Shockley Read Hall Srh Defect Recombination Kinetics

# Shockley-Read-Hall Recombination and Defect-Assisted Carrier Dynamics in Semiconductors

## 1. Introduction: Why Defects Matter at Sub-3nm Scale

In bulk semiconductors, radiative recombination (band-to-band photon emission) dominates at high carrier densities, while Shockley-Read-Hall (SRH) recombination via deep traps governs lifetime at equilibrium and low-injection conditions.

At sub-3nm gate lengths, several factors elevate the importance of trap-assisted recombination:
1. High electric fields accelerate trap capture and emission
2. Interface defects: Atomic-scale roughness at semiconductor-dielectric interfaces creates trap states
3. Finite lifetimes: Device operation (ns timescales) becomes comparable to carrier lifetimes (μs–ms in bulk)
4. Thermal instability: Temperature rise from Joule heating changes trap occupancy and capture rates

Understanding SRH recombination is essential for predicting leakage currents, subthreshold swing degradation, and device reliability.

## 2. Shockley-Read-Hall Recombination: The Mechanism

The SRH process involves a three-step mechanism:

### Step 1: Capture of Electron (Into Trap)
An electron from conduction band is captured by a trap state at energy $E_t$:

where:
- $c_n$ is the electron capture coefficient (typically 10⁻¹⁵–10⁻¹⁶ m³/s for Si)
- $n$ is electron concentration
- $N_T$ is trap density
- $n_t$ is number of occupied traps

### Step 2: Capture of Hole (Into Same Trap)
A hole from valence band is captured by the now-filled trap:

where $c_p$ is hole capture coefficient and $p$ is hole concentration.

### Step 3: Net Recombination
The net result: one electron and one hole have annihilated via the trap-assisted pathway, with energy released as phonons (heat) rather than photons.

## 3. The SRH Recombination Rate Formula

Under steady-state conditions, the net recombination rate through a single trap level is:

where:
- $n_i$ is intrinsic carrier concentration
- $n_1 = N_c \exp\left(\frac{E_t - E_c}{k_B T}\right)$ is electron concentration that would make $E_F = E_t$
- $p_1 = N_v \exp\left(\frac{E_v - E_t}{k_B T}\right)$ is analogous for holes

The denominator represents competition between capture and emission processes.

### Special Case: Mid-Gap Trap ($E_t = (E_c + E_v)/2 = E_i$)

When the trap is located mid-gap, $n_1 = p_1 = n_i$ and the SRH rate simplifies:

For $c_n \approx c_p \approx c$:

Maximum recombination occurs at mid-gap (hence why mid-gap defects are the most harmful).

## 4. Carrier Lifetime and Quality Factor

The carrier lifetime is inversely proportional to trap density and capture coefficient:

where $k_{\text{rad}}$ is radiative recombination coefficient (~10⁻¹⁸ m³/s in Si at 300 K).

In practice, for Si at room temperature:
- Bulk lifetime: $\tau_{\text{bulk}} \sim 100–1000$ μs (limited by bulk defect density ~10¹⁰ cm⁻³)
- Surface lifetime: $\tau_{\text{surf}} \sim 10–100$ ns (interface traps ~10¹² cm⁻² eV⁻¹)

The quality factor of a semiconductor surface is defined as:

Higher $Q$ (lower $N_T$) is desired for fast and reliable devices.

## 5. Auger Recombination: Carrier-Carrier Assisted

At very high carrier densities (>10¹⁸ cm⁻³), Auger recombination becomes important. One recombining carrier (e-h pair) transfers energy to a third carrier instead of emitting a photon:

The Auger coefficient $C_n$ (for electron-initiated) and $C_p$ (hole-initiated):

For Si:
- $C_n \sim 3 \times 10^{-30}$ cm⁶/s
- $C_p \sim 10^{-30}$ cm⁶/s

Auger recombination dominates when:

For $N_T = 10^{12}$ cm⁻³ (surface) and $c_n = 10^{-7}$ cm³/s, this occurs around $n \sim 10^{19}$ cm⁻³.

## 6. Deep Levels and Metastable Defects

Some defects have deep energy levels far from band edges (e.g., $E_t \sim 0.5$ eV from $E_c$ in Si). Examples:
- Gold (Au): $E_t = 0.54$ eV from $E_c$ (strong recombination center)
- Iron (Fe): $E_t = 0.39$ eV (used deliberately for carrier lifetime engineering in solar cells)
- Oxygen (O): $E_t = 0.17$ eV (shallow, weaker recombination)

Metastable defects exist in multiple charge states with different energies:

Each state has different capture coefficients and emission rates, leading to defect memory effects: the recombination rate depends on the defect's charge history.

## 7. Temperature Dependence of Capture Coefficients

Capture is fundamentally a collision process. The capture coefficient depends on:

where $\sigma_n$ is the capture cross-section (~10⁻¹⁴–10⁻¹⁶ cm² for typical defects).

As temperature increases:
- $v_{\text{th}} \propto \sqrt{T}$ (thermal velocity increases)
- $\sigma_n$ may decrease slightly (phonon scattering reduces effective cross-section)
- Net effect: $c_n$ increases ~30–50% per 100 K

This temperature dependence directly impacts device reliability: a 50 K temperature rise at the junction can reduce carrier lifetime by 50%, accelerating leakage.

## 8. Python Implementation: SRH Recombination Rate Calculator

import numpy as np
import matplotlib.pyplot as plt
from scipy.constants import k, e as q_e

def srh_recombination_rate(n, p, n_i, N_T, c_n, c_p, E_t, E_i, k_B, T):
    """
    Compute Shockley-Read-Hall recombination rate.
    
    U_SRH = c_n c_p N_T (np - n_i²) / [c_n(n+n_1) + c_p(p+p_1)]
    
    Parameters:
    -----------
    n, p : float (cm⁻³)
        Electron and hole concentrations
    n_i : float (cm⁻³)
        Intrinsic carrier concentration
    N_T : float (cm⁻³)
        Trap density
    c_n, c_p : float (cm³/s)
        Electron and hole capture coefficients
    E_t : float (eV)
        Trap energy level
    E_i : float (eV)
        Intrinsic level (midgap)
    k_B : float (eV/K)
        Boltzmann constant
    T : float (K)
        Temperature
    
    Returns:
    --------
    U : float (cm⁻³/s)
        Recombination rate
    """
    
    # Reference energies (set E_c = 0)
    dE_t = E_t  # Energy of trap above E_c
    dE_i = E_i  # Energy of intrinsic level above E_c
    
    # n_1 and p_1 (assuming known band structure)
    # For simplicity: use approximations
    # n_1 ≈ n_i * exp((E_t - E_i) / (k_B T))
    # p_1 ≈ n_i * exp((E_i - E_t) / (k_B T))
    
    if T > 0:
        n_1 = n_i * np.exp((E_t - E_i) / (k_B * T))
        p_1 = n_i * np.exp((E_i - E_t) / (k_B * T))
    else:
        n_1 = p_1 = 0
    
    # Numerator
    numerator = c_n * c_p * N_T * (n*p - n_i**2)
    
    # Denominator
    denominator = c_n * (n + n_1) + c_p * (p + p_1)
    
    if denominator == 0:
        return 0
    
    U = numerator / denominator
    
    return max(U, 0)  # Recombination rate is non-negative

def auger_recombination_rate(n, p, C_n, C_p):
    """
    Compute Auger recombination rate.
    
    U_Auger = (C_n n + C_p p) np
    
    Parameters:
    -----------
    n, p : float (cm⁻³)
    C_n, C_p : float (cm⁶/s)
        Auger coefficients
    
    Returns:
    --------
    U : float (cm⁻³/s)
    """
    return (C_n * n + C_p * p) * n * p

def radiative_recombination_rate(n, p, n_i, B):
    """
    Radiative recombination: U_rad = B(np - n_i²)
    """
    return B * (n*p - n_i**2)

def total_recombination_rate(n, p, n_i, N_T, c_n, c_p, E_t, E_i, 
                             C_n, C_p, B, T):
    """
    Total recombination (SRH + Auger + Radiative)
    """
    k_B_eV = 8.617e-5  # eV/K
    
    U_srh = srh_recombination_rate(n, p, n_i, N_T, c_n, c_p, E_t, E_i, k_B_eV, T)
    U_auger = auger_recombination_rate(n, p, C_n, C_p)
    U_rad = radiative_recombination_rate(n, p, n_i, B)
    
    return U_srh, U_auger, U_rad, U_srh + U_auger + U_rad

# Parameters (Si at 300 K)
T = 300  # Temperature
k_B_eV = 8.617e-5
n_i = 1e10  # cm⁻³ (Si intrinsic)
N_T = 1e12  # cm⁻³ (surface trap density)
c_n = 1e-7  # cm³/s (typical)
c_p = 1e-7  # cm³/s
C_n = 3e-30  # cm⁶/s (Auger)
C_p = 1e-30
B = 1e-15  # cm³/s (radiative)
E_t = 0.0  # eV (mid-gap trap, taking E_c = 0, E_v = -E_g)
E_i = -0.71  # eV (mid-gap in Si, E_g ≈ 1.42 eV)
E_g = 1.42

# Carrier concentration array (log scale)
n_array = np.logspace(12, 20, 200)  # cm⁻³

U_srh_array = []
U_auger_array = []
U_rad_array = []
U_total_array = []

for n in n_array:
    p = n  # Assume equal n and p (intrinsic excitation)
    U_s, U_a, U_r, U_t = total_recombination_rate(
        n, p, n_i, N_T, c_n, c_p, E_t, E_i, C_n, C_p, B, T
    )
    U_srh_array.append(U_s)
    U_auger_array.append(U_a)
    U_rad_array.append(U_r)
    U_total_array.append(U_t)

# Lifetime calculation
tau_srh = 1.0 / np.array([max(u, 1e-30) for u in U_srh_array])
tau_total = 1.0 / np.array([max(u, 1e-30) for u in U_total_array])

# Plotting
fig, axes = plt.subplots(2, 2, figsize=(13, 10))

# Panel A: Recombination rates
axes[0, 0].loglog(n_array, np.array(U_srh_array) + 1e-30, 'b-', linewidth=2.5, label='SRH')
axes[0, 0].loglog(n_array, np.array(U_auger_array) + 1e-30, 'r-', linewidth=2.5, label='Auger')
axes[0, 0].loglog(n_array, np.array(U_rad_array) + 1e-30, 'g-', linewidth=2.5, label='Radiative')
axes[0, 0].loglog(n_array, np.array(U_total_array) + 1e-30, 'k--', linewidth=2.5, label='Total')
axes[0, 0].set_xlabel('Carrier Concentration n=p (cm⁻³)', fontsize=11)
axes[0, 0].set_ylabel('Recombination Rate (cm⁻³/s)', fontsize=11)
axes[0, 0].set_title('Recombination Mechanisms (Si, T=300K)', fontsize=12, fontweight='bold')
axes[0, 0].legend(fontsize=10)
axes[0, 0].grid(True, alpha=0.3, which='both')

# Panel B: Carrier lifetime
axes[0, 1].loglog(n_array, tau_srh*1e9, 'b-', linewidth=2.5, label='SRH-only')
axes[0, 1].loglog(n_array, tau_total*1e9, 'k-', linewidth=2.5, label='Total')
axes[0, 1].axhline(y=1e-3, color='gray', linestyle=':', alpha=0.5, label='Sub-ns (device limit)')
axes[0, 1].set_xlabel('Carrier Concentration n=p (cm⁻³)', fontsize=11)
axes[0, 1].set_ylabel('Lifetime (ns)', fontsize=11)
axes[0, 1].set_title('Carrier Lifetime (Dominated by SRH)', fontsize=12, fontweight='bold')
axes[0, 1].legend(fontsize=10)
axes[0, 1].grid(True, alpha=0.3, which='both')

# Panel C: SRH vs trap energy
E_t_array = np.linspace(-0.71, 0.71, 100)  # Energy range (mid-gap at 0)
n_test = 1e17  # Fixed carrier concentration
p_test = 1e17
U_vs_Et = []

for Et in E_t_array:
    U_s, _, _, _ = total_recombination_rate(
        n_test, p_test, n_i, N_T, c_n, c_p, Et, E_i, C_n, C_p, B, T
    )
    U_vs_Et.append(U_s)

axes[1, 0].semilogy(E_t_array, np.array(U_vs_Et) + 1e-30, 'purple', linewidth=2.5)
axes[1, 0].axvline(x=E_i, color='k', linestyle='--', alpha=0.5, label='Mid-gap (max U)')
axes[1, 0].set_xlabel('Trap Energy E_t (eV)', fontsize=11)
axes[1, 0].set_ylabel('SRH Rate (cm⁻³/s)', fontsize=11)
axes[1, 0].set_title('SRH Rate vs. Trap Energy (n=p=10¹⁷ cm⁻³)', fontsize=12, fontweight='bold')
axes[1, 0].legend(fontsize=10)
axes[1, 0].grid(True, alpha=0.3, which='both')

# Panel D: Temperature dependence of lifetime
T_array = np.array([100, 200, 300, 400, 500])  # K
tau_vs_T = []

for T_test in T_array:
    U_s, U_a, U_r, U_t = total_recombination_rate(
        n_test, p_test, n_i, N_T, c_n, c_p, E_t, E_i, C_n, C_p, B, T_test
    )
    tau = 1.0 / max(U_t, 1e-30)
    tau_vs_T.append(tau*1e9)

axes[1, 1].plot(T_array, tau_vs_T, 'o-', linewidth=2.5, markersize=8, color='darkblue')
axes[1, 1].set_xlabel('Temperature (K)', fontsize=11)
axes[1, 1].set_ylabel('Lifetime (ns)', fontsize=11)
axes[1, 1].set_title('Temperature Dependence of Lifetime', fontsize=12, fontweight='bold')
axes[1, 1].grid(True, alpha=0.3)

plt.tight_layout()
plt.savefig('srh_recombination_analysis.png', dpi=150, bbox_inches='tight')
plt.show()

print("=== Shockley-Read-Hall Recombination Analysis Complete ===")
print(f"SRH-dominated region (n~10¹⁷ cm⁻³): τ = {tau_srh[np.argmin(np.abs(n_array - 1e17))]*1e9:.2f} ns")
print(f"Auger-dominated region (n~10¹⁹ cm⁻³): τ_Auger = {1.0/(auger_recombination_rate(1e19, 1e19, C_n, C_p)+1e-30)*1e9:.2f} ns")
print(f"Temperature increase from 300K to 400K: {tau_vs_T[2]/tau_vs_T[3]:.1f}× lifetime reduction")

## 9. Emission Rate and the Thermal Re-emission Processes

An electron trapped in a defect can thermally escape (be re-emitted) with rate:

Similarly for holes:

The emission-capture equilibrium determines trap occupancy at thermal equilibrium. Traps near the mid-gap have very low emission rates at room temperature—they stay filled or empty.

## 10. Interface Trap Density and Surface Recombination Velocity

At semiconductor-oxide interfaces (critical for sub-3nm MOSFETs), interface traps (Dit) with density ~10¹⁰–10¹² eV⁻¹ cm⁻² dominate recombination.

The surface recombination velocity:

Typical values:
- Clean Si (in vacuum): $v_s \sim 100–1000$ cm/s
- Si/SiO₂ interface (well-passivated): $v_s \sim 1–10$ cm/s
- Si/Al₂O₃ (poorly passivated): $v_s \sim 10⁴–10⁵$ cm/s

Passivation (H termination, post-metallization anneal) dramatically reduces Dit and $v_s$.

## 11. Defect-Assisted Leakage in Reverse Bias

Under reverse bias, the space-charge region widens. Generation currents through traps dominate:

Generation current increases linearly with depletion width and trap density, but exponentially with temperature via $n_i(T)$.

At room temperature (~300 K), $J_{\text{gen}}$ can exceed diffusion current by orders of magnitude in heavily trapped devices.

## 12. Reliability: Bias Temperature Instability (BTI)

Negative BTI (NBTI) in p-channel MOSFETs involves:
1. Hole injection into oxide under negative gate bias
2. Trapping in interface states ($N_{it}$) or oxide traps
3. Threshold voltage shift: $\Delta V_t \propto Q_{\text{trapped}} / C_{\text{ox}}$

The trapping/detrapping kinetics follow power-law time dependence:

Lifetime projection to 10-year device operation requires acceleration testing (higher T, higher V) and extrapolation—a major reliability challenge for sub-3nm technology.

## 13. Strategies for Reducing Trap-Assisted Recombination

1. Defect reduction: High-temperature anneals reduce trap density
2. Bandgap engineering: Wide-bandgap overlayers (SiC, diamond) suppress tunneling to traps
3. Passivation: Hydrogen termination, Al₂O₃ coating
4. Doping engineering: Pocket dopants reduce depletion width in critical regions
5. Cryogenic operation: Reduces emission rate (non-practical for most applications)

## 14. Measurement Techniques: Extracting Defect Parameters

• Deep Level Transient Spectroscopy (DLTS): Measure trap energy and capture cross-section
• Charge Pumping: Quantify interface trap density
• Impedance Spectroscopy: Extract Dit vs. energy
• Temperature-Dependent I-V: Reveals activation energies

## 15. Future Outlook: Defect-Tolerant Devices

Emerging strategy: instead of eliminating defects, design devices to be insensitive to them:
- Defect-tolerant ferroelectrics: Traps stabilize remnant polarization
- Self-healing oxides: Defects rearrange under bias to reduce leakage
- Machine learning prediction: Neural networks trained on DLTS data predict device lifetime