Decoding Directional Control in Metal‐Assisted Chemical Etching via Catalyst Architecture

Abstract

Metal‐assisted chemical etching (MaCE) has emerged as a promising technique for fabricating silicon nanostructures, yet the presence of anomalous isotropic etching poses significant challenges for precise dimensional control. Here, it is demonstrated that catalyst morphology, particularly its aspect ratio, plays a crucial role in determining etching directionality. Through systematic investigation of the initial stages of MaCE, it is revealed that significant undercutting occurs within seconds of etching initiation, persisting across all solution compositions. This phenomenon is quantitatively analyzed using the Degree of Undercutting (DoU) and Degree of Anisotropy (DoA) metrics, establishing that conventional solution chemistry control alone cannot suppress lateral etching. These findings reveal that high‐aspect‐ratio dendrite catalysts, formed at elevated AgNO₃ concentrations, undergo physical separation during etching, leading to residual catalysts that promote localized isotropic etching. To address this, a thermal treatment approach is developed that effectively transforms these problematic structures into stable, low‐aspect‐ratio catalysts. A critical transition at 450 °C, where enhanced silver atom mobility coincides with surface defect formation, enables nearly perfect vertical etching. This work not only provides fundamental insights into the relationship between catalyst geometry and etching behavior but also presents a practical solution for achieving precise control over silicon nanostructure fabrication.

Metal‐assisted chemical etching (MaCE) enables silicon nanostructure fabrication but suffers from isotropic undercutting. This study highlights the critical role of catalyst morphology in etching directionality. High‐aspect‐ratio catalysts induce lateral etching, while thermal treatment at 450 °C stabilizes catalyst geometry, promoting vertical etching. These findings offer a strategy for precise silicon nanostructure control, advancing semiconductor and nanofabrication applications.

1. Introduction

Metal‐assisted Chemical Etching (MaCE) has emerged as a powerful technique for silicon nanostructure fabrication, offering unique advantages in the ongoing pursuit of device miniaturization.^{[ 1} , ^{2 ]} The technique has gained significant attention due to its ability to create high‐aspect‐ratio structures with controlled morphologies at relatively low cost, making it particularly attractive for applications in energy harvesting, sensing, and optoelectronics.^{[ 3} , ⁴ , ⁵ , ⁶ , ^{7 ]} However, anomalous isotropic etching behavior during the initial phases of MaCE poses significant challenges for precise nanostructure control, limiting its widespread adoption in industrial applications. Our systematic investigations have revealed a previously overlooked phenomenon: significant undercutting occurs during the early stages of the MaCE process, even within the first few seconds of etching. This isotropic etching behavior, which coexists with the controlled anisotropic etching, leads to structural instability and poor reproducibility in the resulting nanostructures. The presence of undercutting is particularly problematic for applications requiring precise dimensional control, such as in photonic devices or semiconductor manufacturing, where even minor deviations from designed specifications can significantly impact device performance.

The fundamental principle of MaCE involves the localized oxidation and dissolution of silicon in the presence of noble metal catalysts and an etching solution typically composed of hydrofluoric acid (HF) and hydrogen peroxide (H₂O₂). Early models proposed by Li and Bohn established that the metal catalyst acts as a local cathode, facilitating the reduction of the oxidizing agent and injecting holes into the silicon substrate as shown in Figure 1 .^{[ 8 ]} These holes promote the oxidation and subsequent dissolution of silicon in the presence of HF, leading to the formation of various nanostructures including nanowires, nanopores, and complex 3D architectures.^{[ 9} , ^{10 ]} However, several observations have emerged that challenge this conventional understanding. The formation of porous silicon layers beneath the metal catalyst and etching in regions not in direct contact with the catalyst cannot be fully explained by the traditional hole injection model.^{[ 11} , ^{12 ]} More importantly, the significant undercutting we observe during the initial etching stages suggests that the current models are incomplete. Recent studies have attempted to address these discrepancies through various approaches.^{[ 13} , ^{14 ]} For example, Lai et al. proposed that Schottky junctions formed between metal catalysts and silicon play a critical role in controlling hole distribution during MaCE. Their work demonstrated that n‐type silicon confines holes near the surface due to band bending effects, promoting vertical etching, while p‐type silicon experiences hole drift away from the interface, resulting in tapering and lateral etching.^{[ 13 ]} Similarly, Borgohain and Dash investigated post‐deposition thermal annealing of gold catalysts to eliminate pinholes in catalyst layers and improve mass transport during MaCE. They showed that annealed thick gold microstripes enhance uniform etching by preventing bending and ensuring consistent contact with the silicon substrate.^{[ 14 ]} However, these studies are limited in their ability to explain the significant undercutting that occurs during the early stages of the MaCE process, even within the first few seconds of etching. For instance, Huang et al. proposed that the diffusion of reactive species along the silicon‐metal interface could contribute to observed etching patterns.^{[ 15 ]} Their model explains how mass transport mechanisms influence microporous silicon formation but does not fully capture the role of catalyst morphology in directional control or isotropic suppression. Despite these efforts, a comprehensive model that accounts for both the vertical and lateral etching components remains elusive.

Figure 1.

Schematic illustration of the metal‐assisted chemical etching mechanism showing the progression of hole injection and etching processes. Left: Initial stage showing oxidation‐reduction reactions at the Ag catalyst surface, where H₂O₂ reduction generates holes that are injected into the silicon substrate. Middle: Penetration of the Ag catalyst into silicon substrate through localized electrochemical reactions. Right: Formation of extended depletion regions through hole diffusion, leading to both vertical and lateral etching processes.

Our preliminary investigations indicate that catalyst morphology plays a more crucial role in determining etching behavior than previously recognized. We have observed that the initial shape and distribution of the metal catalyst significantly influence the degree of undercutting and subsequent etching directionality. This observation suggests that the physical characteristics of the catalyst, rather than just its electrochemical properties, may be key to controlling the etching process. In particular, we find that the aspect ratio of the catalyst structures, which can be controlled through deposition conditions, strongly correlates with the tendency for isotropic etching. This insight has led us to propose a catalyst‐centric approach to understanding and controlling the MaCE process. Compared to previous studies that primarily focused on solution chemistry and electrochemical reactions, we investigate how catalyst morphology evolves during etching and how this evolution affects the etching behavior. Through careful control of silver nitrate concentration during catalyst deposition and subsequent thermal treatment, we demonstrate that catalyst geometry can be engineered to minimize undesired isotropic etching while maintaining the desired vertical etch characteristics. By analyzing the relationship between catalyst morphology and etching behavior, particularly the effects of thermal treatment on catalyst stability, we provide new insights into the mechanisms governing the MaCE process. Our findings not only advance the fundamental understanding of metal‐assisted etching but also provide practical solutions for improving process control in nanostructure fabrication. This work has significant implications for the broader field of nanofabrication, potentially enabling more precise and reliable production of silicon nanostructures for various technological applications.

2. Results and Discussion

2.1. Anomalous Isotropic Etching Behavior in Initial Stages of MaCE

The precise fabrication of silicon nanostructures through MaCE has gained significant attention due to its capability to produce high‐aspect‐ratio nanowires. While conventionally understood as an anisotropic process,^{[ 10} , ^{11 ]} our investigations reveal a more complex etching behavior that challenges this view. Figure  2 presents a comprehensive study of the initial stages of the MaCE process, uncovering anomalous etching phenomena that significantly impact the resulting nanostructures. As shown in Figure 2a, 20‐min MaCE resulted in unexpected photoresist (PR) delamination and significant roughening of the silicon surface beneath the PR mask, although this duration is typically used to achieve 10–20 µm long nanowires.^{[ 16 ]} The Scanning Electron Microscope (SEM) image clearly shows areas where the PR has been completely stripped. Our control experiment, where PR‐coated silicon was exposed to HF/H₂O₂ solutions without Ag catalysts, showed no PR delamination or silicon etching, confirming that the observed delamination is specifically due to MaCE‐induced undercutting facilitated by the Ag catalysts (see Figure S1, Supporting Information). Given that our etchant solution does not contain any components capable of dissolving the PR, and no physical processes were involved that could cause PR delamination, we conclude that the silicon surface beneath the PR was etched, leading to the PR's detachment. This unexpected result prompted us to investigate shorter etching times to capture the initial stages of the process more accurately. As shown in Figure 2b, a high‐magnification SEM image of 20‐min MaCE shows a striking feature: significant undercutting (red‐colored) beneath the PR mask coexisting with the expected anisotropic etching below (blue‐colored). While the PR remains intact on the surface, we observe clear undercutting beneath it, indicating isotropic etching. To understand the factors influencing this anomalous isotropic etching, we systematically varied the HF and H₂O₂ concentrations in the etchant solution. The etching behavior in MaCE is primarily governed by the ratio of H₂O₂ to HF concentrations (ρ = [H₂O₂]/[HF]+[H₂O₂]). As shown in Figure 2c and Figure S2 (Supporting Information), the etch rate increases with increasing ρ values, consistent with previous reports.^{[ 17 ]} This relationship can be attributed to enhanced hole injection from H₂O₂ reduction at the metal catalyst surface, which accelerates the oxidation and subsequent dissolution of silicon by HF.^{[ 18} , ^{19 ]} To quantitatively analyze the lateral etching behavior, we define the degree of undercutting (DoU) as:

$$\mathrm{DoU}=\left(d_f-d\right)/2$$ (1)

where d_f is the final width after etching and d is the initial pattern width (see Figure S3, Supporting Information). Figure 2d presents DoU measurements across various HF and H₂O₂ concentrations. Notably, significant undercutting (DoU values between 0.2 and 0.45) was observed across all tested solution compositions, showing no clear correlation with the HF and H₂O₂ concentrations. This observation contradicts conventional understanding, where solution chemistry, particularly the balance between hole injection and diffusion rates, was thought to be the primary factor controlling etching isotropy.^{[ 9} , ^{20 ]} It is important to distinguish the isotropic undercutting observed in our study from previously reported directional etching phenomena. Prior work by Huang et al. demonstrated that oxidant concentration can influence etching direction between crystallographically‐preferred <100> directions (at low H₂O₂ concentrations) and surface‐normal directions (at high H₂O₂ concentrations) through different dissolution mechanisms.^{[ 21 ]} Additionally, they showed that catalyst connectivity can affect directional preference, with interconnected catalysts favoring vertical etching.^{[ 22 ]} However, the undercutting we observe occurs on (100) substrates that should inherently favor vertical etching, and persists regardless of solution chemistry adjustments. This suggests a different mechanism is at play, specifically, the physical separation of high‐aspect‐ratio catalysts during etching that leads to simultaneous isotropic etching near the surface, independent of crystallographic preferences. To investigate the temporal evolution of undercutting, we conducted time‐dependent studies focusing on the initial stages of MaCE. As shown in Figure 2e, the DoU characterizations reveal significant isotropic etching behavior even within the first few seconds of the process, with the value rapidly increasing up to 30 s before reaching a plateau. Remarkably, a measurable DoU was observed at the shortest measured time of 10 s, indicating that isotropic etching initiates almost immediately upon contact between the catalyst and the etchant solution (see cross‐sectional SEM images in Figure S4, Supporting Information). Our time‐resolved SEM analysis reveals that directional etching initiates within 5–10 s of immersion in the etching solution. During the first 5–10 s, the catalyst structure becomes fully embedded in the silicon. Beyond 15 s, both lateral and vertical etching processes are clearly observable, with lateral etching reaching its maximum extent ≈30 s, after which the DoU value plateaus as shown in Figure 2e. This rapid onset and subsequent saturation of DoU suggests that the competition between isotropic and anisotropic etching mechanisms exists from the very beginning of the MaCE process. The presence of a plateau in the DoU versus time plot indicates that while isotropic etching is inherent to the initial stages of MaCE, its relative contribution to the overall etching behavior stabilizes over time. This observation challenges the conventional understanding of MaCE as a purely anisotropic process and suggests the existence of fundamental mechanisms promoting isotropic etching that cannot be explained by the traditional hole injection model alone. These findings raised important questions about the relationship between catalyst morphology and etching behavior, leading us to investigate the role of catalyst structure in determining etching characteristics.

Figure 2.

Analysis of isotropic etching behavior during MaCE process. a) Cross‐sectional SEM image of silicon nanostructures after 20‐min MaCE showing PR delamination region. b) High‐magnification image of the selected area in (a), highlighting distinct isotropic and anisotropic etching regions. c) Correlation between etching rate and ρ ([H₂O₂]/[HF]+[H₂O₂]) showing increased etching rate at higher ρ values. d) DoU measurements at various HF concentrations with different H₂O₂ concentrations (0.10 , 0.19 , and 0.40 m). e) Time‐dependent evolution of DoU showing a rapid onset of undercutting within initial 30 s followed by saturation. Scale bars: 3 µm (a) and 1 µm (b).

2.2. Catalyst Morphology‐Dependent Etching Behavior in MaCE

Our observation of persistent isotropic etching, regardless of solution chemistry, led us to investigate the physical aspects of the catalyst behavior during MaCE. Figure 3a schematically illustrates our proposed mechanism for the simultaneous occurrence of vertical and lateral etching. During the MaCE process, catalysts with high aspect ratios experience physical separation: while the main portion of the catalyst proceeds with vertical etching forming nanowires, some parts remain near the surface due to their elongated structure. These residual catalysts continue to promote localized isotropic etching, resulting in the observed undercutting phenomenon. To validate this mechanism, we first needed to rule out alternative explanations. One possibility was that Ag catalysts might dissolve in the HF/H₂O₂ solution and redeposit on the silicon surface, creating new etching sites. We tested this by transferring Ag catalysts to HF‐resistant polyimide tape and exposing them to the etchant solution. SEM images and diameter analysis of the Ag catalysts showed no significant morphological changes in Figure S5 (Supporting Information), indicating that catalyst dissolution is not a major factor during our short etching timeframe.

Figure 3.

Effect of AgNO₃ concentration on catalyst morphology and physical separation mechanism during MaCE. a) Schematic illustration of the proposed physical separation mechanism: high‐aspect‐ratio Ag catalysts experience fragmentation during etching, leading to residual catalysts that promote localized isotropic etching. b–e) Plan‐view SEM images showing the evolution of Ag catalyst morphology with increasing AgNO₃ concentration from uniformly dispersed nanoparticles to complex dendrite structures. f) Quantitative analysis of Ag coverage area as a function of AgNO₃ concentration. g–j) Cross‐sectional SEM images reveal the dramatic increase in catalyst height with increasing AgNO₃ concentration. k) Box plots showing the height distribution of Ag catalysts with various AgNO₃ concentrations. Scale bars: 1 µm (b–e) and 500 nm (g–j).

Based on our observations, we hypothesized that catalyst morphology, particularly its aspect ratio, determines the propensity of physical separation during etching, thereby controlling the extent of undercutting. To test this hypothesis, we systematically investigated how initial catalyst formation conditions affect their structure. Figure 3b–k shows the evolution of catalyst coverage as a function of AgNO₃ concentration during deposition. The accompanying SEM images (Figure 3b–e) reveal a dramatic morphological transition: at low concentrations (0.001 m), we observe uniformly dispersed nanoparticles, but as concentration increases, the catalysts develop increasingly complex dendrite structures. The critical evidence supporting our mechanism comes from the height measurements shown in Figure 3k. Cross‐sectional SEM images (Figure 3g–j) demonstrate that catalysts formed at higher concentrations (>0.01 m) develop substantial vertical dimensions, with heights increasing from ≈15 nm at 0.001 m to over 100 nm at 0.02 m. This increase in aspect ratio directly correlates with our observations of enhanced isotropic etching. These high‐aspect‐ratio catalysts, particularly those formed at higher concentrations, are physically constrained from penetrating through the narrow gaps (tens of nm) between forming silicon nanowires. Instead, they remain near the surface where they continue to catalyze silicon etching, leading to the observed isotropic behavior. This explains why both anisotropic and isotropic etching occur simultaneously: shorter catalysts can penetrate and promote conventional vertical etching, while elongated structures remain near the surface, causing localized undercutting. While previous studies^{[ 23} , ^{24 ]} have examined the relationship between catalyst morphology and etching patterns, primarily focusing on the distinction between random porous structures formed by isolated particles and vertical etching promoted by networked catalysts, our work specifically addresses the isotropic undercutting phenomenon that occurs during the initial stages of MaCE. This undercutting, which we quantify through the DoU metric, persists regardless of solution chemistry adjustments and represents a fundamental challenge for precise dimensional control. Our findings reveal that catalyst aspect ratio, rather than merely network connectivity, is the critical factor determining the extent of undercutting. The physical separation mechanism we propose, wherein high‐aspect‐ratio catalysts fragment during etching, provides a new perspective on how catalyst geometry influences etching directionality beyond the established understanding of network effects. This discovery provides a mechanistic link between initial catalyst formation conditions and the resulting etching behavior. To fully understand the formation mechanism of these different catalyst morphologies and their impact on etching behavior, we need to examine the catalyst growth process in detail.

Figure 4a,b schematically illustrates how AgNO₃ concentration influences catalyst morphology and subsequent etching patterns. At low AgNO₃ concentrations, Ag ions react with numerous nucleation sites on the silicon surface, resulting in the uniform deposition of high‐density, small‐diameter nanoparticles (Figure 4a; Figure S6a–c, Supporting Information). These small, isolated Ag particles possess high surface energy due to large specific surface area, which can lead to irregular etching patterns.^{[ 25 ]} Consequently, this morphology tends to reduce directional etching, resulting in the formation of random porous silicon structures with near‐isotropic etching characteristics (Figure S6d–f, Supporting Information).^{[ 26 ]} In contrast, at higher AgNO₃ concentrations (Figure 4b), we observe a significant shift in catalyst morphology. Ag particles initially form on defect sites or other nucleation points on the silicon surface. Subsequent Ag ion reduction preferentially occurs on these existing silver structures due to their lower energy state, promoting the formation of dendrite‐like structures instead of adhering to the silicon interface. This process involves both vertical growth, forming dendrites, and lateral growth, connecting adjacent silver particles into a networked structure (Figure S7, Supporting Information). Initially, this interconnected network of silver catalysts behaves differently from the isolated particles formed at lower concentrations. Rather than exhibiting random etching patterns due to high surface energy, these networked catalyst structures tend to etch along the (100) direction, which requires the least energy to break silicon back bonds. This directional preference results in the formation of vertical nanowires. However, the high aspect ratio of these dendrite‐like structures introduces a new challenge such as mechanical instability. Due to the high‐aspect ratio of dendrite‐structure, the catalysts formed at the high AgNO₃ concentration are more prone to fragmentation and detachment as shown in Figure S8 (Supporting Information). This can lead to catalyst separation, particularly at the upper portions of the structure, potentially causing isotropic etching at the top of the silicon surface.

Figure 4.

Schematic illustration of AgNO₃ concentration‐dependent catalyst formation and subsequent etching behavior. Comparative mechanism at a) low and b) high AgNO₃ concentrations. c) Proposed solution through thermal treatment: annealing transforms high‐aspect‐ratio dendrite structures into low‐aspect‐ratio catalysts with improved stability, enabling more controlled etching behavior. Scale bars in SEM images: 1 µm.

To address these issues and enhance vertical etching while preventing catalyst detachment, we propose that improving the adhesion between the catalyst and silicon is crucial. This could be achieved by increasing the contact area between the catalyst and silicon, essentially aiming for a catalyst morphology with a lower aspect ratio (Figure 4c). However, our findings suggest that merely adjusting the wet etching process parameters is insufficient to achieve this ideal catalyst morphology – one that maintains a low aspect ratio without inducing random etching behavior. Conventional wet deposition processes tend to favor the formation of high aspect ratio catalysts, resulting in the creation of dendrite‐shaped catalyst. Significantly reducing the aspect ratio would require additional processing steps beyond simply altering solution concentrations. Therefore, our study reveals that eliminating the isotropic etching at the top of silicon nanostructures necessitates a novel approach to catalyst design and deposition. To address this challenge, we explored post‐deposition treatments, specifically thermal annealing, as a method to modify catalyst morphology and improve its stability during the etching process. This approach allows us to manipulate the catalyst structure after deposition, potentially overcoming the limitations of wet deposition methods.

2.3. Enhancement of Etching Anisotropy through Thermal Control of Catalyst Morphology

To overcome the limitations of wet deposition in controlling catalyst morphology, we explored post‐deposition thermal treatment as an alternative approach. Figure 5 demonstrates how thermal treatment effectively modifies both catalyst structure and subsequent etching behavior. The annealing temperature was carefully selected to be below both the melting point of silver (960 °C) and the silver‐silicon mixing temperature (830 °C), allowing for controlled morphological changes while maintaining structural integrity.^{[ 27 ]} Based on cross‐sectional SEM images (Figure 5a–c), we observed distinct temperature‐dependent etching behaviors. While the DoU provides information about the lateral etching extent, it does not fully capture the directional nature of the etching process. Therefore, we introduce the Degree of Anisotropy (DoA) to better quantify the overall etching directionality:

$$\mathrm{DoA}=1-\mathrm{DoU}/\mathrm{h}$$ (2)

where h is the etch depth. This metric provides a comprehensive measure of etching behavior, with DoA = 0 indicating purely isotropic etching and values approaching 1 representing highly anisotropic etching. Figure 5d reveals three distinct regions of etching behavior. Region 1 shows the results using as‐deposited catalysts, where the presence of the PR mask helps maintain relatively high DoA values despite some undercutting. In Region 2 (200–400 °C), we observe a decrease in DoA compared to Region 1. This deterioration occurs because thermal treatment at these temperatures requires PR removal, yet the temperature is insufficient to effectively modify catalyst morphology. Without the PR mask's protective effect and lacking significant catalyst restructuring, the etching becomes less controlled. However, a dramatic improvement appears in Region 3 (>400 °C), where we achieve nearly perfect vertical etching with minimal undercutting, resulting in the highest DoA values. This enhancement coincides with significant changes in catalyst morphology. With corresponding cross‐sectional SEM images (Figure 5e–h) showing the evolution of catalyst morphology, Figure 5i plots both the catalyst aspect ratio and DoU as a function of annealing temperature. The data reveals two distinct regimes: the as‐deposited catalyst region characterized by high aspect ratio structures and significant undercutting, and the annealed catalyst region exhibiting low aspect ratio morphology with suppressed lateral etching. Notably, the trends in catalyst aspect ratio and DoU show remarkable correlation, with both metrics decreasing significantly as temperature increases. A critical transition occurs ≈450 °C, where we observe a substantial decrease in both DoU and catalyst aspect ratio. This transformation can be explained by temperature‐dependent silver‐silicon interactions. To investigate whether crystallographic changes contribute to the observed etching behavior, we conducted grazing incidence X‐ray diffraction (GIXRD) analysis of silver catalysts at various annealing temperatures (see Figure S9, Supporting Information). Interestingly, the as‐deposited catalysts already exhibit a dominant (111) orientation, with minor peaks corresponding to (200), (220), and (311) planes. This crystallographic orientation persists across all thermal treatments, with only a slight enhancement of the (111) peak intensity at higher temperatures. This finding indicates that the electroless deposition process naturally produces silver catalysts with thermodynamically favorable crystal orientations, and the dramatic improvement in etching anisotropy observed at 450 °C is primarily attributed to the geometric transformation of catalysts rather than significant changes in their crystallographic orientation. The hemispherical shapes formed at this temperature better accommodate the lattice mismatch between silver and silicon by maximizing the contact area while maintaining stable crystallographic interfaces. Below 400 °C, insufficient thermal energy limits silver atom surface diffusion, maintaining the original high‐aspect‐ratio structures of the as‐deposited catalysts. However, as the temperature approaches 450 °C, enhanced silver atom mobility following the Arrhenius relationship enables catalyst restructuring. The increased thermal energy allows Ag atoms to overcome the surface diffusion activation barrier, transforming dendrite‐like structures into more stable hemispherical shapes that maximize contact with the silicon surface. Notably, this morphological evolution coincides with the onset of silicon surface defect formation, reported to occur at temperatures exceeding 450 °C. These thermally induced defects serve as anchor points for the thermally modified, low‐aspect‐ratio catalysts, promoting stronger catalyst‐substrate interaction. The improved contact and stability of the annealed catalyst particles effectively suppress undercutting, as evidenced by the decreased DoU values at higher temperatures. This correlation between catalyst aspect ratio and DoU provides strong evidence that catalyst morphology directly influences the extent of lateral etching during the MaCE process. The improvement in etching behavior can be attributed to three key factors: 1) the transformation of catalysts into hemispherical shapes that better accommodate the lattice mismatch (≈25%) between silver and silicon,^{[ 28 ]} 2) the formation of surface defects that provide stable anchor points, and 3) the increased catalyst‐substrate contact area that promotes consistent vertical etching. These findings establish thermal treatment as an effective method for controlling the MaCE process. By optimizing the annealing temperature to achieve both favorable catalyst morphology and strong substrate interaction, we can significantly enhance etching directionality while maintaining efficient vertical etching capabilities.

Figure 5.

Effect of thermal treatment on etching directionality and catalyst morphology. a–c) Cross‐sectional SEM images showing distinct etching behaviors at different thermal treatment conditions. d) Temperature‐dependent DoA evolution for different Ag deposition times (30, 60, 90s). e–h) Cross‐sectional SEM images showing the evolution of catalyst morphology with increasing annealing temperature. i) Correlation between catalyst aspect ratio and DoU as a function of annealing temperature. Scale bars: 1 µm (a–c,f,h), 400 nm (c,g).

To further validate our findings regarding the importance of catalyst morphology, we conducted comparative experiments using vacuum‐deposited silver catalysts of varying thicknesses (10, 20, and 30 nm) as shown in Figure S10 (Supporting Information). These vacuum‐deposited catalysts consistently exhibited reduced undercutting compared to wet‐deposited counterparts, forming relatively flat films with lower aspect ratios. Interestingly, we observed that the pattern edges of vacuum‐deposited samples still showed some isotropic etching due to the formation of isolated nanoparticles at these regions, particularly in thinner films (10 nm). This effect gradually diminished with increasing film thickness (30 nm), where more continuous coverage was achieved. These results provide additional evidence for our proposed mechanism regarding the importance of both catalyst networking and aspect ratio in controlling etching directionality. While vacuum deposition demonstrates superior control over catalyst morphology, our thermal treatment approach offers a more cost‐effective solution that maintains the simplicity advantage of wet processing while addressing its morphological limitations.

3. Conclusion

In this study, we have investigated and addressed a fundamental challenge in MaCE of silicon: the persistent occurrence of isotropic etching during the initial phases of the process. Through systematic investigation, we demonstrated that this phenomenon is inherent to the MaCE process, occurring within seconds of initiation and persisting regardless of solution chemistry adjustments. Our findings reveal that catalyst morphology, particularly the aspect ratio, plays a crucial role in determining etching behavior. The key discovery lies in establishing the relationship between catalyst geometry and etching directionality. High‐aspect‐ratio dendrite structures, formed at higher AgNO₃ concentrations, were found to promote isotropic etching due to their mechanical instability and tendency to separate during the etching process. This understanding led us to develop a thermal treatment approach that effectively transforms these problematic structures into more stable, hemispherical catalysts. The critical temperature of 450 °C was identified as a turning point where enhanced silver atom mobility and surface defect formation combine to create optimal conditions for directional etching.

By quantitatively analyzing the DoU and DoA, we demonstrated that thermal treatment significantly improves etching directionality through three mechanisms: reshaping catalysts to better accommodate lattice mismatch, creating beneficial surface defects that anchor catalysts, and increasing the catalyst‐substrate contact area. This approach successfully suppresses undesired lateral etching while maintaining efficient vertical etching capabilities. Our work not only provides a practical solution for controlling MaCE processes but also advances the fundamental understanding of catalyst‐substrate interactions in wet chemical etching. While vacuum deposition techniques can also achieve controlled catalyst morphologies with reduced undercutting, our thermal treatment approach maintains the cost‐effectiveness and simplicity that make MaCE attractive for industrial applications. Future research directions could explore alternative methods to achieve similar morphological control without thermal treatment, such as plasmonic heating using light sources or refined wet deposition conditions. These insights open new possibilities for precise nanostructure fabrication, particularly for applications requiring high‐aspect‐ratio features with minimal lateral deviation. Looking forward, this approach could be extended to other metal catalyst systems and may find applications in various fields requiring precise control of nanostructure dimensions.

4. Experimental Section

Materials and Sample Preparation

N‐type (100) silicon wafers (1–10 Ω·cm, 525 µm thickness, Czochralski grown) were used as substrates. Prior to processing, the wafers underwent sequential ultrasonic cleaning in acetone, isopropyl alcohol (10 min each), followed by thorough rinsing in deionized water (18.2 MΩ·cm) and drying under nitrogen flow. This study focused specifically on lightly doped n‐type (100) silicon wafers to isolate the effects of catalyst morphology on etching behavior. Future investigations examining how the observed phenomena vary with doping level, wafer orientation, and alternative etching solution compositions would provide valuable complementary insights.

Photolithographic Patterning

Photolithography was employed to define catalyst deposition regions and provide protective masking during the etching process. The wafers were dehydrated at 150 °C for 10 min, followed by spin‐coating of positive photoresist (DONGJIN SEMICHEM., DPR‐i1549) at 4000 rpm for 30 s. Soft baking was performed at 100 °C for 60 s on a hotplate. The resist‐coated wafers were exposed using a mask aligner (Prowin Co., Ltd., M‐100) under vacuum contact mode with an exposure dose of 140 mJ cm^{− 2}. The pattern consisted of circular hole arrays (2 µm diameter, 3 µm pitch) as shown in Figure S11 (Supporting Information). Development was carried out in AZ 300 MIF developer (EMD Performance Materials) for 120 s, followed by rinsing with deionized water and nitrogen drying. Hard baking was performed at 110 °C for 90 s to improve the resist's chemical stability during subsequent wet processing.

Metal‐Assisted Chemical Etching

The MaCE process consisted of two primary steps: catalyst deposition and silicon etching. For Ag catalyst deposition, the patterned wafers were immersed in an aqueous solution containing AgNO₃ (425 mg, DEAJUNG) and HF (49 wt.%, 416.5 mL, SAMCHUN) diluted with deionized water (83.5 mL) at room temperature. The deposition time was 30 s for the experiments shown in Figure 3, 90 s for Figure 4a, and 30 s for Figure 4b. For the thermal treatment experiments presented in Figure 5, varying deposition times of 30, 60, and 90 s were used with a consistent etching time of 3 min. Following deposition, samples were thoroughly rinsed with deionized water and dried under nitrogen flow. For comparative studies, silver catalysts were also prepared using vacuum deposition. Silver films of various thicknesses (10, 20, and 30 nm) were deposited on patterned silicon substrates using thermal evaporation at a base pressure of 3 × 10⁻⁶ Torr and deposition rate of 1 Å s⁻¹. The samples were then subjected to the same etching conditions as the wet‐deposited catalysts to enable direct comparison of etching behaviors. Selected samples underwent thermal annealing in a tube furnace (T‐ENG) under a nitrogen atmosphere (99.99% purity) with a flow rate of 0.5 L min⁻¹. The temperature was ramped at 20 °C min⁻¹ to target temperatures (250, 300, 350, 450, and 550 °C) and maintained for 10 min. After cooling to room temperature under continued nitrogen flow, samples were rinsed with deionized water and dried. The etching process was conducted in solutions containing various concentrations of HF (49 wt.%) and H₂O₂ (30 wt.%) at room temperature. The ρ value ([H₂O₂]/[HF]+[H₂O₂]) was systematically varied to investigate its effect on etching behavior. For the experiments shown in Figure 2a,b, etching was conducted for 20 min, while Figure 2c,d experiments used a consistent etching time of 3 min. To investigate the initial etching behavior and temporal evolution of undercutting shown in Figure 2e, shorter durations (5, 10, 20, 25, 60, and 90s) were tested. Figure 3 presents catalyst deposition results without etching, while the etching experiments in Figures 4 and 5 were conducted for 3 min. Post‐etching, samples were thoroughly rinsed with deionized water and dried under nitrogen flow.

Morphological Characterization

The morphology of the Ag catalysts and etched silicon structures was characterized using field‐emission scanning electron microscopy (FE‐SEM, Hitachi S‐4800) operated at 15 kV. Cross‐sectional samples were prepared by cleaving along the crystal plane. The dimensions of the nanostructures and degree of undercutting were analyzed using Image J software.

Crystallographic Characterization

The crystallographic properties of silver catalysts at various annealing temperatures (as‐deposited, 250, and 450 °C) were analyzed using grazing incidence X‐ray diffraction (GIXRD) with an incident angle of 0.5° to optimize surface sensitivity for the thin silver layers. The measurements were performed using SmartLab, RIGAKU with Cu Kα radiation (λ = 1.5406 Å) operating at 40 kV and 40 mA. The diffraction patterns were collected in the 2θ range of 20–80° with a step size of 1° and scan speed of 1°/min.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Han Y., Jeong J., Cho H., Ahn Y., Park S., Kim H., Shin J. Y., Park M.‐J., Kim T., Um H.‐D., Decoding Directional Control in Metal‐Assisted Chemical Etching via Catalyst Architecture. Adv. Mater. 2025, 37, 2502840. 10.1002/adma.202502840

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.