Semiconductor Manufacturing Process: Epitaxy (Epi) Modeling

1. Introduction to Epitaxy

Epitaxy is the controlled growth of a crystalline thin film on a crystalline substrate, where the deposited layer inherits the crystallographic orientation of the substrate.

1.1 Types of Epitaxy

• Homoepitaxy
• Same material deposited on substrate
• Example: Silicon (Si) on Silicon (Si)
• Maintains perfect lattice matching
• Used for creating high-purity device layers

• Heteroepitaxy
• Different material deposited on substrate
• Examples:
• Gallium Arsenide (GaAs) on Silicon (Si)
• Silicon Germanium (SiGe) on Silicon (Si)
• Gallium Nitride (GaN) on Sapphire ($\text{Al}_2\text{O}_3$)
• Introduces lattice mismatch and strain
• Enables bandgap engineering

2. Epitaxy Methods

2.1 Chemical Vapor Deposition (CVD) / Vapor Phase Epitaxy (VPE)

• Characteristics:
• Most common method for silicon epitaxy
• Operates at atmospheric or reduced pressure
• Temperature range: $900°\text{C} - 1200°\text{C}$

• Common Precursors:
• Silane: $\text{SiH}_4$
• Dichlorosilane: $\text{SiH}_2\text{Cl}_2$ (DCS)
• Trichlorosilane: $\text{SiHCl}_3$ (TCS)
• Silicon tetrachloride: $\text{SiCl}_4$

• Key Reactions:

$$\text{SiH}_4 \xrightarrow{\Delta} \text{Si}_{(s)} + 2\text{H}_2$$

$$\text{SiH}_2\text{Cl}_2 \xrightarrow{\Delta} \text{Si}_{(s)} + 2\text{HCl}$$

2.2 Molecular Beam Epitaxy (MBE)

• Characteristics:
• Ultra-high vacuum environment ($< 10^{-10}$ Torr)
• Extremely precise thickness control (monolayer accuracy)
• Lower growth temperatures than CVD
• Slower growth rates: $\sim 1 \, \mu\text{m/hour}$

• Applications:
• III-V compound semiconductors
• Quantum well structures
• Superlattices
• Research and development

2.3 Metal-Organic CVD (MOCVD)

• Characteristics:
• Standard for compound semiconductors
• Uses metal-organic precursors
• Higher throughput than MBE

• Common Precursors:
• Trimethylgallium: $\text{Ga(CH}_3\text{)}_3$ (TMGa)
• Trimethylaluminum: $\text{Al(CH}_3\text{)}_3$ (TMAl)
• Ammonia: $\text{NH}_3$

2.4 Atomic Layer Epitaxy (ALE)

• Characteristics:
• Self-limiting surface reactions
• Digital control of film thickness
• Excellent conformality
• Growth rate: $\sim 1$ Å per cycle

3. Physics of Epi Modeling

3.1 Gas-Phase Transport

The transport of precursor gases to the substrate surface involves multiple phenomena:

• Governing Equations:

• Continuity Equation:

$$\frac{\partial \rho}{\partial t} + abla \cdot (\rho \mathbf{v}) = 0$$

• Navier-Stokes Equation:

$$\rho \left( \frac{\partial \mathbf{v}}{\partial t} + \mathbf{v} \cdot abla \mathbf{v} \right) = - abla p + \mu abla^2 \mathbf{v} + \rho \mathbf{g}$$

• Species Transport Equation:

$$\frac{\partial C_i}{\partial t} + \mathbf{v} \cdot abla C_i = D_i abla^2 C_i + R_i$$

Where:

• $\rho$ = fluid density
• $\mathbf{v}$ = velocity vector
• $p$ = pressure
• $\mu$ = dynamic viscosity
• $C_i$ = concentration of species $i$
• $D_i$ = diffusion coefficient of species $i$
• $R_i$ = reaction rate term

• Boundary Layer:
• Stagnant gas layer above substrate
• Thickness $\delta$ depends on flow conditions:

$$\delta \propto \sqrt{\frac{ u x}{u_\infty}}$$

Where:

• $

u$ = kinematic viscosity

• $x$ = distance from leading edge
• $u_\infty$ = free stream velocity

3.2 Surface Kinetics

• Adsorption Process:
• Physisorption (weak van der Waals forces)
• Chemisorption (chemical bonding)

• Langmuir Adsorption Isotherm:

$$\theta = \frac{K \cdot P}{1 + K \cdot P}$$

Where:

• $\theta$ = fractional surface coverage
• $K$ = equilibrium constant
• $P$ = partial pressure

• Surface Diffusion:

$$D_s = D_0 \exp\left(-\frac{E_d}{k_B T}\right)$$

Where:

• $D_s$ = surface diffusion coefficient
• $D_0$ = pre-exponential factor
• $E_d$ = diffusion activation energy
• $k_B$ = Boltzmann constant ($1.38 \times 10^{-23}$ J/K)
• $T$ = absolute temperature

3.3 Crystal Growth Mechanisms

• Step-Flow Growth (BCF Theory):
• Atoms attach at step edges
• Steps advance across terraces
• Dominant at high temperatures

• 2D Nucleation:
• New layers nucleate on terraces
• Occurs when step density is low
• Creates rougher surfaces

• Terrace-Ledge-Kink (TLK) Model:
• Terrace: flat regions between steps
• Ledge: step edges
• Kink: incorporation sites at step edges

4. Mathematical Framework

4.1 Growth Rate Models

4.1.1 Reaction-Limited Regime

At lower temperatures, surface reaction kinetics dominate:

$$G = k_s \cdot C_s$$

Where the rate constant follows Arrhenius behavior:

$$k_s = k_0 \exp\left(-\frac{E_a}{k_B T}\right)$$

Parameters:

• $G$ = growth rate (nm/min or μm/hr)
• $k_s$ = surface reaction rate constant
• $C_s$ = surface concentration
• $k_0$ = pre-exponential factor
• $E_a$ = activation energy

4.1.2 Mass-Transport Limited Regime

At higher temperatures, diffusion through the boundary layer limits growth:

$$G = \frac{h_g}{N_s} \cdot (C_g - C_s)$$

Where:

$$h_g = \frac{D}{\delta}$$

Parameters:

• $h_g$ = mass transfer coefficient
• $N_s$ = atomic density of solid ($\sim 5 \times 10^{22}$ atoms/cm³ for Si)
• $C_g$ = gas phase concentration
• $D$ = gas phase diffusivity
• $\delta$ = boundary layer thickness

4.1.3 Combined Model (Grove Model)

For the general case combining both regimes:

$$G = \frac{h_g \cdot k_s}{N_s (h_g + k_s)} \cdot C_g$$

Or equivalently:

$$\frac{1}{G} = \frac{N_s}{k_s \cdot C_g} + \frac{N_s}{h_g \cdot C_g}$$

4.2 Strain in Heteroepitaxy

4.2.1 Lattice Mismatch

$$f = \frac{a_s - a_f}{a_f}$$

Where:

• $f$ = lattice mismatch (dimensionless)
• $a_s$ = substrate lattice constant
• $a_f$ = film lattice constant (relaxed)

Example Values:

4.2.2 In-Plane Strain

For a coherently strained film:

$$\epsilon_{\parallel} = \frac{a_s - a_f}{a_f} = f$$

The out-of-plane strain (for cubic materials):

$$\epsilon_{\perp} = -\frac{2 u}{1- u} \epsilon_{\parallel}$$

Where $ u$ = Poisson's ratio

4.2.3 Critical Thickness (Matthews-Blakeslee)

The critical thickness above which misfit dislocations form:

$$h_c = \frac{b}{8\pi f (1+ u)} \left[ \ln\left(\frac{h_c}{b}\right) + 1 \right]$$

Where:

• $h_c$ = critical thickness
• $b$ = Burgers vector magnitude ($\approx \frac{a}{\sqrt{2}}$ for 60° dislocations)
• $f$ = lattice mismatch
• $

u$ = Poisson's ratio

Approximate Solution:

For small mismatch:

$$h_c \approx \frac{b}{8\pi |f|}$$

4.3 Dopant Incorporation

4.3.1 Segregation Model

$$C_{film} = \frac{C_{gas}}{1 + k_{seg} \cdot (G/G_0)}$$

Where:

• $C_{film}$ = dopant concentration in film
• $C_{gas}$ = dopant concentration in gas phase
• $k_{seg}$ = segregation coefficient
• $G$ = growth rate
• $G_0$ = reference growth rate

4.3.2 Dopant Profile with Segregation

The surface concentration evolves as:

$$C_s(t) = C_s^{eq} + (C_s(0) - C_s^{eq}) \exp\left(-\frac{G \cdot t}{\lambda}\right)$$

Where:

• $\lambda$ = segregation length
• $C_s^{eq}$ = equilibrium surface concentration

5. Modeling Approaches

5.1 Continuum Models

• Scope:
• Reactor-scale simulations
• Temperature and flow field prediction
• Species concentration profiles

• Methods:
• Computational Fluid Dynamics (CFD)
• Finite Element Method (FEM)
• Finite Volume Method (FVM)

• Governing Physics:
• Coupled heat, mass, and momentum transfer
• Homogeneous and heterogeneous reactions
• Radiation heat transfer

5.2 Feature-Scale Models

• Applications:
• Selective epitaxial growth (SEG)
• Trench filling
• Facet evolution

• Key Phenomena:
• Local loading effects:

$$G_{local} = G_0 \cdot \left(1 - \alpha \cdot \frac{A_{exposed}}{A_{total}}\right)$$

• Orientation-dependent growth rates:

$$\frac{G_{(110)}}{G_{(100)}} \approx 1.5 - 2.0$$

• Methods:
• Level set methods
• String methods
• Cellular automata

5.3 Atomistic Models

5.3.1 Kinetic Monte Carlo (KMC)

• Process Events:
• Adsorption: rate $\propto P \cdot \exp(-E_{ads}/k_BT)$
• Surface diffusion: rate $\propto \exp(-E_{diff}/k_BT)$
• Desorption: rate $\propto \exp(-E_{des}/k_BT)$
• Incorporation: rate $\propto \exp(-E_{inc}/k_BT)$

• Master Equation:

$$\frac{dP_i}{dt} = \sum_j \left( W_{ji} P_j - W_{ij} P_i \right)$$

Where:

• $P_i$ = probability of state $i$
• $W_{ij}$ = transition rate from state $i$ to $j$

5.3.2 Molecular Dynamics (MD)

• Newton's Equations:

$$m_i \frac{d^2 \mathbf{r}_i}{dt^2} = - abla_i U(\mathbf{r}_1, \mathbf{r}_2, ..., \mathbf{r}_N)$$

• Interatomic Potentials:
• Tersoff potential (Si, C, Ge)
• Stillinger-Weber potential (Si)
• MEAM (metals and alloys)

5.3.3 Ab Initio / DFT

• Kohn-Sham Equations:

$$\left[ -\frac{\hbar^2}{2m} abla^2 + V_{eff}(\mathbf{r}) \right] \psi_i(\mathbf{r}) = \epsilon_i \psi_i(\mathbf{r})$$

• Applications:
• Surface energies
• Reaction barriers
• Adsorption energies
• Electronic structure

6. Specific Modeling Challenges

6.1 SiGe Epitaxy

• Composition Control:

$$x_{Ge} = \frac{R_{Ge}}{R_{Si} + R_{Ge}}$$

Where $R_{Si}$ and $R_{Ge}$ are partial growth rates

• Strain Engineering:
• Compressive strain in SiGe on Si
• Enhances hole mobility
• Critical thickness depends on Ge content:

$$h_c(x) \approx \frac{0.5}{0.042 \cdot x} \text{ nm}$$

6.2 Selective Epitaxy

• Growth Selectivity:
• Deposition only on exposed silicon
• HCl addition for selectivity enhancement

• Selectivity Condition:

$$\frac{\text{Growth on Si}}{\text{Growth on SiO}_2} > 100:1$$

• Loading Effects:
• Pattern-dependent growth rate
• Faceting at mask edges

6.3 III-V on Silicon

• Major Challenges:
• Large lattice mismatch (4-8%)
• Thermal expansion mismatch
• Anti-phase domain boundaries (APDs)
• High threading dislocation density

• Mitigation Strategies:
• Aspect ratio trapping (ART)
• Graded buffer layers
• Selective area growth
• Dislocation filtering

7. Applications and Tools

7.1 Industrial Applications

7.2 Simulation Software

• Reactor-Scale CFD:
• ANSYS Fluent
• COMSOL Multiphysics
• OpenFOAM

• TCAD Process Simulation:
• Synopsys Sentaurus Process
• Silvaco Victory Process
• Lumerical (for optoelectronics)

• Atomistic Simulation:
• LAMMPS (MD)
• VASP, Quantum ESPRESSO (DFT)
• Custom KMC codes

7.3 Key Metrics for Process Development

• Uniformity:

$$\text{Uniformity} = \frac{t_{max} - t_{min}}{2 \cdot t_{avg}} \times 100\%$$

• Defect Density:
• Threading dislocations: target $< 10^6$ cm$^{-2}$
• Stacking faults: target $< 10^3$ cm$^{-2}$

• Profile Abruptness:
• Dopant transition width $< 3$ nm/decade

8. Emerging Directions

8.1 Machine Learning Integration

• Applications:
• Surrogate models for process optimization
• Real-time virtual metrology
• Defect classification
• Recipe optimization

• Model Types:
• Neural networks for growth rate prediction
• Gaussian process regression for uncertainty quantification
• Reinforcement learning for process control

8.2 Multi-Scale Modeling

• Hierarchical Approach:

Ab Initio (DFT)
    ↓ Reaction rates, energies
Kinetic Monte Carlo
    ↓ Surface kinetics, morphology
Feature-Scale Models
    ↓ Local growth behavior
Reactor-Scale CFD
    ↓ Process conditions
Device Simulation

8.3 Digital Twins

• Components:
• Real-time sensor data integration
• Physics-based + ML hybrid models
• Predictive maintenance
• Closed-loop process control

8.4 New Material Systems

• 2D Materials:
• Graphene via CVD
• Transition metal dichalcogenides (TMDs)
• Van der Waals epitaxy

• Ultra-Wide Bandgap:
• $\beta$-Ga$_2$O$_3$ ($E_g \approx 4.8$ eV)
• Diamond ($E_g \approx 5.5$ eV)
• AlN ($E_g \approx 6.2$ eV)

Common Constants and Conversions