BRENNEN

Industries / Mining and materials / Chemicals and petrochemicals / Fertilizers and agrochemicals

Ammonia and nitrogen fertilizer

This work fixes nitrogen from the air into ammonia, then turns the ammonia into nitric acid, urea and ammonium nitrate, the main nitrogen fertilizers.

6

Milestones achieved, from 1905 to 2019.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1905

    Nitrogen from an electric arc

    Kristian Birkeland noticed a smell of nitrogen oxide when his electric cannon short-circuited into an arc of flame, and with the engineer Sam Eyde he turned the arc into a way to bind nitrogen from the air. Norsk Hydro was founded on 2 December 1905, and its plants at Notodden, Norway, went on to make nitric acid, calcium nitrate and nitrite. Eyde reported that tests at Aas Agricultural College found the calcium nitrate as good as natural Chilean saltpeter.

    hydro.comhydro.com

    Credit

    Kristian Birkeland, Sam Eyde and Norsk Hydro, founded with Swedish and French capital; Marcus Wallenberg chaired its first board

  2. 1913

    Oppau's thirty tons a day

    BASF's first Haber-Bosch ammonia plant, at Oppau in Germany, went on stream on 9 September 1913 with a capacity of 30 metric tons of ammonia a day. Getting there took BASF chemists more than 2,500 catalysts before they settled on a promoted iron catalyst in 1910. An early mild-steel reactor lasted only 80 hours before it failed from decarbonization, the loss of carbon from the steel; lining it with soft iron and cutting grooves to let hydrogen escape solved the problem.

    LessonHot hydrogen under high pressure attacks ordinary steel, so the reactor's metal was as hard a problem as the chemistry.

    aiche.org

    Credit

    Carl Bosch, Alwin Mittasch and other BASF chemists, scaling up Fritz Haber's laboratory process

  3. 1927

    The arc gives way

    Through the 1920s Hydro's research center tried to make its arc process compete with Haber-Bosch, but the work went better in small trials than at large scale. In spring 1927 Hydro's management agreed a partnership with IG Farben that let it expand by the "ammonia method", and its new fertilizer plant at Herøya, near Porsgrunn, began in 1929 running on liquid ammonia sent by rail from Rjukan and Notodden. Chemical production stopped at Notodden in 1967 and at Rjukan around 1990.

    LessonEven cheap hydroelectric power did not keep the arc process competitive with Haber-Bosch.

    hydro.comhydro.comhydro.comhydro.com

    Credit

    Norsk Hydro and IG Farben

  4. 1960s

    One train, one plant

    Until the 1960s an ammonia plant was an assembly of separate units for making, cleaning and compressing gas. In the mid-1960s the American Oil Co. started a single-converter plant at Texas City, Texas, engineered by M.W. Kellogg, with a capacity of 544 metric tons a day. It raised its gas pressure with a four-case centrifugal compressor and recovered the process's own waste heat as steam to drive its turbines. The design won the Kirkpatrick Chemical Engineering Achievement Award in 1967. For ICI's Severnside site Kellogg then proposed a synthesis loop at 152 bar instead of 324 bar, low enough for centrifugal compressors throughout.

    aiche.org

    Credit

    M.W. Kellogg and American Oil Co., with reaction data from Haldor Topsøe for the lower-pressure loop

  5. 1992

    A ruthenium catalyst

    In late 1992 Ocelot Ammonia's plant at Kitimat, British Columbia, added a converter filled with a ruthenium catalyst on a graphite support, the first commercial use of Kellogg's Advanced Ammonia Process (KAAP). Oil & Gas Journal reported that the plant kept full output while using 30% to 40% less steam and 5% to 10% less electricity. Kellogg's product manager called it the first important advance in ammonia catalysts since iron was introduced more than 80 years earlier.

    ogj.com

    Credit

    M.W. Kellogg and BP, which developed the catalyst

  6. 2019

    A test for ammonia claims

    Many labs had reported making ammonia from nitrogen with electricity at room temperature, a possible low-energy alternative to Haber-Bosch, but in amounts so small that traces from air, breath, membranes or nitrogen compounds in the gas could explain them. A 21-author paper in Nature set out a protocol that uses nitrogen gas made of the heavier isotope nitrogen-15, so ammonia truly made from the supplied gas can be told apart from contamination. Following it, the team found no ammonia from the most promising pure-metal catalysts in water, and confirmed ammonia made with lithium in tetrahydrofuran, an organic solvent.

    LessonA tiny yield counts as a result only after controls have ruled out contamination.

    ebi.ac.uk

    Credit

    Suzanne Z. Andersen, Ib Chorkendorff and 19 co-authors

Watch for

  1. Off-grid ammonia at scale

    Envision says its plant in Chifeng, Inner Mongolia, makes 320,000 metric tons of ammonia a year from hydrogen made with power from its own wind and solar grid, without the public grid, and plans 1.5 million tons a year by 2028. It says the plant holds ISCC Plus and Bureau Veritas renewable-ammonia certificates. Caixin reported that during trial operation the plant met three straight days of little wind and sun, had to buy power from the public grid, and ran at a loss.

    Achieved whena buyer, certifier or regulator publishes the plant's measured output over a full year and the share of its power that came from its own wind and solar.

    caixinglobal.comengineers.scot

    Claimed by

    Envision, July 2025

  2. Ammonia that follows the wind

    The ASPIRE demonstrator outside Oxford, UK, came online in December 2025: a 100 kW electrolyzer feeding a containerized plant rated at 300 kg of ammonia a day, built to change its output as the power supply changes. Its team says the design can scale to 200 MW of wind power and turn down by a factor of twenty, to 10 MW.

    Achieved whenmonths of operating data showing output following real wind power are published and checked by someone outside the project, or a scaled-up unit runs.

    ammoniaenergy.org

    Claimed by

    the ASPIRE team at the UK Science and Technology Facilities Council, December 2025

Open

  1. 01

    Scoring electrochemical ammonia claims

    A public, regularly updated table of papers that claim to make ammonia from nitrogen with electricity, scoring each against the checks set out by Andersen and colleagues in 2019 and Choi and colleagues in 2020: whether the yield is large enough to matter, whether quantitative experiments with nitrogen-15 gas were run, and whether contamination from nitrogen oxides was ruled out. Each score links to the page of the paper it rests on.

    The brief
    Credit · unclaimed
  2. 02

    Forecast a green ammonia plant

    An open, hour-by-hour model of a named renewable-powered ammonia plant. It takes public weather data, the plant's announced wind, solar, battery and electrolyzer sizes, and how far its ammonia loop can turn down, and forecasts the plant's yearly output and the power it would need to buy from the grid. The forecast is published before the operator reports a full year of output, then scored against it.

    The brief
    Credit · unclaimed

Open 01 · the brief

Scoring electrochemical ammonia claims

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public, regularly updated table of papers that claim to make ammonia from nitrogen with electricity, scoring each against the checks set out by Andersen and colleagues in 2019 and Choi and colleagues in 2020: whether the yield is large enough to matter, whether quantitative experiments with nitrogen-15 gas were run, and whether contamination from nitrogen oxides was ruled out. Each score links to the page of the paper it rests on.

Why it is still open

Choi and colleagues scored 127 papers on water-based electrochemical nitrogen reduction, published up to April 2020, in a traffic-light table, and concluded that making ammonia this way in water "remains unproven". Searches in November 2026 found reviews that discuss the protocol and no public table that carries their scoring forward to later papers.

pmc.ncbi.nlm.nih.govebi.ac.uk

Built with what exists

Crossref and OpenAlex to list papers; Choi and colleagues' Supplementary Table 1 as the starting set and scoring guide; an AI model that reads long documents and returns quoted evidence, such as Muse Ember (see the Note "The jobs agents repeat"); a spreadsheet or small database with a public web page.

  1. First step

    Download Choi and colleagues' supplementary table, pull the 2020–2026 papers with "nitrogen reduction" and "ammonia" in their titles from Crossref, and score the first 20 by hand.

  2. Done looks like

    At least 300 papers published since May 2020 scored on the three checks, each score with the quoted sentence or figure behind it; a second person re-scores a random 10% without seeing the first scores, and the agreement rate is published.

Open 02 · the brief

Forecast a green ammonia plant

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

An open, hour-by-hour model of a named renewable-powered ammonia plant. It takes public weather data, the plant's announced wind, solar, battery and electrolyzer sizes, and how far its ammonia loop can turn down, and forecasts the plant's yearly output and the power it would need to buy from the grid. The forecast is published before the operator reports a full year of output, then scored against it.

Why it is still open

Design-stage models exist. Samsung C&T used a dynamic model in AVEVA's software to predict that its 1-million-ton-a-year concept would make 951,269 metric tons a year before design changes, and Rosbo and colleagues and Groll and Sin have published models of ammonia loops that run on a varying hydrogen supply. None is compared with a running plant. Caixin reported that Envision's Chifeng plant met three straight days of little wind and sun in its trials and bought grid power. Searches in November 2026 found no published comparison between a model and a named plant's actual output.

aveva.compsecommunity.orgpsecommunity.orgcaixinglobal.com

Built with what exists

ERA5 hourly weather reanalysis from the Copernicus Climate Data Store; pvlib and windpowerlib for solar and wind output; PyPSA, an open power-system optimizer; turndown limits taken from the published loop models; Python.

  1. First step

    Collect Envision's published wind, solar, battery and electrolyzer capacities for Chifeng and download one year of ERA5 wind and sunlight data for the site.

  2. Done looks like

    Dated forecasts for at least three named plants, published before each plant's first full-year output figure; afterwards, each forecast's error against the reported output, with the model and its inputs released.