Defect-Tolerant Nanophotonic Film
A Tidy3D FDTD GUI Research Design for a College Application Portfolio
Project Type: Computational nanophotonics, metasurface robustness, optical simulationPrepared by: [Your Name]Date: September 2026Software: Tidy3D FDTD GUI, parameter sweeps, flux/field/diffraction monitors



.
Academic integrity noteThis document is written as a simulation-ready research report. It includes a complete Tidy3D design, analysis plan, target metrics, and portfolio structure. Any plots labeled as target or expected should be replaced with actual exported Tidy3D results after running the simulations. Do not present target curves as measured data.

Contents
1. Abstract
2. Research Objective and Hypothesis
3. Background and Significance
4. Device Concept
5. Tidy3D GUI Design Specification
6. Defect Model and Simulation Matrix
7. Measurements and Analysis Metrics
8. Expected Results and Portfolio Figures
9. Validation, Error Checks, and Limitations
10. Portfolio Presentation Strategy
11. Appendix A: GUI Build Checklist
12. Appendix B: Script-Object Pseudocode
13. References


1. Abstract
This project proposes and simulates a defect-tolerant nanophotonic film: a subwavelength all-dielectric metasurface made from titanium dioxide nanodisks on a glass substrate. The central question is whether a useful optical film can remain functional even when realistic fabrication defects are introduced, including missing pillars, radius variation, height variation, and positional jitter. The design is built for Tidy3D FDTD GUI using a periodic supercell so that both pristine and defective patterns can be tested under visible-light illumination. The intended application is a transparent or semi-transparent light-management coating for display covers, architectural glass, solar windows, and optical sensors, where low haze and high transmission matter more than an extremely narrow resonance. The project emphasizes robust physics instead of a fragile perfect-pattern design: the nanodisk array behaves like an engineered effective-index layer with broad Mie-type scattering control, so small local defects should perturb the spectrum only weakly. The proposed study outputs transmission, reflection, diffraction/haze proxy, near-field maps, and a quantitative Defect Sensitivity Index. This makes the project valuable for a college portfolio because it combines nanophotonics, computational electromagnetics, materials design, uncertainty analysis, and responsible scientific reporting.
2. Research Objective and Hypothesis
2.1 Research question
Can a subwavelength nanophotonic film designed in Tidy3D maintain high visible transmission and low haze after common fabrication defects are introduced?
2.2 Hypothesis
A film based on broad, low-Q all-dielectric scattering will be more defect-tolerant than a narrow high-Q resonance device. If the period remains below the visible diffraction threshold in the substrate and if the nanodisks are not tuned to an extremely sharp resonance, then moderate radius errors, small positional jitter, and even a small fraction of missing pillars should create only small changes in broadband transmission and reflection.


2.3 Portfolio-level success criteria
Criterion	Target for portfolio version	Why it matters
Average transmission, 450–750 nm	> 85% after 10% missing-pillar defects	Shows that the film remains optically useful.
Average reflection	< 8% across most of visible band	Demonstrates light-management behavior, not just scattering.
Haze proxy / non-zero diffraction power	< 3% for pristine and < 5% for 10% defect case	Makes the film plausible for display/window use.
Defect Sensitivity Index	< 0.06 for 10% defects	Quantifies robustness rather than relying on appearance.
Portfolio evidence	At least 4 exported field maps + 3 spectral plots + one parameter-sweep heat map	Shows technical depth and reproducibility.
3. Background and Significance
Nanophotonic films use structures smaller than the wavelength of light to control reflection, transmission, scattering, color, polarization, and field localization. A perfect periodic structure can produce beautiful optical effects, but real fabricated films contain imperfections. A stronger portfolio project therefore asks not only “Can I make an optical effect?” but also “Does the design still work when reality is imperfect?”
This project is designed for Tidy3D because the FDTD method directly solves the time-domain electromagnetic response of nanoscale structures. The Tidy3D GUI workflow is suitable for creating structures, assigning materials, setting sources, adding field/flux/diffraction monitors, running parameter sweeps, and visualizing fields. The GUI also supports parameterization and script objects, which are useful for creating repeated nanostructures and defect patterns. See References [1]–[5].
3.1 Why defect tolerance is impressive
Basic project	Research-grade upgrade
Simulate one perfect metasurface.	Simulate pristine plus defective supercells and compare performance statistically.
Show one field plot.	Show field maps, spectra, diffraction/haze proxy, and robustness score.
Tune for maximum resonance.	Trade off optical performance against manufacturability and defect tolerance.
Looks visually impressive.	Demonstrates engineering judgment and scientific maturity.
4. Device Concept
The proposed film is an all-dielectric TiO₂ nanodisk array placed on a glass substrate. The array period is chosen to be subwavelength in glass for most visible wavelengths, reducing unwanted diffraction orders. The nanodisk radius and height are chosen to create moderate broad scattering and impedance-matching behavior rather than a narrow fragile resonance.






4.1 Starting geometry
Parameter	Symbol	Starting value	Reason for value
Square-lattice period	P	0.28 µm	Below λ/n_glass at short visible wavelengths, reducing diffraction and haze.
Nanodisk radius	R	0.07 µm	Moderate fill fraction; avoids excessively strong or narrow resonance.
Nanodisk height	H	0.22 µm	Large enough for phase/scattering control but still fabricable.
Supercell size	N × N	6 × 6 cells	Large enough to include defect patterns while remaining GUI-manageable.
Film material	TiO₂	n ≈ 2.35 assumed	High-index, low-loss visible dielectric approximation.
Substrate	Glass/SiO₂	n ≈ 1.45 assumed	Common transparent substrate model.
Ambient medium	Air	n = 1.00	Incident visible light from air side.

5. Tidy3D GUI Design Specification



5.1 Coordinate system
Axis	Meaning	Simulation choice
x	Horizontal periodic direction	Supercell width = N·P = 1.68 µm for N = 6.
y	Horizontal periodic direction	Supercell width = N·P = 1.68 µm for N = 6.
z	Vertical propagation direction	Plane wave travels from air into glass, along -z.
5.2 Simulation domain and boundaries
Setting	Recommended value	Notes
x span	1.68 µm	Six periods of 0.28 µm. Increase to 8 × 8 for stronger statistical disorder if credits allow.
y span	1.68 µm	Same as x for square supercell.
z span	Approx. 2.2–2.8 µm	Includes air buffer, nanodisk height, glass substrate, and PML spacing.
x/y boundaries	Periodic for normal incidence; Bloch for angled incidence	Models an infinite repeated film.
z boundaries	PML	Absorbs outgoing reflected/transmitted waves.
Symmetry	Off for defective supercells	Defects usually break symmetry. Use symmetry only for pristine checks.
5.3 Source setup
Source parameter	Value	Reason
Source type	PlaneWave	Represents external illumination of a film.
Propagation direction	-z	Light enters from air and transmits into glass.
Polarization	Ex first; repeat with Ey	Checks polarization sensitivity.
Wavelength range	450–750 nm	Visible-band response.
Center wavelength	550 nm	Middle of visible range; useful for field snapshots.
Incidence angles	0°, 15°, 30° optional	Tests angular robustness for real use.
5.4 Monitor setup
Monitor	Location	Data collected	Portfolio use
Reflection FluxMonitor	Above source, below top PML	Reflected power R(λ)	Reflection spectrum.
Transmission FluxMonitor	Below substrate, above bottom PML	Transmitted power T(λ)	Transmission spectrum.
DiffractionMonitor	Transmission side	Power in allowed diffraction orders	Haze/diffraction proxy.
FieldMonitor xz	Vertical slice through nanodisks	|E|², Ex, Ez at 550 nm	Beautiful field confinement image.
FieldMonitor xy	Horizontal plane through disks	|E|² at disk mid-height	Shows local field perturbation near defects.
FieldTimeMonitor optional	One point below film	Time-domain decay check	Confirms simulation has settled.
5.5 Mesh and run-time settings
Setting	Starting value	Rationale
Grid	Automatic nonuniform mesh	Efficient for nanostructures.
Minimum steps per wavelength	25–30	Visible wavelengths and high-index TiO₂ require finer resolution.
Local override near disks	5–8 nm if affordable	Improves curved-edge and near-field accuracy.
Courant/stability	Default Tidy3D stable setting	Use solver defaults unless warning appears.
Run time	100–200 fs or auto shutoff	Enough for broadband spectra; increase if residual energy remains.
Shutoff	1e-5 or lower	Avoid stopping before fields decay.

6. Defect Model and Simulation Matrix
A strong portfolio version should not use a single defect case. It should test a matrix of realistic manufacturing deviations. Each case is compared with the pristine film using the same source, monitors, mesh, and wavelength sampling.
Case	Defect type	Value	What it tests
D0	Pristine	No defects	Baseline performance.
D1	Radius variation	σR = 5 nm	Mild lithography or etch-radius variation.
D2	Radius variation	σR = 10 nm	Moderate fabrication error.
D3	Height variation	σH = 10 nm	Etch-depth nonuniformity.
D4	Positional jitter	σx,y = 10 nm	Placement or pattern-transfer error.
D5	Missing nanodisks	5% randomly removed	Particle, mask, or processing failures.
D6	Missing nanodisks	10% randomly removed	Severe but plausible defect tolerance test.
D7	Combined defects	σR = 10 nm, σx,y = 10 nm, 5% missing	More realistic combined-disorder test.
6.1 Parameter sweep plan
Parameter	Sweep values	Purpose
Period P	0.24, 0.26, 0.28, 0.30 µm	Find subwavelength period that reduces haze while maintaining optical effect.
Radius R	0.055, 0.065, 0.070, 0.080, 0.090 µm	Optimize fill fraction and resonance strength.
Height H	0.16, 0.19, 0.22, 0.25 µm	Tune phase delay and scattering strength.
Missing fraction	0%, 5%, 10%, 20%	Quantify robustness to large defects.
Incident angle	0°, 15°, 30°	Test real-world angular performance.
7. Measurements and Analysis Metrics
The central portfolio goal is to convert field pictures into quantitative evidence. The following metrics should be extracted from exported Tidy3D monitor data.
7.1 Power metrics
Transmission and reflection are normalized by incident source power:
T(λ) = P_transmitted(λ) / P_incident(λ)
R(λ) = P_reflected(λ) / P_incident(λ)
A(λ) = 1 − T(λ) − R(λ)  (for nonabsorbing materials, A should be near zero; deviations suggest numerical loss or scattering outside the monitored region).
7.2 Defect Sensitivity Index
Use a single robustness number to compare defects:
DSI = mean over λ of |T_defect(λ) − T_pristine(λ)| / mean over λ of T_pristine(λ)
A smaller DSI means the film is less sensitive to the defect. For this portfolio project, DSI < 0.06 under 10% missing-pillar defects is a strong target.
7.3 Haze / diffraction proxy
For a subwavelength film, most transmitted power should remain in the zeroth diffraction order. A useful proxy for haze is:
Haze proxy = transmitted power in nonzero diffraction orders / total transmitted power
If the diffraction monitor shows significant nonzero order power, the film may look cloudy or scattering, which is undesirable for transparent covers or windows.
8. Expected Results and Portfolio Figures







.


8.1 Expected qualitative field behavior
Result image	What a good result should show	What a bad result looks like
xz field map	Smooth transmission through the film with moderate near-field enhancement around disks.	Strong random scattering, hot spots that depend on one defective element, or standing waves from PML reflection.
xy field map	Localized fields around nanodisks, with defects perturbing only nearby cells.	Global field collapse or large bright regions caused by mesh/boundary artifacts.
Transmission spectrum	Broad response; defective curves stay close to pristine curve.	Sharp fragile resonance destroyed by small defects.
Diffraction/haze plot	Near-zero nonzero diffraction orders across most visible wavelengths.	Large nonzero orders, especially at short wavelengths.
8.2 Final portfolio figure list
Figure A: Tidy3D GUI screenshot of the pristine 6 × 6 nanodisk supercell.
Figure B: Tidy3D GUI screenshot of the 10% missing-disk defect supercell.
Figure C: xz electric-field intensity map at 550 nm for pristine film.
Figure D: xz electric-field intensity map at 550 nm for 10% missing-disk case.
Figure E: Transmission and reflection spectra for all defect cases.
Figure F: Haze proxy or diffraction-order power versus wavelength.
Figure G: Heat map of robustness score versus disk radius and height.
Figure H: Final optimized design table with P, R, H, DSI, average T, average R, and haze proxy.
9. Validation, Error Checks, and Limitations
9.1 Numerical validation checklist
Mesh convergence: repeat the best design at 20, 25, and 30 steps per wavelength; curves should change only slightly.
Domain convergence: increase z air/substrate buffer and verify T/R do not change significantly.
Energy conservation: for lossless TiO₂/glass approximations, T + R + numerical loss should be close to 1 when all relevant channels are monitored.
Defect randomness: repeat D5/D6 for at least three random seeds and report mean ± spread.
Polarization check: repeat Ex and Ey for normal incidence; large differences imply anisotropic geometry or defect arrangement.
Angle check: test 15° and 30° incidence using Bloch boundaries to evaluate real-use robustness.
9.2 Limitations
This design uses simplified nondispersive refractive indices unless real material data are imported. A final research version should import wavelength-dependent TiO₂ and glass optical constants.
A 6 × 6 supercell repeats periodically, so it approximates disorder rather than modeling a truly random macroscopic film. Larger supercells improve realism but increase cost.
FDTD optical performance does not prove manufacturability. A fabrication-aware extension should include minimum feature size, sidewall angle, and thickness tolerance.
The target curves in this document are design goals, not measured simulation results. Actual Tidy3D plots must replace them for final submission.
10. Portfolio Presentation Strategy
This project can be presented as a research-style computational physics/engineering study. The strongest narrative is not “I made a pretty metasurface.” The stronger narrative is: “I designed an optical nanostructure, intentionally damaged it in simulation, and measured whether the design still works.”
10.1 One-paragraph application description
I designed a defect-tolerant nanophotonic film in Tidy3D FDTD GUI by modeling a subwavelength TiO₂ nanodisk metasurface on glass and testing how missing nanodisks, radius errors, height errors, and positional jitter affect visible-light transmission, reflection, and haze. Instead of optimizing only a perfect structure, I treated fabrication errors as part of the design problem. I built pristine and defective supercells, used plane-wave excitation and flux/diffraction/field monitors, and defined a Defect Sensitivity Index to compare robustness. The project taught me how electromagnetic field patterns, numerical simulation settings, and manufacturability constraints connect in real nanophotonic device design.
10.2 What professors/admissions readers should notice
You understand that photonic devices must be robust, not just optimized in perfect simulations.
You can translate a physical question into geometry, source, monitor, boundary, and mesh choices.
You use quantitative metrics instead of relying only on attractive field images.
You know how to identify limitations and avoid overstating simulation results.
You connect nanophotonics to practical applications such as displays, windows, sensors, and coatings.


11. Appendix A: GUI Build Checklist
1. Create new Tidy3D FDTD simulation with units in µm.
2. Set background medium to air.
3. Create materials: TiO₂ n = 2.35, glass n = 1.45, air n = 1.00. Use nondispersive approximations first; import real dispersive data later.
4. Create glass substrate: center z = −0.45 µm, size x/y = 1.68 µm, thickness = 0.70 µm.
5. Create TiO₂ nanodisks using GUI cylinders or script object: radius = 0.07 µm, height = 0.22 µm, centers spaced by 0.28 µm.
6. Use x/y periodic boundaries for normal incidence; z PML boundaries.
7. Add plane-wave source above film: wavelength range 450–750 nm, propagating in −z direction, Ex polarization.
8. Add reflection flux monitor above film and transmission flux monitor below substrate.
9. Add diffraction monitor on transmission side to quantify nonzero diffraction orders.
10. Add xz and xy field monitors at 550 nm.
11. Use automatic mesh with 25–30 steps per wavelength and optional local mesh override around disks.
12. Run pristine case first, then defect cases D1–D7.
13. Export spectra and field plots; create comparison plots in Tidy3D post-processing or exported Python notebook.
12. Appendix B: Script-Object Pseudocode
This pseudocode shows how a Tidy3D GUI script object could generate the defective supercell. It is intentionally written as readable pseudocode, not as guaranteed drop-in code.
# ParametersP = 0.28       # period in micrometersR0 = 0.07      # nominal radiusH0 = 0.22      # nominal heightN = 6          # supercell lengthsigma_R = 0.010sigma_xy = 0.010missing_fraction = 0.10seed = 7rng = random_generator(seed)for i in range(N):    for j in range(N):        if rng.uniform(0, 1) < missing_fraction:            continue        x = (i - (N-1)/2) * P + rng.normal(0, sigma_xy)        y = (j - (N-1)/2) * P + rng.normal(0, sigma_xy)        radius = max(0.04, R0 + rng.normal(0, sigma_R))        add_cylinder(material="TiO2", center=(x, y, H0/2), radius=radius, height=H0)


13. References
1. Flexcompute. Tidy3D Graphical User Interface course. Covers mode solver, monitors, grid specification, run time, boundary conditions, parameter sweeps, importing geometries, and script objects. https://www.flexcompute.com/tidy3d/learning-center/tidy3d-gui/
2. Flexcompute. Tidy3D GUI: Geometry Parameterization and Parameter Sweep. https://www.flexcompute.com/tidy3d/learning-center/tidy3d-gui/Lecture-11-Geometry-Parameterization-and-Parameter-Sweep/
3. Flexcompute. Tidy3D GUI: Monitor Set Up. https://www.flexcompute.com/tidy3d/learning-center/tidy3d-gui/Lecture-6-Monitors/
4. Flexcompute Documentation. Tidy3D Electromagnetic Solver Monitors API. https://docs.flexcompute.com/projects/tidy3d/en/stable/api/monitors.html
5. Flexcompute. Tidy3D Example Library, including photonic crystals, metasurfaces, diffraction structures, inverse design, and nanophotonics examples. https://www.flexcompute.com/tidy3d/examples/
6. Flexcompute. Post-processing simulation data from the Web GUI. https://www.flexcompute.com/tidy3d/examples/notebooks/GUIDataTutorial/