A Bibliometric Study on Integrated Solar Combined Cycles (ISCC), Trends and Future Based on Data Analytics Tools

: In this paper, a bibliometric analysis was performed in order to analyze the state of the art and publication trends on the topic of ISCC (Integrated Solar Combined Cycles) for the period covering 1990 to July 2020. The Web of Science (WOS) database was consulted, and 1277 publications from 3157 di ﬀ erent authors and 1102 di ﬀ erent institutions, distributed among 78 countries, were retrieved as the corpus of the study. The VOSViewer software tool was used for the post-processing of the WOS corpus, and for the network data mapping. Multiple bibliometric indicators, such as the number of citations, keyword occurrences, the authors’ a ﬃ liations, and the authors, among others, were analysed in this paper in order to ﬁnd the main research trends on the ISCC topic. The analysis performed in this paper concluded that the main publication source for ISCC research was Energy Conversion and Management, in terms of the total number of publications (158), but Solar Energy had the highest number of citations on the ISCC topic (4438). It was also found that China was the most productive country in terms of ISCC publications (241), and the Chinese Academy of Sciences was the most productive institution (52). Nevertheless, the author with the most publications on ISCC was I. Dincer, from Ontario Tech University (24). Based on publication keywords, a series of recommendations for future developments in the ISCC topic were derived, as well as the ways in which those ideas are connected to the global state of solar energy research.


Introduction
Human development indicators and societal developments are tightly bounded to extensive energy consumption [1]. That trend is particularly sharpened for emerging economies, where the increase of middling social classes and the access to new technologies is increasing the energy demand. We as a society demand more commodities, and we do not want to give up to our high living standards [2]. On the contrary, keeping the traditional energy scenario that still relies on thermal energy conversion, and that is dependent on fossil fuels (coal, gas and oil) and nuclear energy, has shown a clear impact on global warming and climate change. Therefore, a change of the energy generation paradigm is required if we want to meet society, users and environmental demands. In that scenario, several agreements have been signed over the last few decades regarding climate protection and the reduction of greenhouse gases and pollutant emissions, from the earlier Kyoto protocol in 1997 [3], the Paris Climate Agreement in 2015 [4], and more recently the COP25 Climate Change Conference in 2019 [5] and the United Nations Sustainable Development Goals declaration [6]. Under those 17 UN Goals, three of them have a direct impact on our energy model. Those are; 'Affordable and Clean Energy' (goal 7); 'Industry, Innovation and Infrastructure' (goal 9) and 'Climate Action' (goal 13). On the same line, several countries are making a big effort in the energy transition with the support of expertise, in order to discard, reject or support their assumptions based on similar research works. Usually, this stage is commonly known as the literature review, and it is the cornerstone of any research activity, from Masters theses, to PhD theses, to research papers, and it is even crucial for successful applications for project proposals and funding schemes. Despite the importance of that stage, the depth of that analysis depends in practice on previous expertise and personal experience, since it is typically addressed as a human-based activity that requires many years of experience. Nowadays, that exercise becomes even harder to attain, due to the increasing number of research papers being published, the appearance of new journals and platforms, and the infamous motto "publish or perish". Fortunately, the developments on Big data, data mining and data analytics that have became popular in social networks analysis and in behavioural sciences [38] can be also applied to Energy research and other technical sciences. Big data treatment through nodes and networks analysis has great potential for engineering research applications, and in particular to the topic of ISCC, since it allows researchers and scholars to understand trends and topics related to ISCC technologies, nurture future collaborations among different research groups and researchers, and to increase their awareness on the topic's importance. Last but not least, it establishes data analytics as one of the core activities of the scientific method by providing researchers a powerful, and yet unfamiliar, tool. Besides this, it is proposed as a methodology to thoroughly address a research topic that was rather manual, time consuming and biased, up until this time. The application of bibliometric studies for renewable energy topics is quite recent; some examples can be found, such as the use of community detection tools for scientific collaboration analysis [39], or keywords trend evolution for the study of interactions between the economy, energy and the environment [40]. Bibliometric studies have also shown their potential as a tool for the analysis of the research impact of a country [41], or to analyse the global transition to low-carbon electricity [42]. In particular, and related to solar energy, very few bibliometric studies could be found, with most of them having been published recently [43][44][45]. This paper has been organized according to the following structure; firstly, energy and technology contexts are presented. After that, the works' relevance is discussed, together with the data mining source and the research method employed. Later, corpus data is analysed using VOSviewer software, which is a tool based on the use of network data for the creation of maps, and for visualizing and exploring them [46]. Finally, conclusions are drawn, and recommendations are compared to another hot topic in the CSP field.

Data Source
The data corpus used in this work was retrieved from the Web of Science (WOS) Core Collection using the search questions (keywords) "Integrated Solar Combined Cycle" and "Solar Combined Cycle" for the time period between 1990 and 2020 (only papers published before 14th of July were accounted for in the 2020 analytics). Both works published as journal articles and conference proceedings that can be retrieved from the WOS have been considered for the analysis. Figure 1 shows the distribution of the retrieved publications according to the WOS thematic areas, with the total number of publications for each category indicated between brackets. As it can be observed, the main topic area on ISCC publications refers to Energy Fuels (1161), followed by Thermodynamics (424), Mechanics (255), Engineering Mechanical (234), Green Sustainable Science Technology (228), Engineering Chemical (136), Environmental Sciences (81), Environmental Engineering (61), Chemistry Physical (47) and Electrochemistry (47). It must be pointed out that the number of publications indicated between brackets is not cumulative, since the same work might be classified under different categories in the WOS. Further refinement was applied in order to exclude the retrieved references that were related to different Integrated Solar Technologies, such as PV-only research. After that refinement, the total number of publications retrieved was 1277, which constitutes the data corpus for this study. Further insight into this search is shown in Figure 2, where it can be observed that 75.7% of the retrieved documents were journal articles, 18.9% were proceeding papers, and less than 5% were review papers. Editorial material, early access, corrections and notes accounted for less than 1% altogether.   It must be pointed out that the number of publications indicated between brackets is not cumulative, since the same work might be classified under different categories in the WOS. Further refinement was applied in order to exclude the retrieved references that were related to different Integrated Solar Technologies, such as PV-only research. After that refinement, the total number of publications retrieved was 1277, which constitutes the data corpus for this study. Further insight into this search is shown in Figure 2, where it can be observed that 75.7% of the retrieved documents were journal articles, 18.9% were proceeding papers, and less than 5% were review papers. Editorial material, early access, corrections and notes accounted for less than 1% altogether.   As it can be observed, the number of publications related to ISCC has significantly increased since 2011, and especially since 2016. In fact, a baseline of 100 publications per year was imposed from 2014 onwards, which gives an idea about the interest in that technology. In fact, this trend has already been kept in 2020, in which 100 ISCC publications were reported before 14 th July. A similar trend can be found in the total number of citations, with more than 5500 in 2019. Both the increasing number of ISCC publications and citations agree with the trend observed in CSP's installed capacity, which confirms the interest and deployment of the technology [47]. Figure 4 shows the summary report generated by the ISCC search on the WOS. As it can be observed, for the time period of the analysis, ISCC publications were cited 28,201 times in total in 17,768 different items indexed within the Web of Science Core Collection. Removing self-citation, that number was reduced to 24,934 citations in 17,010 different items. Dividing the sum of the times cited by the total number of publications results in an average of 22.08 citations per item, as can be observed in Figure 4.

Analysis Methodology
For the data analytics, the data corpus retrieved from the WOS search was exported, including the full record and cited references to be analysed using the graphics analysis software. Save to other formats, the full record and cited references, and all of the data from those papers were exported, and were analysed using graphics analysis software. The VOSViewer [48] software tool was chosen, since it is a free software for constructing and visualizing bibliometric networks, and it also offers a text mining functionality that can be used to construct and visualize the co-occurrence networks of the relevant terms extracted from a body of scientific literature. These networks may for instance include journals, researchers, or individual publications, and they can be constructed based on citations, bibliographic coupling, co-citations, or co-authorship relations. A diagram of the methodology followed in this work can be found in Figure 5. As it can be seen, the first stage was comprised of As it can be observed, the number of publications related to ISCC has significantly increased since 2011, and especially since 2016. In fact, a baseline of 100 publications per year was imposed from 2014 onwards, which gives an idea about the interest in that technology. In fact, this trend has already been kept in 2020, in which 100 ISCC publications were reported before 14th July. A similar trend can be found in the total number of citations, with more than 5500 in 2019. Both the increasing number of ISCC publications and citations agree with the trend observed in CSP's installed capacity, which confirms the interest and deployment of the technology [47]. Figure 4 shows the summary report generated by the ISCC search on the WOS. As it can be observed, for the time period of the analysis, ISCC publications were cited 28,201 times in total in 17,768 different items indexed within the Web of Science Core Collection. Removing self-citation, that number was reduced to 24,934 citations in 17,010 different items. Dividing the sum of the times cited by the total number of publications results in an average of 22.08 citations per item, as can be observed in Figure 4.

Analysis Methodology
For the data analytics, the data corpus retrieved from the WOS search was exported, including the full record and cited references to be analysed using the graphics analysis software. Save to other formats, the full record and cited references, and all of the data from those papers were exported, and were analysed using graphics analysis software. The VOSViewer [48] software tool was chosen, since it is a free software for constructing and visualizing bibliometric networks, and it also offers a text mining functionality that can be used to construct and visualize the co-occurrence networks of the relevant terms extracted from a body of scientific literature. These networks may for instance include journals, researchers, or individual publications, and they can be constructed based on citations, bibliographic coupling, co-citations, or co-authorship relations. A diagram of the methodology followed in this work can be found in Figure 5. As it can be seen, the first stage was comprised of filtering the data using the research question, time period and thematic areas applied to the WOS Core Collection. The result of that filtering stage is the corpus data of the ISCC topic. That corpus data was analysed applying different bibliometric indicators and using the VOSviewer tool for networking mapping.
Sustainability 2020, 12, x FOR PEER REVIEW 7 of 30 filtering the data using the research question, time period and thematic areas applied to the WOS Core Collection. The result of that filtering stage is the corpus data of the ISCC topic. That corpus data was analysed applying different bibliometric indicators and using the VOSviewer tool for networking mapping.

Results
Several indicators were considered for the bibliometric network study, including co-authorship analysis, co-occurrence analysis and citation analysis, among others. Each study can be appointed by using different units of analysis, such as authors, sources, organizations, countries, documents or keywords.

Co-Authorship Analysis
For the analysis, papers with a large number of authors were ignored by applying a threshold of 25 maximum authors for a single publication. In addition, full counting criteria were applied in the analysis, which means that if an author co-authors a document with, for example, 10 other authors, each of the 10 co-authorship links has a weight of 1 instead of a 1/10. That has an impact in bibliometric network construction, since links' thickness and node sizes depend on that counting. Applying that criterion, a total number of 3568 different authors have ever published at least one paper related to ISCC topic, as can be found in Table 1. Indeed, this table shows the number of different authors fulfilling a minimum publication criterion applied to the ISCC corpus data.

Results
Several indicators were considered for the bibliometric network study, including co-authorship analysis, co-occurrence analysis and citation analysis, among others. Each study can be appointed by using different units of analysis, such as authors, sources, organizations, countries, documents or keywords.

Co-Authorship Analysis
For the analysis, papers with a large number of authors were ignored by applying a threshold of 25 maximum authors for a single publication. In addition, full counting criteria were applied in the analysis, which means that if an author co-authors a document with, for example, 10 other authors, each of the 10 co-authorship links has a weight of 1 instead of a 1/10. That has an impact in bibliometric network construction, since links' thickness and node sizes depend on that counting. Applying that criterion, a total number of 3568 different authors have ever published at least one paper related to ISCC topic, as can be found in Table 1. Indeed, this table shows the number of different authors fulfilling a minimum publication criterion applied to the ISCC corpus data. As it can be deduced, a total number of 3568 authors have one publication related to the ISCC topic, and this number is reduced down to 576 for authors with two publications, and to 71 authors with five publications. There were only eight authors with 10 publications published on the ISCC topic since 1990. Those numbers do not consider any other requirement regarding the minimum number of citations of an author. In order to account for the relevance of their works, a minimum criterion of at least 10 citations of an author' documents (according to the WOS) was considered. As it can be observed, the impact of that filtering is only noticeable for those authors holding just one publication, which reduced author number by half.
For co-authorship networking mapping, a minimum number of two publications per author with 10 citations was chosen, which gives a significant number of nodes meeting the thresholds (460). For each of the 460 authors, the total strength of the co-authorship links with other authors was calculated. Some of the authors in the network are not connected to each other in the graph, since they may have cited different authors that did not meet the minimum number of publications and citations criteria. The largest set of connected items consists of 241 items that were finally represented in the network map on Figure 6. This graphical information is useful for ISCC researchers and scholars, since it provides a clear idea about who the main authors publishing in this topic are, and the connections (citations) among them. Furthermore, the size of each node is related to the number of publications. Table A1 from Appendix A contains detailed information about the clusters shown in Figure 6.
Sustainability 2020, 12, x FOR PEER REVIEW 8 of 30 As it can be deduced, a total number of 3568 authors have one publication related to the ISCC topic, and this number is reduced down to 576 for authors with two publications, and to 71 authors with five publications. There were only eight authors with 10 publications published on the ISCC topic since 1990. Those numbers do not consider any other requirement regarding the minimum number of citations of an author. In order to account for the relevance of their works, a minimum criterion of at least 10 citations of an author' documents (according to the WOS) was considered. As it can be observed, the impact of that filtering is only noticeable for those authors holding just one publication, which reduced author number by half.
For co-authorship networking mapping, a minimum number of two publications per author with 10 citations was chosen, which gives a significant number of nodes meeting the thresholds (460). For each of the 460 authors, the total strength of the co-authorship links with other authors was calculated. Some of the authors in the network are not connected to each other in the graph, since they may have cited different authors that did not meet the minimum number of publications and citations criteria. The largest set of connected items consists of 241 items that were finally represented in the network map on Figure 6. This graphical information is useful for ISCC researchers and scholars, since it provides a clear idea about who the main authors publishing in this topic are, and the connections (citations) among them. Furthermore, the size of each node is related to the number of publications. Table A1 from Annex A contains detailed information about the clusters shown in Figure 6. As it can be observed in Figure 6, some western authors appeared to be duplicated since they signed their documents using different initials, which is considered as two different authors by the software. The software allowed us to get rid of a researcher's first name and replace it by its initial, which would lead to an incorrect analysis, mainly with regard to Asian researchers. In order to verify the quality of the results, the generated table of authors was analysed, and the duplicate cases were revised, such as the ones found corresponding to renowned authors I. Dincer, who also published as Ibrahim Dincer; A. Steinfeld, who published as A Steinfeld; or D. Yogi Goswami, who also appeared as Dy Goswami. Applying that filter to the co-authorship network, a version of Table 2 including the As it can be observed in Figure 6, some western authors appeared to be duplicated since they signed their documents using different initials, which is considered as two different authors by the software. The software allowed us to get rid of a researcher's first name and replace it by its initial, which would lead to an incorrect analysis, mainly with regard to Asian researchers. In order to verify the quality of the results, the generated table of authors was analysed, and the duplicate cases were revised, such as the ones found corresponding to renowned authors I. Dincer, who also published as Ibrahim Dincer; A. Steinfeld, who published as A Steinfeld; or D. Yogi Goswami, who also appeared as Dy Goswami. Applying that filter to the co-authorship network, a version of Table 2 including the top authors in ISCC was created. Another interesting parameter that can be deduced from Table 2 is the average number of citations each ISCC paper from those renowned authors received. As it can be seen, the highest ratio corresponds to D. Goswami and A. Steinfeld, with more than 50 citations for each ISCC paper, while the lowest ratio for the top 10 author list is 6.7 for B. Laumert at The Royal Institute of Technology, KTH. Regarding the co-authorship analysis by organizations (authors' institutions), there were 1102 different institutions that had published at least one paper on the ISCC topic. However, only 85 of them published at least 5 publications, as can be observed from Table 3. A minimum number of 5 publications (and 10 citations) were chosen for the bibliometric analysis, which resulted in 85 nodes. Some of them were not connected in the network, since they did not meet the publication criteria or did not cite any of the other filtered groups. Therefore, the largest set of connected items (69) was represented instead in Figure 7. Based on that figure, the different collaborations among the institutions can be observed. This kind of analysis is relevant because it allows scholars to identify the main institutions publishing on ISCC (nodes size) and linking relations (citations) among them. For example, Universidad Politécnica de Madrid had joint publications in ISCC together with Universidad Carlos III and UNED. Meanwhile, North China Electric Power University had joint publications in ISCC together with Hunan University, the Chinese Academy of Sciences, Tianjin University, Nanyang Technology University, Huazhong University, the University of Pennsylvania and the Technical University of Denmark. Table A2 from Appendix A contains the detailed information about the clusters shown in Figure 7.
of Pennsylvania and the Technical University of Denmark. Table A2 from Annex A contains the detailed information about the clusters shown in Figure 7. Based on that analysis, it can be observed, in Table 4, that the Chinese Academy of Sciences is the main institution regarding ISCC publications, with 52 contributions in total (and another 28 under the University of Chinese Academy of Sciences affiliation). As it can be observed, five out of the first 10 institutions publishing about ISCC are from China. Another interesting parameter that can be deduced from Table 4 is the ratio between the number of total citations per institutions and the number of published papers. As it can be observed, institutions with fewer publications in Table 4 (National Technical University of Athens and Xi'an Jiao Tong University) exhibit the higher ratio, with almost 40 cites per document; on the contrary, the Chinese Academy of Sciences-which was the institution with more publications (52) and more citations (736)-got the lowest ratio, with around 14 cites per document. Those numbers were also confirmed through a co-authorship analysis based on the different countries. As it can be observed from Table 5, the corpus publication on ISCC came from 78 different countries. That number was reduced to 43 countries with five publications on that topic, and to 31 countries with ten publications. There are seven countries with at least 50 publications, and only four countries published 100 or more papers on the ISCC topic. Based on that analysis, it can be observed, in Table 4, that the Chinese Academy of Sciences is the main institution regarding ISCC publications, with 52 contributions in total (and another 28 under the University of Chinese Academy of Sciences affiliation). As it can be observed, five out of the first 10 institutions publishing about ISCC are from China. Another interesting parameter that can be deduced from Table 4 is the ratio between the number of total citations per institutions and the number of published papers. As it can be observed, institutions with fewer publications in Table 4 (National Technical University of Athens and Xi'an Jiao Tong University) exhibit the higher ratio, with almost 40 cites per document; on the contrary, the Chinese Academy of Sciences-which was the institution with more publications (52) and more citations (736)-got the lowest ratio, with around 14 cites per document. Those numbers were also confirmed through a co-authorship analysis based on the different countries. As it can be observed from Table 5, the corpus publication on ISCC came from 78 different countries. That number was reduced to 43 countries with five publications on that topic, and to 31 countries with ten publications. There are seven countries with at least 50 publications, and only four countries published 100 or more papers on the ISCC topic. As it can be observed, when the criterion about having at least 10 citations by country is introduced, the number of countries meeting the requirement reduces to 68, and if the number of citations is increased to 100, the number is reduced to 38. As the number of publications increases, (above 10) the citation criteria show no effect. For networking mapping representation, a minimum number of five publications with 10 citations by country was chosen, which lead to 43 nodes. However, the largest set of connected items consists of 42 items (Morocco is the only country not meeting the criterion) that were chosen for the graphical representation in Figure 8. This kind of analysis is relevant because it allows an understanding of the main connections (citations) between different countries, and the identification of common collaborations among research institutions. Table A3 from Appendix A contains detailed information about the clusters shown in Figure 8.  As it can be observed, when the criterion about having at least 10 citations by country is introduced, the number of countries meeting the requirement reduces to 68, and if the number of citations is increased to 100, the number is reduced to 38. As the number of publications increases, (above 10) the citation criteria show no effect. For networking mapping representation, a minimum number of five publications with 10 citations by country was chosen, which lead to 43 nodes. However, the largest set of connected items consists of 42 items (Morocco is the only country not meeting the criterion) that were chosen for the graphical representation in Figure 8. This kind of analysis is relevant because it allows an understanding of the main connections (citations) between different countries, and the identification of common collaborations among research institutions. Table A3 from Annex A contains detailed information about the clusters shown in Figure 8.   Table 6 shows the top 10 countries publishing on the ISCC topic; as it can be observed, China is the main contributor in this topic, with 241 papers and 4108 citations in the last 30 years, followed by the United States (202), Spain (115) and Italy (102). Another interesting parameter that can be inferred from Table 6 is the ratio between the total number of citations and the number of documents. At one end, Germany is the 6th country in terms of ISCC research, with 83 publications (most of them from the DLR German Aerospace Centre). However, each of those publications was cited almost 30 times on average. On the contrary, Indian publications on ISCC were cited 12 times on average.

Co-Occurrence Analysis
A co-occurrence analysis was performed based on all of the keywords, with a full counting method. Considering all of the appearing keywords, Table 7 shows the number of occurrences of a keyword. As it can be observed from the table, there are at least 4534 different keywords that appear at least once in the data corpus (1277 total publications). This results in an average number of 3.55 keywords per publication. In order to consider only the more relevant keywords, a minimum number of occurrences was considered. In doing, so it can be concluded that 394 different keywords appeared in at least in five different papers, 71 appeared in at least 25 papers, and six keywords appeared in 100 publications from the corpus. For the network graph representation, the 100 most common keywords were chosen, as can be seen in Figure 9. As it can be observed, most of the keywords are related to modelling topics based on keywords like 'optimization', 'design', 'exergy analysis', 'thermodynamic analysis', 'multiobjective optimization', 'model', or 'simulation'. Apart from those common keywords, some others, such as thermoeconomic analysis, performance analysis, exoergonomic analysis or parametric analysis also appeared in Figure 9. Despite the abundance of keywords related to modelling and simulation, another series of keywords related to different technologies such as direct steam generation, CO 2 capture, hybrid plants, biomass or desalination also appear which gives an idea of combined applications of ISCC with other technologies. This kind of analysis and graphical representation is relevant because it allows us an understanding of the connection (co-ocurrence of keywords) between the topics of ISCC research. As it can be observed, a clear interest for low and medium temperature applications of ISCC is also evident from the occurrence of keywords such as 'parabolic trough', 'organic Rankine cycles', 'combined heat and power' and 'cooling and refrigeration'. Table A4 in Appendix A contains detailed information about the clusters shown in Figure 9.
Sustainability 2020, 12, x FOR PEER REVIEW 13 of 30 medium temperature applications of ISCC is also evident from the occurrence of keywords such as 'parabolic trough', 'organic Rankine cycles', 'combined heat and power' and 'cooling and refrigeration'. Table A4 in Annex A contains detailed information about the clusters shown in Figure  9.  Table 8 summarizes the list of the top 10 keywords and their frequency of occurrence. As was mentioned above, most keywords are related to simulations, such as 'optimization' (249 times), 'performance' (237), 'design' (175), 'exergy analysis '(98) and 'thermodynamics analysis' (94). We can also observe the growing interest in low grade heat energy recovery at ISCC by the high occurrence of the 'Organic Rankine Cycle' keyword (94).

Citation Analysis
Citation bibliometric mapping can be performed through different units of analysis: documents, sources, authors, organizations and countries. Based on the number of documents, it is clear that there are 1277 publications with a minimum number of citations of 0 (the total corpus); meanwhile, 833 publications were cited five times, 392 publications were cited 20 times, 152 publications were cited 50 times, and 52 documents accumulated 100 citations each, as can be observed in Table 9.  Table 8 summarizes the list of the top 10 keywords and their frequency of occurrence. As was mentioned above, most keywords are related to simulations, such as 'optimization' (249 times), 'performance' (237), 'design' (175), 'exergy analysis '(98) and 'thermodynamics analysis' (94). We can also observe the growing interest in low grade heat energy recovery at ISCC by the high occurrence of the 'Organic Rankine Cycle' keyword (94).

Citation Analysis
Citation bibliometric mapping can be performed through different units of analysis: documents, sources, authors, organizations and countries. Based on the number of documents, it is clear that there are 1277 publications with a minimum number of citations of 0 (the total corpus); meanwhile, 833 publications were cited five times, 392 publications were cited 20 times, 152 publications were cited 50 times, and 52 documents accumulated 100 citations each, as can be observed in Table 9. The publications with 50 or more citations were selected for the networking mapping; according to Table 9, there are 152 publications meeting that requirement. Furthermore, the tool suggests the representation of only the largest set of connected items, which numbers 107. In other words, 107 publications with 50 citations (or more) each out of the 152 publications cite at least to one of the other works from that list, so that a link between two nodes will exist, as can be observed in Figure 10. This kind of analysis and graphical representation is relevant because it allows us an understanding of how the most relevant publications (in terms of the total number of citations) are connected to each other. Table A5 in Appendix A contains detailed information about the clusters shown in Figure 10.  Table 9. Citation analysis. The publications with 50 or more citations were selected for the networking mapping; according to Table 9, there are 152 publications meeting that requirement. Furthermore, the tool suggests the representation of only the largest set of connected items, which numbers 107. In other words, 107 publications with 50 citations (or more) each out of the 152 publications cite at least to one of the other works from that list, so that a link between two nodes will exist, as can be observed in Figure 10. This kind of analysis and graphical representation is relevant because it allows us an understanding of how the most relevant publications (in terms of the total number of citations) are connected to each other. Table A5 in Annex A contains detailed information about the clusters shown in Figure 10. As can be observed in Figure 10, there is a high interconnection among the nodes, which means that the most cited articles cited each other. Furthermore, each node contains full reference information and the hyperlink to the internet website hosting the article. The latter is very practical since one can quickly access to the full publication online. From the analysis, it can be concluded that the most cited articles are Mills (2004) [49], with 472 citations, and Behar (2013), with 335 [50]. The detailed list of the most cited articles retrieved from the WOS on the ISCC topic are gathered in Table  10.  As can be observed in Figure 10, there is a high interconnection among the nodes, which means that the most cited articles cited each other. Furthermore, each node contains full reference information and the hyperlink to the internet website hosting the article. The latter is very practical since one can quickly access to the full publication online. From the analysis, it can be concluded that the most cited articles are Mills (2004) [49], with 472 citations, and Behar (2013), with 335 [50]. The detailed list of the most cited articles retrieved from the WOS on the ISCC topic are gathered in Table 10. Based on the sources-which means journals and indexed conference proceedings at the WOS-a total number of 129 different sources were retrieved for ISCC related publications. Figure 11 shows the network mapping for all of the sources with five or more publications in common (41). Nevertheless, the largest set of connected items that is represented in the network numbers 39. That map also contains information about the relevance of each node (publications of each source) and the connection between the journals. This refers to which journals were cited by each node; for example, publications from the Sustainability node cited publications from Renewable Energy, Energy, Applied Energy, Energy Conversion and Management, Sustainable Energy Technologies and Energy Technology, or vice-versa, which is translated as a link between those nodes. This kind of analysis and graphical representation is relevant because it allows us an understanding of which sources are the most cited, and how those sources are connected (citations) amongst themselves. Table A6 in Appendix A contains detailed information about the clusters shown in Figure 11. Based on the sources-which means journals and indexed conference proceedings at the WOSa total number of 129 different sources were retrieved for ISCC related publications. Figure 11 shows the network mapping for all of the sources with five or more publications in common (41). Nevertheless, the largest set of connected items that is represented in the network numbers 39. That map also contains information about the relevance of each node (publications of each source) and the connection between the journals. This refers to which journals were cited by each node; for example, publications from the Sustainability node cited publications from Renewable Energy, Energy, Applied Energy, Energy Conversion and Management, Sustainable Energy Technologies and Energy Technology, or vice-versa, which is translated as a link between those nodes. This kind of analysis and graphical representation is relevant because it allows us an understanding of which sources are the most cited, and how those sources are connected (citations) amongst themselves. Table A6 in Annex A contains detailed information about the clusters shown in Figure 11. Based on the information provided in Figure 11, summary Table 11 is provided in order to analyze the most frequent sources for publications on the topic of ISCC. The information was sorted Based on the information provided in Figure 11, summary Table 11 is provided in order to analyze the most frequent sources for publications on the topic of ISCC. The information was sorted in terms of the total number of documents. As it can be observed, Energy Conversion and Management was the preferred platform for ISCC publications (158 documents) whereas Solar Energy was the source that received more citations (4438 citations). In order to normalize that information, the ratio of citations per document was introduced. As can be observed, papers published in Renewable and Sustainable Energy Reviews received the highest number of citations, with almost 54 cites per published document on average, followed by Solar Energy, with almost 44. On the other side, 'Energies' showed the lowest ratio, where each of their ISCC related publications (33 documents) received an average of four citations.

Discussion
The results presented in Section 3 (Results) and the tables gathered in Appendix A are relevant for solar energy scholars because some trends and research topics can be derived from them. Based on the detailed keyword information presented in Table A4, it can be deduced that most publications addressed modelling and performance optimization, with 857 keyword appearances. In particular, medium and low temperature studies were performed, as is indicated by the 59 keyword appearances for 'parabolic trough studies' and 246 keyword appearances for 'Organic Rankine Cycles' (ORC). Furthermore, a great interest in ISCC applications for combined heat and power generation as related keywords appear in 217 publications. It is also relevant that thermal energy storage studies were applied to ISCC topics, with 165 publications including related TES keywords. Hydrogen production was also related to ISCC publications, with 99 keyword appearances. To a lesser extent, 'life-cycle-analysis' (LCA) was considered for ISCC studies, with 82 keyword appearances. The less studied topics related to ISCC were 'desalination' (26 keyword appearances), 'phase-change material energy storage' (24 keyword appearances), 'CO 2 capture' (22 keyword appearances) and 'hybrid plants' (20 keyword appearances).
In this section, the ISCC topic is also compared to another hot energy topic: supercritical CO 2 cycles (sCO 2 ). As it was referred to in the introduction, sCO 2 cycles for electricity production are one of the hottest research topics, with a clear interest for CSP application. Furthermore, both technologies are highly efficient and advanced solutions for electricity generation, both are of interest for CSP applications, and, in some cases, they were studied together. For this reason, the bibliometric study presented in this work was compared to recent bibliometric study for sCO 2 [31].
For comparison purposes, bibliometric indicators such as the number of publications, the total number of citations, authors, institutions, countries, and the most relevant publications were compared and analyzed for both technologies, and are gathered in Table 12. The bibliometric parameters and indicators were normalized due to both studies covering a different period of time; for the case of this work, the ISCC publications were collected from 1990 to July 2020, while for the sCO 2 bibliometric analysis, the timespan covered 2000 to 2019. As it can be observed in the table, both technologies are hot topics for Energy research, with 1277 ISCC publications and 724 for sCO 2 . Furthermore, those topics have grabbed the attention of a large number of researchers and institutions from all over the world. In general, it could be said that the ISCC topic involved double the number of researchers and institutions compared to the sCO 2 topic. However, if normalized indicators are compared, similar numbers can be observed, such as, for example, the case of average number of publications per year, which was 42.6 for the ISCC topic and to 38.1 for sCO 2 . For example, comparing the ratio between the total number of publications and the number of institutions, the average publishing ratio for sCO 2 (1.33) is higher than that for ISCC (1.16). Comparing the total number of citations, the differences between the ISCC and sCO 2 topics become more evident; for example, the number of citations per year for ISCC papers reaches 940, while for sCO 2 it is 511. One of the reasons behind that trend could be the fact that the research interest for sCO 2 cycles appeared later than for ISCC due to the technical difficulties discussed for supercritical CO 2 compared to the mature technology of Combined Cycles, as was discussed at the introduction. That can be confirmed from both of the publishing evolutions shown in Table 13. As it can be observed, in the year 2010, there were 30 publications on ISCC, but only 11 publications for sCO 2 . Despite the rapid research deployment of sCO 2 in recent years leading to 131 publications by 2019, still more papers were reported for the ISCC topic (174), which explains the lower citation ratio.
Comparing the most relevant countries in terms of publishing, it can be observed in Table 14 that, for both the ISCC and sCO 2 topics, China and United States are the main publishing countries, while Spain is the third highest publishing country for ISCC, as is South Korea for sCO 2 . It can also be observed in Table 14 that the average citation ratio (total number of citations divided by the number of publications) for those countries for the ISCC topic (around 22 citations per publication) is higher than for the sCO 2 topic (around 15 citations per publication). Despite the high research output from those countries, some differences were found regarding the main publishing institutions. For the ISCC topic, two Chinese organisations and one Iranian organization were the most productive, while for the sCO 2 topic, institutions from China, South Korea and United States were relevant. Regarding the most relevant authors, it is interesting that authors in sCO 2 topics had more publications than the authors retrieved for the ISCC topic; however, the citation ratio (the number of citations divided by the total number of publications from an author) was higher for ISCC topics. The relevance of the conference proceedings for sCO 2 topics can also be observed, where the Proceedings of the ASME Turbo Expo were the first publishing platform in terms of documents (113), while the 'Energy' journal was in second place for both ISCC and sCO 2 . Table 15 compares the most cited publications for both the ISCC and sCO 2 topics. As it can be observed, most of the relevant publications for ISCC had more total citations than for the sCO 2 topic, but the ratio of citations/year was half of the ratio observed for the sCO 2 topic. The publishing sources also differed, with the exception of Renewable and Sustainable Energy Reviews. Finally, it can be observed in Table 16 that the most frequent keywords for the ISCC topic were quite general ('optimization', 'performance', 'energy'), while those for sCO 2 were the name of the topic itself ('supercritical carbon dioxide', 'Brayton cycle', 'supercritical CO 2 Brayton cycle'). It was also observed in the detailed keywords information provided in Table A4 that the most common keywords for the ISCC topic were related to modelling and simulations.

Conclusions
In recent years, the growing interest in higher conversion efficiencies for CSP applications has led to an increasing number of papers covering the topic of Integrated Solar Combined Cycle technologies. In particular, ISCC interest is based on its ability to increase the contribution of renewable energy sources into the global energy mix at a very high plant efficiency, whether in a pure solar configuration or in hybrid arrangements. Based on the presented bibliometric study, the following conclusions can be summarized:

•
There is a growing interest in ISCC topics, as can be observed from the increasing number of publications and citations. This trend sharpened in 2011.

•
The most productive countries in terms of the number of publications were China (241) Funding: This research received no external funding.

Conflicts of Interest:
The author declare no conflict of interest.
Appendix A Table A1. Detailed cluster information for co-authorship analysis by the authors (detailed information for Figure 6).    Table A3. Detailed cluster information for co-authorship analysis by countries (detailed information for Figure 8).  Table A4. Detailed cluster information for the co-occurrence analysis of keywords (detailed information for Figure 9).  Table A5. Detailed cluster information for the most cited publications (detailed information for Figure 10).  Table A6. Detailed cluster information for the most common sources (detailed information for Figure 11).