The role of peer review in the pursuit of commercial fusion energy

We present the case here, for the private fusion industry to maintain the practice of publishing scientific findings as peer-reviewed articles in the open literature. While the advantages of this practice are clear in the historical context of fusion as a publicly sponsored research program, questions can arise when considering the strategic and competitive impact on for-profit companies. The authors of this paper, all associated with Commonwealth Fusion Systems (CFS), argue for substantial openness. CFS was founded in 2018 by members of the MIT Plasma Science and Fusion Center, introducing an academic-oriented mindset, which combined with other startup and industry cultures to form the company's values. Those values and the deep technical challenges of commercializing fusion energy lead to a pragmatic approach to problem solving. We need to form and test hypotheses in a rigorous data-driven manner, facilitate an open intellectual environment as the most effective way to find solutions and realize that no amount of money or best story will bend the laws of nature. While we seek to innovate new fusion technology and push the speed at which fusion energy can be deployed on the grid, the peer review process has been a part of the pursuit of fusion energy since its earliest days and remains a necessary means in how we accomplish those goals. This article is organized as a series of questions, which are discussed and answered in each of the subsequent sections.

Defining the degree of openness in sharing technical knowledge is a key question for any private entity in the high-tech sphere, and there is not, by default, one best answer. As an example, consider one of the technical disciplines that CFS has drawn from to grow its team: the successful commercial space and aviation industry. Anecdotally, our staff from those backgrounds have remarked how collaborative and open fusion seems relative to their experience. While such collaboration is encouraged and, in some cases, required for publicly funded fusion research, what is the motivation to maintain this practice as a privately funded fusion company? A straightforward answer is that scientific challenges still exist and need to be overcome. To do this, we need to combine the new approaches and mindsets enabled by private funding with existing best practices. Private funding does not eliminate the research and development needs that the fusion community has identified.^1–3 In part, we are building SPARC⁴ to answer many of those same questions, simply on our schedule, with an emphasis on our priorities. SPARC will not start as a discovery-driven, user-facility as prior tokamaks, though we will need to leverage, and accelerate, a similar science-based evaluation of tokamak plasmas to make swift progress. This is outlined further in a companion Perspective.⁵ While the need for peer review is clear, the impact of peer review on our progress may be different than for publicly funded programs. We welcome efforts to maintain the high quality of the peer review process, since we may make high impact decisions using limited data produced by others, vetted via peer review. We may take more risks, scientifically, and be willing to take the next steps with an assumption that an on-going peer review process will be confirmatory or that a published result may not be as conclusive and universal as suggested.

Internal peer review and self-correction will only get us so far. There is a risk of group-think during design and planning, and an inability to access the diversity of viewpoints needed to overcome road-blocks we encounter. Many decades of experimental and theoretical research in plasma physics and fusion technology have established widely read journals and well-organized and attended conferences where both public and private fusion researchers can share their findings. While we could just draw upon this resource as observers, experience tells us the greatest value will come when engaging with this community as peers. That means paying the entry fee, which is generally your latest results, whether they be triumphs or tragedies. Success can be further strengthened by withstanding public criticism, errors in our methods can be revealed and corrected, and observations from outside of your bubble may lead to a solution to a vexing problem. Privately funded groups may need to take extra steps to ensure regular two-way engagement on scientific topics in order to establish informal methods of peer review. Presently, publicly funded fusion scientists will share results outside of a formal conference setting. When a colleague asks another for guidance or advice, help is often given knowing that in the future, it is likely to be reciprocated. When there is a public/private dividing line between these two colleagues, there can be confusion on what is appropriate and what is allowed in the absence of formal agreements between companies and research institutions.

Having private fusion default to openness and leverage peer review is also aligned with the best interest and long-term viability of CFS as a company. The more open we are, the more trust we build with the community around us, which helps to propel us forward. This trust could be with our own investors, who we want to continue to support us, and their being able to easily cross-check our claims as part of their due diligence. Transparency on scientific results will build confidence with local and national entities who are being asked to regulate fusion energy for the first time. It includes governments that are considering expanding private–public partnerships that want an in-depth, reviewable look into how tax dollars are translating to new science and innovation. This trust also needs to extend to the general public who we want to be excited by fusion and eager to see it deployed to power their communities. While the general public may not be as keen to debate the details of tokamak physics as senior fusion scientists, a fusion company's commitment to peer review should also translate to a willingness to explain to the public about the basic properties and behavior of the plasma confinement facility that shows up in their backyard. Being open allows us to attract the widest range of collaborators from the nominally publicly funded fusion community. This is enabled at CFS by a dedicated, Open Innovation team, with a mission to build the fusion movement external to CFS and connect external technical expertise to those working on tough, unsolved problems internally. The physics basis for the SPARC tokamak⁶ included over 45 authors from 12 institutions, and though these contributions were supported by CFS, they helped to maintain the intellectual diversity of our team. Generally, the shared goals and similar plasma physics basis of a private fusion organization helps to open the door, but execution of collaborations often requires freedom to publish, even for sponsored research. Career progression in universities and national laboratories is driven in part by a person's publication record, and in some cases doing R&D that is restricted from open communication runs against institutional norms or researchers' personal codes of ethics. Hiring within CFS also benefits from the ability to publish in peer-reviewed journals. It attracts staff from the existing fusion community who may want to continue to develop their publication record and grow a new generation of scientists who join companies like CFS early in their careers.

Additionally, while CFS believes we are choosing the surest, fastest way to commercial fusion energy, we do not claim to be the only viable way. Other companies are expected to have success, and we may benefit from their observations. To enable that opportunity, we take the risk that our published results end up benefiting an industry competitor. This is not new to the race for fusion either, as the TFTR and JET teams helped make each other better during the first set of D-T tokamak experiments in the 1990s.⁷ In particular, there are some innovations, such as plasma diagnostics, that are generally agnostic to approaches taken by the industry. If shared, they could further bootstrap open communication by allowing more precise understanding of new plasma physics, reducing the opportunity for misinterpretation of limited observations. Encouraging peer review and open communication also helps to maintain the health of the private fusion community by having a natural feedback response to unintentional misinterpretation of breakthroughs. Producing measurable or even significant neutron flux from D–D or D–T fusion reactions is generally straightforward and repeatable. A single Joule of D–T fusion energy releases over 10¹¹ neutrons, which can be easily detected, even at a distance. However, net energy requires those neutrons be put in context with the energy used to create that fusion pulse. That calculation is not easy, and a company's being open to critique should be seen as a sign of strength and integrity not a sign of weakness or uncertainty. Worse would be intentional bad actors that seek to exploit enthusiasm and the lowering barrier to entry for private investment in fusion. We have a solid, though incomplete, understanding of plasma physics, built by decades of peer-reviewed science. Gaps can be utilized, along with simplification and approximation, to suggest that various plasma confinement approaches may scale favorably when basic research has suggested otherwise.⁸ Sunlight, ironically in the case of fusion, remains the best disinfectant.

We believe that the wide-scale integration of any new energy technologies, including fusion, will be market driven. As such, it is the role of private companies to translate fusion technology into products that can be introduced into markets and spread throughout the world through commercial means. As commercial endeavors, the private fusion companies will operate under different constraints than the publicly funded fusion programs. Unlike publicly funded programs, the private companies must first raise money from investors and then through revenue generation by selling products in the market. Thus, while being open to peer review and sharing results is necessary, private companies also cannot offer universal access to all results if the companies are to remain viable commercial entities.

Much of the value of a private fusion company will rest upon the worth of the company's “core competencies”—defined as the special knowledge, skills, and technological know-how that distinguish the company. Core competencies cannot be bought off the shelf and must be developed over time via investment and information exchange by human capital. Achieving and scaling fusion energy in the private sector will require technical innovation, and some of that innovation related to a company's core competencies will be protected as intellectual property, either through published patents, which are disseminated publicly, or held as proprietary know-how or trade secrets internal to the company. The protection of this intellectual property will preserve the value of the company, allowing it to continue to operate and grow. An example of a core competency for CFS is superconducting fusion magnet assembly. It is important to note that even for technology that constitutes a core competency, our position is to still argue for substantial openness. For example, during the development of the SPARC Toroidal Field Model Coil (TFMC) magnet, we shared information about the testing and performance of the TFMC,⁹ and we have shared some information about the high-level magnet design.¹⁰ However, we have safeguarded the novel aspects of the design, magnet operation, and manufacturing process as patents and/or trade secrets.

As fusion becomes a highly competitive, successful industry, we expect that claims of achievement will need to be transparently shared with the public to increase stakeholder confidence, even if we maintain certain technologies to be proprietary. The “what” that has been achieved, i.e., the discovery, should be reviewable even if aspects of the “how” of the enabling technology, i.e., the invention, are not. At CFS, we do not anticipate keeping the experimental data from SPARC plasmas, e.g., measurements from plasma diagnostics that observe density, temperature, plasma shape, composition, etc., as proprietary information. This is the set of information that other scientists would need to replicate a result and confirm or disprove our claims, either experimentally or through modeling and simulation. These data will likely remain confidential within the SPARC team for a limited period of time to allow for initial scientific investigations to be completed and sufficient scrutiny to be placed on data to ensure they are correct and free of major errors. They will then be presented at conferences and published in peer-reviewed journals by a mix of CFS staff and our international team of scientific collaborators. These are aligned with practices that are normally followed within publicly funded fusion research programs. In fact, fusion can seem relatively closed in comparison to some other research fields where data are generated using large-scale user-facilities. Despite running for decades, devices such as JET or DIII-D still maintain archives that are not publicly available. In contrast, data from space-based telescopes like the new JWST become publicly available after a certain embargo period.¹¹ 

How SPARC experimental data may be released and become accessible for peer review is still under evaluation at CFS, and is also discussed in a companion Perspective.⁵ Initial modeling and simulation results have been made available via GitHub¹² for those that want to model SPARC-like plasmas, but this is a relatively small amount of information. The infrastructure to maintain access to what will be several TBs of raw data, which also require substantial meta-data to contextualize, would take dedicated resources to host and maintain. Enabling the widest, freest accessibility to SPARC data allows for the broadest peer review and most complete elimination of silos within our research program, but the value of increasing accessibility can be non-monotonic. Eventually, a private fusion company will begin to incur non-negligible costs and develop an obligation to correct misinterpretations that are the consequence of unintentional third-party error. Concerning also is how the widest release of data creates an opportunity for intentional bad actors to create false narratives. Reaching thresholds, such as passing Q_fus > 1 (discussed below) or achieving record-breaking peak field and current density in a tokamak, are expected to be robust, but trends and projections of the long-term, commercial viability of the tokamak from limited sets of data can often be made to support a number of hypotheses, given the complexity and newness of burning plasma physics. We hope to stimulate a culture of openness and peer review, but will need to create boundaries to data sharing to continue moving forward with our mission with finite resources.

At CFS, we also do not expect recipes for operational modes or plasma control schemes to be kept as proprietary. We will not patent the discovery of a new enhanced confinement regime, the equivalent of H-mode or I-mode, nor keep the use of any ELM control or disruption prevention avoidance and mitigation routines we develop as a trade secret. The specific code that executes plasma control in real time is unlikely to be publicly released, but the algorithms that underpin them, co-developed by CFS and our collaborators, will be open to peer review along with relevant experimental results from SPARC. Maintaining this openness will allow the freest evaluation of our claims as well as the potential to scale approaches back to existing experiments for confirmation or improvement. SPARC has intentionally been designed to be as close as practically allowable in dimensional and non-dimensional space to ARC in order to provide data to de-risk its design with minimal extrapolation. However, the data from SPARC still need to be placed in context with decades of fusion results, and where novel observations are encountered, replicated where possible on existing tokamaks. While the burning plasma physics SPARC plans to explore will not be replicable, fast-particle physics questions raised by SPARC may be accessible using heating and current drive tools on existing devices. SPARC will field novel disruption mitigation tools such as a 3D coil for deconfining runaway electrons¹³ and deploy massive gas injection with up to six toroidally and poloidal spaced injectors. If these are successful on SPARC, we would be excited to see that success replicated or further explored on existing tokamaks. For the 3D coil, its integration into DIII-D is already being investigated.¹⁴ 

Much of the prior two sections discussed the issue of peer review and openness at a conceptual level, giving some examples from SPARC's recent history. In this section, we give more specific examples, where we expect to share information as part of the peer review process moving forward. This focus is mainly on plasma physics results for SPARC, as policies and practices on the release of information on future magnet technology are still being developed.

A primary way that a specific tokamak's data contribute to a greater understanding is through adding data to existing databases to create multi-device scalings. We drew upon many of these in helping to design SPARC, including those for energy confinement,^15,16 prediction of disruption dynamics,¹⁷ peak,¹⁸ and total¹⁹ ELM energy loss, inter-ELM scrape-off layer heat flux width,²⁰ and thresholds to transition from L-mode to H-mode²¹ and avoid locked modes.²² While it is easy for us to see if predictions match our observations, better still would be to include SPARC data and allow the fusion community to revisit scalings and draw conclusions. This has been done recently for energy confinement, motivated by newer data from metal-walled tokamaks,²³ highlighting the in-depth regression analysis required. One concern for tokamaks is that nearly all of this present work is facilitated by the ITER-centric International Tokamak Physics Activity (ITPA) whose charter²⁴ directs it to be primarily focused on increasing the likelihood of success of ITER. We expect any databases that we contribute SPARC data to, CFS would also have full access to all the data. It may be worth discussing as a community if we need to transition maintenance of tokamak-centric databases elsewhere, such as to the IAEA.

CFS has been open with our modeling and simulation results, with over a dozen CFS-supported journal publications and over 70 SPARC-related presentations at the annual APS-DPP conference alone, summed over 2019–2022. We are generally less forthcoming and have restrictions when interacting with third party teams, when it comes to sharing the engineering design of SPARC, even outside of the magnet technology discussed above. This generally consists of restrictions on sharing detailed CAD models, thermal/structural calculations, vendor and material specifications, and assembly and construction plans for the core tokamak components and other interfacing systems on our site in Devens, MA. Sharing this material outside of CFS, such as when we invite people to be included in design reviews, requires signing non-disclosure agreements (NDA), and restrictions on third-party sharing of these data are included in our collaboration agreements. Following the final design and completion of SPARC assembly some of these details may become more accessible, while others may remain confidential. The detailed engineering design of proprietary components will not be shared. We may need to also restrict sharing of details of interfacing components that one would not think would be restricted because of the ability to potentially infer aspects of proprietary technology through design choices. What we will likely be more open to sharing in more detail are the engineering design of systems that do not translate to ARC and need to be represented at high fidelity in physics modeling tools in order to provide accurate analysis of tokamak physics. These may include details of plasma facing components, properties of our tungsten and tungsten heavy alloy materials, design of our vacuum vessel, ICRF antennas, and plasma diagnostics. Detailed designs of these are already shared with our scientific collaborators in order to verify if in-progress designs meet requirements. Once designs are finalized, we will evaluate sharing of more details in order to facilitate modeling and simulation of SPARC plasmas. Finally, many of the details of those systems listed will be accessible or inferred through media CFS will release, so there is limited value in keeping some data about them restricted. Prior to operations, we expect the building on our Devens campus that will house SPARC to be available for scheduled tours, and we will continue to release selected images on social media during assembly and share videos of early success like first plasma. This will stimulate scientific discussions, and it allows our hard-working team to freely share their accomplishments with friends and family.

The specific details of how CFS will make announcements of SPARC's Q_fus measurements are still being established, but we will not keep it confidential. This follows a precedent that is being established by publicly funded fusion teams. The DTE2 campaign on JET resulted in a press conference³⁰ seven weeks after their 59 MJ record shot in December 2021. The recent achievement of fusion gain at the National Ignition Facility (NIF) in December 2022 was announced³¹ 8 days after the experiment was completed. Both of these top-level headlines were also effectively “known” by the fusion community within hours, so trusting in the discretion of fusion researchers to keep exciting physics achievements on SPARC confidential for very long is likely a fool's errand anyway. However, both those facilities also benefited from multiple years of prior operational time to gain confidence in their measurements prior to the specific high fusion energy experiments, which is something that private companies may not have. For the SPARC team to have high confidence in our results so quickly after experiments so early in the device's lifetime highlights an exciting but stressful challenge. Starting the peer review process of our methodology as soon as possible is one of the key ways that we start to establish that confidence and allow the pace of breakthroughs to be maintained once SPARC begins operations. CFS views the high-field tokamak approach as the surest, fastest path to commercialize fusion energy but still anticipates debate and competition within the fusion community even after we share news of our Q_fus achievement. What we hope to eliminate at that time is the debate that we have indeed achieved the performance metrics we have claimed, because the methods and measurement techniques have been discussed and evaluated ahead of time. We are encouraged that this practice is already starting to be adopted by the privately funded fusion community.³² 

This article outlines the motivation and processes for peer review within private companies by asking and answering the following three questions: “Why maintain a culture of peer review?” “What gets peer-reviewed and what is kept proprietary?” and “How will this translate in practice for SPARC?” Through this discussion, we find that being peer reviewed is an important requirement for ensuring that the advancement of fusion science is continued, but recognize that not all results will be subject to peer review, especially in areas of fusion technology that need to remain proprietary to preserve the value of the private company. Examples are given for results for the SPARC tokamak, specifically outlining how the team intends to measure and communicate fusion gain in SPARC.

This work was supported by Commonwealth Fusion Systems. It has benefited from discussions on values and cultural norms with the entire SPARC team, as well as discussions and debates with the wider fusion community in the past 5–10 years (nominally at pubs around the world) during when the notion of privately funded tokamak research evolved.

The authors have no conflicts to disclose.

M. L. Reinke: Conceptualization (equal); Writing – original draft (equal); Writing – review & editing (equal). B. Sorbom: Conceptualization (equal); Writing – original draft (equal); Writing – review & editing (equal). M. Greenwald: Conceptualization (equal); Writing – review & editing (equal).

Data sharing is not applicable to this article as no new data were created or analyzed in this study.