On July 16, 2026, the US Department of Energy (DOE) announced manufacturing technology for silane (SiH4) enriched to 99.9999% silicon-28 (Si-28), along with germane (GeH4) with germanium-73 (Ge-73)—a source of magnetic noise—suppressed to below 1 ppm. Oak Ridge National Laboratory (ORNL) separates the isotopes, while Pacific Northwest National Laboratory (PNNL) converts and purifies them into precursor gases that can be supplied to semiconductor manufacturing equipment. According to the DOE, contaminating isotopes have been reduced by at least a factor of 100 compared to commercially available materials.
What the US has gained is a track record in high-purification and a domestic process that converts enriched material into deposition gases. The United States halted large-scale stable isotope enrichment in 1998, and has since relied on research stockpiles and foreign products. This time, a pathway has been established that connects domestic processes all the way to the raw materials for the thin films used to make qubits. However, supply capacity, pricing, and the degree of improvement in actual devices have not been disclosed. Whether this materials-science achievement grows into a manufacturing foundation for quantum computers will be determined by verification going forward.
How 99.9999% Purity Eliminates Magnetic Noise from Qubits
While 92.22% of natural silicon is Si-28, the 4.69% that is Si-29 carries nuclear spin. Electron spins confined in quantum dots interact with the tiny fluctuating magnetic fields created by surrounding Si-29 atoms, causing loss of phase in superposition states. Increasing the proportion of Si-28, which has zero nuclear spin, reduces this noise source at the material level.
The difference in purity has already been observed in devices. In a 2024 study, a substrate with Si-29 reduced to 800 ppm showed an electron pure dephasing time 5,000 times longer than natural silicon. In evaluations of 300mm wafers by Intel and others, the free-induction decay time T2* averaged 0.6 microseconds for natural-Si quantum wells versus 5 microseconds for Si-28 quantum wells. Echo-based T2 also extended from 98 microseconds to 205 microseconds, and single-qubit Clifford gate fidelity for Si-28 devices reached approximately 99.9%.
The DOE's silane brings Si-29 down to below 1 ppm—more than two orders of magnitude lower than 800 ppm-class material, and even below the 2.3±0.7 ppm achieved through localized ion implantation in 2024. However, the nominal purity of locally processed crystals and that of thin-film growth gas are not measures that compete on the same scale for performance. The former demonstrates whether a narrow region can be highly purified, while the latter serves as an entry point for supplying material across an entire wafer.
The same mechanism applies to germanium. Ge-73, which makes up 7.75% of natural Ge, is the only stable germanium isotope with nuclear spin. A 2024 paper in Nature Materials found that hole qubits in natural Ge overlap with roughly one million nuclear spins, and when the magnetic field direction deviates from a narrow optimal condition, nuclear spin noise limits coherence. Reducing Ge-73 to below 1 ppm can ease, from the materials side, the burden of precisely aligning the magnetic field direction. However, the same experiment also captured charge noise. Quieting the isotopes alone does not eliminate interface defects or fluctuations in electrode voltage.
Separated at ORNL, Converted Back into Deposition Gas at PNNL
ORNL's electromagnetic isotope separation (EMIS) turns raw material into gas, ionizes it, and bends the ions' trajectories by mass within a magnetic field. Because ions of different mass numbers pass through orbits of different radii, Si-28, Ge-70, Ge-76, and others can be separately collected. Current equipment can collect multiple isotopes of the same element in a single run, allowing more flexible target switching than conventional calutrons.
However, obtaining an enriched element does not immediately produce a thin film for a quantum chip. Manufacturing processes such as chemical vapor deposition flow silane or germane into a reactor to deposit layers of Si or Ge. If ordinary isotopes get mixed in during conversion, the purity achieved at ORNL is lost. If metallic impurities or moisture remain, they introduce crystal defects and electrical noise.
PNNL has built facilities that convert enriched raw materials such as SiF4, GeF4, and GeO2 into silane and germane, removing chemical impurities to levels well below 1 ppm. It also directly enriches the gases using thermal diffusion isotope separation (TDIS). The TDIS technique that PNNL demonstrated with hydrogen chloride combines radial diffusion and vertical convection in a double-walled tube with an internal-external temperature difference, biasing isotopic components toward opposite ends of a column. Multiple columns are connected in series to amplify small separation differences. Because this method directly enriches the target gas, it reduces the number of conversion steps needed after enrichment into a different chemical form, thereby suppressing opportunities for ordinary isotopes to mix in. As of the DOE's March 2026 announcement, additional research was still needed for the design and safe operation of TDIS for silane and germane. In the July announcement, the DOE stated that multiple automated systems had been implemented, advancing to direct enrichment. The change over these four months lies in this process integration.
The ability to handle these gases at all is itself part of the manufacturing capability. Silane is spontaneously flammable in air, and germane has inhalation toxicity. PNNL's control system monitors hundreds of process variables. The supply chain encompasses not only high-purification but also processes for maintaining isotopic ratios, avoiding chemical contamination, and safely containing and shipping the material.
However, it cannot be confirmed that the upstream raw materials themselves are entirely sourced domestically. PNNL's March announcement described its starting point as "commercially available enriched compounds," and the DOE's July announcement also does not indicate the origin of the raw materials used by ORNL. The fact that enrichment and conversion facilities now exist within the United States should be viewed separately from whether raw material procurement has been fully domesticated.
The 1998 Supply Gap and China's 99.99%
Since the United States halted its Manhattan Project-era calutrons in 1998, its stable isotope enrichment capability has carried a gap. The DOE decided in 2009 to rebuild capacity at ORNL, and in 2011 brought a modern EMIS prototype online. Since then, throughput and separation performance have continued to improve. This announcement is not a record that appeared suddenly, but an extension of over 15 years of rebuilding facilities and expertise.
Supply risk has also shown up in quantities. A 2015 report compiled by a DOE advisory committee estimated that, apart from a research market of under 1 kg per year, the global market for bulk enriched stable isotopes—then worth about $100 million—was supplied 85% by Russia and 15% by URENCO of the Netherlands. This included Si-28 and Ge-76. This ratio does not reflect the current market share. Even so, it shows that the United States started from a position of relying on stockpiles and foreign companies.
Competitors have also begun building their supply chains domestically. China National Nuclear Corporation (CNNC) announced on June 30 that it had achieved China's first mass production of Si-28 on June 15, with isotopic abundance exceeding 99.99%. The DOE's announcement came 16 days later. While the digits after the decimal point differ from the US figure of 99.9999%, CNNC has not disclosed the product form, residual Si-29 level, annual production volume, or measurement method. The figures from the two countries cannot be directly translated into a quality ranking.
Still, the fact that the two announcements point in the same direction carries weight. Competition in quantum semiconductors has expanded beyond qubit design and control circuitry into a manufacturing problem: whether a country can repeatedly supply, domestically, materials with nuclear spin removed. Because Si-28 is also used in advanced semiconductors and metrological standards, investment in enrichment facilities will not end with a single application in quantum computing.
What Remains Beyond the Purity Record: Mass Production and Device Verification
The DOE has designated the National Isotope Development Center (NIDC) as the contact point for inquiries about silane and germane, and the center's catalog already lists Si-28 silane and Ge-70/Ge-76 germane as product forms. However, the public page uses a quotation-based system and does not disclose pricing or standard lot sizes for the 99.9999% product. Lead times and annual supply volumes are also unknown. The DOE's comparison of "100 times better than commercial products" is likewise not accompanied by information on the manufacturers used for comparison or the testing procedures.
Throughput deserves particularly careful scrutiny. In 2023, ORNL described EMIS as excelling in high purity and rapid target switching, while noting that the equipment handles quantities on the order of milligrams to grams. This is sufficient for research-scale thin films for silicon quantum chips, but at the stage where multiple companies engage in repeated production on 300mm wafers, the ability to continuously supply uniform gas in volume becomes decisive. The Stable Isotope Production and Research Center under construction at ORNL is scheduled to begin phased operation in 2028, meaning a substantial expansion of supply capacity is still some way off.
The DOE has also set a 2028 demonstration target on the quantum computing side. Quantum Genesis, launched in June, aims to demonstrate by 2028 a fault-tolerant quantum system equipped with somewhat fewer than a few hundred logical qubits, usable for scientific research. Si-28 and Ge-70/76 directly benefit semiconductor spin qubit systems, and not all participating approaches use these materials. Even so, this suggests that the DOE is trying to align the timelines of quantum computing hardware and materials supply, so that candidate architectures are not held up waiting for raw materials when transitioning to mass production.
The next evaluation will be determined not by purity certificates but by three stages: shipment, wafer, and qubit. How much quantity NIDC supplies and on what lead time; whether the grown films maintain sub-1 ppm purity across the wafer surface; and how far coherence and gate fidelity extend in devices built from them. If these three points are demonstrated by the time facilities become operational in 2028, the United States will have reclaimed the stable isotope manufacturing capability it once lost, rebuilt as an industrial foundation for quantum semiconductors.