The co-citation network was divided into many clusters of co-cited references in CiteSpace, so that references are closely connected within the same cluster but loosely connected among different clusters (
Figure 7). The 10 major clusters are listed in
Table 3 by size, which represents the number of members in each cluster. The silhouette score of a cluster reflects its quality, i.e., homogeneity or consistency. If the silhouette value of a cluster is close to 1.0, then it was homogenous [
22]. All the clusters in
Table 3 were highly homogeneous, as indicated by their high silhouette scores. Noun phrases from the terms (e.g., titles or abstracts) used in articles in the cluster were used to label each cluster. Labels selected by the log-likelihood ratio (LLR) test were used in subsequent discussions [
35].
We can identify the average year of the publications in a cluster by their recentness, i.e., Cluster #6 on aquifer remediation had an average year of publication of 1985. The recently formed clusters, Clusters #2 and #3 (nano-zero‑valent iron and metallic iron, respectively), had an average year of publication of 2009 and 2008, respectively.
3.6.1. Analyses of Research Fronts
Price [
36] proposed the concept of the research front, and postulated that a research front can characterize the momentary nature of a research field. Garfield [
37] defined a research front as “a cluster of co-cited articles and all articles that cite the cluster”. Chen [
38] defined a research front as “an emergent and transient grouping of concepts and underlying research issues”. CiteSpace shapes the network knowledge map of research fronts, with mutant terms that can be extracted from the index terms, abstracts, titles, and record indicators of the references. Specific methods include selecting a cited reference as the net node, the g-index (
k = 10) as threshold willing, and the key pathfinder algorithm. We obtained 17 clusters by selecting “Find clusters” and abstracted the names of the clusters by selecting “Label clusters with indexing terms”.
Figure 7 shows the net knowledge map generated.
There were 558 nodes, 874 links, and 17 clusters. Clusters #2, #3, and #4 had a high concentration of nodes with citation bursts, which echoed the fact that these were the most recently formed clusters.
If a cluster has a larger area, it has more bibliographic entries, and large clusters generally indicate main research directions, i.e., each cluster corresponds to a research front. The research fronts and major trends in groundwater extraction, in situ groundwater remediation, permeable reactive barriers, metallic iron, and nanoscale zero‑valent iron particles are shown in
Figure 7 and
Table 3.
3.6.2. Timeline of Research Fronts
Figure 8 shows timelines of the 17 distinct co-citation clusters and their interrelationships. All timelines run from left to right [
39], show the times at which research fronts appear and disappear, and display structural information about the research front clusters [
40]. Analysts can visually identify various characteristics of a cluster, such as its citation classics, historical length, citation bursts, and connection to other clusters.
The following paragraphs provide an interpretation of the research fronts’ timelines. Groundwater remediation research has a long history (
Figure 8). The earliest research front, “aquifer remediation”, provided basic information for subsequent research. Next, technologies to deal with contaminated groundwater were developed. The containment and/or control of contaminated groundwater can generally be accomplished using one, or a combination, of several available techniques, which can be broken down into aquifer rehabilitation, physical containment measures, and withdrawal, treatment, and use [
41].
The second research front, “groundwater extraction”, began around 1979, and pump‑and‑treat as a groundwater extraction technology began at selected sites in 1982 [
42], in response to groundwater pollution control and contamination remediation. Earlier pump‑and‑treat systems, which did not consider the presence of geologic heterogeneity, poor definition of initial condition in source zones, did not clean aquifers to the required level. Many of the original systems worked adequately for a period of time, but, after they were switched off, the contaminant levels at many sites reached values higher than those before remediation [
31]. Subsequent to the pivotal 1989 article by Mackay [
43], the research front became inactive. A number of new technologies for groundwater remediation are under development, and these may accelerate contaminant removal from the subsurface (e.g., injection of steam, surfactants) or destroy the contaminant in situ [
43]. Hence, research fronts are discontinuous, and start and end abruptly when scientists move from one puzzle to the next [
40].
In the 1990s, scientists and engineers had to prepare to deal with recent puzzles, which included residual oils, source zones with non-aqueous phase liquids (NAPLs), and vapours in the unsaturated zone [
31]. During this period, four research fronts, “anionic surfactant remediation (1995)”, “decision analyses (1997)”, “laboratory column test (1999)”, and “in situ groundwater remediation (1999)”, were created in response to growing concern over efficient and cost-effective clean‑up solutions. Among these research fronts, “in situ groundwater remediation” was worthy of note. This research front experienced a period of stability and extends to the present. A growing number of researchers focused on the development of in situ remediation technologies, e.g., in situ chemical oxidation [ISCO]. ISCO, a type of advanced oxidation process technology, has proven useful for in situ remediation technology for the most prevalent organic contaminants in groundwater. The development of in situ remediation technologies led to the formation of three research fronts in the new century: “permeable reaction barriers (PRBs) (2002)”, “metallic iron (2008)”, and “nano zero‑valent iron (nZVI) (2009)”.
The research front, “PRB”, which dates to 1989, had a median publication date of 2002. Over the last two decades, PRBs have been emerging as an effective alternative passive in situ remediation technology. In the 1990s, research on PRBs increased considerably, which led to many new approaches for suitable reactive materials, target contaminants, and PRB design.
“nZVI” is the latest research frontier, showing rapid growth and a professional pattern. Gillham and O’Hannesin [
44] discovered that halogenated aliphatic compounds in groundwater can be reduced using bulk ZVI. This characteristic of iron led to the advanced Fe-PRB, in which vertical trenches were filled with granular ZVI, placed in the flow path of the underground contaminant plumes [
45,
46].
A report [
47] by the Chinese Academy of Sciences indicated that the third top research front in ecology and environmental sciences, entitled “Activation of persulfate for degradation of aqueous pollutants by transition metal and nanotechnology”, is receiving much global attention. The combination of persulfate ion activation technology and nanotechnology will improve the efficiency of polluted water treatment, reduce energy consumption, and promote recycling.
3.6.3. Analyses of the Intelligence Base
Chen [
38] defined “the intellectual base of a research front as its citation and co-citation footprint in the scientific literature, an evolving network of scientific publications cited by research-front concepts”.
(1) Most‑Cited Articles
The most‑cited articles are generally considered landmarks, owing to their ground‑breaking contributions [
22]. Cluster #7 had three of the top 10 landmark articles, and Clusters #3 and #10 each had two (
Table 4). The most-cited articles in the databases were by Blowes (2000), with 154 citations, followed by Gillham (1994), with 144 citations and Matheson (1994), with 135 citations, and the most recent was a review article by Fu (2014). Interestingly, the titles of the most‑cited articles contained the terms “permeable reactive barriers”, “zero-valent iron”, “nanoscale iron particles” (
Table 4), which were in accordance with the research fronts noted above.
Gillham and O’Hannesin [
44] investigated the potential of Fe
0 in the dehalogenation of ethanes, ethenes, and 14 chlorinated methanes. The results demonstrated biotic reductive dechlorination, in which iron serves as the source of electrons. In response to the rapid degradation rates, an application for in situ remediation of contaminated groundwater was proposed.
Blowes, et al. [
48] was cited the most frequently. This paper reviewed the recent research progress in PRBs for the remediation of inorganic contamination of groundwater.
(2) Betweenness Centrality
The betweenness centrality measure that Freeman [
49] proposed is used to give prominence to potential pivotal points in the synthesized network shifts over time. The betweenness centrality of nodes in a network is indicative of the importance of the location of the nodes. We are especially interested in the nodes located between different node groups, because they probably offer insight into emerging trends [
22].
Table 5 shows 10 structurally crucial references in the network, and three of these nodes were in Cluster #3, and five in Cluster #7. These references can be identified as landmark works in the field of groundwater remediation.
(3) Citation Bursts
A reference citation burst may indicate an emergent research front, and the citation-burst-detection algorithm of Kleinberg [
50] is adapted for identifying emergent research front concepts.
Table 6 lists the references that had the strongest metric of citation bursts across the entire database during the period 1950–2018. Among the articles with strong citation bursts (
Table 6 and
Figure 9), Mackay and Cherry [
43] is worthy of note. Their article explored the reasons for the difficulty of groundwater clean-up, noted some implications, and suggested that achieving stringent health-based clean-up standards is unlikely, and the ultimate cost of clean-up is high in many cases. Thus, they suggested that site characterization and remediation have much room for improvement, by both the development of new tools and ongoing training of staff [
43]. Subsequently, the development of permeable reactive barrier technology using zero-valence iron filings has proceeded from recognition, evaluation, technology conceptualization, and proof of concept, to commercialization.
(4) Sigma
The structural centrality and citation burstness of cited references can be measured by the Sigma metric measure, i.e., the Sigma value of a reference that is strong in both measures will be higher than that of a reference that is strong in only one of the two measures [
22] (
Table 7). The pioneering article by Fu, et al. [
51] had the highest Sigma of 101,578.09, indicating it to be structurally indispensable in the field, due to its strong citation burst. This article reviewed the recent advances of ZVI and the progress made in groundwater remediation using ZVI technology.