Decommissioning as Residual Care: Transforming Nuclear Reactors into Waste

Abstract

This paper takes nuclear decommissioning as an entry point in the way we understand and value innovations from what they leave behind. Nuclear decommissioning allows one to capture transformations through which an infrastructure that has lost its initial use and value is transformed into storable nuclear waste. We label this process “residual care,” a form of care in practice dealing with neglected things, and analyze it through the case of graphite-moderated reactors in France. Once at the forefront of nuclear innovation in the 1950s–60s, these machines have turned into poorly known piles of degrading residues that are now planned to be fully decommissioned in 2100. The irradiated graphite they contain, long considered a negligible background element, does not fit the current standards of nuclear waste management and must be turned into proper waste. We argue that residual care changes the nature of the infrastructure—as different facts appear, such as the radiological properties of graphite and the fragility of its containment devices. What was once a demonstration of mastery has become the object of a fragile demonstration of care.

Introduction

The current movement toward a nuclear “revival,” “renaissance,” or “relaunch” illustrates a general tendency of contemporary innovation societies: even in domains where problematic legacies are a visible matter of concern, novelty continues to dominate public valuations of innovation. After years of doubt following the 2011 Fukushima disaster, phasing out decisions and controversies surrounding nuclear waste management, the nuclear industry in many countries has recently regained momentum. This resurgence is driven by forward-looking promises of new technological breakthroughs and solutions to global warming and energy shortages, whether through extending the lifespan of existing reactors or launching a new wave of investments. Nuclear waste, long considered a burdensome legacy of technological hubris, is reframed in this enthusiastic rush to innovation. As Jessica Johnson, a member of Nucleareurope and panel moderator in an international nuclear summit in March 2024, puts it: “With deep geological repositories around the corner, in a way the problem of nuclear-spent fuel has been solved.” ¹ In this framing, little space is left for other matters of concern.

But even if we were to subscribe to this questionable optimism, the fact is that spent nuclear fuel represents only a small portion of the nuclear industry's material legacy. There is much to learn by broadening the scope: the entirety of the current nuclear infrastructure is turning into cumbersome nuclear residues that will have to be taken care of and turned into massive amounts of low-level radioactive waste. This article uses the case of nuclear decommissioning—specifically the handling of irradiated graphite in France's early reactors—to explore how we value innovation through its residues (Pottin 2024; Felt 2025). After outlining key research on nuclear waste, we introduce our conceptual lens of residual care, followed by the case study and methodological approach. Our findings trace three dimensions of residual care: the noticing of irradiated graphite as a matter of concern by experts, its fragile infrastructure, and the process of rendering it storable. We argue for a shift in analytical focus from waste itself to the transformation of residues into waste, examining what it means to become a residue and the value tensions in residual care. Ultimately, we call for attention to residual care and the related arts of noticing that can help us move our attention from mastery to maintenance, or from short-term technological solutionism (Morozov 2013) to long-term care.

From Nuclear Waste to the Transformation of a Nuclear Infrastructure into Waste

The nuclear industry leaves us with a material legacy that is generally considered to be especially hazardous and cumbersome. As part of recent efforts to make nuclear energy appear as a “green” technological solution to climate change, its advocates have apparently needed to demonstrate that they know how to take care of this legacy. ² But what does it mean to take care of nuclear waste?

Attempts to address this question have tended to focus on a very impressive, yet small portion of the nuclear material legacy: high-level and long-lived nuclear waste (HLW), primarily originating from spent nuclear fuel, and the projects to build deep geological repositories that have been launched in several countries to manage HLW. Science and technology studies scholars have engaged in intense debates surrounding HLW over the past two decades. It has served as a paradigmatic case for exploring future-oriented responsibility (Adam and Groves 2011), the democratic tensions arising in the absence of clear scientific expertise (Barthe 2005), the political agency of material things (Gregson 2012), and the democratic governance dilemmas posed by potentially irreversible decisions (Parotte 2018). Others have highlighted the profound epistemic challenges of storing radioactive waste over millennia (Macfarlane and Ewing 2006). Given that some radionuclides have half-lives of several hundred thousand years, HLW has prompted reflections on how societies reckon with the long-term consequences of technology (Ialenti 2020) and how existing institutional and financial systems attempt to account for “deep time” (Saraç-Lesavre 2021).

While these studies have yielded important insights, their exclusive focus on HLW limits their scope: HLW, though undeniably hazardous, constitutes an estimated 1 percent of nuclear waste (IAEA 2022). The focus on deep geological repositories can overshadow the many other networks in which nuclear residues are embedded. Several authors have already shed light on other aspects of the multiscalar problems of nuclear things (Hecht 2018). Some have focused on the different geographies of low-level waste (Garcier 2021; Martinais 2021), and on the evolution of standards of management that were associated with them—from dispersion into the sea (Hamblin 2009) to long-term surface storage. Others have questioned the ambiguous category of “legacy waste” (Kasperski 2019) and how it relates to nuclear remnants not only as a burden, but also as a heritage (Rindzevičiūtė 2021). Finally, some have recently started to study nuclear decommissioning, as a matter of memorial politics (Novac 2019), of material transformation (Rosini 2024), as a prioritization issue (Blanck 2021), as a modification of landscapes (Kalshoven 2023), as a reconfiguration of nuclearized territories (Storm 2020), or as entailing a “politics of impossibility” (Cram 2023).

Our article contributes to this literature by taking the effort to decenter from high-level waste and deep geological repositories one step further. Our main argument is as follows. The issue with nuclear waste is not only a matter of disposal, or of putting it in the right place for a very long time: first, it is a matter of transforming nuclear energy infrastructures into “storable” waste. In other words, nuclear waste is not just waiting for someone to pick it up and store it somewhere; there are a series of operations, practices, regulations, and negotiations that need consideration before nuclear waste can be deemed safely disposable. In this article, we turn our attention to nuclear decommissioning as a process of transforming infrastructure into waste.

In shifting focus from disposal to transformation, we draw on insights from waste and discard studies. Scholars in this field have shown that being waste is not an intrinsic quality but the result of sociomaterial processes (Gregson and Crang 2010), shaped through negotiations over ambiguous entities (Hird 2012) and situated within broader sociotechnical regimes of valuation and discard (Gille 2014; Liboiron and Lepawsky 2022). While this perspective has only recently been applied to nuclear contexts (Garcier 2021; Weber 2022), we build on and extend this work through a close examination of a case of nuclear decommissioning. Yet, this process of transforming an infrastructure into storable nuclear waste still needs further characterization. To do so, we propose the concept of “residual care.”

Residual Care

What is the status of the material things that need to be transformed into storable nuclear waste? To capture the sociotechnical ambiguity of the infrastructure awaiting decommissioning, we will employ the concept of residues. Some scholars have recently borrowed this term from chemistry and given it a broader meaning in social science, shedding light on their “unruliness,” their association with gappy modes of governance and with the treatment of people and places as being disposable (Boudia et al. 2021; Hecht 2023). This concept allows one to depart from pre-existing actor categories such as “waste,” “recoverable materials,” “effluents,” or “externalities” (Pottin 2024). Building on these conceptual reflections, we broaden the lens to understand innovation through what it durably leaves behind (Felt 2025)—not only waste and hazardous materials, but also infrastructural remnants and the epistemic residues shaped by the ways of thinking and knowing that accompanied its construction and operation.

Transforming ambiguous residues into “proper” waste is not only a matter of governance, we argue, but also a “matter of care” (Puig de la Bellacasa 2011). Issues regarding nuclear power tend to be framed in the terms of a “logic of choice” (Mol 2008): debating already designed technical solutions, weighing the risks and advantages of nuclear production in comparison with other energy sources, and so on. This framing often obscures the “logic of care” that emerges in the ongoing engagement with nuclear residues. In this article, we follow the actors involved in different ways in caring for nuclear decommissioning—nuclear safety engineers, decommissioning engineers, concerned citizens, critical experts—and show that they are not involved in discrete decisions concerning what should be done, but in a continuous and relational process between themselves and an evolving infrastructure.

Three main features characterize this process. First, it requires attentiveness to matters long neglected—and still largely disregarded—in the governance of nuclear infrastructures (Puig de la Bellacasa 2011). Second, it conceives of infrastructure not as a stable production device, but as a fragile assemblage of evolving materials and fallible human practices—making residual care closely aligned with the attentive work of maintenance (Denis and Pontille 2024). This perspective also calls for cultivating what Anna Tsing (2015, 24) has evocatively termed the “arts of noticing,” enabling us to fully “appreciate the multiple temporal rhythms and trajectories of the assemblage.” Third, residual care resists standardized procedures and fixed classifications; it demands improvisation, tinkering, and a readiness to navigate the uncertain and unexpected (Mol, Moser, and Pols 2010). In this, we align with work on high-level waste (Kasperski and Storm 2020) and mining residues (Ureta 2016), though our approach adopts a critical care perspective by offering an analysis that is “cautious, thoughtful, and considered” (Martin, Myers, and Viseu 2015, 635); as they remind us, care is also a “selective mode of attention.”

Focusing on residual care does not mean that we minimize the issues of residual governance (Hecht 2023): more often than not, these practices are intertwined with strategies that consist of delaying the bulk of residual care work, maintaining the promise of technological mastery by which the nuclear justifies its cumbersome presence, governing critique (Topçu 2013), or depoliticizing hazardous situations (Polleri 2020). Yet, we follow a slightly different purpose. We believe that unfolding the logic of residual care can prove politically useful in a context of nuclear “revival,” in which residues are at risk of being rendered invisible, neglected, buried as it were, under the promises of future innovation. When we conducted this inquiry, the French nuclear safety agencies were in a process of reform, merging the technical expertise of one agency with the regulatory authority of another. This unsettling context was very present during the interviews as potentially participating in this neglect for the care needed to properly decommission nuclear installations.

Case Study, Materials, and Methods

In what follows, we use a specific case study of decommissioning the oldest industrial-scale nuclear reactors in France to ask the question: what exactly is being decommissioned, that is, what is the status of the residual stuff that is being carefully transformed into different types of waste?

Once emblematic of nuclear innovation, the graphite-moderated and gas-cooled UNGG (uranium naturel graphite-gaz) reactors have become disregarded machines—now perceived as deteriorating remnants embedded in a long-overlooked infrastructure for managing irradiated graphite. This material, in its various forms, fails to conform to existing standards of nuclear waste classification and therefore must be carefully reconfigured into “proper” waste. Below we first trace how graphite, once a peripheral component of nuclear fission, emerged as a focal point of residual care. Next, we examine the fragile containment infrastructures where irradiated graphite currently resides, awaiting transformation into waste. Finally, we analyze the envisioned displacements and classificatory efforts involved in rendering graphite into waste forms deemed suitable for final disposal.

As part of a larger project on innovation residues, we conducted fourteen semidirected interviews between November 2023 and July 2024 with a total of seventeen people engaged in different aspects of residual care for irradiated graphite in France. These interviews allowed us to gather varied insights on the material history of irradiated graphite while engaging in self-reflexive or critical discussions on the industry's practices of residual care. Interviewees included industry professionals from the two French nuclear safety agencies (Institut de radioprotection et de sûreté nucléaire [IRSN] and Autorité de Sûreté Nucléaire [ASN], n = 7), the nuclear waste management agency Agence nationale de gestion des déchets radioactifs (Andra) (n = 2), and a company specialized in decommissioning (n = 1). We also interviewed critical experts (n = 3), two of whom were former industry professionals, as well as local representatives animating local information committees (n = 2), and activists engaged in citizen monitoring (n = 2). Attempts to engage with members of Electricité de France (EDF), the main operator of these reactors, were ultimately unsuccessful. Thus, our analysis is in a similar situation as the actors of residual care that we interrogated: having to deal with the operator's silences. ³

We also collected archival data from the IRSN (now ASNR). We found materials documenting the technical and regulatory transformation of a nuclear reactor into a decommissioning site in the 1990s, and reports concerning the practices and knowledge of irradiated graphite handling from 1968 to 1988. This allowed us to scrutinize the routinized practices of residual care, and to gain historical depth into what the industry knows about irradiated graphite.

Finally, we analyzed various publicly available documents: technical reports issued by nuclear safety and waste management (some of which were indicated or communicated by interviewees), reports in which operators document and justify their decommissioning strategy, international reports documenting decommissioning practices and discussing the best options for residual care, and scientific literature on the properties of irradiated graphite.

Empirical Findings

Noticing Irradiated Graphite

Residual care generally arrives as an afterthought. To understand the concerns that surround the transformation of graphite reactors into waste, one needs to first understand that, for decades, the material agency and the fragility of residues have remained a neglected thing of nuclear innovation.

The nine French graphite-moderated reactors are monumental machines, all commissioned between 1956 and 1972 and shut down between 1973 and 1993. Unlike the now-dominant pressurized water reactors, they use graphite—a solid material derived from coal—instead of water ⁴ to moderate nuclear fission. These reactors are housed in concrete buildings approximately 50 meters tall and contain, on average, 2,500 tons of graphite suspended several meters above ground by a steel framework. Over decades of operation, the graphite has become irradiated, making it one of the main challenges in the complex process of decommissioning. This type of large-scale decommissioning has never been undertaken globally: of the ninety graphite-moderated reactors worldwide, only two small-scale examples—one in the United States and one in the United Kingdom—have been fully dismantled to date (EDF 2017). UNGG decommissioning represents a very complex challenge for the industry, and EDF and the Commissariat à l’Energie Atomique (CEA), their French operators, ⁵ have been for a long time postponing the bulk of decommissioning work.

Why is dismantling UNGG reactors so complex? When they were built, their afterlife was largely overlooked. As former nuclear engineer Bernard Laponche puts it in a 2017 radio interview: “Psychologically it is very strange…the reactors were seen as eternal, and so neither decommissioning nor waste was taken into account” (France Culture 2017). Many engineers and experts confirm that this lack of anticipation is a key reason why dismantling them today is so challenging. But the issue goes deeper: it is not only their future that was neglected, but also a significant part of their materiality. For decades, the thousands of tons of graphite inside remained backstage, seen merely as passive infrastructure supporting the technoscientific project of nuclear fission.

The first UNGG reactors were built for military purposes. In the 1950s, the CEA settled for graphite-moderated reactors against the competing technology of heavy-water reactors as the fastest way to access weapon-grade plutonium. The reactors G1, G2, and G3 were launched in 1956, 1958, and 1960, in a spirit of secrecy and urgency. The most pressing concerns then were to gain access to “the” bomb, to make France a major actor in a Cold War world order, and to thus redefine the national identity of France and its international radiance through technological craft (Hecht 2009). France followed the path Hanford opened in the United States, Maïak in the Soviet Union (Brown 2015), and Windscale (now Sellafield) in the United Kingdom. But the UNGG reactors also found energy uses to secure fuel and energy supply at a time when France was growing more and more dependent on oil (Frost 1991), and had seen its supply capacities challenged by the 1956 Suez crisis and by the independence of Algeria in 1962 (Auzanneau 2015). A partnership with the newly nationalized electricity company EDF was established in 1956. In the following years, EDF built, with the help of CEA engineers, the reactors of Chinon A1 (1961), A2 (1965), and A3 (1966), Saint-Laurent des Eaux A and B (1969 and 1971), and Bugey (1972).

In the rush toward innovation, the priority was to get these reactors up and running as quickly as possible—showcasing postwar France's capacity to master a complex technology imbued with the promise of power, modernity, and national independence. As Bernard Laponche remarked in our interview, “The goal—and this was the mindset inherited from the military programs—was to get these damn machines to work” (November 6, 2023). After a PhD on the neutronic properties of plutonium, Laponche participated in the launch of Chinon A3 before becoming the head of the CEA's UNGG program. ⁶ For the engineers animated by the “pioneering” spirit of the time, graphite remained a black-boxed component—an inert backdrop enabling nuclear fission, plutonium production, and electricity generation. When we asked Laponche, after a detailed explanation of the challenges of using nonenriched uranium, how nuclear-grade graphite was produced, he simply replied: “Well, it's coal. So it's much easier than heavy water.” Yet, as we will see later, the different production processes of graphite at the time had an impact on its radiological properties once it had become a residue.

For engineers, graphite was the tame and easy-to-produce moderator of nuclear fission, the homogeneous milieu that allowed for the impressive technological phenomenon of controlled nuclear fission to take place. This separation between core nuclear activities and lowly technical realities (Le Renard 2021) entails a paradoxical relationship to materiality, in which the very material conditions of technology tend to fade in the background. It is well captured in the novel Combat contre l’invisible (Fighting the Invisible), by Queffélec (1958). This book describes the work at the G1 nuclear reactor in a heroic fashion, laden with sci-fi undertones (Chabot 2015). It is telling that there is only one mention of graphite in this book during the scene describing the reactor start-up:

The invisible bombardment of countless particles spewing from uranium rods, unleashed behind graphite and concrete, had been echoed by the invisible bombardment of thoughts. (Queffelec 1958, 41)

Along with concrete, graphite, the thousands of tons of bulk material that make fission possible, becomes a barrier through which an alchemical marriage of mind and ethereal matter occurs.

Graphite was not only overlooked by engineers but also by those living in and around the nuclear plants. We interviewed a long-standing member of the Comité Local d’Information (Local Information Committee) at the Saint-Laurent des Eaux plant—a regional representative who has followed the facility since the late 1970s and continues to track its decommissioning. When asked whether graphite and its future were a concern back in the late 1970s or early 1980s, he replied:

Absolutely not. About graphite, really, it was resting peacefully in its vessel…In Saint-Laurent, even then, we did not imagine that we would one day need to decommission this installation. It was part of the landscape, it was the asset (manne) that had fallen, not from the sky but from higher up. (Local representative, March 7, 2024)

What mattered with the UNGG reactors was not graphite. But it was not only their plutonium or energy output either: it was their economic value for France and for the region—this “asset” from “higher up.”

But these forms of valuation did not go unchallenged. In 1969, the UNGG reactor stopped being the reference design for French nuclear reactors in favor of the pressurized water reactors licensed by the US company Westinghouse. In the 1980s, the operating company EDF started debating whether to keep the UNGG reactors in operation. The massive dimensions of these reactors made them more difficult to run (safety expert, June 10, 2024); they required the same amount of manpower as the pressurized water reactors for half the energy output. ⁷ To make things worse, the Chernobyl accident in 1986 gave nuclear graphite a bad image—even though the UNGG reactors are a different design than the Soviet ones. Only a minority in the industry, still faithful to the original “French” design, defended them (local representative, March 7, 2024). By the late 1980s, they had lost their original value. Ultimately, they were stopped after only thirty years of service. By comparison, pressurized water reactors have been kept in operation for at least forty years.

The downfall of these once-prominent icons of technological prowess marked the first moment when graphite could be noticed. With the production of energy, plutonium, and wealth halted, attention could finally turn inward—to the residues left behind and the aging infrastructures that contain them. What emerges from this shift, however, is far messier than anticipated.

Irradiated Graphite's Precarious Infrastructure

From the 1990s, graphite residues began to draw the attention of nuclear engineers, experts, and regulators. What emerged was threefold: the evolving properties of irradiated graphite once it became a residue of nuclear production; the fragility of the infrastructure housing it; and the diversity of material contexts in which this once-discrete component had played a role. Irradiated graphite does not sit passively inside the reactor vessel awaiting removal—it continues to degrade along a now-precarious infrastructure of use and storage. It has become emblematic of a broader class of old, bulky, and complex remnants from the early nuclear era—what the industry refers to as “historic waste” or “inherited situations” (safety expert, July 23, 2024). In this section, we follow irradiated graphite's diverse and messy materiality (Ingold 2012) along the lines of the often-unnoticed background infrastructure (Star 1999) that made the functioning of the UNGG reactors possible. We show that the industry's actors understood and tried to organize a change of status of this infrastructure from an infrastructure of production to an infrastructure of precarious and unsatisfactory care for the residues.

There are two main families of irradiated graphite residues that had different material histories: the pilings within the reactors, and the “sleeves” that used to contain nuclear fuel. We describe them one after the other.

Inside the reactor's vessel, graphite comes in the form of pilings of an average of 9,000 cubic meters (EDF 2017). It now contains a range of radioelements, primarily formed through the nuclear activation of impurities present from its original production—its material past directly shaping its properties as a nuclear residue. Among the most problematic are carbon-14 and chlorine-36, with half-lives of 5,730 and 300,000 years, respectively. Yet radiological concerns are only part of the story. Though inherently solid, decades of exposure to high temperatures have rendered the graphite friable. Graphite dust poses risks of fire and even explosion under certain conditions (D’Amico 2016; IAEA 2016). Moreover, it is not only the graphite that poses a challenge—the structure containing it is also fragile and demands careful handling. The graphite piles rest on a corroding steel framework, suspended above ground within a concrete block designed not to be reopened.

In the 1990s, EDF first put into place a “deferred decommissioning” strategy. It can be understood as a practice of partial residual care, in which a once-functioning machine is turned into a residue under surveillance. Two boxes archived at the IRSN ⁸ allow us to follow the transformation of a reactor (Chinon A3) into a new installation or storage site undergoing decommissioning (Chinon A3D) from 1993 to 1997. First, the reactor is shut down: the chain reaction is progressively stopped, the fuel is removed, and most of the radionuclides with it. What is deployed then corresponds to a logic of confinement: remaining radioactive residues outside of the reactor vessel are sorted between low-level and very low-level waste into specific zones on site (DSIN 1996a) before they are sent to a national low-level waste storage center; new confined pathways are designed for the operators (DSIN 1996b); surveillance systems are installed, both visual and radiological (DSIN 1996c).

Most important for our purposes, irradiated graphite is kept inside the vessel, and thus, what was once an operating machine is turned into a confinement device. This requires shutting down the openings that allow gas to circulate in the reactor while installing a ventilation system. This ensures no pressure differential between the inside of the reactor and the outside that could threaten its stability in the long run (DSIN 1994). This process is also regulatory, and nuclear safety agents visit the installation regularly to monitor progress in the work of EDF agents. These visits sometimes provided an opportunity to express concerns about the coordination of operations and to question the viability of delaying the decommissioning of the vessel (DSIN 1996d).

After 2001, EDF adopted the principle of “immediate decommissioning” and declared its intention to convert these reactors into waste. Once the demonstration of mastery over the production of electricity and nuclear materials, UNGG reactors became—at least discursively—instances to demonstrate mastery in the technologies of residual care. Yet, the proposed first strategy—filling up the reactor with water to ensure a radioprotection layer for the workers, a method used to decommission small-scale experimental reactors—did not result, from 2001 to 2016, in any concrete realization. Since 2016, EDF has changed its strategy to a “three-step derisking strategy.” ⁹ Decommissioning would take place from a platform on top of the reactor, with robotic arms cutting and recuperating the graphite underneath. This installation would be developed in three steps: first, a scale 1 model to gain knowledge and train workers; then, a prototype decommissioning of a reactor, Chinon A2; and finally, the decommissioning of the eight other reactors with the tools and knowledge acquired in the first two steps. This strategy acknowledges that residual care requires patience—but can also be seen as a means to keep the actual residual care work waiting.

Indeed, this plan would eventually lead to having the last UNGG transformed into waste in 2100, more than a century after the end of the reactors’ productive life. While the nuclear safety authority criticizes this protracted timeline (ASN 2016), EDF and other decommissioning actors argue that there is still a lot to learn to ensure safe, careful decommissioning of such a fragile infrastructure. There are many uncertainties concerning the degradation of materials; decommissioning on that scale has never been done before and many unexpected problems will likely appear along the way. An engineer from a company created in 2019 to tackle the graphite issue told us: “Chinon being the first reactor of this type to be decommissioned, we are starting from scratch” (decommissioning expert, June 4, 2024). And as preparations progress, new uncertainties emerge. For instance, the protective layer meant to prevent heat from escaping the reactor is made from a material that is no longer produced, and that engineers lack knowledge about how it decays: could it crumble and damage the robots? These uncertainties are important because the situation will be different with each of the nine UNGG reactors, as each of them was built with different designs.

The ongoing decay of the vessel presents experts with a puzzle that captures the ambivalence and tinkering of care practices. On the one hand, there is a need to reinforce the structures to ensure their (relative) stability as a confinement device, and nuclear safety engineers demand that reactors undergoing decommissioning works are subject to the same type of norms as operating reactors (IRSN 2019). On the other hand, reinforcing the structures too much could make decommissioning impossible (safety expert, June 10, 2024). Residual care relies on a delicate pondering of maintenance (Denis and Pontille 2024): degrading buildings require maintenance to avoid accidents, but making decommissioning possible requires limiting maintenance. A possible, yet highly controversial strategy is looming on the horizon: what if the industry chooses to never decommission, to stabilize and leave the building as a permanent storage site for nuclear waste? This practice, referred to as “entombment” (IAEA 2018), is already advocated by some, especially in the United States, but remains “unthinkable” and “unacceptable” in France because it prides itself on its good residual care practices (safety expert, July 23, 2024).

But engineers have warned that the issues raised by irradiated graphite go beyond the reactor vessel. This brings us to the second family of residues. Part of the fuel that was used in the UNGG reactors also contained an additional layer of graphite, called the “sleeves.” To operate, UNGG reactors were thus embedded in a larger infrastructure that handled fuel and the residues that were subsequently generated. This infrastructure mainly consists of silos, buried underground, in which the residues of graphite sleeves have been stored in ways that are now regarded as especially precarious. These instances of what the industry calls “historic storage sites” represent one of the main concerns surrounding the transformation of infrastructure into waste.

Once cut off from the irradiated fuel, graphite sleeves were simply dumped from an opening at the top of the silo—as a nuclear safety expert puts it, “well, it's the old ways” (December 8, 2023). Two of these silos were located on the same site as the reactors, in Bugey and in Saint-Laurent des Eaux. They have had a troubled relationship with the surrounding waters. The Bugey silo was evacuated immediately after a storm in 1999, due to a high risk of inundation (local representative, March 7, 2024), and taken to the Aube storage center for low-activity waste, saturating the Center's capacity for long-lived radioactive elements (safety experts, March 12, 2024). The Saint-Laurent des Eaux silos, on the other hand, underwent a slower degradation: in the 2000s, an infiltration problem was noticed, raising concerns about the waterproofing of the bottom of the silos, where the graphite sleeves had been dumped back in the 1960s. To counter the risks of interaction with the groundwater table nearby, a project to build another confinement barrier was proposed (safety expert, December 8, 2023), but was never completed. By now, EDF's plan is to recondition this graphite and send it to a final disposal site by 2060 (EDF 2019).

Other silos were situated in the reprocessing installations of Marcoule and La Hague, where fuel was chemically treated to extract the materials that could prove useful for energy production (Denoun 2022; Pottin 2024). There, the CEA studied irradiated graphite's properties and made some projects to incinerate it as early as 1968 (CEA 1968). These projects were meant to limit the space graphite takes once stored, but were later rejected because incineration tends to generate new kinds of residues (CEA 1987). For the most part, graphite sleeves were also dumped in half-buried silos there. The most infamous is silo 130 in La Hague—a “textbook case” of the issues with historic storage sites (safety expert, July 23, 2024). There, irradiated graphite was dumped among other types of waste. In 1981, a fire occurred in this silo containing inflammable graphite. To stop it, the residues were soaked in water. By now, silo 130 is the most important reconditioning work taking place in France (safety expert, June 10, 2024), necessitating the construction of new infrastructure (IRSN 2023). It is disrupting the operations of La Hague reprocessing plant, the largest installation of its kind in the world, which must handle a large number of residues (citizen experts, March 24, 2024).

Decommissioning has infrastructural ties to other types of “historic waste” that need to be recuperated and reconditioned. The very need to recuperate and recondition residues stored in the “old ways” points to the fact that standards of residual care evolve, and that the transformation of residues into waste is situated among a set of norms.

Making Irradiated Graphite Storable

Waste is not a given entity but the by-product of evolving standards of care. Although public attention usually focuses on high-level waste and deep geological repositories, low-level waste, which represents 90 percent of the mass of nuclear waste, ¹⁰ exemplifies the evolution of residual care in different ways. In the first decades of the nuclear industry, low-level nuclear waste was dumped and dispersed in the seas, a practice that was eventually forbidden by the 1972 London Convention (Hamblin 2009). ¹¹ The management standard for low-level nuclear waste subsequently shifted from dispersion–dilution to concentration–confinement into long-term disposal sites (Garcier 2014). In France, the Centre de Stockage de l’Aube (CSA) and the Centre industriel de regroupement, entreposage et stockage (Cires) started welcoming low-level and very-low-level waste in 1992 and 2003, respectively. These sites are supposed to be able to welcome packages of waste, confine them, and monitor them for 300 years. Their construction started in a context marked by heavy contestation of nuclear waste disposal projects siting in 1988-9 (Barthe 2005). The nuclear waste disposal sites are meant to demonstrate the industry's ability to responsibly take care of its residues (Martinais 2021) by transforming them into harmless and controlled waste. As a policymaker writes in a 1992 report, “the example of a waste treatment Centre like the one in Aube can be used to play down certain conflicts” (Le Déault 1992, 77).

For nuclear residues to be disposed of in an “exemplary” manner, they must first undergo a series of material and regulatory transformations. Disposal options determine how residues are to be assembled and packaged—what nuclear waste management terms “specifications.” This process involves negotiations in which the interests of different actors, sociotechnical coordination, and the material properties of residues are at stake (Garcier 2021). Nuclear waste is a fuzzy object: its properties, management, and classification are mutually constituted in negotiations (Parotte 2021). Transforming residues into waste thus opens up three interrelated uncertainties: How radioactive is it? How should it be packaged? Where can it be placed? In the language of nuclear waste management, these correspond to the challenges of radioactive inventory, conditioning, and disposal.

The case of irradiated graphite is particularly complex: not only is its extraction from current emplacement technically challenging, but it also lacks a designated disposal path and agreed-upon conditioning specifications. This stems from a key material property—it contains long-lived radionuclides, defined in the French system as elements remaining radioactive for over thirty-one years. As such, it falls into the “faible activité, vie longue” (FA-VL) category, which is an especially blurry category in France's nuclear waste classification. As a technical report notes, “a purely a priori and ontological definition of this waste was not possible” (Orano 2023). FA-VL waste is unified more by its problematic status than by shared physical characteristics: it is not radioactive enough for deep geological disposal, yet its longevity may disqualify it from surface repositories. Since 2006, Andra has explored several disposal options, initiated site evaluations, and held local public debates—yet no solution has materialized (nuclear waste experts, April 12 and 26, 2024).

The discussions concerning irradiated graphite's status as a future nuclear waste mostly revolve around one radioelement: chlorine-36 (³⁶Cl) (safety experts, March 12, 2024). We delve into this aspect, for it illustrates the multiple layers of planned residual care, and with it the numerous uncertainties and negotiations that surround the planned ultimate transformation of a residue into what current standards consider as “proper” waste. ³⁶Cl appears as a matter of care only in the context of current standards of waste disposal. ³⁶Cl did not raise concern in the studies of the radiological properties of irradiated graphite we found in the archive (CEA 1976). Until the 2000s, the focus was rather on the presence of carbon-14, an element that is considered the most hazardous for the health of future decommissioning workers. But ³⁶Cl is a long-lived and “migratory” element that can move through physical barriers. The problem, then, is that if its presence exceeds a certain threshold, it must be stored underground, where a protective clay layer can shield living organisms from its harmful effects.

Yet estimating the quantity of ³⁶Cl in the existing stocks of irradiated graphite is highly difficult due to the heterogeneity of irradiated graphite (safety experts, March 12, 2024). The presence of ³⁶Cl was caused by the nuclear activation of one of the impurities of graphite, chlorine-35. The activation of these impurities depends on the history of the material: In which part of the reactor was it? How was it exposed to neutron fluxes? How different is it between the pilings and the sleeves? The estimation of the quantity of ³⁶Cl raises empirical difficulties: in some cases, accessing samples is impossible, and estimates of average quantities rely on complex mathematical models whose reliability is difficult to assess. These models extrapolate from limited data, often without knowing whether the samples are representative or whether the models adequately capture the problems at stake (PNGMDR 2019). Experts we interviewed acknowledged the complexity of this challenge and described the difficulties of coordinating teams around such highly technical topics.

And that is not all: how can one define a threshold below which the presence of long-lived, migrating radionuclides is still acceptable for surface disposal? The CSA storage facility already applies such a threshold—introduced precisely because nuclear waste is rarely pure, and even short-lived waste contains traces of long-lived elements. However, this limit has already been reached, partly due to irradiated graphite stored after the 1999 storm. Meanwhile, engineers continue to devise increasingly complex disposal concepts to accommodate the heterogeneity of future FA-VL waste. Yet when we did the interviews, the proposed semiburied site in the same region had seen no progress since 2019 (nuclear waste expert, April 26, 2024). As public and political focus remains fixed on deep geological disposal for high-level waste, the final destination of irradiated graphite continues to receive little attention.

This neglect has consequences on the decommissioning process: in the absence of a disposal option, the specification for the future packages of irradiated graphite is not yet clear, and thus EDF has long been reluctant to extract it (expert, December 8, 2023). The transformation of the UNGG reactors into waste is still a projected horizon laden with major uncertainties. This leads us back to the question: how far is residual care performed? To answer that question, one needs to reflect on what residual care means.

Concluding Reflections: Toward Residual Care

In this article, we have pursued an analytical shift from interrogating waste to examining the processes by which residues are (planned to be) transformed into disposable waste. In this conclusion, we not only summarize some aspects of this focal shift but also highlight what it makes salient: once an innovation has become a matter of residual care, its nature changes, as well as the registers of values through which we approach it.

Nuclear waste is certainly a dazzling object: where and how can we safely dispose of a thing that remains highly dangerous for very long or even unimaginable time spans? Indeed, deep geological repositories for high-level waste usually intend to demonstrate safe disposal for millennia, and surface repositories for low-level waste for centuries. The Aube Center in France, for instance, is supposed to monitor its waste for 300 years. Yet, by focusing on the disposed object, one could miss the process that shapes it and gives it its characteristics. Nuclear waste is not waiting for someone to pick it up and store it somewhere. Once the infrastructure of a formerly functioning nuclear power plant has become a residue, it still needs to be analyzed, modified, legally redefined, monitored, deconstructed, partially maintained, repackaged, and ultimately relocated. These processes entail tinkering efforts (Mol, Moser, and Pols 2010) that are laden with uncertainties concerning the risky collateral futures (Felt 2025) they might bring about; it involves knowledge-making that is fragilized by the memory loss and the difficulty of accessing empirical data about hazardous materials sealed in closed-off locations; finally, it is the occasion for regulatory negotiations concerning the “specifications” of future waste, the status of the installation, the best decommissioning practices, the timelines, and so on.

With the shift from waste to the process of carefully transforming residue into waste, other aspects become salient. Public discussion and part of the research in social sciences on nuclear waste has, for good reasons, focused on the question of the emplacement of disposal sites (Barthe 2005), which raises issues of democracy (Parotte 2018) and technical knowledge (Macfarlane and Ewing 2006) in the very long term of 300 years for low-level waste, and millennia for deep geological repositories of high-level waste. Turning to residual care, we also see the many displacements that are necessary beforehand to have objects that can be disposed of somewhere: irradiated graphite would have to be taken out of the reactors and silos and packaged before it would be sent (in a distant and uncertain future) to a storage site. Similarly, public debate has tended to focus on the very long time spans, often counted in millennia, associated with the safe disposal of nuclear waste. Studying residual care processes makes other time spans appear: what will have to be done between now and 2100? What has happened in terms of material degradation since the early 1990s? How long can these reactors hold?

But this shift opens another question: what does it mean for an infrastructure to have become a residue? The story of the UNGG reactors and of the networks of irradiated graphite illustrates the many changes an infrastructure undergoes once it has become a matter of residual care: the matters of care change the matters of fact (Puig de la Bellacasa 2011). What once were sites for plutonium production, electricity generation, wealth creation, energy independence, and national radiance have become concealed piles of crumbling materials, precarious ancient storage sites, and poorly known assemblages of radioactive elements. Graphite, the homogeneous and easy-to-produce material that played a seemingly negligible supporting role in a heroic performance in which pioneering engineers master the powers of nuclear fission, has become irradiated graphite: a fragile, hazardous, blurry, cumbersome, lowly yet durably radioactive remnant of a not-so-distant past that the industry has partly lost the memory of. The efficient infrastructure of the radiance of France (Hecht 2009), now obsolete, testifies to the “old ways,” a set of practices that are seen according to current standards as a bit reckless and expeditive—like constructing reactors without any decommissioning plans in mind, or dumping graphite sleeves into buried silos next to a river.

Residual care reshapes how one perceives both the materialities and temporalities of the infrastructure, demanding sensitivity to the fragility of artifacts once seen as embodiments of technological mastery and control (Jackson 2014). It calls for studying heterogeneity of the material processes that went unnoticed behind the dominant gaze over a supposedly homogeneous matter (Ingold 2012), and to the diversity of situations that residual care will need to face—the UNGG reactors and their different designs, the silos and their histories of waterproofing issues, inundation hazards, and fires. Finally, the temporal properties of the infrastructure changes: the reactors are no longer these “eternal” buildings with no afterlife, but an assemblage of residues that degrade at different paces.

In short, a close examination of residual care obliges us to rethink the very meaning of “life of an installation.” This vitalistic and biographical metaphor is very present in expert language. Decommissioning and waste management are regulated in the name of the mastery of the whole “life cycle” of an installation. Beyond the nuclear case, one can think of the routinized practice of “life cycle assessment.” Residual care for irradiated graphite shows that the life of a reactor as a residual infrastructure can be far longer than the life as a production infrastructure. If the industry does manage to end decommissioning around 2100, these reactors will have “lived” more than a century as a residue, against thirty years as electricity- and plutonium-producing devices. Residual care also challenges the simplifying assumptions that the “life cycle” metaphor often carries. Far from being a closed cycle or a linear trajectory, the residual life of an innovation generates new, unexpected phenomena (including those that could have been expected), open-ended negotiations, uncertainties, and continued neglect.

Residues do not simply emerge on their own; rather, they become visible through a shift in the value registers that frame our relationship to the infrastructure—specifically, when it transitions from a site of production to one of precarious containment and residual management. Therefore, like any practice of care, residual care requires a pondering of values (Mol, Moser, and Pols 2010) and thus opens a space for ambiguity. We saw that, for the irradiated graphite residues to be noticed, the UNGG infrastructure first had to lose its original use and economic value ¹² —plutonium and electricity production, embodiment of a modernized national identity, and wealth. But we do not think that what is then left is simply the opposite of value, as some definitions of waste have (Gille 2014). On the contrary, residual care is laden with new and sometimes conflicting values.

To put it bluntly: why should industry bother to invest time, money, workforce, space, and energy in decommissioning these old and obsolete infrastructures? When we asked the head of the decommissioning department at the nuclear safety authority (ASN) whether “entombment”—reinforcing the reactor to turn it into a storage facility—was an option, he told us: “In France, it is simply unacceptable” (safety expert, July 23, 2024). Meaning, in this country at least, long-term environmental protection requires transforming the reactors into proper waste to be disposed of in a secure centralized location.

But protection and the values of environmental ethics are not the only things at stake. When EDF embraced the principle of “immediate decommissioning,” the goal was to demonstrate its ability to master the complete so-called “life cycle” of nuclear installations. When Andra created the storage centers for low-level and very low-level waste in Aube in the 1980s-1990s, the goal was to “play down certain conflicts.” Residual care is meant as a performance of responsibility and quiet mastery of a technology associated with images of destruction and contamination (Jasanoff and Kim 2009). And such a demonstration is of value when wanting to fuel innovation: if the solution is, as the international summit's organizer we quoted in the introduction put it, “around the corner,” then there is no concern about unleashing a new series of machines—and the residues they will not fail to generate.

Moreover, residual care is not incompatible with continued neglect. As we have seen, the projected transformation of UNGG reactors into waste is arguably intertwined with a tactic of residual governance that mostly consists in waiting: waiting for the moment technologies will be ready, waiting for the moment a disposal site will be on the way, waiting for the moment the industry will accept spending significant amounts of money in transforming old and obsolete reactors into low-level waste. This way of exercising power can be understood, in Pierre Bourdieu's (1997, 270) terms, as “adjourning, deferring, delaying, raising false hopes”—and which is successful if waiting means “delaying without destroying hope.” Which prompts the question: what will the world in this projected residual care look like if one waits for too long? Will the institutions that are responsible—EDF, the CEA, nuclear safety, the Andra—still be there in 2100? What will it entail to handle thousands of tons of irradiated graphite in a world where climate conditions will probably be very different? (Laponche, November 6, 2024).

To conclude, we suggest that residual care invites a profoundly different value register—one grounded in humility. UNGG reactors, active for just a few decades, will remain sites of residual becoming well into the next century, far outliving their productive phase. This temporal asymmetry confronts us with the enduring responsibilities innovations leave behind. If we are to take the task of caring for these technological afterlives seriously, then we must also critically reflect on the dominant values that underpin innovation choices themselves. Rather than celebrating novelty and acceleration alone, innovation must come to terms with its long-term consequences—requiring a shift of our arts of noticing from a focus on mastery to maintenance, and from short-term technological solutionism to long-term care.