Gas Cluster Ion Beam Smoothing Technique: A Review

Abstract

The need for ultraprecision smoothing strategies has been highly emphasized in modern society, and demanding requirements have posed unprecedented challenges to current precision manufacturing techniques. The Gas Cluster Ion Beam (GCIB) is a newly developed technology for materials processing. It is easy to prepare, environmentally friendly, and offers high processing accuracy and efficiency, thus garnering considerable interest and widespread application. This paper provides a comprehensive overview of the GCIB processing technology, emphasizing surface smoothing applications. It reviews various materials processed using GCIBs and analyzes the role of GCIB parameters and their effects on materials processing. Besides, the paper discusses the potential application prospects and future directions of the GCIB technology, and key conclusions and insights are identified. The discussion also includes an evaluation of the challenges and limitations associated with the GCIB, offering a balanced perspective on its practical implementation. Overall, this paper provides a thorough understanding of the GCIB technology and its role in advancing the field of surface smoothing.

Highlights

Provided a comprehensive overview of GCIB processing technology.

Summarized the role of GCIB parameters and their effects on the material processing.

Discussed the potential application prospects and future directions of GCIB technology.

Introduction to the Gas Cluster Ion Beam (GCIB)

Background

The demand for ultraprecision smoothing strategies entails the development of precision instruments, machine tools, and optical applications [1]. Particularly, demands in the fields of photolithography, optical observation, and high-energy laser weapons have enabled manufacturing on a nanometer, sub-nanometer, or even atomic scale [2, 3]. However, demanding requirements have posed unprecedented challenges to the current precision manufacturing industry. As a fundamental industry in a country, ultraprecision manufacturing has been extensively investigated in many countries, including the United States of America, Japan, Russia, Germany, South Korea, and China.

Currently, numerous ultraprecision manufacturing technologies have been developed. However, new problems have been introduced while solving the old ones. Mechanical polishing (Fig. 1a) is one of the most widely used polishing strategies because of its broad application [4], easy preparation, and low cost, but it could not remove scratches and deformations due to applied preload [5]. Notably, a hybrid polishing technology combining chemical and mechanical polishing (CMP) methods has been proposed to achieve components with low surface roughness and suppress surface roughness to sub-nanometer levels [6]. Importantly, Guo et al. [7, 8] proposed a new chemo-mechanical slurry that enabled polishing even at the atomic level for some materials (e.g., LiNbO₃ crystal and KH₂PO₄; Fig. 1b). However, materials to be machined are restricted by chemical reactions, waste is difficult to handle, and some scratches could still be found on machined surfaces.

[See PDF for image]

Fig. 1

Mainstream polishing technology. a Mechanical polishing (MP) [5], b chemical–mechanical polishing (CMP) [7], c dynamic friction polishing (DFP) [9], d plasma-assisted polishing (PAP) [14], e ultraviolet-irradiated polishing (UV) [11], and f magnetorheological polishing (MRP) [12]

Dynamic friction polishing (Fig. 1c) was proposed to offset high pressure on a workpiece but is a heat-sensitive application because it utilizes resistance heating to remove materials [9]. Plasma-assisted polishing (Fig. 1d) is another promising polishing strategy [10] characterized by high precision, low damage, and wide applicability. However, the high cost of equipment, complex operation, slow processing speed, and high surface temperature are inevitable factors that influence its application.

Another polishing strategy, called ultraviolet (UV)-irradiated polishing (Fig. 1e) [11], is essentially a CMP process combined with a UV-induced photochemical reaction. In this process, atoms on the topmost surfaces of materials are oxidized by active species, such as hydroxyl (OH) and oxygen radicals, at localized high temperatures and are finally removed as easily as removable reactants. However, although low surface roughness could be achieved, the removal efficiency is quite low, and the process is only applicable to samples with initially fine surfaces. A strategy using magnetorheological polishing (Fig. 1f) demonstrates good polishing performance but is sensitive to polishing distance, requires large polishing wheels, and is unsuitable for polishing conformal optical components; thus, it is unsuitable for conformal and high-precision deterministic polishing [12, 13].

For the mentioned situations, new processing techniques continued to emerge. Ion beam figuring technologies process materials at the atomic scale but inevitably introduce subsurface damage and have low removal efficiency [15]. Meanwhile, the GCIB smoothing strategy was developed at Kyoto University because of its low energy per atom and unique lateral sputtering effect [16]. It is highly efficient in transferring atoms from a higher position to a lower position and causes a low degree of damage. Insepov et al. [17] analyzed existing advanced surface polishing methods (Fig. 2), predicting that the extra-large GCIB could bring polished surfaces up to the theoretical limits. These methods are considered promising and of great importance to the field of tool finishing because they are superior in smoothing performance to conventional smoothing strategies. However, conventional smoothing strategies have reached their limits in terms of development, and no considerable progress in GCIB technology development has been achieved yet, likely because of the high costs of equipment and technical barriers that confine the adoption of this technology to countries with limited industrial capability. Conventional precision manufacturing processes have not yet necessitated cutting-edge capabilities, diminishing the urgency for its widespread integration. However, escalating demand for atom-scale ultraprecision manufacturing in high-end industries has ushered the development of the GCIB technology owing to the evolving requirements of next-generation advanced manufacturing paradigms.

[See PDF for image]

Fig. 2

Surface roughness obtained with various surface-polishing techniques [17]. Note: CMP, chemical mechanical polishing; MRF, magnetorheological finishing; TCP, tribo-chemical polishing; SP, super-polishing (a technique that combines metal electroplating with magnetron sputtering deposition, i.e., it is not a true smoothening process but rather a multistage process that includes various methods); CIB, cluster ion beam; XL-GCIB, extra-large GCIB

GCIB Processing Technology

GCIB is a newly developed technique applied to the field of materials processing. Yamada et al. [18, 19] reviewed important events that occurred during its development, emphasizing new advancements in recent years (Fig. 3).

[See PDF for image]

Fig. 3

Emergence of industrial ion equipment [18]

A GCIB is formed by the ionization of gas molecules or their mixtures, and strategies for preparing GCIBs [20] include gas aggregation, laser evaporation, gas cooling, and ultrasonic dispersion. Gas aggregation is one of the most commonly used methods to date. The general principle is to turn a gas current into a GCIB by vacuum evaporation, high-voltage ionization, acceleration, magnetic screening and focusing. By adjusting the parameters of a gaseous reaction system, such as gas composition, reaction temperature, pressure, electric field, and other control methods, gas cluster ions with different sizes, shapes, and properties could be obtained.

Figure 4 illustrates the schematic of a GCIB apparatus [21]. Clusters form as source gases pass through the nozzle and skimmer through adiabatic condensation expansion [22]. The skimmer filters out lateral single ions and small clusters, allowing only central cluster beams to pass through directly. Subsequently, the clusters are charged by the ionizer and accelerated by the accelerator, acquiring a certain amount of energy. Upon passing through an analysis magnet, clusters with high specific charge are removed, whereas clusters with low specific charge are largely unaffected. Finally, after further neutralization, the cluster beam is irradiated directly onto a substrate through the aperture.

[See PDF for image]

Fig. 4

Schematic of a GCIB apparatus [21]

A gas cluster typically contains tens to thousands of atoms bonded by Wan der Waals forces, ranging from 1 to 100 nm in size. Each cluster has few charges, and characteristics distinct from those of single-atom ion beams are produced, including increased energy intensity, expanded heat impact zone, lateral sputtering effects, and ability to transport more materials under the same beam conditions compared with monomer ion beam [23]. Additionally, given that the average energy of each atom in an ion beam is low, effects on materials manifest in shallow surface layers, imparting different physical and chemical properties that are comparable to those of atoms, molecules, and crystals [24]. Figure 5 shows the characteristics of monatomic and cluster ion beam irradiation. A monatomic ion beam has a deeper penetration depth than a GCIB, which would easily cause massive damage. A GCIB has tens or thousands of atoms, which makes the per-atom energy quite low. When it is impacted on a surface, its penetration depth is limited, and minimal damage can be achieved on the surface [25].

[See PDF for image]

Fig. 5

Comparison of a monatomic and b cluster ion beam characteristics

Compared with the traditional materials precision processing strategies, the GCIB technology is convenient and environmentally friendly and has high processing accuracy and efficiency, thus garnering widespread attention and application in the field of materials preparation and processing. This technology has been utilized in a wide range of materials processing applications [25], such as sub- or nanoscale polishing, atomic layer etching (ALE), and assisted deposition.

GCIB is a promising technique for smoothing owing to its unique lateral sputtering effect [26]. Different from monatomic ion beams, in which ions penetrate deep into a substrate and cause damage, cluster ion beams consist of tens or thousands of gas atoms (molecules), such as argon, and the small energy per element limits damage to the vicinity of a surface. These features allow for ion beam etching with extremely low energy per atom and surface flattening after etching, which are impossible to achieve with other ion beam technologies. Therefore, companies have used GCIBs to remove damaged layers on surfaces after Ar⁺ ion sputtering, reporting the GCIBs’ good performance [27].

Mechanism of GCIB Smoothing

The mechanism of GCIB smoothing involves physical and chemical aspects. The physical smoothing mechanism is dominant when a chemical inert gas source is used; otherwise, the chemical smoothing mechanism is dominated.

Physical Mechanism

The physical mechanism of GCIB smoothing involves the interactions of large clusters of gas atoms or materials’ surfaces. Large clusters consisting of multiple atoms and accelerating toward a target collide with one another and transfer kinetic energy to surface atoms [28]. The physical smoothing effect of GCIBs could be attributed to (i) efficient energy transfer, (ii) surface atom dynamics, and (iii) selective material removal, which facilitate ultrasmooth surface finishing [29].

Nazarov et al. [30] performed molecular dynamics simulations to study the physical processes of clusters of noble gases (e.g., Ar, Kr, and Xe) interacting with Cu and Mo surfaces. They found the front atoms of light cluster are subjected to backscattering by surface atoms, which affects cluster penetration, energy exchange, thermalization, and energy per cluster atom. Besides, the craters formed by oblique differed considerably from those generated by normal incidence angle [31]. Ieshkin et al. [32] studied the role of cluster size in physical processes, revealing that cluster size considerably influences and increases sputter yield, which is substantially higher than that considered for nonlinear sputtering; they discussed differential characteristics, such as angle, energy, and time distribution, and the dynamics of the collision area. Wang et al. [33] believed the sputtering rate differences and atom relocation–induced selective material removal are key factors in physical smoothing, as shown in Fig. 6. When a cluster under near-normal incidence impacts a surface, atoms are ejected from surface crater valleys and isotropically redeposited around the impact region. Conversely, cluster impacted on sloped surfaces induces the preferential downward dislocation of atoms, filling adjacent valleys. Sputtering rates increase at elevated surface features (hills) relative to those in valleys, progressively reducing topographic height disparities.

[See PDF for image]

Fig. 6

Mode of a cluster impacts the surface and the surface-smoothing mechanism of cluster ions [33]

Therefore, the physical mechanism of surface smoothing occurs owing to two synergistic mechanisms: (i) valley infilling by dislocated atoms (a transient state) and (ii) uniform sputtering to an optimal depth. Transitional dislocated atoms deposited in valleys are subsequently removed via sputtering during prolonged irradiation, ultimately yielding a flat surface with a minimal residual dislocated atomic layer.

Chemical Mechanism

The chemical mechanism of GCIB smoothing is characterized by surface activation, enhanced atomic mobility, selective etching, or deposition driven by gas clusters’ chemical nature and minimized damage to the substrate, contributing to effective surface refinement and functionalization [34].

Etching plays an important role in the GCIB chemical smoothing mechanism [35]. GCIB etching ionizes and accelerates large numbers of molecules into clusters, which are then directed toward a sample for irradiation. The kinetic energy of these clusters could reach several kilo electron volts, but the energy per molecule remains relatively low at a low electron volt per atom. When the clusters collide with a sample surface, many-body collisions occur, resulting in the formation of a high-temperature, high-pressure region and promoting surface chemical reactions. This unique combination allows for ultralow energy irradiation, leading to low-damage etching. Thus, GCIBs are effective for precision material removal in sensitive applications.

Sputtering yield plays a crucial role in a GCIB chemical smoothing mechanism. Sputtering involves the ejection of atoms from the surface of a material bombarded by energetic particles. Toyoda et al. [36] revealed that sputtering yields could be one or two orders of magnitude higher than that of Ar cluster ions (which have only the physical sputtering effect) when reactive gases are used. Mahoney et al. [37] found that for materials highly susceptible to cross-linking during sputtering, such as parylene C, using an O₂ GCIB is more effective than using a conventional Ar GCIB source.

Surface modifications are included in a GCIB chemical smoothing mechanism. When reactive gases are used, sample surfaces are oxidized or reduced, resulting in surface characteristics quite different from those of the matrix [38]. The smooth surfaces are created through the combined action of physical and chemical effects.

In summary, the chemical mechanism of GCIB smoothing involves complex interactions during the bombardment of a material’s surface by clusters of reactive gas atoms or molecules. Etching, enhanced sputtering yields, and surface modifications are the general factors of chemical processing. However, the chemical mechanism of GCIB smoothing has not been completely clarified, for example, (i) whether the unique spatial structure of clusters acts differently from their atomic or molecule state and (ii) whether physical effects would enhance the chemical process as both are working during the process are unclear.

Section Summary

GCIB is a newly developed approach in materials processing. In the context of GCIB, the balance between the kinetic energy of the clusters and their interaction with surfaces facilitates controlled material removal, minimizing damage while enhancing surface smoothness. This mechanism contributes to the effectiveness of GCIB in applications where precision and surface integrity are critical. It is convenient and environmentally friendly and has high processing accuracy and efficiency, thus garnering widespread attention and application.

With the understanding of the GCIB processing technologies, this review proceeds to provide an overview of GCIB smoothing applications employed in previous studies. The scope of this review encompasses target materials processed by GCIB and delves into the functional mechanism of GCIB, scrutinizing processing parameters from existing literature and elucidating the roles of each parameter. Furthermore, previous applications are investigated, and processing parameters applied in various GCIB scenarios are summarized. Finally, the review discusses the application of GCIB processing techniques, identifies research gaps, and draws conclusions.

GCIB Smoothing on Different Materials

GCIB smoothing improves the surface smoothness of nearly any material [25] and has been widely applied in multiple fields [19, 39], including photoelectrocatalysis, reactive GCIB etching and smoothing processes, reactive neutral-cluster-beam-injection etching processes, biomaterial surface processing, analytical instrumentation applications, and nano-electronic device fabrication. Many materials, including metals, semiconductors, and compounds, are feasible for GCIB processing. These materials also include soft (e.g., GaSb [40]) and hard materials (e.g., diamond [33, 41]), see details in Table 1.

Table 1. Target materials processed by GCIB

Materials		Researchers	Years	Purposes
Metal	Ag, Au, Cu, W, Zr, Ti	Yamada et al. [23]	2001	General research
	NiFe	Kirkpatrick et al. [25]	2003	Smoothing
	GaSb	Li et al. [40]	2003	Smoothing
	Nb	Insepov et al. [17]	2007	Polishing
	FeCo	Hinoura et al. [42]	2014	Etching
	Au	Saleem et al. [43]	2016	Nanostructures
	Cu, Co, Ni, Ru, Ta	Toyoda et al. [44]	2019	Etching
	NiPd	Ieshkin et al. [45]	2020	Topography & composition
	Al, Mo	Kireev et al. [29]	2023	General research
	NiTi	Chernysh et al. [38]	2024	Surface modification
Si-based	SiC	Fenner et al. [46]	2001	Smoothing
	Si, SiO2	Bourelle et al. [47]	2005	Smoothing
	Si	Isogai et al. [48]	2008	Surface modification
	SOI	Selvaraja et al. [49]	2011	Uniformity correction
	Fused silica	Korobeishchikov et al. [50]	2019	Smoothing
	Si	Zeng et al. [51]	2020	Cleaning
	Si	Kireev et al. [52]	2022	Origin of nanoripples
	Fused silica	Nikolaev et al. [31]	2023	Crater formation
	Silicon nitride film	Takeuchi et al. [53]	2025	Surface stress modification
Carbon-based	Polycrystalline diamond film	Greer et al. [54]	2000	Polishing
	Carbon overcoat	Nagato et al. [21]	2008	Smoothing
	Diamond	Wang et al. [33, 41]	2022	Polishing
Others	Sapphire	Fenner et al. [46]	2001	Smoothing
	CaF2, glass	Kirkpatrick et al. [25]	2003	Smoothing
	YBCO	Hatzistergos et al. [55]	2004	Defective layer removal
	MgF2	Bourelle et al. [47]	2005	Smoothing
	Lithium niobate	Siew et al. [56]	2018	Smoothing
	ZnO	Zeng et al. [57]	2020	Scratches removal
	Photoresist	Cho et al. [58]	2020	Particles removal
	LiB3O5	Korobeishchikov et al. [59]	2021	Polishing
	Polymer composites	Collinson et al. [34]	2022	Etching
	Polyimide	PHI (CHINA) Limited [27]	2023	Damage layer removal
	Calcite	Norrman et al. [60]	2024	Cleaning
	KGd(WO4)2	Korobeishchikov et al. [61]	2024	Microstructural & chemical effects
	Parylene C	Mahoney et al. [37]	2025	Depth profiling

As observed from Table 1, surface smoothing is the first GCIB method for production use [23]. This expectation arose from the high sputtering yields and lateral distribution of sputtered atoms, which produce surface smoothing effects that are unattainable with monomer ion beams. Within the scope of this paper, the following sections unveil the details of GCIB applications for various materials.

Metal-Based Materials

To date, many metal materials planarized by GCIB have been intensively studied [23], including Ag, Cu, Au, W, and Zr. Toyoda and Uematsu [44] conducted ALE experiments on metal materials (e.g., Cu, Co, Ni, Ru, and Ta) using continuous O₂-GCIB irradiation with and without acetylacetone. The results demonstrate substantial differences in GCIB processing among various materials. The etching depth of Co in acidic environments is 17 times that in acid-free environments, as shown in Fig. 7.

[See PDF for image]

Fig. 7

Etching depth of various metals by continuous O₂-GCIB irradiation with acetylacetone (5, 20 kV, 1 × 10¹⁶ ions/cm²) [44]

Yamada et al. [23] revealed that when the acceleration energies of an Ar cluster at normal incidence varied between 10 and 25 keV, the average roughness of Cu showed no considerable difference in smoothing efficiency (Fig. 8a). The average roughness decreased monotonically with increasing ion dose, from the initial Ra of 6 to 1.3 nm at an ion dose of 8.0 × 10¹⁵ ions/cm² (as shown in Fig. 8b). Toyoda and Uematsu [44] studied the etching of Cu film under acetylacetone ambient with O₂-GCIB, finding that ALE could be achieved using low energy (5 keV) GCIB irradiation.

[See PDF for image]

Fig. 8

Acceleration voltages and ion doses impact the average surface roughness of Cu with Ar cluster ions [23]. a Sputtered depth dependence of the roughness irradiated with 10–25 keV acceleration energy. b Ion-dose dependence of roughness after irradiation with 20 keV acceleration energy

Tseng et al. [62] integrated GCIB with CMP to planarize tungsten replacement metal gates (RMG) on a 300 mm SOI substrate (Fig. 9). They achieved a height variation (3σ) of less than 1.5 nm, satisfying the stringent specifications for the 14 nm technology node and future advancements.

[See PDF for image]

Fig. 9

RMG planarization scheme [62]. a (Top) Conventional CMP approach (i.e., POR); b (bottom) combined GCIB (or RIE) and CMP approach (i.e., CMPG)

Pelenovich et al. [63] explored the smoothing performance of GCIB on titanium coating; they demonstrated that with a 10 keV acceleration energy and ion dose of 1.25 × 10¹⁶ ions/cm² under normal irradiation, the roughness (Rq) of the Ti coating decreased from 3.7 to 0.8 nm. Additionally, microparticles were removed from the surface, as shown in Fig. 10.

[See PDF for image]

Fig. 10

AFM images of Ti coating surface [63]. a Before and b after Ar₁₀₀₀ cluster irradiation (10 keV, 1.25 × 10¹⁶ ions/cm²)

Kirkpatrick [25] investigated the GCIB smoothing effect of nickel–iron film (Fig. 11). The study indicated that Ar-GCIB could effectively process metal alloys, and the roughness (Ra) of NiFe film could be further reduced from 1.01 to 0.23 nm after bombardment. Additionally, GCIB smoothing was observed to be effective in enhancing the surfaces of materials that initially had good surface roughness.

[See PDF for image]

Fig. 11

AFM images of NiFe film smoothed a before and b after GCIB treatment [25]

Si-Based Materials

Compared with metal-based materials, nonmetal materials have a greater demand for GCIB technology because of the rapid development of the semiconductor industry. Mack [64] from Epion Corporation reviewed the development status of GCIB equipment, focusing on semiconductor applications and considering the GCIB technology as a powerful new tool for wafer processing.

Zeng et al. [51] investigated single-crystal silicon wafers processed by Ar-GCIB and found that the Rq value was reduced from 1.92 to 0.5 nm after irradiation and observed a beam-cleaning effect (Fig. 12). Additionally, they conducted the same GCIB treatment on Si, SiC chips, Ti coatings, and Au films, which achieved Rq values of 0.5–1.0 nm.

[See PDF for image]

Fig. 12

AFM surface topography of single-crystal silicon wafer [51]. a Before cleaning; b after chemical cleaning; c after cluster cleaning

Isogai et al. [65] studied the surfaces of Si wafers (doped with phosphorus) irradiated by Ar-GCIB, finding that irradiation altered the surface structures by increasing roughness and forming amorphous Si layers. However, the roughness could be atomically reduced, and the amorphous silicon layers could be eliminated with a subsequent annealing process in hydrogen at high temperatures (as shown in Fig. 13). Besides, the concentration of Ar in the amorphous silicon layer could be reduced by irradiating SF₆-GCIB after Ar-GCIB irradiation [48].

[See PDF for image]

Fig. 13

XTEM images of Si surfaces annealed at 1473 K after irradiation by a Ar-GCIB and b SF₆-GCIB [65]

A similar research was carried out by Selvaraja et al. [49], who used the GCIB technology to correct the nonuniformity of Si thickness on SOI wafers. They demonstrated that CHF₃-GCIB was effective in decreasing the nonuniformity of the silicon surface, and the surface damage caused by GCIB could be reduced by subsequent annealing. Bourelle et al. [47] conducted experiments to study the SF₆-GCIB smoothing effect on optical materials (Si, SiO₂, and MgF₂). The acceleration energy and ion doses were 30 keV and 2.1 × 10¹⁵ ions/cm², respectively. The results show that the surface roughness of the three materials initially increased with the irradiation angle of up to approximately 60°, where the roughness suddenly dropped. The Si and SiO₂ surfaces were eventually smoothed at 0.3 nm, whereas MgF₂ was smoothed to 1 nm.

Bakun et al. [66] proposed a method for fabricating super-smooth glass–ceramic optical surfaces by combining GCIB and accelerated neutral atom beam (ANAB). They demonstrated that the roughness (Ra) decreased to 0.2–0.3 nm after 15 min of 30 keV GCIB treatment. After 180 min of 25 keV ANAB treatment for 180 min, the Ra value further dropped to 0.15 nm (as shown in Fig. 14).

[See PDF for image]

Fig. 14

AFM image of the typical native glass–ceramic surface before and after processing [66]. a Original, b GCIB-treated, and c GCIB + ANAB-treated surfaces

Korobeishchikov et al. [50] studied the smoothing effects of Ar-GCIB on the surface morphology of fused silica. The results demonstrate that using gas clusters of different sizes and specific energies allowed the fused silica surface (Rq) to be smoothed in the middle and high spatial frequencies (the middle spatial frequencies refer to the spatial cycle length within a range of 33–0.12 mm, whereas high spatial frequencies are defined as spatial cycle length of ≤ 0.12 mm, which was defined by Lawrence, USA Livermore State Key Laboratory; Fig. 15).

[See PDF for image]

Fig. 15

AFM images of the fused silica surface obtained at different conditions [50]. a Original surface; after Ar-GCIB consecutive treated at bE = 22.5 keV, N = 1000 atom/cluster, cE = 22.5 keV, N = 600 atom/cluster, and dE = 11 keV, N = 600 atom/cluster at a dose of 1.0 × 10¹⁶ ion/cm². e Effective roughness σ_eff, root mean square roughness R_q, and etching depth D depend on ion fluence

Vasiliy et al. [67] studied the GCIB smoothing effect of 4H-SiC by a two-step high-to-low energy (15 to 5 keV) Ar-GCIB irradiation. They found that the Rq value could be reduced to 0.78 nm, which was lower than that of single-step processing at the same cluster ion dose. Besides, two-step processing had a clear effect on mechanical scratch removal and surface damage repair. Fenner et al. [46] studied the GCIB smoothing effect on CVD-grown SiC through two-step sequential O₂-GCIB processing (step 1: 25 keV, 1 × 10¹⁶ ions/cm² & step 2: 7 keV, 1 × 10¹⁶ ions/cm²), finding that a higher Ra value (0.38 nm) could be achieved (Fig. 16a & b). Interestingly, although Ar-GCIB considerably reduced the roughness of SiC, O₂-GCIB was able to more considerably improve surface smoothness and lower the extent of induced damage than Ar-GCIB.

[See PDF for image]

Fig. 16

AFM images of the SiC wafer surfaces [46]. a Original and b O₂-GCIB processed surfaces

Carbon-Based Materials

Greer et al. [54] studied the polycrystalline diamond film before and after Ar-GCIB processing, using a high dose (1.4 × 10¹⁷ ions/cm²) at 20 keV. They found the film surface roughness (Ra) could be considerably reduced from 29 to 9.4 nm (Fig. 17).

[See PDF for image]

Fig. 17

AFM images for a polycrystalline diamond film [54]. a Before and b after processing with a high dose (1.4 × 10¹⁷ ions/cm²) at 20 keV

Wang et al. [41] found that Ar-GCIB could alter diamond morphology and increase its surface roughness (Ra). A good result was achieved when SF₆-GCIB was used in polishing polycrystalline CVD after rough polishing with Ar-GCIB (Fig. 18).

[See PDF for image]

Fig. 18

AFM images of the diamond surface [41]. a Original surface; surfaces irradiated by b Ar-GCIB and c SF₆-GCIB with an ion dose of 1 × 10¹⁷ ions/cm² at an acceleration voltage of 30 keV

Nagato et al. [21] investigated the Ar-GCIB smoothing effect on a carbon overcoat with nonpatterned regular discrete track surfaces. They observed that the roughness of the patterned surface increased with track pitches. In a typical surface, roughness (Ra) could drop from 0.8 to 0.38 nm after Ar-GCIB treatment at 20 keV acceleration energy and 5 × 10¹⁵ ions/cm² irradiation (Fig. 19).

[See PDF for image]

Fig. 19

Roughness of nonpatterned surfaces [21]. a Before and b after GCIB (acceleration energy: 20 keV; dose: 5 × 10¹⁵ ions/cm²)

Other Materials

In Fig. 20a shows the AFM images of a typical CaF₂ optical surface obtained by the limits of CMP. Although the initial roughness (Ra) was below 1 nm, the surface contained a high density of shallow scratches [25]. However, the surface smoothness could be further improved after Ar-GCIB smoothing (Fig. 20b).

[See PDF for image]

Fig. 20

AFM images of CaF₂ surface [25]. a Before and b after being treated by Ar-GCIB

In Fig. 21 illustrates the glass substrate smoothed by Ar-GCIB, indicating that the surface asperities (as shown in the yellow circle of Fig. 21a) could be effectively eliminated. Although the roughness of the surface changed slightly after irradiation, the surface smoothness was improved [25]. The same smoothing effect was observed in the processing of YBCO films, during which the roughness (Ra) could be reduced to 0.8 nm at 20 keV acceleration energy and 2 × 10¹⁶ ions/cm² Ar₂₀₀₀-GCIB irradiation [68]. Hatzistergos et al. [55] employed GCIB to remove the defective top layers of YBCO films (which are high-temperature superconductors) grown on single-crystal SrTiO₃ substrates, finding that the critical current of the film remained the same and even improved, whereas a part of the defective top layer was removed.

[See PDF for image]

Fig. 21

AFM images of the glass [25]. a Before and b after treatment with Ar-GCIB

Fenner et al. [46] studied the smoothing effect of typical sapphire and silicon carbide surfaces with fine but small, random roughness together with relatively large and sharp asperities. They found that these surfaces could be efficiently smoothed by GCIBs. Specifically, the typical sapphire surface (as shown in the yellow circle of Fig. 22a) after irradiation with Ar-GCIB at 20 keV acceleration energy and 8 × 10¹⁵ ions/cm², showing a decline in average roughness from 0.36 nm to approximately 0.25 nm. The highest asperity dropped to 5 nm (1/6 of its initial value), as shown in Fig. 22b.

[See PDF for image]

Fig. 22

AFM images of the sapphire surfaces [46]. a Original and b O₂-GCIB-processed surfaces

Li et al. [40] used the GCIB technique to smooth a GaSb wafer, which is a relatively soft material prone to defects and scratches. However, after 5 × 10¹⁵ ions/cm² two-step (10 and 3 keV) O₂-GCIB smoothing, the best Rq value could be reduced to 0.18 nm (Fig. 23).

[See PDF for image]

Fig. 23

AFM images of GaSb wafer surfaces [40]. a Before and b after oxygen gas cluster processing

Siew et al. [56] presented a two-step fabrication method for ultralow loss waveguides (< 1 dB/cm) on lithium niobate, utilizing optimized Ar ion milling and GCIB smoothing. The effects of GCIB smoothing were evident in scanning electron microscopy (SEM) images, as shown in Fig. 24.

[See PDF for image]

Fig. 24

SEM images of the sample surface before and after GCIB treatment [56]

Korobeishchikov et al. [59] studied a method for finishing a LiB₃O₅ surface with Ar-GCIB at different incident energies. They found that a GCIB provided a high processing efficiency with minimal material damage, and the final Rq value could be reduced to 0.18 nm after high- and low-energy treatments (Fig. 25). The unique capabilities to improve the performance of nonlinear single-crystal-based optical devices were revealed.

[See PDF for image]

Fig. 25

AFM images of the LiB₃O₅ surface treated by Ar-GCIB under different conditions [59]. a Original surface; surfaces treated by Ar-GCIB at b 22 keV, 6 × 10¹⁵ ions/cm², c 22 keV, 6 × 10¹⁵ ions/cm², and d 22 & 10 keV, 6 × 10¹⁵ ions/cm²

Collinson et al. [34] presented low-energy GCIB bombardment (at an acceleration voltage of 2.5 kV) to prepare composite materials for nanomechanical and nanochemical analyses. They demonstrated that the Ar-GCIB can effectively remove material and produce damage-free surfaces for composites comprising polymethyl methacrylate or polystyrene confined in the 7-nm-diameter pores of an organosilicate matrix. As shown in Fig. 26, in durations over 30 s, the composite structure was revealed, and the exposed nanocomposite surface had an average Ra value of 0.3 nm.

[See PDF for image]

Fig. 26

AFM images of an OCS/PS-50 k composite after etching with a GCIB at 2.5 kV for the labeled durations [34]

Section Summary

This section reviews materials processed by GCIB. This technique is feasible for most of the materials. However, the smoothing mechanisms vary by material. For metals, chemical mechanisms predominate (e.g., enhanced sputtering, low-temperature etching, and surface modification). Therefore, reactive gases are often used. As for Si-based materials, physical bombardment and chemical etching are preferred to achieve a smooth surface. For carbon-based materials, a uniform surface can be achieved via physical bombardment, whereas a smooth surface can be realized by the proper use of chemical gases. For other materials, although physical bombardment is most common, O₂-GCIB processing is quite suitable for some materials (e.g., sapphire and GaSb).

Factors Affecting GCIB Smoothing

Gas Type

Advancements in GCIB technology and equipment have been made over the last few decades. The technology has been applied to simple (e.g., Ar, CF₄, CO₂, F₂, He, N₂, O₂, and SF₆) and mixed clusters (e.g., Ar-CH₄, Ar-H₂, B₂H₆-GeH₄, and glycerine) [69]. These source gases serve various purposes in polishing, etching, and deposition during GCIB processing.

In general, gas clusters formed by Ar, He, O₂, SF₆, and N₂ are often used for polishing. Yamada et al. [23] proved the potential applications of Ar-GCIB for surface planarization by using molecular dynamics simulation methods. The “lateral sputtering effect” was found by subsequent experiments using Ar-GCIB irradiation on Cu surfaces. Similarly, Skryleva et al. [70] used XPS and Ar₂₅₀₀₊ GCIB sputtering to analyze the Li/Nb ratio on a LiNbO₃ single-crystal surface. They indicated that GCIB can reach perfect termination conditions. Fenner et al. [46] found that CMP-induced intensive residual scratches on a SiC surface can be effectively removed by O₂-GCIB. They also found that the extent of subsurface or surface damage did not increase after O₂-GCIB polishing. Notably, O₂-GCIBs with SiC provided results superior to those obtained with Ar-GCIB [46]. Considering O₂ cluster collision, Insepov et al. [17] used molecular dynamics to study crater formation on a niobium surface with Ar and O₂ clusters (as shown in Fig. 27). Their findings indicate that the craters created by chemically active gas clusters were shallower than those created by chemically inactive gas clusters. Additionally, they observed that reflected oxygen and ejected niobium particles exhibited more pronounced lateral features than the impacts from inert gas clusters. Preliminary studies revealed that oxygen cluster impacts could produce a larger lateral sputtering effect than chemically inactive argon beams. Consequently, oxygen cluster beams were expected to achieve a considerably higher surface smoothing effect than nonreactive gases.

[See PDF for image]

Fig. 27

Molecular dynamics simulation for crater formation on a niobium surface [17]. Calculated with a single a Ar₄₂₉ cluster at 125 eV/atom and b oxygen molecular cluster (O₂)₄₂₉ at 100 eV/atom

Toyoda et al. [71] studied the effects of Ar and O₂ cluster ion beam irradiation on CVD diamond films. They found that Ar cluster irradiation decreased the diamond surface roughness, but a thin graphite layer formed as well. By contrast, O₂ cluster irradiation neither increased the roughness nor induced the formation of a graphite layer (Fig. 28). Wang et al. [33, 41] employed Ar and SF₆ GCIB to polish CVD diamond (grown polycrystalline diamond) substrates with a chip size of 10 mm× 10 mm for direct bonding to power devices. The Ra value was reduced from 334 to 0.5 nm, and the polished substrate was successfully bonded to GaN through surface-activated bonding with a Si nanolayer.

[See PDF for image]

Fig. 28

AFM images of CVD diamond films [71]. a Before irradiation, b Ar cluster ion irradiation (20 keV, 1 × 10¹⁷ ions/cm²) and c O₂ cluster ion irradiation (20 keV, 1 × 10¹⁷ ions/cm²)

Interestingly, even though all the mentioned gases were used for polishing, these gas sources functioned quite differently during polishing. Lozano et al. [72] conducted silicon polishing by Ar₃₀₀₀-GCIB irradiation at 30 keV acceleration energy and 1 × 10¹⁶ ions/cm² fluence, finding that roughness continuously increased with irradiation angle (Fig. 29a). Bourelle et al. [47] experimentally demonstrated a different result after performing SF₆-GCIB polishing on optical materials (e.g., Si, SiO₂ and MgF₂) as irradiation angles changed (Fig. 29b). The results show that the Ra value of the surface increased as the angle increased from 7° to approximately 60°, but the value considerably decreased to 0.14 nm when the incline angles increased to nearly 80°.

[See PDF for image]

Fig. 29

Gas type influence on the average roughness of surface at varying irradiation angles. a Ar-GCIB with 30 keV acceleration energy and 1 × 10¹⁶ ions/cm² dose [72]; b SF₆-GCIB with 30 keV and 2.1 × 10¹⁵ ions/cm.² [47]

Gas clusters containing O₂, SF₆, and fluorine are often used for etching. Using oxygen gas clusters resulted in the smoothest surface, the fluorine gas cluster etched the fastest, and the nonreactive argon gas cluster prevented the formation of a thick oxide. The fluoride process leads to excellent smoothing and produces relatively thin oxides and may thus be useful in molecular beam epitaxy, providing a smooth substrate to an epilayer interface and a reasonable temperature for oxide desorption [40]. Notably, combinations of various gases had distinguishable kinetic and chemical lateral sputtering effects, resulting in differences in etching speed, surface smoothness, and oxide thickness. Seki et al. [73] added 6% ClF₃ to Ar to prepare a ClF₃–Ar neutral cluster beam for reactive etching. A high-precision anisotropic etching process with an aspect ratio of 7 was achieved. Hanahara et al. [74] found that when GCIB irradiation was performed in an acetic acid environment, the acid vapor was adsorbed onto copper oxide, lowering the etching threshold and improving adhesion when combined with Ar-GCIB irradiation.

As for other gases, Cornett et al. [75] found that glycerine clusters provide more stable secondary ion signals than Xe ion beams when bombarding glutarimide. Pareek et al. [76] demonstrated that CO₂-GCIB facilitates the study of de novo purine synthesis in HeLa cells through metabolomics and mass spectrometry imaging. Sparvero et al. [77] used 70 keV (H₂O)_n⁺ GCIB-SIMS imaging to achieve high-resolution mapping of PEox in cells post-traumatic brain injury.

Source gases, such as Ar, Cl₂, and CO₂, can easily form neutral cluster beams. However, for N₂, CF₄, and O₂, a mixture of carrier gases is required to assist cooling [78]. Thopan et al. [79] investigated the use of two types of mixed gases (light and inert He and heavy molecule N₂). They showed that the energy distribution and transverse direction of ions were reduced in bombarded large biomolecules. Guo et al. [80] used the Rayleigh scattering method to study the average mixed cluster sizes of Ar-CH₄ and Ar-H₂ in supersonic gas jets, analyzing the effects of back pressure and mixing ratio on cluster characteristics. They found that Ar-CH₄ clusters formed easily at 50% Ar content, whereas Ar-H₂ clusters formed at room temperature, enhancing neutron yield and fusion efficiency. These findings are essential for understanding mixed cluster properties and improving neutron yield and efficiency in laser fusion. As GCIB is a valuable surface analysis tool, Lee et al. [81] evaluated its efficiency using a mixed gas of Ar and CO₂, finding that 96% CO₂ increased secondary ion efficiency.

The functions of different source gases are illustrated in Fig. 30, which indicates that GCIB functioned differently as source changes. Gases with oxidizing properties (e.g., fluorine and O₂) are usually employed for etching, whereas inert gases (e.g., Ar, Kr, and Xe) are often used for determining surface morphology effects, such as smoothing and pore sealing, removing atomic species from surfaces, and generating reactive surface radicals. Other gases are usually used for special applications. For example, He is mixed with other gases to form coolants, and H₂ and Cl₂ are mixed with other gases for the regulation of oxidation or reducibility. In addition, CO₂ is used for biological applications, and H₂O gas is used for GCIB-SIMS imaging. These findings underscore the importance of gas type in GCIB processing, and selecting appropriate cluster sources can considerably influence a material’s response and processing accuracy.

[See PDF for image]

Fig. 30

Applications of different source gases

Acceleration Energy

Acceleration energy affected the energy per cluster. Most of the experiments were carried out at an acceleration voltage of 20 keV, but the highest acceleration voltage (150 keV) was used for studying the impact of Ar₁₀₀ cluster ions on a highly oriented pyrolytic graphite surface [23].

Greer et al. [54] revealed that Ar-GCIB smoothing typically occurs from 15 to 30 keV. Given that an average cluster size of 2000 atoms per cluster is usually employed during smoothing (Fig. 31a), the average energy per atom was of the order of 10 eV/atom upon cluster impact with the surface and subsequent dissociation. Isogai et al. [65] showed that an insufficient acceleration voltage would lower processing efficiency, and excessive acceleration voltage would deteriorate polished surfaces, as shown in Fig. 31b and c.

[See PDF for image]

Fig. 31

Acceleration energy dependence. a Etching depth under typical acceleration energy [54], b amorphous Si layer thickness and c surface roughness observed by AFM [65]

However, Toyoda and Uematsu [44] found that the acceleration energy threshold for GCIB etching could be considerably low under an ambient acetylacetone condition (Fig. 32). They observed no material sputtering at acceleration energies below 5 kV, but clear etching occurred in the presence of acetylacetone. This finding suggests that low acceleration energies can be utilized to achieve ultrasmooth surfaces via ALE.

[See PDF for image]

Fig. 32

Time dependence of Cu thickness [44]. a Measured by quartz crystal microbalance (QCM, Inficon) and b by GCIB-ALE with acetylacetone

Irradiation Dose

Along with acceleration energy, radiation dose is another important parameter that should be considered during processing. Yamada et al. [23] studied the ion dose dependence of sputtered depth, using 20 keV Ar₂₀₀₀ cluster ions on Ag, Cu, Au, W, Zr, and Ti films deposited onto Si. They observed fairly good linearity in dose dependence for different materials, and the cluster ion dose varied from 8 × 10¹⁴ ions/cm² to 8 × 10¹⁵ ions/cm² (Fig. 33a). During their investigation of the dose dependence of SiO₂ films produced by (O₂)₁₀₀₀, they noted a clear saturation at doses over 5 × 10¹⁵ ions/cm² at different acceleration energies (Fig. 33b). During monomer ion irradiation at normal incidence, surface roughness worsened with increasing ion dose because of erosion or bubble formation inside the target. However, no roughening mechanism was observed after Ar cluster ion irradiation as the dose increased. The average roughness of the Cu surface saturated at an ion dose over 3 × 10¹⁵ ions/cm².

[See PDF for image]

Fig. 33

Ion dose dependence of a sputtered depth with 20 keV Ar cluster ions for various materials and b the thickness of SiO₂ films produced by (O₂)_n (average n is 1000) cluster ion beam treatment at total cluster energies of 3, 5, and 7 keV [23]

Zeng et al. [57] investigated the dependence of the ripple formation on the ion fluence at an incidence angle of 60°, accelerating voltage of 10 kV, and fluence range of 1 × 10¹⁶–1 × 10¹⁷ ions/cm². The experiment indicates that the formation of ripples began at low fluence; however, the drift lines were dominant. As the fluence increased, ripple crests formed, and the wavelength and height of the ripples on the bottom drift line surface increased (Fig. 34).

[See PDF for image]

Fig. 34

SEM images of ZnO substrate surfaces after Ar-GCIB irradiation at different ion fluences [57]. a 1 × 10¹⁶ ions/cm², b 4 × 10¹⁶ ions/cm², and c 1 × 10¹⁷ ions/cm² at an incidence angle of 60° and an accelerating voltage of 10 kV

Chen et al. [82] studied the dose dependence of the CO₂-GCIB smoothing effect on a ZnO surface. They showed a strong correlation between surface roughness and irradiation dose. The AFM images show that the average Ra value of the ZnO surface decreased from 16.1 to 0.9 nm as the irradiation dose increased to 1 × 10¹⁶ ions/cm² at 30 keV acceleration energy (Fig. 35). The total removed layer was approximately 65.2 nm.

[See PDF for image]

Fig. 35

AFM images of the ZnO films [82]. a Untreated surface; surface being treated by CO₂-GCIB at 30 keV acceleration energy with b 5 × 10¹⁴ ions/cm², c 5 × 10¹⁵ ions/cm², and d 1 × 10¹⁶ ions/cm²

Lu et al. [83] studied the dose dependence of Ar-GCIB on smooth LaSFN9 glass ball surfaces. Their findings indicate that the Ra value decreased from 0.86 to 0.18 nm as the dose increased from 1 × 10¹⁶ ions/cm² to 8 × 10¹⁶ ions/cm² at an acceleration energy of 10 keV (Fig. 36).

[See PDF for image]

Fig. 36

AFM images of the glass ball before and after Ar-GCIB treated [83]. a Original surface. Surfaces treated with b 1 × 10¹⁶ ions/cm² and c 8 × 10¹⁶ ions/cm² at an acceleration energy of 10 keV

Incidence Angle

Incidence angles have been intensively studied over the past decades, and considerable progress has been achieved. The early work of Toyoda et al. [36, 78] revealed the strong dependence of GCIB irradiation on incident angles and the resulting surface morphology. Yamada et al. [23] further investigated the angular distribution of sputtered Cu atoms under various conditions (Fig. 37). Their finding indicates that the measured angular distribution for 20 keV Ar monomer ions at normal incidence followed the cosine law, suggesting the isotropic ejection of atoms from the target surface.

[See PDF for image]

Fig. 37

Angular distribution of sputtered Cu atoms under different conditions [23]. a 20 keV Ar monomer, 10 and 20 keV Ar₂₀₀₀ cluster ions at normal incidence; 20 keV Ar₂₀₀₀ cluster ions at b 10°, c 30°, and d 60°

By contrast, the angular distribution for Ar cluster ions deviated and showed a shape completely different from the cosine distribution. When the incident energy was reduced to 10 keV, the angular distribution became even more flattened (Fig. 37a). For Ar₂₀₀₀ cluster ions incident at oblique angles ranging from 10° to 60° and fixed acceleration voltage of 20 keV, most of the sputtered particles were distributed in the forward direction even at an oblique incidence angle of only 10°. Additionally, no notable change in angular distribution was observed when the incidence angle varied from 10° to 60°.

Interestingly, Zeng et al. [51, 57] studied the effects of irradiation angle on ZnO nanorods, using 5 and 10 kV clusters with fluences ranging from 1 × 10¹⁶ ions/cm² to 1 × 10¹⁷ ions/cm² at incidence angles of 45°–70°. They observed the formation of ordered nanostairs on the ZnO nanorod surfaces after Ar-GCIB bombardment (as shown in Fig. 38a). Additionally, the wavelengths of nanoripples formed under a 10 kV accelerating voltage were nearly twofold as those formed under 5 kV at the same fluences. Thus, the dependencies of ripple wavelength on accelerating voltage and ion fluence were qualitatively similar to those observed on flat targets. Saleem et al. [43] found that a nanoripple array could be observed by SEM at an incidence angle of 60° after studying the effects of incident angle on the gold surface under Ar-GCIB irradiation with a fluence of 4 × 10¹⁶ clusters/cm² (Fig. 38b).

[See PDF for image]

Fig. 38

SEM images of Ar-GCIB incline angle effect on irradiated surfaces. a ZnO nanorods after irradiation with 30°–60° at an acceleration energy of 10 kV and fluence of 4 × 10¹⁶ ions/cm² [51, 57]. b Gold surface after irradiation with 60° at an acceleration energy of 30 keV [43]

Zeng et al. [51, 57] investigated the effects of oblique incidence angles (0°–80°) on a wide and flat single-crystal ZnO substrate using Ar-GCIB (10 keV, 4 × 10¹⁶ ions/cm²). They found that nanowaves formed at oblique incidence angles of 30°–60° and were oriented perpendicular to the incident direction (Figs. 39c–e). The study indicates that a flattening effect dominated at small incidence angles (Fig. 39b), and large droplet-like grooves replaced ripples, valleys, and drift lines at high irradiation angles (Fig. 39f).

[See PDF for image]

Fig. 39

SEM images of ZnO substrate surfaces irradiated by Ar-GCIB at an acceleration energy of 10 kV and dose of 4 × 10¹⁶ ions/cm² with different incidence [51, 57]. a Before irradiation. After irradiation at b 0°, c 30°, d 45°, e 60°, and f 80° (direction of the projection of incident GCIB is indicated by yellow arrows)

Lozano et al. [72] investigated the evolution of nanoripples on silicon using Ar₃₀₀₀-GCIB irradiation at 30 keV acceleration energy and 1 × 10¹⁶ ions/cm² fluence at various incidence angles. AFM images reveal that the irradiation angle affected the morphology and surface roughness of the bombarded surface. As shown in Fig. 40, the surface roughness increased with the incline angle, and the Rq value doubled as the angle increased from 50° to 70°.

[See PDF for image]

Fig. 40

AFM micrographs of Ar-GCIB irradiation (30 keV, 1 × 10¹⁶ ions/cm²) to Si (111) substrates at different angles [72]. a 30°, b 50°, and c 70° (the arrow indicates the direction of the irradiation)

However, the origin of nanoripples was not revealed until 2022, when Kireev et al. [52] investigated the first stage of nanoripple formation on a silicon surface under inclined GCIB irradiation. They found that these ripples exhibited distinct morphological evolution and protrusion geometry, and ripple characteristics were influenced by ion fluence and incident angle. They assumed that the formation of ordered surface ripples arose from the spontaneous emergence and subsequent self-organization of individual protrusions.

To explore the other effects of the oblique incidence irradiation of GCIB, Aoki and Matsuo [84] used molecular dynamics simulations to study the collisional process of Ar₂₀₀₀ GCIB impacting a Si(100) target. The results show that certain angles caused crater-like damage, whereas large angles allowed the cluster to slide without displacement (Fig. 41). These findings indicate that large irradiation angles could be used to smooth a substrate surface without causing damage.

[See PDF for image]

Fig. 41

Snapshots of Ar₂₀₀₀ clusters with 20 keV impacting on Si(100) target surface at incident angles of 30°, 60°, and 80° [84]

Aoki and Matsuo [85] also used molecular dynamics simulations to study the effects of the large irradiation angles of Ar₂₀₀₀-GCIB on surface modification processes. Their results indicate that the Ar₂₀₀₀ cluster slid on the target surface without causing damage, but it split the Si₄₀₉₆ block. The split Si₄₀₉₆ block resulted in sputtering as large chunks, re-deposition, and deformation, which smoothed the target surface.

Aoki and Matsuo [86] further confirmed the flattening process by studying large glancing angle irradiation on irregularly structured surfaces through molecular dynamics simulations. They revealed that Ar₂₀₀₀-GCIB at an 80° irradiation angle slid on the substrate surface without damage and impacted the walls of the block structures, causing multiple collisions and the dynamic deformation of the blocks (Fig. 42). Notably, a large part of the block was sputtered as a large chunk, and the target substrate was smooth, showing a single cluster impact on the small surface block.

[See PDF for image]

Fig. 42

Snapshots of Ar₂₀₀₀ clusters with 20 keV impacting on Si(100) surface attached with Si₁₆₃₈₄ or Si₆₅₅₃₆ block [86]

Cluster Size and Distribution

Chang et al. [87] revealed that cluster size considerably influences processing results. Figure 43 illustrates how cluster size affects processes. Large clusters deform surfaces, creating craters, whereas small clusters or monatomic ions penetrate surfaces, damaging subsurface layers. Only a suitable size would effectively modify substrate surfaces without causing damage. Companies such as ULVAC-PHI have studied the characteristics of large clusters, demonstrating that Ar₂₅₀₀ GCIB efficiently removes damaged layers and restores the correct chemical structure of polyimide (PI) layers, proving that no chemical damage occurs during sputtering even on polymer samples, such as PI [27].

[See PDF for image]

Fig. 43

Impact depth of sputtering with various ion cluster sizes [87]

Artyushkova [88] measured the sputter rates of SiO₂, NiO_x, and crystal TiO₂ thin films for five different settings of Ar gas clusters. The clusters were identified using a time-of-flight mass separation method, with sizes ranging from 500 atoms/cluster to 2000 atoms/cluster. The results indicate that the sputter rate on inorganics could be increased by using a small GCIB cluster size. However, chemical damage to sputter-sensitive inorganic materials was alleviated with a large GCIB cluster size, albeit at the cost of a reduced sputter rate.

Aoki et al. [89] validated through molecular dynamics simulations that ultra-large-scale Ar_100,000-GCIB had a cleaning effect on surface nanoparticles and caused low surface damage. As shown in Fig. 44, the impurity particles adsorbed on the surface migrated and detached from their original adsorption points after the bombardment of the clusters, thereby achieving the goal of surface cleaning. During the cleaning process, surface damage was related to the energy of each atom, and low energy per atom resulted in small damage.

[See PDF for image]

Fig. 44

Snapshots of Ar_100,000 30 keV impact on Si(100) target with nanoparticle attached on the target [89]

However, cluster size and distribution were difficult to control and were believed to be affected by multiple parameters, such as nose structure [90], backing pressure between the nozzle inlet and outlet, and ionization current [23]. Figure 45 shows the sketch of four common nozzles and the backing pressure dependence of cluster size (Nc) for different nozzles in xenon. The supplemental converging nozzle had the advantage of generating large clusters, and the De Laval micronozzle had the advantage of generating stable small clusters, but they had complex structures. By contrast, although cylindrical and De Laval nozzles cannot easily generate large or stable small clusters, they have relatively simple structures. Therefore, cylindrical and De Laval nozzles are commonly used despite their compromised performance.

[See PDF for image]

Fig. 45

Nozzle types and cluster size dependence [90]. a Sketch of the nozzles and b backing pressure dependence of the cluster size Nc for different nozzles in xenon

As illustrated in Fig. 45b, backing pressure has an evident effect on cluster size. Yamada et al. [23] further revealed the impacts of changes in baking pressure (P₀) on cluster size and count (Fig. 46a). The study shows that as the backing pressure increased, the peak of the size distribution shifted to larger sizes, and the size of clusters decreased to roughly 200 atoms. Notably, a considerable increase in the total number of clusters and size was observed when the baking pressure exceeded 0.2 MPa. Additionally, they found a clear correlation between ionization efficiency and cluster size. As shown in Fig. 46b, the ionization efficiency increased with cluster size at all ionization currents (ionization was performed by electrons with V_e of 300 eV).

[See PDF for image]

Fig. 46

Cluster size dependence [23]. a TOF spectra of Ar large clusters at different source gas pressure of up to 3000 atoms per cluster and b cluster size dependence of the ionization efficiency

Typically, the upstream nozzle pressure is set at about 0.5 MPa [54]. However, the pressure could be changed for various reasons, for example, reducing the gas pressure to lower the retained in the trench of a target surface during etching processes [73] and increasing the gas pressure to promote the generation of desired clusters [44, 91].

More interestingly, Yamada et al. [23] found that specific “magic numbers” lead to an increased abundance of clusters. For example, in Xe, an increase in intensity was observed at atom sizes of 13, 55, and 147. Such an effect was attributed to the icosahedral structures. However, in Ar, no size enhancement was reported, and experiments confirmed this over the range where measurements could be taken. These phenomena could be explained by the stability of certain closely packed shell structures.

Others

Toyoda et al. [20] believed that aperture diameter, beam shape, and distance between the irradiation target and aperture play a fundamental role in GCIB processing. To validate their hypothesis, they demonstrated a novel precision GCIB machining of a small mold (0.56 mm in diameter) and a free-surface mold. By installing a small aperture (1 or 0.1 mm) at the exit of the ionizer and adjusting the distance from the irradiation target to the aperture, the error of etching depth from the designed shape was reduced to 10 nm for the free-surface mold. Toyoda [92] found that the multiply charged clusters and residual gas collision would affect GCIB smoothing. The multiply charged cluster aggravated damage because of the high total energy, and the cluster ions tended to collide with residual gases, resulting in energy loss owing to their large collision cross-section.

Vasiliy et al. [67] found that the surface modified by a two-step high-to-low sequential energy (15 to 5 keV) Ar-GCIB irradiation had a lower roughness (Rq: 0.78 nm) at the same cluster ion dose compared with a surface that underwent a single high-energy cluster treatment at 15 or 5 keV. The Rq values were 1.05 and 0.78 nm, respectively. Besides, clear effects on mechanical scratch removal and surface damage repair of two-step processing were found. Pelenovich et al. [93] investigated decreasing three-step Ar cluster treatment on a Si surface. Their study indicates that using a decreasing three-step Ar cluster treatment at 15, 8, and 5 keV not only can shorten treatment time but also lower the surface roughness (Fig. 47). Solid evidence demonstrates that GCIB irradiation is influenced by energy-stepped processing and the efficiency and quality of a machined surface can be improved [40, 59]. However, the mechanism has not been revealed yet.

[See PDF for image]

Fig. 47

AFM images of Si surface before and after exposure to various Ar-GCIB treatment conditions [93]. a Mechanically treated initial sample; after Ar cluster treatment at b 15 keV, c 8 keV, and d 5 keV; e after three-step Ar cluster treatment at 15, 8, and 5 keV

Section Summary

In this section, factors associated with GCIB processing are investigated. However, numerous parameters can affect processing results, and mainstream research has focused on gas type, acceleration energy, irradiation dose, incidence angles, and cluster size and distribution. Gas type plays a critical role in the smoothing mechanism, and suitable gas types can considerably influence a material’s response and processing accuracy. Acceleration energy is associated with physical aspects, and low acceleration energy with reactive irradiation environments can be used for atomic scale etching. Irradiation dose is generally believed to have an impact on final results and tends to be stable during irradiation saturation. The interaction mechanism between a gas cluster and the target surface can be quite complex and is affected by multiple factors. However, normal irradiation can be effective for most applications.

Despite the importance of cluster size and distribution, they have not been comprehensively researched. In general, they affect the per-atom energy and ionization of clusters and would be affected by gas type, pressure, and nose structures. However, no comprehensive and mature explanation for this influence has been formulated. The main study for this part is mainly performed through experimentation. As for other parameters, solid evidence has been shown to surprise researchers, but the latent mechanism still needs to be discovered.

Discussions and Future Directions

The literature shows that GCIB precision processing is a new technology using gas ion clusters as catalysts and utilizes chemical reaction and physical bombardment principles differently from traditional processing methods to process other materials. This processing technology not only allows for the precise machining of complex structures and shapes but also controls the physical and chemical properties of material surfaces, improving mechanical, optoelectronic, and thermal properties. However, a comprehensive understanding of the technique parameters and corrective matching to different processing requirements are key elements for the application and development of the GCIB technology.

Comprehensive Understanding of Technique Parameters

GCIB processing is affected by multiple factors (e.g., gas type, acceleration energy, irradiation dose, incidence angle, and cluster size and distribution). These parameters play different roles in processing and are potentially entangled with one another, complicating the process. The following key comments can be made:

• Although many gas sources have been intensively studied, they still occupy only an extremely small portion of all the available gases [69]. Additionally, even a relatively stable gas may react with some materials in ways that would not occur in a normal environment (e.g., N₂ reacting with C). The situation becomes complex when mixed gas sources are involved [44]. For the further trends, more gas sources need to be explored, including not only simple pure gas but also complex mixed gases.

• The recent maximum acceleration voltages for smoothing range from 15 to 30 keV because the average energy per atom should exceed 10 eV/atom to make the collisions effective [54]. Owing to the demanding requirements of future development (ultrahard material processing, cross-scale machining, high efficiency, and high precision), high acceleration energy should be employed.

• Irradiation dose is commonly believed to be positively correlated with the roughness of a machined surface [82, 83]. However, these positive correlations are neither completely accepted by research nor theoretically proven because some new features (or stable new materials) might be generated and accumulate and cause an increase in roughness [57]. Besides, the saturated bombardment dose of a material for a specific gas type has not been explored.

• For most of the materials, roughness increased with irradiation angle. However, for Si, SiO₂, and MgF₂, roughness suddenly dropped when the irradiation angle exceeded 60° with SF₆-GCIB [47]. Notably, when processed at an incidence angle over 80°, the obtained surface is smoother than achieved at a normal incidence angle. The role of incidence angle should be intensively investigated for a wide range of materials and gas types.

• Cluster size and distribution considerably affect GCIB processing results (e.g., high penetration depth can be achieved by a small-size cluster, and a high cleaning effect was observed when studying large-size clusters) [87]. However, generating clusters of the same size was different from generating a control. As the cluster beam consists of a series of clusters with different sizes, filtering suitable clusters without decreasing beam intensity is challenging. Owing to an increase in deterministic demand, processes carried out at uniform cluster size and fine beam are expected to become the trend.

Methodology to Target Requirements

GCIB smoothing typically requires the removal of small amounts of surface material and is effective for producing excellent surfaces on materials with relatively good initial surfaces. However, a large removal rate for some rough material surfaces can be obtained when using reactive gases. Additionally, changes in surface treatment methods, even those for aligning the treatment sequence for samples, caused variations in processing results. Therefore, the following comments might be made:

• Given that the use of reactive gases enabled GCIB technology to remove large amounts of materials with high efficiency [41], more material removal models are expected to be explored. The GCIB technology is not limited to smoothing and can be used for figuring samples. Cross-scale GCIB treatments (millimeter-level surface flatness, micron-level waviness, and nanoscale roughness) for materials are anticipated to be a future trend.

• The functions of single-GCIB treatments are quite limited. However, efficiency and surface roughness could be improved with multistep sequential treatments [93]. Therefore, a combination of several treatments is expected to be another trend in GCIB development.

Conclusions

• According to the review above, the following conclusions could be identified:

  • GCIBs are promising and valuable in the field of tool finishing processes because they could provide smoothing performance superior to that of conventional smoothing strategies [17].

  • Gas type plays an important role in processing, and many gas types have been explored over the past decades [69]. Owing to the diversity of GCIB applications, more gas types (including mixed gases or artificially synthesized gases) would be explored.

  • GCIBs with smaller beam diameters and more uniform intensity distributions will cater to the future requirement. As the current GCIBs’ diameters range from several millimeters to tens of millimeters and beam intensity generally follows a Gaussian distribution, meeting the needs of processing microstructure at the millimeter level is challenging [20]. With the future development in ultraprecision machining, finer GCIBs are anticipated to become a trend.

  • Cluster size has a key influence on the process, and this influence has been proposed by many researchers [27, 87, 88–89]. However, the generated clusters consisted of a series of clusters ranging from monatomic clusters to clusters with thousands of atoms, which hindered deterministic processing. GCIB with a narrow cluster size distribution and even specific cluster sizes would be appropriate for processing.

  • As the advantages of GCIB smoothing with multistep sequential treatment have been proven [93], it has been recommended as another trend. During processing, GCIB parameters might change several times in an orderly manner.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (52035009), the Natural Science Foundation of Guangdong Province (2024B1515020027) and Key Research Program of the Department of Education of Guangdong Province (2024ZDZX1022). This work also supported by ShenZhen Engineering Research Center for Semiconductor-specific Equipment (XMHT20230111003). The authors would like to acknowledge the supports of them.

Author contributions

Ke Ge Xie: Conceptualisation, Formal analysis, Investigation, Data curation, Original draft, Review & Editing; Yuan Xie: Formal analysis, Investigation, Review & Editing; Hui Deng: Conceptualisation, Supervision, Funding acquisition, Review & Editing.

Funding

National Natural Science Foundation of China, 52035009, Hui Deng

Availability of Data and Material

Data will be made available on request.

Declarations

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Dai, YF; Zhou, L; Xie, XH; Jiao, CJ; Li, SY. Deterministic figuring in optical machining by ion beam. Acta Optica Sinica; 2008; 28, pp. 1131-1135. [DOI: https://dx.doi.org/10.3321/j.issn:0253-2239.2008.06.021]

2. Wilhelm, U; Hans Juergen, R; Russell, MH. Development of dioptric projection lenses for deep ultraviolet lithography at Carl Zeiss. J Micro/Nanolithogr MEMS MOEMS; 2004; 3, pp. 87-96. [DOI: https://dx.doi.org/10.1117/1.1637592]

3. Rothschild, M; Bloomstein, TM; Curtin, JE; Downs, DK; Fedynyshyn, TH; Hardy, DE; Kunz, RR; Liberman, V; Sedlacek, JHC; Uttaro, RS; Bates, AK; Van Peski, C. 157 nm: deepest deep-ultraviolet yet. J Vacuum Sci Technol B: Microelectron Nanometer Struct Process, Meas, Phenom; 1999; 17, pp. 3262-3266. [DOI: https://dx.doi.org/10.1116/1.591137]

4. Kwon, T-Y; Ramachandran, M; Park, J-G. Scratch formation and its mechanism in chemical mechanical planarization (CMP). Friction; 2013; 1, pp. 279-305. [DOI: https://dx.doi.org/10.1007/s40544-013-0026-y]

5. Luo, H; Ajmal, KM; Liu, W; Yamamura, K; Den, H. Polishing and planarization of single crystal diamonds:state-of-the-art and perspectives. Int J Extreme Manufact; 2021; 3, pp. 44-87. [DOI: https://dx.doi.org/10.1088/2631-7990/abe915]

6. Xu, H; Li, D; Wang, Z; Song, P; Zhao, Y; Su, H. Mullite-based abrasives for chemical mechanical polishing of silicon carbide. Appl Surf Sci; 2025; 692, [DOI: https://dx.doi.org/10.1016/j.apsusc.2025.162714] 162714.

7. Guo, J; Shi, H; Tong, Z; Li, L. A new chemo-mechanical slurry for close-to-atomic scale polishing of LiNbO₃ crystal. CIRP Ann; 2023; 72, pp. 293-296. [DOI: https://dx.doi.org/10.1016/j.cirp.2023.04.076]

8. Guo, J; Yang, Z; Yu, X; Zhang, P; Wang, Z; Yu, N; Tian, Y. A Novel slurry for atomic-scale polishing of potassium dihydrogen phosphate crystals. J Manuf Sci Eng; 2024; 146, pp. 1-30. [DOI: https://dx.doi.org/10.1115/1.4066184]

9. Suzuki, K; Iwai, M; Ninomiya, S; Uematsu, T. Dynamic friction polishing method utilizing resistance heating for efficient removal of electrically conductive diamond. Adv Mater Res; 2009; 76–78, pp. 325-330. [DOI: https://dx.doi.org/10.4028/www.scientific.net/AMR.76-78.325]

10. Luo, H; Ajmal, KM; Liu, W; Yamamura, K; Deng, H. Atomic-scale and damage-free polishing of single crystal diamond enhanced by atmospheric pressure inductively coupled plasma. Carbon; 2021; 182, pp. 175-184. [DOI: https://dx.doi.org/10.1016/j.carbon.2021.05.062]

11. Watanabe, J; Touge, M; Sakamoto, T. Ultraviolet-irradiated precision polishing of diamond and its related materials. Diam Relat Mater; 2013; 39, pp. 14-19. [DOI: https://dx.doi.org/10.1016/j.diamond.2013.07.001]

12. Tricard M, Dumas P and Forbes G Subaperture approaches for asphere polishing and metrology. 5638:284. https://doi.org/10.1117/12.577539

13. Wang, YQ. Study on Magnetorheological Finishing Using Large Polishing Tool for Ultra-smooth Flat Surface; 2016; Doctor, Hunan University:

14. Yamamura, K; Emori, K; Sun, R; Ohkubo, Y; Endo, K; Yamada, H; Chayahara, A; Mokuno, Y. Damage-free highly efficient polishing of single-crystal diamond wafer by plasma-assisted polishing. CIRP Ann; 2018; 67, pp. 353-356. [DOI: https://dx.doi.org/10.1016/j.cirp.2018.04.074]

15. Lin, B; Li, K-l; Cao, Z-C; Tian, H. Modeling of pad surface topography and material removal characteristics for computer-controlled optical surfacing process. J Mater Process Tech; 2019; 265, pp. 210-218. [DOI: https://dx.doi.org/10.1016/j.jmatprotec.2018.10.027]

16. (2014) What is Gas Cluster Ion Beam. Surface Analysis Techniques. ULVAC-PHI. Inc. Janan.

17. Insepov, Z; Hassanein, A; Norem, J; Swenson, DR. Advanced surface polishing using gas cluster ion beams. Nucl Instrum Methods Phys Res, Sect B; 2007; 261, pp. 664-668. [DOI: https://dx.doi.org/10.1016/j.nimb.2007.04.134]

18. Yamada, I. Materials Processing by cluster ion beams; 2016; Boca Raton, CRC Press:

19. Yamada, I; Matsuo, J; Toyoda, N; Aoki, T; Seki, T. Progress and applications of cluster ion beam technology. Curr Opin Solid State Mater Sci; 2015; 19, pp. 12-18. [DOI: https://dx.doi.org/10.1016/j.cossms.2014.11.002]

20. Toyoda, N; Houzumi, S; Mashita, T; Mitamura, T; Mochiji, K; Yamada, I. Novel precision machining using gas cluster ion beams. Surf Coat Technol; 2007; 201, pp. 8624-8627. [DOI: https://dx.doi.org/10.1016/j.surfcoat.2006.09.335]

21. Nagato, K; Tani, H; Sakane, Y; Toyoda, N; Yamada, I; Nakao, M; Hamaguchi, T. Study of gas cluster ion beam planarization for discrete track magnetic disks. IEEE Trans Magn; 2008; 44, pp. 3476-3479. [DOI: https://dx.doi.org/10.1109/tmag.2008.2001618]

22. Lazarev, AV; Semenov, TA; Belega, ED; Gordienko, VM. Dynamics of expanding gas from supercritical state in conical nozzle and cluster formation. J Supercrit Fluids; 2022; [DOI: https://dx.doi.org/10.1016/j.supflu.2022.105631]

23. Yamada, I; Matsuo, J; Toyoda, N; Kirkpatrick, A. Materials processing by gas cluster ion beams. Mater Sci Eng R Rep; 2001; 34, pp. 231-295. [DOI: https://dx.doi.org/10.1016/S0927-796X(01)00034-1]

24. Panfilov Y and Kolesnik L (2018) Vacuum methods of surface treatment with subnanometer roughness. MATEC Web of Conferences 224:01117. https://doi.org/10.1051/matecconf/201822401117

25. Kirkpatrick, A. Gas cluster ion beam applications and equipment. Nucl Instrum Methods Phys Res, Sect B; 2003; 206, pp. 830-837. [DOI: https://dx.doi.org/10.1016/S0168-583X(03)00858-9]

26. Nazarov, AV; Zavilgelskiy, AD; Ieshkin, AE; Kireev, DS; Shemukhin, AA; Chernysh, VS; Nordlund, K; Djurabekova, F. On the angular distributions of atoms sputtered by gas cluster ion beam. Vacuum; 2023; 212, 112061. [DOI: https://dx.doi.org/10.1016/j.vacuum.2023.112061]

27. GCIB, C. I. Limited 2023. https://www.coretechint.com/en/product/detail/16?back=%2Fzh-hans%2Fproduct%2Fcategory2%2F3%2F

28. Chernysh, VS; Ieshkin, AE; Kireev, DS; Nazarov, AV; Zavilgelsky, AD. Interaction of gas cluster ions with solids: Experiment and computer simulations. Surf Coat Technol; 2020; 388, 125608. [DOI: https://dx.doi.org/10.1016/j.surfcoat.2020.125608]

29. Kireev, DS; Pelenovich, VO; Yang, B; Nazarov, AV; Ieshkin, AE. Inelastic processes under gas cluster ion bombardment of metals. Vacuum; 2023; 216, 112473. [DOI: https://dx.doi.org/10.1016/j.vacuum.2023.112473]

30. Nazarov, AV; Chernysh, VS; Zavilgelsky, AD; Shemukhin, AA; Lopez-Cazalilla, A; Djurabekova, F; Nordlund, K. The cluster species effect on the noble gas cluster interaction with solid surfaces. Surf Interfaces; 2021; 26, 101397. [DOI: https://dx.doi.org/10.1016/j.surfin.2021.101397]

31. Nikolaev, IV; Stishenko, PV; Yakovlev, VV; Korobeishchikov, NG. Effect of gas cluster species on crater formation for fused silica. J Non-Crystalline Solids; 2023; 619, 122590. [DOI: https://dx.doi.org/10.1016/j.jnoncrysol.2023.122590]

32. Ieshkin, AE; Nazarov, AV; Bessmertniy, DR; Kutlusurin, IN; Shemukhin, AA. Differential characteristics of the particles sputtered by gas cluster ions at elevated temperatures. Mol Dyn Simul Vacuum; 2024; [DOI: https://dx.doi.org/10.1016/j.vacuum.2024.113064]

33. Wang J, Yamada H, Suga T and Mokuno Y (2023) Polishing of CVD Diamond for Direct Bonding Using Ar and SF6-Gas Cluster Ion Beams. In: 2023 IEEE CPMT Symposium Japan (ICSJ). IEEE, Kyoto, Japan. https://doi.org/10.1109/ICSJ59341.2023.10339595

34. Collinson, DW; Nepal, D; Zwick, J; Dauskardt, RH. Gas cluster etching for the universal preparation of polymer composites for nano chemical and mechanical analysis with AFM. Appl Surf Sci; 2022; 599, 153954. [DOI: https://dx.doi.org/10.1016/j.apsusc.2022.153954]

35. Takeuchi, M; Toyoda, N. Pressure resistance evaluation of an ultrathin SiNx membrane etched by a gas cluster ion beam. Nucl Instrum Methods Phys Res Sect B: Beam Interact Mater Atoms.; 2024; 550, 165317. [DOI: https://dx.doi.org/10.1016/j.nimb.2024.165317]

36. Toyoda, N; Kitani, H; Hagiwara, N; Matsuo, J; Yamada, I. Surface smoothing effects with reactive cluster ion beams. Mater Chem Phys; 1998; 54, pp. 106-110. [DOI: https://dx.doi.org/10.1016/S0254-0584(98)00095-9]

37. Mahoney, CM; Adib, K; Yongsunthon, R. Cluster ion bombardment and depth profiling of parylene C using Ar vs O2 gas cluster ion beams. Appl Surf Sci; 2025; 687, 162282. [DOI: https://dx.doi.org/10.1016/j.apsusc.2024.162282]

38. Chernysh, VS; Ieshkin, AE; Kireev, DS; Tatarintsev, AA; Senatulin, BR; Skryleva, EA. Surface modification of NiTi alloy by ion and gas cluster ion irradiation the role of chemical segregation. Nucl Instrum Methods Phys Res Sect B: Beam Interact Mater Atoms.; 2024; 554, 165463. [DOI: https://dx.doi.org/10.1016/j.nimb.2024.165463]

39. Yamada, I; Toyoda, N. Current research topics and applications of gas cluster ion beam processes. Nucl Instrum Methods Phys Res, Sect B; 2006; 242, pp. 143-145. [DOI: https://dx.doi.org/10.1016/j.nimb.2005.08.111]

40. Li, X; Goodhue, WD; Santeufeimio, C; Tetreault, TG; MacCrimmon, R; Allen, LP; Bliss, D; Krishnaswami, K; Sung, C. Gas cluster ion beam processing of gallium antimonide wafers for surface and sub-surface damage reduction. Appl Surf Sci; 2003; 218, pp. 251-258. [DOI: https://dx.doi.org/10.1016/s0169-4332(03)00681-0]

41. Wang J, Takeuchi K, Kataoka I and Suga T (2022) Polishing Diamond Substrates using Gas Cluster Ion Beam (GCIB) Irradiation for the Direct Bonding to Power Devices. In: 2022 International Conference on Electronics Packaging (ICEP), IEEE, Sapporo, Japan. https://doi.org/10.23919/ICEP55381.2022.9795483

42. Hinoura, R; Yamaguchi, A; Toyoda, N; Hara, KI; Yamada, I. In situ X-ray photoelectron spectroscopy study of gas cluster ion beam etching of FeCo film. Jpn J Appl Phys.; 2014; 53, 2 03DD05. [DOI: https://dx.doi.org/10.7567/JJAP.53.03DD05]

43. Saleem, I; Tilakaratne, BP; Li, Y; Bao, J; Wijesundera, DN; Chu, WK. Cluster ion beam assisted fabrication of metallic nanostructures for plasmonic applications. Nucl Instrum Methods Phys Res Sect B; 2016; 380, pp. 20-25. [DOI: https://dx.doi.org/10.1016/j.nimb.2016.05.002]

44. Toyoda, N; Uematsu, K. Atomic layer etching by gas cluster ion beams with acetylacetone. Jpn J Appl Phys.; 2019; 58, SEEA01. [DOI: https://dx.doi.org/10.7567/1347-4065/ab17c5]

45. Ieshkin, AE; Kireev, DS; Tatarintsev, AA; Chernysh, VS; Senatulin, BR; Skryleva, EA. Surface topography and composition of NiPd alloys under oblique and normal gas cluster ion beam irradiation. Surface Sci; 2020; 700, 121637. [DOI: https://dx.doi.org/10.1016/j.susc.2020.121637]

46. Fenner DB, DiFilippo V, Bennett J, Tetreault T, Hirvonen JK, Feldman LC (2001) Ion beam nanosmoothing of sapphire and silicon carbide surfaces. Eng Films with Ion Beams, Nano Diagn Mol Manuf. In: International Symposium on Optical Science and Technology, San Diego, CA, United States. https://doi.org/10.1117/12.452556

47. Bourelle, E; Suzuki, A; Sato, A; Seki, T; Matsuo, J. Sidewall polishing with a gas cluster ion beam for photonic device applications. Nucl Instrum Methods Phys Res, Sect B; 2005; 241, pp. 622-625. [DOI: https://dx.doi.org/10.1016/j.nimb.2005.07.087]

48. Isogai, H; Toyoda, E; Senda, T; Izunome, K; Kashima, K; Toyoda, N; Yamada, I. Dependence of recovery of Si surface damaged by GCIB irradiation on annealing temperature. Nucl Instrum Methods Phys Res Sect B; 2008; 266, pp. 2533-2536. [DOI: https://dx.doi.org/10.1016/j.nimb.2008.03.040]

49. Selvaraja SK, Rosseel E, Fernandez L, Tabat M, Bogaerts W, Hautala J and Absil P (2011) SOI thickness uniformity improvement using corrective etching for silicon nano-photonic device. In: 8th IEEE International Conference on Group IV Photonics, IEEE, London, UK. https://doi.org/10.1109/GROUP4.2011.6053719

50. Korobeishchikov, NG; Nikolaev, IV; Roenko, MA. Effect of argon cluster ion beam on fused silica surface morphology. Nucl Instrum Methods Phys Res, Sect B; 2019; 438, pp. 1-5. [DOI: https://dx.doi.org/10.1016/j.nimb.2018.10.019]

51. Zeng, X-M; Vasiliy, P; Rakhim, R; Zuo, W-B; Xing, B; Luo, J-B; Zhang, X-Y; Fu, D-J. Design and application of gas cluster accelerator for surface smoothing and nanostructures formation. Acta Physica Sinica; 2020; 69, 9 093601. [DOI: https://dx.doi.org/10.7498/aps.69.20191990]

52. Kireev, DS; Ryabtsev, MO; Tatarintsev, AA; Ieshkin, AE. On the origin of nanoripples on silicon by gas cluster-ion irradiation. Nucl Instrum Methods Phys Res, Sect B; 2022; 520, pp. 8-12. [DOI: https://dx.doi.org/10.1016/j.nimb.2022.03.017]

53. Takeuchi, M; Toyoda, N. Surface stress modification of silicon nitride film via argon gas cluster ion beam. MRS Adv; 2025; 10, 2 pp. 229-233. [DOI: https://dx.doi.org/10.1557/s43580-025-01176-0]

54. Greer, JA; Fenner, DB; Hautala, J; Allen, LP; DiFilippo, V; Toyoda, N; Yamada, I; Matsuo, J; Minami, E; Katsumata, H. Etching, smoothing, and deposition with gas-cluster ion beam technology. Surf Coat Technol; 2000; 133, pp. 273-282. [DOI: https://dx.doi.org/10.1016/S0257-8972(00)00876-8]

55. Hatzistergos, MS; Efstathiadis, H; Reeves, JL; Selvamanickam, V; Allen, LP; Lifshin, E; Haldar, P. Microstructural and compositional analysis of YBa₂Cu₃O₇−δ films grown by MOCVD before and after GCIB smoothing. Physica C; 2004; 405, pp. 179-186. [DOI: https://dx.doi.org/10.1016/j.physc.2004.01.024]

56. Siew, SY; Cheung, EJH; Liang, H; Bettiol, A; Toyoda, N; Alshehri, B; Dogheche, E; Danner, AJ. Ultra-low loss ridge waveguides on lithium niobate via argon ion milling and gas clustered ion beam smoothening. Opt Express; 2018; 26, 4 pp. 4421-4430. [DOI: https://dx.doi.org/10.1364/oe.26.004421]

57. Zeng, X; Pelenovich, V; Xing, B; Rakhimov, R; Zuo, W; Tolstogouzov, A; Liu, C; Fu, D; Xiao, X. Formation of nanoripples on ZnO flat substrates and nanorods by gas cluster ion bombardment. Beilstein J Nanotechnol; 2020; 11, pp. 383-390. [DOI: https://dx.doi.org/10.3762/bjnano.11.29]

58. Cho, Y; Choi, H; Mo, S; Kim, T. Removal of nano-sized surface particles by CO₂ gas cluster collisions for dry cleaning. Microelectron Eng; 2020; [DOI: https://dx.doi.org/10.1016/j.mee.2020.111438]

59. Korobeishchikov, NG; Nikolaev, IV; Atuchin, VV; Prosvirin, IP; Tolstogouzov, A; Pelenovich, V; Fu, DJ. Borate nonlinear optical single crystal surface finishing by argon cluster ion sputtering. Surf Interfaces; 2021; 27, [DOI: https://dx.doi.org/10.1016/j.surfin.2021.101520] 101520.

60. Norrman, K; Al-Yaseri, A. The effects of air plasma and argon cluster ion cleaning on quartz and calcite surfaces – Implications for rock-water-gas wettability. Int J Hydrogen Energy; 2024; 62, pp. 617-627. [DOI: https://dx.doi.org/10.1016/j.ijhydene.2024.03.112]

61. Korobeishchikov, NG; Nikolaev, IV; Atuchin, VV; Gerasimov, EY; Tolstoguzov, A; Abudouwufu, T; Fu, D. Microstructural and chemical effects of the argon cluster bombardment on a single crystal KGd(WO₄)₂ surface. Appl Physi A; 2024; 130, 11 842. [DOI: https://dx.doi.org/10.1007/s00339-024-07995-6]

62. Tseng, W-T; Long, J; Mohan, K; Kagalwala, T; Wu, C; Truong, C. A combined gas cluster ion beam (GCIB) and chemical-mechanical polish (CMP) planarization scheme for tungsten replacement metal gate (W-RMG). ECS J Solid State Sci Technol; 2016; 5, P404. [DOI: https://dx.doi.org/10.1149/2.0161607jss]

63. Pelenovich, VO; Zeng, XM; Ieshkin, AE; Chernysh, VS; Tolstogouzov, AB; Yang, B; Fu, DJ. Development of a gas cluster ion source and its application for surface treatment. J Surf Invest; 2019; 13, pp. 344-350. [DOI: https://dx.doi.org/10.1134/S1027451019020356]

64. Mack, ME. Gas cluster ion beams for wafer processing. Nucl Instrum Methods Phys Res, Sect B; 2005; 237, pp. 235-239. [DOI: https://dx.doi.org/10.1016/j.nimb.2004.12.139]

65. Isogai, H; Toyoda, E; Senda, T; Izunome, K; Kashima, K; Toyoda, N; Yamada, I. Study of Si wafer surfaces irradiated by gas cluster ion beams. Nucl Instrum Methods Phys Res Sect B; 2007; 257, pp. 683-686. [DOI: https://dx.doi.org/10.1016/j.nimb.2007.01.070]

66. Bakun, AD; Gusev, AS; Kargin, NI; Ryndya, SM; Siglovaya, NV. Method of formation of super-smooth optical surfaces using GCIB and ANAB processing. Appl Surf Sci; 2020; 523, [DOI: https://dx.doi.org/10.1016/j.apsusc.2020.146384] 146384.

67. Vasiliy, P; Zeng, X-M; Luo, J-B; Rakhim, R; Zuo, W-B; Zhang, X-Y; Tian, C-X; Zou, C-W; Fu, D-J; Yang, B. Double-step gas cluster ion beam smoothing. Acta Physica Sinica; 2021; 70, pp. 053601–1-053601-7. [DOI: https://dx.doi.org/10.7498/aps.70.20201454]

68. Zhang, ZD; Li, H; Wang, ZS; Fu, DJ. Progress in clusters ion beam with nanoscale manufacturing technology. China Surface Engineering.; 2014; 27, pp. 28-43.

69. Guo, E-F; Han, J-F; Li, Y-Q; Yang, C-W; Zhou, R. Study of argon/hydrogen mixed cluster in supersonic gas jet. Acta Physica Sinica; 2014; 63, [DOI: https://dx.doi.org/10.7498/aps.63.103601] 103601.

70. Skryleva, EA; Senatulin, BR; Kiselev, DA; Ilina, TS; Podgorny, DA; Parkhomenko, YN. Ar gas cluster ion beam assisted XPS study of LiNbO3 Z cut surface. Surf Interfaces; 2021; 26, [DOI: https://dx.doi.org/10.1016/j.surfin.2021.101428] 101428.

71. Toyoda, N; Hagiwara, N; Matsuo, J; Yamada, I. Surface treatment of diamond films with Ar and O₂ cluster ion beams. Nucl Instrum Methods Phys Res, Sect B; 1999; 148, pp. 639-644. [DOI: https://dx.doi.org/10.1016/S0168-583X(98)00769-1]

72. Lozano, O; Chen, QY; Tilakaratne, BP; Seo, HW; Wang, XM; Wadekar, PV; Chinta, PV; Tu, LW; Ho, NJ; Wijesundera, D; Chu, WK. Evolution of nanoripples on silicon by gas cluster-ion irradiation. AIP Adv.; 2013; [DOI: https://dx.doi.org/10.1063/14811171]

73. Seki, T; Yoshino, Y; Senoo, T; Koike, K; Aoki, T; Matsuo, J. Reactive etching by ClF3–Ar neutral cluster beam with scanning. Jpn J Appl Phys.; 2016; 55, 6 06HB01. [DOI: https://dx.doi.org/10.7567/JJAP.55.06HB01]

74. Hanahara S, Takeuchi M and Toyoda N. 2022. Surface preparation of metal films by gas cluster ion beams using organic acid vapor for Cu-Cu bonding. Jpn J Appl Phys 61:SF1004. https://doi.org/10.35848/1347-4065/ac5424

75. Cornett, DS; Lee, TD; Mahoney, JF; Todd, PJ. Matrix-free desorption of biomolecules using massive cluster impact. Rapid communications in mass spectrometry : RCM; 1994; 8, 12 pp. 996-1000. [DOI: https://dx.doi.org/10.1002/rcm.1290081218]

76. Pareek, V; Tian, H; Winograd, N; Benkovic, SJ. Metabolomics and mass spectrometry imaging reveal channeled de novo purine synthesis in cells. Science; 2020; 368, pp. 283-290. [DOI: https://dx.doi.org/10.1126/science.aaz6465]

77. Sparvero, LJ; Tian, H; Amoscato, AA; Sun, W-Y; Anthonymuthu, TS; Tyurina, YY; Kapralov, O; Javadov, S; He, R-R; Watkins, SC; Winograd, N; Kagan, VE; Bayır, H. Direct mapping of phospholipid ferroptotic death signals in cells and tissues by gas cluster ion beam secondary ion mass spectrometry (GCIB-SIMS). Angew Chem Int Ed; 2021; 60, pp. 11784-11788. [DOI: https://dx.doi.org/10.1002/anie.202102001]

78. Toyoda, N; Yamada, I. Gas cluster ion beam equipment and applications for surface processing. IEEE Trans Plasma Sci; 2008; 36, pp. 1471-1488. [DOI: https://dx.doi.org/10.1109/TPS.2008.927266]

79. Thopan, P; Seki, T; Yu, LD; Tippawan, U; Matsuo, J. Gas cooling secondary ions emitted by gas cluster ion beam at the travelling-wave ion guide of a Q-ToF-SIMS system. Nucl Instrum Methods Phys Res, Sect B; 2019; 450, pp. 139-143. [DOI: https://dx.doi.org/10.1016/j.nimb.2018.07.024]

80. Guo, EF; Han, JF; Li, YQ; Yang, CW; Zhou, R. Study of argon/hydrogen mixed cluster in supersonic gas jet. Acta Physica Sinica; 2014; [DOI: https://dx.doi.org/10.7498/aps.63.103601]

81. Lee, SJ; Hong, A; Cho, J; Choi, CM; Baek, JY; Eo, JY; Cha, BJ; Byeon, WJ; We, JY; Hyun, S; Jeon, M; Jeon, C; Ku, DJ; Choi, MC. Characteristics of a mixed-gas cluster ion beam for time-of-flight secondary ion mass spectrometry. Appl Surf Sci; 2022; 572, [DOI: https://dx.doi.org/10.1016/j.apsusc.2021.151467] 151467.

82. Chen, H; Liu, SW; Wang, XM; Iliev, MN; Chen, CL; Yu, XK; Liu, JR; Ma, K; Chu, WK. Smoothing of ZnO films by gas cluster ion beam. Nucl Instrum Methods Phys Res, Sect B; 2005; 241, pp. 630-635. [DOI: https://dx.doi.org/10.1016/j.nimb.2005.07.100]

83. Lu, R; Zhang, H; Mitsuya, Y; Fukuzawa, K; Itoh, S. Friction measurements of nanometer-thick lubricant films using ultra-smooth sliding pins treated with gas cluster ion beam. Appl Surf Sci; 2013; 280, pp. 619-625. [DOI: https://dx.doi.org/10.1016/j.apsusc.2013.05.036]

84. Aoki, T; Matsuo, J. Molecular dynamics study of surface modification with a glancing angle gas cluster ion beam. Nucl Instrum Methods Phys Res Sect B; 2007; 255, pp. 265-268. [DOI: https://dx.doi.org/10.1016/j.nimb.2006.11.071]

85. Aoki, T; Matsuo, J. Molecular dynamics simulations of surface smoothing and sputtering process with glancing-angle gas cluster ion beams. Nucl Instrum Methods Phys Res Sect B; 2007; 257, pp. 645-648. [DOI: https://dx.doi.org/10.1016/j.nimb.2007.01.048]

86. Aoki, T; Matsuo, J. Molecular dynamics study of glancing angle gas cluster irradiation on irregular-structured surfaces. Nucl Instrum Methods Phys Res Sect B; 2007; 261, pp. 639-642. [DOI: https://dx.doi.org/10.1016/j.nimb.2007.04.005]

87. Chang, H-Y; Lin, W-C; Chu, P-C; Wang, Y-K; Sogo, M; Iida, S-i; Peng, C-J; Miyayama, T. X-ray photoelectron spectroscopy equipped with gas cluster ion beams for evaluation of the sputtering behavior of various nanomaterials. ACS Applied Nano Materials; 2022; 5, pp. 4260-4268. [DOI: https://dx.doi.org/10.1021/acsanm.2c00202]

88. Artyushkova K (2022) GCIB Depth Profiling - Do you know the size of the cluster you're using? Physical Electronics. ULVAC-PHI, INC Chanhassen.

89. Aoki, T; Seki, T; Matsuo, J. Molecular dynamics simulations study of nano particle migration by cluster impact. Surf Coat Technol; 2016; 306, pp. 63-68. [DOI: https://dx.doi.org/10.1016/j.surfcoat.2016.04.053]

90. Aladi, M; Bolla, R; Cardenas, DE; Veisz, L; Földes, IB. Cluster size distributions in gas jets for different nozzle geometries. J Instrum; 2017; 12, pp. C06020-C06020. [DOI: https://dx.doi.org/10.1088/1748-0221/12/06/c06020]

91. Suda, T; Toyoda, N; K-i, Hara; Yamada, I. Development of Cu etching using O2 cluster ion beam under acetic acid gas atmosphere. Jpn J Appl Phys.; 2012; [DOI: https://dx.doi.org/10.1143/JJAP.51.08HA02]

92. Toyoda, N. Recent developments in gas cluster ion beam technology. J Vac Soc Japan; 2016; 59, pp. 121-127. [DOI: https://dx.doi.org/10.3131/jvsj2.59.121]

93. Pelenovich, V; Zeng, X; Zhang, X; Fu, D; Lei, Y; Yang, B; Tolstoguzov, A. Surface smoothing by gas cluster ion beam using decreasing three-step energy treatment. Coatings; 2023; 13, 942. [DOI: https://dx.doi.org/10.3390/coatings13050942]