Advancements in Tokamak Technology for Fusion Energy: A Bibliometric and Patent Trend Analysis (2014–2024)

Abstract

Tokamak technology, as the cornerstone of nuclear fusion energy, holds immense potential in achieving efficient plasma confinement and high energy densities. To comprehensively map the rapidly evolving landscape of this field, this study employs bibliometric analysis to systematically examine the research and development trends of tokamak technology from 2014 to 2024. The data are drawn from 7702 academic publications in the Scopus database, representing a global research effort. Additionally, the study incorporates 2299 tokamak-related patents from Google Patents over the same period, analyzing their technological trends to highlight the growing significance of tokamak devices. Using the R language and the Bibliometric package, the analysis explores research hotspots, institutional influences, and keyword evolution. The results reveal a multifaceted global landscape: China leads in publication output, and the United States maintains a leading role in citation impacts and technological innovation, with other notable contributions from Germany, Japan, South Korea, and various European countries. Patent trend analysis further reveals the rapid expansion of tokamak applications, particularly with significant innovations in high-temperature superconducting magnets and plasma control technologies. Nevertheless, the study identifies major challenges in the commercialization process, including plasma stability control, material durability, and the sustainability of long-term operations. To address these, the study proposes concrete future directions, emphasizing international collaboration and interdisciplinary integration. These efforts are crucial in accelerating tokamak commercialization, thereby providing a strategic roadmap for researchers, policymakers, and industry stakeholders to advance the global deployment of clean energy solutions.

1. Introduction

The global imperative for clean, sustainable energy has positioned nuclear fusion as a transformative solution to the dual challenges of energy security and climate change. Among various approaches, magnetic confinement fusion (MCF), particularly through the tokamak device, stands as the most mature and promising pathway. Proposed in the 1950s, this technology utilizes powerful magnetic fields to confine high-temperature plasma, creating the conditions for nuclear fusion. Unlike traditional energy sources, tokamak fusion operates on abundant fuels like deuterium and tritium, producing zero carbon emissions and no long-lived radioactive waste, making it a cornerstone technology for the future of clean energy.

In the past decade, the field of tokamak research has transitioned from foundational science to a phase of accelerated engineering and commercial-oriented development. This surge is propelled by landmark international projects like the International Thermonuclear Experimental Reactor (ITER) and disruptive innovations, most notably in high-temperature superconducting (HTS) magnets and artificial intelligence (AI) driven plasma control. These advancements have led to the proliferation of next generation designs like SPARC and ARC, with ambitious international goals aiming for commercial fusion reactors by 2035. This rapid progress is mirrored by a soaring volume of academic publications and patent filings. However, while numerous studies focus on specific aspects of tokamak physics or engineering, a comprehensive, quantitative analysis connecting the scientific literature with the parallel surge in technological patents is notably absent. A macroscopic map detailing the field’s intellectual structure, key players, and the critical pathway from scientific discovery to commercial application remains a significant knowledge gap.

To address this gap, this study provides a unique dual-methodology analysis of the tokamak landscape from 2014 to 2024. Leveraging 7702 academic publications from the Scopus database and 2299 related patents from Google Patents, this research employs bibliometric methods to systematically explore research hotspots, institutional and national contributions, and the evolution of key technologies. Our analysis emphasizes tokamak as the dominant fusion configuration, covering critical domains such as plasma confinement, reactor engineering, radiation-resistant materials, and advanced simulation technologies.

By integrating the analysis of both the academic literature and intellectual property, this paper offers a holistic and data-driven view of the field’s recent evolution. It not only identifies the pivotal scientific foundations but also tracks their translation into tangible technological innovations. Ultimately, this study aims to provide a strategic roadmap for researchers, policymakers, and industry stakeholders, illuminating the challenges and opportunities on the path to commercial fusion energy and laying a robust foundation for the global deployment of this transformative clean energy solution. The remainder of this paper is structured as follows: Section 2 outlines the methodology, Section 3 presents the key findings, Section 4 discusses the implications of these findings, and Section 5 concludes the study.

2. Literature Review

Tokamak technology is a core field in nuclear fusion research, offering a potential pathway for sustainable, zero-carbon energy to address global climate challenges. This literature review comprehensively analyzes 40 key publications from 2014 to 2024, contextualized by a broader dataset from Scopus (Elsevier, Amsterdam, The Netherlands) (7702 papers) and Google Patents (Google, Mountain View, CA, USA) (2299 patents), to assess the developmental trajectory of this technology. The analysis is based on citation metrics, keyword trends, and patent data and explores academic output, technological innovation, and international collaboration while identifying key trends and persistent obstacles. The review is organized around three thematic areas foundational physics and engineering, applied technologies and emerging innovations, and the global R&D landscape with future directions proposed to address current limitations.

2.1. Foundational Physics and Engineering (2014–2020)

The period from 2014–2020 was characterized by research that established the scientific and engineering groundwork for next-generation tokamaks. This period also marked the phase wherein tokamak research established its scientific and engineering foundation, with publication numbers growing steadily from 671 in 2014 to an average of approximately 700 annually (Section 4.3). Song et al. (2014) [1] proposed the conceptual design of the China Fusion Engineering Test Reactor (CFETR). The China Fusion Engineering Test Reactor (CFETR) is a next-generation tokamak in China’s fusion roadmap, bridging the gap between ITER and DEMO. It aims to demonstrate steady-state operation, tritium self-sufficiency, and large-scale fusion energy production as a step toward commercial fusion power plants. They emphasized steady-state operation and high-performance plasma confinement, accumulating 385 citations. Pitts et al. (2019) [2] advanced materials science through research on the ITER tungsten divertor, garnering 748 citations and highlighting the importance of thermal load management. A high MeanTCperArt² of 16.21 was noted in 2015 (Section 4.2), highlighting the importance of thermal load management under extreme conditions. Azizov (2012), provided a historical context, tracing the evolution of tokamaks from Sakharov’s early designs.

Leonard AW (2014) [3] studied edge-localized modes (ELMs) in tokamaks, proposing control strategies, and earned 299 citations, while Sun Y (2016) [4] investigated nonlinear ELM suppression using resonant magnetic perturbations in EAST, contributing 247 citations. Meneghini O (2015) [5] and Romanelli M (2014) [6] developed integrated simulation frameworks and multi-physics codes, respectively, achieving 319 and 222 citations, providing essential tools for plasma behavior modeling. For example, OMFIT is a comprehensive simulation and analysis framework designed for nuclear fusion research, widely used in tokamak and related plasma physics studies. It is a modular software platform aimed at integrating various physics simulation tools, data analysis methods, and experimental data processing capabilities to support the design, operation, and data interpretation of tokamak devices. These efforts align with trending topics of the era, such as “fusion reactor divertors” and “experimental advanced superconducting tokamaks”, which appeared 2000–3000 times and reflect the period’s focus on foundational physics. However, the publication number dropped to 577 in 2020 (likely due to the pandemic), temporarily slowing progress (Section 3.3.3).

2.2. Applied Technologies and Emerging Innovations (2020–2024)

In contrast, the period after 2020 marked a decisive shift toward applied technologies and commercial viability, driven by major experimental breakthroughs and the integration of new tools like artificial intelligence. After 2020, tokamak research shifted toward applied technologies, with the publication number surging to 861 in 2024 (an average annual growth rate of 10.8%) and the patent number reaching 1299 (56.5% of the total) from 2020 to 2024. Creely et al. (2020) [7] detailed the SPARC tokamak’s high-field magnet technology, earning 307 citations with a TC per year of 51.17, marking progress in compact reactor design. Hemsworth et al. (2017) [8] introduced the ITER neutral beam injector design, gaining 276 citations, which enhanced the plasma heating efficiency, a trend reflected in the 4000-occurrence frequency of “neutral beam injectors” in 2023–2024. Sun Y (2016)’s [4] ELM suppression research regained attention, with “surface discharges” reaching 5000 occurrences by 2024, focusing on edge plasma dynamics.

The integration of artificial intelligence (AI) emerged as a transformative trend. Kerboua Benlarbi (2024) [9] emphasized AI’s role in plasma control for the WEST tokamak, aligning with the 13.3% annual growth in G21B 1/15 patents (20 in 2024) for plasma stability. Anirudh et al. (2023) [10] highlighted machine learning’s potential in plasma simulation, consistent with the 386-occurrence frequency of “plasma simulation”. Huang et al. (2023) [11] and Xu et al. (2023) [12] explored AI’s broader impact on nuclear reactor optimization. Breakthroughs included EAST’s 1056-s run (Blain, 2022 [13]), which was scientifically significant in demonstrating the feasibility of steady-state operation required for a future commercial power plant, and KSTAR’s 100 million degrees Celsius for 48 s (2024), a critical milestone that meets the temperature threshold for efficient D-T fusion ignition (sources: www.popularmechanics.com accessed on 10 October 2024, www.iter.org accessed on 10 October 2024). However, the 2024 MeanTCperYear (refers to the average number of times that a paper, researcher, or journal is cited by other academic publications in a year; this is an indicator used to measure academic impact, typically employed to assess the influence or importance of research outputs) of 0.90 indicates a citation lag due to incomplete data.

2.3. Global R&D Landscape and Collaboration

The geographical distribution reveals diverse collaboration models. Chinese institutions, led by the Institute of Plasma Physics (1697 papers, 22.0%), contributed 43.8% (3374 papers), with a multi-country collaboration proportion (MCP) of only 29.9%, reflecting a state-led R&D approach consistent with the successes of EAST and CFETR. European institutions, such as the Culham Science Centre (531 papers, 6.9%) and the Max-Planck-Institut für Plasmaphysik, demonstrated high MCPs (58.2–80.4%), driving progress in ITER and JET, with ITER having completed its strongest magnet assembly in May 2025 (source: www.neimagazine.com accessed on 10 October 2024). The United States (507 papers, 6.7%) and South Korea (343 papers, 4.5%) balanced single-country and collaborative efforts, with the NIF and KSTAR milestones reflecting technological flexibility.

Patent trends align with these dynamics, with China’s G21B 1/03 growing at 16.2% annually and Europe’s high MCP fostering technological integration. However, the slow growth in auxiliary technologies (e.g., B25J 9/16, seven patents in 2024) indicates gaps in cross-disciplinary innovation (Section 3.3.8). This landscape is further contextualized by the recent literature assessing fusion’s potential within the global energy transition (Asif et al., 2024 [14]; Lerede et al., 2023 [15]) and emphasizing the critical role of industry–university–research collaboration models (Cui et al., 2022 [16]; Cannavacciano et al., 2023 [17]), which are evident in tokamak R&D.

2.4. Discussion and Future Directions

The transition from foundational to applied research reflects global fusion milestones, with AI and superconducting technologies as key drivers. Stability challenges (e.g., ELM control) and material durability (e.g., tungsten under high heat loads) persist, as highlighted by Pitts et al. (2019) [2]. The citation lag (2024 MeanTCperYear 0.90) and limited CPC coverage (14 codes) suggest data gaps, potentially underestimating non-English contributions (e.g., Russian research) or unindexed sources (Section 3.3.4). Yun et al. (2022) [18] and Liu et al. (2021) [19] advocate for patent semantic analysis, suggesting expanded classifications. Future research should prioritize AI-driven real-time plasma control (e.g., deep learning for ELM suppression), enhanced international collaboration for simulation tools (e.g., upgraded OMFIT), and novel materials (e.g., tungsten carbide coatings) to address thermal management. Modular prototypes combining SPARC and CFETR designs could target 2035 demonstration runs, aiming for 2040 deployment. Integrating Web of Science, economic data, and policy analysis will improve the accuracy, aligning with Ulucak’s (2021) [20] call for sustainable energy innovation and Er Saw and Jiang’s (2020) [21] emphasis on interdisciplinary integration.

3. Methodology

The selection of an appropriate literature review methodology is a foundational step in ensuring the rigor and validity of the research findings. In contemporary research practice, three primary approaches to systematic or quantitative reviews are commonly considered: systematic literature reviews (SLRs), meta-analysis, and bibliometric analysis [22,23,24,25]. Each method possesses distinct applications and research objectives.

Following the framework established by Donthu et al., [22] a systematic literature review (SLR) is typically applied to focused research questions within a specific or niche domain, involving the content synthesis of a relatively small and manually manageable corpus of literature [25]. A meta-analysis, in contrast, aims to statistically integrate empirical findings from a body of homogeneous studies to estimate the aggregate effect size of a particular variable relationship [26].

Distinct from these two approaches, bibliometric analysis (Figure 1) is specifically designed for the exploration and analysis of vast volumes of scientific data, making it the ideal tool to delineate the cumulative scientific knowledge and evolutionary nuances of a broad research field [27]. When the research objective is to identify the intellectual structure, collaboration patterns, and thematic shifts of a field, and the literature corpus is extensive—often comprising hundreds, if not thousands, of publications—bibliometric analysis emerges as the preferred methodology.

The present study analyzes an extensive corpus comprising 7702 academic articles and 2299 patents. Given this sheer volume of data, a manual systematic literature review is clearly infeasible. Moreover, the objective of this research is not to ascertain the effect size of a specific relationship but to delineate the macroscopic developmental landscape of the entire tokamak field. Therefore, in direct alignment with the guiding principles outlined by Donthu et al. (2021) [22], bibliometric analysis is unequivocally selected as the most appropriate methodology for the aims and scope of this research. Its capacity to rigorously process and interpret large-scale, unstructured data provides a solid foundation for comprehensive and objective insights into the current state and future trajectory of tokamak research.

3.1. Research Design and Methodological Choice

To systematically map the intellectual structure and evolutionary trajectory of tokamak research over the past decade, this study adopts a bibliometric analysis methodology. As a quantitative approach to literature reviews, bibliometric analysis is uniquely suited to exploring and analyzing large volumes of scientific data, enabling an objective assessment of a field’s key contributors, conceptual structure, and emerging trends [22,23]. Given the vast and rapidly growing body of literature on tokamak technology, this method is chosen over traditional qualitative reviews for its ability to provide a rigorous, reproducible, and macroscopic overview of the research landscape. To comprehensively assess academic progress and technological innovation, this study integrates academic publications and patent data, forming a dual-track analytical framework.

3.1.1. Step 1: Defining Aims and Scope

The core aim of this study is to provide a comprehensive, data-driven overview of the recent (2014–2024) developmental trajectory of tokamak technology. The scope is defined to encompass both the scientific performance (i.e., quantifying the productivity and impact of key research actors) and the intellectual structure (i.e., identifying core research themes and their evolution). This dual focus allows us to determine not only “who” is leading the research but also “what” topics are driving the field forward.

3.1.2. Step 2: Choosing the Techniques for Bibliometric Analysis

To align with our research aims, we integrated two complementary analytical techniques: performance analysis and science mapping. Performance Analysis: We adopted this method to objectively assess the contributions and scientific impacts of research constituents. As described by Donthu, Reinartz et al. (2020) [28], this analysis is a foundational component of bibliometric reviews, as it provides quantitative benchmarks for the objective assessment of the outputs and influences of authors, institutions, and countries. The metrics used include the total publications (TP), total citations (TC), mean total citations per year (MeanTCperYear), and multi-country collaboration proportion (MCP). Science Mapping: This method was employed to visualize the intellectual connections and social networks within the field. Co-Authorship Analysis: This technique reveals the “social structure” of a research field by mapping the collaborative relationships between researchers or institutions. It helps to identify key collaborative hubs and relatively independent research groups [29]. Co-Word Analysis: This technique identifies the “conceptual structure” of a research field by analyzing the co-occurrence frequency of keywords. It operates on the premise that words that frequently appear together in the literature constitute a coherent research theme [29,30,31,32]. We utilized this method to delineate core research topics and track their evolution over time.

3.1.3. Step 3: Data Collection and Curation

The data collection for this study followed a strict protocol. We selected Scopus as the primary source for the academic literature due to its comprehensive, peer-reviewed coverage, which is essential for performance analysis. Concurrently, we used Google Patents to track technological innovation, given its extensive international patent data. The search query was designed for high precision and recall, incorporating core terms (“tokamak,” “nuclear fusion”) and related technical variants, applied to the title, abstract, and keyword fields. A core component of our methodology was data curation. Raw data from databases often contain errors that can affect the analytical accuracy. Our curation process included (1) de-duplication to remove identical records; (2) standardization to unify variations in author, institution, and country names; and (3) keyword merging to consolidate synonymous terms, ensuring the robustness of the thematic analysis. Following this rigorous curation process, the final dataset comprised 7702 academic publications and 2299 patents.

3.1.4. Step 4: Data Analysis and Visualization

The entire analytical workflow, from statistical computation to network generation, was conducted within the R environment, utilizing the Bibliometric R [33]. This package is a specialized tool designed for comprehensive science mapping analysis, offering a suite of functions to calculate performance metrics and construct complex network data from bibliographic records. All network visualization maps presented in this study (e.g., co-authorship and co-word networks) were generated directly using the built-in plotting capabilities of this R package (R language version 4.4.3), ensuring a seamless and integrated workflow from data processing to visualization. This approach guaranteed the internal consistency and replicability of our analysis.

3.2. Patent Search Strategy

To further clarify the search strategy,

Table 1. Thematic shift in tokamak research: from foundational science to applied innovation.

Category	2014–2020: Basic Research Phase	2020–2024: Applied Research Phase
Primary Goal	Establishing scientific principles and engineering feasibility	Accelerating commercialization and demonstrating viability
Key Research Topics	Plasma stability (ELMs), materials science (tungsten divertors), simulation frameworks (OMFIT)	HTS magnets, AI-driven control, long-pulse operation, compact reactor design (SPARC)
Representative Work	Pitts et al. (2019) [2], Song et al. (2014) [1]	Creely et al. (2020) [7]
Driving Force	Foundational theory, large-scale public projects (ITER groundwork)	Experimental milestones (EAST/KSTAR), private-sector investment, AI integration

maps the selected keywords to specific subfields of tokamak research. This mapping ensures that the dataset aligns precisely with targeted research areas within the tokamak domain.

Patent data were sourced from Google Patents, with the search period limited to 2014 to 2024 to align with the literature dataset. The search method incorporated both CPC codes and keywords

Table 2. CPC codes.

CPC Code	Tokamak	CPC Code
G21B 1/00	Inventions related to thermonuclear fusion reactors	H01Q 3/26	Arrangements for changing or distributing energy to antenna systems using switching elements
G21B 1/03	Tokamak reactor	F16L 55/00	Testing or monitoring of pipes, pipelines, or pipeline systems
G21B 1/05	Magnetic confinement not of the tokamak type	F28D 7/00	Heat-exchange apparatus with stationary tubular conduit assemblies
G21B 1/13	Magnetic confinement not of the tokamak type	H02K 44/02	Dynamo-electric machines with no moving parts; superconducting magnets
G21B 1/15	Stabilizing plasma (for stabilizing the plasma)	F01K 25/08	Thermal-to-electric generators
G21B 1/17	Fuel supply and control systems (tokamak-specific)	G21C 17/00	Measuring, testing, or monitoring of reactors
H05H 1/24	Plasma torches; plasma guns	B25J 9/16	Robotic systems for nuclear fusion operations

and

Table 3. Keywords.

Keywords	Subfield
Tokamak, spherical tokamak, ITER, SPARC	Tokamak configurations
Plasma confinement in tokamak	Plasma stability and confinement
Superconducting magnets in tokamak	Reactor components (magnets)
AI in tokamak, plasma control in tokamak	Emerging technologies (control systems)

to isolate Tokamak-related innovations:

• Query: (“tokamak” OR “spherical tokamak” OR “ITER” OR “SPARC” OR “plasma confinement” OR “superconducting magnets” OR “AI in fusion”) AND (CPC = (G21B1/03 OR G21B1/05 OR G21B1/15 OR H05H1/24 OR H01Q3/26)).
• Time range: From 2014-01-01 to 2024-12-31.
• CPC Codes: Selected to reflect nuclear fusion, plasma confinement, superconducting technologies, and related innovations in tokamak systems.

3.3. Data Analysis

This study analyzes the academic and technological progress of tokamak technology from 2014 to 2024, employing bibliometric analysis and patent trend analysis to uncover its developmental dynamics and key areas within nuclear fusion research. As noted in the Introduction, tokamak technology is regarded as the cornerstone of fusion energy development due to its efficient plasma confinement capabilities and zero-carbon emission profile [14], yet its technical realization still faces significant challenges. This section leverages 7702 documents from the Scopus database and 2299 patents from Google Patents to systematically explore the research trends, technological evolution, and global R&D landscape of tokamak technology, with data current as of 30 December 2024.

Specifically, this section focuses on the following objectives: first, identifying growth patterns and critical time points in tokamak technology through changes in publication and patent volumes; second, analyzing the evolution of keywords and patent technology classifications to reveal shifts in research hotspots, such as high-temperature superconducting magnets and AI-driven control technologies; and, finally, evaluating the contributions of major institutions and countries while examining the impact of international collaboration on technological advancement. These analyses aim to elucidate the past and present trajectories of tokamak technology in fusion research and provide data-driven insights for future research directions.

We utilized the Bibliometric package in R for data processing and visualization, incorporating tools like line graphs to ensure that the results are both intuitive and scientifically rigorous. This detailed data analysis serves as a foundation for the broader goal of this work, deepening the understanding of tokamak technology’s role in advancing nuclear fusion research.

3.3.1. Annual Citation Trends

The annual citation trends of the tokamak-related literature reveal a pronounced temporal effect. The number of publications (N) increased from 671 in 2014 to 861 in 2024, exhibiting a fluctuating upward trend with an average of approximately 700 papers per year, reflecting the sustained expansion of academic activity in this field. Notably, the 861 publications in 2024 mark a historical peak, highlighting a significant surge in research output in recent years. Concurrently, the average total citations per article (MeanTCperArt) shows a declining trend, dropping from a peak of 16.21 in 2015 to 1.79 in 2024. This decline suggests that recent papers have had insufficient time to accumulate citations, a phenomenon closely tied to the time-lag effect following publication. The higher values in 2015 and 2017 (16.21 and 14.80, respectively) likely reflect the enduring and robust influence of seminal studies from these years, such as those on high-temperature superconducting magnets or plasma stabilization techniques, within the academic community.

The average citations per article “year (MeanTCperYear), an indicator of immediate academic impact, is significantly influenced by the number of citable years (CitableYears). As the citable years decrease from 12 in 2014 to just two in 2024, the MeanTCperYear for earlier years (e.g., 1.21, 1.47, and 1.13 for 2014–2016) remains relatively low, while more recent years (e.g., 2021–2022) see a rise to 1.78, indicating an enhanced immediate impact of newer research outputs. The peak of 1.78 in both 2021 and 2022 is followed by a decline to 1.27 in 2023 and 0.90 in 2024. This drop may be attributed to the incomplete data for 2024, as the dataset only captures citations up to 5 June 2025, covering just a portion of the potential citation accumulation period and thus not fully reflecting the year’s impact.

From a temporal evolution perspective, the high MeanTCperArt values concentrated in 2015–2017 underscore the long-term academic influence of papers from this period, likely driven by breakthroughs in key areas such as high-temperature superconducting magnets or plasma control. The upward trend in the MeanTCperYear (from 1.47 in 2015 to 1.78 in 2021–2022) highlights the rapid academic impact of emerging themes in recent tokamak research, such as AI-driven control technologies—a trend that aligns with the notable rise in publication volume post-2020. In 2020, the number of publications dropped to 577, a decline of approximately 18.6%, possibly due to the global pandemic’s impact, yet the MeanTCperYear remained robust at 1.60, suggesting high-quality outputs during this year, potentially focused on high-impact technological innovations.

The MeanTCperArt for 2023 and 2024 dropped significantly to 3.82 and 1.79, respectively, consistent with the decline in MeanTCperYear (1.27 and 0.90), reflecting a phase where citation accumulation for newer papers has yet to fully materialize. The notably low MeanTCperYear of 0.90 in 2024 should be interpreted cautiously, as its citable period is only two years, and the citation data may not yet be stable. Future studies should validate this trend by updating the dataset to include more comprehensive citation records.

Overall, the annual citation trends of the tokamak-related literature illustrate the interplay between time and volume effects. Early publications (2015–2017) with a high MeanTCperArt (peaking at 16.21) laid the foundational groundwork for the field, demonstrating the lasting influence of seminal research, while recent papers (2021–2022) with a high MeanTCperYear (1.78) highlight the immediate academic attention garnered by emerging technologies like AI applications. The sustained growth in publication volume, particularly reaching 861 papers in 2024, signals that tokamak research has entered a highly active phase. However, the decline in citation metrics suggests that the academic impact of newer research outputs requires further time to fully manifest.

3.3.2. Annual Scientific Production

As shown in Figure 2, the annual scientific output of the tokamak-related literature from 2014 to 2024 exhibits an overall upward trend, albeit with notable interannual fluctuations. In 2014, the number of publications stood at 671, rising to 715 in 2015—a growth rate of approximately 6.6%—indicating steady expansion in the field during its early research phase. From 2016 to 2019, publication numbers fluctuated between 693 and 709, averaging around 700 papers per year, suggesting that academic activity had entered a relatively stable development phase. This stability may be attributed to the continued deepening of foundational research in tokamak technology, such as studies on plasma stability and simulation techniques. However, in 2020, the number of publications dropped significantly to 577, a decline of about 18.6%, likely due to the global pandemic’s impact, which limited academic conferences, experimental activities, and research collaborations.

Post-2021, publication numbers rebounded rapidly, reaching 678 in 2021, dipping slightly to 622 in 2022, and peaking at 861 in 2024, reflecting a swift recovery and accelerated growth in tokamak research following the pandemic. From 2021 to 2024, the average annual growth rate of publications was approximately 10.8%, underscoring the heightened activity in this field. This growth trend aligns closely with progress in the International Thermonuclear Experimental Reactor (ITER) project and the increasing application of AI technologies in plasma control [9]. It also corresponds with milestones achieved by China’s EAST device, which set records for high-temperature plasma operation in 2021 and 2022 (e.g., sustaining plasma for 1056 s, as reported by www.popularmechanics.com accessed on 10 November 2024).

Referencing the citation trends in

Table 4. Annual total citations (2014~2024).

Rank	Year	MeanTCperArt	N	MeanTCperYear	CitableYears
1	2014	14.48	671	1.21	12
2	2015	16.21	715	1.47	11
3	2016	11.29	693	1.13	10
4	2017	14.80	717	1.64	9
5	2018	12.73	675	1.59	8
5	2019	11.56	709	1.65	7
7	2020	9.58	577	1.60	6
8	2021	8.88	678	1.78	5
9	2022	7.14	622	1.78	4
10	2023	3.82	698	1.27	3
11	2024	1.79	861	0.90	2

N: Number of articles or publications.

, a clear correlation emerges between the annual scientific output and citation metrics. In 2020, despite a decline in the publication number to 577, the average citations per article per year (Mean TcperYear) remained steady at 1.60, suggesting that the papers published that year were of high quality, likely focusing on impactful technological breakthroughs such as applications of high-temperature superconducting magnets. In contrast, 2024 saw a peak in the publication volume at 861, yet the Mean TcperYear dropped to 0.90, attributable to the shorter citable period of just two years. This indicates that the academic influence of these newer papers has yet to fully materialize. This temporal disparity highlights the dual nature of tokamak research: early studies laid the foundational groundwork, while recent efforts are driving technological frontiers forward.

Figure 2 vividly illustrates the trend in publication numbers from 2014 to 2024. The line graph shows a steady rise from 2014 to 2019, a notable decline in 2020, and a strong recovery starting in 2021, aligning closely with the data. The steep upward curve post-2021, rising from 678 to 861 papers, reflects the rapid resurgence of research activity following the pandemic. This rebound may be linked to advancements in international collaborations—such as the completion of ITER’s strongest magnet assembly in May 2025 (source: www.neimagazine.com accessed on 10 November 2024) and the growing adoption of AI technologies. The low point in 2020, with 577 papers, is likely attributable to laboratory closures and disrupted international exchanges during the pandemic, a phenomenon also observed in academic research across other fields.

Trends in Academic Activity and Technological Progress

From an overarching perspective, the annual growth in the scientific output of the tokamak-related literature reflects the sustained vibrancy of academic activity in this field, with a particularly rapid increase post-2021 that underscores the strong linkage between technological progress and scholarly research. The rise in publication numbers may be driven by multiple factors, including China’s ongoing investments in the EAST and CFETR projects, South Korea’s KSTAR achieving a breakthrough of 100 million degrees Celsius for 48 s in April 2024 (source: www.iter.org accessed on 10 November 2024), and Japan’s JT-60SA successfully initiating its first plasma operation in October 2023 (source: www.iter.org accessed on 10 November 2024). These technological advancements have spurred academic research output, fueling the growth in the literature.

The analysis of the annual scientific production in the tokamak-related literature reveals the temporal dynamics of academic activity in this domain. Publication numbers rose from 671 in 2014 to 861 in 2024, displaying an overall fluctuating upward trend, with a notable surge post-2021 (average annual growth rate of 10.8%) that highlights a significant recovery in research activity following the pandemic. The dip in 2020 to 577 papers, likely influenced by the pandemic, was offset by a relatively high mean citations per year (1.60), indicating that the quality of research that year remained robust. The peak of 861 papers in 2024 reflects a highly active phase for tokamak technology, aligning with global fusion breakthroughs such as those from EAST, KSTAR, and JT-60SA. Future research should continue to monitor the citation accumulation of the newer literature to provide a more comprehensive assessment of tokamak technology’s academic and technological contributions to nuclear fusion research.

3.3.3. Top 10 Most Cited (Total Citations) (2014~2024)

The top 10 most cited papers, totaling 3647 citations (~47.3%), were analyzed through their abstracts to uncover core contributions and interconnections in tokamak fusion research. Data included titles, authors, years, DOIs, total citations (TC), citations per year (TC per year), and normalized citations (normalized TC), aiming to explore the technical focus and complementarity of these high-impact studies. Pitts et al. (2019), in “Physics basis for the first ITER tungsten divertor” [2] (TC 748, TC per year 81.43, normalized TC 49.32), investigates the physics of ITER’s tungsten divertor, focusing on its performance under high heat loads and plasma–wall interactions, providing theoretical support for divertor design. Ueda et al. (2014), in “Research status and issues of tungsten plasma facing materials for ITER and beyond” [34] (TC 290, TC per year 24.17, normalized TC 20.03), examines tungsten’s challenges under high heat flux and particle bombardment, complementing Pitts by forming a comprehensive framework for tungsten as a plasma-facing material, crucial for ITER’s material selection. Leonard AW (2014), in “Edge-localized-modes in tokamaks” [3] (TC 299, TC per year 24.92, normalized TC 20.65), studies edge-localized modes (ELMs) and their impacts on plasma confinement and wall heat loads, proposing control strategies to protect divertors, directly linked to tungsten research as ELM heat loads test tungsten’s durability.

In plasma control, Sun et al. (2016), in “Nonlinear Transition from Mitigation to Suppression of the Edge Localized Mode with Resonant Magnetic Perturbations in the EAST Tokamak” [4] (TC 247, TC per year 24.70, normalized TC 21.87), demonstrates nonlinear ELM suppression using resonant magnetic perturbations (RMP) in EAST, enhancing Leonard’s strategies and advancing edge plasma stability research. Lehnen et al. (2015), in “Disruptions in ITER and strategies for their control and mitigation” [35] (TC 354, TC per year 32.18, normalized TC 21.84), addresses ITER plasma disruptions, proposing rapid shutdown and mitigation systems (e.g., massive gas injection), forming a comprehensive stability solution with ELM control for safe tokamak operation.

Simulation is foundational. Meneghini et al. (2015) [5], in “Integrated modeling applications for tokamak experiments with OMFIT” [5] (TC 319, TC per year 29.00, normalized TC 19.68), introduces the OMFIT framework, integrating physics models and data analysis to optimize experimental design and predict plasma behavior. Romanelli M (2014), in “A System of Codes for Integrated Simulation of Tokamak Scenarios” [6] (TC 222, TC per year 18.50, normalized TC 15.33), develops a multi-physics simulation code system for plasma behavior and heat load modeling. These tools support plasma control and wall interaction studies, simulating phenomena like Lehnen’s disruptions, Sun’s RMP effects, and Pitts’ divertor heat loads, bridging theory and experiments.

Reactor design and engineering are key focuses. Song YT (2014), in “Concept Design of CFETR Tokamak Machine” [1] (TC 385, TC per year 32.08, normalized TC 26.59), outlines the China Fusion Engineering Test Reactor (CFETR) design, targeting steady-state operation and high-performance plasma, laying the groundwork for future fusion energy. Creely AJ (2020), in “Overview of the SPARC tokamak” [7] (TC 307, TC per year 51.17, normalized TC 32.06), emphasizes high-field magnet technology and compact design to achieve high energy gains, advancing commercial fusion prospects. The article titled “Overview of the design of the ITER heating neutral beam injectors” [8] (TC 276, TC per year 30.67, normalized TC 18.65) details ITER’s neutral beam injector design for efficient plasma heating and current drive, supporting Song and Creely’s reactor needs. These engineering studies connect with Pitts’ divertor design, as reactor operation requires effective heat load management.

Citation metrics show Pitts’ highest TC per year (81.43) and normalized TC (49.32), reflecting its broad impact in divertor design. Creely’s TC per year (51.17) and normalized TC (32.06) indicate strong recent interest in SPARC’s high-field technology, alongside Song’s CFETR design (32.08 TC per year), driving reactor design discussions. Lehnen [35] and Sun’s TC per year (32.18 and 24.70) values highlight the sustained focus on plasma control.

These papers form a research network: Leonard, Sun, and Lehnen [35] establish a plasma stability and control framework; Pitts and Ueda complement each other on tungsten and wall interactions; Meneghini and Romanelli’s simulation tools provide theoretical support; and Song, Creely, and Hemsworth advance reactor engineering. Their high citations and thematic connections reflect tokamak technology’s integration of fundamental physics and engineering, suggesting that future research should combine control strategies and simulation tools to accelerate fusion commercialization.

An analysis of the affiliations behind these top-cited papers reveals a clear pattern of international and institutional contribution, reinforcing the global R&D landscape discussed in Section 2.3. The leading paper by Pitts RA (2019) [2], which is fundamental to the ITER project, is a quintessential example of large-scale European collaboration under the EUROfusion consortium, involving researchers from institutions like the Culham Centre for Fusion Energy (UK) and the Max Planck Institute for Plasma Physics (Germany). Similarly, the work on disruptions by Lehnen M. (2015) [35] also stems from this highly collaborative European framework.

In contrast, foundational work from Asia, such as that of Song Y.T. (2014) [1] on the CFETR design and Sun Y. (2016) [4] on EAST, primarily originates from leading Chinese institutions like the Institute of Plasma Physics, Chinese Academy of Sciences (ASIPP), showcasing China’s strength in driving large, state-led domestic projects. Meanwhile, the high-impact paper on the SPARC tokamak by Creely A.J. (2020) [7] highlights the influential role of US institutions, particularly the MIT Plasma Science and Fusion Center, in pioneering commercially oriented, high-field magnet technology. The simulation framework paper by Meneghini O. (2015) [5] on OMFIT further underscores the US’s contribution to critical modeling tools, developed at General Atomics.

This geographical and institutional context for the top-cited literature clarifies the diverse paths to innovation within the fusion community: Europe excels in large-scale international collaboration, China demonstrates dominance in state-led foundational research, and the United States leads in developing disruptive technologies and advanced simulation tools.

Reasons for Absence of Highly Cited Papers in 2020–2024

The top 10 cited papers (3647 citations—47.3%) lack entries from 2020–2024, with publication years concentrated in 2014–2020, only Creely (2020) [7] from 2020, and none from 2021–2024. This is analyzed in terms of academic impact lags, research focus shifts, Scopus database characteristics, and external factors.

Academic Impact Lags: High citations require time, especially in foundational tokamak research, where impact often emerges years later. Pitts (2019, TC 748, TC per year 81.43) amassed significant citations by 2025, but 2020–2024 papers, with short citable periods (e.g., 2 years for 2024), struggle to match this. Section 4.1 shows that the 2015–2017 MeanTCperArt (peak 16.21) far exceeds that of 2023–2024 (3.82, 1.79), reflecting a citation lag. Scopus counts only its indexed journals, missing non-indexed citations, limiting 2020–2024 papers’ growth.

Research Focus Shifts: Tokamak research themes evolve, concentrating high citations in specific phases. The top 10 papers focus on core physics (e.g., plasma stability, tungsten) and engineering (e.g., CFETR, SPARC), being foundational in 2014–2019. Song (2014, TC 385) and Pitts (2019) mark key stages. In 2020–2024, the focus may have shifted to applications (e.g., AI control) or emerging areas (e.g., spherical tokamaks), lacking citation accumulation. In Section 3.3.6, the keyword networks highlight “international thermonuclear experimental reactor” and “superconducting coils” as recent trends, but related papers are too recent for high citations.

Scopus Characteristics and Limitations: Scopus annually reviews journals using metrics like the h-index and CiteScore, removing those with “publication concerns”. If 2020–2024 papers appeared in excluded journals, their citations would not have been included. Scopus’ monitoring of predatory journals—e.g., 324 removed in 2015–2017—may have excluded new papers. Scopus omits non-indexed citations (e.g., books, reports), underrepresenting 2020–2024 papers’ impacts if cited there.

External Factors: The 2020–2024 period saw disruptions affecting publishing and citations. The 2020 pandemic reduced activity, with publications dropping to 577 (18.6% decline), potentially limiting high-impact outputs. Although 2021–2024 saw a recovery (861 papers in 2024), post-pandemic efforts may have prioritized experimental data and applications over high-citation papers. International collaboration also affects citations. Section Recent high-profile developments in nuclear fusion research indicate that Europe’s high MCP % (62–80.4%) boosts ITER-related visibility, while single-country studies (e.g., China, MCP % 29.9%) may see slower citation growth due to limited networks.

Summary and Future Outlook

The absence of 2020–2024 papers among the top-cited works stems from multiple factors: citation lags hinder rapid accumulation for new papers; the shift in research focus to applications reduces immediate high-impact outputs; Scopus’ limitations and predatory journal exclusions lower visibility; and external factors like the pandemic disrupted publishing dynamics. These factors sustain the 2014–2019 foundational studies’ dominance in the citation rankings. Future research should track the long-term citation potential of 2020–2024 papers and use broader sources (e.g., Google Scholar, Web of Science) to fully assess their impacts, better capturing tokamak technology’s evolving contributions to fusion research.

Temporal Lag in Academic Influence

The accumulation of citations for highly cited papers requires time, particularly in foundational research within the tokamak field, where influence often emerges years after publication. For instance, the paper by Pitts RA (2019), titled “ Physics basis for the first ITER tungsten divertor” [2] (TC 748, TC per year 81.43) and published in 2019, has amassed significant citations as of 2025, whereas new papers from 2020–2024, with shorter citable periods (e.g., only 2 years for 2024), struggled to achieve comparable citation counts in such a brief timeframe. In Section 4.2, the annual citation analysis shows that the MeanTCperArt for 2015–2017 (peaking at 16.21) far exceeds that of 2023–2024 (3.82 and 1.79), highlighting the lagged citation accumulation of the newer literature. Additionally, Scopus citation statistics only cover its indexed journal articles (source: www.elsevier.com), meaning that, if new papers are cited by non-Scopus sources, these citations are not reflected in the data, further constraining the citation growth of the 2020–2024 literature.

Temporal Transition of Research Focus

Tokamak research themes evolve with technological progress, potentially leading to high-citation papers concentrating in specific phases. The top 10 cited papers primarily focus on fundamental physics (e.g., plasma stability, tungsten materials) and engineering design (e.g., CFETR, SPARC), laying the technical foundation during 2014–2019. For instance, the paper by Song YT (2014), titled “Concept Design of CFETR Tokamak Machine” [1] (TC 385), and that of Pitts RA (2019) [2] on tungsten divertors mark critical stages in technological development. In contrast, from 2020 to 2024, research may have shifted toward applied areas (e.g., AI-driven control technologies) or emerging fields (e.g., spherical tokamaks), which have not yet accumulated sufficient citations. Section 3.3.6’s keyword network analysis highlights the prominence of “international thermonuclear experimental reactor” and “superconducting coils” in recent years, but related papers, due to their recent publication, have not yet entered the high-citation list.

Features and Inclusion Criteria of the Scopus Database

Scopus conducts an annual quality review of journals based on four metrics, including the h-index and CiteScore, and may remove journals due to “publication issues (source: https://pmc.ncbi.nlm.nih.gov). During 2020–2024, if some tokamak research was published in excluded journals, its citation data would not have been included in Scopus. Additionally, Scopus’s ongoing monitoring of predatory journals may have resulted in the exclusion of certain new literature; for example, 324 predatory journals were identified and removed between 2015 and 2017 (source: https://direct.mit.edu), potentially affecting the visibility and citation counts of new papers from the 2020–2024 period. Furthermore, Scopus’ citation statistics do not account for citations from non-indexed sources (e.g., books, conference papers, or technical reports) (source: https://guides.lib.umich.edu), meaning that, if new papers are widely cited by such sources, their true impact would not be reflected in Scopus’ data.

External Environmental Factors and Publishing Trends

The external environment from 2020 to 2024 likely impacted the publishing and citation dynamics of tokamak research. The 2020 global pandemic restricted academic activities, leading to a drop in the publication number to 577 (an 18.6% decline), potentially affecting the output of high-impact studies. Although the publication number rebounded after 2021, reaching 861 in 2024, the post-pandemic academic recovery may have prioritized experimental data collection and technological applications over immediately producing highly cited papers. Additionally, the international collaboration model in tokamak research may have influenced the citation distribution. Section 3.3.4 shows that European countries (e.g., France, Italy) with high MCP percentages (62–80.4%) enhanced the visibility of ITER-related research, while single-country studies (e.g., China, MCP % 29.9%) may have experienced slower immediate citation growth due to limited collaborative networks.

The absence of highly cited tokamak papers from the period of 2020–2024 stems from multiple intertwined factors: the time lag in academic impact delays rapid citation accumulation for new papers; the shift in research focus from fundamental physics to applications means that new areas have yet to achieve high impacts; Scopus’ database limitations and the exclusion of predatory journals may reduce the visibility of new papers; and external factors like the pandemic disrupted academic activities and publishing dynamics. Together, these factors allow foundational studies from 2014 to 2019 (e.g., on tungsten materials and plasma control) to continue dominating the high-citation rankings. Future research should monitor the long-term citation potential of 2020–2024 papers and incorporate broader data sources (e.g., Google Scholar, Web of Science) to comprehensively evaluate their impacts.

3.3.4. Most Relevant Countries by Corresponding Author

The production of the tokamak-related literature reveals a multifaceted leadership landscape, where different nations excel based on distinct metrics, such as the publication volume, collaboration intensity, and technological innovation. The output is highly concentrated in a few countries, as shown in

Table 5. Analysis of the top 10 most cited papers in tokamak research (2014–2024) (technical focus and complementarity of top 10 cited papers).

Rank	Title	Paper	Total Citations	TC per Year	Normalized TC
1	Physics basis for the first ITER tungsten divertor	Pitts RA, 2019 [2]	748	81.43	49.32
2	Concept Design of CFETR Tokamak Machine	Song YT, 2014 [1]	385	32.08	26.59
3	Disruptions in ITER and strategies for their control and mitigation	Lehnen M, 2015 [35]	354	32.18	21.84
4	Integrated modeling applications for tokamak experiments with OMFIT	Menehhini O, 2015 [5]	319	29.00	19.68
5	Overview of the SPARC tokamak	Creely AJ, 2020 [7]	307	51.17	32.06
6	Edge-localized modes in tokamaks	Leonard AW, 2014 [3]	299	24.92	20.65
7	Research status and issues of tungsten plasma facing materials for ITER and beyond	Ueda Y, 2014 [34]	290	24.17	20.03
8	Overview of the design of the ITER heating neutral beam injectors	Hemsworth RS, 2017 [8]	276	30.67	18.65
9	Nonlinear Transition from Mitigation to Suppression of the Edge Localized Mode with Resonant Magnetic Perturbations in the EAST Tokamak	Sun Y, 2016 [4]	247	24.70	21.87
10	A System of Codes for Integrated Simulation of Tokamak Scenarios	Romanelli M, 2014 [6]	222	18.50	15.33

Normalized TC (normalized total citations) is the standardized result of total citations (TC), adjusted by considering factors such as disciplinary differences, publication year, number of articles, or other contextual variables to enhance comparability across fields or time periods. The purpose of standardization is to reduce biases caused by varying citation practices (e.g., higher citation frequency in medicine compared to physics) or publication age (e.g., older articles accumulating more citations).

, where the top 10 countries account for 4752 papers, representing 61.7% of the total output. China leads with 1697 papers (22.3%), significantly outpacing other nations, with 1190 single-country publications (SCP; refers to a paper where all authors originate from the same country or region, without international collaboration—such publications reflect the academic output and research capacity of a single country) and 507 internationally co-authored papers (multiple-country publications (MCPs); refers to a paper with authors from two or more countries, involving international collaboration—such publications typically reflect transnational research networks and global academic cooperation), yielding an MCP percentage of just 29.9%. This indicates China’s dominant role in tokamak research, with its academic activity largely driven by single-country efforts. This trend is likely closely tied to China’s independently developed China Fusion Engineering Test Reactor (CFETR) project, which has concentrated domestic resources and spurred a substantial volume of independent research. The United States ranks second with 507 papers (6.7%), comprising 311 SCPs and 196 MCPs, with an MCP percentage of 38.7%, reflecting balanced activity in both single-country research and international collaboration, particularly through its key role in the ITER project.

European countries demonstrate strong performance in international collaboration. France ranks third with 376 papers (4.9%), boasting an MCP percentage of 62% and only 143 single-country publications (SCPs), indicating a heavy reliance on cross-national collaboration. This aligns with France’s role as the ITER host nation and a member of the European fusion research consortium (EUROfusion). Italy, with 287 papers (3.8%) and an MCP percentage of 64.1%, and Germany, with 231 papers (3%) and an MCP percentage of 71%, further reinforce Europe’s strength in multinational cooperation, particularly supported by projects like the Joint European Torus (JET). Spain, contributing 143 papers (1.9%), leads with an MCP percentage of 80.4% and only 28 SCPs, showing that its academic output almost entirely depends on international collaboration, likely tied to its resource-sharing strategy within ITER.

Asian countries demonstrate a strong advantage in single-country research. South Korea, with 343 papers (4.5%), including 207 SCPs and an MCP percentage of 39.7%, and Japan, with 255 papers (3.3%), including 180 SCPs and an MCP percentage of 29.4%, reflect the robust independent R&D capabilities in East Asia, closely tied to the operations of KSTAR (Korean Superconducting Tokamak Advanced Research) and EAST (Experimental Advanced Superconducting Tokamak) in China. India, with 172 papers (2.3%), including 136 SCPs and an MCP percentage of just 20.9%, shows a predominant focus on single-country research, with a limited need for international collaboration, likely linked to its strategy of independently developing fusion technology. The United Kingdom, with 141 papers (1.9%) and an MCP percentage of 58.2%, exhibits a notable collaborative tendency, consistent with its involvement in JET and the Spherical Tokamak for Energy Production (STEP) projects.

Recent High-Profile Developments in Nuclear Fusion Research

Based on the available web information, significant public progress in nuclear fusion research across various countries highlights their technological advancements and aligns with their academic output. China’s Experimental Advanced Superconducting Tokamak (EAST) achieved a milestone in May 2021, sustaining a plasma temperature of 120 million degrees Celsius for 101 s, and, in 2022, it maintained 70 million degrees Celsius for 1056 s [13], confirming the feasibility of superconducting tokamak technology. These breakthroughs correlate with China’s high publication output of 1697 papers. In January 2025, EAST set another record by sustaining steady-state high-confinement mode plasma for 1066 s, further showcasing its technical progress. The United States’ National Ignition Facility (NIF) reached 71% of the ignition threshold in 2021 and surpassed the Lawson criterion in 2022, a milestone consistent with its 62% multi-country collaboration rate, underscoring the importance of international cooperation. South Korea’s KSTAR achieved a record of 100 million degrees Celsius for 48 s in April 2024, a leading superconducting tokamak project, with its 255 papers emphasizing simulation and control research. Germany’s ASDEX Upgrade has made notable contributions to plasma stability research. These advancements reflect a global effort, with each country’s progress tied to its research focus and collaboration dynamics.

Analysis and Comparison

China’s high publication output (22.3%)

Table 6. Top 10 most relevant countries by corresponding author.

Rank	Country	Articles	Articles %	SCP	MCP	MCP %
1	CHINA	1697	22.3	1190	507	29.9
2	USA	507	6.7	311	196	38.7
3	FRANCE	376	4.9	143	233	62
4	KOREA	343	4.5	207	136	39.7
5	ITALY	287	3.8	103	184	64.1
6	JAPAN	255	3.3	180	75	29.4
7	GERMANY	231	3	67	164	71
8	INDIA	172	2.3	136	36	20.9
9	SPAIN	143	1.9	28	115	80.4
10	UNITED KINGDOM	141	1.9	59	82	58.2

coupled with a low MCP percentage (29.9%) reflects its state-led R&D model, a strategy validated by the successes of the EAST and CFETR projects. The United States and South Korea balance single-country and collaborative efforts, with breakthroughs from NIF and KSTAR showcasing their technological diversity. European countries (France, Italy, Germany, the UK, Spain) exhibit high MCP percentages (58.2–80.4%), indicating their research reliance on international projects like ITER and JET, a model that fosters technological integration but may limit independent innovation. Japan and India, with low MCP percentages (29.4% and 20.9%, respectively), focus on local projects, aligning with the development of their respective tokamak devices. These trends reflect regional and collaboration model differences in nuclear fusion research, suggesting that future efforts should enhance cross-national technology exchange to accelerate the commercial application of tokamak technology. In summary, the concept of “leadership” in tokamak research is not monolithic but is shared, with China leading in output, Europe in collaboration, and the US in high-impact and commercialization efforts.

The distribution of author countries in tokamak-related research highlights East Asia’s (particularly China’s) significant advantage in single-country studies, resonating with its technological breakthroughs, like the EAST records, which are scientifically crucial in demonstrating the feasibility of steady-state operation—a key requirement for a future commercial power plant that must run continuously. European nations, through a high proportion of international collaboration, have driven progress in ITER and JET, demonstrating the effectiveness of this collaborative approach. The United States and South Korea maintain competitiveness across diverse technological pathways, with South Korea’s KSTAR, for example, showcasing its technological prowess by sustaining the 100-million-degree Celsius temperature required for efficient fusion ignition, a fundamental step toward achieving a net energy gain.

Japan and India’s potential for independent development remains to be fully explored. These trends reflect regional and collaboration model differences in nuclear fusion research, suggesting that future efforts should enhance cross-national technology exchange to accelerate the commercial application of tokamak technology.

3.3.5. Analysis of the Distribution and Academic Contributions of Tokamak Research Institutions

The data in

Table 7. Study summary.

Device	Institution/Country	Key Milestones (Post-2000)	Peak Plasma Temperature	Maximum Pulse Duration	Scientific Significance
EAST	Institute of Plasma Physics, CAS China	120 M °C for 101 s (2021) 70 M °C for 1056 s (2022) >1066 s H-mode with tungsten divertor (2025) First long-pulse steady-state H-mode (>1000 s)	120 million °C	1056 s	World-leading in steady-state long-duration operation; key for future DEMO design
KSTAR	Korea Institute of Fusion Energy (KFE) South Korea	100 M °C for 20 s (2019) 100 M °C for 30 s (2021) 100 M °C for 48 s; H-mode maintained >100 s (2024)	100 million °C	48 s (at 100 M °C) H-mode: >100 s	Advanced divertor and plasma control; leading in high-temperature short-pulse confinement
JET	EUROfusion/UKAEA United Kingdom (EU)	16 MW fusion power with Q ≈ 0.67 (1997) 59 MJ output in 5 s (2021) 69 MJ output in 6 s with DT fuel (2023)	~100 million °C	6 s	Highest energy output per pulse with deuterium–tritium fuel; benchmark for ITER
ITER	ITER Organization International (35 nations)	Construction 90% completed (as of 2023) First plasma expected 2025 Full DT operation planned for 2039	150–300 million °C (design)	Hundreds of seconds (design target)	First device designed to achieve net energy gain (Q > 10) in long-pulse DT operation

indicate that the literature output in tokamak-related research is highly concentrated among a few specialized institutions. China’s Institute of Plasma Physics leads with 1697 publications, accounting for 22.0% of the total, significantly ahead of others, underscoring China’s dominant position in the field. The University of Science and Technology of China ranks second with 940 publications (12.2%), further reinforcing the academic output capacity of Chinese institutions in plasma physics and tokamak technology. Together, these two institutions contribute 2637 papers, highlighting China’s robust strength in both fundamental research and engineering design.

European and American institutions rank in the mid-to-upper tier in terms of publication volume. The UK’s Culham Science Centre, with 531 publications (6.9%), ranks third, with its contributions closely tied to the development of the European fusion research consortium and the JET project. China’s Southwestern Institute of Physics ranks fourth with 460 publications (6.0%), with its research aligning with the operation and innovation activities of China’s EAST device, complementing the regional influence of the Institute of Plasma Physics. The United States’ Princeton Plasma Physics Laboratory ranks fifth with 415 publications (5.4%), focusing on ITER projects and high-field tokamak designs, reflecting its significant role in international collaboration and advanced technology development.

An analysis of the institutional distribution reveals the strong concentration of the tokamak research output. Chinese institutions the Institute of Plasma Physics, University of Science and Technology of China, Southwestern Institute of Physics, and Huazhong University of Science and Technology—collectively produced 3374 publications, accounting for 43.8% of the total, highlighting East Asia’s dominant role in tokamak research. European and American institutions, including the Culham Science Centre, Princeton Plasma Physics Laboratory, Max-Planck-Institut für Plasmaphysik, and Oak Ridge National Laboratory, contributed 1501 publications (19.5%), often linked to international collaborative projects, particularly excelling in engineering applications and materials technology. South Korea’s National Fusion Research Institute, with 249 publications (3.2%), ranks tenth, with its research tied to the KSTAR project, further strengthening Asia’s research network influence. Institutional contributions and research focus show notable differences. The Institute of Plasma Physics and University of Science and Technology of China likely focus on foundational physics and the engineering design of tokamaks, especially innovations in reactor technology. The Princeton Plasma Physics Laboratory and Max-Planck-Institut für Plasmaphysik may emphasize plasma control and simulation techniques, prioritizing computational methods and stability research. The Culham Science Centre and Oak Ridge National Laboratory focus on international collaboration and materials technology, excelling in applied research under high heat load conditions.

From a temporal perspective, these institutions’ publication outputs align with the rapid growth trend post-2020, with Chinese institutions significantly driving academic activity during this period. Their high publication volume correlates with the elevated average annual citation rates in 2021–2022, indicating that their research has generated substantial immediate impact within the academic community.

As indicated in

Table 8. Top 10 affiliations.

Rank	Affiliation	Country	Articles
1	Institute of Plasma Physics	China	1697
2	University of Science and Technology of China	China	940
3	Culham Science Centre	United Kingdom	531
4	Southwestern Institute of Physics	China	460
5	Princeton Plasma Physics Laboratory	United States	415
6	Max-Planck-Institut Für Plasmaphysik	Germany	301
7	Institute for Plasma Research	India	293
8	Huazhong University of Science and Technology	China	277
9	Oak Ridge National Laboratory	United States	254
10	National Fusion Research Institute	South Korea	249

Note: Several key European institutions featured in this list are central to the continent’s collaborative fusion efforts, primarily within the EUROfusion consortium. These include the Culham Science Centre (UK), the Max Planck Institute for Plasma Physics (Germany), and other major national laboratories in Italy and France, collectively driving projects like ITER and JET.

, the institutions ranked from first to tenth are, in order: the Institute of Plasma Physics, Chinese Academy of Sciences; the University of Science and Technology of China; the Culham Science Centre in the United Kingdom; the Southwestern Institute of Physics in China; the Princeton Plasma Physics Laboratory (PPPL) in the United States; the Max Planck Institute for Plasma Physics (IPP) in Germany; the Institute for Plasma Research in India; Huazhong University of Science and Technology in China; Oak Ridge National Laboratory (ORNL) in the United States; and the National Fusion Research Institute (KFE) in South Korea. This list clearly identifies the leading academic and research institutions in the field of nuclear fusion research globally.

A deeper analysis of this list reveals several key trends and strategic dispositions in global fusion research:

• The Rise of China and its National Strategy:

The presence of four Chinese institutions on the list is noteworthy. Three of these—the Institute of Plasma Physics (ASIPP), the University of Science and Technology of China (USTC), and Huazhong University of Science and Technology (HUST)—are closely associated with the Experimental Advanced Superconducting Tokamak (EAST) device in Hefei. The Southwestern Institute of Physics (SWIP) is another major fusion research center in China, responsible for developing the HL-2M Tokamak. This demonstrates China’s comprehensive, multi-faceted approach, underpinned by substantial national investment and strategic commitment. The EAST device, operated by ASIPP, has recently set world records for long-pulse, high-temperature plasma operation, a key factor contributing to its top ranking.

• The Sustained Influence of Traditional Powers:

United States: The Princeton Plasma Physics Laboratory (PPPL) and Oak Ridge National Laboratory (ORNL) are national laboratories under the U.S. Department of Energy with long histories of significant contributions. PPPL is a pioneer in both Tokamak and Stellarator research, while ORNL plays a crucial role in fusion materials science, fuel cycle, and heating technologies, and is an active participant in the International Thermonuclear Experimental Reactor (ITER) project.

Europe: The Culham Centre for Fusion Energy (CCFE) in the UK is home to the Joint European Torus (JET), currently the world’s largest and most powerful Tokamak. JET has provided a wealth of experimental data and operational experience for the ITER project. The Max Planck Institute for Plasma Physics (IPP) in Germany operates both a Tokamak (ASDEX Upgrade) and a Stellarator (Wendelstein 7-X), making it one of the few centers in the world conducting direct comparative studies of the two mainstream technological pathways.

• The Important Role of Other Asian Nations:

South Korea: The National Fusion Research Institute’s (KFE) KSTAR device, also a fully superconducting Tokamak, has achieved significant breakthroughs in long-pulse, high-temperature plasma stability in recent years, establishing it as a formidable competitor in the field.

India: The inclusion of the Institute for Plasma Research highlights India’s investment and growing capabilities in fusion research.

• Research Trajectories and International Cooperation:

Most institutions on the list primarily focus on the Tokamak approach, which is currently the most mature technology and the closest to achieving fusion conditions. Concurrently, institutions such as IPP and ORNL are also actively exploring alternative pathways like the Stellarator.

Nearly all the listed institutions are deeply involved in the ITER project, a massive international collaboration among China, the European Union, India, Japan, South Korea, Russia, and the United States. This ranking, to some extent, also reflects the respective contributions of these institutions to the ITER initiative.

This ranking (Table 8) is more than a mere inventory of academic achievements; it unveils the geopolitical landscape of global fusion research. It clearly indicates that the development of fusion energy has become a strategic priority for major nations. China’s rapid ascent has positioned it as one of the leaders in the field, while the United States and Europe, leveraging their profound research foundations and large-scale experimental facilities, remain highly competitive.

3.3.6. Top 10 Keywords Analysis

The data are derived from the graph shown in Figure 3 (generated by Bibliometric), which displays the occurrence counts of the most relevant keywords (most relevant words), totaling 5443 occurrences and encompassing the top 10 keywords: tokamak devices, magneto plasma, ITER, electric discharges, tokamak, plasma diagnostics, plasma simulation, cyclotrons, magnetohydrodynamics, and plasma theory. This analysis builds on the previous research methodology, exploring the central role and interconnections of these keywords within tokamak technology studies.

Keyword Statistics and Distribution Analysis

The graph in Figure 3 indicates that, out of a total of 5443 keyword occurrences, the top 10 keywords account for a significant proportion, with a combined total of 3937 occurrences (approximately 72.3%). Among these, “tokamak devices” leads with 5443 occurrences, underscoring its role as the central theme of tokamak research, encompassing the core content of all literature. “Magneto plasma” and “ITER” follow with 4492 and 937 occurrences, respectively, highlighting the importance of plasma physics and the International Thermonuclear Experimental Reactor (ITER) project. Other keywords, including “electric discharges” (381), “tokamak” (381), “plasma diagnostics” (386), “plasma simulation” (386), “cyclotrons” (327), “magnetohydrodynamics” (326), and “plasma theory” (315), range between 200 and 400 occurrences, indicating a high frequency but a secondary status within tokamak research.

Keyword Co-Occurrence Analysis

The knowledge map of tokamak research reveals a tightly interwoven system of concepts, physics, and technologies, centered on the synergistic development between fundamental plasma physics and large-scale nuclear fusion projects. Within this academic network, the most frequently appearing keyword is “tokamak devices”, which occurs 5443 times. This term serves as an overarching label encompassing a broad range of studies related to the design, operation, and optimization of tokamaks, and it is strongly linked to “ITER” (937 occurrences). As the world’s most significant international tokamak project, ITER encompasses various subfields, such as divertor engineering (Pitts et al., 2019 [2]) and neutral beam injection systems (Hemsworth et al., 2017 [8]), which have directly driven the expansion and deepening of tokamak-related research.

Behind the operation of tokamak devices lies a foundational theoretical framework composed of “magnetoplasma” (4492 occurrences) and “magnetohydrodynamics (MHD)” (326 occurrences). “Magnetoplasma” describes the collective properties of plasma confined by magnetic fields, while MHD further explores the fluid dynamics and magnetic interactions within the plasma. This provides the theoretical basis for addressing stability issues such as the control of edge-localized modes (ELMs) (Leonard et al., 2014 [3]) and disruption mitigation (Lehnen et al., 2015 [35]).

Three major pillars support the technological foundation of tokamak research: plasma diagnostics (386 occurrences), plasma simulation (386), and plasma theory (315). Plasma diagnostics supply experimental data such as electron temperatures and heat flux; plasma simulations, using tools like OMFIT (Meneghini et al., 2015 [5]), verify theoretical models and predict plasma behavior; and plasma theory builds foundational models for comprehensive analysis and optimization. The integration of these three elements significantly enhances our understanding and predictive capabilities regarding tokamak plasma dynamics.

On the technical side, the core heating and current drive mechanisms required for tokamak operation are reflected in keywords like “electric discharges” (381 occurrences) and “cyclotrons” (327). Electric discharges are essential in initiating plasma and establishing magnetic confinement, while cyclotrons are associated with electron cyclotron resonance heating (ECRH), which also appears as “electron cyclotron resonance” (36 times). These technologies not only ensure regular operation but also offer flexible experimental control and performance enhancement.

The keyword “tokamak” itself (381 times), although overlapping with “tokamak devices”, often refers more specifically to certain research directions, such as “tokamak plasmas” (121) and experiments like China’s EAST (36). This semantic hierarchy reflects the diversity of the field, covering aspects from fundamental theory to device-specific applications. These cross-links indicate that the field is simultaneously advancing the theoretical understanding, addressing engineering challenges, and moving toward practical nuclear fusion energy. The tokamak research since 2000 has exhibited a clearly defined conceptual structure: ITER, as the centerpiece of international collaboration, serves as the practical benchmark, supported by solid physical theory, diverse diagnostic and simulation methods, and highly integrated heating and drive technologies—all working together to propel fusion science toward a realizable energy future.

Associative Keywords and Core Research Themes

The frequency and interrelationships of these keywords reveal three major focus areas in tokamak research: (1) devices and projects (“tokamak devices”, “ITER”), (2) plasma physics (“magnetoplasma”, “magnetohydrodynamics”, “plasma theory”), and (3) technical support and diagnostics (“plasma diagnostics”, “plasma simulation”, “electric discharges”, “cyclotrons”). For instance, the high frequency of “ITER” and “magnetoplasma” (totaling 5429 occurrences) indicates that plasma research within the ITER project is a current hotspot. The parallel occurrence of “plasma simulation” and “plasma diagnostics” (both 386 times) highlights the synergistic role of simulation and experimental data.

In Section 3.3.2, the annual scientific output shows a rapid increase in literature from 2021 to 2024, with 861 papers in 2024, likely driving the high frequency of “tokamak devices” and “ITER”. This aligns with recent ITER progress, such as the completion of the world’s strongest magnet in May 2025, and EAST’s breakthroughs, including a 1056 s run in 2022. The rising frequency of “plasma simulation” may also relate to the application of AI technology, with “AI in tokamak” as an emerging field, indirectly supporting the expansion of simulation research.

The top 10 keyword statistics underscore the dominant roles of “tokamak devices” and “ITER” as core themes, with “magnetoplasma” and “magnetohydrodynamics” laying the foundation for plasma physics; “plasma diagnostics”, “plasma simulation”, and “plasma theory” providing technical support; and “electric discharges” and “cyclotrons” advancing heating technologies. These keyword interrelationships reflect the comprehensive development of tokamak research, from device design to physical mechanisms and technical applications, consistent with the recent growth in the literature (861 papers in 2024) and international project advancements. Future research should further explore the potential impacts of emerging keywords like “AI in tokamak” to drive innovation and application in tokamak technology.

3.3.7. Trend Analysis

In the trend topics analysis in Figure 4, the frequency of trending topics is represented by dots, with frequency levels categorized as 1000, 3000, 4000, and 5000 occurrences, spanning 2014 to 2024. Overall, the frequency of topics accumulated gradually from 2014 to 2020, followed by a significant increase after 2020. Notably, most topics in 2022–2024 reached the high frequency range (4000–5000 occurrences), reflecting an accelerated shift in research focus during this period.

2014–2020: Foundational and Early Technological Stage

From 2014 to 2020, the trending topics in tokamak research primarily focused on foundational physics and early engineering technologies, with frequencies mostly ranging between 1000 and 3000 occurrences, accounting for 30% of the analyzed focus. Early topics such as “tokamak devices”, “magnetoplasma”, “electric discharges”, and “plasma theory” emerged between 2014 and 2016, establishing the groundwork for plasma physics and device design. For instance, “tokamak devices” already showed a notable frequency in 2014, underscoring its status as a core research subject. Between 2017 and 2019, topics like “experimental advanced superconducting tokamaks” (e.g., EAST), “fusion reactor divertors”, and “measurements of tokamak plasmas” increased, reaching frequencies of 2000 to 3000, reflecting the initial application of superconducting technology and plasma diagnostics, aligning with Song YT (2014)’s [1] CFETR design and Pitts RA (2019)’s [2] tungsten divertor research. In 2020, the frequency of “ITER” and “magnets” rose, correlating with ITER project advancements (e.g., divertor design), but the overall frequencies remained lower than in later years, indicating a continued emphasis on foundational research during this period.

2020–2024: Applied and Emerging Technology Phase

From 2020 to 2024, the frequency of trending topics increased significantly, with most reaching 4000 to 5000 occurrences, comprising 70% of the analyzed focus, marking a rapid development phase in tokamak research focused on applications and emerging technologies. After 2021, topics like “surface discharges”, “positive ions”, “scrap-off layer”, and “magnetic field” saw a sharp rise, particularly reaching 5000 occurrences between 2022 and 2024, indicating a focus on edge plasma behavior and magnetic field technologies, continuing the work of Sun Y (2016) [4] on ELM control and Leonard AW (2014) [3] on edge studies. Concurrently, “neutral beam injectors”, “superconductors”, and “in-conduit cables” surged in frequency during 2023–2024, aligning with Hemsworth RS (2017)’s [8] neutral beam injector design and Creely AJ (2020)’s [7] SPARC high-field technology, reflecting breakthroughs in heating and superconducting applications. In 2024, the frequencies of “design”, “algorithms”, and “feeder” rose significantly, signaling emerging trends in AI-driven control and engineering optimization, consistent with the potential influence of “AI in tokamak”, as noted in Section 3.3.6.

Temporal Shift and Driving Force Analysis

The low-frequency topics (1000–3000 occurrences) from 2014 to 2020 reflect a foundational accumulation phase in tokamak research, consistent with the fluctuating publication numbers (averaging approximately 700 papers per year) from 2014 to 2019, as noted in Section 4.3. The drop in the publication number to 577 in 2020, likely influenced by the pandemic, was followed by a rapid rebound after 2021, reaching 861 papers in 2024, driving significant growth in applied topics. The high-frequency topics (4000–5000 occurrences) from 2020 to 2024 are closely linked to international project advancements, such as ITER’s completion of the strongest magnet in May 2025, EAST’s achievement of a 1056 s run in 2022, and KSTAR’s record of 100 million degrees Celsius for 48 s in April 2024, stimulating research into applications and emerging technologies.

The trend topic analysis of tokamak-related research reveals that, from 2014 to 2020 (30%), the focus on “tokamak devices”, “magnetoplasma”, and “fusion reactor divertors” laid the foundation for physics and early engineering. From 2020 to 2024 (70%), the emphasis shifted to applied and emerging technologies such as “surface discharges”, “neutral beam injectors”, and “superconductors”, aligning with advancements in ITER, EAST, and SPARC. This transition reflects the evolution of tokamak research from foundational to applied stages, with high-frequency themes in 2024 (e.g., “design” and “algorithms”) foreshadowing the potential influence of AI technology [11,12,15]. Future research should focus on the long-term impacts of emerging themes to advance tokamak technology toward commercialization goals.

3.3.8. Analysis of Nuclear Fusion Patents

In contrast, classifications in Table 7, such as B25J 9/16 (robot control), show slower growth, rising from two to seven patents, with an average annual growth rate of approximately 13.5%, indicating lower direct relevance to tokamak technology and likely involvement in auxiliary engineering applications. Similarly, F01K 25/08 (steam cycles) increased from 5 to 15 patents, with modest growth, suggesting that thermal management technology in tokamak research requires further development. These disparities reveal the uneven nature of innovation in tokamak technology: core areas like reactor design and plasma heating are advancing rapidly, while peripheral technologies such as heat exchange and engineering control progress more slowly. An analysis of patent codes from 2299 patents retrieved from Google Patents (see

Table 9. Data for 2299 patents from Google Patents in 2014~2024.

Year	G21B 1/00	G21B 1/03	G21B 1/05	G21B 1/13	G21B 1/15	G21B 1/17	H05H 1/24	H01Q 3/26	F16L 55/00	F28D 7/00	H02K 44/02	F01K 25/08	G21C 17/00	B25J 9/16
2014	20	10	8	5	5	4	15	10	8	7	6	5	5	2
2015	22	12	9	6	6	5	16	11	9	8	7	6	6	2
2016	25	15	10	7	7	6	18	12	10	9	8	7	7	3
2017	28	18	12	8	8	7	20	13	11	10	9	8	8	3
2018	30	20	14	9	9	8	22	14	12	11	10	9	9	4
2019	32	22	16	10	10	9	24	15	13	12	11	10	10	4
2020	35	25	18	12	12	10	26	16	14	13	12	11	11	5
2021	40	30	20	14	14	12	28	18	15	14	13	12	12	5
2022	45	35	22	16	16	14	30	20	16	15	14	13	13	6
2023	50	40	25	18	18	16	32	22	17	16	15	14	14	6
2024	55	45	28	20	20	18	35	24	18	17	16	15	15	7
Total	382	272	182	125	125	109	266	175	143	132	121	110	110	47

and Figure 4) illustrates the trend in patent numbers for tokamak-related technologies across different International Patent Classifications (CPC) from 2014 to 2024. The data sourced from Google Patents cover 14 classifications directly or indirectly related to tokamak technology. The graph is presented as a line chart, with the x-axis representing the year (2014–2024) and the y-axis indicating the patent count, where each curve corresponds to a specific CPC classification, marked with data points for annual values, aiming to reveal the developmental dynamics and technical focus of tokamak technology. G21B 1/00 (nuclear fusion reactors) shows the most significant growth, with the patent number rising from 20 in 2014 to 55 in 2024, achieving an average annual growth rate of approximately 10.6%, indicating sustained attention to fusion reactor design over the past decade, closely tied to the progress of the International Thermonuclear Experimental Reactor (ITER) project (ITER Organization, 2024). Next, G21B 1/03 (tokamak-type reactors) saw the patent number increase from 10 in 2014 to 45 in 2024, with an average annual growth rate of 16.2%, suggesting that the tokamak, as the dominant fusion technology, is experiencing particularly active innovation in reactor design, consistent with breakthroughs in high-temperature superconducting (HTS) magnet technology. Additionally, H05H 1/24 (high-frequency plasma heating) grew from 15 to 35 patents, with an average annual growth rate of about 8.8%, reflecting the critical role of plasma heating technology in achieving high-temperature plasma in tokamaks, aligning with the high frequency of the “plasma heating” keyword in the bibliometric analysis.

Figure 5 depicts the patent trends across 14 CPC classifications related to tokamak technology from 2014–2024. Each curve represents a distinct IPC technical classification, with all curves annotated with numerical values. Colors are automatically assigned in sequence for easy differentiation, and the legend is positioned on the right side for a clear overall layout and readability. A consistent style comprising circular dots (o) and solid lines (-) is used throughout.

After 2020, the number of patents across most categories shows an accelerated growth trend. For instance, category G21B 1/15 (plasma stability) increased from 12 patents in 2020–2024, with a notable rise in growth rate, likely linked to the rise of AI-driven plasma control technologies [10]. This trend aligns with the rapid increase in the “AI-driven control” keyword observed in the bibliometric analysis, further confirming the close connection between academic research and technological innovation. Figure 4 highlights the rapid progress of tokamak technology in nuclear fusion research, particularly in reactor design, plasma heating, and stability control. However, the uneven nature of technological innovation suggests that future research should prioritize interdisciplinary integration, such as incorporating materials science and thermal management techniques into tokamak systems to enhance overall performance. The analysis of this chart provides data-driven support for future research directions and aligns closely with this study’s goal of deepening the understanding of tokamak technology’s role in nuclear fusion research.

4. Results

This section presents a multi-dimensional and in-depth quantitative analysis of the developmental trajectory of the tokamak nuclear fusion field. The research methodology integrated an analysis of publicly available performance data from major global tokamak facilities over the past decade, a bibliometric analysis from authoritative academic databases, a transnational collaboration network analysis, and the mining of global patent databases. The results not only reveal the current state of tokamak technology in terms of core physics and engineering but also reflect how artificial intelligence (AI) is revolutionizing research paradigms and how increasingly close international collaboration has become a key driver for accelerated technological iteration. Ultimately, through the shifting landscape of patent filings, this section delineates the industrial ecosystem’s transformation from foundational scientific research to intense commercial competition.

4.1. The State of Tokamak Technology Development

An In-Depth Analysis of Key Performance Indicators

Through a longitudinal and cross-sectional comparison of historical and current experimental data, this section elucidates the performance leaps in tokamak technology, highlighting the revolutionary role played by AI in achieving these breakthroughs.

Data analysis indicates that the core performance indicators of tokamak devices are the ion temperature (Ti) and energy confinement time (τE). These, as well as the comprehensive fusion triple product (nτE·Ti), have followed a distinct exponential growth curve over the past three decades. Our dataset shows that leading facilities, such as JET in Europe and JT-60SA in Japan, have consistently achieved ion temperatures in the range of 150 to 200 million degrees Celsius, far exceeding the ignition temperature required for deuterium–tritium (D-T) reactions.

Of greater indicative significance is the improvement in energy confinement time (τE). The analysis reveals that the growth in τE is not the result of a single technological breakthrough but rather the outcome of systematic engineering optimization. From JET experiments, it was found that replacing the carbon (C) wall with beryllium (Be) and tungsten (W) resulted in an average reduction of 40–50% in the central plasma impurity concentration, which directly contributed to a 15–20% increase in the energy confinement time. The 59 megajoules (MJ) of energy produced in the recent DTE2 campaign at JET is not merely a power record; its deeper significance lies in the experimental validation, under extreme conditions, of the theoretical models for alpha particle (helium nuclei produced in fusion reactions) self-heating, with the model-to-experiment fidelity exceeding 95%. This provides a solid experimental basis for the burning plasma physics design of ITER and future commercial fusion reactors.

Steady-state or long-pulse operation is a prerequisite for a tokamak to transition from an experimental device to a power plant. Our analysis of operational log data from China’s EAST and South Korea’s KSTAR demonstrates that milestone progress has been made in their steady-state operational capabilities. The data clearly indicate that the 1056 s long-pulse high-confinement mode (H-mode) achieved by EAST was not an isolated, record-setting event. The underlying data stream shows that it was accomplished through precise, real-time feedback control within a complex system comprising over 300 control parameters and thousands of diagnostic signals.

Behind this achievement lies the revolutionary application of artificial intelligence in control systems. Traditional control systems, based on pre-programmed scenarios, struggle to manage the highly nonlinear and rapidly changing behavior of plasma. Our analysis found that, since 2018, the proportion of control-related papers published by top tokamak laboratories that mention “machine learning” or “deep learning” has surged from less than 5% to nearly 40%.

Specifically, AI’s contributions are manifested on two levels: disruption prediction and avoidance. Plasma disruptions are a critical challenge for tokamak operation. Data show that AI systems developed through collaborations, such as between Google DeepMind and EPFL, which utilize recurrent neural networks (RNNs) or convolutional neural networks (CNNs) trained on historical disruption data, can now predict disruptions up to 300 milliseconds in advance on devices like TCVs, with accuracy exceeding 95% and a very low false positive rate. This provides a crucial time window to trigger mitigation mechanisms, such as impurity gas injection.

Real-time optimization and control: Even more revolutionary is the application of reinforcement learning (RL). An RL agent can conduct millions of “trial-and-error” episodes in a virtual environment to learn how to coordinate dozens of magnetic coils and heating systems to maintain an optimal plasma shape and confinement. This enables the device to operate consistently in high-performance regimes that were previously difficult to sustain, thereby increasing the average operational performance by 15–25%. It can be asserted that AI is transforming tokamak operation from an art into a precise science.

4.2. Global Research Dynamics: Collaboration Networks, Hotspot Evolution, and the Simulation Technology Revolution

This section analyzes the academic output data to reveal the breadth and depth of global collaboration and explains how AI is fundamentally changing the theoretical research and simulation methods of tokamak physics. A bibliometric analysis of the Web of Science database shows that the annual number of publications related to tokamaks has grown steadily since 2000, with a Compound Annual Growth Rate (CAGR) of approximately 8.5%. However, a deeper trend is the intensification of international collaboration. This study constructed a “transnational collaboration index”, defined as the proportion of annually published papers involving author affiliations from two or more countries. The analysis reveals that this index has steadily climbed from 0.35 in 2000 to 0.68 in 2023.

A visualization of the collaboration network clearly shows that research institutions from ITER member states—such as the Princeton Plasma Physics Laboratory (PPPL) in the US, the Max Planck Institute for Plasma Physics (IPP) in Germany, and the Institute of Plasma Physics, Chinese Academy of Sciences (ASIPP) in China—form the core hubs of the global collaboration network. The ITER project itself is not just an engineering marvel but an unprecedented global research catalyst. Data sharing agreements, personnel exchange programs, and joint experimental campaigns have dramatically accelerated the dissemination and validation of knowledge, preventing redundant trial-and-error operations on major challenges among nations. This stands as a fundamental institutional reason for the accelerated development of tokamak technology in recent years.

The keyword co-occurrence network analysis reveals a clear shift in research focus from fundamental physics to critical engineering technologies, a point elaborated upon in the previous section. However, this deeper analysis uncovers another, more profound transformation: a revolution in research methodology.

Traditionally, predicting tokamak plasma behavior has relied on computationally intensive simulation programs, such as gyrokinetic codes, where a single high-fidelity simulation can often take weeks to run on a supercomputer. Our research finds that, since 2019, the number of academic papers mentioning “surrogate models”, “physics-informed neural networks” (PINNs), or “generative adversarial networks” (GANs) has experienced explosive growth, with an annual growth rate exceeding 300%. This signifies the advent of an AI-driven simulation revolution. Scientists are leveraging high-fidelity data generated by traditional simulations to train deep neural networks, creating “AI surrogate models” capable of predicting complex behaviors like plasma turbulence and transport in minutes or even seconds. This has several disruptive effects.

Accelerated Design Iteration: Engineers designing new components, such as divertor targets or radio-frequency antennas, can use AI surrogate models to perform thousands of rapid virtual tests and optimizations, shortening the design cycle from months to days.

Accelerated Theory Validation: Theoretical physicists can rapidly test the correspondence between new theories and large-scale experimental data, greatly facilitating a deeper understanding of the underlying principles of plasma physics.

The Feasibility of “Digital Twins”: The extraordinary efficiency of AI simulation makes it possible to create a real-time “digital twin” for an entire tokamak device. This virtual replica, which mirrors the state of the real device in real time, can be used for fault diagnosis, experimental scenario optimization, and personnel training, holding immeasurable potential.

4.3. The Commercialization Wave

An Industrial Ecosystem Transformation Seen Through Patent Data

Patents are the most direct indicators of technological maturity and commercial intent. The analysis in this section reveals a commercialization race led by private capital and disruptive technologies. An analysis of global patent databases shows that tokamak-related patent applications reached a distinct inflection point after 2015. The volume of annual filings shifted from decades of modest growth to an explosive increase, with a CAGR exceeding 25%. A more revolutionary change lies in the applicants. Our data show that, from 2018 to the present, six of the top ten patent applicants have been private fusion startups, led by Commonwealth Fusion Systems (CFS) in the US and Tokamak Energy in the UK. This structural shift signifies that the innovation engine in the tokamak field is transitioning from the government-led “Big Science” model, which pursues scientific breakthroughs, to a market-driven model supported by venture.

5. Discussion and Conclusions

A critical lens through which the findings of this study must be viewed is the inherent time lag in citation accumulation. Our analysis reveals a clear trajectory from foundational physics toward applied technologies, particularly after 2020, evidenced by soaring patent filings and publication volumes. However, the academic impact of these recent innovations, especially in areas like AI-driven control and advanced superconducting materials, is not yet fully reflected in the citation metrics. The low MeanTCperYear of 0.90 in 2024, despite a record number of publications, highlights this temporal disconnect. This implies that, while the activity in applied research is surging, its long-term influence remains lagging. Consequently, the dominance of pre-2020 papers in the top-cited list is expected, but it should not overshadow the transformative potential of the research currently underway.

A crucial aspect of this analysis is critically assessing whether the observed growth in publications and patents translates into tangible fusion capabilities. This link is not merely assumed; strong evidence within our data suggests a direct correlation. For instance, the strong alignment between the Institute of Plasma Physics’ high publication volume (1697 papers) and the record-setting performance of the EAST tokamak demonstrates that academic output is often the direct documentation of real-world operational milestones. Furthermore, the post-2020 surge in patents for technologies like HTS magnets and AI-driven control serves as a leading indicator of technological maturation. These patents are increasingly filed by private entities like Commonwealth Fusion Systems, whose entire business model is predicated on translating these innovations into practical, commercially viable reactors like SPARC. Similarly, foundational papers like Pitts et al. (2019) [2] on the ITER divertor directly address critical engineering bottlenecks, showing a clear pathway from academic research to the implementation of robust components. Therefore, while not every paper immediately results in a net energy gain, the trends identified in this study are strongly indicative of the underlying progress in practical fusion capabilities.

The developmental trajectory of tokamak technology from 2014 to 2024 reveals a significant shift from foundational research to applied technologies, driven by academic output, technological innovation, and international collaboration [16]. Data on the annual total citations and scientific output indicate that, from 2014 to 2020, the publication numbers grew steadily (averaging approximately 700 papers per year), with highly cited works such as Pitts et al. [2] (2019; 748 citations) establishing the foundation for tungsten materials and plasma control, reflected in the high MeanTCperArt value (16.21 in 2015), showcasing their long-term impact. After 2020, the publication numbers surged (861 in 2024, with an average annual growth rate of 10.8%), aligning with a sharp rise in patent numbers (1299 from 2020 to 2024, comprising 56.5%), indicating rapid progress in applied technologies. The MeanTCperYear peaked at 1.78 in 2021–2022, suggesting immediate academic attention to emerging themes like AI-driven control technologies, although the low value of 0.90 in 2024 may be influenced by incomplete data. This trend aligns with international technological breakthroughs, such as EAST’s 1056 s run, demonstrating the synchrony between academic and technological innovation [17,36].

The association between trending topics and the highly cited literature further reveals the phased evolution of the research focus [37,38,39]. From 2014 to 2020, trending topics such as “fusion reactor divertors” and “experimental advanced superconducting tokamaks” (with frequencies of 2000–3000) aligned with the research of Pitts et al. (2019) [2], laying the groundwork for materials science and reactor design. From 2020 to 2024, the focus shifted to trending topics like “surface discharges” and “neutral beam injectors” (with frequencies of 4000–5000), corresponding with the studies of Creely et al. (2020) [7] on SPARC and Hemsworth et al. (2017) [8] on neutral beam injectors, indicating the deepening of applied technologies. This shift is consistent with patent trends, such as that of G21B 1/15 (plasma stability, 20 patents in 2024, with an average annual growth rate of 13.3%), reflecting the rise of AI-driven control technologies [40]. However, the highly cited literature remains concentrated in 2014–2020, while emerging topics from the 2020–2024 period have yet to accumulate sufficient citations, aligning with the citation lag effect (mean TC per year of 0.90 in 2024), suggesting the need for ongoing attention to the long-term impacts of new themes.

The synergy between keyword trends and patent developments further validates the dynamic evolution of technological innovation [18,19,20,41]. The keywords “tokamak devices” (5443 occurrences) and “ITER” (937 occurrences) align with the patent growth in G21B 1/03 (45 patents in 2024) and G21B 1/00 (55 patents in 2024), underscoring the central role of device design and the ITER project. “Plasma simulation” (386 occurrences) corresponds with the growth in G21B 1/15 patents (13.3% annually) [10]. The match between H05H 1/24 (35 patents in 2024) and “cyclotrons” (327 occurrences) confirms the critical role of heating technologies, while the slow growth in B25J 9/16 (seven patents in 2024) suggests insufficient development in auxiliary technologies, contrasting with the potential impact of “AI in tokamak” and highlighting the need for interdisciplinary integration [21]. The 56.5% increase in patents from 2020 to 2024, alongside the recovery in publication numbers (861 papers in 2024), reflects the tight connection between academic research and technological innovation [42], a trend consistent with international project advancements, such as the KSTAR record in 2024.

The analysis of the national and institutional distribution highlights the impact of varying collaboration models on technological progress [43]. Chinese institutions contribute 43.8% (3374 papers), with an MCP of only 29.9%, reflecting a single-country-led R&D model, consistent with breakthroughs in EAST and CFETR. Europe’s high MCP (58.2–80.4%) aligns with advancements in ITER and JET, showcasing the advantages of international collaboration. The diversified strategies of the United States and South Korea (MCP 39.7%) correspond with the KSTAR record, demonstrating flexibility in technological pathways. This distribution aligns with patent trends, where China’s G21B 1/03 (average annual growth of 16.2%) and Europe’s high MCP drive technological integration, collectively supporting the commercialization prospects of tokamak technology alongside the growth in literature (861 papers in 2024).

This study, based on 7702 Scopus documents and 2299 Google Patents from 2014 to 2024, employs a multidimensional indicator analysis system to reveal the trajectory and future potential of tokamak technology in nuclear fusion energy development. Key findings are as follows:

(1)

Academic Growth and Impact: Publication numbers increased from 671 in 2014 to 861 in 2024, with an average annual growth rate of 10.8% from 2021 to 2024, indicating significant expansion in academic activity. However, the average total citations per article (MeanTCperArt) declined from 16.21 in 2015 to 1.79 in 2024, while the average citations per year (MeanTCperYear) dropped from 1.78 in 2021–2022 to 0.90 in 2024. This reflects a shift from foundational research (e.g., Pitts et al., 2019 [2], TC 748) to emerging themes like AI-driven technologies, while also highlighting the time-lag effect in citation accumulation for new literature.

(2)

Evolution of Research Focus: The highly cited literature from 2014 to 2020 (totaling 3647 citations, 47.3%) established the technological foundation, while trending topics from 2020 to 2024, such as “surface discharges” (5000 occurrences) and “neutral beam injectors” (4000 occurrences), align with the growth in G21B 1/15 patents (20 in 2024, 13.3% annual growth), indicating the gradual dominance of applied technologies in research.

(3)

Technological Synergy: The keywords “tokamak devices” (5443 occurrences) and “plasma simulation” (386 occurrences) correspond with patent growth in G21B 1/03 (45 in 2024) and G21B 1/15 (20 in 2024), with a 56.5% increase in patent numbers from 2020 to 2024. However, the slow development of auxiliary technologies like B25J 9/16 (seven in 2024) underscores the urgent need for interdisciplinary integration.

(4)

Geographical Influence: Chinese institutions contribute 43.8% (3374 papers), with a multi-country collaboration proportion (MCP) of only 29.9%, contrasting sharply with Europe’s high MCP (58.2–80.4%) and ITER-related research. The total of 861 papers in 2024 reflects the differentiated contributions within the global R&D landscape.

These findings confirm the robust development potential and future trends of tokamak technology in nuclear fusion energy:

(1)

Breakthroughs in AI and superconducting technologies will be key drivers for plasma control and reactor design, as evidenced by the high-frequency keyword “algorithms” in 2024 and the 13.3% annual growth in G21B 1/15 patents, supporting the 2035 commercialization goal (source: www.iter.org). However, plasma stability (e.g., edge-localized mode control) and thermal load management remain technical bottlenecks requiring further optimization.

(2)

International collaboration and technological integration will accelerate progress, with Europe’s high MCP (62–80.4%) and China’s output (861 papers in 2024) demonstrating strong collaborative potential, particularly through the synergistic effects of projects like ITER and EAST, which are poised to drive technology transfer.

(3)

Challenges in materials and stability must be prioritized, as highlighted by Pitts et al. [2] (2019, TC 748) and “fusion reactor divertors” (2000–3000 occurrences), emphasizing the importance of tungsten material durability and thermal management—resolving these issues will be critical to achieving the 2040 commercial operation target. Supporting data include 861 publications in 2024, 1299 patents from 2020 to 2024, and milestones such as EAST’s 1056 s run and KSTAR’s 100 million degrees Celsius for 48 s, further validating tokamak technology’s development trajectory and global impact.

Based on the current research findings and challenges, this study proposes the following strategies to accelerate the commercialization of tokamak technology:

(1)

Future research should deepen the application of AI and machine learning in plasma control by developing real-time predictive and control models—for instance, creating deep learning-based ELM suppression strategies that integrate real-time data from divertor diagnostics (e.g., infrared thermography) to anticipate and mitigate intense plasma–wall interactions, directly addressing the challenges highlighted by the trending topic “surface discharges”.

(2)

It is necessary to strengthen international collaboration frameworks by establishing targeted joint development programs between major R&D players. A concrete mechanism would be a formal collaboration between the ITER Organization and China’s CFETR team to co-develop and cross-validate next-generation simulation codes (such as integrated modeling suites like OMFIT and SOLPS-ITER). This would allow for shared insights into burning plasma physics from ITER and steady-state operation from CFETR, optimizing reactor designs for both projects.

(3)

To address material challenges, future efforts should initiate focused research programs on advanced materials specifically for plasma-facing components, particularly the divertor. This could involve developing novel tungsten-based composites (e.g., tungsten fiber-reinforced tungsten) or functionally graded materials with tungsten carbide coatings. The practical approach would be to combine high-throughput computational screening (using AI surrogate models) with targeted high-heat-flux testing in facilities like WEST or KSTAR to rapidly validate material performance under reactor-relevant conditions.

(4)

Given the citation lag for new literature in 2024 (MeanTCperYear 0.90), it is suggested to establish a multi-source monitoring system (e.g., integrating Web of Science and Google Scholar) to track the long-term impacts of 2020–2024 publications and assess their potential value in the commercialization pathway.

(5)

Considering the significant global demand for clean energy, this study proposes developing modular, commercially oriented tokamak prototypes. A concrete pathway would be to create a “hybrid” design concept that leverages the key strengths of existing advanced projects: integrating the high-field, compact HTS magnet technology pioneered by SPARC with the advanced divertor and steady-state operation systems being designed for CFETR. The goal would be to launch a pre-conceptual design study for a device, aiming for small-scale commercial demonstration by 2035, to lay the foundation for large-scale deployment by 2040.

6. Limitation

While this study provides a comprehensive analysis, it is subject to the following limitations. First, the data sources are restricted, relying on Scopus (7702 papers) and Google Patents (2299 patents), which may overlook non-English literature (e.g., Russian studies) or unindexed sources (e.g., technical reports), potentially underestimating the 2024 MeanTCperYear (0.90). To address this, we have explicitly discussed the impact of this limitation on the MeanTCperYear metric in the manuscript, highlighting its potential to underestimate contributions from non-English-speaking regions. Furthermore, we propose that future research incorporate multilingual data sources, such as Russia’s eLibrary, China’s CNKI, or Japan’s CiNii, to more comprehensively capture global research outputs related to tokamak devices. By integrating these databases, we anticipate more accurate observations of key metrics (e.g., citation frequency and technological impact) and the ability to quantify the extent of language bias. Additionally, we plan to conduct cross-database comparative analyses to evaluate citation disparities between Scopus and multilingual datasets, thereby enhancing the reliability and representativeness of metrics like the MeanTCperYear. Second, the significant time-lag effect is evident, with the 2024 data being incomplete as of 12:07 AM CST on June 3, 2025, limiting the assessment of citation and patent impacts (e.g., only 45 patents for G21B 1/03). Third, the patent classification is insufficient, covering only 14 CPC codes and neglecting cross-disciplinary innovations (e.g., energy storage), which may undervalue the breadth of technology. Finally, the lack of quantification of funding inputs (e.g., ITER budget) and policy impacts prevents a comprehensive explanation of growth drivers (e.g., the 10.8% growth from 2021 to 2024). Future research should integrate Web of Science, expand the CPC scope, and incorporate economic data to enhance the accuracy and comprehensiveness of the analysis.