Electron‐Induced C─F Bond Activation in Sn₆‐oxo Cluster Resist for Enhanced Sensitivity and Sub‐10‐nm Patterning

ABSTRACT

Metal‐oxo cluster (MOC) resists are promising candidates for extreme ultraviolet lithography (EUVL), but optimizing the resolution, line‐edge roughness (LER), and sensitivity (RLS) trade‐off remains challenging. Leveraging dissociative electron attachment (DEA) to activate C─F bonds provides a controllable handle to improve this RLS trade‐off, yet remains underexplored. Here, we report a room‐temperature, scalable synthesis of a tunable series of Sn‐oxo clusters and demonstrate hundred‐gram‐scale, single‐batch production. The fluorinated FPAA demonstrates enhanced sensitivity without compromising resolution or LER. Under electron beam lithography (EBL), FPAA achieves a critical dimension (CD) of 9.1 nm and an LER of 2.2 nm, and enables high‐fidelity complex patterning. Additionally, FPAA exhibits robust performance under deep ultraviolet (DUV) lithography and EUVL, achieving a CD of 20.9 nm under EUVL. FPAA also shows exceptional plasma resistance, achieving a high estimated Si:resist etch selectivity of >50:1. Mechanistic analyses indicate that irradiation‐generated low‐energy secondary electrons (LESEs) promote electron‐induced C─F activation, consistent with a DEA‐mediated pathway. Concurrently, organic ligand bridging and Sn─O─Sn network densification occur, forming a crosslinked network that drives the solubility switch. This work elucidates the lithographic mechanisms of Sn‐oxo clusters and provides design principles for MOC resists, supporting progress toward sub‐10‐nm lithographic resolution.

A fluorinated Sn‐oxo cluster resist (FPAA) leverages dissociative electron attachment (DEA) to trigger C─F cleavage under electron beam lithography (EBL) and extreme ultraviolet lithography (EUVL), enabling high‐fidelity writing of sub‐10‐nm lines and scannable QR‐code patterns for information storage. High Si:resist etch selectivity (>50:1) enables robust pattern transfer.

1. Introduction

Extreme ultraviolet (EUV, 13.5 nm) lithography, serving as the core driver for advancing Moore's Law, enables the production of semiconductor devices at sub‐7‐nm technology nodes and beyond [1, 2, 3, 4]. However, the limited photon flux and high photon energy (92 eV) in EUV exposure systems impose stringent requirements on photoresist materials [5, 6, 7, 8]. These resists must not only exhibit high sensitivity, i.e., efficient photon utilization, but also maintain excellent resolution and minimal line‐edge roughness (LER) [9]. This interdependence of resolution (R), LER (L), and sensitivity (S) constitutes the fundamental and often intractable “RLS trade‐off” [10, 11, 12]. Conventional organic chemically amplified resists (CARs), the industry mainstay, suffer from several limitations for this task [13]. Their low EUV absorption, coupled with acid diffusion‐induced blurring effects and stochastic polymer damage, fundamentally limit their performance at the most advanced nodes [14, 15]. This has driven the exploration of new resist platforms that combine high‐photon‐absorption of elements with molecular‐level structural precision [16, 17, 18, 19, 20, 21].

In this context, hybrid inorganic‐organic materials, particularly metal‐oxo clusters (MOCs), have emerged as a promising alternative [22, 23]. Unlike polymers, MOCs generally feature atomically precise sub‐3 nm molecular structures with an inorganic core composed of high‐atomic‐number elements such as tin (Sn), surrounded by organic ligands. Their inorganic cores confer high EUV absorption and enable efficient generation of low‐energy secondary electrons (LESEs) upon exposure [24, 25, 26, 27, 28]. These electrons can trigger chemical reactions in the surrounding organic ligands, such as decarboxylation or crosslinking, leading to a solubility switch that enables pattern formation. Despite these advantages, a major challenge remains: Can the sensitivity of MOCs be significantly enhanced without compromising resolution or LER? Conventional approaches to boost sensitivity, such as introducing highly reactive functional groups (e.g., carbon‐carbon double bonds) [29, 30], often lead to excessive or poorly controlled reactions, undermining pattern fidelity and exhibiting poor storage stability [31, 32]. Consequently, there is an urgent need to develop a more advanced ligand engineering strategy that introduces a controllable “trigger” to enhance sensitivity while maintaining stability.

The carbon‐fluorine (C─F) bond may offer a promising and underexplored solution to this design challenge. As one of the strongest chemical bonds in organic chemistry, the C─F bond exhibits exceptional strength, with bond dissociation energies typically in the range of 480–530 kJ mol⁻¹, which underpins the chemical and thermal stability of fluorinated materials (e.g., Teflon) [33, 34, 35]. However, the C─F bond can undergo selective cleavage under specific chemical conditions, including strong nucleophilic attack, Lewis acid catalysis, and high‐energy radiation [36, 37, 38]. Importantly for EUV and electron beam (EB) lithography, LESEs can effectively induce C─F bond cleavage through a process of dissociative electron attachment (DEA) [39, 40, 41, 42, 43].

When integrated into the ligand shell of MOCs, C─F bonds can function not merely as passive stabilizers, but as multifunctional “smart switches” that synergistically modulate several critical properties of photoresists. First, the inherent stability of C─F bonds promotes resist integrity during storage and processing. Notably, their susceptibility to DEA under irradiation introduces additional reaction pathways beyond conventional decarboxylation. This process generates reactive fluorine species and aryl radicals, which can promote crosslinking and densification of the M─O─M network, amplifying the solubility contrast between exposed and unexposed regions. Thus, this unique combination of stability and controllable reactivity under irradiation presents a novel avenue to decouple the RLS trade‐off.

In this work, we validate this hypothesis through the rational design and synthesis of a series of Sn‐oxo clusters functionalized with phenylacetic acid derivatives (FPAA, PAA, MePAA, MeOPAA, NAA). We systematically varied the ligand substituent structures to examine empirical relationships between ligand structure and lithographic performance in the Sn₆‐oxo cluster resist series. The fluorinated cluster, FPAA, was specifically designed to exploit the DEA‐mediated cleavage of the C─F bond. Our integrated approach combined a room‐temperature, scalable synthesis with comprehensive lithographic evaluation and mechanistic investigation. We demonstrated that the fluorinated FPAA achieves the best performance, exhibiting a significantly lower critical dose (D₀) than the non‐fluorinated series, without compromising resolution or LER. FPAA achieved sub‐10‐nm critical dimensions (CDs) under EBL, and it further demonstrated robust patterning capability under both deep ultraviolet (DUV, 254 nm) and EUV exposure. Furthermore, FPAA exhibited strong etch resistance during pattern transfer, with a Si:resist etch selectivity exceeding 50:1. Mechanistic investigations using x‐ray photoelectron spectroscopy (XPS) and time‐of‐flight secondary ion mass spectrometry (ToF‐SIMS) support that the enhanced performance is associated with DEA‐related electron‐induced C─F activation, which operates in parallel with decarboxylation to promote network crosslinking. Overall, this study not only introduces a high‐performance resist material but also elucidates a general design principle, i.e., leveraging DEA‐active motifs like C─F bonds for developing next‐generation photoresists that simultaneously meet the demanding requirements of resolution, LER, sensitivity, and etch resistance for sub‐10‐nm semiconductor manufacturing.

2. Results and Discussion

2.1. Molecular Design and Synthesis

To systematically investigate the influence of ligand structure on lithographic performance, we designed a series of Sn‐oxo clusters functionalized with phenylacetic acid derivatives (Figure 1a–c). Phenylacetic acid (PAA) was selected as the base ligand due to its known susceptibility to photo‐ or electron‐induced decarboxylation, which generates reactive benzyl radicals [44, 45]. To examine the influence of non‐fluorinated ligand substituents, we introduced para‐substituted variants including para‐methyl (MePAA), para‐methoxy (MeOPAA), and α‐naphthylacetic acid (NAA), thereby generating a ligand series with varied substituent structures. To examine the effect of fluorination, we included a para‐fluorinated ligand (FPAA), which introduces the possibility of LESE‐induced C─F bond cleavage via DEA (Figure 1d). In addition to their reactivity, aromatic ligands often have a higher carbon‐to‐hydrogen (C/H) ratio compared with aliphatic chains, enhancing etch resistance during pattern transfer processes [46].

FIGURE 1.

Sn₆‐oxo cluster structure and DEA‐related electron‐induced C─F activation. (a) Schematic illustration of Sn₆‐oxo cluster with carboxylate ligands R. (b) Top view of the drum‐shaped Sn─O─Sn core. (c) Side view of the drum architecture, shown in ball‐and‐stick and polyhedral views. (d) Schematic of LESE‐induced C─F bond cleavage via DEA in FPAA. Atom color code: Sn, cyan; O, red; C, gray; F, green; H atoms omitted for clarity.

Our synthetic approach represents a significant departure from conventional methods that typically require high‐temperature reflux or hydrothermal conditions [28, 47, 48, 49]. Instead, we employed a room‐temperature (≈25°C) one‐pot condensation process in toluene between n‐butylstannoic acid and the corresponding carboxylic acids, which rapidly afforded the target Sn‐oxo clusters without the need for reflux, pressurization, or complex workup procedures. This green and energy‐efficient protocol not only reduces environmental impact but also offers excellent scalability. For example, a single‐batch synthesis successfully produced 106 g of FPAA (Figure S1), demonstrating the potential for large‐scale manufacturing and seamless integration into industrial lithographic workflows.

Detailed synthetic methods for the Sn‐oxo clusters are provided in the Supporting Information. The synthesized clusters were characterized by nuclear magnetic resonance (NMR) spectroscopy, single‐crystal x‐ray diffraction (SCXRD), Fourier transform infrared (FTIR) spectroscopy, high‐resolution mass spectrometry (HRMS), and elemental analysis (Figures S2–S23).

Single‐crystal x‐ray diffraction (SCXRD) revealed that all five clusters share a common drum‐shaped Sn─O─Sn core, approximately 2 nm in diameter, with variations only in the peripheral carboxylate ligands (Figure S24; Tables S1 and S2). This structural uniformity is critical for isolating the effects of ligand substitution on lithographic performance. The “drum” architecture features top and bottom faces formed by six‐membered Sn₃(μ₃‐O)₃ rings (Figure 1b), connected by six Sn₂(μ₃‐O)₂ rectangular side units (Figure 1c). Each Sn center is six‐coordinate, bonded to three μ₃‐oxygen atoms, two μ₁‐oxygen atoms from carboxylate groups, and one butyl carbon (Sn─C). Hirshfeld surface analysis [50] based on the SCXRD structure of FPAA showed that the crystal packing is dominated by H···H and H···F/F···H interactions (Figure S25), with only minor polar contributions, consistent with dispersion‐dominated (nonpolar) packing around the organic periphery. Powder x‐ray diffraction (PXRD) patterns matched the simulated data (Figure S26), indicating high phase purity and structural consistency. All Sn‐oxo clusters exhibited excellent long‐term stability, with no detectable decomposition over 24 months under ambient storage. Thermogravimetric analysis (TGA) showed thermal stability up to 260°C for all clusters (Figure S27), ensuring compatibility with subsequent lithographic processing.

2.2. Film Quality and Lithographic Sensitivity

The patterning performance of the Sn‐oxo clusters was evaluated using a standard lithography workflow (Figure 2a). Atomic force microscopy (AFM) showed that FPAA formed smooth and uniform films with a root‐mean‐square roughness (Rq) of only 0.266 nm (Figure 2b). AFM results of the other cluster films (Figure S28) showed consistently low surface roughness (Rq < 1 nm for most; PAA was the exception at 2.68 nm), indicating excellent film‐forming properties suitable for high‐resolution lithography. Energy‐dispersive x‐ray spectroscopy (EDS) elemental mapping further confirmed uniform distributions of C, O, Sn, and F in FPAA film (Figure S29), with no signs of elemental enrichment or phase segregation. These results collectively indicate good morphological and compositional uniformity, an essential prerequisite for reliable lithographic evaluation. Film thickness was measured by the AFM step‐height method, with values of 33–43 nm across all formulations (Figure S30).

FIGURE 2.

Film preparation, wetting, and sensitivity of Sn‐oxo photoresists. (a) Schematic of the spin‐coating and lithography process. (b) AFM image of FPAA film (Rq = 0.266 nm). (c) Contact angles of the FPAA film with different solvents. (d) Static contact angles of PAA, FPAA, MePAA, MeOPAA, and NAA films. (e) AFM image of the FPAA film dose matrix (2 kV area‐exposure, EA development). (f) Overlaid contrast curves of all films. (g) D₀ and D₁₀₀ extracted from contrast curves. (h) D₀ as a function of ligand molar mass, with an empirical trendline for the non‐fluorinated series.

Static water contact angle measurements were used to assess film hydrophobicity under ambient conditions [16, 19]. All cluster films exhibited hydrophobic character, with water contact angles exceeding 90°. This hydrophobicity may help reduce moisture uptake from ambient environments and may benefit film stability in air. Among these clusters, FPAA exhibited the highest water contact angle (104.0°), consistent with the known surface‐segregating behavior of fluorinated moieties. In parallel, preliminary contact angles of candidate organic developers were measured to check their wetting compatibility with the cluster films [51]. All films were readily wetted with common organic solvents (Figure 2c, d), including toluene, ethyl acetate (EA), 2‐heptanone, cyclohexanone, and isopropanol (IPA) (Figures S31–S34; Table S3). Considering practical aspects such as availability, cost, handling, and environmental profiles, EA was selected as the standard developer for subsequent lithographic evaluation.

Sensitivity was evaluated under 2 kV area‐exposure with EA as the developer. A dose matrix was applied, and the remaining film thickness was measured by AFM to construct normalized contrast curves (Figures 2e, f, S35, and S36). From these curves, we extracted the critical dose (D₀), saturation dose (D₁₀₀), and contrast (γ) for each resist. The sensitivity trend observed followed the order: FPAA > PAA > MePAA > MeOPAA > NAA (Figure 2g). Parameter definitions, data normalization, and fitting methods are detailed in the Supporting Information. Notably, plotting D₀ against the ligand molar mass (Figure 2h) revealed an empirical trendline for the non‐fluorinated clusters. Ligand molar mass is not a direct measure of steric hindrance, and the exposure response can also be affected by steric, electronic, packing, and other related effects. Therefore, this trendline should be interpreted as an empirical structure‐performance correlation rather than a quantitative model governed solely by steric effects. These coupled factors may modulate ligand dissociation and Sn─O─Sn condensation during exposure, thereby changing the dose required for insolubilization. FPAA deviated below this empirical trendline, exhibiting higher sensitivity than the non‐fluorinated ligand series. The enhancement is associated with DEA‐related electron‐induced C─F bond activation, which introduces additional reaction pathways such as C─F cleavage and radical generation, beyond conventional decarboxylation. These pathways promote crosslinking and densification of the Sn─O─Sn network, a mechanism corroborated by subsequent XPS and ToF‐SIMS analyses.

2.3. EBL Patterning

In EUVL, the resist absorbs 92 eV photons and generates secondary electrons (SEs), which initiate chemical reactions responsible for the solubility switch and subsequent pattern formation. EBL involves similar SE‐driven mechanisms but offers greater experimental accessibility, making it a widely accepted surrogate platform for evaluating EUV resist candidates [52, 53, 54]. Here, we assessed the patterning performance of Sn‐oxo cluster resists using 30 kV EB exposure (Figure 3).

FIGURE 3.

Patterned Sn‐oxo cluster resists. Unless otherwise noted, all resist films were exposed at 30 kV in line‐exposure mode and developed for 30 s. (a) SEM images of FPAA and MePAA at pitches of 100, 50, and 40 nm after development in EA, with annotated dose, CD, and LER. (b) CD versus dose for FPAA, MePAA, MeOPAA, and NAA at a pitch of 50 nm after development in EA. (c) SEM images of FPAA line patterns obtained after development in 2‐heptanone at pitches of 100, 50, and 40 nm. (d) AFM (3D) image of the 40 nm pitch pattern in c. (e) CD‐dose curve at 50 nm pitch for the FPAA resist in different developers. (f) FPAA after area‐exposure at a dose of 3000 µC cm⁻² and development in EA: SEM image and EDS elemental maps (C, O, Sn, F, Si) of the patterned lines.

Line‐exposure experiments were conducted to evaluate the resolution limits of each resist. Both FPAA and MePAA successfully produced well‐defined line patterns at pitches ranging from 100 nm down to 40 nm (Figure 3a). PAA, in contrast, failed to resolve features below 80 nm pitch due to excessive line broadening and bridging (Figure S37). While MePAA and MeOPAA maintained pattern integrity at 50 nm pitch, their performance degraded significantly at 40 nm pitch, with LER increasing to 5.0 nm and 7.6 nm, respectively. NAA suffered from severe bridging at 40 nm pitch, precluding reliable LER extraction. Notably, FPAA demonstrated superior performance, achieving well‐defined 40 nm pitch lines at a dose of only 9 nC cm ^{− 1}, with a CD of 15.9 nm and an LER of 2.9 nm. These results highlight FPAA's exceptional balance of sensitivity, resolution, and LER control. Comprehensive dose‐pitch datasets are provided in Figures S38–S44.

At a fixed pitch of 50 nm, CD increased monotonically with line dose for all resists (Figure 3b). PAA was excluded from this comparison due to severe line broadening and bridging at 50 nm pitch. Consistent with area‐exposure results, resists bearing ligands with higher molar mass generally required higher doses to achieve comparable CDs, shifting the CD‐dose curves to the right. This behavior can be rationalized by increased ligand steric effects hindering ligand dissociation and Sn─O─Sn condensation, with possible contributions from electronic effects and molecular packing. FPAA achieved stable 50 nm pitch lines at lower doses than the non‐fluorinated resists, reflecting its enhanced sensitivity and supporting the additional DEA‐related electron‐induced C─F activation pathway discussed above. Analysis of LER as a function of dose showed that the non‐fluorinated resists exhibited a sharp increase in LER (up to 6–8 nm) at higher doses, whereas FPAA maintained a consistently low LER (≈3 nm) across the entire dose range (Figure S45). This behavior is associated with DEA‐mediated C─F bond activation, which may introduce additional reaction pathways, facilitate ligand removal, and promote uniform crosslinking within the Sn─O─Sn network. These effects result in a sharper solubility transition and significantly reduced LER.

To assess the practical development performance of different organic developers, we systematically evaluated EA, toluene, 2‐heptanone, cyclohexanone, and IPA under 30 kV line‐exposure mode. Dose‐pitch matrices for all developers are shown in Figures S46–S52. Among these, 2‐heptanone emerged as the optimal developer. At a line dose of 9 nC cm⁻¹, FPAA film developed in 2‐heptanone produced continuous, bridge‐free lines across pitches from 100 down to 40 nm (Figure 3c). At 50 nm pitch, this condition delivered a CD of 9.1 nm and an LER of 2.2 nm, representing the best performance achieved in this study. AFM images showed uniform topography and sharp line‐trench boundaries (Figure S49), with 3D profiling at 40 nm pitch confirming excellent pattern quality (Figure 3d).

As shown in Figure 3e, the relationship between CD and line dose at a 50 nm pitch is plotted for each effective developer. Across the tested dose range, 2‐heptanone consistently yielded the smallest CDs, followed by EA and IPA, while toluene produced the largest features. The corresponding LER‐dose curves (Figure S53) further reinforce this ranking: 2‐heptanone (2.2–3.1 nm), EA (2.5–3.4 nm), IPA (2.5–4.2 nm), and toluene (4.5–6.3 nm). It is worth noting that cyclohexanone did not produce regular lines under the conditions tested, likely due to poor solubility contrast or swelling (Figure S50).

After area‐exposure at 30 kV, EDS elemental mapping (C, O, Sn, F, and Si) across both lines and trenches showed uniform composition without detectable phase separation or residual material (Figure 3f), indicating uniform reactions during exposure and clean development. Under the same conditions, FPAA successfully resolved line/space patterns at 80 nm pitch (L/4S) and 60 nm pitch (L/3S), with CD/LER of 16.6/3.2 nm and 15.5/3.6 nm, respectively, at a dose of 8000 µC cm⁻² (Figure S54). Additionally, 100 nm pitch (L/3S) and 75 nm pitch (L/2S) patterns were reliably generated at doses of 5000–7000 µC cm⁻² (Figures S55 and S56). We also evaluated process flexibility by performing the area‐exposure experiment at 2 kV under otherwise identical conditions. Sensitivity increased by approximately one order of magnitude, while resolution degraded and LER increased (Figures S57–S59). This behavior is consistent with the accelerating‐voltage dependence of electron penetration and energy deposition in EBL: at 2 kV, the electron penetration depth decreases while the energy deposition per unit path length increases, enabling more efficient near‐surface exposure and lowering the required dose; however, the broader effective exposure distribution increases line broadening and LER [24, 55]. Overall, FPAA also enables robust area‐exposure patterning.

To demonstrate the application potential of FPAA for high‐resolution, high‐fidelity patterning of non‐repeating, complex geometries, we selected the Dalian University of Technology (DUT) emblem as a test pattern. Under 30 kV line‐exposure mode, SEM images showed sharply defined isolated features with no observable bridging or pattern collapse. The minimum CD was 16 nm at a line dose of 11 nC cm⁻¹ (Figure 4a). The AFM characterization further confirmed uniform height profiles and well‐defined contours, corroborating the high pattern quality (Figure S60). A comprehensive dose matrix demonstrated reproducible patterning down to 17 nm CD at a line dose of 16 nC cm⁻¹ (Figure S61), underscoring FPAA's robustness for complex design rules.

FIGURE 4.

Complex patterning in FPAA resist (30 kV EB; developed in EA for 30 s). (a) SEM images of the DUT emblem written in line‐exposure mode at a dose of 11 nC cm⁻¹. (b–d) Patterns written in area‐exposure mode at 5000 µC cm⁻², imaged by SEM and AFM (2D/3D): (b) Chinese characters representing the name of DUT; (c) DUT emblem; (d) QR code (State Key Laboratory of Fine Chemicals website; AFM confirms continuous, well‐defined pixels and machine readability); (e) Schematic demonstration of machine readability.

Area‐exposure mode was subsequently employed to pattern large‐area, data‐rich structures. At a dose of 5000 µC cm⁻², FPAA successfully reproduced three distinct complex patterns: (1) Chinese characters spelling “Dalian University of Technology” with a minimum CD of 31 nm (Figures 4b and S62); (2) the DUT emblem with a minimum CD of approximately 25 nm (Figure 4c); and (3) a fully functional QR code encoding the URL of the State Key Laboratory of Fine Chemicals, verified to be machine‐readable under optical scanning (Figure 4d,e). SEM and AFM (2D/3D) analyses confirmed uniform topography, sharp edge definition, and the absence of scumming or collapse across all fabricated patterns.

In summary, the ability of FPAA to resolve isolated features below 20 nm, coupled with uniform area‐exposure performance, positions it as a strong candidate not only for conventional semiconductor patterning, but also for emerging applications such as nanoscale data storage, optical security features, and functional metasurfaces. Its compatibility with both vector (line) and raster (area) exposure modes further enhances its versatility across research and industrial patterning platforms.

2.4. DUV and EUV Patterning

To evaluate the versatility of FPAA across lithographic platforms, we first tested its performance under DUV (254 nm) exposure (Figure 5a). The patterned features included line arrays, the acronym “DUT”, a simplified circuit layout, and an institutional emblem. All features were faithfully patterned, with uniform contrast and no signs of scumming, bridging, or collapse, confirming that FPAA retains its photosensitivity and solubility‐switching behavior under DUV irradiation. These results establish a process baseline and validate the resist's compatibility with lower‐energy exposure systems prior to EUV implementation.

FIGURE 5.

DUV and EUV patterning of FPAA resist film (developed in EA for 30 s). (a) Optical micrographs of DUV (254 nm) test patterns (line arrays, “DUT” lettering, simple circuit, and emblem). Scale bars in panel a, 500 µm. (b) Schematic of EUVL for 75 nm pitch (L/S) patterning. (c) SEM images of 75 nm pitch (L/S) patterns fabricated at the SSRF with doses of 80, 100, and 120 mJ cm⁻². (d) AFM (3D) images of the corresponding EUV patterns. The symbol “/” indicates that LER could not be reliably extracted due to line bridging or irregular features.

Subsequently, EUV patterning was performed using a mask‐projection exposure tool (λ = 13.5 nm) at the Shanghai Synchrotron Radiation Facility (SSRF) [56]. The wafer optical micrograph showed a well‐defined EUV exposure matrix (Figure S63), and selected dose fields were examined by SEM and AFM for pattern characterization. Figure 5b illustrates the imaging process for a 75 nm pitch (L/S) pattern. SEM analysis revealed that FPAA resolved 20.9 nm lines at a dose of 80 mJ cm⁻² (Figure 5c). At a higher dose of 120 mJ cm⁻², the line quality improved, yielding a CD of 31.7 nm and an LER of 4.0 nm. The corresponding AFM (3D) topography confirmed uniform line height and sharp edge definition (Figure 5d). These results demonstrate the EUV patterning capability of FPAA and support its potential as a non‐chemically amplified Sn‐oxo cluster resist for nanoscale EUV patterning.

2.5. Radiation‐Induced Chemistry of Sn‐oxo Clusters

After establishing the patterning performance of FPAA, we proceeded to examine the radiation‐induced chemical transitions that enable pattern formation and the underlying mechanism. As an auxiliary thermal‐fragmentation analysis, thermogravimetry‐mass spectrometry (TGA‐MS) provided insight into the fragmentation behavior of FPAA by revealing a three‐stage degradation pathway (Figure 6a): (i) thermal robustness up to 260°C; (ii) abrupt mass loss at 260–350°C; (iii) plateau above 350°C, with a minor secondary loss near 700°C. The second stage was accompanied by m/z = 56 (C₄H₈) and 58 (C₄H₁₀), diagnostic of butyl‐chain homolysis. The signal at m/z = 154 (C₈H₇FO₂) reflected desorption of the p‐fluorophenylacetic acid ligand, while peaks at m/z = 44 (CO₂) and 110 (C₇H₇F) indicated ligand decarboxylation to CO₂ and p‐fluorotoluene. The emergence of m/z = 182 (C₁₄H₁₄) is reasonably assigned to dimethylbiphenyl, likely formed via self‐recombination of tolyl radicals generated after defluorination of the C₇H₇F species. Collectively, these TGA‐MS results reveal thermal fragmentation fingerprints involving ligand cleavage, CO₂ elimination, and radical recombination, which help clarify the fragmentation propensity of FPAA [19].

FIGURE 6.

Reaction mechanisms induced by light and radiation in FPAA resists. (a) TGA‐MS of FPAA. (b) EPR (DMPO) of FPAA after DUV exposure. (c) UV–vis spectra of FPAA film exposed to DUV for various exposure times. (d) Solid‐state 19F NMR of FPAA before and after 1 h of DUV exposure (equal mass, identical acquisition). (e) FTIR spectra of FPAA film exposed to DUV for various exposure times. (f) Raman spectra of FPAA film exposed to various irradiation sources and doses.

Following this thermal‐fragmentation analysis, we further examined the photochemical susceptibility of the fluorinated ligand shell under DUV irradiation. These results provide auxiliary photochemical evidence and are not intended to establish mechanistic equivalence with SE‐driven EB/EUV chemistry. To capture transient radicals, electron paramagnetic resonance (EPR) spin‐trapping experiments were conducted using 5,5‐dimethyl‐1‐pyrroline N‐oxide (DMPO) as a radical scavenger under 254 nm exposure (Figure 6b). The EPR spectrum was resolved into three components: the alkoxyl adduct DMPO‐OR (a _N = 13.53 G, a _H = 7.98 G), the alkyl adduct DMPO‐R (a _N = 14.26 G, a _H = 22.90 G), and the diagnostic signal of DMPO oxidation by strongly oxidizing radicals, DMPO‐X (a _N = 13.84 G) [23, 57, 58, 59]. The hyperfine constants matched literature values, and the simulated spectrum closely replicated the experimental data. We therefore infer that exposure initially generates alkyl radicals (•R), most of which are rapidly oxidized to alkoxyl species (•OR) and trapped by DMPO. DMPO‐X arises from oxidation of DMPO by strongly oxidizing radicals. These findings support a sequential oxidation model and anchor the proposed cascade of ligand dissociation, decarboxylation, and radical‐induced condensation.

UV–vis spectroscopy was used to track the consequences of ligand loss on the Sn‐oxo core (Figure 6c). The electronic transition from O (2p) → Sn (5d) band at 200 nm decreased monotonically with irradiation time, consistent with replacement of carboxylate by Sn─O─Sn bridges [60, 61].

¹⁹F NMR spectroscopy was performed to further track the evolution of the local chemical environments of fluorine. After 1 h of DUV (254 nm) exposure, the solid‐state ¹⁹F NMR spectra exhibited no resolvable chemical shift change, whereas the total integrated signal intensity decreased by 42.1% (Figure 6d). These observations indicate that fluorine is either volatilized or transferred to NMR‐invisible sites (e.g., surface Sn‐F species). In solution‐state ¹⁹F NMR, as the exposure time increased from 0 to 105 min (Figure S64), the parent C─F resonance at −116.5 ppm decayed progressively with the concurrent growth of new peaks, indicating transformation of fluorinated ligand environments and formation of new fluorinated fragments. Thus, C─F activation is not merely destructive but redistributes fluorine within the material network.

FTIR spectra exhibited clear time‐dependent changes under DUV (254 nm) irradiation (Figure 6e). The synchronous decay of ν_as(COO⁻) (1599, 1546 cm⁻¹) and ν_s(COO⁻) (1426, 1408 cm⁻¹) [62] indicates loss of carboxylate ligation and decarboxylation, consistent with the CO₂ detected by TGA‐MS. The ν(Ar─F) [16] band at 1226 cm⁻¹ decreased by 54.0% after 30 min of DUV irradiation, indicating partial activation of aryl C─F bonds; the released F⁻ is likely captured by neighboring Sn sites to form Sn─F. Concurrent evolution of the framework modes ν(Sn─O─Sn) (604 cm⁻¹), ν(Sn─C) (561 cm⁻¹), and ν(Sn─O) (447 cm⁻¹) (Figure S65), together with a decrease in the ν(C─H) manifold at 2960–2850 cm⁻¹, points to network densification through ligand loss and Sn─O─Sn condensation. The appearance and growth of a broad ν(O─H) band at 3400–3600 cm⁻¹ indicate minor moisture uptake and the formation of Sn─OH during irradiation [63]. Taken together, the FTIR data support a solubility‐switch mechanism driven by carboxylate loss and decarboxylation, aryl C─F activation with Sn─F formation, and Sn─O─Sn condensation, collectively yielding a denser, more inorganic network that enables high‐fidelity patterning.

Under 30 kV EB exposure at doses of 2000, 4000, and 10000 µC cm⁻², as well as DUV exposure for 2 min, the Raman spectra of FPAA films exhibited clear dose‐ and time‐dependent intensity attenuation (Figure 6f). The ν(Ar─H) [64] band at 3070 cm⁻¹ decreased markedly, whereas the 1603 cm⁻¹ band, attributed to the aromatic C═C skeletal vibration, decreased only slightly, suggesting that most detached aromatic ligands remain but are edge‐functionalized. Parallel attenuation of 1292 cm⁻¹ (Ar─F) [65] and 2906/1443 cm⁻¹ (aliphatic C─H stretching and ─CH₂─ bending) [66] supports partially overlapping chemical transformations, including side‐chain ablation, C─F activation, and subsequent Sn─F formation.

To simulate electron‐induced reactions at the film surface, high‐resolution electron ionization mass spectrometry (EI‐MS) was conducted on an Orbitrap Exploris GC equipped with a direct insertion probe. Two prominent Sn isotopic clusters centered at m/z = 2012.9557 and 1916.9923 were assigned to C₆₈H₈₁F₆O₁₈Sn₆ ⁺ (FPAA‐Bu, butyl scission product) and C₆₄H₈₄F₅O₁₆Sn₆ ⁺ (FPAA‐C₈H₆FO₂, carboxylate‐loss product), respectively (Figure S66). The experimental isotopic envelopes closely matched the simulations (Figures S67 and S68). Meanwhile, low m/z fragments were identified at 58.0654 (C₄H₁₀ ⁺), 154.0433 (C₈H₇FO₂ ⁺), 109.0453 (C₇H₆F•⁺), and 43.9894 (CO₂ ⁺), which mirrored the TGA‐MS results. These observations indicate that electron‐induced fragmentation produces several characteristic fragments also observed in TGA‐MS, suggesting partially overlapping labile bonds under electron ionization and thermal fragmentation conditions.

To visualize the evolution of bond types and valence states after irradiation, XPS was performed with peak deconvolution and quantitative component analysis of C 1s, O 1s, F 1s, and Sn 3d (Figure 7a–e). Elemental atomic percentages, used as a reference for overall compositional changes, were summarized, and the composition of the initial material was consistent with its molecular formula (Figure S69). Using Sn as an internal standard (six atoms per cluster), the peak areas corresponding to the specific chemical states of C, O, and F before and after exposure were quantified. These values were corrected using relative sensitivity factors and normalized relative to Sn (Table S4).

FIGURE 7.

XPS and ToF‐SIMS of irradiation‐induced surface chemistry in FPAA resists. (a) C 1s, (b) O 1s, (c) F 1s, (d) Sn 3d, and (e) XPS survey spectra of FPAA resist films (unexposed; DUV exposed, 15 min; EB exposed, 2000 µC cm⁻² at 30 kV). (f–h) ToF‐SIMS spectra of FPAA films (unexposed; DUV, 15 min; EB dose matrix, 5–245 µC cm⁻² at 2 kV; EUV, 100 mJ cm⁻²). Intensities are normalized to the total ion intensity of each spectrum. (i–k) ToF‐SIMS images of FPAA resist films after EB exposure (dose matrix, 5–245 µC cm⁻² at 2 kV): (i) total ions, (j) SnOH+, (k) SnF+. EUV exposure (100 mJ cm⁻²): (l) SnOH⁺.

In the C 1s region, the C─C/C─H component (284.8 eV) decreased from 56.36 (unexposed) to 2.85 (DUV) and 8.85 (EB), reflecting extensive ligand fragmentation and desorption. The marked decline in C─F intensity (286.4–286.9 eV) from 4.37 to 0.88 (DUV) and 2.87 (EB), coupled with the depletion of the O─C═O component (288.8–289.2 eV) from 4.16 to 0.45 (DUV) and 2.37 (EB) [16, 60], signifies not only decarboxylation but also cleavage of the strong C─F bond, likely via DEA.

The O 1s spectra reveal a systematic reorganization of the oxygen chemical environment. The total O content increased from 15.28 to 19.04 (DUV) and 18.09 (EB), while the Sn─O─Sn component (530.5–530.8 eV) increased from 4.93 to 6.60 (DUV) and 7.45 (EB), directly evidencing radiation‐triggered oxo‐cluster condensation. Concurrently, the O─C═O component (532.0–532.3 eV) decreased, corroborating progressive ligand decarboxylation. Meanwhile, Sn─OH (531.7–532.9 eV) emerged as a new component, reaching 2.51 (DUV) and 0.76 (EB) [67, 68]. This points to hydrolysis of coordinatively unsaturated Sn sites formed during ligand ejection, a process facilitated by ambient moisture, particularly under DUV exposure.

Fluorine redistribution is equally dramatic. The depletion of the C─F signal (686.6–687.2 eV) from 5.85 to 0.14 (DUV) and 0.27 (EB) is accompanied by the appearance of the Sn─F component (684.2–684.7 eV), rising from 0 to 1.00 (DUV) and 0.43 (EB). This semi‐quantitative chemical‐state evolution suggests that part of the fluorine released via C─F cleavage is trapped by Lewis‐acidic Sn centers to form Sn─F bonds. The total fluorine content declined from 5.85 to 1.14 (DUV) and 0.70 (EB), indicative of partial volatilization of small fluorinated fragments. In addition, the Sn 3d_5/2 binding energy shifted slightly from 486.8 eV to 487.3 eV (DUV) and 487.1 eV (EB) (Figure 7d), consistent with an elevation of the Sn oxidation state [68].

ToF‐SIMS analysis further corroborated the surface chemical evolution (Figure 7f–l) [24, 60]. In the low m/z region (Figure 7f), a marked reduction in the C7H6F⁺ signal at m/z = 109 after exposure confirmed the efficient decomposition of the fluorinated aromatic ligands. Concurrently, an H₂O⁺ signal at m/z = 18 indicated surface hydration. After exposure, the emergence of Sn₃O_y species at m/z = 380–430 reflected the formation of oxygen‐rich Sn‐oxo frameworks (Figure 7g), consistent with the increased Sn oxidation state observed by XPS. In the high‐m/z region (Figure 7h), the unexposed film displayed characteristic parent‐cluster fragments at m/z 1917 and 2013, assigned in EI‐MS to C₆₄H₈₄F₅O₁₆Sn₆ ⁺ (FPAA‐C₈H₆FO₂) and C₆₈H₈₁F₆O₁₈Sn₆ ⁺ (FPAA‐Bu), respectively. These signals nearly disappeared after exposure, consistent with extensive fragmentation and reorganization of the cluster framework.

Spatially resolved ToF‐SIMS imaging further visualized the distribution and pattern selectivity of chemical species (Figure 7i–l). For the EB dose‐matrix sample, the total ion image (Figure 7i) reproduced the dose array with sharply defined boundaries, indicating confined reaction‐diffusion. The intensities of SnOH⁺ (Figure 7j) and SnF⁺ (Figure 7k) signals increased with dose and approached saturation, reflecting the progressive formation of Sn─OH and Sn─F species. In addition, detection of C₄H₁₀Sn₂ ⁺ fragment ion (Figure S70) corresponding to Sn─(CH₂)₄─Sn bridge implied radical recombination between Sn‐bound butyl chains, providing an additional densification route. The residual C₈H₆FO₂ ⁺ signal (Figure S70) indicated the incomplete removal of carboxylate ligands under the present exposure conditions. Under EUV exposure, SnOH⁺ (Figure 7l) and SnF⁺ signals (Figure S71) were observed, confirming that similar mechanistic pathways are activated across different irradiation modalities.

Based on the comprehensive experimental evidence, we propose a set of reaction pathways governing the chemical evolution and solubility switching of the Sn‐oxo clusters under EB/EUV irradiation (Figure 8). In pathway (a), SEs generated by high‐energy irradiation trigger decarboxylation of carboxylate ligands and homolytic cleavage of the Sn‐bound butyl chains. This process releases CO₂ and butyl radicals, while simultaneously creating active sites on the Sn‐oxo core and terminal Sn‐butyl radicals. Pathway (b) involves radical‐mediated grafting of butyl radicals at reactive Sn sites to form Sn─butyl─Sn bridges between neighboring clusters, promoting network insolubilization. In pathway (c), LESEs cleave C─F bonds via DEA, yielding fluoride ions (F⁻) and aryl radicals. These aryl radicals can undergo radical coupling to form aryl‐aryl crosslinks. Concurrently, F⁻ is either captured by neighboring Sn active sites to re‐coordinate as Sn─F, or removed as volatile fluorinated fragments, resulting in partial defluorination. Pathway (d) describes oxidative condensation and hydration: Sn‐oxo cores bearing reactive sites undergo oxidation to form dimers, trimers, and higher‐order Sn─O─Sn oligomers. Simultaneously, hydrolysis by ambient moisture generates Sn─OH species, further promoting structural rearrangement. Collectively, these parallel and synergistic pathways drive surface reconstruction of the Sn‐oxo clusters and progressive densification of the Sn─O─Sn network, ultimately enabling solubility switching required for high‐resolution patterning.

FIGURE 8.

Proposed reaction mechanisms induced by EB/EUV irradiation. (a) Decarboxylation and butyl‐chain homolysis release CO₂ and butyl radicals, generating reactive Sn sites. (b) Radical‐mediated grafting of butyl radicals at reactive Sn sites forms Sn─butyl─Sn bridges between neighboring clusters, promoting network insolubilization. (c) LESE‐driven DEA enables C─F cleavage, producing F⁻ and aryl radicals; aryl radicals may form aryl‐aryl crosslinks, while F⁻ re‐coordinates as Sn─F or is removed in volatile fluorinated fragments (partial defluorination). (d) Oxidative condensation and hydration yield higher‐order Sn─O─Sn oligomers and Sn─OH species, driving network densification and solubility switching.

2.6. Etch Resistance

To be practically viable, a resist must withstand aggressive plasma to serve as an effective silicon etch mask. Therefore, the etch durability of FPAA resist lines was evaluated using a Bosch deep‐Si etch (ICP‐RIE). AFM line profiles showed an initial line height of 17.2 nm before etching, while the underlying Si was etched to a depth of 881 nm (Figure 9a–c). FIB‐SEM cross‐sectional analysis showed no discernible residual FPAA layer after etching (Figure S72); thus, the consumed FPAA thickness was approximated by its initial line height, giving an estimated Si:resist etch selectivity of 51.2, demonstrating strong etch resistance under the tested Bosch ICP‐RIE conditions. For comparison, the PAA reference showed an estimated Si:resist etch selectivity of 39.5 under the same Bosch ICP‐RIE conditions (Figure S73). Related values from the literature are summarized in Table S5 for reference. The vertical sidewall profile observed in SEM images (Figure 9d) confirmed minimal transfer of LER, indicating that pattern fidelity was maintained during the etching process.

FIGURE 9.

Deep‐Si etching using FPAA resist. (a) AFM (2D/3D) images of 1 µm‐wide lines (L/2S) exposed at 30 kV with a dose of 3000 µC cm⁻² before etching; (b) AFM (2D/3D) images after etching. (c) AFM line profiles before (black) and after (red) etching; the estimated Si:resist etch selectivity was 51.2 (881/17.2). (d) SEM images of etched Si arrays (Au‐sputter‐coated): 1 µm L/S, 1 µm L/2S; 500 nm L/S, 500 nm L/2S; 200 nm L/S, 200 nm L/2S, 200 nm L/3S. Scale bars: 2 µm.

The remarkable etch selectivity observed is primarily attributed to the Sn─O─Sn inorganic framework and the exposure‐induced crosslinked network, which together help suppress etching and lateral erosion under Bosch ICP‐RIE conditions. The higher etch selectivity of FPAA compared with PAA further suggests that fluorination‐assisted network densification may contribute to the improved etch resistance. These results validate FPAA as a dual‐function material capable of functioning as both a high‐resolution imaging layer and a robust Si‐etch mask. This combination makes FPAA suitable for applications requiring high‐aspect‐ratio micro/nanostructures, such as silicon photonics and micro‐electro‐mechanical systems (MEMS).

3. Conclusion

We report a room‐temperature, scalable synthetic route to access a library of ligand‐tunable Sn‐oxo cluster resists and identify empirical relationships between ligand substituent structure and lithographic performance. By systematically modulating ligand substituent structures, we identified an empirical sensitivity‐ligand molar mass trendline in the non‐fluorinated homologues. FPAA deviates below this trendline and exhibits enhanced sensitivity, suggesting that fluorination may introduce an additional DEA‐related electron‐induced C─F activation pathway that promotes radical generation and network densification. FPAA achieved CD/LER of 9.1/2.2 nm under EBL, and 20.9 nm resolution under EUV exposure.

The resist further demonstrates high‐fidelity complex patterning capability, including functional QR codes, Chinese characters, and institutional emblems, validating its potential for optical security and information encoding. Moreover, FPAA exhibits exceptional plasma resistance, achieving a high estimated Si:resist etch selectivity of 51.2 in Bosch deep‐Si etching, enabling high‐fidelity pattern transfer with an ultrathin mask layer.

Mechanistic studies reveal a multi‐channel crosslinking network: (1) LESE‐driven decarboxylation and butyl scission, promoting crosslinking and densifying the Sn─O─Sn network; (2) DEA‐mediated C─F cleavage, followed by F⁻ recoordination to Sn, and coupling of aryl radicals; and (3) butyl radicals grafting at Sn sites to form Sn─butyl─Sn bridges. These findings provide design principles for next‐generation MOC resists that incorporate high‐EUV‐absorption cores (e.g., Sn, Zr, Hf) for enhanced photon utilization and use fluorinated ligand motifs capable of electron‐induced C─F activation to introduce additional exposure‐driven reaction pathways. Overall, this work positions ligand‐engineered MOCs as a versatile and scalable resist platform capable of simultaneously satisfying the resolution, LER, sensitivity, and etch resistance demands of sub‐10‐nm patterning.

Author Contributions

Wenzheng Li: methodology, investigation, writing – original draft, data curation, visualization, formal analysis. Min Zhang: methodology, investigation, visualization. Yingdong Zhao: investigation, methodology, formal analysis. Riyao Cong: methodology, investigation. Runfeng Xu: investigation. Saihe Yang: methodology. Xiaofeng Gong: methodology. Daohan Wang: investigation. Jiangli Fan: funding acquisition, project administration, resources. Jun Zhao: investigation, resources. Pengzhong Chen: writing – review and editing, supervision, funding acquisition, project administration, resources. Xiaojun Peng: supervision, project administration, writing – review and editing, funding acquisition, resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie73192‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article.