1. Introduction
In modern ultra-large-scale integration (ULSI) semiconductor manufacturing, defining features at sub-nanometer scales requires an intricate dance between lithography, plasma etching, and surface cleaning . As pattern dimensions shrink, the materials used to construct metal interconnects and insulating layers undergo a continuous evolution . The transition from aluminum (Al) metallization to copper (Cu) metallization, and the introduction of delicate organosilicate glass (OSG) as an ultra-low-k (ULK) dielectric, have significantly narrowed the process window for back-end-of-line (BEOL) integration [P3, A3]. Among the most critical yet challenging steps in this integration is the clean step immediately following dry etching: post-etch residue removal (PERR) [P2, P4].
During dry etching, reactive plasma gases chemically volatilize target materials while simultaneously generating complex, non-volatile polymeric byproducts . These byproducts, collectively referred to as post-etch residues, deposit along the sidewalls and bottoms of trenches and vias . If left unaddressed, these residues cause severe yield-limiting defects, such as high contact resistance, poor adhesion of subsequent barrier layers, and advanced reliability issues like electromigration (EM) or time-dependent dielectric breakdown (TDDB) [T1, A1, A2].
To safely eliminate these contaminants without damaging the surrounding delicate metal and dielectric structures, the industry relies heavily on specialty wet chemical formulations . Historically developed by EKC Technology (now part of Dupont), EKC post-etch residue removal (EKC) represents a benchmark family of semiaqueous organic solvent mixtures designed to selectively dissolve, chelate, and lift off highly crosslinked post-plasma polymers . Understanding the physical, chemical, and thermodynamic principles behind EKC cleaning is essential for any process engineer tasked with designing yield-viable BEOL metallization flows .
2. Physics & Mechanism
To appreciate how EKC wet cleans operate, one must first examine how dry etching and subsequent bulk photoresist (PR) strip steps alter the surface chemistry of the wafer .
Origin and Chemical Nature of Post-Etch Residues
In a typical dual-damascene etching sequence, fluorocarbon-based plasmas (e .g., carbon tetrafluoride ($CF_4$), octafluorocyclobutane ($C_4F_8$), and trifluoromethane ($CHF_3$)) are utilized to pattern dielectric trenches . These gases dissociate in the radio frequency (RF) plasma to yield fluorine ($F$) radicals that etch the silicon-oxygen framework, and carbon-fluorine ($CF_x$) radicals that polymerize on sidewalls to provide anisotropic profile control [T1, P4].
Simultaneously, ion bombardment sputters underlying metals (such as copper, aluminum, or titanium nitride hard masks) into the plasma, where they react with the fluorocarbon species [P1, P4]. This results in a highly complex, heterogeneous residue consisting of: 1 (Engineering Practice). Halogenated polymers: Chemically robust carbon-fluorine ($CF_x$) and carbon-chlorine ($CCl_x$) chains [P1, P4]. 2. Organometallic complexes: Metallic atoms (such as $Al$, $Cu$, or $Ti$) chemically incorporated into the polymer matrix [P1, A3]. 3. Inorganic oxides/halides: Non-volatile metal fluorides ($AlF_3$, $CuF_2$) or silicon oxide ($SiO_x$) species [P1, A3].
When the wafer is subjected to downstream oxygen ($O_2$) plasma ashing to remove the remaining bulk photoresist, the high thermal load and active oxygen radicals cause further crosslinking and carbonization of the residue, creating a hardened, "skin-like" organic shell that is highly resistant to standard organic solvents [P1, P2].
[Dry Etch Plasma (CFx, Cl)] ──> [Sputtered Metals & PR Fragments] ──> [Mixed Halogenated/Organometallic Residue]
│
(O2 Plasma Ashing)
▼
[Highly Crosslinked Hardened Skin]
Chemical Reaction Principles of EKC Semiaqueous Cleaners
EKC formulations are engineered as "semiaqueous organic mixtures" to exploit synergistic chemical mechanisms . A typical formulation includes:
- Organic Polar Solvents: Solvents such as dimethyl sulfoxide (DMSO) or N-methylpyrrolidone (NMP) act to swell the crosslinked organic polymer matrix (Engineering Practice). Swelling expands the intermolecular spaces within the polymer network, reducing its mechanical integrity and allowing other active chemical species to penetrate deep into the residue layer (Engineering Practice).
- Alkanolamines (Alkaline Agents): Amines provide two critical functions: they maintain an alkaline pH to promote nucleophilic attack on the polymer backbone, and they act as basic ligands that coordinate with transition metal ions [P1, A1]. By forming soluble chelate complexes with oxidized metal residues (e .g., $Cu^{2+}$ or $Al^{3+}$), amines pull insoluble organometallic complexes into the solution [P1, A1].
- Water (Semiaqueous Component): Water acts as a solvent for inorganic species and facilitates the dissociation of active chemicals, enabling hydrolysis reactions that break down silicon-oxygen-fluorine networks .
- Fluoride Ions (Optional/Specific Formulations): In formulations designed to target silicon-rich or refractory metal residues (such as titanium or tungsten), dilute fluorides (e .g., ammonium fluoride) are added to chemically etch the underlying oxide or metal-halide skeleton, thereby "undercutting" and lifting off the residue [A1, A3].
Corrosion Inhibition and Surface Passivation
A major thermodynamic challenge during wet cleaning is preventing the simultaneous chemical dissolution of exposed metal interconnects . Since EKC chemistries contain highly active amines and water, they can easily cause galvanic or chemical corrosion of copper, aluminum, or cobalt .
To mitigate this, advanced EKC formulations incorporate organic corrosion inhibitors . Molecules such as benzotriazole (BTA) or long-chain phosphonic acids (e (Engineering Practice).g., dodecylphosphonic acid) are introduced to selectively adsorb onto exposed metal surfaces . These molecules self-assemble into a hydrophobic, dense monolayer that acts as a physical barrier, suppressing anodic and cathodic electrochemical reactions on the metal while allowing the bulk chemical solution to dissolve the adjacent post-etch residues .
3. Process Principles
The performance of an EKC post-etch residue removal process is determined by several highly interactive physical and chemical parameters . Optimizing these parameters involves managing delicate tradeoffs between residue cleanliness, material loss, and electrical reliability .
Temperature Effects
Chemical reaction rates within the wet cleaning bath follow the Arrhenius relationship, where the rate constant increases exponentially with absolute temperature [A1, A2]. Elevating the temperature of the EKC chemistry significantly enhances:
- The diffusion rate of solvent molecules into the dense polymeric residues .
- The hydrolysis rate of crosslinked polymers .
- The solubility limit of chelated metal complexes in the chemical bath (Engineering Practice).
However, an excessively high process temperature narrows the operating window by accelerating the diffusion of corrosive species through the inhibitor passivation layer, leading to severe metal corrosion [A1, A3]. Furthermore, high temperatures can cause organic solvent components to evaporate prematurely, altering the chemical ratio of the bath and destabilizing the process (Engineering Practice).
Concentration and Bath Lifetime
The active components of the EKC solution—particularly the water-to-solvent ratio and the concentration of active amines or fluorides—must be kept in a steady-state equilibrium . Over time, chemical usage leads to: 1 (Engineering Practice). Water absorption or evaporation: Semiaqueous solvents are often hygroscopic, absorbing moisture from the cleanroom atmosphere, which shifts the pH and chemical activity (Engineering Practice). 2. Bath loading: As residues are dissolved, the concentration of complexed metal ions in the bath increases, which can eventually lead to redeposition of contaminants onto the wafer surface .
Rinse Chemistry and Surface Tension
Once the residues are dissolved or detached from the surface, they must be completely rinsed away without redepositing or collapsing the high-aspect-ratio trenches .
[Polymer Swelling/Dissolution] ──> [De-bonding & Mass Transport] ──> [Surfactant Penetration] ──> [IPA/DIW Displacement]
Due to the extreme hydrophobicity of fluorinated post-etch residues and the low surface energy of porous low-k dielectric materials, standard deionized water (DIW) rinses can fail due to high surface tension . Under Young's equation, liquids with high surface tension cannot wet or penetrate narrow, hydrophobic structures .
To overcome this, surfactants are often added to the clean or rinse steps to lower the static and dynamic surface tension . Additionally, a post-clean solvent rinse using isopropyl alcohol (IPA) is frequently integrated prior to the final DIW rinse . IPA has an exceptionally low surface tension, allowing it to displace the organic EKC chemistry from within ultra-fine trenches and completely dissolve any remaining surfactant molecules, preventing them from being trapped inside the porous dielectric network .
Mechanical and Acoustic Assistance
Relying solely on diffusion-limited chemical dissolution is often insufficient for sub-22nm nodes . Megasonic agitation or high-pressure spray technology is applied during the wet process to introduce mechanical energy . This physical action thins the boundary layer of the liquid at the wafer surface, accelerating the mass transport of fresh reactants to the trench bottom and aiding in the physical dislodgement of loosened polymer skins .
4. Challenges & Failure Modes
Designing a robust EKC process requires navigating several physical failure modes that directly impact device reliability and yield .
Low-k Dielectric Damage (k-Value Degradation)
Porous organosilicate glasses (SiCOH) achieve a low dielectric constant by incorporating hydrophobic organic methyl groups ($-CH_3$) into a porous $SiO_2$-like matrix [P3, P4]. During dry etching or conventional oxygen plasma ashing, vacuum ultraviolet (VUV) photons and active oxygen radicals break the silicon-carbon ($Si-C$) bonds, leaving behind a highly hydrophilic, damaged $SiO_x$ surface enriched with silanol ($-SiOH$) groups [P3, P4].
If an EKC chemistries has a poorly optimized pH or contains aggressive amine concentrations, it will chemically attack this damaged hydrophilic layer, causing further isotropic etching of the trench sidewalls [P2, P3]. This chemical attack leads to:
- Critical dimension (CD) blow-out: The trench width increases, causing severe capacitance variation .
- Moisture absorption: The hydrophilic silanol groups readily absorb atmospheric moisture (water has an extremely high k-value of approximately 80), leading to a drastic increase in the effective k-value of the dielectric stack [P3, P4].
Chemical and Surfactant Entrapment
When surfactants or polar organic solvents are used to enhance wetting in porous low-k materials, the pore channels (often 1-2 nm in diameter) can physically trap these organic molecules . If the subsequent rinse sequence (e (Engineering Practice).g., using DIW instead of IPA) fails to completely extract these trapped species, they remain inside the dielectric film . Upon subsequent thermal processing, these trapped organics decompose, causing a dramatic increase in the refractive index, leakage current, and dielectric loss, ultimately leading to premature TDDB failure .
Cross-section of Porous Low-k:
┌──────────────────────────────────────────────┐
│ _ _ _ _ _ _ _ _ │ <-- Trapped Surfactants / Amines
│ (─)(─) (─) (─) (─) (─) (─) (─) │ in porous matrix
└──────────────────────────────────────────────┘
Galvanic and Localized Metal Corrosion
In structures where dissimilar metals are electrically connected and exposed to the same conductive wet chemical solution (such as copper lines in contact with a titanium or tungsten nitride barrier layer), a galvanic cell is established . The metal with the lower electrochemical potential acts as an anode and undergoes rapid, localized dissolution (Engineering Practice).
If the corrosion inhibitors in the EKC formulation are depleted or fail to form a continuous, defect-free passivating monolayer, severe galvanic corrosion will occur at the metal-barrier interface [A1, A3]. This manifests as "voiding" or "notching" of the metal lines, drastically reducing the cross-sectional area of the conductor and leading to early electromigration failures during device operation [A1, A3].
Incomplete Residue Removal (Residue Blockage)
If the dry etch plasma or oxygen ashing temperature is too high, the photoresist and sidewall polymers become excessively carbonized and crosslinked . If the EKC process window is too conservative (e (Engineering Practice).g., due to low temperatures or short process times designed to protect the low-k dielectric), these hardened polymers will not be completely dissolved [P1, A3]. Leftover residues at the bottom of contact vias act as highly insulating electrical barriers, resulting in open-circuit failures or erratic contact resistance .
5. Technology Node Evolution
The strategy for post-etch residue removal has undergone dramatic shifts as the industry scaled from planar transistors to 3D FinFET architectures .
28nm Planar Node
At the 28nm Planar Flow, the interconnect pitch was relatively wide, and the low-k dielectrics used were relatively dense (with low porosity) (Engineering Practice). Under these conditions, the process window for wet cleaning was wide . Wafers typically underwent a high-temperature oxygen plasma ash step to remove bulk photoresist, followed by a standard single-wafer EKC clean to dissolve the remaining halogenated residues . The low porosity of the dielectric prevented significant chemical uptake, and standard DIW rinses were sufficient to clean the features .
14nm FinFET Node
With the introduction of the 14nm FinFET architecture, pitch scaling forced the adoption of porous ultra-low-k (ULK) materials . At this node, conventional high-temperature oxygen plasma ashing became highly destructive, as the plasma damage extended deep into the trench sidewalls, causing severe dielectric degradation [P2, P3].
To counter this, the industry shifted toward metal hard mask (MHM) integration schemes, where the photoresist is stripped before the final trench etch, or dry-strip-free processes are utilized . EKC formulations had to be re-engineered to operate at lower temperatures and with highly selective chemistries that could differentiate between organic photoresist residues and the carbon content of the porous SiCOH dielectric [P2, P4].
7nm FinFET Node and Beyond
At the 7nm FinFET node and beyond, the half-pitch of the tightest metal lines shrank to sub-30 nm levels . At this scale:
- The "Zero-Loss" Requirement: Any physical loss of the dielectric or metal during the clean step is unacceptable due to the tight CD budgets .
- Plasma-less stripping: Standard plasma-based resist stripping must be entirely replaced or combined with selective photochemical treatments . For instance, UV radiation is used to selectively break bonds in the resist, followed by highly targeted ozone ($O_3$) gas or advanced EKC semiaqueous solvents to dissolve the residue without using a plasma .
- In-situ dielectric repair: Post-etch residue removal is now immediately coupled with surface modification steps . After the EKC clean removes the polymers, silylation agents are introduced to react with damaged silanol groups, chemically grafting hydrophobic hydrocarbon chains back onto the dielectric sidewalls to restore the original k-value and form a barrier against moisture and metal diffusion .
| Node | Interconnect Material | Dielectric Type | PERR Scheme | Primary Cleaning Risk |
|---|---|---|---|---|
| 28nm | Copper / TaN / Ta | Dense Low-k ($k \approx 3.0$) | High-Temp Ash + Batch EKC | Polymer residue retention |
| 14nm | Copper / TaN / TiN | Porous ULK ($k \approx 2.5$) | Low-Temp Ash + Single-Wafer EKC | Dielectric k-value degradation |
| 7nm & beyond | Cu, Co, Ru / Barrierless | Highly Porous ULK ($k < 2.2$) | Plasma-free UV/$O_3$ + EKC + Silylation | CD loss, metal corrosion, moisture uptake |
6 [P2]. Related Processes
An EKC wet clean does not exist in isolation; its chemistry and execution are highly dependent on upstream and downstream process steps .
[Dry Etching (ICP/RIE)] ──> [Plasma Ashing (Bulk PR Strip)] ──> [EKC Post-Etch Wet Clean] ──> [ALD Barrier/Seed Deposition]
Dry Etching and Plasma Ashing
The plasma gas composition and bias power used during dry etching dictate the chemical composition and thickness of the sidewall polymer [T1, P1]. If the etch process uses high fluorocarbon ratios, the resulting residue will be highly fluorinated and require solvent formulations with high swelling power . Similarly, the choice of downstream ashing—whether in-situ low-temperature O2 plasma ashing or remote helium/hydrogen ($He/H_2$) plasma—directly determines the degree of polymer crosslinking that the EKC chemistry must overcome [P1, P3].
Chemical Mechanical Planarization (CMP)
Following the EKC wet clean, metal barrier layer deposition, and copper electroplating, the excess metal is polished away using chemical mechanical planarization (CMP) . Post-CMP cleaning represents another critical wet chemical step (Engineering Practice). While CMP cleans focus primarily on removing silica/alumina nanoparticles and organic slurry residues from a planar surface, they share similar thermodynamic requirements with EKC cleans, specifically the need for precise corrosion inhibition to protect the freshly exposed metal lines .
Barrier/Seed Layer Deposition
The ultimate test of an EKC clean's effectiveness is the quality of the subsequent barrier and seed layer deposition . Advanced nodes utilize atomic layer deposition (ALD) to deposit ultra-thin conformal barrier layers (e .g., $TaN$ or $TiN$) (Engineering Practice). If any microscopic post-etch residue remains at the bottom of the vias, it will disrupt the nucleation of the ALD precursors, leading to local film discontinuity, high contact resistance, and poor adhesion that causes the metal stack to delaminate under thermal stress .
7. Future Outlook
As the semiconductor industry marches toward the sub-3nm era—marked by the adoption of Gate-All-Around (GAA) nanosheet transistors and Backside Power Delivery Networks (BSPDN)—the chemistry of post-etch residue removal must undergo further evolutionary leaps .
With the potential transition from copper to alternative metals such as cobalt ($Co$) or ruthenium ($Ru$) for the tightest metal levels, traditional EKC formulations designed for copper-dielectric systems must be entirely reformulated (Engineering Practice). Ruthenium and cobalt exhibit vastly different electrochemical properties and corrosion kinetics compared to copper, requiring new chelating ligands and highly tailored corrosion inhibitors that can remain stable across wide pH ranges [A1, A3].
Furthermore, as aspect ratios of contacts and vias exceed 10:1 or even 20:1, supercritical carbon dioxide ($scCO_2$) drying and cleaning systems are being actively researched to complement or replace traditional wet solvents . Supercritical fluids possess gas-like diffusivity and zero surface tension, enabling them to carry active organic cleaning agents into ultra-fine structures without the risk of capillary-force-induced pattern collapse [P2, P4]. Ultimately, the future of post-etch residue removal lies in the co-design of dry-etch plasma chemistries, non-damaging photochemical strip processes, and highly selective, zero-damage wet chemical formulations that treat the wafer surface not just as a materials-to-be-removed boundary, but as a delicate chemical system requiring atom-scale precision [P2, A2].