As a typical emerging and converging technology field, nanoscience & nanotechnology (N&N) has attracted tremendous governmental funds and scientific efforts. Articles in the field of N&N have grown explosively. With the rapid development of N&N, studies on N&N have been widely conducted by information scientists worldwide.
Disciplinary analysis of N&N as a major research area has drawn many scholars’ interests. Numerous topics of N&N have been studied, such as impact evaluation of N&N (Bartol & Stopar, 2015; Kostoff, Barth, & Lau, 2008; Leydesdorff, 2013), nano-competition or “nanorace” among countries or regions or institutions (Gorjiara & Baldock, 2014; Guan & Wei, 2015; Leydesdorff & Wagner, 2009; Wong, Ho, & Chan, 2007), technological life cycle of N&N (Anick, 2007; Milanez et al., 2013), and mapping of N&N (Bartol & Stopar, 2015; Kostoff, Koytcheff, & Lau, 2007; Mohammadi, 2012).
Investigations into the interdisciplinarity of N&N have been explored from a wide range of aspects. From the view of author collaboration, hypotheses, such as whether the collaboration in the area of N&N is of an obvious nature, have been proposed (Schummer, 2004), yet the results have not verified the assumption. As far as the toxicology and environmental risks of N&N are concerned, some approaches from interdisciplinary angles have been presented; some examples are an interdisciplinary approach for a comprehensive analysis of the impacts and ethical acceptability of nano technologies (Patenaude et al., 2015); and an interdisciplinary challenge for nanotoxicology has also been pointed out (Krug & Wick, 2011). Actually, various fields related to N&N from the perspective of interdisciplinarity have been studied, such as environmental areas (Bottero et al., 2015), chemistry and physics (Lindquist, 2014), material science (Mody & Choi, 2013), and biotechnology & genomics (Heimeriks, 2013).
Studies concerning the disciplinary structure of N&N are warranted to help set context for analyses of N&N research patterns and knowledge exchange. Porter and Youtie have explored the disciplinary structure of N&N by using Science Citation Index (SCI) journals’ Subject Categories (SCs). They selected nano-related papers by means of a Boolean search in SCI: “nano*,” less exclusions, then plus seven additional modules, detailed by Porter et al. (Porter et al., 2008; Porter & Youtie, 2009). Following this approach, we note that Subject Categories (SCs, WoS version 4) have been supplanted by “Web of Science Categories” (WCs, WoS version 5) launched in August, 2011. We address WCs to accomplish the analysis of the disciplinary structure of N&N in this paper. Compared to SCs, the 222 ISI Subject Categories (SCs) for SCI & Social Sciences Citation Index (SSCI)’s two databases in version 4 of Web of Science (WoS) were renamed and extended to 225 WoS Categories (WCs) (also, a new set of 151 Subject Areas were added, but a higher level of aggregation) (Leydesdorff, Carley, & Rafols, 2013). Thus, we use WCs to detect the disciplinary structure of N&N, further conducting a comparison with conclusions of Porter and Youtie (2009) with the predecessor SCs.
Besides analysis from the perspective of the social network analysis of the disciplinary structure of N&N, cluster analysis by employing cliques embedded in Ucinet software has also been conducted in this paper. This can help understand the disciplinary structure evolution of N&N.
It is of great significance to study the disciplinary structure of N&N both for theory and practice. Theoretically, this study can help us understand the disciplinary and knowledge origins from the beginning of N&N development and trace the trajectory of related subjects’ convergence over time. Practically, it will support research and development (R&D) policy-makers to formulate decisions according to a perspective of converging sciences and technologies.
This paper is organized as follows: Following the introduction, Section 2 introduces data and methods; Section 3 shows the analyses and results; Section 4 states the discussions and conclusions.
Data in this study are retrieved from the database of Science Citation Index- Expanded, SCI-E. N&N has been listed as a WC in SCI-E nowadays, so it is convenient for us to capture articles belonging to the research area of N&N. Articles in the WC of N&N have been searched in the SCI-E database. Our search strategy is as follows: document types = article; WC for nanoscience nanotechnology; time span: 1900–2014; limited to SCI-E. The date of data search and download is July 1, 2015. We retrieved 249,596 resulting records. The yearly distribution of N&N articles is shown in Figure 1.
The WC called
In order to gain an insight into the evolution of the disciplinary structure of N&N in different developmental phases, Statistical Product and Service Solutions (SPSS) software is employed to do the phase-dividing work. Three variables, (different years, the amount of N&N articles published in each year, and the number of distinct WCs of N&N articles in each year) are selected according to the method of Hierarchical Cluster Analysis embedded in SPSS, combining significant events during the N&N developing history, such as the Scanning Tunneling Microscope invented in 1981 (Tersoff & Hamann, 1983; Tersoff & Hamann, 1985), the Atomic Force Microscope invented in 1986 (Binnig, Quate, & Gerber, 1986; Martin, Williams, & Wickramasinghe, 1987), and the US National Nanotechnology Initiative (NNI) taken in 2000 (Roco, 2001; Jung & Lee, 2014). Three stages have been obtained: Stage I: 1966–1980, the infancy phase; Stage II: 1981–1999, the preliminary development phase; and Stage III: 2000–2014, the fast development phase (Figure 1).
We recognize that nanoscience does not really get started in any reasonable way until the advent of the Scanning Tunneling Microscope in 1981 and the Atomic Force Microscope in 1986, so we begin at the second stage timeframe. The evolution of the N&N disciplinary structure during stage II and stage III will be explored, respectively.
Disciplinary co-occurrence matrix reports the relationship among different disciplines of N&N, as operationalized as WCs. The matrix construction is the basic work for analyzing disciplinary network structure and disciplinary cliques here. WCs provide an effective level of measurement of discipline for the study of interdisciplinary processes (National Academies Committee on Facilitating Interdisciplinary Research, 2005). The 225 or so WCs (the number is adjusted slightly over time) reflect sub-disciplines (e.g. Organic Chemistry). WCs have been selected to map science disciplines (Leydesdorff, Carley, & Rafols, 2013), and to do many other bibliometric analyses (Fu & Ho, 2015; Garner, Porter, & Newman, 2014; Lin & Ho, 2015).
An article may involve contributions from two or more disciplines. Keep in mind that the classification into WCs is based on the journal of publication, not on analysis of the individual article. In the SCI-E database, some 40% of journals are associated with multiple WCs; for example, there are six WCs in the following article in the area of N&N.
These six WCs in this record (this is an extreme example; recall that nearly 60% of journals are associated with a single WC) represent the co-occurrence relationship. That is, the record is associated with multiple disciplines (WCs). Bibexcel (Persson & Dastidar, 2013) and Ucinet (Borgatti & Everett, 1999; Freeman, Borgatti, & White, 1991) are jointly employed to get the WC co-occurrence matrix as follows (Table 1). Take the cell crossed by 5 and 8 with value of 1,794 for an example, it means that the co-occurrence frequencies of 5 (Chemistry_Applied) and 8 (Chemistry_Physical) are 1,794 times.
Web of Science category, WC co-occurrence matrix (Partial).1 2 3 4 5 6 7 8 9 10 0 22 0 0 0 0 800 0 0 0 22 0 1,685 1,685 0 0 0 0 0 1,685 0 1,685 0 1,685 0 0 0 0 0 1,685 0 1,685 1,685 0 0 0 0 0 0 1,685 0 0 0 0 0 0 0 1,794 0 0 0 0 0 0 0 0 0 0 0 0 800 0 0 0 0 0 0 7,878 0 0 0 0 0 0 1,794 0 7,878 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1,685 1,685 1,685 0 0 0 0 0 0
After obtaining the WC co-occurrence matrix, we can map the disciplinary network by employing Netdraw (Johnson et al., 2009). The WC co-occurrence matrix we used here is the original matrix derived from the bibliographic data, and the Jaccard index method proposed by Leydesdorff (2008) has not been employed here, for the total disciplines (WCs) concerning N&N are not more than 40, and the disciplinary network structure can be visualized clearly by mapping directly from the original WC co-occurrence matrix. The disciplinary networks help us identify the ties among disciplines engaged in N&N and the evolution of the disciplinary network structure over time. It is simple for us to find out which disciplines have connections with a specific discipline in the network (Figure 2).
The indicator of betweenness centrality (Equation 1) is applied to measure each discipline’s mediating effect according to the path of Network-Centrality-Freeman Betweenness-Node Betweenness, embedded in the Ucinet program, and to further help us understand the mechanism of the evolution of N&N. Betweenness is a centrality measure of a vertex within a graph. Betweenness centrality quantifies the number of times a node acts as a bridge along the shortest path between two other nodes (Freeman, 1977).
If
The higher the betweenness centrality of a specific discipline concerning N&N is, the more contribution this discipline has made to the development of N&N.
Cliques analysis embedded in the Ucinet program is employed to do the disciplinary cluster analysis in the study according to the path of “Network- Subgroup-Cliques.” Cliques are one of the basic concepts of graph theory and are used in many other mathematical problems and constructions on graphs. Since the numbers of disciplines concerning N&N are comparatively limited (no more than 40), and the disciplinary network is composed of a whole component, cliques analysis is found out doing well in a subgroup, and cluster analysis is found after exploring several other schemes (e.g. N-Clan, K-Plex, Lambda Set, Factions, f-Groups, etc.).
The disciplinary network structure of N&N, mapped by employing the social network analysis tool, helps us better understand the dynamic evolution of N&N over time from the perspective of the evolution of the network structure, such as nodes and links added over time. Tree diagrams of N&N disciplines inform the dynamic evolution of N&N from a logical view by showing relationships among different disciplines near or far.
According to the methods illustrated in Section 2.2, we first mapped the disciplinary network structure in two stages: Stage II, the preliminary development phase (1981–1999), and Stage III, the fast development phase (2000–2014); and then we measured each discipline’s mediating effect by selecting the indicator of betweenness centrality in each network.
In this technology’s early development phase (1981–1999), there are a total of 22 WCs participating in the N&N related disciplinary network (Figure 3), and these WCs are connected with each other, forming a whole network.
The map of the disciplinary network of Stage II, 1981–1999, in Figure 3 seems very dense, and the disciplinary network structure appears to be comparatively obvious, with concentrations relating to
In Figure 3, not only can we easily identify those WCs connected with N&N directly or indirectly but also determine if those WCs are also linked to any specific discipline (WC) directly or indirectly. In fact, each WC in Figure 3 can be mapped as an ego-network of disciplinary structure, as shown in Figure 2.
In Figure 3, the relationships among
In Stage II, during 1981–1999, a total of seven WCs have played mediating effects (Table 2). The discipline of nBetweenness refers to normalized betweenness centrality. But keep in mind that the set of records was determined by a search on the N&N WC as such.
Values of betweenness and nbetweenness centrality over 0 of each discipline in Stage II: 1981–1999.Rank ID Betweenness nBetweenness 1 Nanoscience_&_Nanotechnology 143.000 68.095 2 Physics_Applied 11.167 5.317 3 Materials_Science_Multidisciplinary 10.333 4.921 4 Engineering_Electrical_&_Electronic 3.000 1.429 5 Instruments_&_Instrumentation 3.000 1.429 6 Chemistry_Physical 1.000 0.476 7 Physics_Condensed_Matter 0.500 0.238
In Stage III, the fast development phase from 2000 to 2014, more disciplines are added into the disciplinary network (Figure 4). The core N&N network expands to 34 WCs. These WCs are connected with each other, forming a whole network.
The density of the disciplinary network in Stage III is much higher than that of Stage II, especially the left part of the network. 12 new WCs are added in Figure 4 compared to Figure 3; they are
Values of betweenness and nbetweenness centrality over 0 of each discipline in Stage III: 2000–2014.Rank ID Betweenness nBetweenness 1 Nanoscience_&_Nanotechnology 394.833 74.779 2 Materials_Science_Multidisciplinary 35.000 6.629 3 Physics_Applied 9.000 1.705 4 Chemistry_Multidisciplinary 6.500 1.231 5 Engineering_Electrical_&_Electronic 5.000 0.947 6 Instruments_&_Instrumentation 4.500 0.852 7 Biophysics 3.000 0.568 8 Physics_Condensed_Matter 3.000 0.568 9 Chemistry_Physical 2.333 0.442 10 Materials_Science_Characterization_&_Testing 2.000 0.379 11 Biotechnology_&_Applied_Microbiology 1.500 0.284 12 Physics_Fluids_&_Plasmas 1.000 0.189 13 Biochemical_Research_Methods 1.000 0.189 14 Engineering_Manufacturing 0.333 0.063
It is notable that
In order to have an overview of the two development stages, Stage II and Stage III, three indicators (density, average distance, and mean nbetweenness (m-nbetweenness)) are selected to do the comparison, and the results are shown in Table 4.
Comparison between two stages: Density, avg. distance, and m-nbetweenness.Density (Avg. value) Avg. distance M-nbetweenness Stage II 518.636 1.745 3.723 Stage III 804.373 1.836 2.613
Table 4 shows that the value of density is higher in Stage III than in Stage II, indicating closer relationships among N&N related disciplines over time. As far as average distance is concerned, the average distance becomes further as time goes on, which is mainly due to more and more subjects that have appeared. In terms of mean nbetweenness, the value is smaller over time, which is also because of more WCs participating in the arena of N&N over time.
Though the disciplinary networks of N&N in Section 3.1 can tell us what the whole network structure is like, they also help us identify which WCs are connected to a specific discipline. Yet, sometimes the clustering of WCs is not so clear. Thus, clique analysis embedded in the Ucinet program is employed to do cluster analysis, and this will further help us detect the main domains of N&N by selecting the tree diagram display of a subgroup analysis.
According to the method illustrated in Section 2.2, and following the path of Network-Subgroup-Cliques of Ucinet, selecting
Figure 5 shows that the cliques of N&N are clearly identified in Stage II: 1981–1999. There are six branches from the root in the tree diagram, and they are (1)
By using the same method, we get 18 cliques (Figure 6) in Stage III, the fast development phase, 2000–2014, as follows.
Figure 6 demonstrates two imbalanced domains of N&N: A smaller one is about
In this study, some extant studies pertinent to N&N were reviewed, and it was proposed that studies regarding the disciplinary structure of N&N were insufficient. Next, the research purpose and significance were stated, as well as data sources and methods. The data in this paper are from SCI-E with a WC of N&N; a total of 249,596 results of N&N articles are obtained. The methods in this study mainly involve social network analysis and cluster analysis by employing the Ucinet program and Bibexcel software.
The disciplinary network structure reveals relationships among different disciplines in the N&N developing process. We identify the disciplines that are connected with N&N directly or indirectly (and even the disciplines that are linked to a specific discipline). In general, more N&N related disciplines converge into the N&N developing process over time in stages; also, the density of the disciplinary network is closer as time goes on and the average distance is further over time. The value of mean nbetweenness is also smaller. More WCs play a mediating effect with the evolution of different phases of N&N;
The results of N&N cluster analysis show logical relationships among different disciplines related to N&N. The analysis can reveal the original knowledge source at the beginning stage of N&N, the dynamic evolution of N&N over time and also show us relative strength of connections among the different disciplines. With the development of N&N, besides
The novelty of this research lies in mapping the disciplinary network structure of disciplines related to N&N, based on a search using WC in SCI-E. That is also both the strength in focusing on one version of an N&N core, and the limitation in that it does not address the wider swath of R&D that can be identified by a broad, term-based search in such databases. Here, we identify the WCs playing a mediating effect in two stages (especially, analyzing clusters among disciplines related to N&N, revealing close or distant relationships among distinct areas pertinent to N&N). The results help better understand the knowledge sources of N&N at the beginning stage, and also the dynamic evolution of N&N over time.
Compared to similar previous research, core data of the domain of N&N have been selected and analyzed in this paper. There are many studies concerning the interdisciplinary structure of N&N and various subfields and further research could compare results and their implications with such studies to better understand the disciplinary network structure and dynamics (c.f. Porter & Youtie, 2009; Souminen, Li, & Youtie, 2016; Wang & Shapira, 2011).
Another point in this paper is that the WCs (version 5) launched by Thomson Reuters in August 2011 are selected to accomplish the analysis of the disciplinary structure of N&N, supplanting the ISI Subject Categories (SCs) for SCI & SSCI (two databases in version 4 of Web of Science).
Cluster analysis of disciplines related to N&N by employing cliques function embedded in the Ucinet program helped understand the evolutionary mechanics of N&N. The results help illuminate how the area of N&N developed, and which disciplines have converged into N&N over time.