Ask HN: Environmental Cost of Semiconductors?
Is there a recent review of the environmental cost of semiconductor electronics covering the full life-cycle from resource acquisition (including mining footprints) through manufacture (including release of greenhouse gases) through disposal or recycling?
18 comments
[ 3.5 ms ] story [ 39.2 ms ] thread* The environmental impact of mining, refining and shipping of the ultra pure silicon needed for semiconductor development.
* The energy required to research/development then produce next generation semiconducting and at what yield or adoption rate would it make sense to push Moore's Law.
* Percentage of electronics that end up in landfill verse what is recycled.
I am having trouble trying to think of other clear examples beyond the fab due to too many variables. How do distinguish processors and GPUs in someone's gaming rig using power from a local coal plant to a cloud farm that is using wind and solar to power it? To throw another wrench into this there is also youhttps://en.wikipedia.org/wiki/Jevons_paradox.
If you were looking for a project it would probably be a good one, I for one would be interested in reading something that was backed up by solid research and objectivity.
[1] http://svtc.org/resources/reports/
1. Water
Water is extensively used and it has to be as pure as you can imagine. This requires a regular investment in filters and energy. I wouldn't be surprised to see hundreds of water wash steps for a wafer of chips, plus water wash steps for the equipment.
2. Acids and bases
You can dump these in the third world, sure, but neutralizing them tends to be favored, and also very expensive. Sulfuric or Hydrofluoric acid may require almost zero energy input, but the caustic soda, soda ash, or baking soda you need to add to passivate their waste requires a lot.
3. Silicon itself
This one's a bit of a surprise. A gram of hyperpure silicon doesn't really take that much energy, or even generate all that much waste. There's the reduction cost where you burn out the oxide with coal and wood chips, plus the follow-up stages of purification and crystallization. You lose half again by wafering. But even if it's 10 grams waste per gram of chip, you're not doing all that poorly. An idling car engine makes 10 grams of waste in no time. Plus the waste is almost entirely mg-Si, CO, and CO2; very manageable. China has a bit of a silicon-tetrachloride problem from some of its purification steps, but that stuff is avoidable.
4. Photoresists
These are the absolute worst. Carcinogenic and complicated to manufacture.
5. Specialty gases
What's a little diborane, phosphine, or arsine among friends? These players are bad news, but you use such trivially small quantities of them that their impact is pretty light. You still have to scrub their waste gases out before you release them, and that's a bit of a pain.
6. Bulk gases
Hyperpure nitrogen, oxygen, and argon get used in large amounts as well but they're almost as cheap as air in bulk. Release at will.
7. Air handling
Cleanrooms require more airflow than some bitcoin farms, and specialty filters have to be changed out occasionally. It's just electricity though.
8. Packaging
All that copper, tin, silver, and gold, plus whatever weird plastics they're using these days may end up being far worse for the environment than the chips they encase.
9. Actual use
Unless they're generating power (PV) they're going to suck down power, whether in your cable box set to off or the switchgear. This is by far the biggest energy consumption associated with semiconductors.
10. Vacuum!
Along with to a lesser extent compressed air, vacuum greases and machinery can get up there in terms of cost. Roughing pumps will spew oil, and turbo pump and cryopump units cost a lot to build, run, and maintain.
11. Organic solvents
Cheap to make, expensive to dispose, doubly so when you mix them with some of the other expensive-to-dispose stuff.
So in the meantime I write apps.
That's disappointing! For an industry that's at the heart of so much modern technological advancement, it's surprisingly conservative. And that goes double for choosing suppliers, unfortunately. Myself, I got worn down by the hours and left the industry.
The good news is that electronic components often obey commodity pricing. And for commodities there is a very good approximation for LCA: their market price.
An LCA is a sum environmental impacts, weighted by importance. Market price represents some sort of consensus weighting. You can argue for other weightings, but such arguments are at least as subjective as the market weights and seldom give a significantly different result.
Interestingly, even non-commodities reveal their environmental impact through their price. The difference is, their prices include all downstream environmental impacts, whereas traditional LCA and commodity prices only reflect impacts within some "system boundary".
If you mean "secret", then we're not going to progress much. Impact on environment is not like impact on your finance. Your finance are private, the environment is everyone's. Therefore, everyone's must have access ton the information needed to assess if the impact is acceptable or not.
In the USA, corporations fight right to repair laws. That's the current state of the affairs. It has to transition from that into a more thoughtful usage driven buying/selling process.
There is nothing wrong in buying newer, better electronics - but junking the old ones as e-waste is absolutely wrong.