Tag: by-product

  • Scandium: The Element Too Scarce to Use

    Global scandium production is approximately 25-40 tonnes per year. Projected demand is 117 tonnes per year by 2026. That gap — roughly three to four times more demand than supply — is not the result of a sudden crisis, an export control, or a geopolitical shock. It is the normal state of the scandium market. It has been the normal state for decades. Scandium is the critical mineral that has never had enough supply to discover how much demand actually exists, because the industries that would use it — aerospace, automotive, fuel cells, 3D printing — have never been able to buy it in quantities large enough to justify designing it into their products. Adding 0.1-0.2% scandium to aluminum creates an alloy that is 15-20% lighter than conventional alternatives, weldable without losing strength, corrosion-resistant, and suitable for aircraft fuselages, EV frames, and naval vessels. Approximately $2 million of scandium in a single airliner yields an estimated $27 million in net present fuel savings over the aircraft’s life. The economics are spectacular. The supply doesn’t exist to act on them. The Soviet Union discovered this first — the MiG-21 and MiG-29 used aluminum-scandium alloys in their airframes starting in the 1960s — and the West has been trying to replicate the supply chain ever since. As of 2026, it still hasn’t.

    Why there isn’t enough

    Scandium is more abundant in the Earth’s crust than silver, lead, or mercury. It is not geologically rare. It is economically rare because it has almost no affinity for the common anions that form concentrated ore deposits — meaning it is spread thinly across the lithosphere rather than concentrated into mineable veins. There is, at the time of this writing, essentially one dedicated scandium mine on Earth: Scandium International Mining’s operation in New South Wales, Australia. Everything else is by-product recovery.

    Scandium is recovered in small quantities from the processing of other metals — iron ore, rare earths, titanium, zirconium, uranium, and nickel laterite tailings. China produces the most, primarily from titanium dioxide production and rare earth processing. The Philippines, Kazakhstan, Russia, and Ukraine produce smaller amounts from nickel laterites and uranium operations. None of these producers are mining for scandium. They are recovering it as a residue from processes designed for other purposes — the same by-product supply ceiling that constrains indium, tellurium, rhenium, hafnium, and the noble gases.

    But scandium adds a dimension the other by-product metals don’t have. Rhenium is scarce, but the aerospace industry has designed around it — jet engine superalloys are formulated to use rhenium because the supply, though small, has been stable enough for turbine manufacturers to commit. Hafnium is scarce, but Intel adopted hafnium oxide in 2007 because 75 tonnes per year was enough for the semiconductor industry’s needs at the time. Scandium has never reached the supply threshold where a major industry could commit to using it at scale. The result is a chicken-and-egg problem that has persisted for half a century: manufacturers won’t design products around scandium because supply is unreliable, and miners won’t invest in scandium production because manufacturers haven’t committed to buying it. The market is stuck at 25-40 tonnes per year, with latent demand estimated at 5-10 times that level, and no mechanism to bridge the gap.

    What it would do if you could get it

    Aluminum-scandium alloys account for roughly 45% of scandium oxide consumption. The metallurgy is straightforward: adding 0.1-0.2% scandium by weight to aluminum refines the grain structure, eliminates the heat-affected zone weaknesses that make conventional aluminum alloys difficult to weld, and produces a material that can be reliably joined without post-weld heat treatment. That weldability property alone could transform aircraft manufacturing — currently, aluminum aircraft structures are largely riveted because the welding of conventional aerospace aluminum degrades the metal’s strength at the weld. Aluminum-scandium alloys can be welded without strength loss, which means fewer fasteners, lower weight, faster assembly, and simpler structural designs. The weight reduction translates directly into fuel savings for every flight the aircraft makes for 30 years.

    The automotive sector sees the same economics. Net aluminum content per light-duty vehicle is projected to increase from 459 pounds in 2020 to 570 pounds by 2030. If just 10% of that aluminum used 0.1% scandium, annual scandium demand from automotive alone would reach 700 tonnes — roughly 20 times current global production. The EV industry has an even stronger incentive: every kilogram removed from an EV extends its range, and range is the constraint that determines consumer adoption. The aluminum-scandium value proposition in EVs is not theoretical. It is purely a supply problem.

    Solid oxide fuel cells are the other growth engine — currently representing roughly 15-55% of global scandium consumption, depending on which estimate you use. Bloom Energy, the leading commercial SOFC manufacturer, uses scandia-stabilized zirconia electrolytes because scandium is a better ionic conductor than yttrium in this application — it allows the fuel cell to operate at lower temperatures, extending operational lifetime and reducing maintenance costs. A typical 100-kilowatt Bloom Energy server box contains 13-15 kilograms of scandium oxide. SOFC deployment is growing at roughly 23% compound annual growth rate. If scandium supply doesn’t grow with it, SOFC manufacturers will either pay dramatically more for feedstock or switch back to yttrium-stabilized zirconia — a substitution that trades performance for availability, at a moment when yttrium’s own supply chain is experiencing a 4,400% price spike.

    The supply response that might be coming

    Rio Tinto opened a scandium oxide plant in Sorel-Tracy, Quebec, in 2021, producing up to 3 tonnes per year from its existing titanium dioxide feedstock. In 2024, Rio Tinto acquired Platypus Alloys — an Australian company producing aluminum-scandium master alloy — signaling that the world’s second-largest mining company sees a market worth vertically integrating into. NioCorp Developments holds a scandium resource of 11,000 tonnes at its Elk Creek site in Nebraska and projects production capacity of 100-135 tonnes per year of scandium oxide, though the project has not yet reached construction. In Europe, the ScaVanger project in France targets 21 tonnes per year of scandium oxide from titanium dioxide coproduction, with production projected to begin in 2026. Clean TeQ (now Sunrise Energy Metals) in Australia has significant scandium resources in its nickel-cobalt laterite deposits.

    The project pipeline exists. The production doesn’t — not yet. If NioCorp, ScaVanger, and Rio Tinto all deliver on their stated timelines, global non-Chinese scandium supply could triple by 2028. That would still leave the market short of projected demand, but it would break the chicken-and-egg cycle by giving aerospace and automotive OEMs enough material to design aluminum-scandium into production platforms rather than test programs. The question is whether the projects get built before the demand window closes — before alternative lightweighting technologies (carbon fiber, magnesium alloys, advanced high-strength steel) lock in the market share that aluminum-scandium could have captured if the supply had existed five years earlier.

    Why it’s in the course

    Scandium is the Rare Earth Elements course’s case study in suppressed demand — the mineral whose scarcity has prevented the market from discovering its own size. Every other mineral in the course has a functioning market: lithium has a price, a supply chain, a demand curve. Copper has a shortage. Antimony had a price spike. Terbium has export controls. Scandium has a hypothetical market that is 5-20 times larger than the actual market, with the gap explained entirely by supply that has never existed in sufficient quantities for demand to materialize. The gallium/germanium export controls disrupted an existing supply chain. Scandium’s disruption is that the supply chain was never built.

    China classified scandium as a national strategic material in the April 2025 export controls — the same announcement that restricted terbium, samarium, and yttrium. For most of those elements, the controls created a crisis. For scandium, the controls restricted a supply that was already too small to matter. The crisis isn’t the export control. The crisis is that the element the Battlefields of the Future course identifies as capable of making fighter jets 15-20% lighter and the energy transition identifies as capable of making fuel cells more efficient has been stuck at 25 tonnes a year for a generation because nobody built the mine.

    This is the kind of supply chain our Rare Earth Elements course was built to map — where the most economically valuable aluminum additive ever discovered, deployed by the Soviet military sixty years ago, remains a curiosity rather than a commodity because global production has never exceeded two shipping containers, the industries that would use it can’t commit because supply doesn’t exist, and the mines that would produce it can’t justify the investment because demand hasn’t materialized — the purest chicken-and-egg trap in the critical minerals landscape.

  • Hafnium: The Element Inside Every Advanced Chip That Nobody Mines

    Every semiconductor chip manufactured at process nodes below 45 nanometers — which, as of 2026, includes every processor in every smartphone, every data center server, every AI accelerator, and every advanced military system on Earth — uses hafnium. Not in the packaging. Not in the wiring. In the transistor itself. Hafnium oxide replaced silicon dioxide as the gate dielectric in 2007 when Intel introduced its 45-nanometer process, because silicon dioxide at that thickness — just a few atoms wide — leaked too much current for the transistor to function. Hafnium oxide has a higher dielectric constant, meaning it can be physically thicker while remaining electrically thinner, which solved the leakage problem and made every subsequent generation of chip miniaturization possible. No hafnium, no chips below 45 nanometers. No chips below 45 nanometers, no modern computing. Global hafnium metal production is approximately 75 tonnes per year. Total production including oxides is roughly 85 tonnes. Demand is projected to reach 150-200 tonnes per year. Prices have risen 400% in recent years. And hafnium has no dedicated mine anywhere on Earth — it is produced exclusively as a by-product of zirconium refining, at a ratio of roughly 50 tonnes of zirconium for every 1 tonne of hafnium. The by-product supply ceiling that constrains indium, tellurium, the noble gases, and rhenium applies here — but hafnium’s version has a twist. Rhenium is a by-product of a by-product. Hafnium is a by-product of a by-product’s purification: it only exists as a separated element because the nuclear industry requires hafnium-free zirconium for reactor fuel cladding, and the process of removing hafnium from zirconium produces hafnium as a residue. If the nuclear industry didn’t need ultra-pure zirconium, nobody would be separating hafnium at all.

    Three sectors, one element, 75 tonnes

    Hafnium’s demand splits across three sectors that are all growing simultaneously and all drawing from the same 75-tonne annual pool.

    The first is semiconductors. Hafnium oxide — HfO₂ — is the gate dielectric in every advanced transistor. Intel, TSMC, Samsung, and every other foundry running sub-45nm processes deposit hafnium oxide films measured in angstroms — fractions of a nanometer — onto billions of transistors per wafer, millions of wafers per year. As transistor geometries shrink to 3 nanometers, 2 nanometers, and eventually below, hafnium oxide remains the standard gate dielectric. DRAM memory cells use hafnium oxide for the same reason: its high dielectric constant allows capacitors to store charge in smaller physical spaces. The semiconductor supply chain has its famous chokepoints — TSMC in Taiwan, ASML in the Netherlands, neon from Ukrainian air separation units, yttrium coating the etching chambers. Hafnium is the chokepoint nobody talks about because nobody has had to — the 75 tonnes has been enough, barely, with no margin. The question is what happens when demand reaches 150 tonnes and production can’t follow.

    The second is nuclear energy. Hafnium absorbs neutrons more effectively than almost any other material — a property that makes it essential for reactor control rods, the components that regulate the chain reaction by absorbing excess neutrons when inserted into the reactor core. Every pressurized water reactor, every boiling water reactor, every naval nuclear propulsion system uses hafnium control rods. The nuclear renaissance the fusion companies post documented — Microsoft restarting Three Mile Island, Google’s Kairos Power deal, Amazon’s Talen Energy acquisition, China’s plan to build six to eight reactors annually — is accelerating hafnium demand from the nuclear sector at the same time the semiconductor sector is accelerating demand from the electronics side. Both sectors drawing from the same 75-tonne pool. The pool doesn’t get larger because neither sector demands more zirconium.

    The third is aerospace superalloys. Hafnium is added to nickel-based single-crystal superalloys — the same CMSX-4 and René N5 alloys that contain rhenium — to improve grain-boundary strength and oxidation resistance at extreme temperatures. Hafnium-enriched superalloys tested in gas turbines and jet engines deliver 10-15% improved creep resistance at 1,200°C compared to baseline compositions. The Boeing and Airbus backlog of 15,000+ aircraft that drives rhenium demand drives hafnium demand through the same turbine blades. Every jet engine blade that needs rhenium inside it also needs hafnium inside it.

    And then there’s the emerging sector: plasma cutting. Hafnium-tipped plasma torch electrodes survive 6,000°C arcs and last 30% longer than conventional designs. Industrial metal cutting, shipbuilding, heavy fabrication — every plasma cutter in every shipyard and fabrication shop uses hafnium electrode tips that are consumed during operation and must be replaced. It’s a small market in tonnage terms. In a 75-tonne global pool, small markets matter.

    The zirconium dependency

    The hafnium supply chain is the most structurally constrained by-product chain in the Rare Earth Elements course — more constrained than rhenium (by-product of a by-product via copper-molybdenum), more constrained than indium (by-product of zinc), because hafnium’s separation is tied not just to a host metal’s production economics but to a specific purification process that exists for a specific customer: the nuclear fuel industry.

    Zirconium is mined as zircon — a mineral found in beach sand and alluvial deposits in Australia, South Africa, Brazil, and other countries. Most zircon is consumed directly as a ceramic material in tiles, foundry molds, and refractory linings — applications that do not require removing the hafnium. Only the nuclear industry requires hafnium-free zirconium, because hafnium’s neutron-absorbing properties are exactly the opposite of what you want in a fuel rod cladding material — zirconium is chosen precisely because it is transparent to neutrons, but natural zirconium contains 1-2.5% hafnium, which must be removed to achieve nuclear-grade purity.

    The separation process — typically a solvent extraction or extractive distillation operation — is concentrated in a handful of facilities worldwide. Orano (formerly Areva) operates five plants in France through its Cezus subsidiary. ATI and Western Zirconium operate in the United States. Chepetsky Mechanical Plant operates in Russia. Chinese facilities serve China’s domestic nuclear and semiconductor demand. The total global capacity for zirconium-hafnium separation produces roughly 75 tonnes of hafnium per year. That capacity was built to serve the nuclear fuel industry’s demand for pure zirconium. The hafnium was the waste product that someone figured out how to sell. Expanding hafnium production requires expanding nuclear-grade zirconium separation — a capital-intensive process justified by nuclear fuel demand, not by hafnium demand.

    China’s internal consumption of hafnium — for both its nuclear reactor construction program and its semiconductor industry — absorbs most of its domestic output, leaving little for export. After 2024 restrictions tightened the flow, Japan’s semiconductor supply chain, South Korea’s foundries, and India’s aerospace programs became increasingly dependent on French and American separation facilities. The CHIPS Act invests billions in semiconductor fabrication. The hafnium oxide that goes inside the transistors those fabs produce comes from a separation process built to serve the nuclear fuel industry, at volumes that were never designed to support the semiconductor industry’s growth trajectory.

    Why it’s in the course

    Hafnium is the Rare Earth Elements course’s purest demonstration that a material can be simultaneously indispensable and invisible. Every chip below 45 nanometers uses it. Every nuclear reactor control rod uses it. Every advanced turbine blade uses it. Global production is 75 tonnes. Nobody mines it on purpose. It exists as a separated element only because the nuclear industry needs it removed from something else. And the three sectors that need it — semiconductors, nuclear energy, aerospace — are all growing at the same time, all pulling from the same pool, with no mechanism to expand the pool independently of nuclear fuel demand.

    The lithium supply chain has a demand problem that can be solved by building lithium mines. The copper supply chain has a capacity problem that can be solved by building copper mines. The rhenium supply chain has a by-product problem that can only be solved by expanding copper-molybdenum mining. Hafnium has a by-product problem that can only be solved by expanding nuclear-grade zirconium separation — a process that exists to serve one industry, produces a residue consumed by three others, and cannot be scaled by any of the three industries that need the residue, because none of them control the process that produces it.

    This is the kind of supply chain our Rare Earth Elements course was built to map — where every transistor manufactured below 45 nanometers depends on 75 tonnes a year of a metal that nobody mines, separated from a mineral the nuclear industry needs purified, at a ratio of 50 to 1, by a handful of facilities built for a different purpose, with three industries competing for the output and none of them able to increase it.

  • Rhenium: The Rarest Metal in Your Jet Engine

    Global rhenium production in 2024 was approximately 62,000 kilograms — 62 tonnes. That’s it. The entire world’s annual output of the metal that allows jet engines to operate at 1,700 degrees Celsius without the turbine blades melting would fit inside two shipping containers. Rhenium prices rose 91.5% in 2025, then added another 34% in the first quarter of 2026, reaching roughly $2,400 per kilogram in Europe. The price action isn’t speculative froth — it’s a structurally tiny market absorbing the largest commercial aviation production ramp-up in history. Boeing and Airbus have combined backlogs of more than 15,000 aircraft. Each wide-body jet engine contains several kilograms of rhenium concentrated in the turbine blades that spin at 40,000 RPM in exhaust gas temperatures that would liquefy any unalloyed metal on the periodic table. A single company — Molymet S.A. in Chile — controls 60% of global rhenium production. Rhenium has no primary mine anywhere on Earth. It is recovered exclusively as a by-product of molybdenum roasting, which is itself a by-product of copper mining. The by-product supply ceiling that constrains indium, tellurium, and the noble gases applies here in its most extreme form: rhenium is a by-product of a by-product. Two layers of someone else’s business model between you and the metal your jet engine needs.

    What rhenium does

    Rhenium has the third-highest melting point of any naturally occurring element — 3,186°C — and the highest boiling point of all. When alloyed with nickel at 2-6% by weight, it forms single-crystal superalloys — materials like CMSX-4 and René N5 — that retain their structural integrity under the combination of extreme heat, extreme mechanical stress, and corrosive combustion gases that exist inside a jet engine’s high-pressure turbine section. Without rhenium, turbine blades deform under the sustained loading at operating temperature — a failure mode called creep — and the engine loses efficiency, thrust, and eventually structural integrity. The addition of rhenium doesn’t just improve performance. It makes the performance possible. Modern high-bypass turbofan engines like the GE9X (which powers the Boeing 777X) and the Rolls-Royce Trent XWB (which powers the Airbus A350) were designed around rhenium-containing superalloys. You cannot substitute the rhenium out of these engines without redesigning the turbine, recertifying the engine, and accepting lower operating temperatures that reduce fuel efficiency — a process that takes years of testing and hundreds of millions of dollars in development cost.

    Roughly 80% of global rhenium consumption goes into aerospace superalloys. The remaining 20% splits between petroleum refining catalysts — rhenium-platinum catalysts are essential for producing high-octane lead-free gasoline — and a growing list of niche applications including medical isotopes (rhenium-186 and rhenium-188 for cancer therapy), radiation shielding, and high-precision surgical instruments.

    The double by-product problem

    The rhenium supply chain is the purest expression of the by-product constraint the Rare Earth Elements course has been documenting across indium/tellurium, noble gases, and iridium within the PGM complex. But rhenium adds a layer of complexity the others don’t have.

    Copper is mined from porphyry deposits in Chile, the United States, Peru, and Kazakhstan. Some of those copper ores contain molybdenite — a molybdenum sulfide mineral — which is recovered as a by-product of copper processing. The molybdenite concentrate is then roasted to produce molybdenum trioxide, and during that roasting process, rhenium volatilizes out of the concentrate as rhenium heptoxide in the flue gas. Specialized scrubbers in the roasting facility’s exhaust system capture the rhenium. If the scrubbers aren’t installed, or aren’t optimized for rhenium recovery, the rhenium exits the smokestack and disperses into the atmosphere. Gone.

    Chile produced 29,000 kilograms of rhenium in 2024 — nearly half the global total — because Chile has the world’s largest copper-molybdenum porphyry deposits (Escondida, Chuquicamata, El Teniente) and because Molymet, the Santiago-based molybdenum processor, installed rhenium recovery equipment decades ago and has since acquired downstream processing capacity including Rhenium Alloys in Ohio, now operating as Molymet Alloys USA. The United States produced 9,500 kg, mostly from Freeport-McMoRan’s copper-molybdenum operations in Arizona and New Mexico. Poland produced 9,400 kg from KGHM’s copper operations. China produced 5,300 kg. The entire rest of the world produced the remainder.

    The structural constraint is identical to indium’s but twice compounded: if the world needs more rhenium, it needs more molybdenum roasting capacity with rhenium recovery equipment installed, which requires more molybdenum concentrate, which requires more copper mining. The decision to expand copper production is made by copper companies based on copper economics. The decision to install rhenium scrubbers is made by molybdenum processors based on rhenium prices relative to scrubber capital costs. At each layer, the entity making the production decision is optimizing for a different metal than the one the aerospace industry needs. Supply elasticity is, in the language of the economics textbooks, structurally constrained. In the language of the Battlefields of the Future course: the fighter jet’s turbine blade depends on a molybdenum roaster in Chile deciding the flue gas scrubber is worth the investment.

    The aerospace bottleneck

    The demand side of the rhenium market is defined by a single fact: the global aviation industry is attempting the largest production ramp-up in its history, and every new engine needs rhenium. The pandemic cleared factory floors, grounded aircraft, and delayed maintenance cycles. The recovery has produced backlogs that will take a decade to work through. Every widebody engine, every military fighter jet engine, every industrial gas turbine that uses rhenium-containing superalloys draws from a 62-tonne global pool. Demand is projected to grow at 2.1% annually through 2034 — a modest growth rate in percentage terms, but applied to a base so small that even modest demand growth tightens supply.

    Turbine blades have 15-to-25-year service lives before they’re retired and recycled. The recycling rate for rhenium from spent superalloys and foundry revert material is 25-50% — significantly better than most critical minerals but constrained by long collection cycles and the capital intensity of superalloy recycling. GE Aerospace and Rolls-Royce have both established closed-loop product stewardship programs to recover rhenium from spent blades. The rare earth recycling infrastructure the course has documented is nascent for most minerals. For rhenium, it’s more mature — but still limited by the physics of 15-year wait times between engine installation and blade retirement.

    GE and Rolls-Royce have also invested in “rhenium-lean” and “rhenium-free” superalloy formulations using combinations of ruthenium, tantalum, hafnium, tungsten, and molybdenum. Some newer engine designs have successfully reduced rhenium loading. But the installed fleet and the engines currently in production were designed around rhenium-containing alloys, and redesigning a turbine blade is not a materials-substitution exercise you complete in a quarter. Each new alloy formulation requires years of creep testing, oxidation testing, fatigue testing, engine testing, and FAA certification before it can be deployed in a commercial aircraft. The semiconductor supply chain can redesign a chip in 18 months. A turbine blade alloy substitution takes a decade.

    Why it’s in the course

    Rhenium is the Rare Earth Elements course’s most extreme case study of by-product dependency — and the one that connects the critical minerals thesis directly to the aerospace and defense industrial base. The antimony post documented a mineral whose supply was disrupted by deliberate export controls. The nickel post documented a market restructured by deliberate resource nationalism. The noble gases post documented a supply chain that concentrated by accident and broke because of a war. Rhenium is none of those things. Rhenium’s supply constraint is the most permanent kind: the element is genuinely one of the rarest on Earth — 1 part per billion in the crust, rarer than gold, rarer than platinum, rarer than all but a handful of stable elements — and it can only be obtained as a by-product of a by-product, from mines built for a different purpose, processed by companies optimizing for a different metal, at volumes that cannot be meaningfully increased without expanding copper mining globally.

    The lithium supply chain has a demand problem — consumption is growing faster than supply can respond. The copper supply chain has a capacity problem — there aren’t enough mines being built. Rhenium has a physics problem. There is no rhenium mine to build. There is no rhenium deposit to discover. There is only the copper-molybdenum system, the flue gas, the scrubber, and the 62 tonnes per year that represents the ceiling the aerospace industry is building its production ramp-up under.

    This is the kind of supply chain our Rare Earth Elements course was built to map — where the metal that allows a turbine blade to survive 1,700 degrees at 40,000 RPM is recovered from the smokestack exhaust of a molybdenum roaster in Santiago, at 62 tonnes a year, with a single company controlling 60% of global output, and the entire commercial aviation industry’s production backlog competing for every gram.

  • Neon, Krypton, and Xenon: The Invisible Gases That Make Every Chip on Earth

    On February 24, 2022, Russia invaded Ukraine. Within days, two companies in Odessa and Mariupol — Cryoin and Ingas — shut down their operations. Together, they had been producing roughly 50% of the world’s semiconductor-grade neon. Ukraine as a whole supplied approximately 70% of global neon, 40% of krypton, and 25-30% of xenon — the three noble gases that power the lasers used to etch circuits onto every advanced semiconductor chip manufactured on the planet. Neon prices in China rose tenfold within weeks. Krypton prices in Japan quadrupled. Xenon, which had traded at $15 per liter in 2020, spiked above $100. The world’s most sophisticated industry — semiconductor fabrication — discovered that it was dependent on gases captured from Soviet-era steel mills in a war zone, purified by two mid-sized companies in cities that were being bombed. The concentration wasn’t the result of geological scarcity. Neon, krypton, and xenon exist everywhere — they’re in the air you’re breathing right now. The concentration was an accident of industrial history, and the fact that the accident had never been corrected in three decades of post-Soviet globalization tells you something about how supply chains actually work: nobody fixes a single point of failure until it fails.

    What the gases do

    Neon, krypton, and xenon are noble gases — chemically inert elements that don’t react with other materials, which is precisely what makes them useful in environments where contamination would destroy the product.

    Neon’s critical application is semiconductor photolithography. The excimer lasers used to etch circuit patterns onto silicon wafers — the deep ultraviolet (DUV) systems that still produce the majority of the world’s chips — use gas mixtures that are approximately 96% neon, with small amounts of argon, krypton, fluorine, or xenon depending on the wavelength required. ArF (argon-fluorine) lasers at 193 nanometers and KrF (krypton-fluorine) lasers at 248 nanometers are the workhorses of the semiconductor industry. Every fab that runs DUV lithography consumes neon. The gas mixtures degrade during use and must be regularly replaced. TSMC, Samsung, Intel, and every other chipmaker on Earth — including the Chinese fabs the CHIPS Act was designed to compete against — need a continuous supply of ultra-high-purity neon to keep their lasers firing.

    Krypton serves double duty. In semiconductor manufacturing, it’s a component of the laser gas mixtures. Outside the fab, krypton fills the gap between panes in energy-efficient triple-glazed windows — a growing market as building energy codes tighten globally. It’s also used in high-intensity lighting for airports and stadium illumination.

    Xenon has the broadest application portfolio of the three. It’s an anesthetic in medicine — safer than nitrous oxide, with faster recovery times, though dramatically more expensive. It fills the flash tubes in high-end photography equipment. It’s used as a contrast agent in CT imaging. And — increasingly — it fuels the ion propulsion systems on communications satellites and Earth observation spacecraft. SpaceX’s Starlink constellation and Amazon’s Project Kuiper are driving xenon demand as satellite constellations proliferate. When a Starlink satellite adjusts its orbit, it’s expelling ionized xenon. The space economy’s growth curve is, unexpectedly, a noble gas demand curve.

    Why Ukraine had 70% of global neon

    The answer is Soviet military planning. During the 1970s and 1980s, the Soviet Union treated neon as a strategic material for high-powered laser weapons research — the kind of Cold War physics that the Battlefields of the Future course covers from the other side. Every major air separation unit in the Soviet Union was equipped with neon, krypton, and xenon enrichment facilities. Air separation units produce oxygen — which steel mills need in enormous volumes — and the noble gases are captured as by-products of the oxygen production process. The Soviet Union had massive steel mills. The massive steel mills had massive air separation units. The massive air separation units captured massive quantities of noble gases. When the Soviet Union collapsed, the steel mills ended up in Ukraine — particularly in the industrial cities of the Donbas and Black Sea coast — and the noble gas capture equipment went with them.

    For three decades, Ukrainian companies collected these gases, purified them to semiconductor grade (99.999% purity for neon), and exported them to the global chip industry at prices that made building competing production capacity uneconomical anywhere else. The same by-product supply structure that constrains indium and tellurium applies here: neon is a by-product of oxygen production, which is a by-product of steelmaking. You cannot produce more semiconductor-grade neon without operating air separation units at steel mills, and the economics of operating those units are determined by steel demand, not neon demand. Ukraine’s dominance wasn’t because Ukrainian neon was better. It was because Ukrainian steel mills had the gas capture equipment installed, nobody else had bothered to install it, and the resulting supply was cheap enough to discourage competition.

    What happened after the shock

    The predicted catastrophe — fabs shutting down, chip shortages deepening, economic disaster — largely didn’t materialize. The semiconductor industry responded faster than most analysts expected, for several reasons.

    First, the major chipmakers had prepared. After neon prices spiked 600% during the 2014 Crimean annexation, TSMC, Samsung, and others diversified suppliers and built strategic stockpiles. Most large fabs had 3-6 months of gas reserves when the 2022 invasion began.

    Second, recycling technology scaled rapidly. Modern DUV scanners can recover and purify over 90% of the neon used in each laser pulse, dramatically reducing virgin neon consumption per wafer. TSMC’s neon recycling program became a model for the industry.

    Third, new production came online. Linde had invested $250 million in a neon production facility in La Porte, Texas, after the 2014 scare. Chinese air separation companies expanded noble gas capture. South Korean and Japanese producers increased output. By 2023, the acute shortage had eased. Prices retreated from their peaks. The industry congratulated itself on resilience.

    Fourth — and this is the detail that changes the long-term picture — the technology is shifting. ASML’s extreme ultraviolet (EUV) lithography systems, which are required for the most advanced 5-nanometer and 3-nanometer chips, do not use neon. EUV lasers vaporize tin droplets rather than exciting noble gas mixtures. As EUV adoption expands and DUV’s share of leading-edge production declines, neon demand from the semiconductor industry will structurally decrease. The gas that nearly crippled the chip industry in 2022 may become less critical to the chip industry by 2030 — not because the supply chain was fixed, but because the technology moved on.

    Where it stands in 2026

    The noble gas market in 2026 is more diversified than 2022 but still structurally fragile. Ukrainian production has partially recovered — Cryoin’s Odessa facility has resumed operations, though at reduced capacity, and the Mariupol facilities remain destroyed. China, Japan, South Korea, and the United States have all expanded noble gas production and purification capacity. The acute price crisis is over.

    But the underlying architecture hasn’t fundamentally changed. Noble gases remain by-products of air separation at steel mills and industrial gas plants. The decision to capture them — rather than venting them into the atmosphere — is discretionary, driven by the economics of the gas market relative to the cost of operating the capture equipment. When neon was $100 per liter, everyone captured it. At lower prices, the incentive weakens. The antimony supply chain showed that price normalization after a crisis doesn’t mean the structural vulnerability has been resolved — it means the market has priced in the assumption that the crisis won’t recur.

    Xenon faces its own emerging constraint. Satellite constellation demand is growing faster than xenon supply. SpaceX’s Starlink alone operates over 6,000 satellites, each requiring xenon for station-keeping maneuvers. Amazon’s Kuiper constellation will add thousands more. If the space economy’s xenon demand outgrows the industrial gas industry’s xenon capture, the same by-product ceiling that constrains indium, tellurium, and iridium will constrain the propellant supply for the satellite industry. SpaceX has already begun testing krypton as a cheaper, more abundant xenon substitute in some Starlink applications — a substitution that trades performance for supply security, and that shifts demand pressure from one noble gas to another rather than relieving it.

    Why they share a lecture

    Neon, krypton, and xenon are the Rare Earth Elements course’s case study in accidental concentration — the supply chain vulnerability that exists not because of geological scarcity or deliberate resource nationalism, but because of industrial inertia. Nobody cornered the noble gas market. Nobody imposed export controls. The Soviet Union installed gas capture equipment at steel mills for laser weapons research. The equipment ended up in Ukraine. Ukraine supplied the world for thirty years. Then a war started and the supply vanished overnight.

    The nickel case is about deliberate resource nationalism — Indonesia’s export ban was a strategic decision. The gallium/germanium and antimony cases are about deliberate export controls — China’s licensing regime is an instrument of state policy. The noble gas case is about none of those things. It’s about a supply chain that concentrated by accident, stayed concentrated through inertia, and broke because of a war that had nothing to do with semiconductors. That’s a different category of supply chain risk — one that no critical minerals policy is designed to prevent, because it’s not the result of any policy at all. It’s the result of nobody looking at the map and asking what happens when the cheapest supplier is in a country that borders Russia.

    This is the kind of supply chain our Rare Earth Elements course was built to map — where the gases that power every chip laser on Earth were by-products of Soviet steel mills repurposed for Cold War laser weapons, concentrated in two cities in a war zone, and the industry that needed them only found out when the bombs started falling.