Research · PhD, University of Edinburgh 2022 · Postdoc, IQS Barcelona 2022 to 2024

Singlet oxygen, from a mouse to a molecule

Photodynamic therapy treats tumours and infections with light, a drug and oxygen. I built a way to measure the molecule that does the damage.

Part of Work · What I've done

Before AI, I was a physicist. My PhD in Edinburgh measured singlet oxygen, the short-lived molecule that photodynamic therapy uses to kill cells, when the drug is switched on by two photons at once. I did it in water, with a few drops of drug inside a hollow glass fibre thinner than a hair. It was the first quantitative measurement of its kind.

Much of how I work now comes from those years: measure the thing directly and fix one problem at a time. Scroll down to zoom from a mouse to a molecule, then through the detector, the fibre and the seven versions of the experiment. The last section is my postdoc in Barcelona, measuring singlet oxygen inside bacteria.

Terms used here

Singlet oxygen (¹O₂): An excited form of oxygen. It lives about 3.5 µs in water and damages whatever is within about a micrometre.
Photosensitiser: The drug. It absorbs light and passes the energy to oxygen.
Photodynamic therapy: Treatment with light, a light-sensitive drug and oxygen. The drug only switches on where the light shines.
Two-photon absorption: Excitation by two particles of light arriving together. It only happens where the light is most intense, at the focus.
Hollow-core fibre: A glass fibre with an empty core. Here the core holds the sample and guides light along the whole length.
Pump-probe: One laser drives the reaction (the pump, 800 nm) and another reads the result (the probe, 470 nm).
Log-log slope: Signal against laser power on logarithmic axes. A slope of 2 means two photons are needed.
nL and M: A nanolitre (nL) is a millionth of a millilitre. M (molar) measures concentration, and 10⁻⁸ M is very dilute.
Cross-section (GM): How strongly a molecule absorbs two photons, in Göppert-Mayer units.

The ideaLight switches the drug on where it shines. The drug turns oxygen into singlet oxygen (¹O₂), which destroys the cell. How much it makes is a measure of the drug's strength.

The questionNear-infrared light reaches deeper, and two-photon excitation acts only at the focus. Nobody had measured how much singlet oxygen drugs make this way in water, which is what matters in the clinic.

The approachOne tiny focus gives too little signal. So I filled a hollow glass fibre, thinner than a hair, with the drug and sent the laser down it. The light then works along the whole 15 cm.

first: two-photon ¹O₂ measured in water
≈ 40 nL: of sample in the fibre
10⁶×: less drug than a 1 mL cuvette (about 40 nL at 10⁻⁸ M)

The resultThe first quantitative measurement of two-photon singlet oxygen in water, in any optical system, using only drops of drug.

The diagrams are schematics. Below is the illustrated version, then the detail.

3 cm

Scroll to zoom

Powers of ten

This shows photodynamic therapy in general. My own thesis work was in solution. Switch between one and two photons at each scale.

3 cm · The animal

Light from outside

One photon acts along its whole path. Two near-infrared photons add up only at the focus, which can sit inside the tumour.

3 mm · The tissue

Skin, fat, muscle, tumour

Blood and scattering stop short wavelengths near the surface. Near-infrared light, 750 to 950 nm (the phototherapeutic window), gets further.

100 µm · The cells

The drug waits in the tumour

The drug, a photosensitiser (amber dots), builds up in tumour cells over hours to days and does nothing in the dark.

10 µm · One cell

A one-micrometre reach

Singlet oxygen lives about 3.5 µs in water and acts within roughly 1 to 1.5 µm (dashed red), so it destroys whatever sits next to the drug.

1 nm · The molecules

Absorb, cross, transfer

The drug absorbs, flips to a long-lived triplet state and hands its energy to oxygen. Out comes singlet oxygen, which oxidises the nearest lipid.

10 µm · The outcome

Damage where the light was

With two photons only the cell at the focus is hit. That precision is the reason to measure how much singlet oxygen each drug makes with two photons.

01 · Seeing it

1270 nm photon counting

Detecting a molecule that barely glows

Singlet oxygen gives off a faint infrared glow at 1270 nm. In water only about one molecule in a million emits a photon, and the detector catches about 3% of those. Williams and colleagues in our group first detected the glow inside a hollow fibre in 2018, and my thesis started from there.

photons arriving at the NIR detectortime after the laser pulse

Look for a quick rise, then a slow fade. The fade is singlet oxygen's lifetime. Heavy water stretches it about twenty times, which is why plain water, the one cells live in, is so much harder.

photons detected0

photons emitted (3% QE)0

singlet oxygen made0

fitted lifetimecollecting…

Simulation: arrival time = triplet delay (2 µs, estimated for air-saturated water) + singlet-oxygen lifetime (thesis values), with 3% dark counts; lifetime fitted by maximum likelihood on the tail. It reads slightly long (about 3.8 and 71 µs against 3.5 and 68 µs) because the tail still holds some of the rise and the dark counts.

02 · The PhD

hollow-core photonic crystal fibre

A lab inside a fibre the width of a hair

An anti-resonant hollow-core fibre: a ring of seven thin glass capillaries around a hollow core. The core holds the sample and guides both lasers through it.

fibre cross-section

Look for the light staying in the central hole, switching between the 470 nm probe and the 800 nm pump.

guidance windows

membrane thickness 300 nm

Each dip is a colour that leaks out. Drag the slider until both dashed lines sit in a clear window.

How it guides. The thin capillary walls act as Fabry-Perot resonators. Away from their resonances they reflect light back into the core; at a resonance, λ = 2t·√(n_glass² − n_core²)/m, light leaks out. Filling the core with water moves every resonance to about 0.55 times its wavelength in air, so the fibre carries both a 470 nm probe and an 800 nm pump in water.

Light-matter interaction from my thesis, for a 20 µm core 15 cm long, on a log scale. The fibre holds about 2.8 nL per cm, around 40 nL in total.

The fibre drawing is schematic, and the mode is the J₀² profile of a hollow waveguide. The membrane thickness is a free parameter for you to set.

03 · The method

two-colour pump-probe

Counting singlet oxygen with a light switch

In water, inside a fibre, the 1270 nm glow is too weak to count. So I used a chemical reporter, Singlet Oxygen Sensor Green (SOSG). It stays dark until singlet oxygen adds across its anthracene, then fluoresces about ten times brighter. Two lasers share the core. An 800 nm femtosecond pump excites the photosensitiser by two photons for 60 s, then a 470 nm probe reads the reporter for 5 s, and the cycle repeats twenty times or more.

How fast the reporter brightens tells you how much singlet oxygen was made. Plot that rate against laser power on log-log axes and two photons give a slope of 2. In the final design each photosensitiser landed between 1.85 and 2.09. Divide the rate by concentration and you get the number that matters for two-photon therapy, cross-section times singlet-oxygen yield. The agent on my lab-agent page arrived at the same log-log test by itself.

04 · Seven designs

the engineering

From 350 mW to 10 mW, one problem at a time

Design	Reporter / photosensitiser	Fibre	800 nm power	What changed and what it fixed
1	50–100 / 1–10 µM	30 cm	350 mW	Proof of concept. Weak signal, re-absorption, early plateau from oxygen running out.
2	5 / 0.5 µM	30 cm	150–450 mW	Twenty times less reporter, more signal. First log-log proof of two photons (slope 2.03).
3	1 / 0.05–0.5 µM	30 cm	150–450 mW	More photosensitisers. Found residues on the fibre walls skewing late measurements.
·	study	30 cm	up to 300 mW	Photobleaching study: the reporter self-sensitises, its product bleaches, photosensitisers bleach above about 200 mW.
4	1 / 0.05–0.1 µM	30 cm	50–150 mW	Light moved from the cladding into the core, with both lasers in the same fundamental mode.
5	1 / 0.05–0.1 µM	15 cm	50–150 mW	Shorter fibre, less micro-bending loss, cleaner spectra.
6	1 / 0.1–1 µM	15 cm	50–150 mW	Probe down to 50 µW: ten times more counts, no oxygen plateau.
Final	2 / 0.1–1 µM	15 cm	10–50 mW	Optical isolator (back-reflections were breaking the laser's mode-locking), telescopes matching both beams, autocorrelator checks. The cleanest modes of any design.

singlet-oxygen yield in water: two-photon (this work) vs one-photon literaturereference: Rose Bengal 0.76; Ruth G1, G2: AlPcS4

Amber is the two-photon yield I measured in the fibre, grey the one-photon value from the literature. Most pairs agree within error, which validates the method. *Ruthenium complexes designed for two-photon use by the Gasser group, with literature values in PBS rather than water. They have the highest cross-sections, about 330 to 360 GM.

05 · What was new

The first use of an anti-resonant hollow-core fibre in photochemistry, and of the SOSG reporter ratiometrically inside a hollow-core fibre.
Photosensitiser concentrations down to 10⁻⁸ M, a hundred times lower than earlier fibre work and low enough to avoid aggregation. At that level a copper phthalocyanine reported to make no singlet oxygen turned out to make plenty as a monomer.
Two-photon cross-sections of clinical photosensitisers measured in water, many for the first time.

06 · The postdoc

Light4Lungs · IQS Barcelona

Breathing in the light

Light4Lungs was an EU Horizon 2020 FET Open project, coordinated at IQS by Prof. Santi Nonell, whose singlet-oxygen probes I had used throughout my PhD. The aim was to treat drug-resistant lung infections with an inhaled photodynamic therapy, using light-emitting particles breathed in as an aerosol.

I measured the singlet oxygen made inside bacteria with photodiodes and photomultiplier tubes, both sensitive enough for very faint light (try both detectors). I also coordinated the optics, biochemistry and modelling teams.

Try the two-photon physics Watch an agent repeat my method

Sources

S. Adan Bermudez, Detection of Two-Photon Photosensitised Singlet Oxygen in Hollow-Core Photonic Crystal Fibres, PhD thesis, University of Edinburgh, 2022 (supervisor Prof. Anita C. Jones).
G. O. S. Williams et al., ChemPhotoChem 2018: direct 1270 nm detection of singlet oxygen in a hollow-core fibre, the group's earlier work this thesis built on.
light4lungs.eu: EU H2020 FET Open, grant 863102.