-
Bibliometric Mapping: Eight Decades of Analytical Chemistry,
WithSpecial Focus on the Use of Mass SpectrometryIn this Feature we
use automatic bibliometric mapping tools to visualize the history
of analyticalchemistry from the 1920s until the present. In
particular, we have focused on the application of massspectrometry
in dierent elds. The analysis shows major shifts in research focus
and use of massspectrometry. We conclude by discussing the
application of bibliometric mapping and visualizationtools in
analytical chemists research.
Cathelijn J. F. Waaijer*, and Magnus Palmblad
Centre for Science and Technology Studies, Faculty of Social and
Behavioural Sciences, Leiden University, P.O. Box 905,2300 AX
Leiden, The Netherlands
Center for Proteomics and Metabolomics, Leiden University
Medical Center, 2333 ZA Leiden, The Netherlands
*S Supporting Information
Bibliometrics is the study of interrelated bodies of docu-ments,
a prime example being the scientic literature. Oneof its best known
applications is the comparative evaluation ofcountries,
universities, research institutes, and individualresearchers, but
it may also be used for other purposes, such asgaining a better
understanding of a elds structure or deter-mining developments in
research topics. It is the latter applica-tion that we will
highlight in this Feature, by using automaticbibliometric mapping
tools to map developments within ana-lytical chemistry.Compared to
more traditional historical methods, automatic
bibliometric mapping of scientic literature has the advantage
ofrelative ease and low laboriousness. Furthermore, the eld
struc-ture is established by (almost) automatic methods, producing
amore objective result than manual mapping could.Mapping of
networks visualizes multiple items (nodes) and
their underlying relationships (edges). The nodes can be
dif-ferent entities, e.g., authors, journals, or key terms
occurring inresearch papers. In addition, the edges can be based on
dierenttypes of data, e.g., in a network of authors one could
determinewhich authors co-author papers (a co-authorship network)
orwho cites whom (a citation network).The rst bibliometric maps
were manually constructed cita-
tion networks.1 Gareld, Sher, and Torpie studied a book on
thehistory of genetics and compared the dependencies betweendierent
studies as described by the author to the citationalpatterns
between the studies, and found that the two methods
closely mirror each other. Hence, citation networks are able
toshow structures in knowledge ows.Ipso facto citation networks
have a temporal aspect: a
publication can only refer to earlier published work.
However,the temporal aspect is often not explicit in citation
networks astime is not explicitly shown in the visual
representation of thenetworks. Exceptions include main path
analysis for citationnetworks2 and the HistCite3 and CitNetExplorer
software tools.4
All of these show chronological maps of the main lines
ofresearch through time, but by means of dierent methods.Although
citation networks do represent an underlying
structure of (elds of) scientic knowledge, they do not
directlyrepresent the content of papers. To this end, co-word maps
canbe constructed. In this approach terms are extracted from
papers(e.g., from titles and abstracts) and for each pair of terms
thenumber of papers in which they both occur is determined.
Termsthat appear often together are likely to concern the same
subjectmatter, whereas terms that never appear together are
unlikely tobe related subject-wise. Counting the co-occurrences for
everypair of terms yields a co-occurrence matrix of terms. Further
stepsinclude the normalization of this matrix and its
visualization.57
In this Feature we use co-word and citation networks ofdierent
bodies of documents concerning analytical chemistry toshow shifts
in research topics within this eld. Special attentionis given to a
method that became increasingly important inanalytical chemistry,
mass spectrometry (MS).
EVOLUTION OF TOPICS IN ANALYTICALCHEMISTRY 19292012
In this study we use three sources of data as input for
bibliometricanalysis (Figure 1). The rst is this journal, founded
in 1929as Industrial & Engineering Chemistry Analytical Edition
andrenamed to Analytical Chemistry in 1947. Most if not
allscientometric studies employing mapping methods to analyzethe
development of research elds have used titles and abstractsobtained
from large bibliographic databases, of which the best
Published: March 6, 2015
Feature
pubs.acs.org/ac
2015 American Chemical Society 4588 DOI: 10.1021/ac5040314Anal.
Chem. 2015, 87, 45884596
-
known are Web of Science (WoS), founded in 1964 by Gareldas the
Science Citation Index and acquired by Thomson Reutersin 1992, and
Scopus, launched in 2004 by Elsevier. However,abstracts were not
regularly included before 1990. To map theevolution of research
topics within analytical chemistry over along period, it was
therefore impossible to use ready-to-usedatabases such as WoS or
Scopus. Instead we extracted the titlesand abstracts of all
Analytical Chemistry papers publishedbetween 1929 and 2013, using
this journal as a proxy forthe eld of analytical chemistry.8 A
detailed description of theabstract extraction procedure, and of
the other methods used inthis paper, can be found in the Supporting
Information.
A glossary of the discussed terms, techniques, and softwaretools
is given in Table 1.First, we constructed co-word maps of the eld
of analytical
chemistry in each decade: 19291940, 19411950, 19511960,19611970,
19711980, 19811990, 19912000, and 20012012 and visualized these in
the software VOSviewer (clickon a decade to download the
correspondingmap as an interactiveJava application; JavaScript 6 or
higher required).9 On thesemaps, terms that occur together often
are positioned closeto each other, whereas terms that co-occur less
often arepositioned further apart. Furthermore, clustering of terms
intofour to eight clusters with dierent colors is applied using
analgorithm that nds the clustering solution that ts the
co-occurrences between the dierent terms best
(modularity-basedclustering).10
The 19291940 map shows four dierent clusters:apparatuses (in
green), gases (in pink), inorganic chemistry (inred), and
industrial applications, hydrocarbons, and food analysis(in cyan)
(Figure 2). For the other decades, the description of theclusters
is given in Table 2. The maps show that inorganicchemistry (red)
has been an important topic within analyticalchemistry for a long
time; from 1929 until 1990 there were one ormore clusters on
inorganic chemistry. In the 19912000 periodit was merged with the
topics of electrochemistry and sensors.Much attention was given to
(the development of) dierentapparatuses between 1929 and 1980
(green). A cluster ongeneral and editorial issues can be found in
almost every period(yellow). Topics that have developed over time
includeelectrochemistry, chromatography, and mass
spectrometry.Electrochemistry shows up as its own cluster in the
19511960 period (sea green), but terms relating to the subject can
alsobe found in the inorganic chemistry and metals cluster
from1941. This suggests the topic of electrochemistry has
developedfrom inorganic chemistry and metals to form its own
subeld.Chromatography is apparent in the maps from the
19511960period onward (cyan); mass spectrometry from the
19711980period. The maps suggest the widespread use of mass
spec-trometry in analytical chemistry primarily developed through
itscoupling to chromatography (cyan); for the 19711980 periodterms
relating to mass spectrometry can be discerned in themaps, but the
cluster is still dominated by chromatographictechniques and
applications. However, from the 19811990period, mass spectrometry
broke o and formed its own subeld(blue). Finally, from 2001 a
cluster on separations and micro-uidics emerged (mustard). This
cluster also contains termsrelating to theory and simulations (of
such microuidic systems).Next, we analyzed the development and use
of a number of
techniques within analytical chemistry. As a proxy, we
deter-mined how many articles mentioned the technique in their
titlesduring the 19292012 period. It is important to note that this
isonly a proxy and as we only look for the mention of techniques
inthe article titles, there is bias toward novel uses and
developmentof the technique.This approach shows that titration
techniques reached their
publication peak in the 1950s, gas chromatography in the
1960s,and liquid chromatography in the 1980s (Figure 3). Of
thesetechniques, only the latter was still mentioned in the titles
of over5% of papers published in the 20012012 period. On the
otherhand, microuidics is an example of a technology not
mentionedbefore 1990 that has really taken o in this 20012012
period.A technique not mentioned to a great extent in the titles
ofAnalytical Chemistry papers is nuclear magnetic resonance(NMR),
despite the fact that according to historical studies, it
Figure 1. Overview of data and methods. The rst data source was
allarticles published in Analytical Chemistry between 1929 and
2012. Titleswere extracted from metadata in XML-format and these
titles were usedto nd the start of each article on the scanned and
OCRed rst pages ofeach article. By nding the start of each article
the abstracts could beextracted. For 19962012, the abstracts were
available in XML-formatand extracted from these les. Titles and
abstracts were used to make co-word maps using the VOSviewer and to
determine the use of severaltechniques. The second data source was
the Centre for Science andTechnology Studies (CWTS) version of the
Web of Science, whichapplies the NOWT classication to group
journals into scientic elds.This source was used to determine the
contribution of dierent scienticelds in MS research and determine
the relative use of MS in eachscientic eld. The third source was
also the Web of Science, but theonline version (Web of Knowledge),
which holds the metadata ofscientic articles going back to 1945.
This source was used to construct alongitudinal citation network of
MS research.
Analytical Chemistry Feature
DOI: 10.1021/ac5040314Anal. Chem. 2015, 87, 45884596
4589
-
became an important physics-based analytical method in thesecond
half of the 20th century.11 Presumably, chemical researchusing NMR
was published in journals other than AnalyticalChemistry. As the
co-word maps already suggested, the mentionof mass spectrometry
increased throughout the entire period.
Whereas between 1929 and 1940 none of the AnalyticalChemistry
papers mentioned mass spectrometry in their title,the fraction of
papers that did increased to 18% in the 20012012 period (Figure 3),
revealing a continuous increase inrelative importance of mass
spectrometry in analytical chemistry.
Figure 2. Evolution of the eld of analytical chemistry. Maps
based on all texts published in Analytical Chemistry except for
advertisements (19291995)or on all articles, letters, and reviews
published in Analytical Chemistry (19962012). The colors depict the
cluster the term belongs to (cf. Table 2). Thesize of the circle is
proportional to the number of occurrences. The distance of two
terms on the map reects the relatedness of the two terms, i.e.,
howoften they co-occur.
Analytical Chemistry Feature
DOI: 10.1021/ac5040314Anal. Chem. 2015, 87, 45884596
4590
-
Our bibliometric ndings are by and large in accordance
withhistorical studies on the development of analytical
chemistry.Historical studies have also shown that analytical
chemistry hasundergone major changes during the 20th century. The
mainchange has been a shift of focus from chemistry- to
physics-basedanalytical methods.11,12 Before the instrument era,
analyticalchemists would rst have to separate their compound of
interestfrom the sample using chemical reactions, after which they
couldqualitatively identify the elements present in the compound.
Therst (now called classical) methods of quantitative
determi-nation involved gravimetric and volumetric analysis. In the
1940sand the early 1950s, these methods were still the most
used,together with colorimetric methods.13,14 The classical
methodswere still used in the late 1950s and 1960s but
increasinglysupplanted by chromatography, electrophoresis, and
MS.15,16
Another development during the investigated period was
thedecrease in the share of papers on inorganic and
organicchemistry due to a surge in biochemical research, a
ndingmirrored in our maps.
DEVELOPMENT AND USE OF MASSSPECTROMETRY
As mass spectrometry became an important technique for
ana-lytical chemistry, we now zoom in on the use of this
technique.Co-words maps of all Analytical Chemistry papers
mentioningMS in their title or abstract were constructed (Figure
4). Topicsof clusters in the map are described in Table 3. The maps
show ashift from analysis of smaller molecules (gases,
hydrocarbons,metals) and isotope analysis to the analysis of larger
and morecomplicated molecules (polymers, proteins). Main topics in
the
Table 1. Glossary of Terms, Techniques and Software Tools
Bibliometrics The quantitative study of literatures as they are
reected in bibliographies22
Citation network Network of citation relations between items
(e.g., publications, authors or journals)CiteSpaceII Software tool
developed by Chen for detecting and visualizing emerging trends and
transient patterns in scientic literature5
CitNetExplorer Software tool developed by Van Eck and Waltman
for visualizing and analyzing citation networks of scientic
publications4
Mapping Positioning of a subset of the publications in a
citation network (usually selected based on their citation
frequency) in a two-dimensional mapin which the vertical dimension
indicates time (i.e., the year of publication) and the horizontal
dimension indicates the closeness ofpublications in the citation
network.
Clustering Partitioning of the publications in a citation
network into a number of groups (clusters). Publications assigned
to the same group are closelyconnected to each other in the
citation network.
Co-word map Map of words (or terms), usually extracted from the
titles and abstracts of scientic publications, showing the
co-occurrence relations of thewords (i.e., the number of
publications in which two words occur together).
HistCite Software tool developed by Eugene Gareld to generate
chronological maps of scientic literature based on WoS input3
Sci2 Software tool developed by a team led by Borner and Boyack
that is a modular toolset specically designed for the study of
science. It supportsthe temporal, geospatial, topical, and network
analysis and visualization of scholarly datasets at the micro
(individual), meso (local), andmacro (global) levels.
VOSviewer Software tool developed by Van Eck and Waltman for
analyzing bibliometric networks,9 in particular networks based on
citation and co-occurrence relations
Mapping Positioning of the items in a network in a
two-dimensional map in such a way that strongly connected items
tend to be located close to eachother while weakly connected items
tend to be located further away from each other. The horizontal and
vertical axes have no specialmeaning. Only the relative distances
between items carry meaning in a map.
Clustering Partitioning of the items in a network into a number
of groups (clusters). Items assigned to the same group are closely
connected to each otherin the network.
Web of Science (WoS) Multidisciplinary bibliographic database
produced by Thomson Reuters
Table 2. Main Topics in Analytical Chemistry (cf. Figure 2)
Color Description Color Description Color Description Color
Description
19291940 19411950 19511960 19611970Green Apparatuses Green
Apparatuses Cyan Chromatography Cyan ChromatographyPink Gases Pink
Inorganic chemistry:
gases/halogensSea green Electrochemistry Red Inorganic
chemistry
Red Inorganic chemistry Red Inorganic chemistry:metals
Red Inorganic chemistry: metals Sea green Electrochemistry
Cyan Industrial applications,hydrocarbons and food
Dark blue Organic and foodchemistry
Green Apparatuses Yellow General/editorial andinformatics
Yellow General/editorial Yellow General/editorialCyan Industrial
applications
and hydrocarbons
19711980 19811990 19912000 20012012Cyan Chromatography Yellow
General/editorial Cyan Chromatography Sea green Detection,
electrochemistry and
(bio)sensorsRed Inorganic chemistry Sea green Electrochemistry
Purple Electrophoresis Brown Small molecules and
quantitationSea green Electrochemistry Red Inorganic chemistry
Sea green Inorganic chemistry, electrochemistry
and (bio)sensorsBlue Mass spectrometry
Yellow General/editorial Cyan Chromatography Yellow
General/editorial Mustard Separations, microuidics, andtheory and
simulations
Green Apparatuses Blue Mass spectrometry Blue Mass spectrometry
and proteomicsPink Gases
Analytical Chemistry Feature
DOI: 10.1021/ac5040314Anal. Chem. 2015, 87, 45884596
4591
-
19411960 period are hydrocarbons, structural
analysis,quantitation, and gases (again, click the period to
download aninteractive Java application of the map). The 19611970
and
19711980 periods are characterized by the emergence ofsoftware
and by the development of apparatuses and interfaces.In addition,
the 19711980 period saw the establishment ofchromatography and
chemical ionization as important ancillarytechnologies. Terms
relating to secondary ionmass spectrometry(SIMS) can rst be
distinguished for the 19711980 period (aspart of a cluster that
also includes quantitation) and also formed acluster in the
19811990 period (also including laser and plasmadesorption) and the
19912000 period (also including theanalysis of polymers). Another
main topic that emerged in the19912000 period is proteomics. In
this period it formed acluster with matrix-assisted laser
desorption ionization (MALDI)while also being positioned close to
the cluster on electrosprayionization, quadrupoles, ion traps,
Fourier transform ioncyclotron resonance (FTICR), and tandem mass
spectrometry(MS/MS) but became its own cluster in the 20012012
period.MALDI then formed a cluster with imaging mass
spectrometry.The main nding from these co-word maps is again the
shift
from the analysis of simple to more complex molecules, as
was
Figure 4. Evolution of MS within analytical chemistry based on
co-word maps. Maps based on all texts with the term mass spectro*
in the title and/orabstract published in Analytical Chemistry
except for advertisements (19291995) or on all articles, letters,
and reviews with mass spectro* publishedin Analytical Chemistry
(19962012). The colors depict the cluster the term belongs to (cf.
Table 3).
Figure 3. Use of dierent techniques in Analytical Chemistry.
Searchterms used were mass spectro*, nuclear magnetic resonance
orNMR, titration, gas chromato*, liquid chromato*, and micro-uid*,
searched against the titles of Analytical Chemistry papers.
Analytical Chemistry Feature
DOI: 10.1021/ac5040314Anal. Chem. 2015, 87, 45884596
4592
-
the case for analytical chemistry generally. The maps also
providethe likely explanation that enabled this shift: the
improve-ment of equipment, interfaces and software, and especially
thedevelopment of new physical techniques. Examples are
thedevelopment of chemical ionization and SIMS in the 1970s,
andMALDI in the 1990s. In addition, joining of older techniques
andMS is evident from these maps, such as the incorporation
ofchromatography in the 1970s. These developments in turnenabled
the applications ofMS in a wider area of research, such
asproteomics (enabled by several ionization techniques) andpolymer
analysis (enabled by SIMS), which is also visible in themaps.Next,
we set out to estimate how often MS was used over time
in all research elds, insofar as they are covered by the
WoSdatabase. We determined how many articles in the WoS had theterm
mass spectrometry in their title or abstract. The plottedgraph of
this absolute number shows a large increase from 1981until 2013
(Figure 5a, pink plot). The graph is discontinuousbetween 1990 and
1991 due to abstracts being regularly includedinto the WoS database
only from 1991, including for journalspublishing many papers using
MS, such as Journal of BiologicalChemistry, Journal of
Chromatography, and Rapid Communica-tions in Mass Spectrometry.
After the 1991 discontinuity, theincrease in use of MS is still
considerable. In 1991, about 2 600papers on or including MS were
published; by 2013, this gurehad increased to around 16 000.
However, the WoS databaseexpanded tremendously between 1981 and
2013, due to anincreasing number of journals being covered and each
journal onaverage publishing more articles per year (Figure S-1 in
theSupporting Information). Therefore, we also determined
therelative number of articles with MS in their titles or
abstracts.This analysis shows that MS was indeed increasingly used
inrelative terms (Figure 5a, yellow).This raises the question which
scientic disciplines work on
(and with) MS. We determined which disciplines mainly
con-tribute to research involving MS. To this end we used theNOWT
classication of journals, which classies journals
according to scientic disciplines (see Table S-1 in
theSupporting Information for a complete list).17 This reveals
thatmost research using MS has been published in journals from
thechemistry, physics, and astronomy category (Figure 5b).18
However, the emphasis of these disciplines has decreasedover
time as the use of MS in the life sciences increaseddramatically
between 1991 and 2013. Furthermore, therehas been a slight decrease
in the share of papers published inengineering journals.This
analysis, however, only measures the share of scientic
disciplines in the total output of research using MS.
Therefore,we also determined to what extent MS was used per
scienticdiscipline. Results of the latter analysis show that the
use of MS inthe chemistry, physics, and astronomy category has
still beenincreasing over the past 15 years, albeit slowly (Figure
5c). Incomparison, there has been a large increase in the use of
MSin the life sciences, from about 0.75% of all papers in life
sciencesjournals to over 2.5%. The use of MS in the medical
sciences hasalso increased, as it has in earth and environmental
sciences, butin the latter, the rate of growth has decreased. As
mentionedabove, it is important to keep in mind that text mining
from titlesand abstracts is more likely to pick up new applications
or noveltechnologies rather than routine, established use, where
theymayonly appear in the methods section.Finally, we investigated
which lines of research have been the
most important in research employing MS and which papershave
been most inuential. To this end, we visualized alongitudinal
citation network using CitNetExplorer.4 All articles,reviews, and
letters with the term mass spectrometry in thetitle, abstract, or
listed keywords published in the online versionof the WoS, along
with their cited references, were included intothe analysis. The
inclusion of cited references makes it possible toalso include
scientic work that is not included in the WoS, suchas textbooks,
older articles, and articles not employing MS butcited by mass
spectrometrists, in the citation network. For ashort explanation on
mapping and clustering, see the glossary(Table 1).
Table 3. Main Topics in Mass Spectrometry within the Field of
Analytical Chemistry (cf. Figure 4)
Color Description Color Description Color Description
19411960 19611970 19711980Yellow General and editorial Mustard
Software Cyan ChromatographyPurple Hydrocarbons Green-brown Sample
preparation, separations and
derivatizationBrown Compound quantication and
secondary ion MSDark blue Structural analysis Purple
Hydrocarbons and organic chemistry Green Apparatuses and
interfaces
(incl. informatics)Brown Quantitation Green Apparatuses and
interfaces Sea green Chemical ionizationPink Gases Blue General
MSGray Nondiscernible Red Inorganic chemistry, metals and isotope
ratio MS
Yellow Editorial
19811990 19912000 20012012Brown Compound quantication Dark blue
MALDI-TOF and proteomics Blue MALDI and imaging mass
spectrometryCyan Chromatography Purple Chromatography,
quantitation and isotope ratio
MSSea green Direct analysis (DART etc.),
ESI and ICPMSSea green Chemical ionization Red SIMS, surfaces
and polymers Brown Quantitation (GCMS,
LCMS)Pink FAB, FD/mass analyzers and
MS/MSSea green Electrospray ionization, quadrupoles, ion
traps,
FTICR and MS/MSGreen-brown Sample preparation (labeling,
enrichment, purication)Green-brown Secondary ion mass
spectrometry, laser desorptionand plasma desorption
Dark blue Proteomics
Gray Nondiscernible
Analytical Chemistry Feature
DOI: 10.1021/ac5040314Anal. Chem. 2015, 87, 45884596
4593
-
The citation network displays 10 clusters, of which the
largestis on peptides and proteins (Figure 6). The network showsa
tightly knit group of clusters on peptides and proteins (darkblue)
and metabolites (orange) and part of a cluster on technicalaspects
of MS (dark green). Clusters of research on carbo-hydrates (red)
and lipids, plants and neonatal metabolism(brown) are also quite
closely related to this group. The foodanalysis cluster (pink) is
connected both to the aforementionedmetabolite cluster and to a
cluster on environmental research(purple). Further right on this
map is part of the cluster onsurfaces and polymers (bright green);
the other part on imagingmass spectrometry is mapped more closely
to the biologicalgroup. Finally, there are discrete clusters on
atmospheric andgeological science (cyan) and isotopes (yellow).In
addition we determined which research paper per cluster
is the most cited (excluding reviews and book chapters).
Thisanalysis attempts to nd the main papers inuencing
applicationsof mass spectrometry in dierent elds. It turns out that
fora considerable number of clusters, the most cited research
paperis not one specically on MS, e.g., Laemmlis paper on
proteinquantitation, Arthur and Pawliszyns on the solid
phasemicroextraction of organochlorides, Bligh and Dyers on
lipid
extraction, Van den Dool and Kratzs on gas chromatography,and
Guenther et al.s on gas emission (Figure 5b), againillustrating MS
is frequently combined with other methods.In conclusion, analysis
of the scientic papers in which MS is
mentioned reveals a shift from development of the technology
bythe physics and chemistry communities to application in the
life,medical, and earth and environmental sciences. This shift
isevident from a slight increase in the use, or further
development,of MS within physics and chemistry but a much larger
increasein the more applied elds. The most cited papers in
researchapplying MS are often not concerned with MS, but withallied
technologies, illustrating the interdisciplinary natureof much of
the research using MS. The results from thisbibliometric analysis
trace the common historical narrative ofmass spectrometry, e.g., as
described in Graysons MeasuringMass: From Positive Rays to
Proteins.19 This book also cites thecommercialization of mass
spectrometers as the main enablingfactor behind the expansion of
applications of MS. Thiscommercialization was rst fueled by the oil
industry and theManhattan Project but later driven by a need for
pharmaceuticaland environmental analysis.
Figure 5. Use of mass spectrometry in scientic literature and by
scientic discipline, 19812013. (a) Number of MS papers in WoS
database. Thesearch term used was mass spectrometry, which was
searched for in the title and/or the abstract. (b) Share of
scientic disciplines in MS research.(c) Percentage of papers using
the term mass spectrometry in title and/or abstract, per scientic
discipline. Scientic disciplines are based on NOWTmedium
categories, fractionally counted.
Analytical Chemistry Feature
DOI: 10.1021/ac5040314Anal. Chem. 2015, 87, 45884596
4594
-
CONCLUSIONSBibliometric mapping reveals clear shifts in
analytical chemistryresearch topics from inorganic and
(small-molecule) organicchemistry to biochemistry and complex
biomolecules. Fur-thermore, a sequence of new methods emerging and
sometimesreplacing older ones is apparent in the time series
analyses. It isimportant to note that our ndings are, for the most
part, neithernew nor even unexpected but rather support historical
researchand the results from other bibliometric methods that have
beenused to investigate the development of analytical
chemistry.However, we show here how these ndings can be obtained
usingsemiautomatic bibliometric mapping methods to visualize
theevolution of research elds in an unsupervised manner.
BIBLIOMETRIC VISUALIZATION TOOLS FOR YOUROWN RESEARCH
The use of bibliometric visualization tools is not limited
tohistorical analyses such as the one presented here. The tools
canalso be used to obtain a comprehensive view of other
researchelds. This is especially useful for junior researchers who
aregetting started in the eld and would like to have a rst glance
atits structure. Furthermore, by charting time series new topics
ofpotential interest to the researcher can be found. Co-word
mapsare especially well-suited to obtain an overview of a eld, as
theyshow the main terms used in that eld.However, there are
limitations to co-word analysis, e.g.,
researchers use dierent writing styles, and terminology,homonyms
and synonyms all aect the co-occurrence of terms.
In addition, mapping by denition is a simplication, whichcauses
loss of information.20 For example, electrospray ionizationhas
arguably been a major development in the development ofMS.21 It was
rst developed for recording mass spectra of largebiomolecules by
Yamashita and Fenn already in 1984. However,the terms electrospray
ionization and ESI are not depictedon our 19811990 maps on MS, but
only from 1991 to 2000,presumably because the number of occurrences
had not reachedthe set threshold of minimum occurrences for the
19811990period. Hence, only the largest subelds are generally
visible onthe map, whereas the smaller ones (that might actually
hold themost exciting developments) are not. Furthermore, the
divisionof terms into clusters is to some extent subject to the
chosenclustering parameters, making clustering dependent on a
certainsubjectivity. Still, the main structure of a research eld is
easilymapped.Bibliometric visualization tools may be useful to
uncover
unexpected linkages to other elds and scientic literature
aswell. Citation network visualization is useful to analyze how
abody of documents is related and which other work it drawsupon.
The ability to uncover linkages in the scientic literaturemay be
especially of use when writing a review of the
literature.CitNetExplorer is a useful tool in this regard, as it
can also showreferences to articles not included in the input
data.In this work we used two dierent bibliometric
visualization
tools that suited our purposes, VOSviewer and
CitNetExplorer.4,9
However, there are many other mapping tools (also
freelyavailable) that have more extensive functionality for
otherpurposes. Examples include CiteSpaceII and Sci2.5,6
CiteSpaceII
Figure 6. Longitudinal citation network of mass spectrometry
research, 19452013. The colors represent the cluster a publication
belongs to thecolored numbers represent the cluster numbers in the
table. Labels show the last name of the last author of a
publication.
Analytical Chemistry Feature
DOI: 10.1021/ac5040314Anal. Chem. 2015, 87, 45884596
4595
-
provides a built-in database for data handling as well as
geospatialanalysis features. Sci2 has functionality for geospatial
analysis andnetwork analysis, such as calculation of in-degree,
k-core, andcommunity detection.
ASSOCIATED CONTENT*S Supporting InformationAdditional
information as noted in text. This material is availablefree of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATIONCorresponding Author*Phone: +31 71 527 6072.
Fax: +31 527 3911. E-mail:[email protected]
authors declare no competing nancial interest.BiographiesCathelijn
Waaijer works as a doctoral researcher at the Centre forScience and
Technology Studies at Leiden University. She did herM.Sc. in
Biomedical Sciences and did a research project on the
massspectrometric analysis of cartilage tumors at the Leiden
UniversityMedical Center (LUMC).
Magnus Palmblad is Associate Professor at the LUMC Center
forProteomics andMetabolomics specializing in clinical applications
of andinformatics solutions for mass spectrometry based
proteomics.
ACKNOWLEDGMENTSWewould like to express our gratitude to the
American ChemicalSociety for making data available for this study
and technicalsupport. In particular, we would like to thank
Catherine Boylan,Emma Moore, David Martinsen, and Jerey Krugman. We
alsothank Rob Marissen (LUMC) and Bjorn Victor (Institute
forTropical Medicine, Antwerp) for technical assistance. Finally,
wewould like to thank Nees Jan van Eck and Ludo Waltman (bothCWTS)
and Michael Grayson for fruitful discussions.
REFERENCES(1) Gareld, E.; Sher, I. H.; Torpie, R. J. The Use of
Citation Data inWriting the History of Science; Institute for
Scientic Information Inc.:Philadelphia, PA, 1964.(2) Hummon, N. P.;
Doreian, P. Soc. Networks 1989, 11, 3963.(3) Garfield, E. J. Inf.
Sci. 2004, 30, 119145.(4) van Eck, N. J.; Waltman, L. J. Informetr.
2014, 8, 802823.(5) Chen, C. M. J. Am. Soc. Inf. Sci. Technol.
2006, 57, 359377.(6) Sci2 team. Science of Science (Sci2) Tool;
Indiana University andSciTech Strategies, https://sci2.cns.iu.edu,
2009.(7) van Eck, N. J.; Waltman, L.; Dekker, R.; van den Berg, J.
J. Am. Soc.Inf. Sci. Technol. 2010, 61, 24052416.(8) Naturally, the
use of a single journal as a representation of a eld isan
oversimplication of all research in a eld. However, working with
asingle journal also has clear advantages, such as the continuity
of a single,uninterrupted stream of publications in a journal
managed by oneeditorial board (which changes slowly over time), a
nite number ofarticle types and formats and the practical
possibility of accessing allarticles published over nearly a
century. For 19291995, only scannedrst pages of articles processed
by Optical Character Recognition(OCR) to TXT-les were available.
Thus, we extracted nwords after thearticles title, with n being the
average abstract length for a ve-yearperiod. Therefore, a
limitation of our study is that for 19291995 wewere not able to
precisely extract abstracts, and that OCR errors werepresent in the
texts.(9) van Eck, N. J.; Waltman, L. Scientometrics 2010, 84,
523538.(10)Waltman, L.; van Eck, N. J.; Noyons, E. C. M. J.
Informetr. 2010, 4,629635.
(11) Reinhardt, C.; Shifting and Rearranging: Physical Methods
and theTransformation of Modern Chemistry; Science History
Publications/USA: Sagamore Beach, MA, 2006; pp 321.(12) Baird, D.
Ann. Sci. 1993, 50, 267290.(13) Strong, F. C. Anal. Chem. 1947, 19,
968971.(14) Fischer, R. B. Anal. Chem. 1956, 28, 9A14A.(15) Brooks,
R. R.; Smythe, L. E. Talanta 1975, 22, 495504.(16) Berezkin, V. G.;
Chernysheva, T. Y. J. Chromatogr. 1977, 141,241246.(17) Nederlands
Observatorium van Wetenschap en Technologie.Wetenschaps- en
Technologie-Indicatoren 2010; Ministry of Education,Culture and
Science: The Hague, Netherlands, 2010; SupplementaryTable B.(18)
Most research using MS in this category has been published
inchemistry and physics journals, not in astronomy journals.(19)
American Society for Mass Spectrometry. Measuring Mass:
FromPositive Rays to Proteins; Grayson, M. A., Ed.; Chemical
Heritage Press:Philadelphia, PA, 2002.(20) van Eck, N. J.
Methodological Advances in Bibliometric Mapping ofScience. Ph.D.
Thesis, Erasmus University Rotterdam, Rotterdam, TheNetherlands,
2011; p 13.(21) Braun, T.; Zsindely, S. TrAC, Trends Anal. Chem.
1992, 11, 307309.(22) White, H. D.; McCain, K. W. In Annual Review
of InformationScience and Technology; Williams, M. E., Ed.;
Elsevier: Amsterdam, TheNetherlands, 1989; pp 119186.(23) Laemmli,
U. K. Nature 1970, 227, 680685.(24) Tanaka, K.; Waki, H.; Ido, Y.;
Akita, S.; Yoshida, Y.; Matsuo, T.Rapid Commun. Mass Spectrom.
1988, 2, 151153.(25) Arthur, C. L.; Pawliszyn, J. Anal. Chem. 1990,
62, 21452148.(26) Matuszewski, B. K.; Constanzer, M. L.;
Chavez-Eng, C. M. Anal.Chem. 2003, 75, 30193030.(27) Marechal, C.
N.; Telouk, P.; Albarede, F. Chem. Geol. 1999, 156,251273.(28)
Bligh, E. G.; Dyer, W. J. Can. J. Biochem. Physiol. 1959, 37,
911917.(29) van den Dool, H.; Kratz, P. D. J. Chromatogr. 1963, 11,
463471.(30) Guenther, A.; Hewitt, C. N.; Erickson, D.; Fall, R.;
Geron, C.;Graedel, T.; Harley, P.; Klinger, L.; Lerdau, M.;
Mckay,W. A.; Pierce, T.;Scholes, B.; Steinbrecher, R.; Tallamraju,
R.; Taylor, J.; Zimmerman, P. J.Geophys. Res.: Atmos. 1995, 100,
88738892.(31) Weibel, D.; Wong, S.; Lockyer, N.; Blenkinsopp, P.;
Hill, R.;Vickerman, J. C. Anal. Chem. 2003, 75, 17541764.(32)
Domon, B.; Costello, C. E. Glycoconjugate J. 1988, 5, 397409.
Analytical Chemistry Feature
DOI: 10.1021/ac5040314Anal. Chem. 2015, 87, 45884596
4596