Comprehensive Engineering User Guide & Operational Workflow
A practical reference manual outlining the complete 10-step process engineering workflow: recipe formulation, multi-nozzle allocation, parameter sweeps, result interpretation, and experimental qualification on 300mm single-wafer wet tools.
1. The 10-Step Simulation Workflow
To achieve reliable, reproducible results when evaluating semiconductor wet processing recipes, process engineers should follow this structured 10-step protocol:
- "Does increasing spin speed from 800 to 1500 RPM induce edge dewetting or thin the film below the safety margin?"
- "Can adding a mid-radius nozzle at $r = 75\,\text{mm}$ compensate for the $r^{-2/3}$ centrifugal film depletion?"
- "What is the minimum DIW rinse flow rate required to maintain continuous substrate wetting across all dies?"
- Wafer Spin Speed (RPM): Range: 100 to 3000 RPM (default: 800โ1200 RPM for spin-rinse/clean).
- Process Temperature (ยฐC): Range: 10ยฐC to 180ยฐC. Temperature updates fluid dynamic viscosity via the Andrade thermal model and modifies surface tension.
- Chemical Species: Select from 8 semiconductor chemicals: DIW, SC1 ($NH_4OH/H_2O_2$), SC2 ($HCl/H_2O_2$), DHF, SPM (Piranha), IPA, BHF, or DSP+.
- Nozzle Coordinates $(x, y)$: Place nozzles at center ($r \approx 0\,\text{mm}$), mid-radius ($r \approx 50-90\,\text{mm}$), or outer radius ($r \approx 100-140\,\text{mm}$). You can also click and drag nozzle icons directly inside Card 4 (Wafer Side View).
- Individual Flow Rates ($Q_i$): Set volumetric discharge rates in mL/min or mL/s. Enable or disable individual nozzle arms to simulate single-arm or multi-arm configurations.
- Discharge Footprint Width: Adjust spray/stream dispersion width on the substrate surface.
- Solver Status: Should display green โ RUNNING.
- Frame Rate & Timestep ($\Delta t$): Confirms real-time temporal integration (typically $\Delta t = 5\,\text{ms}$, 60 FPS).
- Convergence: Allow 2 to 5 seconds of simulated time for the liquid film to achieve hydraulic steady state.
- Center Thickness ($h_0$): Typically $200 - 800\,\mu\text{m}$. Excessive thickness indicates stagnation puddling.
- Edge Thickness ($h_{150}$): Liquid film boundary layer at the 300mm wafer bevel. Values below $15\,\mu\text{m}$ signal dry spot risk.
- Thickness Range ($\Delta h = h_{\max} - h_{\min}$): Lower range indicates uniform distribution.
- Wafer Uniformity (%): Evaluated via standard half-range metric: $\text{Uniformity} = (1 - \Delta h / 2h_{\text{avg}}) \times 100\%$. Production targets usually exceed 85%.
- Trace the radial profile curve from $r = 0\,\text{mm}$ (wafer center) to $r = 150\,\text{mm}$ (bevel edge).
- Verify whether mid-radius nozzle arms successfully eliminate local profile dips.
- Check the theoretical Emslie-Bonner-Peck reference dashed curve to assess how closely current dispense conditions follow ideal $r^{-2/3}$ scaling.
- Click "ํ์ฌ ์ผ์ด์ค ์ ์ฅ" (Save Current Case) to snapshot baseline metrics into browser LocalStorage.
- Modify a parameter (e.g. increase RPM by 300 or shift Nozzle 2 outward by 20mm) and run again.
- Compare uniformity, dewetting risk score, and thickness range side-by-side in the comparison table.
- Specify your constraint: Constant Total Chemical Flow Rate (e.g. $2500\,\text{mL/min}$).
- Select optimization objective: MAXIMIZE FILM UNIFORMITY, MINIMIZE DEWETTING RISK, or BALANCED OPTIMIZATION.
- Click "๐ง Run Flow Ratio Sweep" to evaluate candidate distributions and click "โ Apply Best Flow Ratio" to instantly adopt the top configuration.
- Review the dynamic Korean/English engineering conclusions generated automatically from active recipe physics.
- Click "๐ ๋ฆฌํฌํธ ์ถ๋ ฅ" (Export Engineering Report) or "๐พ JSON / Markdown Export" to export a structured technical report containing recipe parameters, boundary metrics, and optimization findings.
- Program the candidate recipe parameters (RPM, nozzle arm coordinates, flow ratios, chemical temperature) into the physical wet station controller.
- Perform test runs with blanket monitor wafers (e.g. thermal oxide or silicon nitride film thickness measurement via ellipsometry/reflectometry).
- Correlate experimental etch rate or cleaning uniformity with the simulated film thickness profile.
2. Practical Application Studies
Objective: Quantify the effect of increasing spin speed from 500 RPM to 1500 RPM on liquid film thickness and chemical consumption efficiency during an etching step.
Theoretical Context: Centrifugal body force accelerates as $F_{\text{cent}} \propto \omega^2 r$. According to the Emslie-Bonner-Peck scaling, steady-state film thickness scales as $h \propto \omega^{-2/3}$, while radial convective velocity scales as $u_r \propto \omega^{2/3}$.
Simulation Observation:
- At 500 RPM: Wafer center thickness $h_0 \approx 520\,\mu\text{m}$, edge thickness $h_{150} \approx 62\,\mu\text{m}$. Chemical boundary layer is thick and well-stabilized against dewetting, but chemical residence time is prolonged.
- At 1000 RPM: Center thickness drops to $\approx 328\,\mu\text{m}$ ($-36.9\%$), edge thickness drops to $\approx 39\,\mu\text{m}$. Uniformity improves moderately.
- At 1800 RPM: Edge thickness drops to $\approx 26\,\mu\text{m}$. Radial liquid velocity accelerates dramatically, causing localized thinning near the bevel that requires auxiliary outer nozzle replenishment.
Objective: Compare chemical distribution uniformity between a traditional single center nozzle and a 2-nozzle distributed arrangement under an identical total flow constraint of $2400\,\text{mL/min}$.
| Configuration | Nozzle Locations | Flow Allocation | Edge Thickness ($h_{150}$) | Wafer Uniformity (%) | Dewetting Risk |
|---|---|---|---|---|---|
| Single Center Nozzle | N1: $r = 0\,\text{mm}$ | N1: $2400\,\text{mL/min}$ (100%) | $34.2\,\mu\text{m}$ | $64.8\%$ | Medium (0.42) |
| Dual Distributed Nozzle | N1: $r = 10\,\text{mm}$, N2: $r = 75\,\text{mm}$ | N1: $1400\,\text{mL/min}$, N2: $1000\,\text{mL/min}$ | $58.7\,\mu\text{m}$ | $88.4\%$ | Low (0.16) |
Engineering Insight: Reallocating $42\%$ of the chemical flux to a mid-radius nozzle directly counters the natural $r^{-2/3}$ thinning, boosting wafer-wide uniformity by $+23.6\%$ and significantly mitigating edge starvation without increasing total chemical volume.
Objective: Evaluate the hydrodynamic consequences of raising process temperature from 25ยฐC to 65ยฐC during hot chemical cleaning.
Physics Mechanism: Water dynamic viscosity drops from $\approx 0.89\,\text{mPa}\cdot\text{s}$ at 25ยฐC to $\approx 0.43\,\text{mPa}\cdot\text{s}$ at 65ยฐC according to Andrade's law. Because viscous resistance drops by over $50\%$, liquid flings off the wafer edge much more rapidly.
Simulation Guidance: When operating at elevated temperatures, engineers must increase dispense flow rates by approximately $25-35\%$ to maintain an equivalent boundary layer thickness and prevent hot-spot evaporation dewetting.
3. How to Interpret Simulation Outputs
The simulator provides multiple synchronized display modes to analyze process behavior:
Thickness Deviation Mode
Visualizes deviations from mean wafer thickness ($\Delta h = h(x, y) - \bar{h}$). Blue regions denote localized liquid starvation or excessive thinning; orange/red zones denote puddle accumulation beneath dispense streams.
Marangoni Vector Field Overlay
Overlays tangential shear arrows driven by surface tension gradients ($\tau_M = \nabla \sigma$). Arrows point toward higher surface tension zones. Convergence of arrows signifies liquid accumulation, while diverging vectors indicate thinning or impending film rupture.
Dewetting Risk Shader
Calculates the dimensionless Dewetting Risk Index $i_{\text{Dewet}} \in [0, 1]$ across the wafer grid:
- $i_{\text{Dewet}} < 0.30$ (Normal / Green): Continuous, stable fluid boundary layer.
- $0.30 \le i_{\text{Dewet}} < 0.65$ (Moderate / Yellow): Noticeable boundary layer thinning; inspect nozzle alignment.
- $i_{\text{Dewet}} \ge 0.65$ (Critical / Red): Immediate danger of dry spot formation or dewetting defect. Immediate flow increase or RPM reduction recommended.
4. Operational Scope & Model Limitations
This simulator is designed specifically for depth-integrated thin-film screening. It is not an alternative to empirical wafer tool qualification or multi-phase 3D CFD solvers.
Key boundary assumptions include:
- Laminar Flow Regime: Assumes film Reynolds number $Re_h < 100$. Highly turbulent high-pressure jetting is outside the model scope.
- No Interfacial Wave Breakup: Explicit droplet splashing, mist formation, and hydraulic jumps are approximated by continuous source footprint equations.
- Chemical Properties: Dynamic properties are evaluated based on pure/standard mixtures; in-situ etching reaction products or surfactant decomposition are not tracked.
- Experimental Validation Required: All optimization configurations must be confirmed on production equipment before applying recipes to production lots.
5. Engineering Recipe Pre-Flight Checklist
Before finalizing a single-wafer recipe, verify the following engineering criteria:
| Check Item | Recommended Engineering Target | Status / Action |
|---|---|---|
| Wafer Uniformity | > 85% half-range uniformity | Adjust mid-radius nozzle position if < 85% |
| Edge Thickness ($h_{150}$) | > 25 ยตm minimum margin | Increase outer nozzle flow if edge drops < 20 ยตm |
| Center Thickness ($h_0$) | < 650 ยตm stagnation cap | Shift center nozzle slightly off-axis ($r \approx 10-15\,\text{mm}$) |
| Dewetting Risk Index | < 0.40 across entire surface | Check for opposing Marangoni gradients during chemical transitions |
| Total Flow Rate | Within tool dispense pump capacity | Validate pump line delivery pressures |