The project was allocated rather than chosen. I had no prior background in biosensors or clinical diagnostics. It turned out to be more useful than the topics I would have picked.
The project involved cleanroom work, biomedical device characterisation, and enough MATLAB to have real opinions about signal processing. It also, indirectly, led to Morph.
Lung cancer has poor survival rates partly because it is usually detected late. Current diagnostic methods — CT scans, bronchoscopy, biopsy — are invasive, expensive, or both. There is clinical interest in non-invasive alternatives.
Exhaled breath condensate (EBC) is collected by having a patient breathe through a cooled tube. The condensate contains trace concentrations of ions, proteins, and volatile organic compounds that reflect the biochemical state of the lungs. The question this project addressed: can a graphene-coated Ion-Sensitive Field-Effect Transistor (ISFET) detect the pH and ionic changes in EBC that are associated with lung pathology?
An ISFET is a MOSFET with an ion-sensitive gate. Instead of a metal gate contact, the gate surface is exposed to an electrolyte solution, and the threshold voltage shifts in response to ion concentration at the surface. They are used in pH sensing, DNA sequencing platforms, and biosensor arrays.
Graphene as a gate material offers higher sensitivity than conventional oxide surfaces. It has exceptional carrier mobility, a high surface-to-volume ratio, and can be functionalised to respond selectively to specific analytes. The hypothesis was that a graphene-coated ISFET would show measurable, repeatable response to the ionic composition of EBC, and that this response would differ between healthy and pathological samples.
The graphene coating was applied to CMOS ISFET arrays in a cleanroom environment. The process involved surface preparation, graphene transfer, and verification of electrical contact. Consistency across sensor batches was the central challenge — small variations in the transfer process produced measurable differences in baseline sensitivity. Multiple batches were prepared and characterised to separate process variability from genuine sensor response.
EBC was collected using an apparatus built for the project. A subject breathes through a cooled condensation tube; the breath moisture condenses and is collected for analysis. The challenge is thermal: the condensation has to be efficient enough to collect a usable sample volume — a few hundred microlitres — within a practical timeframe. Below ten minutes was the target. The collection volume directly constrains what measurements are possible, since the same sample is used for multiple sensor characterisations.
The test protocol measured sensor response across a range of pH-buffered solutions and EBC samples of varying ion concentration. The MATLAB analysis pipeline processed the raw current-voltage data into transfer characteristic curves, extracted threshold voltage shifts, and plotted sensitivity across the measurement range.
The graphene-coated ISFETs showed measurable response to pH variation in both buffer and EBC samples. The transfer characteristic curves showed the expected shift in threshold voltage with changing ion concentration, with sensitivity values in line with published graphene-ISFET literature. Variability between sensor batches was present and traceable to the coating process — a manufacturing consistency problem, not a fundamental limitation of the sensing mechanism.
The result demonstrates that the approach is viable as a proof of concept. The gap between this finding and a clinical device is substantial: selectivity for specific biomarkers rather than general pH, reproducibility at scale, and regulatory validation are all unsolved. But the sensor responds, and the response is interpretable.
[ from MSc thesis ]
Transfer characteristic curves showing threshold voltage shift with ion concentration in EBC samples. Graphene-coated vs uncoated control.
[ from MSc thesis ]
Sensor sensitivity across pH range. Variability between batches shown; traceable to graphene coating consistency.
The project introduced biomedical device constraints — the regulatory environment, the tolerance for variability, the distance between a working prototype and something deployable — in a way that has been directly relevant since. It also reinforced what the CDO2 placement had started: that the most interesting hardware problems are at the interface between electronics and a physical system that doesn't behave predictably.