Visualizing Sustainable Supply Chain Management: A Systematic Scientometric Review

: Sustainable supply chain management (SSCM) has been attracting extensive attention from both practitioners and scholars. The main objective of this paper is to visualize and conduct a systematic scientometric review on 9151 articles and reviews published from 2007 to 2021. Research techniques of co-author analysis, co-word analysis, and co-citation analysis are applied to reveal the social structure, conceptual structure, and intellectual structure of the SSCM ﬁeld, identify main concepts and research hotspots, and illuminate major specialties and emerging trends. The results of this work show that: (1) the top ﬁve most productive scholars are Joseph Sarkis, Kannan Govindan, Minglang Tseng, Angappa Gunasekaran, and Charbel Jose Chiappetta Jabbour. The top ﬁve most productive institutions are Hong Kong Polytech University, Islamic Azad University, University of Southern Denmark, Dalian University of Technology, and University of Tehran. (2) The main concepts include sustainable supply chain management, green supply chain management, circular economy, corporate social responsibility, and reverse logistics. The research hotspots of the SSCM ﬁeld, currently, are game theory and circular economy related topics. (3) The leading researchers and inﬂuential journals are also identiﬁed. The emerging trends include sustainable supplier selection, circular economy, cap-and-trade regulation, blockchain technology, big data analytics, the COVID-19 pandemic, and the best-worst method and logistics performance. Finally, limitations and future researches are discussed. We expect this paper will show a big picture of the SSCM ﬁeld for researchers as well as practitioners. supply chain supply sustainable supply chain reverse strategy, corporate supply chain environmental management, social responsibility, transportation, eco-efﬁciency, logistics, strategy, and corporate sustainability. include supply chain management, environmental social of the impact of data can for dynamic decision-making, evaluation of procurement channel strategic partnership in ﬁrms based on


Introduction
The last decade has witnessed more and more attention from both practitioners and scholars on the sustainable supply chain management (SSCM) domain, which takes environmental, social, and economic outcomes into consideration across a focal firm's supply chain process. More and more enterprises realize the importance of sustainable development and construct sustainable supply chain systems to implement SSCM. To tackle the severe threat and high level of uncertainty due to disasters such as the COVID-19 pandemic [1], demand uncertainty [2,3], a challenging market, combined with pressure from stakeholders (competitors, end-users, or governments, etc.) [4], reputational risk [5], and corporate social responsibility [6], firms make SSCM a strategy so as to ensure long term benefits and achieve a competitive position in the market. SSCM practices, such as adopting green human resource management [7][8][9], applying environmental management systems [10,11], and evaluating and selecting sustainable suppliers [12][13][14], are implemented by firms in their internal management and external supply chains' operations.
Therefore, it is no wonder that SSCM has aroused the interest of scholars. The empirical and conceptual papers concerning SSCM topics have been growing significantly, which indicate that SSCM has evolved into a significant separate stream of supply chain management or a new stage of supply chain management, which was believed by the study

Retrieval Strategy
The literature review of SSCM using CiteSpace software takes a bibliographic dataset, whose quality measured by recall and precision from the perspective of information retrieval as the basis of subsequent analysis. Therefore, formulating an optimal retrieval strategy in advance is of prime importance and significantly challenging, especially when we attempt to develop an overview and identify emerging trends of SSCM. Overall, there are two retrieval methods, namely a query-based search and a citation expansion search in electronic databases to attain documents for literature review [42].
The query-based search method typically requires terms or phrases to find records of publications relevant to a research topic of interest, which is perhaps the preferred way. For instance, the search query of Nimsai et al. [38] consisted of two phrases about SSCM: "sustainable development and supply chain" or "sustainability in supply chain" in the first search. Other valid search terms may include "sustainable supply chain" "supply chain and sustainab*", or "green supply chain management" to search for bibliographic records in the field of SSCM [34,43,44]. However, the detection of latent semantic relations or closely related concepts is the main drawback of the query-based search method, by which recall takes precedence over precision [42]. Some relevant papers may not appear by search term sieve, which poses profound challenges to the quality of the dataset. Based on the view of citation index that papers citing the source material deserve further consideration, the citation expansion search method can uncover potentially valuable and relevant papers that may be overlooked by the method of query-based search. For example, Li et al. [45] conducted a bibliometric analysis on hospitality management research, in which citation expansion was carried out on the core dataset using citation expansion function of the Web of Science database. Chen et al. [40] applied the method of topic term search and citation expansion, in order to cast a wider net in the bibliometric analysis of orphan drugs and rare diseases.
Based on the above analysis, the query-based search combined with the citation expansion search method is applied to construct a representative and comprehensive dataset concerning SSCM research field in our retrieval process, which can maximize recall and precision. Specifically speaking, we firstly retrieve scholarly publications of SSCM as an initial dataset, namely the core dataset by the query-based search approach. Subsequently, the citation expansion search method is applied to expand the core dataset through citation links.

Retrieval Process
In order to source documents of SSCM for this review, we followed the retrieval process step by step: Web of Science (WoS) core collection, core dataset, expanded dataset, duplicates removal, and finally, total records. Figure 1 shows an explicit sequence of the retrieval process and corresponding results, which can mitigate subjectivity in selecting papers and ensure procedural transparency [46]. search and citation expansion, in order to cast a wider net in the bibliometric analysis of orphan drugs and rare diseases. Based on the above analysis, the query-based search combined with the citation expansion search method is applied to construct a representative and comprehensive dataset concerning SSCM research field in our retrieval process, which can maximize recall and precision. Specifically speaking, we firstly retrieve scholarly publications of SSCM as an initial dataset, namely the core dataset by the query-based search approach. Subsequently, the citation expansion search method is applied to expand the core dataset through citation links.

Retrieval Process
In order to source documents of SSCM for this review, we followed the retrieval process step by step: Web of Science (WoS) core collection, core dataset, expanded dataset, duplicates removal, and finally, total records. Figure 1 shows an explicit sequence of the retrieval process and corresponding results, which can mitigate subjectivity in selecting papers and ensure procedural transparency [46]. First of all, the WoS Core Collection was selected for the reason that the database covers the world's leading peer-reviewed journals about SSCM research domains. In light of the interdisciplinarity of SSCM, such as environmental sciences, operations research, and business, we delimited the Science Citation Index Expanded (SCI-E) and Social Sciences Citation Index (SSCI) in the WoS Core Collection. Secondly, we referred to the study of Patel and Desai (2019) [47] who used "sustainable supply chain management" to retrieve peer-reviewed research articles. The search term "sustainable supply chain management" is also a professional term. An advanced search was conducted to construct the core dataset on 8 March 2021, using the topic search (TS)TS = "Sustainable Supply Chain Management" and restricting results by the English language and article/review document types with an open-ended timespan. Any papers published before the retrieval time deserve further investigation. It does not matter whether our dataset includes the whole year or not. According to the study of Bouazzaoui et al. [46], we delimited peer-reviewed articles and reviews, and excluded books, proceeding papers, and other unpublished works in order to control for the quality of publications. A total of 592 records, including 524 articles and 68 reviews, resulted from the query-based search approach. Then, we expanded the initial dataset through citation links by using the "Create Citation Report" function of the WoS database to construct the expanded dataset. This citation expansion search uncovered 8131 articles and 973 reviews that cite one or more publications of the core dataset. In addition, there are 545 duplicates including 480 articles and 65 reviews needing discarding when these two datasets were merged into one dataset. Eventually, as indicated in Figure 1, after refinement of excluding these duplicates, total unique records of 9151 peer-reviewed papers composed of 8175 articles and 976 reviews with 320,376 valid references were obtained. These papers published from 2007-2021 are considered for later scientometric review. Figure 2 shows the distribution of publications of the SSCM domain over the years, which can be an indicator reflecting the macro development trend [48] and scholars' interests. As is showed in Figure 2, there was a rapid increasing trend from 2007 to 2020, First of all, the WoS Core Collection was selected for the reason that the database covers the world's leading peer-reviewed journals about SSCM research domains. In light of the interdisciplinarity of SSCM, such as environmental sciences, operations research, and business, we delimited the Science Citation Index Expanded (SCI-E) and Social Sciences Citation Index (SSCI) in the WoS Core Collection. Secondly, we referred to the study of Patel and Desai (2019) [47] who used "sustainable supply chain management" to retrieve peer-reviewed research articles. The search term "sustainable supply chain management" is also a professional term. An advanced search was conducted to construct the core dataset on 8 March 2021, using the topic search (TS)TS = "Sustainable Supply Chain Management" and restricting results by the English language and article/review document types with an open-ended timespan. Any papers published before the retrieval time deserve further investigation. It does not matter whether our dataset includes the whole year or not. According to the study of Bouazzaoui et al. [46], we delimited peer-reviewed articles and reviews, and excluded books, proceeding papers, and other unpublished works in order to control for the quality of publications. A total of 592 records, including 524 articles and 68 reviews, resulted from the query-based search approach. Then, we expanded the initial dataset through citation links by using the "Create Citation Report" function of the WoS database to construct the expanded dataset. This citation expansion search uncovered 8131 articles and 973 reviews that cite one or more publications of the core dataset. In addition, there are 545 duplicates including 480 articles and 65 reviews needing discarding when these two datasets were merged into one dataset. Eventually, as indicated in Figure 1, after refinement of excluding these duplicates, total unique records of 9151 peer-reviewed papers composed of 8175 articles and 976 reviews with 320,376 valid references were obtained. These papers published from 2007-2021 are considered for later scientometric review. Figure 2 shows the distribution of publications of the SSCM domain over the years, which can be an indicator reflecting the macro development trend [48] and scholars' interests. As is showed in Figure 2, there was a rapid increasing trend from 2007 to 2020, especially after 2015, which indicates that SSCM research is getting more and more academic attention. The reason why publications in 2021 are only 430 is that our dataset only covers the first three month of the year. In 2007, there were only two papers, one of which [49] studied aspects of SSCM and introduced the concepts of first-, second-, and n-order supply chains. The other paper [50] studied the pinch analysis approach combined with multi-criteria analysis to realize a more sustainable production. In 2017, a decade later, the quantity increased to 909, to be 9.93% of total records. The number of publications hit the peak in 2020 when 2571 papers were published. Besides this, 79.29% of the publications (7256 out of 9151) were published in the last five years (2017-2021), suggesting that the SSCM research has entered a booming period.
Sustainability 2021, 13, x FOR PEER REVIEW 5 of 26 especially after 2015, which indicates that SSCM research is getting more and more academic attention. The reason why publications in 2021 are only 430 is that our dataset only covers the first three month of the year. In 2007, there were only two papers, one of which [49] studied aspects of SSCM and introduced the concepts of first-, second-, and n-order supply chains. The other paper [50] studied the pinch analysis approach combined with multi-criteria analysis to realize a more sustainable production. In 2017, a decade later, the quantity increased to 909, to be 9.93% of total records. The number of publications hit the peak in 2020 when 2571 papers were published. Besides this, 79.29% of the publications (7256 out of 9151) were published in the last five years (2017-2021), suggesting that the SSCM research has entered a booming period.

Methods
Currently, there are a variety of tools that can perform bibliometric analysis, such as CiteSpace, VosViewer, Bibexcel, etc. Each tool has different features and the choice depends on the requirements and intended purpose of the analysis. Compared with other tools, CiteSpace can be used to generate knowledge maps, show the knowledge structure and its dynamic changes, evaluate research status, reveal research hotspots, and predict research trends by analyzing literature in a certain research field. It is widely used in COVID-19 research [51], regenerative medicine [52], hotel management [45], innovation system [36], and other disciplines. Therefore, this paper conducts a bibliometric review on the SSCM field using CiteSpace 5.7 R2.
According to the study of [53], research techniques of co-author analysis, co-word analysis, and co-citation analysis are applied in this study to analyze different aspects of the SSCM field including social structure, conceptual structure, and intellectual structure. Firstly, using the method of co-author analysis, we map networks of co-authorship and co-authors' institutions in the SSCM field, so as to reveal the social structure. Secondly, using the method of co-word analysis, we map the network of co-occurring keywords to reveal the conceptual structure and identify main concepts and research hotspots. Thirdly, we carry on co-citation analysis, taking authors, journals, and documents as analysis units to reveal the intellectual structure of the SSCM field. Specifically, document co-citation analysis is based on the assumption that when two documents are cited by the third one at the same time, there is some correlation between the two documents. We also carry out clustering about references. The cited references in the cluster are cited by the academic group, which play the function of knowledge base, whereas the citing articles may be derived from the corresponding clustering and are considered as the research fronts [39]. Clustering quality can be measured by network modularity and weighted mean silhouette [51]. The former indicator reflects the clustering effect of the whole network structure. The

Methods
Currently, there are a variety of tools that can perform bibliometric analysis, such as CiteSpace, VosViewer, Bibexcel, etc. Each tool has different features and the choice depends on the requirements and intended purpose of the analysis. Compared with other tools, CiteSpace can be used to generate knowledge maps, show the knowledge structure and its dynamic changes, evaluate research status, reveal research hotspots, and predict research trends by analyzing literature in a certain research field. It is widely used in COVID-19 research [51], regenerative medicine [52], hotel management [45], innovation system [36], and other disciplines. Therefore, this paper conducts a bibliometric review on the SSCM field using CiteSpace 5.7 R2.
According to the study of [53], research techniques of co-author analysis, co-word analysis, and co-citation analysis are applied in this study to analyze different aspects of the SSCM field including social structure, conceptual structure, and intellectual structure. Firstly, using the method of co-author analysis, we map networks of co-authorship and co-authors' institutions in the SSCM field, so as to reveal the social structure. Secondly, using the method of co-word analysis, we map the network of co-occurring keywords to reveal the conceptual structure and identify main concepts and research hotspots. Thirdly, we carry on co-citation analysis, taking authors, journals, and documents as analysis units to reveal the intellectual structure of the SSCM field. Specifically, document co-citation analysis is based on the assumption that when two documents are cited by the third one at the same time, there is some correlation between the two documents. We also carry out clustering about references. The cited references in the cluster are cited by the academic group, which play the function of knowledge base, whereas the citing articles may be derived from the corresponding clustering and are considered as the research fronts [39]. Clustering quality can be measured by network modularity and weighted mean silhouette [51]. The former indicator reflects the clustering effect of the whole network structure. The larger the value is, the easier it is to divide the whole network into several groups with close connections among members but loose connections between groups. Modularity between 0.4-0.8 is acceptable. The latter indicator measures the similarity among members, and the larger the value, the more significant a certain cluster division is [42].
This paper mainly uses metrics of frequency or count, betweenness centrality and citations burst to identify important nodes, which may be authors, journals, keywords, and documents [39]. The sizes of nodes with higher frequencies or counts are bigger than those with lower frequencies. For example, cited references with high citation frequency have been widely recognized by the scientific communities and have large citation rings in the figure [51]. The betweenness centrality, as a structural indicator, indicates the position of a particular node in the network. Nodes with high betweenness centrality (greater than or equal to 0.1) are located in structural holes and covered by purple circles in the figure, which have the potential to link various research themes and may bring transformative findings. Knowledge "turning points" can be identified by this metric [36]. From the perspective of information flow, scholars in the structure hole are supposed to connect different researchers and get in touch with various ideas, perspectives and viewpoints, which can make them more open-minded and creative. The citations burst, as a temporal metric, can be applied to identify particular nodes such as keywords or cited references, which have attracted wide attention of scholars in a certain period [40].

Results
In this study, we reveal social structure (RQ1), conceptual structure (RQ2), and intellectual structure (RQ3) of the SSCM field by the methods of co-author analysis, co-word analysis, and co-citation analysis.

Social Structure of SSCM Field
We focus on network of co-authors and their institutions in this section to analyze social structure.

Co-Authorship Network
In order to study key authors and their collaborative relationships in the field of SSCM, this paper analyzes the authors of 9151 literatures related to SSCM. The co-authorship network was generated by CiteSpace whose parameters are set up as follows: top 50 Per year (2007-2021), LRF = 3, LBY = 5, and e = 2. To make the map easier to visualize, we pruned the sliced networks and merged networks based on a pathfinder algorithm, resulting in a network with 607 nodes and 585 links whose largest subnetwork is shown in Figure 3. Each node is labeled by the corresponding author. The linkage between two nodes indicates that two authors collaborated to research in the same paper. The thicker the linkage is, the high the level of cooperative relationship is. In addition, the nodes with frequency greater than or equal to 15 are displayed. From Figure 3, we can identify cooperative relationships among scholars and productive scholars clearly in the largest subnetwork.
In terms of cooperative relationships, the density of the network is 0.0037, indicating that authors' group has not yet formed strong relationships of collaboration. There are only two scholars for which the betweenness centrality is more than 0.1, including Joseph Sarkis

Co-Authors' Institutions Network
In order to explore core institutions and relationships of cooperation in the field of SSCM, we generated a network of co-authors' institutions. CiteSpace parameters are similar to the co-authorship network. The font size of an organization is proportional to the number of publications. The connection and thickness between nodes represent the cooperation relationship and frequency between two institutions.
The nodes and linkages number of the whole cooperation network of research institutions are 291 and 283, respectively. The network density is 0.0067. The largest subnetwork has 226 node network members, accounting for 77% of the total network nodes, shown in

Co-Authors' Institutions Network
In order to explore core institutions and relationships of cooperation in the field of SSCM, we generated a network of co-authors' institutions. CiteSpace parameters are similar to the co-authorship network. The font size of an organization is proportional to the number of publications. The connection and thickness between nodes represent the cooperation relationship and frequency between two institutions.
The nodes and linkages number of the whole cooperation network of research institutions are 291 and 283, respectively. The network density is 0.0067. The largest subnetwork has 226 node network members, accounting for 77% of the total network nodes, shown in

Conceptual Structure of SSCM Field
In this section, we apply the method of co-word analysis to reveal the conceptual structure, and identify main concepts and research hotspots.

Co-Occurring Keywords Network
Keywords, as representative words of a paper, are high-level summary of content [48], which can enable researchers to gain an understanding of the core and essence of the study. High frequency and betweenness centrality of co-occurrence of keywords can reflect main concepts in its research fields. Therefore, the method of co-word analysis is applied to reveal the conceptual structure of SSCM fields. For the purpose of conducting a co-word analysis utilizing CiteSpace software, the configuration parameters were set up as follows: top 30 Per year (2007-2021), LRF = 3, LBY = 5, and e = 2. The sliced networks and merged network map of co-occurrence were pruned by a pathfinder algorithm. The network of co-occurring keywords, with 129 nodes and 156 links, is shown in Figure 5. As is shown in Figure 5, each keyword is represented by a circle-node whose size is a sign of the frequency of a keyword. The color of linkages between keywords indicates the first time when two keywords occurred in the same paper. The brighter the color of linkage is, the closer the first year of co-occurrence is to the present. In addition, the nodes with frequency greater than or equal to four are displayed. We can also obtain the count and centrality of keywords from CiteSpace. The importance of keywords cannot only be judged by the frequency or count, but also its betweenness centrality in the network. Table 1 lists

Conceptual Structure of SSCM Field
In this section, we apply the method of co-word analysis to reveal the conceptual structure, and identify main concepts and research hotspots.

Co-Occurring Keywords Network
Keywords, as representative words of a paper, are high-level summary of content [48], which can enable researchers to gain an understanding of the core and essence of the study. High frequency and betweenness centrality of co-occurrence of keywords can reflect main concepts in its research fields. Therefore, the method of co-word analysis is applied to reveal the conceptual structure of SSCM fields. For the purpose of conducting a co-word analysis utilizing CiteSpace software, the configuration parameters were set up as follows: top 30 Per year (2007-2021), LRF = 3, LBY = 5, and e = 2. The sliced networks and merged network map of co-occurrence were pruned by a pathfinder algorithm. The network of co-occurring keywords, with 129 nodes and 156 links, is shown in Figure 5. As is shown in Figure 5, each keyword is represented by a circle-node whose size is a sign of the frequency of a keyword. The color of linkages between keywords indicates the first time when two keywords occurred in the same paper. The brighter the color of linkage is, the closer the first year of co-occurrence is to the present. In addition, the nodes with frequency greater than or equal to four are displayed. We can also obtain the count and centrality of keywords from CiteSpace. The importance of keywords cannot only be judged by the frequency or count, but also its betweenness centrality in the network. Table 1 lists the top 20 keywords with high count or centrality. From Figure 5, we can identify keywords with high frequency and/or with high betweenness centrality and draw out main concepts. the top 20 keywords with high count or centrality. From Figure 5, we can identify keywords with high frequency and/or with high betweenness centrality and draw out main concepts.    In terms of frequency, sustainability with the frequency of 1359 ranks first of all the keywords. Supply chain management is the second high count keyword, followed by supply chain (446). Scholars also pay great attention to circular economy (322) and corporate social responsibility (203). The keywords sustainable supply chain management (184), green supply chain management (170), and green supply chain (158) represent the subdivided classification of supply chain management studied by scholars. The keyword literature review (196) and case study (143) represent the methods of SSCM research. Supplier selection (155) and reverse logistics (132) are also high frequency keywords.
On the other hand, keywords with high betweenness centrality indicate that they are in the center of network and important in linking other keywords or other research topics. The centrality of nodes in a purple circle are greater than 0.1. For example, the centrality of supply chain is 0.47, greater than 0.1, which links corporate social responsibility, circular economy, resilience, green product, and so on. Table 1 shows the top 10 high betweenness centrality value of keywords, in which the keywords supply chain, sustainable supply chain management, and reverse logistics show the highest betweenness centrality among all others at 0.47, 0.42 and 0.42, respectively. Other keywords such as strategy, corporate social responsibility, sustainability, corporate sustainability, and supplier management at betweenness centrality values range from 0.3 to 0.4. Keywords whose centrality is greater than 0.1 deserve to be studied further.
Considering both criteria of frequency and betweenness centrality, the values of following keywords are high: sustainability, supply chain, sustainable development, circular economy, corporate social responsibility, sustainable supply chain, sustainable supply chain management, green supply chain, case study, and reverse logistics. These keywords can be summarized into five main concepts: sustainable supply chain management, green supply chain management, circular economy, corporate social responsibility, and reverse logistics.

Keywords Burst Analysis
The notable increase in the frequency of a keyword during a relatively short period of time usually reflects research hotspots, which are paid special attention by the scientific community. We carried out keywords burst detection to identify research hotspots of the SSCM domain using CiteSpace [54]. Table 2 shows 25 keywords with bursts of at least two years. In chronological order, the burst keywords in the SSCM field have been changing over the years from 2007 to 2021.  T  T  T  T  T  T  S  S  S  S  S  S  S  S  2  environmental management  22.05  2009  2015  S  S  T  T  T  T  T  T  T  S  S  S  S  S  S  3  social responsibility  11.21  2010  2016  S  S  S  T  T  T  T  T  T  T  S  S  S  S  S  4  , which indicate that these topics got more attention and are more influential than other keywords, thus, they became research hotspots of the SSCM domain in corresponding periods. Besides, the game theory and circular economy beginning bursting in 2018 and 2019, respectively, continue bursting to the present, which are research hotspots currently.
From the point of view of burst strength, supply chain management (25.39) is the strongest burst, followed by circular economy (22.22), environmental management (22.05), social responsibility (11.21), transportation (4.99), and supplier management (4.79), which are research hotspots in their corresponding periods. It is noteworthy that circular economy is not only bursting to present, but also has high burst strength. This keyword appeared in 322 records including 50 records in 2018, 87 records in 2019, and 126 records in 2020 in our dataset. On the whole, we argue that the research hotspots of the SSCM field currently are game theory and circular economy related researches.

Intellectual Structure of SSCM Field
In this section, we apply the method of co-citation analysis to reveal intellectual structure by analyzing cited authors, journals, and documents.

Author Co-Citation Network
Author co-citation network is generated to identify highly cited scholars whose publications are widely recognized by research communities in SSCM research. When two scholars are cited in the same publications, the relationship of author co-citation occurs. We take 9151 peer-reviewed papers' references as analysis objects. The highly cited authors may not be SSCM scholars, but their contributions certainly have a great impact on the development of the SSCM field. CiteSpace configurations were set up as follows: Top N (N = 50) Per year (2007-2021), LRF = 3, LBY = 5, and e = 2. Sliced and merged networks were pruned according to the pathfinder algorithm, which resulted in 233 nodes and 419 links. The bigger the size of each node, the more citations the scholar has. The thicker linkage between two nodes is, the more times two authors are cited in the same papers. The nodes with citations over 200 are labeled by the corresponding first author. The nodes with betweenness centrality no less than 0.1 are covered by a purple circle. From Figure 6, we can identify cited authors with high citations and/or with high betweenness centrality and draw out leading researchers.   As is shown in Figure 6, the top five most highly cited authors are Stefan Seuring (with 2158 citations), Kannan Govindan (1514), Craig R. Carter (1478), Qinghua Zhu (1453), and Joseph Sarkis (1328). Figure 7 shows the citation distribution of top five most cited authors. From Figure 7, we can know that on the whole, citations of these five scholars show a rapidly increasing trend from 2008 to 2020. Professor Stefan Seuring of Universität Kassel in Germany is cited the most every year, except for 2020. It is worth mentioning that Kannan Govindan, who works at China Institute of FTZ Supply Chain in Shanghai Maritime University and Centre for Sustainable Supply Chain Engineering in University of Southern Denmark, was cited the first time in 2014, and in 2020 his citations exceeded Stefan Seuring and ranked first. Professor Kannan Govindan is the 2018 Highly Cited Researcher (Clarivate Analytics) and his research interests mainly include reverse logistics, closed-loop supply chain, SSCM, and GSCM. Professor Craig R. Carter works at Arizona State University, whose expertise areas include SSCM, supply chain management decisionmaking and negotiation. Professor Qinghua Zhu works at Antai College of Economics and Management, Shanghai Jiao Tong University, whose research interests include GSCM, corporate social responsibility, and remanufacturing management. Professor Joseph Sarkis works at the School of Business, Worcester Polytechnic Institute, whose research interests include supply chain management, multi-criteria decision-making, and so on.
Arizona State University, whose expertise areas include SSCM, supply chain management decision-making and negotiation. Professor Qinghua Zhu works at Antai College of Economics and Management, Shanghai Jiao Tong University, whose research interests include GSCM, corporate social responsibility, and remanufacturing management. Professor Joseph Sarkis works at the School of Business, Worcester Polytechnic Institute, whose research interests include supply chain management, multi-criteria decision-making, and so on.  It is noteworthy that professor Kathleen M. Eisenhardt is not a SSCM scholar, but she is in Figure 6. Many of her papers are cited by SSCM scholars and her contributions can be considered as the knowledge base (theory or method) for the SSCM field. For example, It is noteworthy that professor Kathleen M. Eisenhardt is not a SSCM scholar, but she is in Figure 6. Many of her papers are cited by SSCM scholars and her contributions can be considered as the knowledge base (theory or method) for the SSCM field. For example, Sarkis et al. (2011) [55] referred to Eisenhardt's paper titled "Agency theory: an assessment and review". Govindan et al. (2014) [56] referred to Eisenhardt's paper titled "Better stories and better constructs: the case for rigor and comparative logic". Her paper titled "Theory Building From Cases: Opportunities And Challenges" [57] is highly cited by SSCM scholars.
In terms of betweenness centrality, there are 16 scholars whose betweenness centrality are no lower than 0. When authors have both high citation and betweenness centrality, they can be considered as influential or leading scholars [36,58]. In this paper, taking citations count and betweenness centrality simultaneously, the following scholars, whose citation and centrality exceed 500 and 0.  In addition, we can also identify influential scholars from the point of citation bursts, that is, a scholar is cited much during a short period. The citations of several authors have been bursting to present, including Sunil Luthra (with a burst strength of 50.88, from 2018), David J. Teece (36.45, 2018) The publications of these authors are worth studying because of their significant impact on SSCM research.

Journal Co-Citation Network
In this section, we first analyze the source of publications and then detect the most representative cited journals in the SSCM field.
It is found that 9151 papers related to the SSCM field are published in 889 journals, the top 10 of which are listed in Table 3. These 10 journals published 4029 papers, which account for 44.03% of the total records, which contribute greatly to progress in this field and reflect a high concentration. As for frequency, the most productive journal is the Journal of Cleaner Production publishing 1434 SSCM related papers and accounting for 15.67% of papers of the total records, followed by Sustainability , which indicate papers in these journals are fully refereed according to accepted standards and conventions. Besides, these journals have a pluralistic nature, such as environmental sciences, engineering, operations research, business, management, and so on, which indicate that the academic field of SSCM is interdisciplinary [59]. We then generated a journal co-citation map to detect and evaluate influential journals that contribute to the development of SSCM research and serve as the knowledge base to some degree, as shown in Figure 8    In terms of betweenness centrality (nodes with purple circle) of cited journals, the following journals whose centrality exceed 0.

Document Co-Citation Network
When a group of documents is frequently cited in conjunction with other documents, this cluster may represent a certain research theme. Compared with other clusters, each cluster member is cited more frequently by a group of the same citing articles. In this section, based on 320,376 valid references cited in the 9151 records in our dataset, we applied the method of document co-citation analysis to visualize the landscape view of the SSCM field and analyze underlying knowledge base and research fronts. CiteSpace parameter settings are set up as follows: g-index (k = 40) Per year (2007-2021), LRF = 3, LBY = 5, e = 2 and Pruning = None. The synthesized network of co-cited references in SSCM research, with 2327 nodes and 12,165 links, is shown in Figure 9. We then carried out clustering, which generated 220 clusters, which are labeled with title terms extracted from citing articles through the Log-likelihood ratio (LLR) algorithm. Compared with the latent semantic indexing (LSI) algorithm focusing on identifying common theme, the LLR algorithm tends to emphasize unique topics [39]. Figure 9 shows 19 clusters, including #0 Chinese manufacturer, #1 sustainable supply chain management, #2 green supply chain management, #3 sustainable supplier selection, #4 circular economy, #5 and-trade regulation, #6 closed-loop supply chain, #7 sustainable production network, #8 blockchain technology, #9 supply chain management profession, #10 big data analytics, #11 corporate sustainability strategies, #12 COVID-19 pandemic, #13 supply chain resilience, #14 green human resource management, #15 best-worst method, #16 logistics issue, #21 energy efficiency, and #26 logistics performance, which are major specialties of the SSCM field. Each cluster signifies distinct aspects of SSCM issues and topics. For instance, the brilliant yellow-colored area at the lower right quadrant is labeled as #8 blockchain technology, which indicates that cluster #8 is cited by papers about blockchain technology related topics. The color of the convex hull of each cluster indicates mean year calculated on publication year of the cluster's members. In addition, the brighter the color is, the closer average year of one cluster is to the present. The quality of co-citation clusters is supposed to meet both criteria of modularity and weighted mean silhouette, which deserves to be thoroughly investigated. The modularity of the network is 0.7664, which is considered as a higher value, denoting that a well-structured network is developed and the specialties in SSCM fields are clearly defined. The weighted mean silhouette, as an indicator measuring the internal homogeneity of each cluster, is 0.8659, signifying the clustering is highly reliable and the members of corresponding cluster are more similar than other clusters' members.   A timeline view of clusters is generated to show the origin, evolution, and time span of each cluster in Figure 10. Disappearance of a cluster may not mean that scholars have lost interest in this field, but may mean that they continue to explore new research directions [39]. As is depicted in Figure 10, each cluster's members are showed in chronological order along the horizontal axis, whereas clusters are displayed vertically from top to down according to their sizes.   Then, we list detailed information of these 19 clusters in Table 4, including cluster ID, size, percentage of the network, silhouette value, start-stop time, duration, mean year, and labels based on the LLR algorithm. Silhouette values of all 19 clusters are greater than 0.7, indicating that clusters are highly reliable and members have high internal consistency. The largest cluster is #0 Chinese manufacturer composed of 361 nodes, which accounted for 15.51% of the whole network. Each of the seven largest clusters has over 100 members representing 64.12% of cited references of the entire network. We are particularly interested in the duration of one cluster. There are eight clusters whose durations exceed 10 years, including #0 Chinese manufacturer, #1 sustainable supply chain management, #2 green supply chain management, #3 sustainable supplier selection, #5 and-trade regulation, #6 closed-loop supply chain, #9 supply chain management profession, and #11 corporate sustainability strategies, of which cluster #3 spans the longest period, lasting 16 years. In terms of activeness, we find nine clusters remaining active, including #1 sustainable supply chain management, #3 sustainable supplier selection, #4 circular economy, #5 and-trade regulation, #8 blockchain technology, #10 big data analytics, #12 COVID-19 pandemic, #15 best-worst method, and #26 logistics performance, which can be considered as emerging trends of SSCM, except #1. Taking size and activeness of clusters into consideration, this study mainly focuses on the following specialties: #1 Sustainable supply chain management, #3 Sustainable supplier selection, #4 circular economy, #5 and-trade regulation, and #8 blockchain technology. By reading core publications that include highly cited references as knowledge base and representative citing articles as research fronts, we can understand these clusters in depth.
The cluster #1, labeled by sustainable supply chain management, is the second largest cluster containing 244 members and with a range of a 11-year duration from 2010 to 2020. A study by Ahi and Searcy (2013) is identified as the most cited reference and also has the strongest burst strength in the cluster. Ahi and Searcy [31] proposed a concise and comprehensive definition of SSCM, which captured key characteristics of both supply chain management and business sustainability. They argued SSCM was the extension of GSCM. As the second most cited article, the study of [15] used the term SSCM in a broad sense, which included all the environmental or social research of supply chain management, such as GSCM. The most representative citing article for cluster #1 is that of Sánchez-Flores [44] covering 36 cited articles in the cluster, which performed a systematic literature review about SSCM research in the context of emerging economies, and showed that when compared to developed countries, the research in emerging economies lagged and was still in its infancy. The study of Mardani [60], covering 33 cited article in the cluster, further presented a systematic review regarding applications of the structural equation modeling (SEM) in the evaluation of GSCM and SSCM.
The cluster #3, labeled by sustainable supplier selection, is the fourth largest cluster across a 16-year period from 2005 to 2020, which is the longest period of time among all clusters. Supplier selection plays a crucial part in SSCM. This cluster focuses on how to select and evaluate suppliers to achieve "triple bottom line" benefits. There are several novel integrated approaches to sustainable supplier selection and evaluation operations. The study of [61] presented a novel integrated fuzzy PIPRECIA-interval Rough SAW model for green supplier selection. Chattopadhyay et al. [62] employed an integrated D-MARCOS method, which can address the uncertainty in the supplier selection process. Durmi [63] applied the Full Consistency method (FUCOM) to define the evaluation of criteria for sustainable supplier selection. The study of [64] combined FUCOM-Rough SAW approach for supplier selection. When it comes to research fronts, the top two citing articles to this cluster offer literature reviews of state-of-the-art studies concerning sustainable supplier selection [65,66]. Based on previous researches, Ecer and Pamucar [67], Hendiani et al. [68], and Kannan et al. [69] cited more than 20 members in cluster #3 and proposed the integrated F-BWM and fuzzy CoCoSo'B multi-criteria model, the multi-stage hierarchical fuzzy index-based approach, and the hybrid approach combining the fuzzy best-worst method and the interval VIKOR technique, respectively.
The cluster #4, labeled by circular economy, contains 157 references, with a silhouette value of 0.972, which is considered a high value. The average year of all members in this cluster is 2017, with an eight-year duration (2013-2020). The most cited article in cluster #4 is published by Genovese et al. who asserted that in view of an environmental point, integrating principles underlying circular economy into SSCM can provide potential enhancement [70]. The study of Geissdoerfer, with the strongest burst strength in the cluster, distinguished the terms circular economy and sustainability explicitly, therefore, clarifying their conceptual contours [71]. The most active citing article covering 37 papers is the paper of [72], which first researched circular economy in the leather industry context and found that financial facility played a vital role in the implementation of circular economy practices by using the best worst method.
The cluster #5, labeled by and-trade regulation, has 151 members with a silhouette value of 0.945. The label can be revised as cap-and-trade regulation through reviewing citing articles. Cap-and-trade regulation is a carbon policy aimed at combating global warming. The most representative article in cluster #5 was published by Ghosh and Shah [73] who explored the influence of cost sharing contract on green supply chains. With regard to representative citing articles, some studies conducted supply chain researches under cap-and-trade regulation, such as channel coordination in a two-echelon sustainable supply chain [74,75], stochastic dual-channel supply chain [76], production and carbon emission reduction strategies [77], and low-carbon production [78]. Tang and Yang [79] analyzed the influence of power structure and financing mechanism on a low-carbon supply chain.
The cluster #8, labeled by blockchain technology, has 84 members, with a silhouette value of 0.99. As a distributed digital ledger technology, blockchain technology has completely overturned the traditional idea of centralized organization, and is expected to solve some problems of some global supply chain management problems due to the globalization of supply chains. The most cited article in cluster #8 is the paper titled "Blockchain technology and its relationships to sustainable supply chain management" [17], which is also hot paper in the WoS Core Collection. This paper examined the potential application of blockchain technology to overcome some barriers of supply chain sustainability. The most citing article of this cluster is the study of Chang and Chen [80] who conducted a literature review of blockchain-based supply chain research and argued that the application of blockchain can facilitate distributed governance and process automation for supply chain operations. Kayikci et al. [81] applied blockchain technology into a food supply chain and proposed a blockchain-enabled food supply chain framework, which can resolve traceability and accountability problems and ensure transparency. The study of Orji et al. [82] and Ar et al. [83] evaluated the critical factor of blockchain adoption in freight logistics industry and the feasibility of blockchain in logistics industry.

Bursts in the Network of Document Co-Citations
In this section, we identify the top 20 references with the strongest citation bursts lasting more than five years between 2007 and 2021 (see Table 5), which can be considered as major milestones in the development and of evolution SSCM search [39].
From 2007 to 2021, the focus of these major milestones can be summarized as follows. (1) Literature review of SSCM or GSCM. Sarkis et al. [55] conducted an organizational theoretic literature review of GSCM, whereas Carter and Easton [84] conducted a systematic review of SSCM. Other papers provided reviews from various angles. For example, using the method of content analysis, Gold et al. [85] highlighted that supply-chain-wide collaboration can facilitate inter-organizational resources and, thus, maintain inter-firm competitive advantage. Seuring [86] reviewed modeling approaches used in SSCM research and founded multi-criteria decision-making, equilibrium models, and analytical hierarchy processes were dominant approaches. (2) Building a conceptual framework of SSCM or GSCM. Seuring and Muller [29], and Carter and Rogers [30] presented a framework of SSCM, respectively, whereas Sarkis [87] provided a framework of GSCM based on a boundaries and flows perspective. It is noteworthy that the study of Seuring and Muller [29] had the strongest burst value of 139.32. (3) Evaluation or process of supplier selection. The study of [88] has experienced the longest burst period of six years, which introduced a method combining grey system with rough set theory for the supplier selection. Based on fuzzy analytic network process, Buyukozkan and Cifci [89] developed a fuzzy multicriteria decision framework with incomplete information. These scholars [90], then, in 2012, proposed a hybrid fuzzy multiple criteria decision-making model for evaluating green suppliers, which integrated DEMATEL, ANP, and TOPSIS in a fuzzy context. (4) Other topics related to SSCM, such as corporate social responsibility [91], firm performance [92], and sustainable sourcing [93].

Main Findings and Contributions
SSCM has been attracting extensive attention from both practitioners and scholars. More and more firms realize the importance of the concept of sustainable development and implement SSCM practices in order to realize an organization's social, environmental, and economic goals. The significant increasing number of empirical and conceptual papers on SSCM topics indicates that SSCM research is getting more and more academic attention. However, there are limited bibliometric literature reviews on SSCM to help us gain a better understanding and overview of the evolution of this field. The main objective of this study is to visualize and conduct a systematic scientometric review on 9151 articles and reviews published from 2007 to 2021. Research techniques of co-author analysis, co-word analysis, and co-citation analysis are applied in this study to address three research questions: RQ1: What is the social structure of the SSCM field? RQ2: What is the conceptual structure of the SSCM field? RQ3: What is the intellectual structure of the SSCM field? We mainly focus important nodes with high frequency, high betweenness centrality, or high burst strength. We reveal social structure, conceptual structure, and intellectual structure, identify main concepts and research hotspots, and illuminate major specialties and emerging trends, which are

Limitations and Future Researches
Our papers have several limitations from the following aspects: databases, publication types, search terms, method, and citation manipulation. Firstly, the bibliometric dataset is retrieved from SCI-E and SSCI in the WoS Core Collection. Various results and conclusions may appear if the dataset is collected from other databases. Therefore, future research can retrieve data from other databases such as the Scopus database, which also includes comprehensive scientific data that is rigorously vetted and selected. Secondly, we delimit peer-reviewed articles and reviews in the English language, and exclude publications from other document types such as books or proceeding papers and other languages such as the Chinese language, which may result in incomplete records. Therefore, future research should cover more various publication types and other languages papers. Thirdly, while we ensure a rigorous retrieval process to construct a representative and comprehensive dataset, we also agree to the point that there may exist many articles, which lack the search term but still focus on the SSCM. Other search terms (e.g., green procurement, RSCM/GSCM, sustainable development and supply chain, sustainability in supply chain), which capture some constructs of SSCM, may also be used as search terms to cover a wider range of articles in future study. Fourthly, we mainly used the scientometric review approach to conduct a review. We can integrate the scientometric review approach with the traditional systematic review approach in future research. For example, we have revealed cluster #8 blockchain technology, which has 84 members, with a silhouette value of 0.99. We can identify representative cited references as the knowledge base and citing articles as research fronts of this cluster. Then we can use a traditional systematic review approach to consolidate these core publications. Therefore, we can get a comprehensive review and synthesis of extant SSCM studies with respect to theoretical perspectives, methodologies, gaps, and potential research avenues. In addition, citation manipulation may occur due to author self-citation, editor or journal self-citation requirements, which may influence the accuracy of co-citation analysis on SSCM research.
Though this study has its limitations, we remain confident that our systematic and comprehensive review offers valuable insights and guidance to readers including both researchers and practitioners in the field of SSCM. The findings presented in this paper provide insights regarding productive scholars and institutions, influential journals, main concepts, research hotspots, major specialties, and emerging trends for readers to better understand the state of the art of the SSCM field. Moreover, this study can facilitate SSCM practitioners to select appropriate institutions for SSCM consulting or cooperation. Based on our findings, we could suggest several related topics for future researches, which include sustainable supplier selection, circular economy, cap-and-trade regulation, blockchain technology, big data analytics, and the COVID-19 pandemic. These topics are emerging trends identified by the analysis of co-citation clusters. However, scholars pay more attention to the first three ones, rather than the rest. Specialties of blockchain technology, big data analytics, and the COVID-19 pandemic deserve more in-depth studies and further exploration. Several future research avenues or research questions are proposed to better understand these three themes.
(1) Blockchain technology. The application of blockchain technology to the supply chain can pose more challenges and opportunities for SSCM and has the potential to transform practices [16,94]. As a distributed database system, blockchain technology can be made use of to obtain competitive advantages and enhance market positions. There are several research questions that need to be further explored. RQ1: How does the introduction of blockchain technology significantly transform SSCM practices, such as advancing inventory management and replenishment, reducing the cost of supply chain transactions, or reducing the need for intermediaries? RQ2: How can blockchain technology be effectively implemented in complex supply chain networks? RQ3: What are the implementation challenges of the supply chain finance driven by blockchain technology, and how can they be overcome?
(2) Big data analytics. The big data processing technology can enable data integration [95], information feedback, and decision-making coordination in modern supply chains more and more effectively, which have become a possible source of competitive advantage. There are extensive research values and significance in the SSCM field within the context of big data. Considering the application of big data analytics, the following research questions need to be well answered. RQ1: How does the theory application and development of big data analytics challenge existing theory on SSCM? RQ2: What are the mechanism and path of the impact of big data analytics on SSCM? RQ3: How can big data analytics be applied for dynamic decision-making, evaluation of procurement risk, channel coordination, and strategic partnership in SSCM? RQ4: How can firms predict irresponsible supply chain issues, such as child labor, unethical behavior, and environmental pollution, through the information management practices based on big data analytics?
(3) COVID-19. With the increasing complexity of the supply chain across the world, the outbreak of the COVID-19 pandemic has disrupted global supply chains and caused chaotic situations in SSCM, which can pose survivability challenges to many enterprises [96].