Bioaerosol Detection With Light Scattering and Fluorescence

ISEF Category: Embedded Systems

Subcategory: Sensors  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A single dust speck can look very different depending on what it is made of. Pollen, mold spores, and soot all travel through the air, but they do not reflect and glow in the same way. That means light can help tell them apart. You can turn that idea into a real sensor project.

What Is It?

This project asks a simple question with a tricky answer, can you tell what kind of airborne particle just passed through a sensor? The detector uses a 405 nm laser for forward scattering, which means it measures how light bounces off a particle, and a 365 nm UV-LED for fluorescence, which means it measures light the particle gives off after being excited by ultraviolet light.

Think of each particle like a tiny fingerprint. Soot, pollen, and mold spores can all trigger a signal, but their shapes, sizes, and chemistry change the pattern. A microcontroller, or MCU, can read those signals and sort them with a simple classifier. Your job is to test whether the patterns really separate the particle types well enough to predict them.

Why This Is a Good Topic

This is a strong science fair topic because it connects a clear engineering challenge, sensor design, to a real problem, indoor air quality. You can test whether different particle types produce different optical signatures, then measure how well a classifier separates them. You also get to learn signal processing, calibration, and confusion matrix analysis, which are the kinds of skills judges like to see.

Research Questions

• How does particle type affect the ratio of scattering to fluorescence signal?
• What is the effect of particle size on forward-scatter peak shape and time-of-flight signature?
• Does adding fluorescence data improve classification accuracy compared with scattering alone?
• To what extent do humidity changes alter the sensor's ability to separate pollen, mold spores, and soot?
• Which simple classifier gives the best separation for these particle signals on an MCU?
• How does sampling angle affect the signal gap between biological and combustion particles?
• What is the effect of background dust on false positive classifications?

Basic Materials

• 405 nm laser diode module with stable mount.
• 365 nm UV-LED with driver circuit.
• Photodiode with amplifier circuit.
• Microcontroller board with analog inputs.
• Breadboard or prototyping PCB.
• Resistors, capacitors, and hookup wire.
• Optical enclosure or dark box.
• Sample chamber with inlet and outlet ports.
• Reference particle samples such as pollen, mold spores, and soot.
• Digital multimeter.
• Smartphone camera for setup checks.
• Laptop for data logging and analysis.

Advanced Materials

• Optical breadboard or rigid alignment rail.
• Calibrated photodiode and low-noise transimpedance amplifier.
• Narrow-band optical filters for scatter and fluorescence channels.
• Time-of-flight sensor or fast sampling ADC.
• MCU development board with higher sampling speed.
• Aerosol sampling chamber with flow control.
• Laser power meter.
• Optical fiber components for signal routing.
• Reference microscopy slides for particle verification.
• Environmental sensors for humidity and temperature.
• Standard aerosol generator or controlled particle source.
• Particle counter or reference instrument for validation.

Software & Tools

• Python: Cleans raw sensor data, plots signal features, and trains simple classifiers.
• ImageJ: Helps compare particle images or reference microscopy data if you verify sample identity.
• Excel: Organizes trials, labels samples, and makes first-pass graphs.
• MIT OpenCourseWare notes: Gives free background on sensors, signal processing, and basic machine learning.
• PubMed: Helps you find review articles on bioaerosols, fluorescence sensing, and optical particle detection.

Experiment Steps

1. Define the particle classes you can collect reliably, and choose a sensor setup that should separate them.
2. Decide which signal features matter most, such as peak height, pulse width, scatter-to-fluorescence ratio, and arrival timing.
3. Build a calibration plan that links raw sensor output to known reference samples.
4. Plan controls that test false alarms, background dust, ambient light, and humidity drift.
5. Choose a classifier strategy that fits your hardware, then set up train-test splits before collecting large datasets.
6. Map out how you will judge success using accuracy, precision, recall, and confusion matrices.

Common Pitfalls

• Mixing particle sources that overlap too much, which makes the classes impossible to separate cleanly.
• Letting ambient light leak into the optical chamber, which distorts fluorescence readings.
• Using one sample batch for both training and testing, which inflates classifier performance.
• Ignoring humidity and flow changes, which can shift scattering signatures across trials.
• Trusting raw signal peaks without normalizing sensor drift, which hides real differences between particle types.

What Makes This Competitive

A stronger version of this project does more than build a working detector. It tests whether the classifier still works under messy real-world conditions, like mixed particles, changing humidity, and sensor drift. You can stand out by comparing multiple feature sets, or by showing which optical signal carries the most useful information. Good validation and honest error analysis matter more than flashy hardware.

Project Variations

• Compare indoor dust, pollen, and candle soot instead of only three clean reference samples.
• Test whether fluorescence adds more value than scatter data when the detector runs on battery power.
• Swap the classifier style, then compare a rule-based MCU model with a small machine learning model trained offline.

Learn More

• NASA Earth Observatory: Search for articles on aerosols, particulate matter, and how light interacts with airborne particles.
• NOAA Air Resources Laboratory: Look for free background on atmospheric particles and air monitoring methods.
• PubMed: Search review articles on bioaerosols, fluorescence detection, and optical particle sensing.
• USGS Water Science School: Use the optics and calibration background to review how scientists measure tiny suspended particles.
• MIT OpenCourseWare: Search for free lecture notes on sensors, signal processing, and embedded systems design.