Poisson Statistics

Semiconductor Manufacturing Process: Poisson Statistics & Mathematical Modeling

1. Introduction: Why Poisson Statistics?

Semiconductor defects satisfy the classical Poisson conditions:

Rare events — Defects are sparse relative to the total chip area
Independence — Defect occurrences are approximately independent
Homogeneity — Within local regions, defect rates are constant
No simultaneity — At infinitesimal scales, simultaneous defects have zero probability

1.1 The Poisson Probability Mass Function

The probability of observing exactly $$k$$ defects:

P(X = k) = \frac{\lambda^k e^{-\lambda}}{k!}

where the expected number of defects is:

\lambda = D_0 \cdot A

Parameter definitions:

$$D_0$$ — Defect density (defects per unit area, typically defects/cm²)
$$A$$ — Chip area (cm²)
$\lambda$ — Mean number of defects per chip

1.2 Key Statistical Properties

Property	Formula
Mean	$E[X] = \lambda$
Variance	$\text{Var}(X) = \lambda$
Variance-to-Mean Ratio	$\frac{\text{Var}(X)}{E[X]} = 1$

> Note: The equality of mean and variance (equidispersion) is a signature property of the Poisson distribution. Real semiconductor data often shows overdispersion (variance > mean), motivating compound models.

2. Fundamental Yield Equation

2.1 The Seeds Model (Simple Poisson)

A chip is functional if and only if it has zero killer defects. Under Poisson assumptions:

\boxed{Y = P(X = 0) = e^{-D_0 A}}

Derivation:

P(X = 0) = \frac{\lambda^0 e^{-\lambda}}{0!} = e^{-\lambda} = e^{-D_0 A}

2.2 Limitations of Simple Poisson

Assumes uniform defect density across the wafer (unrealistic)
Does not account for clustering of defects
Consistently underestimates yield for large chips
Ignores wafer-to-wafer and lot-to-lot variation

3. Compound Poisson Models

3.1 The Negative Binomial Approach

Model the defect density $$D_0$$ as a random variable with Gamma distribution:

D_0 \sim \text{Gamma}\left(\alpha, \frac{\alpha}{\bar{D}}\right)

Gamma probability density function:

f(D_0) = \frac{(\alpha/\bar{D})^\alpha}{\Gamma(\alpha)} D_0^{\alpha-1} e^{-\alpha D_0/\bar{D}}

where:

$\bar{D}$ — Mean defect density
$\alpha$ — Clustering parameter (shape parameter)

3.2 Resulting Yield Model

When defect density is Gamma-distributed, the defect count follows a Negative Binomial distribution, yielding:

\boxed{Y = \left(1 + \frac{D_0 A}{\alpha}\right)^{-\alpha}}

3.3 Physical Interpretation of Clustering Parameter $\alpha$

$\alpha$ Value	Physical Interpretation
$\alpha \to \infty$	Uniform defects — recovers simple Poisson model
$\alpha \approx 1-5$	Typical semiconductor clustering
$\alpha \to 0$	Extreme clustering — defects occur in tight groups

3.4 Overdispersion

The variance-to-mean ratio for the Negative Binomial:

\frac{\text{Var}(X)}{E[X]} = 1 + \frac{\bar{D}A}{\alpha} > 1

This overdispersion (ratio > 1) matches empirical observations in semiconductor manufacturing.

4. Classical Yield Models

4.1 Comparison Table

Model	Yield Formula	Assumed Density Distribution
Seeds (Poisson)	$Y = e^{-D_0 A}$	Delta function (uniform)
Murphy	$Y = \left(\frac{1 - e^{-D_0 A}}{D_0 A}\right)^2$	Triangular
Negative Binomial	$Y = \left(1 + \frac{D_0 A}{\alpha}\right)^{-\alpha}$	Gamma
Moore	$Y = e^{-\sqrt{D_0 A}}$	Empirical
Bose-Einstein	$Y = \frac{1}{1 + D_0 A}$	Exponential

4.2 Murphy's Yield Model

Assumes triangular distribution of defect densities:

Y_{\text{Murphy}} = \left(\frac{1 - e^{-D_0 A}}{D_0 A}\right)^2

Taylor expansion for small $$D_0 A$$ :

Y_{\text{Murphy}} \approx 1 - \frac{(D_0 A)^2}{12} + O((D_0 A)^4)

4.3 Limiting Behavior

As $D_0 A \to 0$ (low defect density):

\lim_{D_0 A \to 0} Y = 1 \quad \text{(all models)}

As $D_0 A \to \infty$ (high defect density):

\lim_{D_0 A \to \infty} Y = 0 \quad \text{(all models)}

5. Critical Area Analysis

5.1 Definition

Not all chip area is equally vulnerable. Critical area $$A_c$$ is the region where a defect of size $$d$$ causes circuit failure.

A_c(d) = \int_{\text{layout}} \mathbf{1}\left[\text{defect at } (x,y) \text{ with size } d \text{ causes failure}\right] \, dx \, dy

5.2 Critical Area for Shorts

For two parallel conductors with:

Length: $$L$$
Spacing: $$S$$

A_c^{\text{short}}(d) = \begin{cases} 2L(d - S) & \text{if } d > S \\ 0 & \text{if } d \leq S \end{cases}

5.3 Critical Area for Opens

For a conductor with:

Width: $$W$$
Length: $$L$$

A_c^{\text{open}}(d) = \begin{cases} L(d - W) & \text{if } d > W \\ 0 & \text{if } d \leq W \end{cases}

5.4 Total Critical Area

Integrate over the defect size distribution $$f(d)$$ :

A_c = \int_0^\infty A_c(d) \cdot f(d) \, dd

5.5 Defect Size Distribution

Typically modeled as power-law:

f(d) = C \cdot d^{-p} \quad \text{for } d \geq d_{\min}

Typical values:

Exponent: $p \approx 2-4$
Normalization constant: $C = (p-1) \cdot d_{\min}^{p-1}$

Alternative: Log-normal distribution (common for particle contamination):

f(d) = \frac{1}{d \sigma \sqrt{2\pi}} \exp\left(-\frac{(\ln d - \mu)^2}{2\sigma^2}\right)

6. Multi-Layer Yield Modeling

6.1 Modern IC Structure

Modern integrated circuits have 10-15+ metal layers. Each layer $$i$$ has:

Defect density: $$D_i$$
Critical area: $A_{c,i}$
Clustering parameter: $\alpha_i$ (for Negative Binomial)

6.2 Poisson Multi-Layer Yield

Y_{\text{total}} = \prod_{i=1}^{n} Y_i = \prod_{i=1}^{n} e^{-D_i A_{c,i}}

Simplified form:

\boxed{Y_{\text{total}} = \exp\left(-\sum_{i=1}^{n} D_i A_{c,i}\right)}

6.3 Negative Binomial Multi-Layer Yield

\boxed{Y_{\text{total}} = \prod_{i=1}^{n} \left(1 + \frac{D_i A_{c,i}}{\alpha_i}\right)^{-\alpha_i}}

6.4 Log-Yield Decomposition

Taking logarithms for analysis:

\ln Y_{\text{total}} = -\sum_{i=1}^{n} D_i A_{c,i} \quad \text{(Poisson)}

\ln Y_{\text{total}} = -\sum_{i=1}^{n} \alpha_i \ln\left(1 + \frac{D_i A_{c,i}}{\alpha_i}\right) \quad \text{(Negative Binomial)}

7. Spatial Point Process Formulation

7.1 Inhomogeneous Poisson Process

Intensity function $\lambda(x, y)$ varies spatially across the wafer:

P(k \text{ defects in region } R) = \frac{\Lambda(R)^k e^{-\Lambda(R)}}{k!}

where the integrated intensity is:

\Lambda(R) = \iint_R \lambda(x,y) \, dx \, dy

7.2 Cox Process (Doubly Stochastic)

The intensity $\lambda(x,y)$ is itself a random field:

\lambda(x,y) = \exp\left(\mu + Z(x,y)\right)

where:

$\mu$ — Baseline log-intensity
$$Z(x,y)$$ — Gaussian random field with spatial correlation function $\rho(h)$

Correlation structure:

\text{Cov}(Z(x_1, y_1), Z(x_2, y_2)) = \sigma^2 \rho(h)

where $h = \sqrt{(x_2-x_1)^2 + (y_2-y_1)^2}$

7.3 Neyman Type A (Cluster Process)

Models defects occurring in clusters:

1. Cluster centers: Poisson process with intensity $\lambda_c$
2. Defects per cluster: Poisson with mean $\mu$
3. Defect positions: Scattered around cluster center (e.g., isotropic Gaussian)

Probability generating function:

G(s) = \exp\left[\lambda_c A \left(e^{\mu(s-1)} - 1\right)\right]

Mean and variance:

E[N] = \lambda_c A \mu

\text{Var}(N) = \lambda_c A \mu (1 + \mu)

8. Statistical Estimation Methods

8.1 Maximum Likelihood Estimation

8.1.1 Data Structure

Given:

$$n$$ chips with areas $A_1, A_2, \ldots, A_n$
Binary outcomes $y_i \in \{0, 1\}$ (pass/fail)

8.1.2 Likelihood Function

\mathcal{L}(D_0, \alpha) = \prod_{i=1}^n Y_i^{y_i} (1 - Y_i)^{1-y_i}

where $Y_i = \left(1 + \frac{D_0 A_i}{\alpha}\right)^{-\alpha}$

8.1.3 Log-Likelihood

\ell(D_0, \alpha) = \sum_{i=1}^n \left[y_i \ln Y_i + (1-y_i) \ln(1-Y_i)\right]

8.1.4 Score Equations

\frac{\partial \ell}{\partial D_0} = 0, \quad \frac{\partial \ell}{\partial \alpha} = 0

> Note: Requires numerical optimization (Newton-Raphson, BFGS, or EM algorithm).

8.2 Bayesian Estimation

8.2.1 Prior Distribution

D_0 \sim \text{Gamma}(a, b)

\pi(D_0) = \frac{b^a}{\Gamma(a)} D_0^{a-1} e^{-b D_0}

8.2.2 Posterior Distribution

Given defect count $$k$$ on area $$A$$ :

D_0 \mid k \sim \text{Gamma}(a + k, b + A)

Posterior mean:

\hat{D}_0 = \frac{a + k}{b + A}

Posterior variance:

\text{Var}(D_0 \mid k) = \frac{a + k}{(b + A)^2}

8.2.3 Sequential Updating

Bayesian framework enables sequential learning:

\text{Prior}_n \xrightarrow{\text{data } k_n} \text{Posterior}_n = \text{Prior}_{n+1}

9. Statistical Process Control

9.1 c-Chart (Defect Counts)

For constant inspection area:

Center line: $\bar{c}$ (average defect count)
Upper Control Limit (UCL): $\bar{c} + 3\sqrt{\bar{c}}$
Lower Control Limit (LCL): $\max(0, \bar{c} - 3\sqrt{\bar{c}})$

9.2 u-Chart (Defects per Unit Area)

For variable inspection area $$n_i$$ :

u_i = \frac{c_i}{n_i}

Center line: $\bar{u}$
Control limits: $\bar{u} \pm 3\sqrt{\frac{\bar{u}}{n_i}}$

9.3 Overdispersion-Adjusted Charts

For clustered defects (Negative Binomial), inflate the variance:

\text{UCL} = \bar{c} + 3\sqrt{\bar{c}\left(1 + \frac{\bar{c}}{\alpha}\right)}

\text{LCL} = \max\left(0, \bar{c} - 3\sqrt{\bar{c}\left(1 + \frac{\bar{c}}{\alpha}\right)}\right)

9.4 CUSUM Chart

Cumulative sum for detecting small persistent shifts:

C_t^+ = \max(0, C_{t-1}^+ + (x_t - \mu_0 - K))

C_t^- = \max(0, C_{t-1}^- - (x_t - \mu_0 + K))

where:

$$K$$ — Slack value (typically $0.5\sigma$ )
Signal when $$C_t^+$$ or $$C_t^-$$ exceeds threshold $$H$$

10. EUV Lithography Stochastic Effects

10.1 Photon Shot Noise

At extreme ultraviolet wavelength (13.5 nm), photon shot noise becomes critical.

Number of photons absorbed in resist volume $$V$$ :

N \sim \text{Poisson}(\Phi \cdot \sigma \cdot V)

where:

$\Phi$ — Photon fluence (photons/area)
$\sigma$ — Absorption cross-section
$$V$$ — Resist volume

10.2 Line Edge Roughness (LER)

Stochastic photon absorption causes spatial variation in resist exposure:

\sigma_{\text{LER}} \propto \frac{1}{\sqrt{\Phi \cdot V}}

Critical Design Rule:

\text{LER}_{3\sigma} < 0.1 \times \text{CD}

where CD = Critical Dimension (feature size)

10.3 Stochastic Printing Failures

Probability of insufficient photons in a critical volume:

P(\text{failure}) = P(N < N_{\text{threshold}}) = \sum_{k=0}^{N_{\text{threshold}}-1} \frac{\lambda^k e^{-\lambda}}{k!}

where $\lambda = \Phi \sigma V$

11. Reliability and Latent Defects

11.1 Defect Classification

Not all defects cause immediate failure:

Killer defects: Cause immediate functional failure
Latent defects: May cause reliability failures over time

\lambda_{\text{total}} = \lambda_{\text{killer}} + \lambda_{\text{latent}}

11.2 Yield vs. Reliability

Initial Yield:

Y = e^{-\lambda_{\text{killer}} \cdot A}

Reliability Function:

R(t) = e^{-\lambda_{\text{latent}} \cdot A \cdot H(t)}

where $$H(t)$$ is the cumulative hazard function for latent defect activation.

11.3 Weibull Activation Model

H(t) = \left(\frac{t}{\eta}\right)^\beta

Parameters:

$\eta$ — Scale parameter (characteristic life)
$\beta$ — Shape parameter
$\beta < 1$ : Decreasing failure rate (infant mortality)
$\beta = 1$ : Constant failure rate (exponential)
$\beta > 1$ : Increasing failure rate (wear-out)

12. Complete Mathematical Framework

12.1 Hierarchical Model Structure

12.2 Summary of Key Equations

Concept	Equation
Poisson PMF	$P(X=k) = \frac{\lambda^k e^{-\lambda}}{k!}$
Simple Yield	$Y = e^{-D_0 A}$
Negative Binomial Yield	$Y = \left(1 + \frac{D_0 A}{\alpha}\right)^{-\alpha}$
Multi-Layer Yield	$Y = \prod_i \left(1 + \frac{D_i A_{c,i}}{\alpha_i}\right)^{-\alpha_i}$
Critical Area (shorts)	$A_c^{\text{short}}(d) = 2L(d-S)$ for $$d > S$$
Defect Size Distribution	$f(d) \propto d^{-p}$ , $p \approx 2-4$
Bayesian Posterior	$D_0 \mid k \sim \text{Gamma}(a+k, b+A)$
Control Limits	$\bar{c} \pm 3\sqrt{\bar{c}(1 + \bar{c}/\alpha)}$
LER Scaling	$\sigma_{\text{LER}} \propto (\Phi V)^{-1/2}$

12.3 Typical Parameter Values

Parameter	Typical Range	Units
Defect density $$D_0$$	0.01 - 1.0	defects/cm²
Clustering parameter $\alpha$	0.5 - 5	dimensionless
Defect size exponent $$p$$	2 - 4	dimensionless
Chip area $$A$$	1 - 800	mm²

Poisson Statistics

Take Poisson statistics further