The data analysis for insulating transformer oil research was divided into two main segments: descriptive analysis and scientometric analysis [23, 24].

The Document Co-citation Analysis (DCA) of transformer oil research has identified 12 distinct co-citation clusters, ranked by their sizes, with the largest cluster designated as #0. In this analysis, the size of the circles within each cluster represents the influence of the publications, with larger circles indicating higher citation rates. This effectively highlights the key papers that have significantly impacted their fields. Additionally, the lines extending from each cluster illustrate the duration or "lifetime" of the cluster, providing insights into the development and evolution of research themes over time.

The size of each cluster is directly proportional to the number of publications it contains, with six out of the twelve clusters including more than fifty publications each. Figure 8 illustrates the top 12 clusters, displayed horizontally with cluster labels positioned on the right. Additionally, Fig. 8 depicts the network of all the articles. The clusters are numbered and ranked by size, beginning with #0 as the largest cluster and #86 as the smallest. The size of each circle represents the magnitude of a publication's influence, where larger circles correspond to greater impact. This highlights significant engagement within these thematic areas, with Cluster #0 emerging as the most prominent based on the number of publications.

The red rings surrounding the circles indicate the "bursts" of articles, revealing when bursts occurred and their relative strength. Furthermore, the silhouette scores for these clusters range from 0.919 to 0.998, reflecting a high level of homogeneity within each cluster's publications. Silhouette scores measure the consistency of objects within their own cluster compared to other clusters, with values above 0.900 demonstrating strong internal coherence.

The clusters were named using four different methods: (i) Latent Semantic Indexing (LSI), (ii) Term Frequency–Inverted Document Frequency (TF-IDF), (iii) log-likelihood ratio (LLR), and (iv) Mutual Information (MI). The Log-Likelihood Ratio (LLR) is generally considered superior to Mutual Information (MI) and Latent Semantic Indexing (LSI) due to its higher discriminative power and statistical robustness. LLR identifies terms that are uniquely and significantly associated with a specific cluster, enabling more precise and representative labeling [31, 32]. Unlike MI, which often highlights frequently occurring terms that may appear across multiple clusters—resulting in vague or generic labels—LLR emphasizes the distinctiveness of terms, thereby enhancing the clarity and specificity of each cluster's thematic identity. Similarly, while LSI is valuable for uncovering latent semantic structures through dimensionality reduction techniques, it tends to produce abstract or less interpretable terms that are not optimal for direct cluster labeling. LLR, by contrast, utilizes a statistical framework that evaluates the likelihood of a term’s association with a cluster beyond random chance, making it particularly effective for generating meaningful and contextually grounded descriptors [33]. As such, LLR is preferred in scientometric studies that prioritize clarity, specificity, and analytical rigor in the interpretation of knowledge domains. Following the approach outlined in [34],this paper focuses on the clusters identified using the log-likelihood ratio (LLR), as the outputs of the other methods occasionally produced less coherent results. Table 6 provides a comprehensive overview of research clusters, characterized by Cluster ID, Size, Silhouette score, Label (LSI), Label (LLR), Label (MI) and Average Year. The analysis identified the most influential clusters, which are summarized in Table 6. The clusters range in size from 4 to 176 articles, with silhouette scores indicating the cohesion within clusters, where a score of 1 denotes perfect cohesion. The clusters represent diverse research themes, with Cluster 0 (Dielectric Strength) being the largest (176 publications, average year 2017), highlighting its longstanding importance in transformer insulation studies. Cluster 1 (DFT Study), while smaller (101 publications), focuses on emerging computational methods like Density Functional Theory (average year 2020) with a high silhouette score (0.970), indicating strong thematic coherence. Notable older themes include Cluster 7 (Copper Sulfide Deposition) with an average year of 2008 and Cluster 18 (Hydrothermal Synthesis) from 2009, which emphasize historical research trends. Conversely, newer themes like Cluster 8 (Health Index) and Cluster 86 (Alternative Liquid Dielectric) reflect contemporary innovations in transformer diagnostics and sustainable materials, both with an average year of 2020. This distribution highlights the field's evolution, balancing foundational studies with modern advancements.

The most cited papers in the field shows in Table 7. The citation statistics were collected from WoS database. The journal is a mixed source from mainstream business economics journals through field journals. The top three articles by [7, 35, 36] have been very heavily cited, with more than 20 citations. The results indicated that [36] received the highest citation count which is published in the most recognized journal, Applied Surface Science. The study titled “Pd-doped MoS₂ monolayer: A promising candidate for DGA in transformer oil based on DFT method” stands out as the most cited work among the top 10 publications in Cluster #1. Despite Cluster #1 having fewer publications than Cluster #0, it focuses on cutting-edge methods like DFT (Density Functional Theory), which has significant practical relevance. This emphasis on innovative approaches has made studies in this cluster highly influential in both academic and industrial circles.

Article citation is an indicator that shows the impact of a study in its research field. Based on previous study, the direction of one research field is associated with its frequently cited articles [37–39]. H. Cui, X. Zhang, G. Zhang and J. Tang [36] study uses Density Functional Theory (DFT) to explore Pd-doped MoS₂ monolayers as advanced materials for dissolved gas analysis (DGA). Its high citation count likely stems from its innovative approach to diagnosing transformer oil degradation, combining computational methods with practical implications in improving transformer reliability. The focus on nanomaterials for DGA highlights the study's relevance to emerging trends in diagnostics for energy systems. H. Cui, D. Chen, Y. Zhang and X. Zhang [35] similar to the first study, this research emphasizes DFT to evaluate Pd-decorated MoSe₂ monolayers for gas sensing applications in transformer oil. Its high impact is due to its alignment with the ongoing interest in nano-engineered solutions for precise and efficient fault detection in transformers. The theoretical insights into gas adsorption mechanisms contribute to both material science and transformer diagnostics. Z. Shen, F. Wang, Z. Wang and J. Li [7] This comprehensive review of plant-based insulating fluids reflects a growing interest in sustainable alternatives to mineral oils. The high citation rate is likely due to its coverage of technological advancements, environmental benefits, and challenges in implementing plant-based fluids. The article's relevance is amplified by the global push towards greener technologies in the energy sector.

In summary, the high citation counts of these articles are due to their cutting-edge focus on nanotechnology for transformer diagnostics [35, 36] and sustainable insulating fluids [7]. Their combination of theoretical rigor, practical applications, and alignment with current research trends makes them highly influential in the transformer oil research domain.

Table 8 provides a comprehensive overview of most cited article in the top 3 research clusters. Cluster #0 is the largest cluster in this analysis, comprising 176 articles with a silhouette value of 0.930, indicating clear differentiation from other clusters. Labelled as “dielectric strength” through the Log-Likelihood Ratio (LLR) method, this cluster emphasizes improving transformer insulation performance, focusing on enhancing the dielectric properties of insulating fluids. In this cluster, [7, 41, 44, 47, 48] were identified as the most influential publications. It includes studies on plant-based insulating fluids, nanoparticle-enhanced transformer oils, and alternative dielectric fluids, reflecting a broad spectrum of approaches to address both technical and environmental challenges.

Key articles in this cluster, such as [7] review of plant-based insulating fluids and [41] work on TiO₂ nanoparticle-enhanced transformer oil, illustrate the diversity and relevance of research. These studies influence academia and industry by highlighting sustainable solutions, advancing material science, and addressing performance optimization in transformer insulation systems. The focus on bridging experimental results, theoretical modeling, and practical applications makes this cluster pivotal in shaping the future of transformer technology.

Cluster #1 is the second-largest cluster, comprising 101 articles with a high silhouette value of 0.970, labeled as “DFT study.” This cluster gained prominence through the application of Density Functional Theory (DFT) in addressing challenges related to transformer oil diagnostics. The most influential publications, such as [35, 36, 40, 43, 49] focus on leveraging advanced nanomaterials as sensors for dissolved gas analysis (DGA). These studies introduced novel methods for detecting critical fault gases, including C₂H₂, H₂, and CO, by utilizing DFT to predict material behavior at the atomic scale. In particular, publications [35, 36] emphasize the catalytic and adsorption properties of Pd-doped materials, showcasing their potential for enhanced gas detection in transformer diagnostics. These articles collectively demonstrate innovative approaches involving nanomaterial design, adsorption behavior, and catalytic properties. They align in their methodology (DFT) and objectives (improving transformer oil fault detection), forming a cohesive narrative within the cluster and highlighting their collective impact on advancing transformer diagnostics.

Cluster #2 is the third-largest cluster, consisting of 70 articles with a silhouette value of 0.976, indicating clear differences from other clusters in the network. It is labeled as “Power Transformer” by the Log-Likelihood Ratio (LLR) test method. This cluster emphasizes advancements in dissolved gas analysis (DGA), focusing on fault diagnosis and condition monitoring in power transformers. The most influential articles in this cluster include [42, 45, 50–52], which highlight the importance of innovative diagnostic approaches.

Research within this cluster discusses the evolution of DGA techniques, starting from traditional measurement and interpretation methods to cutting-edge applications like advanced sensors and artificial intelligence. Articles [42, 45, 52] provide comprehensive reviews of DGA methods, stressing the importance of real-time monitoring and precision. Meanwhile, [50, 51] explore novel approaches such as Pd-decorated ZnO nanorod sensors and genetic algorithms for optimizing diagnostic accuracy. Collectively, this cluster underscores the need for continuous innovation to enhance transformer reliability, extend service life, and reduce downtime, making DGA a cornerstone of transformer maintenance strategies.

A burst detection analysis was performed to identify the most influential or landmark publication in the field that had drawn researcher’s attention. The top 10 major bursts in citations, with the duration shows in Table 9. The top 3 articles with the highest burst are (N. A. Bakar, A. Abu-Siada and S. Islam, 2014; E. Atiya, D.-E. Mansour, R. Khattab, and A. Azmy, 2015; W. Sima, J. Shi, Q. Yang, S. Huang, and X. Cao, 2015). (N. A. Bakar, A. Abu-Siada and S. Islam, 2014) [42], explores the use of Dissolved Gas Analysis (DGA) for diagnosing faults in transformers by monitoring gas concentrations in insulating oil. This method is recognized for its reliability and effectiveness in detecting incipient faults, making it widely influential in both research and practical applications.

(E. Atiya, D.-E. Mansour, R. Khattab, and A. Azmy, 2015) [41] investigates the enhancement of transformer oil insulation properties using TiO₂ nanoparticles. It emphasizes improvements in breakdown strength and dielectric performance, making it a significant study in the adoption of nano-dielectric fluids for modern transformers. The third articles with high burstiness is (W. Sima, J. Shi, Q. Yang, S. Huang, and X. Cao, 2015) [44] provides both experimental and theoretical insights into how nanoparticle characteristics influence insulation performance. It underscores the role of nanoparticle conductivity and permittivity in optimizing oil properties, contributing to a deeper understanding of nano-enhanced transformer oil.

The publication trends demonstrate a significant and steady growth in research output, signaling increasing global interest in insulating transformer oils, particularly over the last two decades. The environmental concerns surrounding traditional mineral oils, such as their non-biodegradability and fire hazards, have further propelled interest in bio-based insulating fluids The biodegradability of insulating liquids affects how long they remain in the environment before breaking down. Mineral oils typically take between 1 and 10 years to decompose, whereas vegetable-based oils like canola oil degrade much faster, typically within 4 to 48 weeks, demonstrating their lower environmental persistence [58]. Biodegradable natural esters are capable of meeting the performance criteria required for insulating oils in power equipment applications [59]. This surge is closely tied to advancements in nanotechnology, which have enabled the development of nano-enhanced transformer oils with superior electrical and thermal properties, as well as the push for more sustainable and high-performance insulation solutions. A review paper [13] reports that most studies observe an enhancement in the flash point of mineral oil when nanoparticles are incorporated. Studies from [60], investigated the impact of nanocomposites and single nanoparticles on key thermal parameters such as the convective heat transfer coefficient (CHTC), Nusselt number (Nu), Grashof number (Gr), and Rayleigh number (Ra). Their findings revealed that titanium dioxide (TiO₂​) nanoparticles had a more pronounced effect on enhancing both CHTC and Nu compared to (ZnFe₂​O₄​) nanoparticles, highlighting the superior thermal performance of TiO₂​​-doped nanocomposites in transformer oil applications [60]. Palm-based transformer oil is recognized for its enhanced fire safety properties, exhibiting a flash point above 280°C and a fire point exceeding 300°C, both of which are considerably higher than those of mineral oil [61]. This improvement suggests that nanoparticle additives can increase the thermal stability and safety margins of transformer oil-based insulating fluids. Between 2015 and 2024, several research landscapes shifted toward exploring bio-based and nano-enhanced insulating oils, a trend supported by the urgent need for environmentally friendly and efficient alternatives to mineral oil in transformers. This shift is reflected in studies like those by Zakaria et al., who investigated palm-based oils as biodegradable alternatives for power transformers in Malaysia [29]. Additionally, research by Peppas et al. highlighted the improved performance of nanofluids, such as those incorporating nanoparticles, when compared to traditional oils [62]. Moreover, Mohd et al. explored the potential applications of palm oil products as electrical insulating mediums, further supporting the move towards sustainable alternatives in transformer technology [61].

The IEEE Transactions on Dielectric and Electrical Insulation stands out as the leading journal in this field, contributing 15.39% of total publications. Its dominance is attributed to its niche focus on dielectric materials, its global visibility as part of the IEEE ecosystem, and its role as a trusted platform for publishing cutting-edge research in transformer insulation technology. The journal also bridges practical engineering solutions with theoretical advancements, making it a preferred venue for academics and industry practitioners alike.

Geographically, India and China emerge as prominent contributors to research on insulating transformer oils, driven by their substantial investments in energy infrastructure and policies that foster innovation. In India, government-backed institutions such as the Indian Institutes of Technology (IITs) and the Indian Institute of Science (IISc) play a pivotal role in advancing this field. Studies such as the investigation into thermal properties of graphene-silicone oil nanofluids [63] and the enhancement of ester-based insulating oils with nanocomposites [64] exemplify India’s commitment to applied research in this domain. These works are often carried out in collaboration with industry leaders like the Power Grid Corporation of India, underlining the importance of academic-industry partnerships in addressing national energy challenges [63–65].

Similarly, China has emerged as a prominent leader in transformer insulation research, supported by substantial government funding and policies aimed at advancing technological innovations in the energy sector. Research contributions include innovative approaches to modifying insulating oils for enhanced electrical and thermal performance, such as the development of KH550-TiO₂-enhanced natural ester oils [66] and investigations into nano TiO₂-modified transformer oils for extreme cold conditions [67]. These efforts reflect China’s emphasis on leveraging nanotechnology to create sustainable and efficient alternatives to traditional mineral oils, supported by policies encouraging academia-industry collaboration [59, 66–68].

The prioritization of transformer insulation research by both countries highlights their strategic focus on modernizing power systems to meet rising energy demands while addressing environmental sustainability. Additionally, these countries lead global efforts in nanotechnology research, which has direct applications in insulating fluids, further reinforcing their dominance in this area.

The multidisciplinary nature of this research field is underscored by its wide-reaching impact across journal quartiles, ranging from Q1 to Q3, reflecting the integration of diverse fields such as electrical engineering, material science, molecular physics, and environmental science. This underscores the need for cross-disciplinary collaboration to address the challenges of modern transformer insulation systems, particularly with the incorporation of nano-enhanced and bio-based insulating fluids. Such efforts are essential for meeting both performance and sustainability goals in the energy sector.

In this scientometric review, Cluster 1, categorized as “Density Functional Theory (DFT) Study,” has emerged as a dominant and influential research area. The primary reason for its prominence lies in the ability of DFT methods to provide molecular-level insights into gas adsorption and sensing mechanisms. These studies focus on gases like C₂H₂, H₂, CH₄, SO₂, and NO₂, which are key indicators for Dissolved Gas Analysis (DGA)—a critical diagnostic technique for transformer health monitoring [35, 36, 40, 43, 49]. By using computational approaches, these studies uncover detailed adsorption behaviors, providing predictive and cost-effective solutions compared to experimental methods.

The articles within Cluster 1 highlight the development of advanced nanomaterials to enhance gas detection sensitivity. For example, Pd-doped MoS₂ and Pd-doped HfSe₂ demonstrate the role of palladium doping in improving adsorption selectivity for fault gases like C₂H₂ and H₂ [36, 49]. Meanwhile, modifications to monolayers such as MoSe₂ and SnS₂ with sulfur defects explore strategies for selective detection of gases [35, 43]. Expanding beyond monolayers, CuO-modified boron nitride nanotubes (BNNTs) showcase their potential due to their high surface area and strong adsorption capability [40]. These innovations provide transformative improvements for DGA, offering practical solutions to enhance transformer fault diagnosis.

The practical impact of Cluster 1 is significant because transformer failures caused by undetected faults can lead to severe economic and operational consequences. DFT-based studies contribute to improving the sensitivity, reliability, and accuracy of gas detection methods, bridging the gap between theoretical research and real-world applications. By advancing materials like Pd-doped MoS₂ and CuO-BNNTs, these studies propose efficient, next-generation diagnostic systems that support better transformer maintenance and grid reliability [36, 40, 43].

Another reason for the dominance of this cluster is its recent publication trend, with an average year of 2020. These recent studies reflect ongoing advancements in transformer oil diagnostics and computational materials research. By incorporating cutting-edge technologies and novel materials, these articles are more likely to attract attention and citations from the research community, maintaining their high influence [35, 43, 49]. The alignment of these studies with current technological needs highlights their relevance in the energy sector.

Moreover, the interdisciplinary nature of DFT studies enhances their appeal to a broader audience. These studies connect key disciplines such as materials science, computational chemistry, and electrical engineering, which increases their visibility and applicability. Compared to older, more specialized clusters like Cluster 3 (average year: 2011), DFT studies remain at the forefront of transformer oil research by pushing the boundaries of innovation.

Last but not least, Cluster 1 stands out due to its methodological importance, practical relevance, and alignment with current research trends. The top-cited articles demonstrate the potential of DFT-based studies to revolutionize transformer fault diagnostics through advanced material design and computational precision. By addressing key challenges in DGA, these studies contribute to both academic advancements and industry applications, solidifying their influence in transformer insulation research [35, 36, 40, 43, 49].

The dominant research themes in transformer oil studies include dielectric strength (Cluster 0), alongside emerging trends such as Density Functional Theory (DFT) studies and power transformer advancements. Through document cluster analysis, Cluster 0 emerged as a major focus due to its relevance to improving insulation performance and overall reliability in power systems. The studies within this cluster focus on traditional but critical aspects, including the breakdown strength of transformer oils, the integration of nanotechnology, and the development of sustainable insulating fluids.

The analysis identified five highly cited articles [7, 41, 44, 47, 48] that contribute significantly to this research direction. These publications are categorized into four primary areas. First, dielectric strength studies [69–71] focus on enhancing the breakdown voltage of insulating oils, which remains central to improving the efficiency and safety of transformers. Dielectric strength is a fundamental property for insulating fluids, as it directly determines their ability to withstand high voltages without failure.

Second, the innovation of nanofluids has emerged as a promising trend. Studies such as [72–74] explore the role of nanoparticles—including TiO₂, ZnO, and hybrid nanomaterials—in enhancing the electrical and thermal performance of transformer oils. By dispersing nanoparticles in traditional oils, researchers have achieved notable improvements in breakdown voltage, thermal conductivity, and dissipation factor, paving the way for nano-enhanced insulating fluids. These advancements address the limitations of conventional oils while improving transformer efficiency and performance.

The third key focus within Cluster 0 is the shift toward natural esters and bio-based fluids as sustainable alternatives to mineral oil. Articles like [9, 14, 75] align with global sustainability efforts by investigating the use of plant-based and biodegradable insulating fluids. These studies highlight the environmental benefits of natural esters, such as reduced carbon footprint, biodegradability, and safety, while demonstrating comparable or superior dielectric properties to mineral oils.

Finally, comprehensive review studies [4, 7, 8, 11] play a critical role in this cluster. These highly cited reviews consolidate experimental findings, theoretical advancements, and emerging trends in transformer oil research. By synthesizing data across multiple studies, they provide clear insights into existing challenges and highlight future research directions, ensuring their continued relevance and impact.

In summary, Cluster 0 reflects a balance between traditional research themes and emerging innovations. The interconnected focus on dielectric strength, nanofluid enhancements, and sustainable insulating fluids highlights a comprehensive approach to advancing transformer technology. These studies have received significant short-term attention and demonstrate the critical need to improve reliability, efficiency, and environmental sustainability in transformer insulation systems.

The document burst analysis reveals a clear trajectory of emerging research topics, with older works being succeeded by newer, impactful studies. The article by [26] stands out as an early foundational work that gained widespread attention from 2013 to 2017. This study focused on enhancing transformer oil properties using Fe₃O₄ nanoparticles, demonstrating a significant improvement in breakdown voltage under power frequency and lightning impulse conditions. By aligning with the global trend of replacing mineral oils with eco-friendly alternatives, such as vegetable oils, it addressed pressing concerns about biodegradability, safety, and performance, thus cementing its relevance.

Following this foundational work, the research landscape evolved to encompass broader advancements. [42] emphasized the development of AI-based techniques for Dissolved Gas Analysis (DGA), a critical diagnostic tool for transformer health monitoring. This shift toward intelligent systems reflects the increasing demand for reliable and automated fault detection methods. Similarly, [41] highlighted the role of TiO₂ nanoparticles in enhancing transformer oil's dielectric strength, emphasizing nanotechnology's potential to improve breakdown voltage and stability, which are directly linked to system reliability and longevity. Complementing these advancements, [44] explored the interplay between nanoparticle conductivity, permittivity, and insulation performance, offering both experimental insights and theoretical frameworks for optimizing nanofluids.

Together, these articles showcase a cohesive evolution in transformer research, from focusing on performance improvement to incorporating resilience and advanced diagnostics. This convergence of diagnostics, materials science, and intelligent systems not only addresses critical challenges but also provides a roadmap for integrating cutting-edge technologies into transformer applications, ensuring reliability and sustainability in the energy sector

This study acknowledges several limitations, including the exclusive use of the Web of Science (WOS) database, which may have led to the omission of relevant works indexed in Scopus or Google Scholar. Additionally, reliance on specific keywords and the inclusion of only English-language publications may have introduced selection bias. To strengthen future research, broader database coverage and manual screening are recommended. Furthermore, based on bibliometric trends and cluster analysis, future directions include promoting international collaborations, especially among leading countries such as China and India, and encouraging interdisciplinary efforts that connect material science, nanotechnology, and smart grid systems. Long-term performance evaluations of green insulating oils, standardized testing methods, and the expanded use of scientometric tools are also vital to enhance data reliability and support strategic research planning.

Conceptualization, Data Analysis and Manuscript Writing - original draft: Dzhahirul Zamri. Software, Research Methodology and Writing - review and editing: Mohd Iqbal Mohd Noor. Conceptualization and Formal Analysis: N. H. Nik Ali. Conceptualization and Supervision: M. H. Mamat. Manuscript Review and Intellectual Rigor: Noor Khairin Mohd. Manuscript Review and Intellectual Rigor: Mohammad. A. Hamdan.