
Time that ties up million-dollar machinery is big money.
Apparently, the researchers do not plate or deposit the indium and gallium. The original abstract
states that they are using core-shell nanorods. There is little detail in this, I assume since they are
furiously trying to patent anything they can. Universities are not bastions of pure research anymore.
They are start-up incubators. This is because engineering, not science, is where the money is.
Everybody likes money, even tweed-jacket suede-elbow-patch college professors. No matter, the only
salient fact is that using GLAD to make core-shell gallium-silver and indium-silver nanowires has to
be even more expensive than making pure silver nano-rods.
This also raises the engineering concern of process control. You have to ensure that the silver wires
are the proper size and proper spacing (Figure 3). You have to confirm the repeatability of the
indium and gallium coating of the nanowires. You also have to ensure that the initial shear
movement is perfect so the wires stand on end and intertwine. You have to control the purity of the
materials, and the time, vacuum, energy, and other variables in making the coatings.
Figure 3 MesoGlue uses silver nanowires made with glancing angle deposition (Source:
Northeastern University).
To the credit of the team at Northeastern, they use the process for the application of CPU cooling
(Figure 4). They state that the MesoGlue is 10- to 20-times more thermally conductive than thermal
grease. The Advanced Materials & Processes article notes the Northeastern researchers achieve a
CPU temperature reduction of 8°C. This would increase device lifetime, or more likely, reduce the
size of the heatsink to offset the MesoGlue cost. If the MesoGlue is as good or better than silverfilled epoxy, but can be removed and reworked, well that is a big deal.
Figure 4 One application the researchers posit for MesoGlue is the replacement of thermal grease
in CPU-cooling heat sinks (Source: Northeastern University).
This CPU cooling application raises its own engineering questions. I assume under the heat of
operation the indium-gallium alloy re-liquifies. This might be great if it would allow for different
coefficients of thermal expansion (CTE). But those silver fingers are a mechanical interlock, so now I
would worry about this liquid stew allowing movement that breaks the wires or ruins the interstitial
mating of the two sets of wires.
Also a mystery is the ability to repair or rework this system. If I heat the interface so the indiumgallium alloy melts, it may still take a ton of force to separate the two components. And since the
eutectic alloying is a one-time event, it's obvious that you can’t reuse the bond; you have to scrape it
off, put the parts back in the vapor deposition machine and start over.
There are a lot of good engineering principles on display here. After decades in the industry, I was
astonished to learn that the thermal conductivity of copper is 8 times better than tin-lead solder.
That is the brilliance of MesoGlue. It lets silver wires carry the heat, and the wires present a lot of
surface area to the indium-gallium alloy to transfer the heat from one set of wires to the other.
Mentioned tangentially by the discoverers, hermaticity is one feature of this process that may be the
killer application. At room temperature there will be voids in the metallic silver-indium-gallum stew,
but at 100°C and a pressure of 9MPa (1300PSI), the bond becomes void-free (Figure 5). A
conductive hermetic seal that can be created at lower temperatures might be a great niche for some
applications.
Figure 5 At room-temperature bonding there are voids in the MesoGlue (a). These voids disappear
at a temperature of 100°C (b) (Source: Northeastern University).
Glue is great, no doubt. Glue is replacing spot-welds in cars. You can use thinner, lighter sheet metal
and the glue distributes the load transfer so there is better structural integrity between panels. This
also has downsides as the glue for cars is not conductive, so now there is less RF shielding and you
have to provide ground wires instead of using the spot-welded chassis. With engineering, it's always
some complex interrelated set of compromises and tradeoffs. It's another reason I consider
engineering to be harder than science.
While I applaud the science of MesoGlue, it's still not ready for primetime as an engineering
breakthrough. When I can order it from Digi-Key or Arrow, and when its cost will justify the
improvement, then it's of interest to engineers. Time will tell, and keep your eyes peeled, but don’t
think glue is replacing solder any time soon.
*In the 1960s, at the height of the Cold War and after the embarrassment of Sputnik, my brother had
a book with the blurb, “Every Russian schoolchild can understand this book. Can your children?” I
think the book might have been Mathematics for the Million, but I can’t find an example. I love the,
“we’re falling behind the Russkies,” Dr. Strangelove aspect of this. Interestingly, when I worked in
Silicon Valley, I met several Russian engineers and scientists, and they do have a better education in
math and the basic sciences. If anyone remembers the book that had this blurb, please leave a
comment below.
Also see:
●
●
●
●
Cold solder joints
Flux residues can cause corrosion on PCB assemblies
Failed solder joint makes car clock go dark
Hot, cold, and broken: Thermal-design techniques