Two consoles sit on the same bench in the same hospital, both labelled 30 W, both fed by the same disposable fiber out of the same cupboard. One of them is aimed at a tumour and left running for twenty minutes. The other fires bursts at a kidney stone and is finished inside a minute. If you put the two engineers who built them in a room, they would find almost nothing to talk about.
Both machines are described in the literature as “laser therapy.” Neither is doing anything the other would recognise as using a laser. One is running a photochemical reaction that needs oxygen and a drug; the other is punching a cavitation bubble into water. Their power densities are three orders of magnitude apart. Their characteristic times are five or six.
That gap is the single most useful thing to understand about laser–tissue interaction, and it is the thing most often collapsed into one word. This article is the coordinate system for everything else on this site: once you can place a clinical application on the time axis, the wavelength, the pulse structure and the fiber specification mostly fall out of it.
The sorting variable is time, not wavelength
Ask most people what decides how a laser interacts with tissue and they will say the wavelength. Wavelength decides which chromophore absorbs the light — that part is true, and we will get to it in the next article in this series. But it does not decide the mechanism. A 1064 nm pulse can coagulate a vessel or shatter a stone, at the same wavelength, by changing two other numbers.
Those two numbers are power density (W/cm², not watts) and time. Put the three branches on those two axes and they separate immediately:
| Branch | Power density | Characteristic time | What carries the effect |
| Photochemical | 0.1–0.2 W/cm² | ~10³ s | Photon + photosensitiser + oxygen |
| Photochemical | ≥ 10 W/cm² | 10⁻³–10⁻¹ s | Heat |
| Photomechanical | 10⁶–10⁷ W/cm² | 10⁻⁸–10⁻⁶ s | Stress confinement |
Read the columns rather than the rows. Between photochemical and photothermal there is a 50- to 100-fold difference in power density and a 10⁵- to 10⁶-fold difference in timescale. Between photothermal and photomechanical, another five orders of magnitude in peak power and another three to five in duration. These are not variations on a theme. They are different physics that happen to share a delivery cable.
Note that the title of this article lists the branches in the order they are conventionally taught. It is not the order of the timescale — photochemical sits at the slow end, not the fast one. Keep that reversal in mind; it is the axis the rest of this piece is built on.
Photothermal: the branch that is really a thermometer
Photothermal interaction is what almost everyone means by “laser surgery,” and it is refreshingly simple to describe: absorbed photons become heat, the tissue warms, and what happens next is decided by a temperature ladder rather than by anything optical.
Niemz’s textbook treatment puts the threshold at 10 W/cm² — below that, in continuous wave or in pulses longer than roughly a microsecond, nothing is heated meaningfully. Above it, the ladder runs:
- 45 °C — hyperthermia; cells are stressed but recover
- 50 °C — enzyme activity falls, cells begin to be fixed
- 60 °C — protein and collagen denature; irreversible after about 6 seconds (Henriques, 1947)
- 80 °C — membrane permeability rises sharply
- 100 °C — vaporisation, thermal decomposition, ablation
- above 100 °C — carbonisation
- above 300 °C — melting
Two things about this ladder are routinely underestimated. The first is that the rungs are time-dependent, not just temperature-dependent. Sixty degrees is a modest fever for six seconds and a burn for six minutes. The second is that the ladder ends in a physical wall: the latent heat of vaporisation of water is about 2260 kJ/kg. Once tissue reaches 100 °C, the next increment of energy goes into turning water into steam rather than into raising temperature further. That is why doubling the power on a device already producing steam does not double the ablation — it buys a little more steam and a lot more char.
Where the photothermal branch gets genuinely subtle is dose, and the key was named in 1983. Anderson and Parrish, writing about selective photothermolysis, defined the thermal relaxation time — how long a target takes to cool to half its peak temperature — and showed that it scales with the square of the target’s diameter. A vessel a few hundred microns across cools in milliseconds; a melanosome less than a micron across cools in hundreds of nanoseconds.
That gives the rule the whole field is built on. If your pulse is shorter than the thermal relaxation time, the heat cannot leave the target before the pulse ends, and damage stays where you wanted it. If your pulse is much longer, the heat diffuses out into everything nearby and you get a non-selective burn. Same wavelength, same average power, opposite outcome. A port-wine stain wants 1–13 ms. A melanosome wants a sub-microsecond pulse. There is no single “correct” laser for both, and no way to describe the difference in watts.
Photochemical: the branch measured in photons, not joules
Photodynamic therapy breaks the pattern in a way that is easy to miss. Everything above is about putting enough energy into a small enough space. PDT is about a chemical reaction that happens to be powered by light, and its dose numbers look like they belong to a different field entirely.
A clinical Photofrin protocol runs at 100–200 mW/cm² — that is 0.1 to 0.2 W/cm², one to two orders of magnitude below the photothermal threshold — for around 1000 seconds, accumulating a total fluence of 100–200 J/cm². ALA-PDT for skin lesions is smaller still: 160 mW/cm² for 8 minutes, 75 J/cm². The light has to arrive slowly because the mechanism is not thermal at all. Photosensitiser molecules absorb the photons, transfer energy to molecular oxygen, and generate singlet oxygen and other reactive species that damage cells directly. The sensitiser has to be present, the oxygen has to be present, and the reaction has to be given time.
This is the only branch in which more power can make the result worse. Run the light too hard and you deplete the local oxygen before the reaction can use it, or photobleach the sensitiser out of the tissue. The published protocols actually interrupt the illumination — five minutes on, two to three minutes off — specifically to let tissue oxygen recover between cycles. A photothermal surgeon raising power to shorten a case would be doing something reasonable; a PDT practitioner doing the same would be degrading his own treatment.
It also means the dosimetry is genuinely unsolved in the way the other two branches are not. A 2019 study using explicit dosimetry found that measured reactive-oxygen concentration predicted outcome better than the delivered light dose did. In other words, joules per square centimetre is a proxy for the real quantity, and a lossy one. When a clinical paper reports a PDT failure, the interesting question is usually not “was the dose right” but “was there oxygen there to receive it.”
The delivery hardware for this branch is a different animal from the fibers used everywhere else. Because 630–750 nm light is needed and the treatment target is a cavity wall rather than a point, PDT is built around the diffuser fiber — a fiber whose distal section emits sideways along a 5–30 mm length rather than out of a tip. It is also the one place where the fiber’s own material choice runs counter to the rest of the catalog, which we will come back to.
Photomechanical: the branch that does not need to heat anything
The third branch is the one that surprises people who learned the photothermal model first. In lithotripsy — and in any application using a submicrosecond pulse — the tissue never has time to conduct heat anywhere, because the energy is deposited faster than sound can cross the absorbing volume. The criterion is not a temperature, it is stress confinement: the pulse must be shorter than the time it takes an acoustic wave to traverse the absorption depth.
Practically, the literature draws the line around a pulse duration of 10 µs. Longer than that and the acoustic contribution is negligible next to the thermal one. Shorter, and the thermal mechanism stops being the primary actor at all. At 10⁶–10⁷ W/cm² delivered in 10⁻⁸–10⁻⁶ s, what leaves the fiber tip is not a temperature rise but a pressure wave and a bubble.
The bubble is where this gets interesting, and it is the reason a 2100 nm holmium laser works at all. Holmium light is absorbed so strongly by water that the effective tissue penetration is around 0.4 mm — a safety feature, since it means energy can’t travel far into the wrong structure. But the same property means that if the fiber tip is more than a few millimetres from the target, the water in between absorbs everything before any light arrives. Past roughly 4 mm, the treatment simply fails.
The solution is not a better laser. It is a bubble the light makes for itself. The leading edge of the pulse vaporises a small volume of water at the fiber tip; that vapour bubble has a much lower refractive index and absorption than the surrounding water, so it acts as a temporary tunnel, and the rest of the pulse travels through it to reach the stone. Isner and colleagues photographed this in 1986 (they called the structures “dynamic optical cavities”) and published in 1988; Royston’s group first quantified the transmission gain; van Leeuwen’s group later measured Ho:YSGG pulses penetrating saline and blood 3.6 times deeper than the water absorption coefficient would predict. The effect carries Moses’s name for obvious reasons.
Once you accept that the bubble is part of the mechanism, a set of clinical numbers stops looking arbitrary:
- The bubble is roughly 2 mm across.
- The optimal fiber-tip-to-target distance is 1–2 mm. Closer than 1 mm, the bubble cannot form properly and the mechanism degrades into “drilling” rather than fragmentation. Much further, and the shockwave decays before it arrives.
- The maximum bubble size depends on the fiber diameter and on the pulse energy — which is the part suppliers rarely say out loud.
- A longer pulse lowers peak power and stretches the bubble from a sphere into a pear shape, reducing the recoil impulse transmitted back to the fiber, at the cost of ablation efficiency.
That third bullet is the one to hold onto. In this branch, the holmium fiber is not a passive conduit carrying light to a place where physics happens. Its core diameter is one of the inputs that sets the bubble geometry, the working distance and the recoil force. Change the fiber and you have changed the mechanism, not just the brightness.
One more set of numbers, because they rarely appear in vendor material and they explain why cavitation is enough. The thermal decomposition thresholds of the common stone compositions are 100 °C for struvite, 206 °C for calcium oxalate monohydrate, 264 °C for cystine and 360 °C for uric acid. Every one of them sits below the critical point of water at 374 °C and 22 MPa — so the conditions inside a collapsing vapour bubble are sufficient to break down essentially any stone chemistry. The photomechanical branch does not need to win on temperature. It wins on pressure and time.