Notes from the bench and the courtyard

Three programs, two file conversions and one undocumented baseline setting stood between a measurement and a figure. The story of how that became RNSpectraX — and then RNBeamX, an optical-setup designer where the light is traced — told one upgrade at a time, with an overview of what both tools can do today.

Why removing the recognition element from a biosensor is the hard part — and what a receptor-free spectral assay has to prove before anyone should trust it in a clinic. Tag-based assays and lateral-flow tests, imaging in surgery, biofluid, breath and pathogen fingerprints; predictive values, data leakage in machine learning, reproducibility and the evidence ladder, with simulations and references.

What an antibody is and which parts of it matter, how gold is prepared with self-assembled monolayers and dextran, and every common way of attaching the antibody — adsorption, EDC/NHS amine coupling, Protein A/G, streptavidin–biotin, glycan and thiol chemistry, tags. Then the physics: why orientation sets the signal, what R_max to expect, and how mass transport bends a sensorgram. With reaction mechanisms, protocols, simulations and references.

Why we need vacuum, how rotary, diffusion and turbo pumps move gas, and why roughing, backing and hi-vac must happen in one order and no other. What GH1, GH2, LT, HT and the DTM on the front panel really mean; how thermal evaporation differs from electron-beam evaporation, and both from electron-beam lithography. Every equation explained, with a coating-unit simulator and other figures you can play with.

When does SERS read the analyte itself, and when does it read a reporter tied to it? The physics both share, why distance decides every design, the adsorption chemistry and Langmuir isotherms behind label-free calibration, partition layers and SHINERS; then SERS tags, Raman reporters, SERRS and silent-region alkyne tags, sandwich assays, multiplex unmixing, hot spots and internal standards — with advantages, disadvantages and how to choose.

Spontaneous Raman is a superb fingerprint and a poor signal. Each variant exists to remove one of its limits: confocal Raman says where, FT-Raman escapes fluorescence, resonance Raman, SERS and TERS amplify the signal (TERS down to nanometres), CARS and SRS make imaging fast, and SORS reads beneath the surface. Why each was needed, how it works from first principles, and what it gains over the one before.

Why a bond vibrates at the frequency it does; stretching, scissoring, rocking, wagging and twisting; how infrared absorption and Raman scattering work and why they see different things; the Raman and IR bands of lipids, proteins and DNA; the fingerprint region, and why the silent region is silent. Then plasmonics — the Drude model, surface plasmons, Mie theory — and how it turns Raman into SERS. With the equations, simulations and references.

What light is — wave, particle, photon — and Maxwell's equations term by term, the wave equation and the speed of light, polarisation; light in glass, at interfaces, in metals and in nonlinear crystals; then how light is trapped and guided: slab modes, optical fibres, silicon photonic wires and couplers, metallic, plasmonic and photonic-crystal waveguides. With simulations, diagrams and references.

Like every other morning I was walking back from the gym, already thinking about breakfast, when something unusual caught my attention. A boy was running across the courtyard, frantically brushing at his head.