Industries / Mining and materials / Chemicals and petrochemicals / Biofuels
Fermentation and ethanol
Turning the sugar in cane, the starch in corn and other grains, and lately the fiber in plants, into ethanol with yeast and enzymes, and selling what is left as animal feed and corn oil.
7
Milestones achieved, from 1975 to 2024.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1975
Brazil's national alcohol program
After the 1973 oil shock raised Brazil's oil import bill from US$600 million in 1973 to US$2.5 billion in 1974, Decree 76,593 of 14 November 1975, signed by President Ernesto Geisel, created the National Alcohol Program, Proálcool. It promoted alcohol from sugarcane, cassava "or any other input" by expanding raw-material supply, modernizing distilleries, building new ones attached to sugar mills or standing alone, and building storage. Production grew from about 600 million liters a year in 1975/76 to 3.4 billion in 1979/80 and 12.3 billion in 1986/87, 15% above the government's goal of 10.7 billion.
Credit
the Geisel government; the program grew from the 1974 study "Photosynthesis as a source of energy" and engine work by Urbano Ernesto Stumpf at the Aeronautics Technology Center in São José dos Campos (Alcarde, University of São Paulo, 2008)
1989
An ethanol shortage in Brazil
By 1985, cars that ran only on ethanol made up more than 95% of Brazil's light-vehicle sales. Then oil fell from about US$40 to US$12 a barrel, prices paid to alcohol producers stayed too low for them to expand, and public money for subsidies ran short, while lower taxes kept ethanol cars attractive to buyers. Supply could not keep up, and pumps ran short in the 1989–1990 season between cane harvests. Brazil filled the gap with a "MEG" mix of 60% ethanol, 34% methanol and 6% gasoline and imported more than 1 billion liters of ethanol and methanol from 1989 to 1995; by 1998–2000, ethanol-only cars were under 1% of new vehicles built.
LessonEncouraging demand with tax breaks while paying producers too little to grow ends in shortage, and buyers then abandon the product.
Credit
a failure; account by André Ricardo Alcarde of the University of São Paulo (2008)
2004
Oil from leftover stillage
Thin stillage is the watery liquid left after ethanol is distilled out of fermented corn mash. In May 2004 a centrifuge was installed at Agri-Energy's dry-mill ethanol plant in Minnesota to recover corn oil from it, using a method developed by David Cantrell and David Winsness of Vortex Dehydration Technology with Alfa Laval equipment. Oil recovery spread across the industry: by Scott Irwin's count from USDA data, US dry mills recovered 0.48 pounds of corn oil per bushel in December 2014 and 1.00 pound in October 2025, and the oil is now sold as a feedstock for biodiesel and renewable diesel.
LessonIn 2020 a federal appeals court upheld rulings that their patents could not be enforced, because the inventors had offered the system to Agri-Energy in July 2003, more than a year before they filed, and kept that offer from the patent office.
cases.justia.comfarmdocdaily.illinois.edufarmdocdaily.illinois.edu
Credit
David Cantrell and David Winsness
2008
Yeasts picked from the mills
Brazilian cane distilleries reuse their yeast. They run two or three fed-batch fermentations a day for six to eight months without a stop, in plants that are not sterile; after each run the yeast is spun out, washed in dilute sulfuric acid at pH 2.0 to 2.5 to kill bacteria, and put back. From 1993 to 2005 a team from Fermentec and the University of São Paulo sampled the yeasts that survived this treatment in the mills and selected strains for industry use. By 2007/08 two of them, PE-2 and CAT-1, were used in about 150 distilleries that made about 60% of Brazil's fuel ethanol. In the team's tests, PE-2 reached 92.0% of the theoretical ethanol yield from sugar, against 88.1% for baker's yeast.
LessonThe best industrial yeasts came from the plants themselves, where years of recycling had selected strains that tolerate the stress.
Credit
Luiz Carlos Basso of the University of São Paulo, with Henrique V. de Amorim, Antonio J. de Oliveira and Mario L. Lopes of Fermentec (FEMS Yeast Research, 2008)
2019
Cellulosic plants go idle
Cellulosic ethanol is made from the fiber in stalks, leaves and wood rather than from grain starch or cane sugar. Three large US plants built to make it, mainly from corn stover (the stalks and leaves left after harvest), opened in 2014 and 2015: Abengoa's at Hugoton, Kansas, POET-DSM's Project Liberty at Emmetsburg, Iowa, and DuPont's at Nevada, Iowa. Abengoa's plant was sold when its Spanish parent went bankrupt, DuPont closed its plant in 2017, and in November 2019 POET-DSM ended commercial production at Liberty, cut 30 of 70 jobs and turned the plant to research, blaming the EPA's handling of the Renewable Fuel Standard. IEA Bioenergy's tracking study puts US operating capacity at a peak of almost 200,000 tonnes a year in 2015 and 2016 and about 12,000 tonnes a year in 2020.
LessonPlant fiber is far harder to process than grain; the IEA authors name equipment wear from handling solids, costly enzymes, high production costs and unsteady policy.
Credit
a failure; tracked by Bacovsky and colleagues for IEA Bioenergy Task 39 (Biofuels, Bioproducts and Biorefining, 2025)
2023
Romanian cellulosic plant closes
Clariant, a Swiss chemical company, developed its own cellulosic ethanol process, sunliquid, decided in 2017 to build a commercial plant at Podari, Romania, and began producing ethanol there in the second quarter of 2022. In July 2023 it started a review after the plant did not reach its targeted operating parameters, and on 6 December 2023 it announced the closure, with about CHF 110 million of write-downs and CHF 60–90 million of closure costs. IEA Bioenergy's study lists the plant at 50,000 tonnes a year and says European cellulosic capacity, which peaked at about 165,000 tonnes a year in 2022, fell again after it closed.
Credit
a failure; Clariant's own announcement, recorded in IEA Bioenergy's tracking study
2024
Fiber in the kernel counts
Dry mills ferment the starch in corn; enzymes can also break the kernel's fiber into sugars that ferment in the same tanks, so no new tanks are needed. In March 2024 the US Environmental Protection Agency issued guidance on how plants can test and document ethanol made this way from corn kernel fiber, which earns D3 credits, the Renewable Fuel Standard's credits for cellulosic fuel, worth an average $2.34 a gallon in 2025 against $1.64 for ethanol at Iowa plants. The EPA had approved such pathways for 85 plants, nearly half of US producers. D3 credits for cellulosic ethanol reached 84.3 million gallons in 2025, 2.5% of US ethanol output, and the average dry-mill yield stayed above 3 gallons per bushel for an extended period for the first time, peaking at 3.07 in July 2025, up from about 2.80 in late 2014.
LessonMost of the starch in corn is already turned into ethanol; Irwin cites a limit of 3.0 to 3.1 gallons per bushel from starch alone and up to 3.5 with the fiber, so further gains have to come from the fiber.
Credit
the EPA's 2024 guidance and the plants that adopted the enzymes; measured by Scott Irwin of the University of Illinois (farmdoc daily, February 2026)
Watch for
Jet fuel from ethanol
LanzaJet says its Freedom Pines Fuels plant in Soperton, Georgia, which it calls the world's first commercial-scale alcohol-to-jet plant, can make 10 million gallons a year of sustainable aviation fuel and renewable diesel from ethanol, and has already made both to ASTM specifications, the standards these fuels must meet. In October 2026 it announced at least $20 million of new financing to scale up production and deliver fuel to customers.
Achieved whenthe plant reports a full year of steady output near its rated 10 million gallons, confirmed by an outside party such as an airline buyer's published purchases, a fuel certification body or US EPA fuel-credit data.
Claimed by
LanzaJet, October 2026
Open
01
Count bacteria under the microscope
An image model that counts, in stained samples from fuel-ethanol fermentations, live yeast, dead yeast and contaminating bacteria (mostly rod-shaped lactic acid bacteria), from photos taken through an ordinary lab microscope or a phone adapter, and reports yeast viability and bacteria per milliliter.
The briefCredit · unclaimed02
Rated versus real output
A public ledger, plant by plant and year by year, of every large cellulosic ethanol plant built since 2010: rated capacity, start and stop dates, and the ethanol it actually made, with each figure linked to its source and marked as company-reported or independently reported.
The briefCredit · unclaimed
Open 01 · the brief
Count bacteria under the microscope
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An image model that counts, in stained samples from fuel-ethanol fermentations, live yeast, dead yeast and contaminating bacteria (mostly rod-shaped lactic acid bacteria), from photos taken through an ordinary lab microscope or a phone adapter, and reports yeast viability and bacteria per milliliter.
Why it is still open
Bacteria in fuel-ethanol fermentations cost an estimated 3% of ethanol yield, over 960 million liters a year in Brazil alone, and mills wash recycled yeast in acid to keep them down (Lino and colleagues, Nature Communications, 2024). The tools that automate microscope counts count yeast only. Oculyze's bioethanol app reports yeast concentration, viability and budding, and UCLA's lens-free counter was tested on distillers' yeast for viability and concentration (Feizi and colleagues, Lab on a Chip, 2016). The yeast maker Lallemand began work with the image-analysis firm Aiforia in 2021 on AI detection of contamination in its yeast products and said then that it was "not ready yet"; no results were found. A November 2026 Europe PMC and web search for automated image counting of bacteria in fuel-ethanol samples found no published tool.
Built with what exists
A lab microscope with a phone adapter or USB camera; methylene blue or methylene violet stain; Label Studio for labeling; Cellpose or an Ultralytics YOLO11 model for finding cells, or a small vision model fine-tuned on one graphics chip (see the Note "One GPU to fine-tune").
First step
Mix baker's yeast with a Lactobacillus culture from a home-brewing supplier at five known ratios, stain with methylene blue, photograph 20 fields of each mix through a microscope, and label them in Label Studio.
Done looks like
At least 2,000 labeled fields of view from at least three sources, released openly. On 200 held-out samples, the model's counts of live yeast, dead yeast and bacteria are compared with two trained people's counting-chamber counts and with plate counts for bacteria, and the error for each class is published.
Open 02 · the brief
Rated versus real output
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public ledger, plant by plant and year by year, of every large cellulosic ethanol plant built since 2010: rated capacity, start and stop dates, and the ethanol it actually made, with each figure linked to its source and marked as company-reported or independently reported.
Why it is still open
IEA Bioenergy's 2025 tracking study lists about 20 large plants with their capacity and status, not their output, and says it could not tell whether China's largest plant was running. The last plant-by-plant review of output, by the campaign group Biofuelwatch in 2018, covered ten plants and found that companies often published no production figures; it reported, for example, that DuPont's Iowa plant closed in November 2017 "without having produced any cellulosic ethanol". US output is published only as national totals of D3 credits.
Built with what exists
The plant list in IEA Bioenergy's study and the Task 39 database; US EPA's monthly credit (RIN) data by fuel type; company annual reports and press releases; Brazil's ANP production data; an AI agent to read hundreds of reports and pull out figures with their page numbers (see the Note "The jobs agents repeat"); a spreadsheet or SQLite file published on GitHub.
First step
Fill in the three large US plants of 2014–2015 for 2014 to 2020 from company statements, news reports and Biofuelwatch's review, and compare their sum with EPA's cellulosic ethanol credit totals for those years.
Done looks like
Every plant in the IEA study's table entered for each year from start-up to 2025, with at least one sourced output figure for 70% of plant-years and every gap marked, and the US rows checked against EPA's yearly cellulosic ethanol credit totals.