Semiconductor Nanofabrication Sub-2nm Node Ready

Atomic Layer Etching (ALE) & Quasichemical Nanopatterning Lab

Model self-limiting directional plasma and thermal isotropic ALE, Langmuir halogen chemisorption kinetics, low-energy ion-assisted desorption energy windows ($\Delta E = E_{\text{th,phys}} - E_{\text{ion}}$), Aspect Ratio Independent Etching (ARIE), and 20-year semiconductor fab cleanroom DCF economics.

Etch-Per-Cycle (EPC)
1.28 Å
Self-limiting atomic removal
ALE Synergy ($S_{\text{ALE}}$)
96.4%
Ideal self-limitation index
Energy Window Margin
+24.5 eV
Buffer below physical sputtering
Surface Roughness ($R_q$)
0.082 nm
Atomically pristine planarization
Cycle Time ($t_{\text{cycle}}$) 9.7 s
Total Etch Depth (30 Cycles) 3.84 nm
ARIE Uniformity Index 98.2%
Lattice Defect Depth 0.7 Å
ALE Cyclic Process Parameters
1.2 s
0.1 s (Under-saturated) 5.0 s (Full saturation)
2.0 s
38 eV
0 eV (Thermal) Eth,chem (18 eV) Eth,phys (62 eV)
4.5 s
45 °C
18:1
Ion Energy Window & Physical Sputter Thresholds
Process Window
Langmuir Step A Chemisorption Saturation
Nanostructure Cross-Section Profile
20-Year Semiconductor Foundry Cleanroom DCF Financial Model
Quantify ALE chamber CapEx, yield preservation, wafer scrap reduction, and discounted cash flows
Semiconductor Cleanroom ROI
Annual Scrap Savings
$33.70M
Yield recovery gain
Net Annual Operating Cash Flow
$33.49M
After tool OpEx
20-Year Net Present Value (NPV)
$318.52M
Discounted at 8.0%
Internal Rate of Return (IRR)
796.4%
Payback: 0.13 Years
Timeline Scrap Yield Savings ($) Chamber OpEx ($) Net Cash Flow ($) Discounted PV ($) Cumulative Cash Flow ($)
Comprehensive Engineering Guide: Atomic Layer Etching & Quasichemical Nanopatterning
1. The Self-Limiting Quasichemical Paradigm

As semiconductor dimensions shrink below the 2-nanometer logic node, traditional continuous plasma Reactive Ion Etching (RIE) reaches fundamental physical limitations. Continuous RIE couples radical flux and ion bombardment simultaneously, leading to severe microloading, aspect-ratio-dependent etching (ARDE), line edge roughness (LER) transfer, and crystal lattice damage. Atomic Layer Etching (ALE) resolves these challenges by subdividing the etch process into two cyclic, self-limiting half-reactions:

  • Step A (Surface Functionalization / Chemisorption): The wafer is exposed to a reactive gas (e.g., molecular $\text{Cl}_2$, $\text{CHF}_3$, or anhydrous $\text{HF}$). Active surface bonding sites chemisorb a self-terminating monolayer ($\theta \to 1.0$) through Langmuir adsorption kinetics. Once all surface dangling bonds are passivated into volatile precursor complexes (e.g., $\text{SiCl}_x$ or $\text{AlF}_x$), chemical uptake halts spontaneously. Spontaneous thermal etching is zero at standard chamber temperatures ($\alpha \approx 0$).
  • Purge Step 1: An ultra-high-purity inert gas pulse (Ar or $\text{N}_2$) evacuates non-adsorbed precursor molecules, eliminating gas-phase parasitic reactions.
  • Step B (Desorption / Product Removal): The functionalized monolayer is selectively removed using directional low-energy $\text{Ar}^+$ ion bombardment or thermal ligand exchange. The kinetic energy is tuned strictly inside the ALE energy window ($E_{\text{th,chem}} < E_{\text{ion}} < E_{\text{th,phys}}$), desorbing the halogenated layer without sputtering underlying pristine substrate atoms ($\beta \approx 0$).
  • Purge Step 2: Desorbed volatile reaction products ($\text{SiCl}_4$, $\text{Al(acac)}_3$) are pumped out to prepare the atomically pristine surface for the subsequent cycle.
2. Directional Plasma vs. Isotropic Thermal ALE

Two major technological families dominate modern ALE applications:

  • Directional Plasma ALE: Employs RF substrate biasing to accelerate ions normal to the wafer surface. Unchlorinated sidewalls remain untouched, enabling anisotropic vertical profile transfer with near-infinite vertical-to-lateral selectivity. Crucial for fin trimming, Gate-All-Around (GAA) nanosheet inner spacer cavity formation, and high-aspect-ratio contact hole etching.
  • Isotropic Thermal ALE: Uses sequential thermal gas exposures (e.g., fluorination followed by metal ligand exchange or metathesis) without any charged particle bombardment. Removes material uniformly in all spatial directions, making it ideal for conformal nanosheet channel release, atomic defect stripping, and gate dielectric pinhole mitigation.
3. The ALE Synergy Parameter & Energy Window Mechanics

The operational health of an ALE process is quantified by the ALE Synergy Parameter ($S_{\text{ALE}}$):

S_ALE = [EPC - (α + β)] / EPC × 100%

where $\text{EPC}$ is the total Etch-Per-Cycle (typically $0.8\text{ to }1.4\,\text{\AA/cycle}$), $\alpha$ is spontaneous chemical etch depth during Step A, and $\beta$ is physical sputtering depth during Step B. Achieving $S_{\text{ALE}} > 95\%$ requires precise RF synchronization:

  • Chemical Sputtering Threshold ($E_{\text{th,chem}}$): The lower energy bound, typically $15\text{ to }22\text{ eV}$ for silicon. Below this threshold, ion impact cannot overcome the surface binding energy of the halogenated species.
  • Physical Sputtering Threshold ($E_{\text{th,phys}}$): The upper energy bound, typically $60\text{ to }75\text{ eV}$ for pristine silicon. Above this threshold, ions induce direct collisional sputtering of unreacted substrate atoms, causing uncontrolled erosion and loss of synergy.
  • Narrow Energy Window ($\Delta E$): Operating in the $30\text{ to }45\text{ eV}$ range maximizes desorption yield while preventing subsurface lattice defects and atomic displacement cascades.
4. Fab Economics & Cleanroom Yield Protection

While ALE cycle times ($8\text{ to }15\text{ s}$) result in lower wafer throughput compared to high-power continuous RIE, the financial return is driven by wafer scrap prevention in cutting-edge 2nm and 3nm foundry nodes. In a typical 300mm logic fab processing 6,500 wafers/month valued at $18,000/wafer:

  • A $2.4\%$ reduction in device scrap (achieved by eliminating channel puncture and gate oxide leakage) recovers over $33.7M in annual gross value.
  • Deducting chamber operating expenses ($210k/yr for specialty halogens, turbo-pump overhauls, and RF maintenance), the net annual cash flow yields an investment payback period under 2 months.
Frequently Asked Questions: Atomic Layer Etching & Advanced Nanopatterning

Conventional Reactive Ion Etching (RIE) exposes the wafer to continuous simultaneous flux of reactive radicals and energetic ions, coupling chemical etching and physical sputtering into a continuous, non-self-limiting process susceptible to aspect-ratio-dependent etching (ARDE), microloading, and lattice damage. In contrast, Atomic Layer Etching (ALE) decouples the process into two sequential, self-limiting half-cycles: Step A functionalizes or chemisorbs a self-terminating monolayer (e.g., chlorination), followed by an inert purge, and Step B selectively desorbs the modified layer using low-energy inert ions (or thermal ligand exchange) operating strictly below the physical sputtering threshold of the underlying substrate. This provides monolayer-level precision, zero microloading, and atomically smooth surfaces.

The ALE synergy parameter quantifies the degree of self-limitation and process ideality, defined as $S_{\text{ALE}} = [\text{EPC} - (\alpha + \beta)] / \text{EPC} \times 100\%$, where $\text{EPC}$ is the etch-per-cycle, $\alpha$ is the spontaneous chemical etching from the precursor pulse alone, and $\beta$ is the physical sputtering from the ion bombardment pulse alone. In an ideal ALE regime, $\alpha = 0$ and $\beta = 0$, giving $S_{\text{ALE}} = 100\%$. In sub-2nm Gate-All-Around (GAA) nanosheets and 3D NAND contact hole patterning, an $S_{\text{ALE}}$ exceeding 90-95% is vital to avoid unwanted spontaneous lateral etching, preserve delicate nanosheet channels, and prevent punch-through into source/drain junctions.

Directional plasma ALE relies on an energy window defined by two distinct sputtering thresholds: $E_{\text{th,chem}}$ (the chemical sputtering threshold of the halogenated or modified surface layer, typically 15-25 eV) and $E_{\text{th,phys}}$ (the physical sputtering threshold of the pristine, unmodified substrate crystal, typically 50-75 eV for silicon and silicon dioxide). By tuning the pulsed RF bias to accelerate argon ions strictly between $E_{\text{th,chem}}$ and $E_{\text{th,phys}}$ (for example, at 35-45 eV for silicon), the ions carry sufficient kinetic energy to desorb the chlorinated $\text{SiCl}_x$ complexes through collisional cascade without damaging or sputtering the underlying pure silicon lattice.

Isotropic thermal ALE achieves atomic-scale layer removal purely through sequential, self-limiting thermal chemical reactions without any ion bombardment. A classic example is $\text{Al}_2\text{O}_3$ or $\text{MoS}_2$ etching: Step A fluorinates or oxidizes the surface using $\text{HF}$ or $\text{O}_2$ vapor to form a thin surface layer (e.g., $\text{AlF}_3$ or $\text{MoO}_3$). Step B exposes the modified layer to a metal precursor such as tin(II) acetylacetonate ($\text{Sn(acac)}_2$) or trimethylaluminum (TMA) that undergoes ligand exchange (metathesis) or coordination complex formation. The resulting reaction products are volatile coordination compounds that spontaneously desorb into the vacuum exhaust at moderate process temperatures (150-250°C), terminating once the modified surface layer is consumed.

A 300mm ALE cluster tool platform represents an initial capital expenditure of approximately $3.5M to $4.8M per chamber. While ALE cycle times (typically 8 to 15 seconds per cycle) are slower than continuous RIE, the economic justification in advanced 2nm logic and 3D NAND foundries stems from catastrophic yield loss prevention. At wafer values exceeding $18,000 to $25,000 per wafer, eliminating nanosheet channel thinning, pinhole punch-through, and line edge roughness (LER) provides a 2-4% reduction in overall wafer scrap, generating $15M to $35M in annual gross savings and delivering an investment payback period under 0.4 to 0.8 years with internal rates of return (IRR) exceeding 100%.
Related Semiconductor & Nanotechnology Laboratories
Atomic Layer Deposition Lab

Model self-limiting ALD surface chemisorption, thermal windows, high-k dielectric growth, and fab DCF.

High-NA EUV Scanner Lab

Model 13.5 nm LPP sources, 0.55 High-NA anamorphic optics, and stochastic shot noise.

Resonant Ultrasound Lab

Model Rayleigh-Ritz Lagrangian mechanics, full elastic stiffness tensors Cijkl, and laboratory DCF.

Muon Spin (μSR) Lab

Model spin-polarized muon implantation, vortex lattice London penetration depths, and quantum foundry DCF.