Technical Principles of Photoresist Removal in Advanced Semiconductor Manufacturing: Mechanisms, Challenges, and Node Evolution

Introduction

In modern integrated circuit (IC) fabrication, lithography serves as the primary mechanism for spatial patterning, defining the critical dimensions of transistors, contacts, and interconnects [T2]. After a photoresist (PR) layer is exposed and developed, it acts as a temporary sacrificial mask during subsequent steps such as dry etching or high-dose ion implantation [T1, P2]. Once these structural or chemical modifications of the underlying material are complete, the temporary mask must be entirely eliminated—a critical process step known as photoresist removal, resist strip, or ashing [T1, P1].

The absolute removal of polymeric resist materials is paramount to securing high device yield and electrical reliability [P3]. Any residual organic contamination can block subsequent deposition steps, such as the application of an interfacial capping layer, resulting in voiding, high contact resistance, or catastrophic electrical opens [P3, A1]. Consequently, the photoresist removal process must achieve extreme selectivity: it must rapidly and completely strip the organic mask without causing material loss, oxidation, or structural damage to the delicate active layers, ultra-low-k dielectrics, or metal interconnects underneath [P1, P2]. Historically dominated by wet chemical dissolution, modern photoresist removal relies on highly engineered plasma ashing, specialized wet stripping formulations, and emerging dry-chemical techniques tailored to meet the demanding requirements of advanced technology nodes [T1, P3, A1].

Physics & Mechanism

The physical and chemical mechanisms of photoresist removal are divided into two main categories: dry stripping (plasma ashing) and wet chemical stripping [T1, A1]. Each relies on breaking down the complex hydrocarbon chains of the photoresist polymer into smaller, soluble, or volatile species [P1, A1].

Dry Stripping: Plasma Ashing and Radical Kinetics

Dry resist strip, or plasma ashing, typically utilizes a low-pressure discharge of oxygen-based ($O_2$) or hydrogen-based ($H_2$) gas mixtures to generate highly reactive atomic radicals [T1, P1]. In conventional oxygen plasma ashing, a radio frequency (RF) or microwave generator excites $O_2$ gas to produce oxygen radicals [T1, P4]. These radicals diffuse to the wafer surface and react with the organic polymer chains of the photoresist via oxidation [P1, P4]. The basic chemical reaction converts the solid hydrocarbon polymer into volatile gaseous byproducts such as carbon monoxide ($CO$), carbon dioxide ($CO_2$), and water vapor ($H_2O$), which are subsequently evacuated by the vacuum system [P4]:

$$C_xH_y + O(\text{radical}) \rightarrow CO + CO_2 + H_2O$$

The ashing rate ($R$) in a downstream plasma system, where purely chemical, radical-driven processes dominate over physical ion bombardment, is a thermally activated process governed by the Arrhenius reaction rate equation [P1, T1]:

$$R = R_0 \exp\left(-\frac{E_a}{kT}\right)$$

Where $R_0$ is the pre-exponential factor, $E_a$ is the apparent activation energy of the chemical reaction, $k$ is the Boltzmann constant, and $T$ is the absolute temperature of the substrate [P1].

While oxygen-based plasma ashing is highly efficient, it can cause severe oxidation of underlying silicon, silicon nitride ($Si_3N_4$), and silicon germanium ($SiGe$) surfaces [P1, P2]. To mitigate this, advanced processes leverage downstream hydrogen-based ($H_2$) plasmas [P1]. In these systems, reactive hydrogen radicals break the $C-C$ and $C-H$ polymer bonds via chemical reduction, generating volatile methane ($CH_4$) and other light hydrocarbons [P1]. The addition of nitrogen ($N_2$) to $H_2$ downstream plasmas changes the plasma reaction pathways, suppressing radical recombination and significantly increasing the density of reactive hydrogen radicals [P1]. This kinetic optimization enhances the photoresist removal rate without risking the oxidation of sensitive substrate materials [P1].

Wet Stripping: Swelling and Dissolution Chemistry

In wet chemical stripping, photoresists are removed using organic solvents or highly alkaline aqueous formulations [A1]. Wet stripping relies on three sequential physical steps: solvent penetration, polymer swelling, and dissolution [A2].

Water-soluble polar aprotic solvents, such as dimethyl sulfoxide (DMSO) or N-methylpyrrolidone (NMP), diffuse into the bulk photoresist matrix, loosening intermolecular forces and causing the polymer to swell [A1]. Concurrently, alkaline components such as quaternary ammonium hydroxides—most notably tetramethylammonium hydroxide (TMAH)—provide a highly basic environment (often exceeding pH 13) that hydrolyzes ester groups and breaks down crosslinked polymer backbones [A1]. For metallized substrates, wet stripping solutions must include specialized corrosion inhibitors, such as triazine-based compounds, which selectively coordinate to exposed copper ($Cu$) or aluminum ($Al$) surfaces, forming a protective passivation layer that prevents metal consumption in the highly alkaline media [A1, A2].

Process Principles

Optimizing a photoresist removal process requires balancing several interacting process parameters to maximize the strip rate while minimizing substrate damage and defectivity [A1].

Substrate Temperature

Temperature acts as the primary accelerator for chemical downstream ashing, dictated by the Arrhenius relationship [P1]. Increasing the substrate temperature exponentially enhances the reaction rate of oxygen or hydrogen radicals with the photoresist polymer [P1]. However, in oxygen-based plasmas, elevated temperatures also accelerate the diffusion-limited oxidation of the underlying silicon, silicon nitride, or metal silicide substrates, leading to material loss [P1]. Furthermore, excessive temperatures can cause volatile components in the bulk resist to gasify rapidly, leading to structural failures like blistering or popping (Engineering Practice).

Gas Chemistry and Gas Mixing Ratios

The choice and ratio of process gases directly dictate the plasma chemistry [P1, P4].

• $O_2/N_2$ Systems: Adding a small percentage of $N_2$ to an $O_2$ plasma modifies the electron temperature in the discharge, maximizing the generation of atomic oxygen radicals and leading to a significant increase in the photoresist ashing rate [P1]. However, if the $N_2$ concentration exceeds a critical threshold, the reactive radicals are diluted, causing the ashing rate to drop sharply [P1].
• $H_2/N_2$ Systems: In non-oxidizing chemistries, introducing $N_2$ into a downstream $H_2$ plasma can yield up to a threefold increase in the photoresist removal rate compared to pure $H_2$ by preventing the recombination of hydrogen radicals on the chamber walls [P1].
• Fluorinated Gas Addition: Incorporating trace amounts of fluorinated gases (such as $CF_4$ or $NF_3$) into an oxygen plasma dramatically increases the strip rate [P4]. The highly electronegative fluorine radicals abstract hydrogen from the polymer backbone, creating reactive radical sites that accelerate oxidation [P4]. However, this approach introduces a trade-off: fluorine radicals readily etch silicon dioxide ($SiO_2$) and silicon substrates, severely degrading material selectivity [P1].

Chamber Pressure and Plasma Configuration

Chamber pressure directly influences the mean free path of species and the dominant etching regime [P4, T1].

• Direct Plasma Mode: In parallel-plate capacitively coupled plasma (CCP) or inductively coupled plasma (ICP) configurations, the wafer is exposed to both reactive neutral radicals and high-energy ions accelerated across the plasma sheath [T1, P4]. Increasing the bias power enhances ion bombardment, which physically sputters away highly crosslinked surface crusts, but increases physical substrate damage [P2, P4].
• Downstream/Afterglow Mode: Generating the plasma upstream and allowing only neutral radicals to diffuse to the substrate minimizes ion-bombardment damage [T1]. In this downstream regime, increasing pressure increases the radical flux but decreases ion energy to zero, ensuring a highly selective, purely chemical isotropic strip process [T1].

Parameter Direction	Effect on Ashing Rate	Effect on Substrate Damage / Loss	Effect on Residue Removal
Increase Temperature	Exponential Increase [P1]	Increases Oxidation/Loss [P1]	Improves Dissolution/Desorption (Engineering Practice)
Increase Bias Power	Moderate Increase (Engineering Practice)	Increases Sputtering Damage [P2, P4]	Breaks High-Dose Crusts [P2]
Increase Pressure	Moderate Increase [T1]	Decreases Ion Bombardment Damage [T1]	Decreases Crust Breaking Efficiency [P2]
Add Fluorine Gas	Dramatic Increase [P4]	High Oxide/Silicon Etch Risk [P4]	Excellent for Inorganic Residues [P4]

Challenges & Failure Modes

Photoresist stripping is plagued by physical and chemical failure modes, particularly when dealing with modified resists or delicate underlying structures [P2, P3].

High-Dose Implantation (HDI) Crust Formation

One of the most severe challenges in front-end photoresist stripping occurs after high-dose ion implantation (HDI) [P2]. During high-dose doping steps, energetic dopant ions (such as arsenic, phosphorus, or boron) bombard the photoresist mask [P2]. This high-energy ion bombardment induces intense bond-breaking, hydrogen abstraction, and carbon-carbon cross-linking within the top layer of the photoresist, forming a dense, carbon-rich, hydrogen-depleted modified surface layer known as the "crust" [P2].

               [ Energetic Dopant Ions (As+, P+, B+) ]
                             │   │   │
                             ▼   ▼   ▼
     ┌─────────────────────────────────────────────────────────┐
     │  Carbon-Rich, Cross-linked, Dopant-Accumulated CRUST    │  ◄── Highly Resistant Layer
     ├─────────────────────────────────────────────────────────┤
     │                                                         │
     │  Unmodified Bulk Hydrocarbon Photoresist                │  ◄── Contains Volatile Solvents
     │                                                         │
     └─────────────────────────────────────────────────────────┘

The crust is highly resistant to standard chemical dissolving agents and oxidizing downstream plasmas [P2]. Standard $O_2$ plasmas preferentially diffuse through and ash the unmodified bulk resist underneath the crust [P2]. As the bulk resist ashes, it generates volatile $CO_2$ and $H_2O$ gases that build up pressure beneath the impermeable crust (Engineering Practice). When the pressure exceeds the mechanical strength of the crust, the layer ruptures violently—a failure mode known as "resist popping" or "blistering" [P2] (Engineering Practice). This event scatters highly contaminated, dopant-rich polymeric particles across the wafer, creating severe defectivity [P2] (Engineering Practice). To prevent this, $H_2$-based reducing plasmas or specialized solvent-swelling wet chemistries must be employed to gently break down and remove the crust layer prior to high-temperature bulk stripping [P2, A1].

Substrate Loss, Oxidation, and Material Damage

During front-end-of-line (FEOL) processing, stripping photoresist from active device regions can result in unwanted substrate consumption [P1, P2]. For example, in silicon germanium ($SiGe$) channels, oxidizing plasmas ($O_2/N_2$) selectively oxidize the germanium, leading to surface roughness and active material loss [P2]. Similarly, standard oxygen ashing can oxidize exposed silicon and silicon nitride surfaces [P1].

In back-end-of-line (BEOL) copper interconnect fabrication, the introduction of porous ultra-low-k dielectrics creates a highly vulnerable interface [P3]. During plasma ashing, reactive oxygen radicals readily diffuse into the pores of the low-k dielectric, attacking the methyl ($–CH_3$) groups that maintain the material's hydrophobic nature and low dielectric constant [P3]. This carbon depletion leaves hydrophilic silanol ($–SiOH$) groups, which absorb atmospheric moisture (Engineering Practice). Because water has a high dielectric constant, this moisture absorption dramatically increases the overall k-value of the dielectric, destroying interconnect performance [P3]. Additionally, this plasma damage causes structural collapse of the low-k film near the trench edges, leading to line-to-line leakage and reliability failures [P3].

Technology Node Evolution

As semiconductor technology scaled from planar devices to 3D architectures, photoresist removal transformed from a straightforward cleaning step into a highly selective, damage-free surface engineering process [P2, P3].

28nm Planar Node

At the planar 28nm Planar Flow, photoresist stripping relied heavily on high-temperature, downstream $O_2/N_2$ plasma ashing [T1, P1]. At this node, gate oxides and bulk silicon substrates were robust enough to tolerate moderate oxygen radical exposure [P1]. Wet stripping utilized standard organic solvents or sulfuric acid-hydrogen peroxide mixtures (SPM) to remove bulk resist and post-etch residue (PER) without significant concern for substrate consumption, as feature aspect ratios remained relatively low [T1, P3].

14nm FinFET Node

With the transition to the 3D transistor architecture in the 14nm FinFET node, photoresist stripping encountered major integration bottlenecks [T2]. The source/drain regions featured highly sensitive, epitaxially grown $SiGe$ structures [P2]. Traditional $O_2$ ashing could no longer be used for post-implant strip because the intense oxidation consumed the thin $SiGe$ fins, altering the active device dimensions [P2].

This node drove the widespread adoption of non-oxidizing, downstream $H_2/N_2$ and $He/H_2$ plasmas [P1, P2]. These reducing chemistries successfully removed high-dose implant crusts and bulk photoresist without oxidizing the underlying $SiGe$ or silicon fins [P2]. However, because hydrogen-based ashing is chemically slower, it required highly optimized downstream microwave or inductively coupled plasma sources to maintain acceptable manufacturing throughput [P1].

7nm FinFET and Beyond

At the 7nm FinFET node and beyond, the introduction of extreme ultraviolet (EUV) lithography necessitated the use of exceptionally thin photoresist layers coupled with advanced organic bottom anti-reflective coating (BARC) materials [T2]. The extremely small pitches and high aspect ratios meant that even minor plasma exposure could distort the fragile photoresist profiles or cause the collapse of adjacent low-k dielectric lines [P3].

To prevent plasma-induced damage underneath metal hard masks (MHM), non-plasma alternatives were introduced [P3]. These advanced schemes utilize ultraviolet (UV) pre-treatment to chemically modify the resist polymer, followed by highly selective ozonolysis ($O_3$) to cleave the polymer backbone at low temperatures [P3]. This chemical modification renders the photoresist soluble in mild organic solvents, completely bypassing plasma exposure and preserving the structural and electrical integrity of the underlying low-k trenches and ultra-shallow junctions [P3].

Related Processes

Photoresist removal is deeply integrated into the surrounding wafer fabrication sequence, and its performance directly dictates the success of adjacent processes [A1].

┌────────────────────────────────────────────────────────┐
│ 1 *(Engineering Practice)*. LITHOGRAPHY & PATTERNING                            │
│    Coating of BARC and Photoresist Mask                │
└───────────────────────────┬────────────────────────────┘
                            │
                            ▼
┌────────────────────────────────────────────────────────┐
│ 2 [A2]. ETCH / ION IMPLANTATION                             │
│    Pattern Transfer or High-Dose Doping (Crust Formed)  │
└───────────────────────────┬────────────────────────────┘
                            │
                            ▼
┌────────────────────────────────────────────────────────┐
│ 3 [P2]. PHOTORESIST REMOVAL (Ashing / Resist Strip)         │  ◄── Focus of This Article
│    Radical Oxidation/Reduction or Wet Dissolution      │
└───────────────────────────┬────────────────────────────┘
                            │
                            ▼
┌────────────────────────────────────────────────────────┐
│ 4 [A2]. POST-ASH CLEANING                                   │
│    Wet clean (DHF / Solvent) removes residue & oxides   │
└───────────────────────────┬────────────────────────────┘
                            │
                            ▼
┌────────────────────────────────────────────────────────┐
│ 5 *(Engineering Practice)*. METALLIZATION & CAPPING                             │
│    Deposition of Capping or Liner Layers               │
└────────────────────────────────────────────────────────┘

Lithography and BARC Etching

Before photoresist stripping can occur, the resist must act as a barrier during the transfer of the circuit pattern into the underlying films [T2]. This pattern transfer stack typically includes an organic bottom anti-reflective coating (BARC) to control light reflection during lithographic exposure [T2]. Because BARC is also an organic polymer, it is frequently stripped simultaneously with the photoresist during the plasma ashing or wet stripping step, requiring a process that strips both organic materials with uniform kinetics [A2].

Wet Chemical Cleaning (Post-Ash Clean)

Dry plasma ashing is rarely a standalone solution; it is almost always paired with a subsequent wet cleaning step [P2, P4]. Ashing removes the bulk organic components but often leaves behind non-volatile inorganic residues, such as dopant oxides (e [P1].g., arsenic or phosphorus oxides) from implantation, or fluorinated polymer residues from the preceding etch step [P2, P4]. A subsequent wet clean, utilizing dilute hydrofluoric acid (DHF) or proprietary organic solvent mixtures, is required to dissolve these oxidized residues and clear the surface for subsequent processing [P2, P4].

Interfacial Capping Layers

Following the complete stripping and cleaning of the photoresist, the open structures are filled or coated with subsequent films [P3]. In BEOL metallization, for example, a dielectric capping layer is deposited directly over the polished copper lines and low-k dielectric surfaces [P4]. If any trace organic photoresist residue or ash-induced defect remains on the sidewalls or top surface, it will degrade the adhesion of the capping layer, leading to delamination, void formation, and electromigration failures under high current densities [P2].

Future Outlook

As the semiconductor industry transitions from FinFETs to gate-all-around (GAA) nanosheets and stacked complementary field-effect transistors (CFETs), photoresist removal faces unprecedented physical constraints [T2].

The extreme geometries of nanosheet architectures present a severe high aspect ratio process challenge [T2]. When stripping resist from deep, narrow trenches or from underneath suspended nanosheet channels, radical transport becomes severely diffusion-limited [P3]. Standard downstream plasmas cannot deliver sufficient neutral radical flux to the bottom of these deep structures, leading to incomplete stripping (Engineering Practice).

To address this, the industry is actively researching gas-phase atomic layer etching (ALE) of polymers (Engineering Practice). This approach employs self-limiting sequential reactions: a modification gas first adsorbs uniformly along the high-aspect-ratio polymer surface, followed by a localized exposure to a reactive gas or thermal energy that selectively desorbs only the modified monolayer (Engineering Practice). This gas-phase ALE-type stripping guarantees highly conformal, damage-free, and complete photoresist removal in the most complex 3D nanostructures [A1]. Simultaneously, environmental, health, and safety (EHS) regulations are driving the development of green wet-stripping solvents with lower volatility and higher biodegradability to replace conventional polar aprotic solvents in high-volume manufacturing [A2].

---