Role in QLT Fabrication
Why Ellipsometry Matters for Quantum Photonics
The ellipsometer measures polarization change of reflected light to determine thin film thickness and optical constants (refractive index n, extinction coefficient k). For QLT's quantum photonic chip, refractive index accuracy is mission-critical because the entire device operation depends on precise phase matching.
PHASE-MATCHING CONDITION FOR OPTICAL PARAMETRIC CONVERSION (OPC):
Δk = 2k_pump - k_signal - k_idler = 0
k = (2π/λ) × n_eff(λ, n_core, n_clad, geometry)
where n_eff depends on:
- n_core = f(n_SiN, n_As₂S₃) → MUST know these precisely
- n_clad = f(n_SiO₂) → MUST know this precisely
- waveguide geometry (width, height) → from profilometer/AFM
SENSITIVITY:
If As₂S₃ refractive index n deviates from design value of 2.43 by Δn = 0.02:
→ Phase-matching bandwidth collapses
→ OPC efficiency drops by > 10 dB
→ Quantum state generation FAILS
REQUIREMENT: Know n of As₂S₃ to ± 0.01 at 1550 nm
→ Only spectroscopic ellipsometry can provide this accuracy
Specific Measurements
Critical Wavelength Coverage Issue
The J.A. Woollam Alpha-SE only covers 380–900 nm. It CANNOT directly measure refractive index at 1550 nm.
For QLT, this matters because:
- SiN and SiO₂ are well-behaved dielectrics → Cauchy extrapolation from visible to 1550 nm gives accuracy ~± 0.01 (acceptable)
- As₂S₃ has absorption features near 800–1000 nm → Cauchy extrapolation is LESS reliable → accuracy degrades to ± 0.02–0.05 (marginal for phase matching)
Options:
1. Alpha-SE + Cauchy model → Good enough for SiN/SiO₂; marginal for As₂S₃ (± 0.02)
2. M-2000 (covers to 1690 nm) → Direct measurement at 1550 nm → accuracy ± 0.001 (definitive)
3. Validate Alpha-SE accuracy by measuring first As₂S₃ films on a university M-2000, then compare Cauchy extrapolation
Ellipsometry Physics for QLT Materials
ELLIPSOMETRY PRINCIPLE:
Linearly polarized light → Sample → Elliptically polarized light
Measure: Ψ (amplitude ratio) and Δ (phase difference)
tan(Ψ) × e^(iΔ) = r_p / r_s
where r_p, r_s = Fresnel reflection coefficients for p and s polarization
From Ψ and Δ at multiple wavelengths:
→ Fit optical model (Cauchy, Sellmeier, Tauc-Lorentz, Cody-Lorentz)
→ Extract: n(λ), k(λ), thickness (t)
→ For QLT: extract n at 1550 nm with ± 0.001 accuracy (M-2000)
MODEL SELECTION FOR QLT MATERIALS:
SiN (Si₃N₄): Cauchy model (transparent in visible-NIR)
SiO₂: Cauchy model (transparent everywhere)
As₂S₃: Tauc-Lorentz or Cody-Lorentz model (has absorption edge near 500 nm)
PVDF-TrFE: Cauchy model (transparent polymer)
Au: Drude model (metallic; free electron response)
Ti: Drude-Lorentz model (metal + interband transitions)
Process Integration
System Installation and Daily Operation
DAILY STARTUP:
1. Power on ellipsometer (M-2000 or Alpha-SE)
2. Wait 15 min for lamp and detector stabilization
3. Launch CompleteEASE software
4. Run system calibration:
├── Straight-through (no sample): verifies beam alignment
└── Reference wafer: verifies Ψ, Δ accuracy
5. System ready for measurements
DAILY SHUTDOWN:
1. Remove sample
2. Close CompleteEASE
3. Power down (or leave in standby ● extends lamp life)
First As₂S₃ Film Characterization (Critical Measurement)
FIRST As₂S₃ CHARACTERIZATION ● DETAILED PROCEDURE:
This is the most important ellipsometer measurement for QLT.
It validates the evaporation process and establishes the As₂S₃ optical model.
STEP 1: Prepare witness samples
├── During As₂S₃ evaporation (G3), place 2 bare Si witness wafers alongside chip
├── Witness sample 1: for ellipsometry (thickness + n,k)
└── Witness sample 2: backup / for Raman spectroscopy
STEP 2: Measure witness sample
├── Place witness on ellipsometer vacuum chuck
├── Measure at 65°, 70°, 75° (multi-angle for robust fit)
├── Spectral range: full (193–1690 nm for M-2000; 380–900 nm for Alpha-SE)
└── Acquire data (< 5 seconds total)
STEP 3: Build and fit optical model
├── Layer structure: As₂S₃ / SiO₂ (native, ~2 nm) / Si
├── As₂S₃ model: Start with Tauc-Lorentz
│ ├── Initial guesses: E_g = 2.4 eV, A = 100, E_0 = 4.5 eV, C = 2.5, ε₁(∞) = 3.0
│ └── Fit all parameters + thickness
├── Check MSE < 5
├── If MSE > 10:
│ ├── Add surface roughness layer (Bruggeman EMA, 50% As₂S₃ / 50% void)
│ ├── Try Cody-Lorentz model instead
│ └── Check for crystalline As₂S₃ optical constants in database
└── Once good fit achieved: SAVE MODEL as "QLT_As2S3_v1.mod"
STEP 4: Extract critical parameters
├── n @ 1550 nm: ______ (target: 2.43 ± 0.02)
├── k @ 1550 nm: ______ (target: < 0.0001)
├── E_g (bandgap): ______ (target: ~2.4 eV = 516 nm)
├── Thickness: ______ nm (target: 500 ± 5 nm)
└── Export n(λ), k(λ) as CSV for photonic simulation
STEP 5: Validate (if using Alpha-SE)
├── Send same witness sample to university with M-2000
├── Compare Alpha-SE Cauchy extrapolation to M-2000 direct measurement at 1550 nm
├── If Δn < 0.02: Alpha-SE is sufficient for routine QC
└── If Δn > 0.02: Must upgrade to M-2000 for production
Vendor Options & Pricing
New System Pricing
Used/Refurbished Market
Used market sources: CAE Online (caeonline.com), LabX (labx.com), Used-Line (used-line.com)
Vendor Contact Information
Our Budget Recommendation
PHASE 1 ● IMMEDIATE (thickness QC only):
- FilMetrics F20 (new): $7,000; lead time 1–2 weeks
- Measures film thickness instantly; no refractive index
- Adequate for PVDF-TrFE and SiO₂ thickness QC during initial process development
PHASE 2 ● STANDARD (visible ellipsometry + extrapolation):
- Woollam Alpha-SE (new): $40,000; lead time 4–8 weeks
- Measures n,k from 380–900 nm; extrapolates to 1550 nm via Cauchy/Sellmeier model
- Sufficient for SiN/SiO₂ characterization; marginal for As₂S₃
- CompleteEASE software is industry-standard
PHASE 3 ● PRODUCTION (direct 1550 nm measurement):
- Woollam M-2000 (new): $120,000; lead time 8–14 weeks
- Direct n measurement at 1550 nm ● definitive for phase-matching verification
- Required if Alpha-SE Cauchy extrapolation for As₂S₃ proves insufficient
Budget: $7,000 (FilMetrics immediate) + $40,000 (Alpha-SE later) = $47,000
Or: $120,000 (M-2000 if budget allows ● eliminates extrapolation uncertainty)