An excerpt from Our Web of Inconvenient Truths:
Semiconductors
To process and store data and provide memory and apps, a smartphone’s low power microprocessors, optical, GPS, accelerometers and other sensors, transmitters and receivers (for cellular, Wi-Fi and Bluetooth signals), and noise filtering microphones all require semiconductors.[i] Semiconductors are special materials (usually silicon) that can control the flow of electricity. Transistors, the basic building blocks of a computer, are made from semiconductors.
Integrated circuits (ICs or computer “chips”) are made on wafers of highly purified silicon. Because manufacturers can create much smaller electronic components than they could two decades ago, the industry now markets hand-held and wearable computers that each contain millions of transistors.
Every microprocessor or memory chip begins on the top of a silicon wafer, “one of the most highly refined artifacts ever created by humans. Converting quartz to electronics-grade silicon wafers ‘guzzles’ electricity[ii] and involves highly toxic intermediate compounds.”[iii] In 1997, the Silicon Valley Toxics Coalition reported that processing a single, eight-inch, silicon wafer into microprocessors (each containing thousands to millions of transistors) required 4,267 cubic feet of bulk gasses, 27 pounds of chemicals, 29 cubic feet of hazardous gasses and 3,023 gallons of de-ionized water. In 1997, production of an eight-inch wafer generated 3,787 gallons of waste water,[iv] negatively impacting the health of waterways and communities near factories.[v]
In 2013, manufacturers began producing far more transistors than farmers grow grains of wheat or rice.[vi]
How are silicon wafers manufactured?
Tom Troszak, machine and process designer The microprocessors (integrated circuits) in electronic devices and the little squares in solar panels are both made from highly purified silicon. The silicon used for electronics can only have about one (non-silicon) impurity atom per billion. Solar-grade silicon can handle about one impurity atom per million—although purer silicon can make solar photo voltaic (PV) cells more efficient.[vii] (Read more about solar PV systems in Chapter 10.) Manufacturing ultra-pure silicon wafers for electronics or solar panels takes five steps:
Step One The raw materials needed to manufacture pure silicon wafers are collected, processed and transported. Pure quartz gravel (not sand) is mined. Slow-burning wood is harvested and ground into chips.[viii],[ix] A carbon source (usually coal, charcoal, petroleum coke or metallurgical coke) is mined or manufactured.[x] Coal is washed to minimize its sulfur content; residual sulfurous coal dust is sold to utilities to burn as fuel. The quartz and wood are washed to minimize contamination of the silicon. Trucks, trains, and/or ships transport all of these materials—often internationally—to a facility with a submerged-arc furnace.
Step Two produces metallurgical grade, ~98% pure silicon. Since 1905, silicon has been smelted in a submerged-arc furnace, where quartz (silicon dioxide) is melted along with wood chips and carbon at greater than 3000 degrees Fahrenheit (1649 degrees Celsius) with a continuous electric arc.[xi] Since there is not enough air in the molten puddle cavity for the carbon to “burn” normally, the white-hot carbon “steals” oxygen from the silicon dioxide, leaving carbon monoxide and a liquid silicon metal mixed with slag (melted impurities; residue from smelting).
Several chemical reactions happen in the furnace in Step Two. If all goes well, the overall reaction is SiO2+2C=Si+2CO. Most of the carbon monoxide (CO) burns to carbon dioxide (CO2) in the air above the melt, and is released to the atmosphere.
Usually, a silicon furnace runs continuously, and the “pot” is rotated occasionally to distribute the heat and slag evenly. Raw materials are regularly added to the top of the pot. The molten silicon settles to the bottom of the pot, where it’s tapped out into a ladle, fluxed to separate the slag, and poured into molds every few hours. After it solidifies, the silicon is broken into pieces, then bagged for shipment to a polysilicon manufacturer.
Typically, Step Two takes up to six metric ton of raw materials to make one metric ton (t) of silicon. A typical furnace consumes about 15 megawatt hours of electricity per metric ton (MWh/t)[xii] of silicon produced, plus 4 MWh/t (for ventilation and dust collection), and generates tremendous amounts of CO2.[xiii]
Step Three converts metallurgical grade silicon—(mg)Si—into polysilicon—(poly)Si—also known as “electronic grade” silicon (for computers) or “solar grade” silicon (for photo voltaic panels) by a vapor deposition process. Step Two’s (mg)Si is crushed and mixed with hydrogen chloride (HCL) to synthesize trichlorosilane (TCS). Once purified, the TCS is sent with pure hydrogen to a “bell jar” reactor. Inside the reactor, the gas mixture passes over very slim rods (filaments) of previously-made, ultra-pure polysilicon pre-heated to about 2012 degrees Fahrenheit (1100 degrees Celsius). Some of the silicon in the trichlorosilane “clings” to the heated filaments, causing them to grow slowly into large rods of ultra-pure polycrystalline silicon. This process is kind of like growing rock candy on a string.
Typically, each batch of rods takes 80 to 105 hours, roughly three to four days, to produce polysilicon. It can take anywhere from 1.5 to 4.5 tons of metallurgical grade silicon to generate one ton of polysilicon, depending on how much (if any) of the reactor’s silicon tetrachloride ‘exhaust gases’ are distilled and converted back into reusable TCS.[xiv]
The large polysilicon rods are removed from the reactor, then sawed into sections or fractured into chunks. The polysilicon is etched with nitric acid and hydrofluoric acid[xv] to remove surface contamination. Then, it is bagged and boxed in a chemically clean room and sent to a crystal grower for Step Four.[xvi]
Step Four is a single-crystal growing process. First, the polysilicon chunks from Step Three must be re-melted in an ultra-pure quartz crucible in an inert atmosphere. Then, a small seed crystal of pure silicon is lowered into the rotating crucible. As the seed crystal is slowly withdrawn, a single crystal forms from the tip of the seed. It continues to grow into a cylindrical ingot, leaving most of the non-silicon impurities in the pot, and aligning the crystalline structure of the silicon. After it cools, the ingot’s (contaminated) crown and tail are cut off. The ingot’s pure silicon center is then ground down to remove more surface contamination. The remaining portion is labeled as electronic grade, solar grade, or scrap. Ingots of sufficient quality are sent to a slicer.
Step Five Like a loaf of bread, the silicon ingot is sliced into wafers. More than 50% of the ingot is lost in this process: it becomes “sawdust,” which cannot be recycled.[xvii] Electronic grade silicon wafers are left round to produce as many integrated circuits (computer silicon chips) as possible. Solar silicon wafers are cut square, so that the greatest number of “cells” can fit into a solar panel. The electricity required for the ingot/wafer/cell process is often higher than the process of making polysilicon.
Next, electronic and solar-grade wafers will have key materials and components chemically embedded into the silicon, layer by layer. Dozens of different layers must be produced during hundreds of additional processing steps to turn each “electronic grade” wafer into microprocessors—using a great deal more energy and toxic chemicals. Solar cells also require many additional processing steps.
Why does manufacturing silicon require coal?
Silicon is a very reactive element, so usable “metallic” silicon cannot be found anywhere in nature. Quartz containing silicon dioxide (silicon ore) is common in nature, but the commercial “carbothermic process” of separating silicon from oxygen requires a prodigious amount of carbon and a lot of heat—hence the name. Coal is a commonly used source of carbon for the process; but raw coal, pure enough for silicon smelting, is only available from a few of the world’s mines. Metallurgical coke is “coked” (cooked) from lower grade coal to remove sulfur and other impurities. The cooking process releases these impurities to the atmosphere; and about 40% of the original coal is lost in the process. Petroleum coke (made from Canadian “tar sand” bitumen (or as a byproduct of crude oil refining) is a rapidly growing source of carbon for silicon production. Producing, refining, and transporting “tar sand” “petcoke” amount to ecological nightmares.[xviii]
Why does smelting silicon from quartz require trees?
In the silicon smelting process (Step Two), the mixture of materials in the smelter must be kept porous so that the gases produced during the chemical reaction will escape upward from the furnace, and the silicon metal will settle to the bottom. The only commercially available product that is sufficiently pure, slow burning and moist…is wood. Also, many silicon producers use wood charcoal as a carbon source (along with coal or coke), requiring many tons of trees to be burned—and lending some carbon to the smelting process.[xix]
How much electricity does it take to make polysilicon?
A large, modern polysilicon plant can require up to 400 megawatts of continuous power to produce up to 20,000 tons of polysilicon per year (~175MW/hrs per ton of polysilicon).[xx] Per ton, that’s more than ten times the energy needed for the initial smelting of raw silicon from ore. But few older, existing plants are this efficient.
As John Bradley, a senior vice president for the Tennessee Valley Authority power company (fueled by a combination of coal, nuclear and hydro power) explains, polysilicon manufacturers “just can’t allow for a hiccup” in electricity delivery. “Reliability in this case is more important than price, because the financial implications are much higher. Think about the loads—100 to 130 megawatts in phase one. A nuclear plant may generate 1200 megawatts. Fully built out, (a polysilicon facility) could use one third of the electricity produced by a nuclear plant.”[xxi]
Only electricity (not oil or gas) can heat the reactor filaments. This electricity must be delivered continuously,[xxii] or the batch and the reactor(s) can be ruined. Because of the hazardous materials involved, interrupting Step Three can result in dangerous fires or explosions.
At present, no form of “renewable” energy can satisfy the power demanded by silicon smelting or polysilicon manufacturing, because “renewables” cannot generate electricity 24/7; and commercial energy storage systems large enough to power a polysilicon plant do not yet exist.
What natural resources does manufacturing polysilicon demand?
In 2016, HiTest Silicon, a Canadian company now called PacWest Silicon, proposed building a new facility in Newport, Washington. Annually, this plant would consume approximately 170,000 tons of quartz, 150,000 tons of coal and charcoal combined, and 130,000 tons of wood chips to produce 73,000 tons of silicon. It would consume (continuously) 105 megawatts of electricity, and emit 320,000 tons of CO2 per year.[xxiii],[xxiv] Daily, it would consume about 10,000 gallons of water. It would also emit substantial amounts of nitrogen oxides, carbon monoxide and sulfur dioxide (components in smog and acid rain).
PacWest’s president, Jason Tymko, considers the smelter neither highly polluting nor damaging to the environment. His PowerPoint about his proposed silicon metal facility shows that based on 2015 values, its CO2 emissions would reach about 320,000 metric tons per year—making it the state’s 15th-largest emitter of greenhouse gases.[xxv] (Many scientists report that man-made GHGs trap heat within our atmosphere and lead to a significant rise in sea level temperatures and climate disruption.[xxvi]) Curiously, this figure does not include the GHG emissions generated by the power plant that will provide the smelter with electricity. The power plant is the state’s primary CO2 emitter.
Iceland’s silicon metal production facility, United Silicon, in Helguvik, has significantly increased the amount of coal that the country burns. In 2015, Iceland burned 139,000 tons of coal. By 2018, the country was expected to burn, annually, 224,000 tons of coal,[xxvii] including 66,000 tons transported to Iceland from El Cerrajon, Columbia and 45,000 tons of imported wood…in order to produce silicon.[xxviii]
In Asia, about 14,000 soccer fields worth of CO2-absorbing Myanmar forests go up in flames each year in response to Chinese demand for charcoal, which “feeds” the smelters. (These burns do not satisfy a fraction of Chinese demand.) Myanmar villagers began producing charcoal by burning trees and stumps around 2014, when demand increased.[xxix] The charcoal is used for producing stainless steel and silicon metal for electronics and photo voltaic solar panels. China now produces over two thirds of the world’s silicon.
In 2015, 8000 hectares (31 square miles) of forests were cut down in Dehong, Myanmar. Kevin Woods, who researches Myanmar’s politics related to resources, explained that the country’s charcoal production is “like taking the house that you live in, burning it down (to charcoal) and selling it for firewood.” The motivation, Woods explained, is that the wider economy is in shambles. Myanmar farmers can sell a bag of charcoal (destined for Chinese silicon manufacturers) for about 90 cents.[xxx]
References
[i]. https://semiengineering.com/the-limits-of-the-lifecycle-2/
[ii]. Schwarzburger, Heiko, “The trouble with silicon,” https://www.pv-magazine.com/magazine-archive/the-trouble-with-silicon_10001055/ September 15,2010.
[iii]. http://semiengineering.com/making-chip-manufacturing-sustainable/
[iv]. www.svtc.org
[v]. Smith, Ted, David Sonnenfeld and David N. Pellow, Eds., Challenging the Chip: Labor Rights and Environmental Justice in the Global Electronics Industry, Temple Univ. Press, 2006.
[vi]. Hayes, Brian, “The Memristor,” American Scientist, 2011.
How are silicon wafers manufactured?
[vii]. Goodrich, A., P. Hacke, et al., “A wafer-based monocrystalline silicon photovoltaics road map: Utilizing known technology improvement opportunities for further reductions in manufacturing costs,” Solar Energy Materials and Solar Cells, 114, pp.110-135, 2013.
[viii]. Wartluft, Jeffrey L., “The Use and Market for Wood in the Electrometallurgical Industry,” USDA Forest Service Research Paper NE-191, 1971.
[ix]. USDA Forest Service, “The Hardwood Resource on Nonindustrial Private Forest Land in the Southeast Piedmont,” Research Paper SE-236 https://www.srs.fs.usda.gov/pubs/rp/rp_se236.pdf
[x]. https://www.profor.info/knowledge/brazil-scaling-renewable-charcoal-production
[xi]. June 5, 2018 letter from Jason Tymko, President/CEO of PacWest Silicon to Grant Pfeifer, Regional Director, Eastern Regional Office, Washington Department of Ecology re PacWest’s proposal to construct and operate a silicon manufacturing facility near Newport, Washington.
[xii]. Kramer, Becky, “Northeast Washington silicon smelter plans raise concerns,” The Spokesman-Review, 11.1.17.
[xiii]. Thorsil Metallurgical Grade Silicon Plan; Helguvik, Reykjanes municipality (Reykjanesbaer), Reykjanes peninsula, Iceland, Environmental Impact Assessment, February, 2015.
[xiv]. https://www.wacker.com/cms/media/documents/investor-relations/cmd16/cmd16_polysilicon.pdf
[xv]. Schwarzburger, 2010
[xvi]. Kato, Kazuhiko, A. Murata and K. Sakuta, “Energy Pay-back Time and Life-cycle CO2 Emission of Residential PV Power System with Silicon PV Module,” Progress in Photovoltaics: Research and Applications, John Wiley & Sons, 1998.
[xvii]. The Society of Chemical Engineers of Japan (ed.), “Production of silicon wafers and environmental problems,” Introduction to VLSI Process Engineering, Chapman & Hall, 1993.
Why does manufacturing silicon require coal?
[xviii]. Stockman, Lorne, “Petroleum Coke: The Coal Hiding in the Tar Sands,” Oil Change International, January, 2013; www.priceofoil.org. Coloarulli, Kate, et al., “FAIL: How the Keystone XL Tar Sands Pipeline Flunks the Climate Test,” Oil Change International, January, 2013.
Why does smelting silicon from quartz require trees?
[xix]. Jungbluth, Niels, Matthias Stucki et al, Life Cycle Inventories of Photovoltaics Version: 2012 jungbluth-2012-LCI-Photovoltaics.pdf; “Use of charcoal in silicomanganese production – July 2003 DOI: 10.13140/RG.2.2.32879.84645 Report number: SINTEF STF24 A03535Affiliation: SINTEF Materials and Chemistry.
How much electricity does it take to make polysilicon?
[xx]. Bruns, Adam, “Wacker Completes Dynamic Trio of Billion-Dollar Projects in Tennessee: ‘Project Bond’ cements the state’s clean energy leadership,” 2009, www.siteselection.com.
[xxi]. ibid.
[xxii]. Schwarzburger, 2010.
What natural resources does manufacturing polysilicon demand?
[xxiii]. June 5, 2018 letter from Jason Tymko to Grant Pfeifer.
[xxiv]. Kramer, 11.1.17.
[xxv]. http://www.electronicsilentspring.com/wp-content/uploads/2018/11/Power-Point.pdf
[xxvi]. Intergovernmental Panel on Climate Change, “Summary for Policy Makers of IPCC Special Report on Global Warming of 1.5C, October, 2018; http://www.ipcc.ch/report/sr15; Wuebbles, D.J., D.W. Fahey, et al, 2017: Executive Summary of the Climate Science Special Report: Fourth National Climate Assessment, Vol. 1, U.S. Global Change Research Program, Washington, DC, USA.
[xxvii]. https://grapevine.is/news/2017/01/05/iceland-burns-thousands-of-tonnes-of-coal-each-year-thanks-to-heavy-industry/
Why does manufacturing silicon require coal?
[xxviii]. ibid.
[xxix]. https://news.mongabay.com/2017/10/burning-down-the-house-myanmars-destructive-charcoal-trade/
[xxx]. ibid.