Boron arsenide crystals can dissipate the heat generated in electronic devices | Hacker News Reader

Boron arsenide crystals can dissipate the heat generated in electronic devices

cacm.acm.org

54 pointsrbanffy19 comments

"In the future, further studies which will enhance the growth of the crystal, as well as explore potential large-scale applications"

Hinting at awesome GPU cooling and such, with not one mention of the materials exceptional hazard.

Boron Arsenide

GHS Hazard Statements

H301 + H331 - Toxic if swallowed or if inhaled

H410 - Very toxic to aquatic life with long lasting effects. Safety Precautions

P261 - Avoid breathing dust

P273 - Avoid release to the environment

P301 + P310 - If swallowed: Immediately call a Poison Center or doctor/physician

P311 - Call a Poison Center or doctor/ physician

P501 - Dispose of contents/ container to an approved waste disposal plant.

https://www.azom.com/article.aspx?ArticleID=8425

This material is hardly exceptionally dangerous. Its close cousin Gallium Arsenide has been nearly the exclusive source of Yellow, Red, and Infrared LED's since the 70's. The screen you're looking at now would not exist without arsenide compounds, both for lights and for other circuit components.

While you could probably make yourself sick by taking old chips of this nature, grinding them up, and inhaling or eating them, or engaging in an accident with their raw materials during their manufacture, they rank pretty low on the list in terms of practical toxicity. Arsenic is a common trace element, and even (in small quantities) has been used to ward off disease in chicken and cattle.

It's not good for you, but this is a positive development, not a poisonous watershed.

If not limited to trace amounts for microscopic components, in the heat spreading role there could be demand for it in much greater quantity than elementally similar LED pigments.

Most manufacturing materials do not require such careful treatment and disposable -thankfully. The concern is not just about unusual behavior such as 'grinding up and eating'. This is a particularly hazardous material - as documented.

You're exaggerating how dangerous BAs would be in this situation. As far as I know, useful single crystal boron arsenide is very expensive to make in industrial quantities and it's too brittle to be used as a large heatsink - at best you can grow a several micron thick layer of it on a semi-flexible substrate like for solar panels. Instead, I'm guessing their design uses small quantities of single crystal BAs to interface between the wafer and the rest of the chip package so it would better transfer heat from the internal silicon to a regular heatsink. Thermal conductivity at that interface is a hard limit to how fast the heatsink can draw heat from the chip package and since its contained, toxicity is largely a concern for recycling.

Its a concern for manufacturing, containment and disposal. In few places public disposal and recycling facilities guarantee complete compliance. The materials hazard specs are not exaggerated, they are intended to be taken seriously as they are stated.

"Instead, I'm guessing their design..." There were no designs indicated here, only research on the crystals thermal properties with the vague mention of "electronic cooling" and "potential large-scale applications"

Absolutely, arsenic-containing materials have the potential to be extremely dangerous. Unscrupulous GaAs manufacturers have put their employees at risk in the past.

[0]: https://www.eastbayexpress.com/oakland/the-axt-way/Content?o...

There are many compounds that dangerous and it isn’t like you are eating it. It stays inside the tubes.

When electronics are thrown out and they either degrade or are burned for precious metals, those compounds are released into the water, soil and air.

While this is a real problem, this is not unique or even new. Electronic waste contains a number of toxic substances, and must be disposed of separate from the normal household waste stream. While acknowledging that this may make the problem worse, it also promises significant benefit.

Is the consumption or grinding and inhalation of GPU coolers a pressing problem?

The smaller the problem, the bigger the outrage.

Isn’t everyone flipping out over drinking straws these days? I mean non disabled people for which straws are extremely helpful of course.

Plastic straws really aren't a 'small problem'. They're everywhere, meant to toss after 1 use, cannot be recycled in the traditional sense, and do not break down in nature. Is it currently the largest pressing issue humans face? Not really. Is it an issue that is within our means to still address? Yes.

I come across maybe half a dozen straws a year.

But then maybe the real problem is most people eat way too much fast food? Starbucks is fast food too.

Yeah, it’s the straws we need to stop... /s

The original article and the "full" article don't actually list any numbers. Looks like the performance is 1000 ± 90 W/m/K [0]

For comparison, a traditional high-end thermal paste may have 10 W/m/K, a liquid-metal thermal paste is up to 73 W/m/k [1]. Copper is 385, Aluminum is 200, Diamond is 1000.

[0] http://science.sciencemag.org/content/early/2018/07/03/scien... [1] http://forum.notebookreview.com/threads/liquid-metal-showdow... [2] http://hyperphysics.phy-astr.gsu.edu/hbase/Tables/thrcn.html

> Diamond is 1000.

The paper itself gives diamond as 2200 and isotopically pure diamond has even higher conductivity.

The question is whether it's any easier to manufacture than a diamond layer. Edit: The paper claims as much.

There are interesting non-metallic materials already in large-scale production which serve as excellent heat sinks and pipes. As an example, pyrolytic graphite has a thermal conductivity several times greater than that of Cu: https://www.mouser.com/m_new/panasonic/panasonicthermalgraph...

Heat sinks and pipes don't matter all that much since you can simply increase the area to achieve higher conduction. The bottleneck is getting it from small chip hotspots to the heat spreader on top of the chip. So this would sit as a layer directly on top of the chip beneath the more conventional heat spreader.

So how would such crystals be used? Fairly large crystals as shims between the tiny silicon chip and a larger heat spreader? Or lots of microscopic crystals as part of a TIM paste?

Does this mean we can now have fully 3-dimensional ICs? (I imagine with stacked layers of silicon crystal).

Reply on news.ycombinator.com