Deep Ion Etching: Principles and Applications

Updated 1 February 2026

Deep ion etching is a plasma etching process that produces high-aspect-ratio features with nearly vertical sidewalls using energy-assisted chemical and physical removal.
Techniques such as the Bosch process and cryogenic etching leverage cyclic passivation or low-temperature operation to optimize etch depth, anisotropy, and surface roughness.
This method is critical in MEMS, photonics, and advanced imaging, enabling reproducible fabrication with controlled etch parameters and integration with post-processing steps.

Deep ion etching encompasses a class of plasma etching processes capable of producing high-aspect-ratio features in semiconductor and dielectric substrates, typically with vertical or near-vertical sidewalls and micron- to submicron-scale critical dimensions. The method is foundational for advanced micro- and nanofabrication in fields such as MEMS, photonics, phase-contrast imaging, and photonic integrated circuits. The most widely implemented variants are deep reactive ion etching (DRIE) based on cyclic or continuous plasma chemistries for silicon, glass, and related materials, achieving aspect ratios exceeding 60:1 and etch depths from 1 µm to several hundred microns (Jefimovs et al., 2018, Gerlt et al., 2021, Lomonte et al., 2023, 0802.3085, Dirdal et al., 2020).

1. Physical Principles and Process Mechanisms

Deep ion etching exploits energetic ion–assisted plasma processes for directional material removal, relying on a combination of chemical and physical etching mechanisms. The prevailing approaches fall into three main categories:

Bosch process: Cyclic alternation between isotropic silicon etch (SF₆ plasma) and polymer passivation (C₄F₈), enabling nearly vertical etch profiles via differential bottom-sidewall polymer removal and anisotropic ion bombardment. This cycle creates scalloped sidewalls with periodic roughness features (Jefimovs et al., 2018, Gerlt et al., 2021, Dirdal et al., 2020).
Cryogenic etching: Continuous plasma operation at substrate temperatures near –100 °C using SF₆/O₂ chemistries, which form in-situ passivating SiOₓ or fluorocarbon films on sidewalls, suppressing lateral etching and yielding smooth, low-roughness profiles with vertical sidewalls (Lomonte et al., 2023).
Fluorocarbon/glass etching: Use of C₄F₈/CHF₃-based plasmas (sometimes with Ar addition or interleaved mechanical cleaning) for borosilicate and other dielectrics, with polymer passivation and process modulation to control sidewall angle and profile (0802.3085).

Process mechanics are governed by competition between polymer deposition (protection), physical sputtering (removal of passivation at feature bottoms), and chemical etch rates. Achievable aspect ratio, depth, and sidewall quality depend critically on tuning gas flows, plasma powers, cycle times, temperature, and mask selectivity.

2. Process Architectures and Workflow

The general deep ion etching workflow comprises the following steps, with tool-specific adaptations:

Masking: Photolithographically or nanoimprint-defined resist or hard mask (e.g., Cr, SiO₂, Si) is patterned on the substrate, often with features in the 1–20 µm regime or finer (Jefimovs et al., 2018, Dirdal et al., 2020, Lomonte et al., 2023).
Etching: The substrate is mounted on a temperature-controlled platen (static or electrostatic chuck; edge clamping or He backside cooling), exposed to pulsed (Bosch) or continuous (cryo) RF plasma in an ICP-DRIE reactor.
Passivation/etch cycling: In cyclic processes, consecutive passivation and etch steps are implemented, with timing, gas flows, and power precisely balanced. For smooth sidewalls, etch step duration must be minimized relative to passivation (Jefimovs et al., 2018, Gerlt et al., 2021).
Profile and sidewall control: Sidewall angle and roughness are managed through process parameter optimization and, in glass etching, through Ar addition or sequential ultrasonic cleaning to remove excess polymer (0802.3085).
Post-etch cleaning and back-end: For certain applications, wet etch (TMAH for residual Si), oxidation/strip (to reduce scallops), or metallization (Al back-reflector for photonics) are integrated (Lomonte et al., 2023, Dirdal et al., 2020).

3. Quantitative Performance Metrics

Key quantitative outcomes of deep ion etching, as reported in current literature, include:

Substrate	Max Depth (µm)	Max AR	Sidewall Angle	Scallop/Surface Roughness	Etch Rate (µm/min)	Reference
Si (Bosch)	30–42	60:1	~90° ±1°	tens of nm scallops (not quantified)	≈1 (est.)	(Jefimovs et al., 2018)
Si (3-step Bosch)	>450	>20:1	89.6° ±0.13°	80–150 nm (scallop)	(not given)	(Gerlt et al., 2021)
Borosilicate glass	300	~3:1	up to 85° ±1°	t_pol <50 nm (with cleaning)	0.9–1.0	(0802.3085)
Si (cryo DRIE)	≥140	0.9	89° ±1°	<10 nm rms	1.8–2.5	(Lomonte et al., 2023)
Si metasurface	1.2–1.6	2.3–5.7	~89–91°	14–86 nm scallops (cycle-dependent)	2–4 (typ.)	(Dirdal et al., 2020)

Process control enables depth uniformity across full wafer scales (±2–3%), precise sidewall verticality (within 1°), and aspect ratios up to 60:1 in sub-2 µm-width features (Jefimovs et al., 2018, Gerlt et al., 2021, Lomonte et al., 2023, 0802.3085, Dirdal et al., 2020). Selectivity to hard masks may exceed 350:1 in optimized three-step Bosch recipes (Gerlt et al., 2021).

4. Etch Profile Control and Optimization Strategies

Profile tailoring and roughness minimization are essential for device function and integration:

Scallop suppression (Bosch): Shortening the etch step within each cycle reduces vertical scallop depth but lowers net etch rate; optimization is application-driven. Scallops can be partially mitigated by post-oxidation/strip or compensated for in design (e.g., metasurface dimension scaling) (Jefimovs et al., 2018, Dirdal et al., 2020).
Etch lag and aspect-ratio effects: Etch rate decreases with increasing aspect ratio (ARDE/lag), empirically modeled as $R(AR) = R_0 / (1 + k \cdot (AR)^\gamma)$ . Optimized passivation/etch ratios can reduce lag to <1.5% at 50 µm depth across feature sizes (Gerlt et al., 2021).
Sidewall angle tuning: In glass, CF₄ polymer thickening is counteracted by Ar addition (raising Ar⁺ sputter yield, thinning passivation) or sequential ultrasonic DI water cleaning, realizing sidewall angles up to 85° independent of mask width (0802.3085).
Thermal and mask management: Consistent substrate cooling (platen/chuck at 15–20 °C for Bosch, –100 °C for cryo) is critical. Mask materials and thickness must be chosen to survive long etches; Cr, SiO₂, or highly crosslinked SU-8 are typical (Jefimovs et al., 2018, Gerlt et al., 2021, Lomonte et al., 2023).

5. Applications and Process Integration

Deep ion etching is central to the fabrication of:

MEMS and microfluidics: Enables microchannels, valves, pumps, and high-fidelity fluidic architectures in Si with uniform channel depths and stable high-AR walls, critical for batch-processing and pressure-tolerant microfluidics (Gerlt et al., 2021).
X-ray phase-contrast gratings: High-AR Si gratings (>30 µm deep, 1–2.4 µm pitch) support advanced imaging modalities, enabled by DRIE pattern transfer from displacement Talbot lithography (Jefimovs et al., 2018).
Photonics: Cryogenic DRIE is used for through-Si membrane formation, facilitating integration of reflective backplanes in grating couplers (e.g., Al mirrors for SiN waveguides with >97% directionality) (Lomonte et al., 2023).
Metasurface metalenses: UV nanoimprint lithography coupled with Bosch DRIE enables scalable fabrication of Si pillars for diffractive optics, with design adjustments allowing for compensation of etched sidewall scallops without deteriorating lens quality (Dirdal et al., 2020).

Process integration often requires tight alignment tolerances (sub-micron), membrane engineering, and sequential metallization for functional device assembly (Lomonte et al., 2023).

6. Process Challenges, Limitations, and Adaptation

Critical factors influencing deep ion etching performance include:

Temperature control and stress: Variations of ±5 °C at low substrate temperatures can induce sidewall taper or nonuniformity (especially in cryo-DRIE); robust thermal design is mandatory (Lomonte et al., 2023).
Mask erosion and selectivity: Mask durability sets maximum etch depth and feature fidelity. Thin resist masks may be consumed, demanding multi-stage or hard mask approaches (Gerlt et al., 2021, Dirdal et al., 2020).
Scallop roughness and compensation: Bosch process sidewall roughness is intrinsic; post-processing or intelligent design compensation (e.g., upscaling pillar dimensions) restore device function (Dirdal et al., 2020). In glass, thick polymer films may cause profile taper, mitigated by Ar addition or cleaning (0802.3085).
Etch lag/loading effects: Uniformity across diverse feature sizes requires process adaptation—optimized gas flow switching, adjusted cycle times, or tool zone control (Gerlt et al., 2021, Dirdal et al., 2020).
Membrane stability and integration: Large suspended membranes risk mechanical failure unless width and thickness are balanced; trade-off exists between optical access and mechanical robustness (Lomonte et al., 2023).

Adaptation to new materials or applications (e.g., LiNbO₃, AlN) can leverage the same plasma etch infrastructure, with chemistry and mask adjustments as required (Lomonte et al., 2023).

7. Analytical Models and Empirical Relations

While no original analytical models are introduced in the cited literature, standard empirical and phenomenological relations appear, such as:

Etch rate averaging (Bosch):

$R_\mathrm{etch} = (\Gamma_\mathrm{SF6} \cdot \sigma_\mathrm{etch}) \cdot t_\mathrm{etch} / (t_\mathrm{etch} + t_\mathrm{pass})$

Aspect-ratio dependent rate fall-off:

$R_\mathrm{etch}(AR) = R_0 \cdot \exp(-\alpha \cdot AR)$

Etch lag quantification:

$\mathrm{lag} = \frac{d(w_1) - d(w_2)}{d(w_1)} \times 100\%$

Sidewall angle (glass):

$\theta \approx 90^\circ - \arctan \left( \frac{R_\mathrm{lateral}}{R_\mathrm{vertical}} \right)$

with $R_\mathrm{lateral}$ dropping as sidewall passivation is reduced.

These models, along with fitting constants and in-situ metrology, guide recipe optimization and device yield improvement (Gerlt et al., 2021, Jefimovs et al., 2018, 0802.3085).

The current state of deep ion etching enables reproducible fabrication of high-AR features across silicon and dielectric substrates with controllable sidewall profiles, moderate to high throughput, and feature sizes suitable for leading-edge MEMS, optics, and photonics. Ongoing challenges center on etch uniformity, sidewall quality, and process transferability across diverse toolsets and material systems (Jefimovs et al., 2018, Gerlt et al., 2021, Lomonte et al., 2023, Dirdal et al., 2020, 0802.3085).