Bibliometric Analysis-Based Review of Fused Deposition Modeling 3D Printing Method (1994–2020)

Abstract

This study aimed at the detailed bibliometric analysis (BA) of fused deposition modeling (FDM) to understand the trend and research area. Web of Science database was used for extracting data using keywords, and 2793 documents were analyzed. From the analysis, the most influential and productive authors, countries, sources, and so on were identified and corresponding interrelations were represented by a three-field plot. Lotka's law was derived for author productivity and its reliability was verified by the Kolmogorov–Smirnov (K–S) test. Bradford's law was used for identifying the core sources contributing to the field of FDM. From the trend topic analysis, it was found that initially the research was focused upon removing error related to deposition as well as part orientation, but with the course of time, it diversified to include topics such as optimization of printing parameters, materials, and applications. Based on the inferences from BA, the article also discusses on current research trend and highlights certain future areas for research work.

Introduction

Fused deposition modeling (FDM) was invented in 1988, commercialized in 1990,¹ and patented in 1992² by Steven Scott Crump, cofounder of Stratasys. As per ISO/ASTM classification, FDM comes under the category of the material extrusion process of additive manufacturing (AM) shown in Figure 1. The FDM printer consists of filament spool, extruder head assembly, base or platform³ over which material is deposited, as shown in Figure 2. The extruder head assembly further consists of the liquefier, heat block containing temperature sensor, heating element, and nozzle. The liquefier or heat sink is connected with the heater block with the help of heat break, which is a small cylindrical element having threads on both sides. Separate heating element and temperature sensor are also connected to the bed so that the required temperature can be maintained at the time of deposition. Three stepper motors are used each for X, Y, and Z directions and one for extruding filament. Generally, the bed moves in Z direction, while the extrusion head assembly in X–Y direction. The stepper motor used for extruding filament can be in-house with extruder head known as direct type, while in some cases it is at a distance and fitted on to the body of printer known as bourdon type. The filament of material is inserted into extruder assembly through force exerted by the rollers of the stepper motor, heated to semisolid stage in the liquefier, and deposited in heated bed through nozzle.⁴ The filament part above the extruder assembly acts as a piston for the heated part, which is inside the liquefier and forces its way out of the nozzle. The quality of deposit largely depends on the input parameter settings such as printing speed, bead height, raster/part orientation, and bead angle. Optimization of the process parameters largely influences mechanical properties and physical appearance resulting in better performance of FDM parts.

FIG. 1.

ISO/ASTM 52900:2015 (International Organization for Standardization/American Society for Testing and Materials) Classification of additive manufacturing highlighting FDM. FDM, fused deposition modeling.

FIG. 2.

FDM or material extrusion process.

The various steps involved in the FDM process are shown in Figure 3, which is like other AM processes. The very first step is the computer aided design (CAD) model design in software such as solid works, creo, and fusion 360. The next step is to save that file in STL (Standard Triangle Language) format, which is simply the conversion of CAD model into several triangles. The STL file is then imported to slicing software such as CURA and 3D Slicer, in which various input printing parameters such as layer height, infill density, temperature, and print speed, value can be set accordingly and then the file is sliced in the form of layers.⁵ The sliced file information is extracted in the form of GM-codes, fed into the FDM machine, and the required geometry can be three-dimensional (3D) printed. The postprocessing in the form of machining, dipping in solution, and so on for removing support structures or eliminating minor defects can be done once the printing is completed. The advantage of FDM is its low cost, reduced time for production,⁶ and wide acceptance in prototyping application. However, it is limited by low temperature and a limited range of materials.¹ FDM as in 3D printing is one of the revolutionary technologies, which is also the key enabling element of industry 4.0 and can achieve the target of smart and lean manufacturing.⁷ Since 3D printing is an automation process that can print customized geometry on demand, also it is easy to deploy and maintain. The on-demand print feature of 3D printing helps in reducing the inventory and production in small batches leading to pull systems, which are the requisites of lean manufacturing. The other objectives such as 5S and man/machine separation are achieved by automation and easy maintenance feature of 3D printing. Isolation and easy deploy feature of 3D printing lead to reduction in lead time, downsizing of factory capacity, just in time, effective logistics system, and so on. One such example is illustrated by the industrial case study⁸ in which AM led to considerable saving of time in maintenance, repair, and overall activities by simply 3D printing the part and replacing it with the defected part. The only time required is in designing/downloading the STL file and 3D printing of the part, which is negligible compared with ordering and waiting time of the new part. Another example is the case of academic research done through case studies in Finland,⁹ which demonstrated the use of AM for reducing the material and inventory, minimizing the wastages and energy usages, making supply chain more flexible, simplified, efficient, or compressed, bringing on more customization.

FIG. 3.

Flow diagram representing process flow of 3D printing model using FDM printer, each subsequent process acts as an input to the next step. 3D, three-dimensional.

Bibliometric analysis (BA) means exploring the past, current, and future trend in a particular field. It deals with both qualitative and quantitative analysis of literature in terms of the number of articles (productivity),¹⁰ number of citations (influential), a particular country or university contributing majorly in specific research, and so on, and therefore helps in identifying the area, which is unexplored or requires sufficient amount of research. There are various techniques described for BA in literature for understanding the research done on existing work dynamically and structurally,¹¹ which include citation and productivity analysis, Bradford analysis, research front identification, bibliographic coupling, and so on.

FDM advantage is its wide variety of application, low cost, customized products, lower lead times, and so on. Major research is going on with regard to the materials used in FDM and the probable application they bring with them. So, BA will help in focusing on the countries, universities, authors, and so on leading in FDM technology, other keywords associated with it, and probable area of future research. The following steps have been followed for BA in this article.

• Searching of one of the most widely used and reputed databases, Web of Science (WOS), using keywords.

• Data collection

• BA of the data using software and app such as Biblioshiny, VOSviewer, and HistCite, and presenting in a meaningful way.

• Results and Conclusion

• Future scope

Scientific productivity is an important aspect of BA and is the measure of published output of a domain with respect to number of authors.¹² Although there are various approaches for measuring the same, one such is Lotka's law for finding out the author's productivity, which simply states that in every field whether scientific, humanity, and so on, the number of published articles is inversely proportional to the number of authors who have contributed. To ensure whether the observed author's published data distribution fits the theoretical distribution, a goodness-of-fit test is performed.¹³ The test for the same should be nonparametric, since author's data are non-normally distributed. The two hypotheses, null and alternate, are tested between expected and observed frequency. The two most widely used tests for hypothesis testing are the chi-square (χ²) test and Kolmogorov–Smirnov (K–S) test.¹⁴ The prerequisites for the χ² test are that data should be categorized in discrete class and the value of expected frequencies should be sufficiently large. While the nature of the author's data is such that most of the publications are done by a handful of authors, a major number of them have a small contribution such as one article, which is a small-frequency case. In the case if more than 20% of the expected frequencies are lesser than 5, then their information is clubbed to form a larger frequency. So, some of the information is lost in combining the frequencies, and therefore, the χ² test does not prove to be effective in this case, unlike the K–S test that treats each observation separately and does not lose any information in the process. The detailed explanation of Lotka's law and its correlation with the K–S test are given further under the Author Impact and Productivity section. Similar to this, sources' (journal) contribution generally follows the Bradford law founded by Bradford in the year 1948,¹⁵ which states that for a particular domain if the sources are divided in the decreasing order of the articles contributed, then this result in the formations of zones of sources having equal number of articles and number of zones follows the geometric progression. The first zone is generally the nucleus and sources in it are known as core sources, which are lesser in number but contributed the most articles in that domain. Lotka's and Bradford laws are important in the context of FDM literature as it helps in finding the trend in authors and source productivity, respectively. The K–S test ensures the reliability of the Lotka's law, that is, how effectively the given author's data of FDM domain fit in it. The Bradford law also helps in identifying the core sources corresponding to FDM literature. The details of which are given in the Journal Contributions and Bradford's Law section.

Data Extraction

WOS is selected as it is one of the powerful research tools for searching different types of article-related information in a time phase manner. The search query in WOS can be general to very specific, supported by IF, OR, AND, NOT, and so on. statement, depending on the information required. WOS provides access to various reputed databases around the world containing diversified and peer-reviewed articles. The keywords used for extracting data from the WOS database are mentioned in Table 1. It was found that combining FDM and fused filament fabrication (FFF) by using OR statement results in a larger number of documents. The idea behind this is not to leave any document using FDM or FFF technology. The final search was performed on November 30, 2020, on the WOS database, which resulted in a total of 3483 articles (articles, review articles, proceedings articles, etc.), out of which 149 review articles and 361 other articles, which were included in the search due to the keywords “fused,” “deposition,” “modeling,” and so on, but had no relevancy to FDM, were discarded. So, 2973 articles were considered for BA, the details of which are given in Table 2.

Table 1.

Keywords Used in Web of Science Database

S. No.	Keywords	No. of research articles
1.	FFF	966
2.	FDM	2887
3.	“Fused Deposition Modelling” OR “Fused Filament Fabrication”	3483

FDM, fused deposition modeling; FFF, fused filament fabrication.

Table 2.

Details of Documents Used for Bibliometric Analysis

Description	Results
Main information about data
Time span	1994:2021
Sources (Journals, Books, etc.)	676
Documents	2973
Average years from publication	2.78
Average citations per document	18.69
References	58,943
Document types
Article	2790
Article; Book chapter	1
Article; Early access	92
Article; Proceedings article	67
Article; Retracted publication	1
Correction	8
Editorial material	5
Meeting abstract	9
Total	2973
Document contents
Keyword Plus (ID)	3574
Author's keywords (DE)	6550
Authors
Authors	9259
Author appearances	13,909
Authors of single-authored documents	54
Authors of multiauthored documents	9205
Authors' collaboration
Single-authored documents	63
Documents per author	0.321
Authors per document	3.11
Coauthors per document	4.68
Collaboration index	3.16

Bibliometric Analysis

The Biblioshiny app in the library of bibliometrix package of RStudio software and HistCite and VOSviewer were used for analysis and to extract meaningful information. The information includes top authors, journals, countries, universities, and so on, having a major impact on FDM and paving the way for others.

FDM growth

The first article in WOS using FDM or rapid prototyping technology goes back to the year 1994. Since then, the rate remained slow till 2013 with an annual growth rate of 6.14%. The reason for this can be traced back to the expiration of FDM patent on October 30, 2009², as it takes 4–5 years for the technology to be popularized and developed. Therefore, the major chunk of articles in this field was published post-2014 with a maximum of 720 in the year 2020. The publications on FDM over the years have been depicted in Figure 4 and as far as average article citation is concerned, it shows an increased trend over the years (Fig. 5) with a patch of decrement in between. The maximum average citation was in the year 2015.

FIG. 4.

Annual production over the years in WOS (x axis shows the number of articles and y axis shows the year). WOS, Web of Science.

FIG. 5.

Annual average citation of documents over the years (x axis shows the average number of citations of documents and y axis shows the year).

Author impact and productivity

The author contribution and its impact can be categorized in two ways, most cited, that is, influential, and the number of articles, that is, productive. The top 20 authors with maximum citations and other details are shown in Table 3, while the ones having the highest number of articles are shown in Table 4. Hutmacher leads the list with the highest number of citations with a relatively low number of 20 articles, while Rupinder leads the list for the highest number of articles. The first three columns in Tables 3 and 4 represent the following:

Table 3.

Top 20 Authors with the Maximum Number of Citations (Influential)

S. No.	Author	h_index	g_index	m_index	TC	NP	FPY
1	Hutmacher	14	20	0.70	3055	20	2001
2	Teoh	5	5	0.25	2333	5	2001
3	Tan	2	2	0.10	2053	2	2001
4	Zein	2	2	0.10	2053	2	2001
5	Basit	15	17	2.14	1692	17	2014
6	Gaisford	15	17	2.14	1692	17	2014
7	Goyanes	15	17	2.14	1692	17	2014
8	Masood	18	30	0.72	1446	30	1996
9	Ahn	6	8	0.32	1249	8	2002
10	Chin Tze Ng	1	1	0.05	1004	1	2001
11	Schantz	1	1	0.05	1004	1	2001
12	Wright	4	4	0.21	989	4	2002
13	Singh	17	23	2.12	949	104	2013
14	Montero	1	1	0.05	905	1	2002
15	Odell	1	1	0.05	905	1	2002
16	Roundy	1	1	0.05	905	1	2002
17	Alhnan	11	12	1.83	853	12	2015
18	Mahapatra	7	9	0.58	842	9	2009
19	Wicker	8	11	1.14	815	11	2014
20	Sood	8	9	0.67	800	9	2009

FPY, first publication year; NP, number of articles; TC, total citations.

Table 4.

Top 20 Authors with the Maximum Number of Articles (Productive)

S. No.	Author	h_index	g_index	m_index	TC	NP	FPY
1	Singh	17	23	2.12	949	104	2013
2	Singh	9	15	1.5	267	33	2015
3	Masood	18	30	0.72	1446	30	1996
4	Wang	7	15	0.58	229	22	2009
5	Kumar	7	11	2.33	137	21	2018
6	Hutmacher	14	20	0.70	3055	20	2001
7	Ahuja	7	11	1.4	142	19	2015
8	Kumar	7	11	0.54	128	18	2008
9	Basit	15	17	2.14	1692	17	2014
10	Fu	10	17	1.67	477	17	2015
11	Gaisford	15	17	2.14	1692	17	2014
12	Goyanes	15	17	2.14	1692	17	2014
13	Jain	9	14	1.8	200	17	2016
14	Zhang	5	10	1.25	106	16	2017
15	Bhowmik	9	15	1.50	547	15	2015
16	Mohamed	9	15	1.50	547	15	2015
17	Batish	5	11	1	132	14	2016
18	Belhabib	5	10	0.83	118	14	2015
19	Guessasma	5	10	0.83	118	14	2015
20	Pearce	7	14	1.16	212	14	2015

• (1)h_index¹⁶ of any author is equal to the minimum number of articles having “h” citation.
• (2)g_index¹⁷ of any author is the largest number of top articles “g” when arranged in decreasing order of citations, which have at least “g²” citations.
• (3)m_index of any author is $\frac{h}{n}$ where n is the number of years after the first article has been published.

Consider the example of Hutmacher (Table 3), having h_index of 14, it means out of his total articles 20, he has 14 articles that have at least 14 citations each, g_index of 20, which is nothing but equal to his total number of articles, since 20² is 400, which is less than his total citations 3055, m_index of 0.70, which is $\frac{14}{20}$, since the first articles was published in 2001, so “n” is 20 (2001–2020).

Hutmacher is the most influential author in the field of FDM it seems. The 20 articles related to FDM literature are mainly in the medical domain. The research interest includes scaffold development, bone and tissue regeneration, breast cancer, tissue reconstruction, and so on using FDM technology. Two articles among the list, published in the years 2001¹⁸ and 2002,¹⁹ are having more than 1000 citations each, in which the polycaprolactone (PCL) filament is prepared from the one-shot extruder using PCL pellets, which is subsequently used for fabricating a honeycomb-structured porous scaffold using FDM. The mechanical properties of the same are measured and evaluated for enhanced applications. The research work also includes in vivo and in vitro studies on both human and animal tissues. The latest published work involves fabrication of scaffold for breast reconstruction, bone model for metastasis in vivo for breast cancer, evaluation of bioresorbable filament, and using an advanced AM technique known as melt electrowriting (MEW) for depositing and fabricating fibrous scaffolds that are anatomically relevant.

Lotka's law for author productivity

Lotka was an American statistician who published the first article²⁰ on the productivity of scientific documents related to frequency distribution in 1926, where he found out that author's productivity over an entire life follows a power law. Potter in his article “Lotka's law revisited”²¹ illustrated that the first time Lotka's article²⁰ was cited was in 1941 and it was termed as “law” in the year 1949. The first attempt made to test its applicability in other disciplines was in the year 1973. Macroberts in the year 1982 published an article on revaluation of Lotka's law,²² in which it was demonstrated that the law is applicable only to the productivity of an author within a particular field (area) instead of an author's or scientist's lifetime productivity. From all these discussions, it was inferred that Lotka's law is used for finding out the productivity of authors in a particular domain; it also gives a sense of understanding and awareness among researchers related to that field.

According to Lotka's law,¹³ a large percent of a number of articles in a particular field are written by a small fraction of authors, and therefore, as the number of articles²³ by a single author increases, the number of authors in that area decreases. The law is given by “xyⁿ = C” where y is the number of authors publishing x articles in a particular domain, and n and C are constants represented by the Equations (1) and (2), respectively. Of the total 2793 documents, it was found that 76.6% (Table 5) of the total 9259 authors have published one article. The graphical result obtained is shown in Figure 6, and it is clearly visible that as the number of documents by an individual author approaches 100, the percentage of authors approaches zero and vice versa. The Lotka test was applied for the current observation and its reliability was verified by the K–S test. Table 5 shows the number of documents (x) written against the number of authors (y), so here n = 24, which is the number of rows or pairs, X = log x and Y = log y. All these terms are utilized in Equation (1) to find the value of exponent “n.” The value of p, which is the limit of x in Equation (2), is taken as 20 for getting the minimum residual error in least number of iterations and n = −2.401 calculated from Equation (1) is in the range according to Pao (1985) and its absolute value is taken in Equation (2) for calculation of C = 0.7272. The function is represented by Equation (3).

Table 5.

Documents Written Corresponding to the Number of Authors^*

S. No.	Documents written (x)	No. of authors (y)	X = log(x)	Y = log(y)	XY	X 2
1	1	7089	0.00	3.85	0.00	0.00
2	2	1247	0.30	3.10	0.93	0.09
3	3	429	0.48	2.63	1.26	0.23
4	4	226	0.60	2.35	1.42	0.36
5	5	103	0.70	2.01	1.41	0.49
6	6	42	0.78	1.62	1.26	0.61
7	7	27	0.85	1.43	1.21	0.71
8	8	27	0.90	1.43	1.29	0.82
9	9	19	0.95	1.28	1.22	0.91
10	10	11	1.00	1.04	1.04	1.00
11	11	9	1.04	0.95	0.99	1.08
12	12	4	1.08	0.60	0.65	1.16
13	13	4	1.11	0.60	0.67	1.24
14	14	6	1.15	0.78	0.89	1.31
15	15	2	1.18	0.30	0.35	1.38
16	16	1	1.20	0.00	0.00	1.45
17	17	5	1.23	0.70	0.86	1.51
18	18	1	1.26	0.00	0.00	1.58
19	19	1	1.28	0.00	0.00	1.64
20	20	1	1.30	0.00	0.00	1.69
21	22	2	1.34	0.30	0.40	1.80
22	30	1	1.48	0.00	0.00	2.18
23	33	1	1.52	0.00	0.00	2.31
24	104	1	2.02	0.00	0.00	4.07
Total		9259	24.74	24.99	15.86	29.63

FIG. 6.

Author productivity and Lotka's law (x axis represents the percentage of authors, while y axis represents the number of documents written by them, as the number of documents increases, the percentage of authors contributing the same decreases).

$$C=\frac{1}{{\sum }_{x=1}^{p-1}\frac{1}{x^n}+\frac{1}{{\left(n-1\right)}^{p^{\left(n-1\right)}}}+\frac{1}{2P^n}+\frac{n}{24{\left(p-1\right)}^{n+1}}}$$ (2)

$$f\left(x\right)=\frac{C}{x^n}$$ (3)

$$f\left(x\right)=\frac{0.7272}{x^{2.401}}$$

K–S test for Lotka's law

As already discussed in the introduction, the goodness-of-fit test is used to check how well the observed frequency matches with the theoretical one. The χ² test power is compromised since it combines the low adjacent frequencies. While in the K–S test, the theoretical distribution is known and its cumulative frequency is compared with the observed frequency. The two hypotheses are formulated as follows: (1) null hypothesis, which states that the observed distribution is not different with the expected and (2) alternate hypothesis, the expected and observed distribution is different. The difference between the cumulative frequencies is expected to be small and test helps in determining the probability associated with the maximum divergence occurring between observed and expected frequencies within the limits. As discussed earlier, each observation is considered separately in this test resulting in no loss of information and therefore provides better result than the χ² test.

Table 6 shows the value corresponding to the K–S test, the last column is the difference between observed and expected accumulated values, if D_max is greater than the critical value of K–S test, then null hypothesis is rejected and vice versa for alternate hypothesis. For a sample size greater than 35 at 0.01 level of significance, Equation (4) is used for calculating the critical value, in the current case it is 24, which is less than 35 and so the direct value from the K–S table can be taken, which is 0.270.

Table 6.

K–S Fit Test

S. No.	Documents published (x)	No. of authors (y)	Observed value of contribution of authors fo(yx) = yx/∑ yx	Observed accumulated value ∑ fo(yx)	Expected value of contribution of authors fe(yx) = yx/∑ yx	Expected accumulated value ∑ fe(yx)	Diffrence (D)|∑ fo(yx)−∑ fe(yx)|
1	1	7089	0.7656	0.7656	0.7276	0.7276	0.0380a
2	2	1247	0.1347	0.9003	0.1378	0.8654	0.0350
3	3	429	0.0463	0.9466	0.0520	0.9174	0.0292
4	4	226	0.0244	0.9711	0.0261	0.9435	0.0276
5	5	103	0.0111	0.9822	0.0153	0.9587	0.0234
6	6	42	0.0045	0.9867	0.0099	0.9686	0.0181
7	7	27	0.0029	0.9896	0.0068	0.9754	0.0142
8	8	27	0.0029	0.9925	0.0049	0.9803	0.0122
9	9	19	0.0021	0.9946	0.0037	0.9841	0.0105
10	10	11	0.0012	0.9958	0.0029	0.9870	0.0088
11	11	9	0.0010	0.9968	0.0023	0.9892	0.0075
12	12	4	0.0004	0.9972	0.0019	0.9911	0.0061
13	13	4	0.0004	0.9976	0.0015	0.9927	0.0050
14	14	6	0.0006	0.9983	0.0013	0.9939	0.0043
15	15	2	0.0002	0.9985	0.0011	0.9950	0.0035
16	16	1	0.0001	0.9986	0.0009	0.9960	0.0026
17	17	5	0.0005	0.9991	0.0008	0.9968	0.0024
18	18	1	0.0001	0.9992	0.0007	0.9975	0.0018
19	19	1	0.0001	0.9994	0.0006	0.9981	0.0013
20	20	1	0.0001	0.9995	0.0005	0.9986	0.0008
21	22	2	0.0002	0.9997	0.0004	0.9991	0.0006
22	30	1	0.0001	0.9998	0.0002	0.9993	0.0005
23	33	1	0.0001	0.9999	0.0002	0.9995	0.0004
24	104	1	0.0001	1.0000	0.0000	0.9995	0.0005
Total		9259

^a D_max.

From Table 6 it can be observed that D_max 0.0380 is smaller than the critical value 0.270 at 0.01 level of significance. Therefore, the alternate hypothesis is rejected, which states that the above data hold to Lotka's law (Fig. 6).

$$Criticalvalue=\frac{1.63}{\sqrt{\sum y}}$$ (4)

Journal contributions and Bradford's law

The source impact includes the top journals in the field of FDM in terms of citations and numbers of articles. The top 20 journals having the maximum number of citations over the years are shown in Table 7, while the ones with the maximum number of articles are shown in Table 8. The Rapid Prototyping Journal is leading in both the lists, while Material Design is second in the list having a high number of citations with a relatively low number of articles, and Additive Manufacturing is second in the list of maximum articles. Interestingly, the Additive Manufacturing Journal by the Elsevier group started in the year 2015 has performed tremendously well in both the fields in a short span of time and therefore having the highest m_index of 4.33.

Table 7.

Top 20 Journals with the Maximum Number of Citations (Influential)

S. No.	Source	h_index	g_index	m_index	TC	NP	FPY
1	Rapid Prototyping Journal	38	74	1.46	6270	246	1995
2	Materials & Design	30	60	1.58	3712	69	2002
3	International Journal of Advanced Manufacturing Technology	28	46	1.22	2445	126	1998
4	Additive Manufacturing	26	42	4.33	2344	157	2015
5	International Journal of Pharmaceutics	22	42	3.14	1993	42	2014
6	Journal of Materials Processing Technology	20	23	0.80	1950	23	1996
7	Biomaterials	8	8	0.42	1693	8	2002
8	Composites Part B-Engineering	19	37	3.17	1388	45	2015
9	Materials	18	29	3.00	1081	116	2015
10	Journal of Biomedical Materials Research	1	1	0.05	1004	1	2001
11	Journal of the American Ceramic Society	6	6	0.25	855	6	1997
12	International Journal of Machine Tools & Manufacture	3	4	0.13	841	4	1998
13	Scientific Reports	8	20	1.60	604	20	2016
14	Journal of Manufacturing Science and Engineering-Transactions of the ASME	12	24	0.55	593	25	1999
15	European Journal of Pharmaceutics and Biopharmaceutics	9	10	1.50	584	10	2015
16	European Journal of Pharmaceutical Sciences	9	15	1.50	552	15	2015
17	ACS Applied Materials & Interfaces	10	19	1.43	545	19	2014
18	Composites Part A-Applied Science and Manufacturing	8	12	1.60	536	12	2016
19	Proceedings of the Institution of Mechanical Engineers Part B-Journal of Engineering Manufacture	12	19	0.63	518	19	2002
20	Polymer	12	19	1.71	507	19	2014

Table 8.

Top 20 Journals with the Maximum Number of Articles (Productive)

S. No.	Source	h_index	g_index	m_index	TC	NP	FPY
1	Rapid Prototyping Journal	38	74	1.46	6270	246	1995
2	Additive Manufacturing	26	42	4.33	2344	157	2015
3	International Journal of Advanced Manufacturing Technology	28	46	1.22	2445	126	1998
4	Materials	18	29	3.00	1081	116	2015
5	Polymers	13	18	2.60	481	88	2016
6	Materials & Design	30	60	1.58	3712	69	2002
7	Composites Part B-Engineering	19	37	3.17	1388	45	2015
8	International Journal of Pharmaceutics	22	42	3.14	1993	42	2014
9	Journal of Applied Polymer Science	8	21	0.44	465	38	2003
10	3D Printing and Additive Manufacturing	8	16	1.14	300	34	2014
11	Applied Sciences-Basel	6	10	1.50	130	31	2017
12	Journal of Manufacturing Processes	12	21	2.00	490	29	2015
13	Polymer Testing	10	22	1.25	488	28	2013
14	Journal of Manufacturing Science and Engineering-Transactions of the ASME	12	24	0.55	593	25	1999
15	Virtual and Physical Prototyping	11	18	2.20	363	25	2016
16	Journal of Materials Processing Technology	20	23	0.80	1950	23	1996
17	Journal of Thermoplastic Composite Materials	6	8	2	81	22	2018
18	Scientific Reports	8	20	1.60	604	20	2016
19	ACS Applied Materials & Interfaces	10	19	1.43	545	19	2014
20	Polymer	12	19	1.71	507	19	2014

According to Bradford's law²⁴ when journals or sources are arranged in decreasing order of the number of articles on a particular domain, then for the same number of articles, sources are divided in the form of zones, which follow the geometric progression 1:n:n²:n³. The very first zone in this outcome is basically the core or nucleus of that domain and has a major contribution to it. When applied to 676 sources with their respective number of articles, it was found that the first 11 sources out of top 20 in Table 7 are the core sources and are in zone 1. Table 9 shows the 11 core sources in zone 1 with their corresponding rank, frequency (number of articles), recent impact factor, and publishing house. Elsevier is the publishing house of four core source journals, while Emerald is publishing for the Rapid Prototyping Journal, which is leading the list. The graphical result of the same is represented in Figure 7.

Table 9.

Core Sources Contributing Toward Fused Deposition Modeling (Zone 1)

Rank	Journal	NP	Impact factor	Publishing house
1	Rapid Prototyping Journal	246	3.714	Emerald Group Publishing Ltd.
2	Additive Manufacturing	157	7.002	Elsevier
3	International Journal of Advanced Manufacturing Technology	126	2.633	Springer London
4	Materials	116	3.057	MDPI AG
5	Polymers	88	3.426	MDPI AG
6	Materials & Design	69	6.289	Elsevier
7	Composites Part B-Engineering	45	7.635	Elsevier
8	International Journal of Pharmaceutics	42	4.845	Elsevier
9	Journal of Applied Polymer Science	38	2.520	John Wiley and Sons, Inc.
10	3D Printing and Additive Manufacturing	34	3.579	Mary Ann Liebert, Inc.

FIG. 7.

Sources impact using Bradford's Law (x axis represents the number of articles, while y axis represents the sources contributing the same).

Institutions and countries

The top 20 institutions working in the field of FDM in terms of the number of articles and citations are shown in Tables 10 and 11, respectively. The citations in the tables are shown as Total Local Citation Score (TLCS) and Total Global Citation Score (TGCS). TLSC is the number of times an article is referred within the articles of the same collection,²⁵ while TGCS is the number of times it is referred in the WOS collection. Guru Nanak Dev's Engg Coll is first in the list of articles and is at the 12th position in the global citation list. The National University of Singapore is at the top in the list of the number of global citations and is also third when it comes to the number of articles. The countries leading in the same fields are shown in Tables 12 and 13, respectively. The United States is at the top of both lists, China is more productive than the United Kingdom, but on the contrary, the United Kingdom is more influential than China with a relatively lesser number of articles.

Table 10.

Top 20 Most Institutions with the Maximum Number of Global Citations (Influential)

S. No.	Institution	NP	TLCS	TGCS
1	National University of Singapore	35	299	3313
2	University College of London, United Kingdom	23	515	1793
3	FabRx Pvt. Ltd, London, United Kingdom	16	515	1692
4	Swinburne University Technology, Australia	36	593	1584
5	University of Texas El Paso, U.S.	33	366	1432
6	National University Hospital, Singapore	4	129	1282
7	University of California, Berkeley, U.S.	11	415	1265
8	Gyeongsang National University, South Korea	5	524	1169
9	Temasek Polytechnic, Singapore	3	191	2176
10	Nanyang Technological University, Singapore	28	244	1019
11	University of Central Lancashire, United Kingdom	12	310	853
12	Guru Nanak Dev Engineering College, Ludhiana, India	92	326	848
13	Natl Inst Foundry & Forge Technol, India	9	404	810
14	Rutgers State University, New Jersey, U.S.	16	195	667
15	Zhejiang University, Hangzhou, China	33	131	627
16	Chinese Academy of Sciences, Beijing, China	32	176	592
17	Texas Tech University, Texas, U.S.	4	245	588
18	University of Minho, Gualtar, Portugal	12	206	573
19	University of Illinois, Chicago, U.S.	9	72	570
20	Georgia Institute of Technology, U.S.	24	130	562

TGCS, Total Global Citation Score; TLCS, Total Local Citation Score.

Table 11.

Top 20 Most Institutions with the Maximum Number of Articles (Productive)

S. No.	Institution	NP	TLCS	TGCS
1	Guru Nanak Dev Engineering College, Ludhiana, India	92	326	848
2	Swinburne University Technology, Australia	36	593	1584
3	National University of Singapore	35	299	3313
4	University of Texas El Paso, U.S.	33	366	1432
5	Zhejiang University, Hangzhou, China	33	131	627
6	Chinese Academy of Sciences, Beijing, China	32	176	592
7	Nanyang Technological University, Singapore	28	244	1019
8	Georgia Institute of Technology, U.S.	24	130	562
9	Punjabi University, Patiala, Punjab	24	77	172
10	University College of London	23	515	1793
11	Virginia Tech, U.S.	23	99	412
12	University of Chinese Academy of Sciences, China	22	146	404
13	University of Delaware, U.S.	22	120	413
14	Warsaw University of Technology, Poland	22	15	301
15	University of Alberta, Canada	21	44	140
16	University of Salerno, Italy	21	115	399
17	University of Wisconsin-Madison, U.S.	21	73	252
18	Massachusetts Institute of Technology, U.S.	20	183	433
19	Sichuan University, China	19	30	221
20	University of Leoben, Austria	19	71	222

Table 12.

Top 20 Most Countries with the Maximum Number of Global Citations (Influential)

S. No.	Country	Recs	TGCS	TLCS
1	The United States	716	15710	4232
2	The United Kingdom	218	6857	1477
3	People's Republic of China	434	6386	1518
4	Singapore	78	4683	618
5	Italy	172	4167	1339
6	India	240	3909	1388
7	Australia	113	2985	746
8	Germany	170	2357	328
9	Spain	146	2156	392
10	South Korea	123	1715	452
11	Canada	108	1421	504
12	Wales	18	1393	532
13	Poland	95	1344	309
14	France	112	1087	213
15	Denmark	26	914	324
16	Portugal	43	829	269
17	New Zealand	29	714	205
18	Taiwan	60	709	179
19	Japan	42	703	86
20	Brazil	59	695	186

Table 13.

Top 20 Most Countries with the Maximum Number of Articles (Productive)

S. No.	Country	NP	TGCS	TLCS
1	The United States	716	15710	4232
2	People's Republic of China	434	6386	1518
3	India	240	3909	1388
4	The United Kingdom	218	6857	1477
5	Italy	172	4167	1339
6	Germany	170	2357	328
7	Spain	146	2156	392
8	South Korea	123	1715	452
9	Australia	113	2985	746
10	France	112	1087	213
11	Canada	108	1421	504
12	Poland	95	1344	309
13	Singapore	78	4683	618
14	Taiwan	60	709	179
15	Brazil	59	695	186
16	Iran	57	607	141
17	Portugal	43	829	269
18	Japan	42	703	86
19	Switzerland	42	558	70
20	Belgium	41	499	149

Interconnections among authors, countries, sources, and universities

Interrelations between three different fields are represented by three-field plots and are also known as Sankey diagram. Figure 8 represents the interrelations among the top 10 authors, countries, and sources in terms of productivity. From the figure, it can be seen that the United States and China have articles in every source. The major chunk of articles from the United States is in the Rapid Prototyping and Additive Manufacturing Journals. Singh from India has single handedly contributed toward a number of articles, which means that in India, a major number of research articles are contributed by a handful of authors, while in the United States, it has been uniformly distributed. Therefore, although the United States tops the list with regard to the number of articles, none of the authors of the United States is in the top 10 (Fig. 8) list of individual production. The plot also shows the collaborations among authors of different countries. Figure 9 represents author with university and countries. The United States leads the way in production and diversity, with its 4 universities (Texas, Virginia, Michigan, and Delaware) out of 10. While in China and Australia, it is mainly associated with a single university (Zhejiang and Swinburne, respectively) and author (Wang and Masood). Interestingly, Hutmacher, who is currently working in Queensland University, Australia, has many publications associated with the National University of Singapore, from where he obtained his PhD. The University of Leoben, Austria, made it to the list due to its association with countries such as Germany and Spain, same goes with the Singapore universities National and Nanyang with their respective associations with the United States, the United Kingdom, France, Korea, Italy, Germany, India, and Spain. While in India, the publications are dominated by one group of authors and university (Guru Nanak Dev).

FIG. 8.

Three-field plot representing authors, countries, and sources (Sankey diagram representing three distinguished fields).

FIG. 9.

Three-field plot representing author, university, and countries.

Most cited documents

The documents under this category are top globally and locally cited documents. Local and global citations are already explained before. So, the top 20 documents with their citations and digital object identifier (DOI) are shown in Table 14 (decreasing order of globally cited) and Table 15 (decreasing order of locally cited). From the table, it can be seen that the article written by Zein in 2002 has the highest global citations, but 10th in the list of local citations. Surprisingly, the article by Hutmacher (2001), which is second in the list of highest global citations, has zero local citations, and the article by Ans (2002) is highest in the list of local citations and also third highest in the list of global citations.

Table 14.

Top 20 Globally Cited Documents

S. No.	Document	DOI	Global citations	Local citations
1	Zein (2002), Biomaterials	10.1016/S0142-9612(01)00232-0	1049	117
2	Hutmacher (2001), J Biomed Mater Res	10.1002/1097-4636(200105)55:2 ≤ 203::AID-JBM1007 ≥ 3.0.CO;2-7	1004	0
3	Ahn (2002), Rapid Prototyping J	10.1108/13552540210441166	905	400
4	Pham (1998), Int J Mach Tool Manu	10.1016/S0890-6955(97)00137-5	460	66
5	Ning (2015), Compos Part B-Eng	10.1016/j.compositesb.2015.06.013	440	175
6	Sood (2010), Mater Design	10.1016/j.matdes.2009.06.016	417	229
7	Sun (2008), Rapid Prototyping J	10.1108/13552540810862028	409	235
8	Lewis (2006), J Am Ceram Soc	10.1111/j.1551-2916.2006.01382.x	355	8
9	Zocca (2015), J Am Ceram Soc	10.1111/jace.13700	301	9
10	Jones (2011), Robotica	10.1017/S026357471000069X	293	51
11	Mohamed (2015), Adv Manuf	10.1007/s40436-014-0097-7	290	134
12	Anitha (2001), J Mater Process Tech	10.1016/S0924-0136(01)00980-3	284	131
13	Bellini (2003), Rapid Prototyping J	10.1108/13552540310489631	280	138
14	Zhong (2001), Mat Sci Eng A-Struct	10.1016/S0921-5093(00)01810-4	270	110
15	Kalita (2003), Mater Sci Eng C Mater Biol Appl	10.1016/S0928-4931(03)00052-3	268	42
16	Thrimurthulu (2004), Int J Mach Tool Manu	10.1016/j.ijmachtools.2003.12.004	261	101
17	Espalin (2014), Int J Adv Manuf Tech	10.1007/s00170-014-5717-7	258	46
18	Chacon (2017), Mater Design	10.1016/j.matdes.2017.03.065	255	142
19	Lee (2005), J Mater Process Tech	10.1016/j.jmatprotec.2005.02.259	253	110
20	Macdonald (2014), IEEE Access	10.1109/ACCESS.2014.2311810	249	24

DOI, Digital Object Identifier.

Table 15.

Top 20 Locally Cited Documents

S. No.	Document	DOI	Local citations	Global citations
1	Ahn (2002), Rapid Prototyping J	10.1108/13552540210441166	400	905
2	Sun (2008), Rapid Prototyping J	10.1108/13552540810862028	235	409
3	Sood (2010), Mater Design	10.1016/j.matdes.2009.06.016	229	417
4	Ning (2015), Compos Part B-Eng	10.1016/j.compositesb.2015.06.013	175	440
5	Chacon (2017), Mater Design	10.1016/j.matdes.2017.03.065	142	255
6	Bellini (2003), Rapid Prototyping J	10.1108/13552540310489631	138	280
7	Mohamed (2015), Adv Manuf	10.1007/s40436-014-0097-7	134	290
8	Carneiro (2015), Mater Design	10.1016/j.matdes.2015.06.053	132	245
9	Anitha (2001), J Mater Process Tech	10.1016/S0924-0136(01)00980-3	131	284
10	Zein (2002), Biomaterials	10.1016/S0142-9612(01)00232-0	117	1049
11	Sood (2009), Mater Design	10.1016/j.matdes.2009.04.030	115	240
12	Wu (2015), Materials	10.3390/ma8095271	111	204
13	Zhong (2001), Mat Sci Eng A-Struct	10.1016/S0921-5093(00)01810-4	110	270
14	Lee (2005), J Mater Process Tech	10.1016/j.jmatprotec.2005.02.259	110	253
15	Dawoud (2016), J Manuf Process	10.1016/j.jmapro.2015.11.002	102	166
16	Thrimurthulu (2004), Int J Mach Tool Manu	10.1016/j.ijmachtools.2003.12.004	101	261
17	Nikzad (2011), Mater Design	10.1016/j.matdes.2011.01.056	100	196
18	Masood (2004), Mater Design	10.1016/j.matdes.2004.02.009	98	220
19	Lee (2007), J Mater Process Tech	10.1016/j.jmatprotec.2006.11.095	97	219
20	Ahn (2009), J Mater Process Tech	10.1016/j.jmatprotec.2009.05.016	96	194

Keywords and keyword plus in WOS database related to FDM articles

Author keywords and keyword plus are the two different word features provided by the WOS database. The former represents the keywords used in one's article, while the latter is extracted from the title of cited references in the WOS database.²⁶ Although keyword plus can be used for BA for gaining useful insights, it is less effective in representing the content of an article. The top 20 keywords extracted with the help of VOSviewer software from the articles and their co-occurrence network are shown in Figure 10. The size of the node represents the frequency of the words used in articles. “3d printing” is the most used keyword followed by “additive manufacturing.” Therefore, the dark lines in the figure represent the network of 3D printing with respect to other keywords, while the light lines show the interconnections of the rest of the keywords. Some words are repeated multiple times due to changed spelling and acronyms used by different researchers. So, to avoid this ambiguity, the keyword plus feature was used to better understand the research and application area of FDM. Figure 11 represents the growth of keyword plus over the years, represented by the y-axis, while the x-axis represents the annual occurrences, the software drew the best fitted curve, there is an increase from 2013 due to the increase in the number of articles (Fig. 4). Figure 12 represents the three-field plot representing authors, keyword plus, and countries. From the figure, it can be inferred that top authors and countries are uniformly working in all the fields represented by keyword plus.

FIG. 10.

Co-occurrence network of keywords (size of node represents the frequency of keywords).

FIG. 11.

Growth of keyword plus over the years (represents the occurrences of keyword plus over the years).

FIG. 12.

Three-field plot representing authors, keyword plus, and countries.

Bibliometric coupling

Bibliometric coupling is defined as the connection between two articles from respective countries, which have used a common reference. This coupling was broadly defined in two categories by Kessler in 1962.²⁷ Category 1 is the group A in which the different articles are coupled by one particular article (P) in that group and strength of the coupling units is defined as the number of connections between P and other articles in the group. Category 2 is group B in which all the articles are connected with each other at least once. The bibliometric coupling between the top 20 countries in terms of the number of citations (Table 12) is shown in Figure 13. The dark lines show the interconnections of the United States with respect to other countries as it is the one having maximum citations and link strength, while the light lines show the interconnections between the rest. The diameter of the node represents the link strength of the coupling units, which means that articles of the United States are cited majorly all over the world.

FIG. 13.

Bibliometric coupling of top 20 countries having highest citations (size of node represents the link strength of countries).

Trend topics

The trend topic represents the sequence of topics explored with respect to the phase of past, present, and their connection with the future. Basically, it helps in identifying the current research area and potential domain, to be focused and explored upon. Figure 14 represents the trend topics over the years. In the initial phase from 2003 to 2005, the research was mainly focused on part orientation, and volumetric error, the corresponding frequency that is represented in x-axis was also low due to the lesser number of articles. From 2007 to 2011, it was focused upon rapid prototypes or rapid prototyping with medium frequency. Similar to word growth described above, from 2013, there is an increase in topics of research due to greater exploration of FDM. The major research in terms of frequency is mainly in the mechanical properties of the 3D-printed product, optimization of printing parameters, dimensional accuracy of the end product, testing of different materials as filament for improved properties, and so on. The other field in which FDM is highly used and explored upon is the medical field, especially for tablets, scaffolds, tissues, and so on. Based on the inferences from the trend topics, the materials and their corresponding application with respect to highly cited articles are further discussed in the next section.

FIG. 14.

Trend topics over the years (represents the logarithmic frequency of the topic trend over the years).

Materials and Application of FDM

Materials used since the inception of FDM were mostly polymers such as acrylonitrile butadiene styrene (ABS),^28–35 PCL,^18,19,36 and polylactic acid (PLA),^37,38 initially due to various properties such as the ease with which they can be printed, the low range of melting temperature of the order of 250°C, low cost, biocompatible, and chemical resistance. Table 16 shows the extrusion, bed, and glass transition temperature of the frequent filament materials used in FDM. Extrusion temperature is the temperature at which a material is extruded from the nozzle also known as hot end temperature of the heater block. Bed temperature is the temperature of the platform on which a material is deposited, and bed heating helps in easy removal of the part, once it is fully printed. While glass transition temperature represents the temperature at which transition of the material from glassy to rubbery state starts to take place.³⁹ The material is in stiff solid phase below this temperature and subcooled to liquid or elastic solid above this temperature. Elastic modulus of the material changes at this temperature resulting in phase transition. From the table it can be clearly seen that maximum extrusion temperature is about 315°C, while average is around 250°C. Therefore, the material is well suited for the FDM printer and all the materials in the table have a range of extrusion or melting temperature since the materials are amorphous, which have no fixed melting temperature.

Table 16.

Extrusion, Bed, and Glass Transition Temperature of the Fused Deposition Modeling Filament Material

S. No.	Filament material	Extrusion temperature °C (Te)	Bed temperature °C	Glass transition temperature °C (Tg)
1	ABS	220–240	80–110	105–110
2	PLA	180–220	20–55	50–80
3	HIPS	220–230	50–60	100
4	TPU	230–260	40–60	60
5	PETG	230–255	55–70	70–80
6	Nylon	235–270	60–80	47–60
7	PC	270–315	90–120	145–150

ABS, acrylonitrile butadiene styrene; HIPS, high-impact polystyrene; PC, polycarbonate; PETG, polyethylene terephthalate glycol; PLA, polylactic acid; TPU, thermoplastic polyurethane.

With the course of time, various other materials such as carbon fibers,^40–44 glass fibers,⁴⁵ carbon nanotube (CNT),^46–48 graphene,^49–51 and other polymers,^52–59 and metal^60–62 were reinforced in polymer filaments (ABS and PLA) to form composites to enhance their properties, resulting in a wide range of applications. Some researchers have also successfully deposited low melting point metal alloys especially solders^1,63–65 using a 3D printer, and this technology is known as FDM for metals (FDMm). The application of FDMm is in the printing of 3D electronic circuits,^1,44,66 complex prototypes of tool, mold and dies, and so on. FDM is also currently being utilized for extruding and depositing the industrial-grade plastics such as PEEK (polyether ketone), having melting temperature around 350°C.⁵² Another important application of FDM is in the medical field in 3D printing tablets,^67–80 drug loading and delivery,^81–92 scaffolds,^{18,37,58,93–114} implants,^115,116 and so on. Therefore, it can be clearly inferred that apart from applications such as rapid prototyping and rapid tooling, FDM is playing an important role in the medical field. Table 17^* shows the summary of the top globally cited articles (citations >200) on FDM with materials, temperatures, and applications. The articles are divided into two major headings based on the end application: (1) Medical domain and (2) Aerospace, automobile, prototyping, and so on. Although there are some more application areas of FDM such as electronics, fashion, and food, the articles under those categories are not highly cited. Therefore, these application areas are not discussed in detail in the following sections. The articles are arranged in decreasing order of their citations under each heading in Table 17.^† Under the medical domain, among the top cited articles, PCL and polyvinyl alcohol are widely used as a filament material, while ABS, PLA, and other polymers composited were dominant in other headings. All the articles that are in Table 14 are not included in Table 17, since some of the top articles have discussed the technology of FDM in their articles in general.

Table 17.

Summary of Most Globally Cited Articles on Fused Deposition Modeling with Materials, Temperature, and Applications Under Two^* Heads (Citations >200)

S. No.	Author and year	Filament materials used in FDM	Extrusion temperature	Applications
Medical domain
1	Zein, et al. (2002)19	PCL	125°C	Tissue engineering
2	Hutmacher et al. (2001)18	PCL	120°C	Scaffolds
3	Kalita et al. (2003)93	TCP and PP	Not mentioned	Scaffold
4	Goyanes et al. (2014)67	PVA loaded with fluorescein	220°C	Tablets
5	Goyanes et al. (2015)68	PVA mixed with paracetamol	180°C	Tablets of different shapes and geometry affecting drug release
6	Skowyra et al. (2015)83	PVA loaded with prednisolone	230°C	Drug release
7	Goyanes et al. (2015)69	PVA loaded with 5-ASA and 4 ASA mixed with ethanol and covered with parafilm	210°C	Customized medicines and modified oral drug release
8	Goyanes et al. (2015)70	PVA loaded with paracetamol	180°C	Devices containing multiple drugs and unique release properties
Aerospace, automobile, prototyping, and so on
9	Ning et al. (2015)40	ABS with CF infills	220°C	High strength and load-bearing components in aerospace, and so on
10	Sun et al. (2008)117	P400-ABS	270°C	Rapid prototyping and development of new materials for FDM
11	Lewis et al. (2006)118	Ceramic (filament-based direct ink)	Not mentioned	Advanced composites
12	Carneiro OS, et al. (2015)119	PP and glass fiber-reinforced PP	190°C–230°C	High temperature and stiffness applications in automobile, and so on
13	Tian et al. (2016)120	PLA reinforced with CF to form composites	180°C–240°C	Aerospace and aviation
14	Matsuzaki et al. (2016)121	PLA—carbon or jute fiber bundle composites	210°C	Automobile and aerospace
15	Masood and Song (2004)60	P301 Nylon/iron composite	Notmentioned	Rapid tooling, injection molding inserts

ASA, aminosalicylic acid; CF, carbon fibers; PCL, polycaprolactone; PP, polypropylene; PVA, polyvinyl alcohol; TCP, tricalcium phosphate.

Printing Parameters and Mechanical Properties

The optimum input printing parameter setting leads to improved mechanical properties, which result in superior performance of the FDM parts. Table 18 summarized the top cited articles (citations >200) that have explored important printing parameters and the subsequent improvement in mechanical properties of the printed part. It can be clearly inferred that raster orientation is the frequently explored printing parameter followed by layer thickness and air gap. These parameters also have a profound impact on the mechanical properties such as tensile, compressive, and flexural strength of the FDM-printed part. Raster or part orientation of (45°,45°) and (0°,90°) are the optimum ones for improving the mechanical properties, while negative air gap increases stiffness and strength. Reinforcing polymer filament with other polymers and metals also has a significant effect in improving the mechanical properties.

Table 18.

Summary of Most Globally Cited Articles on Fused Deposition Modeling with Reference to Material Properties and Printing Parameters (Citations >200)

S. No.	Author and year	Filament material	Printing parameters (input)	Mechanical properties (output response)	Inferences
1	Ahn et al. (2002)29	P400-ABS	Air gap Raster orientation Extrusion temperature Filament color Road width	Tensile strength Compressive strength	FDM-printed part properties compared with injection molding component Raster orientation and air gap influence tensile strength, while air gap influences compressive strength Tensile strength 65–72% and compressive strength 80–90% of injected molded ABS
2	Bellini and Güçeri (2003)30	ABS	Raster orientation Deposition path	Tensile strength Bending strength	Both selected input parameters have significant effect on mechanical properties
3	Zhong et al. (2001)45	ABS reinforced with short glass fiber	Different compositions of ABS composite	Hardness Strength	Hardness and strength of ABS increased after reinforcing it with short glass fiber, while there is reduction in flexibility and handleability Plasticizer and compatibilizer are added to regain the lost properties.
4	Thrimurthulu et al. (2004)31	ABS	Part orientation	Surface finish Deposition time	Genetic code developed for obtaining optimum part orientation during deposition, resulting in good surface finish and lesser deposition time
5	Caminero et al. (2017)122	PLA	Raster/build orientation Layer thickness Feed rate	Tensile strength Bending strength	Upright orientations of the part have the poorest of mechanical properties, while flat and on-edge have the highest stiffness and strength For upright orientations on increasing layer thickness, tensile and flexural strength increases, while it does not have significant effect in the case of flat and edge-on For upright orientations on increasing feed rate, tensile and flexural strength increases, while it has a little significant effect in the case of flat and edge-on
6	Lee et al. (2005)32	ABS	Air gap Raster orientation Raster width Layer thickness	Throwing distance (elastic performance)	Air gap, raster orientation, and layer thickness have significant influence on the flexibility of the ABS
7	Lee et al. (2007)33	ABS	Air gap Raster orientation Extrusion temperature Filament color Road/bead width	Compressive strength	Build direction has the most significant effect on the mechanical property Axial specimen has more compressive strength than transverse specimen Compressive strength of FDM-printed part is lower compared with conventional process
8	Wu et al. (2015)52	Polyether ketone (PEEK) and ABS	Raster orientation/angle Layer thickness	Tensile strength Compressive strength Bending strength	Raster angle 0° and layer thickness 300 μm resulted in optimum mechanical properties of 3D-printed PEEK The tensile, compressive, and bending strength of PEEK is superior to ABS
9	Shofner et al. (2003)46	ABS mixed with VGCF'S	Raster orientation	Tensile strength Stiffness	VGCF incorporation in ABS resulted in increased tensile strength, stiffness, and modulus compared with plain ABS However, fracture mode changed to brittle from ductile due to reduced interlayer fusion, which is further improved by magnum ABS

3D, three-dimensional; PEEK, polyether ketone; VGCFs, vapor grown carbon fibers.

Research Trends and Future Scope

The first point covered under research trend and future scope is “printing parameters and mechanical properties” which is the basic of FDM process and forms the basis of the application. The second and third points are two important application domains of FDM. The articles reviewed for writing this section have citations greater than 200.

Printing parameters and mechanical properties

Optimization of printing parameters and improvement in mechanical properties of the material used in FDM are a field that required continuous research. Raster orientation, layer thickness, and airgap are the parameters mostly explored. The need is to find the effect of more input parameters such as infill density, print speed, and shell thickness in improving the mechanical properties. Also, the effect of lower layer thickness in the case of different raster orientation must be explored upon. Research on new polymers or metals to be added to the existing filament for improving mechanical properties is of vital interest. Once the parameters are optimized, keeping the application area in mind, the parameter setting can be done, which helps in improving the mechanical properties.

Medical domain

A lot has been achieved in the medical field through FDM technology, authors have done great work in the field of 3D printing of tablets, drug loading on filament, drug delivery, design and 3D printing of devices with multiple containers, complex scaffold fabrication for bone tissue, breast reconstruction, and so on. The current research and results will open a new window of scope for the future. The future scope in this domain lies in the innovative design of tablets that can facilitate better drug release with new specific pharmacokinetic characteristics and also developing a flexible dose with multiple containers for drugs. Coming to implants and scaffolds, research is going on to develop biodegradable and customized implants for breast reconstruction with modified mechanical and architectural properties. Advanced AM techniques such as MEW can be utilized for a wide variety of biomedical applications.

Aerospace, automobile, prototyping, and so on.

The lightweight material with significant mechanical properties is of great use in the aerospace and automobile sector, as it leads to better performance, fuel economy, and ultimately results in cost saving. The FDM machine utilizes the properties of two different materials, which are formed by combining them in a single filament and 3D printing the customized geometry. Research has been carried out in this direction by adding carbon fibers in existing polymer filaments such as ABS and PLA, which resulted in improved mechanical properties such as Young's modulus and tensile strength, and desirable for lightweight applications. However, addition of carbon fibers after a certain percentage (around 10%) resulted in porosity issue and reduction in ductility, toughness, and yield strength. So future research should be focused upon developing new FDM materials by experimentally combining existing FDM materials with new polymers, carbon fibers, CNT, metals, and so on for getting the desired properties without compromising the existing mechanical properties. Another important area of research pertaining to lightweight application is the optimization of FDM printing parameters of high melting point industrial-grade material. The mechanical properties of industrial grade plastic obtained through FDM are almost half when compared with injection molding. So, a lot of scope of improvement lies in it.

Conclusion

The BA of 2793 documents, extracted from WOS, was done. The FDM technology can be traced back to 1988 with the first article in 1994, while the surge in the number of articles started close to 2014 due to expiration of the patent in 2009. There was constant growth in the number of articles with maximum number in the year 2020. Among the authors of articles extracted, Hutmacher is the author having the maximum citations (3055), while Singh leads in the number of articles (104). The K–S test was performed on author's data to verify its conformity to Lotka's law. Interestingly, it was also inferred that the Lotka's law was not felicitous to countries such as the United States, which leads in the number of articles because in the United States, the articles are collectively contributed by a number of authors. In the Journal Contributions and Bradford's Law section, Rapid Prototyping Journal leads in both categories of number of citations (6270) and articles (246). Bradford's law helps in identifying the 11 course sources in zone 1, with the Elsevier publishing house leading the way with four journals. Coming to the institutions, the National University of Singapore has the highest number of citations (3313), while Guru Nanak Dev Engineering College has the highest number of articles (92). In countries, the United States leads in both categories (15,710 citations and 716 articles). Due to the highest number of citations, the United States is also the one having a maximum bibliometric coupling of articles. The most globally cited documents (1049 citations) belong to Zein, published in the year 2002, in Biomaterials Journal. “3d Printing” is the most used author keyword, while “Mechanical properties” is the one in the keyword plus category. Mechanical properties' improvement depends on the printing parameter orientation. Raster orientation, air gap, and layer thickness are the widely explored parameters. In a nutshell, FDM has various application domains, which include medical, aerospace, automobile, electrical, food, fashion, and so on. However, the highly cited works are in the area of mechanical properties, printing parameters, medical, aerospace, automobile, and so on. Based on research trend and future scope, the following research problems are to be focused upon in the future. First, exploring different printing parameters, and their optimization and effect on mechanical properties for new materials. Second, the development of a complex and customized scaffold design, and an innovative design for 3D printing capsule with multiple containers with reference to medical domain. Third, parameter optimization of high melting point industrial-grade material in the FDM printer, and mixing and reinforcing of new materials in the existing filament for inheriting the properties of multiple materials to be utilized for light weight applications in aerospace and automobile industries.

Authors' Contributions

Mr. Rishi generated the data base, filtered it and analyzed the data. He also prepared all the figures and wrote all the relevant explanations in the manuscript. Dr. Prateek was responsible for doing the review of papers related to the field of Mechanical and materials. He also contributed in the section related to the material used in related technology. Dr. Varun prepared the review section related to the biomedical area. He was responsible for drafting sections related to biomedical covering different materials used in related area.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

The work presented in this article is funded by the Science & Engineering Research Board (SERB). Grant no. (CRG/2019/001320).