Heavy water, chemically known as deuterium oxide (D₂O), is the foundational technology that enables the CANDU (CANada Deuterium Uranium) reactor series to operate with unparalleled fuel flexibility. Unlike light-water reactors (LWRs) that require enriched uranium, CANDU stations use heavy water both as a neutron moderator and as the primary heat transport coolant. This dual role creates a neutron economy so efficient that fission can be sustained using natural uranium fuel (0.7% U-235). The production of high-purity heavy water, the management of its inventory, and the infrastructure supporting its supply chain are therefore directly linked to the operational viability and economic performance of the global CANDU fleet. Understanding the physics of deuterium, the industrial processes used to separate it from ordinary water, and the logistics of maintaining its purity provides a complete picture of how this unique material supports reliable, low-carbon electricity generation.

The Physics of Heavy Water Moderation

The core advantage of heavy water over ordinary light water lies in the nuclear properties of the deuterium atom. The nucleus of deuterium contains one proton and one neutron, giving it a mass approximately double that of ordinary hydrogen (protium). This additional mass fundamentally alters the way neutrons interact with the medium. The key parameter is the thermal neutron absorption cross-section. For hydrogen (H-1), this cross-section is approximately 0.33 barns. For deuterium (H-2), it drops to just 0.00033 barns — a factor of 1,000 lower.

In a nuclear reactor, neutrons born from fission travel at extremely high speeds (fast neutrons). To maintain a chain reaction, these neutrons must be slowed down, or moderated, to thermal energies. A moderator performs this task by allowing neutrons to collide with its nuclei. In light water, the collisions are effective at slowing neutrons, but the protium atoms absorb a substantial fraction of the neutron population, rendering them unavailable for further fission. This parasitic absorption makes it necessary to enrich the uranium fuel to around 3–5% U-235 to compensate for the lost neutrons. Heavy water, by contrast, provides excellent slowing-down power while absorbing very few neutrons. This high neutron economy is what allows CANDU reactors to load natural uranium directly into the core. The ability to operate without enrichment simplifies the front end of the fuel cycle, strengthens proliferation resistance, and insulates operators from the geopolitical and economic volatility associated with enrichment services. Furthermore, the excellent neutron economy allows the reactor to convert U-238 into fissile plutonium in situ, deriving significant energy from the fertile component of natural uranium.

The slowing-down power of a moderator is quantified by the moderating ratio, which combines the logarithmic energy decrement per collision with the scattering cross-section divided by the absorption cross-section. For light water, this ratio is about 70; for heavy water, it exceeds 6,000. This stark contrast explains why a heavy water reactor can operate with natural uranium: the moderator does not steal the precious neutrons that are needed to maintain the chain reaction. The practical consequence is that a CANDU core requires a larger moderator volume than a comparable LWR, but the payoff in fuel cycle independence is enormous.

Industrial Production of Reactor-Grade D₂O

Deuterium exists in natural water at a concentration of roughly 150 parts per million (0.015%). To function effectively as a moderator and coolant, this concentration must be raised to a minimum of 99.75% D₂O. This enrichment process is thermally and technologically demanding, making heavy water a high-value strategic material. The physical separation exploits the slight mass differences between H₂O and D₂O, which manifest in different vapor pressures, reaction rates, and equilibrium constants in chemical exchange reactions.

The Girdler Sulfide (GS) Process

The workhorse of industrial heavy water production is the Girdler sulfide process, which was deployed at an enormous scale in Canada, the United States, India, and elsewhere. The GS process exploits the temperature-dependent equilibrium of the isotopic exchange reaction between water and hydrogen sulfide gas: H₂O(l) + HDS(g) ⇌ HDO(l) + H₂S(g). At lower temperatures, deuterium favors the liquid phase. At higher temperatures, it favors the gas phase. By operating a dual-temperature tower cascade, deuterium can be systematically stripped from natural water and concentrated.

In a typical GS plant, water flows down a column while H₂S gas is circulated upward. At the bottom of the column, operating around 130°C, deuterium migrates into the gas. The gas then flows to the top of a cold section, operating around 30°C, where the deuterium transfers back into a fresh water stream, now enriched. Hundreds of such stages are required to reach a concentration of 10–20% D₂O. This intermediate product is then fed to a vacuum distillation unit for final enrichment to reactor grade. The GS process consumes substantial amounts of thermal energy and requires the handling of large volumes of toxic hydrogen sulfide gas. The Bruce Heavy Water Plant in Ontario was the world's largest GS facility, at its peak producing over 2,400 tonnes of D₂O annually before its permanent closure in 1997 due to changing market dynamics and lower-than-expected demand for new CANDU builds. The combination of high capital costs, energy intensity, and unique safety hazards associated with H₂S has motivated research into alternative production methods.

An often-overlooked aspect of the GS process is its environmental footprint. The large water flows and energy requirements mean that a plant producing 1,000 tonnes of D₂O per year typically consumes around 500–700 MWt of thermal input and discharges millions of cubic metres of depleted water. Modern design improvements have reduced steam consumption, but the process remains capital intensive. Nevertheless, the GS process produced the majority of the world's current heavy water inventory, and its legacy plants in India and Argentina continue to operate.

Alternative and Emerging Separation Technologies

Several other technologies exist for deuterium separation, each with specific applications. Water distillation exploits the slightly lower vapor pressure of D₂O compared to H₂O. While the separation factor is very small (around 1.05 at atmospheric pressure), it requires an enormous number of stages and high energy input. It is primarily used today for upgrading moderately enriched heavy water to reactor grade rather than for primary production from natural water.

Electrolysis leverages the kinetic isotope effect; because of the higher zero-point energy of the O–H bond compared to the O–D bond, protium is preferentially released as hydrogen gas at the cathode. This enriches deuterium in the remaining liquid. Electrolysis was famously used at the Norsk Hydro plant in Vemork, Norway. However, the extreme electricity consumption makes it uneconomical for bulk production. It is often used as a final polishing step or in small-scale facilities.

Advanced methods currently under investigation include laser isotope separation (LIS), which targets the slight shift in absorption spectra between H₂O and HDO molecules. While laboratory-scale success has been achieved, scaling LIS to the thousands of tonnes required for a reactor core remains a significant engineering challenge. Other techniques include bithermal hydrogen-water exchange using platinum catalysts, which avoids hazardous H₂S, and membrane-based separation technologies that rely on differential permeation rates. The Combined Electrolysis and Catalytic Exchange (CECE) process is one of the most promising next-generation methods. It integrates water electrolysis with catalytic isotopic exchange in a continuously recirculating loop. CECE can achieve single-stage separation factors of 5–10, compared to roughly 2–3 for the GS process, potentially reducing the number of stages and the energy footprint. Pilot plants in Canada and Japan have demonstrated the technical feasibility, but commercial-scale deployment has not yet occurred.

The Symbiotic Relationship Between CANDU Design and D₂O

The design of the CANDU reactor is fundamentally inseparable from the properties of heavy water. The reactor core consists of a large, low-pressure tank called the calandria, which holds the heavy water moderator at a relatively low temperature. Penetrating the calandria are hundreds of horizontal pressure tubes, each containing fuel bundles. The primary coolant — separate from the moderator — is also heavy water, which passes through the pressure tubes to extract the heat generated by fission. This separation of moderator and coolant is a critical design feature that contributes to the reactor's high neutron economy and operational flexibility.

Natural Uranium Fuel Cycle

The ability to use natural uranium without enrichment is the most significant economic and strategic advantage of the CANDU system. It decouples the reactor operator from the international enrichment market, which is dominated by a small number of countries and requires complex safeguards arrangements. Natural uranium fuel bundles are simpler and cheaper to fabricate than the enriched pellets required for LWRs. For countries without enrichment infrastructure, such as Argentina, Romania, and South Korea at the time of their initial builds, the CANDU design offered an independent path to nuclear energy. This autonomy continues to be a powerful driver for the selection of the PHWR design for new nuclear programs.

The natural uranium cycle also has implications for spent fuel management. Since the initial enrichment is low, the spent fuel from a CANDU contains about 0.3% fissile material, compared to about 0.8–1% from an LWR. Less plutonium is produced per unit of energy, but the total volume of spent fuel per MWh is higher because of the lower burnup (around 7–8 GWd/t vs. 40–50 GWd/t for LWRs). This trade-off is acceptable given the proliferation resistance benefits and the freedom from enrichment services.

Online Refueling and Capacity Factors

Heavy water moderation is a direct enabler of online refueling. In an LWR, the entire core must be shut down and depressurized for refueling, typically every 18–24 months. In a CANDU, the horizontal pressure tubes allow fueling machines to access each channel independently while the reactor remains at power. Because the heavy water moderator maintains a stable neutron flux distribution and the natural uranium fuel has significant reactivity reserves (converting U-238 to Pu-239 over its lifespan), the reactor can accommodate a continuous reshuffling of fuel bundles. This results in excellent capacity factors, often exceeding 85–90%, as there are no forced outages for routine refueling. The operator can also tailor the fueling scheme to specific power demands, flatten the flux profile, or burn advanced fuel cycles without a major core redesign.

Online refueling also enables flexible power maneuvering. A CANDU station can adjust its output to match grid demand while maintaining a nearly constant refueling rate. This load-following capability is increasingly valuable in grids with high penetration of variable renewable energy sources. The heavy water moderator acts as a thermal buffer, providing stability during transients and allowing the reactor to respond quickly to load changes.

On-Site Heavy Water Management

Operating a CANDU station requires dedicated systems and procedures for managing the heavy water inventory. The D₂O inventory represents a substantial capital asset, and its loss or degradation has direct economic consequences. A typical 700 MWe CANDU 6 reactor contains approximately 450 to 500 tonnes of heavy water. At current replacement costs of several hundred dollars per kilogram, the on-site inventory can be valued in the hundreds of millions of dollars. Therefore, minimizing leaks, preventing isotopic dilution, and controlling radiological contamination are central operational priorities.

Maintaining Isotopic Purity

Ingress of light water from various sources — including steam generator leaks, condensed moisture from the atmosphere, and makeup water additions — will degrade the D₂O concentration over time. Even a small drop in isotopic purity degrades the neutron economy and can force a reduction in reactor power. To counter this, every CANDU station is equipped with a heavy water upgrading plant. These units typically employ vacuum distillation, operating at reduced pressure to decrease the boiling point of water and minimize the energy required for separation. The upgrader takes in downgraded D₂O from the moderator and coolant circuits and outputs high-purity D₂O (above 99.75%), while rejecting the light water as effluent. Continuous operation of the upgrader is essential for maintaining the reactor's design performance and extending the life of the primary inventory.

The upgrading process also removes chemical impurities that may accumulate, such as dissolved gases, corrosion products, and organic compounds. These contaminants can affect the pH of the heavy water, increase radiolytic decomposition, or interfere with the neutronics. The upgrader is therefore a combined isotopic and chemical purification system, often preceded by filters, ion-exchange columns, and degassifiers. The design and operation of the upgrading plant are tailored to the specific isotopic degradation rates of each station, which depend on leakage rates, makeup water sources, and the age of the fuel.

Tritium Control and Regulation

The most significant radiological concern associated with heavy water operations is the production of tritium (H-3). Tritium is generated via the neutron activation of deuterium: ²H (n, γ) ³H. Over time, the heavy water in the moderator and coolant accumulates tritium, which can reach high specific activity levels. Tritium is a beta emitter with a half-life of 12.3 years. While the beta particles are low-energy and do not penetrate the skin, tritium is biologically hazardous if ingested or inhaled as tritiated water (HTO).

CANDU stations manage tritium exposures through a combination of engineering controls and active removal. Tritium Removal Facilities (TRFs) use a combination of catalytic exchange and cryogenic distillation to extract tritium from the heavy water stream. The extracted tritium is stabilized as a metal hydride and can be stored, or purified and sold for commercial applications such as self-luminous lighting and medical tracers. The Darlington Nuclear Generating Station operates one of the largest TRFs in the world, significantly reducing radiation fields in the plant and minimizing environmental emissions. Regulatory oversight ensures that tritium discharges are kept as low as reasonably achievable, typically well below regulatory dose limits for the public.

The tritium concentration in CANDU moderator systems can reach several tens of curies per kilogram after years of operation. To keep radiation exposures to workers as low as possible, the TRF at Darlington has been extracting tritium since the late 1980s, reducing the specific activity of the moderator by more than a factor of 10 compared to stations without such facilities. This not only lowers personnel dose rates but also reduces the risk of inadvertent releases. The Canadian Nuclear Laboratories continue to innovate in tritium handling and isotope separation, maintaining Canada's position at the forefront of heavy water science.

Leak Detection and Economic Recovery

Given the high cost of D₂O, leak detection is a continuous operational task. CANDU stations employ extensive networks of dew point sensors, tritium-in-air monitors, and humidity detectors to identify even minuscule leaks. Heavy water leaks are categorized as "visible" or "invisible" (vapor). A steam generator tube leak, for example, can allow light water to ingress into the primary coolant, degrading the isotopic purity. Leak detection and recovery systems are sized to handle design-basis events and ensure that the economic losses are contained. Any recovered heavy water is filtered, cleaned, and sent to the upgrading plant for isotopic and chemical purification before being returned to the reactor systems.

The financial incentive for tight leak management is significant. A loss rate of just 1 tonne per year at $300/kg represents an ongoing cost of $300,000 annually. Over a 30-year operating life, that is $9 million of lost inventory. Modern CANDU stations aim for unrecoverable losses of less than 0.5 tonnes per year, achieved through aggressive leak detection programs, improved valve seal designs, and vapor recovery systems. The recovered heavy water from ventilation systems can represent a substantial fraction of the total makeup requirement, reducing the need to purchase fresh D₂O.

Global Supply Chain and Economic Impact

The global heavy water supply chain is relatively mature and consolidated. The initial inventory for a new CANDU is typically supplied under a lease arrangement from a national stockpile, such as that managed by Atomic Energy of Canada Limited (AECL) and now Canadian Nuclear Laboratories. This avoids the massive upfront capital expenditure required to construct a new production facility. The lease model also provides an economic incentive for operators to minimize losses and maintain inventory quality.

For the existing fleet, demand for fresh heavy water is primarily driven by make-up requirements for losses and for new builds. India operates its own GS plants to supply its domestic PHWR fleet, having overcome historical shortfalls to become self-sufficient. Argentina produces heavy water at the PIAP plant in Neuquén, which supplies the Embalse and Atucha stations. The supply chain includes not just primary production but also the recycling of heavy water from decommissioned reactors. As older CANDU units reach the end of their operating lives, their heavy water inventories can be recovered, upgraded if necessary, and redeployed to other reactors or placed into strategic reserves. This circular economy reduces the need for new production and provides a buffer against supply disruptions. The World Nuclear Association maintains detailed data on global production capacity and the economics of the heavy water market.

The cost of heavy water has fluctuated over the decades. In the 1970s, when GS plants were being built, the price was around $100/kg in inflation-adjusted terms. After the closure of the Bruce plant and the end of Canadian production, prices rose to over $300/kg. Today, the market is relatively stable, with Indian and Argentine production meeting most demand. The lease rate for heavy water is typically around 5–7% of the replacement value per year, which covers the cost of capital and the risk of loss. For a 500-tonne inventory at $300/kg, the annual lease fee is $7.5–10.5 million, a manageable expense for a station generating several hundred million dollars in revenue.

Future Trajectories: Production, Novel Reactors, and Advanced Fuel Cycles

Looking ahead, the role of heavy water in nuclear energy is likely to evolve beyond the traditional CANDU 6 fleet. The existing inventory is sufficient to support the current operating fleet for decades. However, emerging reactor concepts and advanced fuel cycles may drive new demand for heavy water or for deuterium itself.

Advanced Production Technology

While the GS process is mature, its high capital and operating costs have spurred interest in more efficient methods. The Combined Electrolysis and Catalytic Exchange (CECE) process is a promising advanced separation technique that offers the potential for lower energy consumption and higher single-stage separation factors compared to the GS process. The CECE process integrates hydrogen electrolysis with catalytic isotopic exchange, allowing for the efficient concentration of deuterium. Research into membrane-based separations and nanomaterial filters could eventually provide a step-change in production efficiency. A recent study by the IAEA's programs on heavy water reactors reviewed these technologies and concluded that CECE is the most mature alternative for future plants.

Another promising avenue is the use of geothermal or waste heat to drive the GS process, which could significantly reduce its carbon footprint. The integration of heat pumps and advanced heat exchangers might also improve the overall energy efficiency. However, none of these technologies have yet been deployed at commercial scale, and the GS process will likely remain the benchmark for the next decade.

Advanced Fuel Cycles

The excellent neutron economy of heavy water reactors makes them ideal platforms for burning advanced fuels. Considerable research has been undertaken into using CANDU reactors to consume surplus plutonium from dismantled nuclear weapons or from light water reactor spent fuel. The flexibility of the online refueling system allows for the precise management of the reactivity and flux profiles required for these mixed-oxide (MOX) fuels. Furthermore, the heavy water moderator is uniquely suited to supporting a thorium fuel cycle. Thorium is three to four times more abundant than uranium, and in a CANDU reactor, Th-232 can be converted to fissile U-233 with high efficiency. The IAEA has coordinated multi-year studies on the thorium fuel cycle in PHWRs, showing that a self-sustaining thorium-U-233 cycle is feasible with careful neutron management.

The use of heavy water reactors to burn minor actinides from LWR spent fuel is another active area of research. The high thermal neutron flux and the ability to adjust the fuel composition on the fly make CANDU reactors attractive for reducing the long-term radiotoxicity of nuclear waste. These advanced fuel cycles would require modifications to the fuel handling and possibly to the coolant chemistry, but the existing heavy water infrastructure provides a solid foundation.

Small Modular Reactors (SMRs) and Fusion

Several SMR designs, including those based on evolved CANDU pressure-tube technology, consider heavy water as a reference moderator or coolant. The use of heavy water in SMRs could provide the same fuel flexibility and proliferation resistance advantages at a smaller scale. For instance, the Canadian Nuclear Laboratories' concept for a small pressure-tube reactor leverages the existing heavy water expertise to deliver a low-cost, factory-buildable unit. Separately, deuterium is a primary fuel for fusion reactors (D-D and D-T reactions). While fusion is still in the development phase, a future fusion industry would create a new and significant demand for deuterium separation technologies, potentially driving innovations that also benefit CANDU operations. The ongoing work at Canadian Nuclear Laboratories on tritium extraction and handling is directly relevant to fusion fuel cycles.

The heavy water supply chain may also see changes if fusion becomes commercial. A 1 GWe fusion plant using D-T fuel would consume about 50 kg of tritium per year, which would need to be bred from lithium. The deuterium required would be about 5 kg per year, a negligible fraction of current production. However, the start-up inventory for a fusion reactor could require several hundred kilograms of tritium, which would need to be produced in fission reactors or accelerators. This could create a symbiotic relationship between CANDU reactors (which produce tritium) and fusion reactors (which consume it).

Conclusion

Heavy water is much more than a coolant or a moderator; it is the defining technological choice that shapes the economics, safety profile, and strategic value of the CANDU reactor system. The unique nuclear properties of deuterium enable the use of natural uranium, providing fuel cycle independence that is rare in the global nuclear industry. The industrial production of heavy water, historically dominated by the Girdler sulfide process, built the national inventories that continue to support the fleet today. Careful on-site management, including isotopic upgrading and tritium removal, ensures the long-term operational viability of these plants. As the industry looks toward advanced fuel cycles, thorium utilization, and small modular reactors, heavy water will likely remain a central element of the reactor physics and engineering landscape, providing a proven pathway to clean, sustainable, and secure nuclear energy. The continued stewardship of existing inventories and investment in advanced separation science will ensure that this strategic resource is available to power the reactors of the future. The international community of CANDU operators, coordinated through organizations like the CANDU Owners Group, continues to share best practices for optimizing heavy water management and extending the life of this critical asset.