Slashdot Mirror
The Costs of Making a DRAM Chip
Anonymous Coward writes "Researchers at the United Nations University in Tokyo studied the physical and environmental costs to produce one 32-megabyte DRAM chip. Their conclusion? The UNU team found that to make every one of the millions manufactured each year requires 32 kg of water, 1.6 kg of fossil fuels, 700 grams of elemental gases (mainly nitrogen), and 72 grams of chemicals (hundreds are used, including lethal arsine gas and corrosive hydrogen fluoride)."
dont eat any old simms or dimms then kids...
MilkMiruku
No wait, then it'll sit in a landfill.... I know, I'll BURN it!
Ever needed a better reason to avoid throwing away old hardware? Just recycle it and improve both social justice and the enviromental impact.
"To any truly impartial person, it would be obvious that I am right."
The publication itself:
Here.
Those numbers may be "used to make" a single microchip, but it doesn't say those numbers are what is CONSUMED. That's what's important... how much of that material is consumed in making a single chip.
I suspect that 32kg of water is reused for many, many chips. Same with the other material. Obviously, you'll have SOME material consumed when making a single chip, but I find it difficult to believe all that is CONSUMED when creating a single chip.
More info needs to be presented about the consumption of materials to make a chip that what is "used" to make a chip.
I would like to see *all* products analyzed like this. A producer would be required to put a sort-of "nutirition-information-style" label on all its products detaililng the environmental impact of its manufacture.
this would enable the advocates of "vote with your wallet" environmentalism to properly inform people to the point where their (ill-conceived (imho)) idea would require. I mean, what is the environmental cost of the plastic toy in your kids-fast-food meal? what about the CDs we buy? what about the thousands of other pcs of consumer garbage your average consumo-bot purchases each year..
The semiconductor business is a filthy one. As mentioned, a LOT of toxic substances are required to produce the computers that we enjoy. I don't like that fact one bit, but...
This is certainly the most effective & least expensive method to produce these things. Would you pay $129 for a piece of memory that claimed to be manufactured in an environmentaly friendly way, when the "regular" memory of the same type and size was only $59? I didn't think so. Do you think that corporations or government would pay a much higher price for what amounts to the same product? Doubt it. The key would be to produce "clean" computer components in a cost effective way. If someone could pull this off, I think that it could signal the beginning of government mandates and corporate policies requiring that all procured components come from "clean" manufacturers. But that isnt going to happen any time soon.
I'm not advocating the filthy practices, just viewing them from a practical point of view. It would take some serious R&D to come up with a cost effective and "clean" chip fab facility.
Just my 2 cents.
I'd rather be a conservative nutjob than a liberal with no nuts and no job.
Well-known flaw of the (unadorned) capitalist system. The manufacturing cost of a product does not include the costs incurred by its manufacture which are borne collectively. Example: if there was no fuel tax (and there is no significant fuel tax in the US), the cost of petrol and cars would not include the environmental cost of pollution, because it wouldn't be paid for by oil companies and car makers, but by everyone.
It's not a case of "brushing it under the carpet", more a case that the system isn't in place to make manufacturers accountable for the collective costs caused by their products. Such a system would be an extra tax on polluting products or tax incentives for less polluting ones. This is all old hat but governments tend to be too scared of being accused of being "anti-business" (or "anti-american" ;-) ) to actually implement such measures.
700 grams of nitrogen? But....the atmosphere is, like, 78% nitrogen or something. What happens if we use it all up? We'll have only 22% of the atmosphere left, and that's mostly oxygen, which is a lethally corrosive and highly flammable gas. We'll all die!
(as the actual paper requires an ACS registration, which I don't have...)
C oreMem ory.html
The total weight of secondary fossil fuel and chemical inputs to produce and use a single 2- gram 32MB DRAM chip are estimated at 1,600 grams and 72 grams, respectively. Use of water and elemental gases (mainly N2) in the fabrication
stage are 32,000 and 700 grams per chip, respectively.
Plain english:
Energy consumed to create chip: approx 1,600g of fossil fuel.
72g of "chemicals", unknown recoverability.
Nitrogen and Water use (resuable), 32,000g and 700g.
So, it takes energy, reusable chemicals, and some (potentially) non-reusable chemicals.
As miniturization increases, so will the mass ratio (what is being compaired in the article) of the output versus the necessary inputs to manufacturing.
What do you thing the product weight of a 32M magnetic core memory (old school memory) would be? Pretty darn high. Manufacutring cost, not as high.
Core memory ref:
http://www.science.uva.nl/faculteit/museum/
this sounds like a 'worst case scenario' type of analysis.
I'm not denying that the chip industry isnt doing Mother Nature any favors, but what exactly do these numbers mean?
I mean, I hear from environmentalist types that every glass of water you drink takes 2 glasses to wash and another 2 to rinse it. But, the water doesnt dissappear or become unusable. It makes its way back into the system.
So of 32 kg of water 'used', how much of that becomes contaminated to the point that it cant be re-used? If its a coolant that evaporates as steam, then I don't see the big deal. If its turned into toxic sludge with a half life measured in eons, then it probalby is.
And WRT to fossil fuels, are they directly used in manufacturing, or are we talking how much needs to be burned to create the electricity needed to manufacture? And why talk about fossil fuels, and not Uranium or solar/hydro/wind power? Because it gets more attention? Wouldnt kW/h would be a better measure? What matters is how much energy is expended.
I understand that we need to better watch and control our impact on the environment, but infactual data and meaningless statements like 'it takes 300 bananas to make a wingnut' don't help.
I don't need no instructions to know how to rock!!!!
My god, at this rate RAM production will consume all of Earth's resources! They must be stopped before Earth is transformed into a floating mass of DRAM!
...programmers consumed 2700 hamburgers, 1503 pizzas (276 vegetarian, 1227 containing meat or meat by-products), 16204 cans of soda (assorted), 790 bags of microwave popcorn and 1 office chair. Three programmers were temporarily blinded while downloading movies & images for research. No animals were harmed...Oh, hang on, I forgot about the ones in the burgers & pizza.
Modest doubt is called the beacon of the wise. - William Shakespeare
In 97' I worked at Samsung's fab in Austin, Texas as a chemical technician, troubleshooting and maintaining the pumps that sent liquid chemicals up to the fab. I also pushed a lot of drums and hooked up tanker trucks of sulfuric and other nasties to the hungry fab.
As the average slashdotter knows, every chip is composed of multiple layers, each masked and etched, bathed in various acids and bases and then neutralized and cleaned before the next layer can be applied.
Then these waste chemicals are pumped out, neutralized (in theory) and diluted before being dumped into the same waste water stream that eventually hits streams, rivers and ground water.
There's a whole lot of water indirectly consumed in the manufacturing process - but a whole order of magnitude more water consumed and dumped to dilute the hopefully neutralized (ie, salts) waste products.
So I believe the numbers - kgs (ie, liters!) of water per MB does not set off my bullsht detector.
To me, it also brings into question the whole drive of chip research. It's all focused on performance. There are some articles on research into environmentally friendy chips. But when did you hear of a chip marketed as enviro-friendly? We're tempted into buying the another chip just a tick faster but not even given the choice. For consumers to even be able to make the choice for a more sustainable product we have to have the information.
But companies don't even want us to know what we're injesting - that isn't important to them and is contrary to their creation of demand for more stuff. Why would we think they would tell us something against their own short-term interest?
. This sig unintentionally left blank. I meant to put something here, but I'm busy.
The article provides some details -- the most vital of which were echoed by the submitter -- but doesn't give us any clear idea of how good or bad this fact is. How does the environmental impact of microchip production compare to other goods?
Fortunately, the study itself -- linked to by another poster first -- provides some more useful details.
This is more useful than the article, but still does not give a clear idea how microchip fabrication stacks up against lower-tech items in terms of environmental impact. I mean, that automobile that he uses as an example is an non-trivial machine. More to the point, all modern cars incorporate microchips. In order to properly compare the environmental impacts of car and microchip fabrication, you'd have to factor in the environmental costs of all of their respective parts. I'll bet that a car has a much higher environmental impact once you add in all its microchips, pieces of plastic, and so on.Furthermore, both microchips and cars have a greater environmental impact than merely that caused during their production. In both cases, you should also consider what sort of impact their use will entail. Microchips require electricity to function; that electricity has to be generated somehow, and the methods of its production have an environmental impact. Microchips also need to be disposed of once they are no longer useful, as happens all to frequently. I personally have found a good computer recycler, but lots of other pieces of equipment are thrown into landfills, where they remain indefinitely. They may also leak toxic substances as they begin to fall apart (Lead from CRTs, for instance.) Likewise cars have a HUGE environmental impact during their use -- just think how much gasoline a car can burn in a year of normal use.
But I digress. The study did not consider the entire lifetime of the chip, merely the circumstances of its production. In which case, I find it less than satisfactory. It's a good starting place, but doesn't follow through.
The production of microchips is not environmentally friendly. This is true. What we need to know now is how dirty the process is, and how great of a problem it is compared to other areas of production. Comparison with a car alone isn't too useful, especially as it doesn't figure in the environmental costs of the car's components. What would be useful would be a comparison with lots of other objects, ranging in complexity from a table knife to a bicycle to, say, the space shuttle, with the environmental costs of the components of the more complex items figured in. Then we could use that study to see what areas are worst, and where we most need to improve.
Lastly, lest I sound too harsh, the article does mention that this is only the first installment of research that has taken several years to complete. It is entirely possible that the team will put out exactly the sort of report I envision here sometime in the future. So overall, I'd have to say this is a good start, but needs a lot more analysis to be especially useful.
Are we supposed to feel guilty because of how expensive we or our tools are in terms of environmental impact?
Guilty, no. Responsible, yes. There are a bunch of non-human, low-intelligence animals on this planet which don't have the capabilities of protecting themselves from us. Free exchange of information is nobel; being responsible caretakers and guardians of the environment is also nobel.
Do you think an environmental impact study was done before the Mona Lisa was painted?
Yep. 2000 years ago, the Romans had environmental impact studies.
Pliny reports on ecological disasters and effects of pollution from refining of metals in his Natural History (check books 8, 11, 19, & 33).
Strabo reports on the effects of clearcutting forests for fuel and on pollution from refining in his Geography. (14.6.5; 3.2.8)
Xenophon reports on pollution from refining of silver in Memorabilia. (3.6.12)
Lastly, Plato talks about the deforestation of Greece in Critias. (111b-c)
Yeah, but what's the cost of not making the chips?
Suppose we threw out all the chips - went back to pencil and paper? How many kilowatt-hours would we consume in heating and lighting the rooms full of green-hatted accountants scratching figures onto paper with pens?
OK, perhaps that's a little too far. (But lots of enviros really hate it when we take their premises - that chipmaking is Evil - to their logical conclustions.)
Suppose we just threw out the 32M chips and 8-inch wafers and 0.13u processes. No new fabs after 1995. We'll stick with 4M chips on 4-inch wafers and 0.35u instead. That would give us a quarter of the memory (and our CPUs would top out at about 300 MHz), and (ta-dah!) use pretty much the same amount of resources as we're using today.
Throwing away that fab that builds 80486 chips and 16M sticks of FPM RAM (and throwing away the products it produced), and replacing it with a fab capable of cranking out 2.4G P4s and 512M sticks of DDR is a good thing, because you can do more with the P4 than you could have dreamt of doing with the 486s.
For running Office, maybe a '486 would be OK. Forget about Doom III, though. Or rendering Lord of the Rings.
And if those aren't "green enough" things to justify building faster/better computers (because, after all, if it's not Greener Than Thou, you Just Shouldn't Do It, Ever!), I'll remind you that you can also forget about the climate simulations and ozone hole analysis, and image processing for weather prediction and crop analysis. Scrap the weapons technology that turns "dumb" 500-pounders into GPS-guided missiles so that one bomb can do the job of 100 - back to carpet-bombing a whole city to powder with a fleet B-52s to hit just one bunker. No more passenger airplanes with wing and engine designs for low fuel consumption and low noise. No more fuel-efficient combustion chambers that help you get more power out of your 4-cylinder than your uncle got from his '68 Malibu. Gotta save the environment, y'know!
32 gallons of water: Needed to make an ounce of beef
1.6 kg of fossil fuel: needed for 3 pounds of beef
72 grams of chemicals: Needed to produce 2-5 grams of beef
So may i ask, "where is the beef"?
With the upcoming superbowl, I sure do appreciate seeing folks warming up their armchair quarterback skills.
Short of weather, taxes, sports and personal hygiene, it seems like environmentalism just brings out the stupidest and hastiest when it comes to holding-forth-like-an-expert.
I mean, I've just read comments from people that worked in a fab (who claim to therefore know all the details of the fab's environmental remediation processes), people inventing an environmental impact metric based on goods/fuel ratio comparisons between cars (largely steel and plastic, with a per-device weight in the tons, and ironically containing many microchips) and microchips (which weigh tens of grams... the comparison is ABSURD), and lots of people advocating all sorts of half-assed remedies.
It's good to explore ideas, but frankly I haven't seen this much evidence at how unscientific techies can be since I taught a freshman physics lab. C'mon, be as critical of your own methodology as you are of the facilities involved.
The fabs I have toured or audited all had room for improvements, but seemed to:
- Have existing and prototype materials-reuse mechanisms implemented to minimize environmental impact. Solvents, the most obvious and arguably the most hazardous, almost always cost so much in terms of purchasing and RCRA-compliant disposal, that a distillation or recovery mechanism costing six figures (dollars) pays for itself easily. This means there are financial benefits and PR benefits, so companies are very open/willing to clean things up.
- Admittedly use an insane amount of water. A large chunk of this is a byproduct of Reverse Osmosis distillation to get water to Megohm pure and better. My point is, the water isn't just pumped thru their wastewater stream to dilute things. It comes in, is superduper-distilled (basically), and then used at an insane rate for processes & rinsing. Water consumption is the biggest environmental problem of most fabs, but the problem isn't how dirty they make it... it's the regional impact of so much water being consumed.
- Either directly treat all wastewater (including their own special steps to precipitate out metals or other problem materials, and are constantly testing/evaluating water quality) or discharge it to a community-owned facility that they work extensively with (to get all the above items). My experience is that much of the water pollution is precipitated out, sludge-pressed, and shipped/handled as low-grade hazardous waste.
- Are, by all the environmental engineers I've ever worked with, greener in most every sense of the word than most other industries. By this I mean the staffs always seem to be proactively reducing their environmental impact. They've started since the US's environmental awakening around 1970, so they don't have to struggle to keep up with competitors grandfathered in doing things some old/cheap/dirty way, etc.
Last of all, the head story mentions HF and arsine. I've been out of this a long time, but if memory serves both are very reactive in a way that they readily degrade into safer compounds and are generally considered to have *NO* long-term environmental impact. They can't survive in the wild enough to be a community/wastewater/landfill concern. The moment I hit this part, I felt like I was reading an econut's rant about highly-radioactive long-lived isotopes... all scientific credibility goes to hell when you spout off half-truths to make a headline. The only people that need to worry about HF or AsH3 are people in the room when it leaks and emergency responders. Anyone else (even a block away) has zero risk short- or long-term to these. Nasty? Hell, yes. Silane (common in fabs) scares me even more (it absorbs thru tissue and makes swiss cheese out of your bones, I'm told). But a community's worst fear from their local fab should be DNAPL's (Dense Non-Aqueous Phase Liquids). TCE, Perc and other DNAPL's can pollute a town's groundwater for a few hundred years, costing the town tens of millions of dollars for air-scrubbers or other remediation hardware.Just to dodge the karma damage a bit, I'm very very much an environmentalist. But I'm an engineer. And I feel environmental protections suffer when people use half-truths and poor science like this. We need to treat it like racism or other societal ills... question everything (including proposed remedies) and stick to an ethical high road that demands that we NEVER sneak by a scientific half-truth. Otherwise, we risk losing our credibility and accidentally creating a legal framework that strangles the innovations and self-improvements we need to advance.
</soapbox>
---advaitavedanta