Industries / Mining and materials / Chemicals and petrochemicals / Fertilizers and agrochemicals
Emissions
This work measures and cuts what fertilizer and agrochemical plants release to the air: carbon dioxide and methane from making ammonia, nitrous oxide from nitric acid, ammonia itself, and fluorides from phosphate plants.
6
Milestones achieved, from 1964 to 2019.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1964
Fluoride caught at the stack
As phosphate plants in Polk and Hillsborough counties, Florida, made more triple superphosphate fertilizer, their fluoride emissions rose. Cattle nearby got fluorosis and citrus trees showed chlorosis, a yellowing of the leaves, worst in winter and early spring. Joint government and industry action cut fluoride emissions from 33,000 pounds a day in the winter of 1961–62 to 13,000 pounds a day in the winter of 1963–64, and $16 million of control equipment was added from 1959, according to T. D. Crocker in the Bulletin of the Atomic Scientists (1965).
Credit
the area's phosphate companies and government agencies
1982
Fertilizer carbon to oil fields
Since 1982 the Enid Fertilizer project has captured carbon dioxide from an ammonia plant in Enid, Oklahoma, and piped it to oil fields in southern Oklahoma, where it is injected to recover more oil. The Center for Climate and Energy Solutions puts the amount at about 600,000 tons a year, and notes that capturing carbon dioxide at fertilizer plants is fully commercial, using the same compression and drying equipment as natural gas.
Credit
the Chaparral and Merit Enid Fertilizer Project, as C2ES names it
2013
A price on nitrous oxide
Nitric acid plants, which make a fertilizer ingredient, give off nitrous oxide (N2O), a strong greenhouse gas. Directive 2009/29/EC brought nitric acid production (carbon dioxide and nitrous oxide) and ammonia production (carbon dioxide) into the EU Emissions Trading System from 2013, so plants in the EU must surrender emission allowances for these gases.
Credit
the European Parliament and Council of the European Union
2018
Ammonia hotspots from space
Van Damme and colleagues used almost a decade of daily observations from the IASI satellite instrument to map 248 ammonia hotspots less than 50 km across, each tied to a single source or a cluster of farms and factories. The EDGAR emissions inventory underestimated two-thirds of them by at least a factor of ten, and industrial emitters were often in the wrong place or missing.
LessonAn inventory built from estimates can leave out whole factories.
Credit
Martin Van Damme, Lieven Clarisse and five co-authors (Nature, 2018), using IASI, a joint instrument of EUMETSAT and the French space agency CNES
2019
Plant reports fall short
Dammers and colleagues used the CrIS satellite instrument to estimate ammonia emissions from 249 large sources worldwide, including fertilizer plants such as ZMU at Kirovo-Chepetsk in Russia and plants near Togliatti and Shiraz. Their total was about 2.5 times the HTAPv2 inventory, and 55 sources were missing from it or underestimated more than tenfold. The pollutant registers of the United States, Canada, Europe and Australia compared poorly with the satellite figures; only Europe's E-PRTR had most sites within a factor of ten, and the authors note that most industrial figures in it are estimated, not measured.
Credit
Enrico Dammers and colleagues (Atmospheric Chemistry and Physics, 2019)
2019
Methane from fertilizer plants
Zhou and colleagues drove a Google Street View car fitted with a methane analyzer along public roads downwind of six US ammonia plants, more than a quarter of the country's gas-based plants. Assuming those plants were typical, they estimated the industry's methane emissions at 29 ± 18 thousand metric tons a year in 2015–16, against 0.2 thousand reported in the EPA's FLIGHT database, and put the share of gas lost at 0.34%.
LessonA plant's reported methane can be a small fraction of what is measured downwind.
Credit
Zhou, Passow, Rudek, von Fisher, Hamburg and Albertson (Elementa, 2019)
Watch for
Stored carbon at Donaldsonville
CF Industries says a unit it started at its Donaldsonville complex in Louisiana lets ExxonMobil take and permanently store up to 2 million metric tons of carbon dioxide a year. Storage is at first through enhanced oil recovery, moving later to ExxonMobil's dedicated Rose site, and CF expects about 1.9 million tons a year of low-carbon ammonia and US tax credits under Section 45Q. The EPA approved the monitoring plan for Rose in April 2026.
Achieved whenthe EPA's FLIGHT data, or another regulator's report, shows tonnes actually stored from the plant.
Claimed by
CF Industries, July 2025
Ammonia plumes from orbit
Researchers at Carbon Mapper, NASA's Jet Propulsion Laboratory and MIT report in PNAS that Planet's Tanager-1 satellite, which reads reflected sunlight, can map ammonia plumes from single plants. They estimate 837 kg an hour from a fertilizer plant in Pakistan, where coarser satellite instruments had given 3,212 to 8,035 kg an hour for 2013–2017, and 1,689 kg an hour from two nearby sources at a fertilizer plant in Uzbekistan.
Achieved whenan independent measurement, such as an aircraft survey or the plant's own monitored emissions, agrees with the satellite rate within the stated uncertainty.
Claimed by
Carbon Mapper, June 2026
Open
01
Plume check at ammonia plants
A public table of ammonia plants that lists, for each plant, every pass of the aircraft and satellites in Carbon Mapper's public catalog, which includes NASA's EMIT instrument, with any methane or carbon dioxide plume detected and its estimated rate, beside the methane and carbon dioxide the plant reports to its regulator. Start with the US plants in the EPA's Greenhouse Gas Reporting Program.
The briefCredit · unclaimed02
Fertilizer carbon, claimed and stored
A public ledger of every project that captures carbon dioxide from a fertilizer plant, listing for each year the capture the company claims, where the gas went (oil fields or dedicated storage), and the amount stored, each figure graded by how well it was checked: by a regulator, by an outside audit, or not at all, the three grades Zhang, Jackson and Krevor used.
The briefCredit · unclaimed
Open 01 · the brief
Plume check at ammonia plants
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of ammonia plants that lists, for each plant, every pass of the aircraft and satellites in Carbon Mapper's public catalog, which includes NASA's EMIT instrument, with any methane or carbon dioxide plume detected and its estimated rate, beside the methane and carbon dioxide the plant reports to its regulator. Start with the US plants in the EPA's Greenhouse Gas Reporting Program.
Why it is still open
Zhou and colleagues' 2015–16 road survey of six US ammonia plants put the industry's methane at about 145 times what plants reported, and no repeat using the satellite and aircraft data now public was found. Satellite work on fertilizer plants has measured ammonia, not methane: Tanager-1 in 2026 and an automated EMIT detector extended to ammonia (arXiv 2605.03372). A test query of Carbon Mapper's catalog in November 2026 around CF Industries' Donaldsonville complex returned carbon dioxide plumes at the plant's coordinates, labeled as fuel burned in manufacturing, and methane plumes several kilometers away, none attributed to ammonia making.
Built with what exists
Carbon Mapper's free plume API; the EPA's Envirofacts API and FLIGHT tool for plant locations and reported emissions; Python and QGIS.
First step
Pull the coordinates of US ammonia plants from the EPA's Envirofacts API and query Carbon Mapper's catalog in a box around each one.
Done looks like
Every US ammonia plant in the EPA program listed with its overpass count, detections and estimated rates since 2022, beside its reported annual figures; a published count of plants where any methane plume rate exceeds the reported annual average hourly rate; code and table released.
Open 02 · the brief
Fertilizer carbon, claimed and stored
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public ledger of every project that captures carbon dioxide from a fertilizer plant, listing for each year the capture the company claims, where the gas went (oil fields or dedicated storage), and the amount stored, each figure graded by how well it was checked: by a regulator, by an outside audit, or not at all, the three grades Zhang, Jackson and Krevor used.
Why it is still open
Zhang, Jackson and Krevor (2022) estimated how much carbon dioxide 20 capture facilities of all kinds actually stored up to 2020, and found the capture capacity usually quoted ran 19–30% above the stored amounts. IEEFA's 2022 review of 13 flagship projects included the Coffeyville fertilizer plant in Kansas and found no public data on its yearly capture over its lifetime. Neither is kept up to date or covers fertilizer projects started since, such as CF Industries' Donaldsonville unit (2025). The EPA's list of approved monitoring plans in November 2026 includes ExxonMobil's Rose site, which is to take Donaldsonville's carbon dioxide, but no oil field fed by the Enid or Coffeyville plants.
Built with what exists
The EPA's FLIGHT data and subpart RR monitoring plans; company sustainability reports and releases; an AI model for pulling figures from long reports, such as Muse Ember (see the Note "The jobs agents repeat"); a spreadsheet with a source for every cell.
First step
Build the rows for Enid, Coffeyville and Donaldsonville from company releases, the IEEFA case study and the EPA's FLIGHT data, and note what each year's public record shows.
Done looks like
At least ten fertilizer-plant capture projects worldwide, with every year since start-up filled in (claimed capture, destination, and stored amount with its grade, or "none published"), each cell linked to its source, and a total of claimed against regulator-confirmed tonnes.