Atomic layer deposition thin film techniques and its bibliometric perspective

Atomic layer deposition (ALD) is known for depositing ultra-thin film materials that enable control of composition, highly conformal film, desirable thickness, self-saturating, and uniform deposition, and this review has established its evolvement in recent times. The ALD techniques have made more device applications possible in energy storage, solar cells, memory storage, catalysis, sensors, and many more. Its advantages and disadvantages for different modes were emphasized and the precursors used for several ALD processes were highlighted. The bibliometric approach used in this review has also revealed how ALD has evolved through the assessment of published documents, journals, authors, organizations, sponsors, and countries. The method also revealed that ALD research is limited in Africa, however, the first two ALD facilities were confirmed to be acquired by T.C. Jen at the University of Johannesburg, which will in turn burst ALD material research in Africa. The current study has provided researchers with a choice when considering using the ALD technique and in terms of research collaborations. It concluded by highlighting the challenges and future perspectives of ALD and bibliometric technique.


Introduction
Atomic layer deposition (ALD) involves the process of depositing thin films on the surface of a substrate. These films are conformal with a desirable thickness which are self-saturated due to precursors' gaseous interaction on the growth surface. The foundation of ALD started by developing atomic layer epitaxy (ALE) in 1970 by Dr. Tuomo Suntola and his associates. They used the technology in depositing thin film materials in the industry and was first used in electroluminescent (TFEL) displays [1]. After Intel introduced ALD into their industry in 2007, it quickly became the focus of the semiconductor production industry [2,3]. Beyond microelectronics, the choice of applications for ALD may have increased, including catalysis [4], energy storage [5], phase-change memory materials (PCMMs) [6], photovoltaics [7], ovonic threshold switches (OTSs) [8], and many more.
Several researchers have written different reviews that covered the ALD fundamentals, scientific growth, and applications [1,2,[9][10][11][12][13]. Also, ALD's historical evolution has been studied. For example, Puurunen focused on the early years of the development of ALE [1]. As the development of ALE continues, reactors were designed to integrate metals and their oxides, numerous thin films were fabricated non-epitaxially, and the adoption of a common name ALD was established [11]. In the twentieth century, Parsons and co-workers described the backgrounds of ALD and also reported its growth in the American Vacuum Society [14]. Nevertheless, the ALD techniques and its research evolvement is ripe for a review.
The evolvement of ALD thin film research can also be established by exploring the use of bibliometric analysis, which can provide information about the ALD community. Bibliometric is a powerful tool that offers the opportunity for reviewing the outputs of science, structure, and dynamics of scientific disciplines [15]. Content is one of the primary methods used in bibliometric analysis [16]. The content review tried to pinpoint the most current hotspots by examining the frequency of the authors' keywords and other dispersions. In general, a bibliometric study is a useful tool for connecting authors and organizations to various countries around the world. By using such an approach, it is also possible to establish the research areas and current topics in the study area [17][18][19][20][21].
However, this review focused on the ALD techniques and how its research has evolved in recent times. In this present study, we have utilized both qualitative and quantitative approaches (two key factors of literature review) to establish a broad perspective about ALD technique in terms of history, processes, research, and collaborative network. To the best of my knowledge, no work has presented the ALD technique with such a prospect. It also concluded by emphasizing the challenges and future perspectives of advancing the ALD technique.

Brief history of ALD
The concept of ALD was initially presented by Prof. Valentin Aleskovsky during his doctoral research [22]. Thereafter, he was able to publish an article based on ALD principles titled "Molecular Layering." The study was carried out with Prof. Stanislav Koltsov's at Leningrad Technological Institute (LTI) in 1960 [22]. In 1970, the creation of atomic layer epitaxy (ALE) technique was established by Dr. Tuomo Suntola and his associates. The technique was used in the industry to develop thin film [1]. The technique was initially employed electroluminescent (TFEL) displays [1]. Thereafter, this technology was presented in 20 nations [1]. The technology was ground-breaking when Suntola and coworkers came up with an inert gas reactor that enables more inclusion of compound reactants such as water vapor (H 2 O), metal chloride (MCl 3 ), and hydrogen sulfide (H 2 S) in the process of ALE [1]. The research started at the Helsinki University of Technology in 1980; the ground-breaking technology was presented at the SID conference [1]. Subsequently, in Microchemistry Ltd, the ALE technology was developed for other applications such as heterogeneous catalysts and photovoltaic devices by Suntola. The ALD technology and manufacturing company Microchemistry Ltd was sold and became ASM Microchemistry Oy in 1999, which was the only manufacturing company of ALD reactors commercially. In the early stage of the 2000s, Picosun Oy and Beneq Oy manufacturing companies sprout up their expertise in Finland based on ALD reactors. The manufacturers of ALD reactors started increasing rapidly and the ALD technology became the industrial revolution of semiconducting applications, as a furtherance of Moore's law [1]. This innovation led to the European SEMI award and Millennium Technology Prize received by Tuomo Suntola in 2004 and 2018 [1], respectively.

General process of ALD
The ALD procedure is known as a bottom-up method. As seen in Fig. 1, the necessary ALD thin film can be created in two or more procedures with various purge stages. Subjecting a substrate to a precursor or reactant is the first phase in the four-step ALD process (i.e., reactant 1). Reactant 1 is injected, generating chemical absorption on the surface of the substrate, and surface contacts between the substrate's functional group and reactant 1 is being initiated. The method is repeated till all the functional groups on the substrate have been utilized. Surface contacts cause by-products to be released from the process of the first step. The by-products and unreacted reactant 1 are eliminated in the second purge phase. In the third phase, chemical absorbed reactant 1 and the co-reactant are involved in a self-limiting surface reaction on the surface of the substrate. In the Fig. 1 Schematic of ALD general process [24] 1 3 second purge phase, the elimination of the by-products and unreacted co-reactants occurred, followed by the fourth step. This results in the formation of new generated active sites with functional groups on the substrate surface. Repeating this method on the substrate's surface results in layer-bylayer film deposition. This layer-by-layer process is shown in the right side of Fig. 1. The ALD cycle can be repeated countless of times to achieve the required thickness. As a result of its capacity to generate uniform deposition, desired thickness, and very conformal coatings on the surface of substrate, ALD has become a viable process for fuel cells, electronics, catalysis, and other applications [23]. Figure 1 depicts the general process of the ALD deposition technique.
To be as effective as feasible, ALD should be operated within a propitious temperature window to minimize sluggish thermal disintegration, growth rates, and inadequate reaction precursor condensation [2]. As a result, the phrase "temperature window" refers to the range of temperature where the film growth is saturated (see Fig. 2). ALD techniques include plasma-enhanced (PE) ALD, flash-enhanced ALD, thermal (T) ALD, and photo-assisted ALD. TALD and PEALD are the most frequent ALD techniques among these ALDs.

Thermal ALD (TALD)
TALD is a surface reaction method that allows for conformal, precise thickness control and substrate geometry. It also permits conformal coating in complicated structures with large aspect ratios. The temperature process of TALD runs from 150 to 350 °C and is considered a somewhat high-temperature technique, which limits its applicability. PEALD was created in order to solve this constraint [24].
An example of TALD film deposition was Al 2 O 3 fabrication from trimethylaluminum (TMA) and water (H 2 O) as precursor and reactant, respectively. While introducing TMA, the process showed dissociative chemisorption of TMA on the substrate surface, pumping out the leftover TMA from the compartment and allowing AlCH 3 to cover the surface. Then, H 2 O vapor is injected to react with the surface, thereby forming a reaction by-product of CH 4 and leading to a hydroxylated Al 2 O 3 surface [24]. The Al(CH 3 ) 3 and H 2 O ALD reaction is displayed in Fig. 3.

Plasma-assisted ALD (PA-ALD)
Another term for PA-ALD is plasma-enhanced ALD (PEALD). This type of ALD technique comprises plasma species with strong reactivity, allowing the deposited film temperature to be reduced without impacting the film's quality. As a result, PEALD is another process mode for improving a material's quality and attributes. Furthermore, PEALD offers a diverse spectrum of precursors that may be aided by plasma and enables for deposition of material that is difficult for TALD [25].
PEALD permits for comparatively deposition of low-temperature, and the film qualities are improved over TALDdeposited films [26]. PEALD also produces highly reactive catalysts or species at lower temperatures to boost the process. This reaction opens up a broader range of deposition precursors and substrates [24]. PEALD may therefore be used to build films on a heat-sensitive substrate. Furthermore, PEALD can generate more pure films than TALD [24]. Again, PEALD has the benefit of working at a faster pace of growth because to a diverse method. Again, PEALD has the benefit of working at a faster pace of growth through a diverse method [27].
Despite its benefits, PEALD has several unfavorable side reactions. These drawbacks include restricted conformal with a high aspect ratio or surface area, low throughput, difficult reaction chemistry, possibly poor conformal, extra growth factors, and complicated reactor designs [24]. There are four (4) types of low-pressure configurations of PEALD, namely, radical enhanced, direct plasma, remote, and direct plasma with mesh (see Fig. 4). The advantage of radical enhanced configuration over other configurations (direct and remote plasma) is because energetic plasma species do not damage the substrate or film growth and in the process of deposition, merely neutral plasma species are involved [24]. The direct plasma with mesh was developed to control the plasma surface interaction [12].
PEALD technique has been previously reported for the depositions of several nitride compounds, metal, and metal oxide films [28].

Spatial ALD (SALD)
When the volume and surface area of an ALD reactor increases, this results in lower throughput of purge time and longer pulse. The lower throughput can be controlled by the use of spatial ALD (SALD), which addresses the quality of films produced by spatially resolving the head by exposing the substrate to a precise precursor based on position and eliminating the pulse-purge compartment [29]. SALD method is capable of reaching the rate of deposition of about 3600 nm/h [2].
The working principle of SALD is different from that of TALD and PEALD. In TALD and PEALD processes, the precursors are introduced in sequential pulses separated by purge steps and therefore involve time separation, whereas SALD involves the introduction of precursors through the different inlets into the reactor as depicted in Fig. 5. In this process, the substrate moves to the different inlets for film deposition. The process also reaches surface limiting and self-terminating due to chemical reaction which makes it equivalent to ALD. Thus, SALD guarantees a high-quality film deposition at low temperatures, a specific growth per cycle, and a high aspect ratio conformal coating. SALD can be 10 2 magnitudes faster and it retains the distinctive properties of ALD. Also, SALD is a cheaper process due to the possibility to fabricate thin films at atmospheric pressure (AP-SALD) and it is easy to scale up because it does not involve an expensive and complex vacuum process. This process also faces some challenges as there is no wide range of precursors to be used [29]. Also, there are challenges such as impurity and non-desirable gas-phase reaction.

Photo-assisted ALD (PA-ALD)
Photo-assisted ALD, also known as ultraviolet (UV)enhanced ALD was recently developed. As compared to PEALD, UV-enhanced ALD has a weak activation, but it can be controlled at ease when the intensity, illumination timing by UV light, and wavelength are been modified [24]. Exposure to UV does speed up the surface reactions on the substrate [24] and thereby enhancing the quality of the film [24,25]. In another word, UV supplies energy to deposition reactions. More recently, a similar PA-ALD was developed called flash-enhanced ALD (FEALD). The only difference is that the light is used to transport heat that is of short pulses on the substrate surface and therefore drives the reactions on the surface of the substrate [30]. PA-ALD has been previously reported to deposit Al 2 O 3 , ZrO 2 , TiO 2 , ZnO, BN, Ta 2 O 5 , and GaAs using alkoxide precursors [31], and some of these precursors are tabulated in Table 2. Limited articles have reported PA-ALD research due to the challenge of constructing photo-ALD reactors [31].

Flash-enhanced ALD (FEALD)
FEALD comprises the process of exposing a substrate to a precursor. The molecules of the precursor chemically or physically absorb on the substrate surface. The substrate is next exposed to a flashlight (an arrangement of xenon flash lamps) of high intensity. This light's spectral  [27] scattering is primarily visible and near-infrared, with merely a small portion extending into the near-ultraviolet range [32]. As a result of direct interactions, the photolytic effects between the precursor molecules and light are negligible, and the flash treatment merely causes brief heating of the illuminated surface, resulting in activating the chemical reaction between the molecules of the adsorbed precursor and the second reactant or the thermally decomposing the molecule of the adsorbed precursor. Atoms from the precursors or compounds produced from the reactions form a layer on the substrate surface, followed by the purging of the volatile by-product from the compartment where the reaction took place. This process operates sequentially and it is also a layer-by-layer deposition process. Furthermore, the periodic annealing of the previously developed film caused by flash heating in each cycle may result in better film quality [30]. This technique has been reported for the depositions of Al 2 O 3 , Si, and Ru [31]. Figure 6 displays the configuration process of FEALD.
The advantages and disadvantages of different modes of the ALD process are presented in Table 1.

ALD precursors
Solids, gasses, and volatile liquids can all be used as ALD precursors. For effective mass passage, the vapor pressure must be high enough; some liquid and solid precursors must be heated. Because of the self-limiting growth, the requirement for consistent evaporation rates is less severe than in other chemical vapor deposition (CVD) processes, making solid precursors easier to use. The inert-gas valving solves the difficult problem of pulsing precursors evaporating at high temperatures [10,33]. The precursors need to be thermally stable at the growth temperature so that the decomposition of the precursor will not take away the self-saturated growth process and the benefits that come with it. Only mildly self-saturated growth can be achieved for thermally unstable precursors, that is, if the decomposition is delayed and contributes just a minor amount to total growth.
The surface sites must adsorb or react with the precursors. The adsorption type, whether dissociative or molecular, as well as reaction mechanisms, has been investigated recently in situ using data from mass spectrometry and quartz-crystal microbalance. For various precursors, different mechanisms have been discovered [10,33]. Halides, particularly chlorides, alkyl compounds, and alkoxides are common metal precursors utilized in ALD (see Table 2). Organometallic compounds like silyl and alkyl amides and cyclopentadienyl have recently received increased consideration [10,34] (see Table 2).
It was realized that precursor combination for novel material with improved properties is very important. Table S1 presents the precursors and reactants used in the previous study for ALD deposition. More of these used precursors and their reactants can also be found in previous reports [12,23,[36][37][38][39][40][41][42][43]. Most of these depositions were achieved employing thermal and plasma ALDs.

Bibliometric analysis of ALD thin film research
In this notion, bibliometric study entails the collection of methods for looking into the technical and scientific analysis of various published documents [67]. Through a series of procedures, including various publishing reports, organizations, authors, countries, and classification methodologies, the method examines several scientific journals with a focus on evolution and analysis [68]. H-index and impact factor (IF) are two (2) key metrics used to evaluate a journal's quality and are related to bibliometric analysis. IF is a method for measuring how frequently a typical journal work has been cited over a given year. The Institute of Scientific Information (ISI) recognized this instrument as a standardized technique for assessing high-quality journals [69,70]. Hirsch introduced the h-index in 2005 as a quantitative method for evaluating the overall influence of academic's performance and output. This index can be used by both institutions and periodicals [71,72].
Although Alvaro and Yanguas [3] have studied the research growth of ALD using the bibliometric technique, but their findings was not up to date. The authors investigated the growth of ALD up to 2015. Hence, this study demonstrates how rapidly ALD has expanded recently (2016-2021).

Methodology
To assess ALD published literature, including reviews, book chapters, conference papers, journal articles, books, conference reviews, and other mediums of communication, this part used the bibliometric analysis method [73]. To complete the bibliometric assessment, R-studio from the R-package software was used along with the command languages bibliometric and biblioshiny. By providing details about a published document such as authors' names, journals, affiliations, nations, h-index, and publication year, the software makes it possible to do in-depth analyses.

Data collection
Numerous data were retrieved from the Scopus database, a recognized repository for academic publications. The data from 2016 to 2021 is retrieved using the keyword TITLE-ABS-KEY [atomic-layer-deposition]. We then analyzed the published documents based on the retrieved data.  Thermal ALD Conformal coating in high aspect ratio and complex structure

Published documents
High deposition temperature and slow deposition process Plasma-enhanced ALD Low deposition temperature, low impurity, and a wide choice of precursors Low throughput and possibly poor conformal on the large surface area Spatial ALD The fast deposition rate and low deposition temperature No wide range of precursors. Impurity and non-desirable gas-phase reaction Photo-assisted ALD High growth rate Materials for photo-ALD are scarce. Challenge in construction of the photo-ALD reactors Flash-enhanced ALD Fast surface reaction High deposition temperature  (7722) articles were peer-reviewed, one thousand one hundred and ninety-seven (1197) conference papers, three hundred and fifteen (315) review articles, forty-four (44) conference review paper, eighty (80) book chapter, twenty (20) Erratum, seventeen (17) Letter, fourteen (14) Note, four (4) data paper and short survey each, three (3) book, two (2) editorial, one (1) retracted, and 3 others. Figure 8 displays the percentage distribution of these articles. The documents were also found to cross a variety of academic fields, as shown in Fig. 9.   Table 4 shows the evaluation of the top twenty (20) of these journals based on several performance indicators. It was noted that ten (10) of these journals are from the USA and 6 from the UK, indicating the most considered by several authors. Of the top 20 presented journals, the first five (5)  Considering the work of Alvaro and Yanguas [3], Thin Solid Films, Applied Surface Science, Journal of Vacuum Science and Technology A Vacuum Surfaces and Films, Chemistry of Materials, Journal of Physical Chemistry C, and ACS Applied Materials and Interfaces were still considered by authors based on the number of published documents as they are listed among the top 10 in Table 4. ACS Applied Materials and Interfaces was ranked tenth in Alvaro and Yanguas's previous study, but came first in this current investigation, implying more consideration of the journal by authors, translating to the improvement of the quality of the journal from 2016 to 2021.

Authors and institution performance
ALD material research was analyzed based on corresponding authors of published documents from 2016 to 2021 and 159 authors were determined. Figure 11 demonstrates the number of documents published by the top ten (10) productive authors. From the figure, it was noted that Kessels WMM. and Detavernier C. have the highest numbers of articles 110 and 89, respectively. The observed highest publication translates to the activeness and effectiveness of the author in the field of ALD research. It also indicates that Kessels WMM. is the most productive author among the top 10 authors. Ritala M. and Hwang CS. were ranked third and fourth with 81 and 73 documents, respectively. Followed  [3], in terms of the productive authors, it was observed that Leskela M. and Ritala M. are still at the forefront of ALD research as they fell into the first ten productive authors in this present review (Fig. 11). Kessels WMM. came tenth most productive author in their work, which was ranked the first in this current review. Combining the previous work of Alvaro and Yanguas in 2018 and this current study, Ritala M. is the most productive author.
A total number of 160 affiliations were revealed by Scopus and Table 5 displays the number of publications and the most pertinent institutions in their country. The top 20 most productive institutions are spread throughout the top 7 most populous nations, according to the table. Additionally, it was discovered that South Korea and the USA each had five institutions, indicating that these countries are actively conducting research using the ALD technique. China also accounted for 4 institutions, while Finland and Belgium accounted for 2 each, and others (France and Netherlands) accounted for one each. Even though France and Netherlands showed the least activity among the 20 most active institutions, they publish more and are ranked among the top 8 countries for the number of papers published, with 257 and 177 articles, respectively. However, China also held the top spot in terms of publications (389) and appeared in the top 5 most-read articles. The results obtained above indicate that the Chinese researchers focus more on research using ALD technology owing to their exceptional film depositions, followed by the French and the Americans.

Countries and sponsors of scientific research production
A total number of 167 countries of scientific research production were identified and the top 10 were demonstrated in Fig. 12. The figure also displayed the frequency of each  Based on ALD research from 2016 to 2021, 159 sponsors were identified, and the top 20 sponsors have been displayed in Table 6. From the same table, the top 20 sponsors came from China, the USA, South Korea, the European Union, Germany, and Taiwan. The National Natural Science Foundation of China was found to have sponsored 1366 ALD research, implying their great interest in the ALD community. National Science Foundation in the USA also sponsored 838 research documents. South Korea was ranked third with sponsored ALD research of 672 documents, followed by the U.S. Department of Energy (fourth) and European Commission (fifth) with sponsored research of ALD deposited films of 599 and 394 documents. As seen in the table, the USA, China, and the European Union appeared five times each, followed by South Korea (appeared three times), Germany and Taiwan appeared once each among the top 20 sponsors. These results have shown that the USA, China, and the European Union sponsored most of ALD thin film research.

Conclusions, challenges, and perspectives
The ALD techniques have been studied, its advantages and disadvantages for different modes have been established, and the precursors used for several ALD processes have been highlighted. Currently, thermal and plasma ALD are the most used techniques due to their large availability of precursors. The plasma ALD has the limitation of low throughput and possible poor conformal that can be achieved on a The University of Texas at Dallas USA 98 Fig. 12 Countries of scientific production substrate with a high aspect ratio or high surface area. Thermal ALD on the other hand is a technique of choice based on temperature but can deposit on such substrate. Despite the advancement of ALD in recent years, there remain some challenges to its future improvement. Fundamental research on precursor chemistry (especially on non-toxic precursors) and film deposition will continue to be crucial, particularly as ALD's application to novel materials and technologies is expanding. Precursor combination is also an important perspective; therefore, a thorough review should be done to have a great knowledge based on the choice of precursors or considering simulations using both density functional theory (DFT) and Ansys Fluent (prediction of future outcome) to avoid trial and error which may lead to wastage and to reduce cost and adverse environmental impacts. Though the simulation challenges have been pointed out by Sibanda et al. and the most crucial is the complexity of the simulation software and the simulation robust structure [74] which needs simplicity and high computing power, respectively, to avoid time wastage. Furthermore, the development of new reactors for optimization of ALD processes is essential as conventional ALDs have different disadvantages limiting material production and applications. The evolvement of ALD research was also investigated using bibliometric analysis in recent times. From 2016 to 2021, the technique revealed a total number of 9426 published documents based on ALD research. It was confirmed that 154 journals (sources) published the above number of documents. It was also affirmed that most of these journals are from the USA and the UK, indicating their consideration by authors and the quality of the journals. The Chinese Academy of Sciences and the National Natural Science Foundation of China were the most productive institution and sponsors of ALD research, respectively, suggesting that they focus more on ALD research. Considering the work of Alvaro and Yanguas [3] and this current review, it was noted that Ritala M. with a University of Helsinki affiliation from Finland is the most productive author. The bibliometric analysis also revealed that ALD research is limited in Africa, nevertheless, the first two ALD facilities (thermal and plasma ALD) have been acquired by Professor T.C. Jen at the University of Johannesburg, South Africa in 2019 [75], which will, in turn, burst ALD material research in Africa. The bibliometric method has been demonstrated to be a crucial instrument for the quantitative study of ALD research, but the method still faces some challenges which lie in the author's names and initials. Authors with the same name and initial may be presented as one and vice versa. Again, content evaluation is necessary as some articles could be listed in the investigation in which their main text might not complement the interested keywords, but because their abstract quoted these keywords. Thus, there is a need to investigate and resolve these issues for an excellent presentation of authors and documents.