The pulsed dye laser has been the workhorse of vascular lesion treatment for four decades, and it does not operate at the wavelength its physics would suggest.
Hemoglobin’s absorption spectrum has a sharp peak at 577 nm. If “chromophore matching” meant what it sounds like — find the peak, set the laser there — every pulsed dye laser would be built at 577 nm. Almost none of them are. The clinical wavelengths cluster around 585 to 595 nm, deliberately off-peak, and the reason is not manufacturing convenience. Move the wavelength from 577 to about 590 nm and hemoglobin absorbs slightly less light. But melanin absorbs considerably less, and dermal scattering falls too — and if you compute the ratio that actually governs the treatment, the energy absorbed by the deep vessel divided by the energy absorbed by the epidermis, that ratio improves. The laser got worse at absorbing, and better at treating.
That single example contains the whole principle, and it is the reason this article exists. Wavelength selection is not a lookup against an absorption spectrum. It is a negotiation between three curves that all move in different directions, and the peak of one of them is only the starting point of the conversation.
What “matching” actually takes
In laser applications, chromophore absorption sounds like a single number to look up and match. It is not. The phrase gets used as though it describes one condition; it describes three, and a wavelength is only matched when all three hold:
- Light has to arrive. The photons must reach the target depth with enough remaining intensity to matter. This is decided by absorption and scattering along the path — not by the target’s own spectrum.
- The target has to out-absorb its competitors. Every tissue volume contains several chromophores absorbing simultaneously. What matters is not absolute absorption but the ratio between the one you want and the ones you do not.
- Something has to convert the absorption into an effect. Heat, a chemical reaction, a pressure wave. This is the mechanism layer from the previous article in this series, and it can veto a wavelength that satisfies the first two conditions perfectly.
The 577 nm case fails condition two. It is the best wavelength for hemoglobin in isolation and a mediocre one in a face that also contains melanin. Most of the errors in wavelength selection are errors of looking at one curve when the decision requires three.
Water: a few sharp overtones and one rule that is usually quoted backwards
Water is the largest chromophore in soft tissue by mass — around 60 % of body weight — and its near-infrared spectrum is not a smooth ramp. It is a set of O–H vibrational overtones and combination bands at roughly 760, 970, 1190, 1450 and 1940 nm, and the absorption coefficient climbs steeply across them:
| Water band | Absorption coefficient (approx.) | Relative to 970 nm |
| 970 nm | ~0.5 cm⁻¹ | 1× |
| 1190 nm | ~1 cm⁻¹ | 2× |
| 1450 nm | ~28.6 cm⁻¹ | 60× |
| 1940 nm | ~129 cm⁻¹ | 260× |
Put those numbers beside the clinical penetration figures and a familiar rule collapses. Effective optical penetration in water-dominated tissue runs about 3.0 mm at 980 nm, 220–400 µm at 1470 nm, and 48–100 µm at 1940 nm. In other words, within the infrared, longer wavelength means shallower, not deeper — the exact opposite of what “longer wavelengths penetrate further” implies.
That folk rule is not wrong, it is misfiled. It holds inside the 600–1100 nm therapeutic window, where scattering dominates and reduced scattering falls with wavelength, so longer wavelengths genuinely do reach deeper. Above roughly 1100 nm, water absorption takes over and drives penetration down by an order of magnitude over a few hundred nanometers. Two adjacent pieces of spectrum, two opposite rules, and the transition between them is the single most common source of confusion in laser wavelength selection.
This is why 2100 nm behaves the way it does in lithotripsy. At that wavelength water absorbs so strongly that tissue penetration is roughly 0.4 mm — which is a safety property and a constraint at the same time, because it means the holmium fiber has to be brought into contact or near-contact with the stone for any of the energy to arrive at all. There is no standoff option at 2100 nm. The wavelength made that decision before the surgeon entered the room.
Scattering is the other half of the distance problem
Everything above is about absorption, which is only half of what stops light from getting somewhere. The other half is scattering, and it is the half that chromophore matching conveniently forgets.
Tissue is not a clear medium with absorbers dissolved in it. It is a dense suspension of collagen fibers, cell membranes and other structures whose dimensions are comparable to the wavelength of visible light. Photons do not travel straight through it; they ricochet, and the ricochets cost them. Reduced scattering in tissue is strongest at the blue end — which is why a hand held over a torch glows red rather than blue, and why blue light is confined to superficial work despite hemoglobin absorbing it well. Two separate mechanisms conspire at short wavelengths: the chromophores absorb more, and the medium scatters more.
Scattering falls with wavelength across the visible and near-infrared, in the Rayleigh-like λ⁻⁴ regime at the short end and flattening toward the red. That decline is the entire physical justification for the “longer wavelengths reach deeper” rule inside the therapeutic window — and note that it is a scattering argument, not an absorption one. The reason 810 nm reaches several millimeters while 532 nm does not is mostly that the tissue stopped getting in the way, not that the target became hungrier.
The practical consequence is that penetration depth cannot be computed from absorption alone. The quantity that governs it is the effective attenuation coefficient, which combines absorption and reduced scattering and is what the half-attenuation depths below actually measure. A wavelength can have a disappointingly weak absorption coefficient for its target and still outperform a strongly absorbed one, if the medium happens to be less hostile to it. 1064 nm is the standard example: hemoglobin is a poor absorber there, which is exactly why it was historically considered a poor vascular wavelength — but melanin absorption is low, water absorption is low, and scattering is low, so the light reaches depths that green light never will. The treatment becomes possible not because the target got better, but because two obstacles stepped aside at once.
Hemoglobin: one word, two curves, and a blind spot
“Hemoglobin” is not one chromophore. Oxygenated and deoxygenated hemoglobin have different spectra — close enough to share absorption peaks around 418, 542 and 577 nm, different enough that the two curves cross at roughly 800–805 nm, the isosbestic point, where they absorb equally.
That crossing point is worth understanding because it is counter-intuitive in a useful way. At 800 nm, a change in oxygenation produces no change in absorption at all, because whatever oxyhemoglobin loses, deoxyhemoglobin gains. A measurement at 800 nm is blind to oxygenation and sensitive only to total hemoglobin concentration. Below 800 nm, deoxyhemoglobin absorbs more; above it, oxyhemoglobin does. This is precisely the logic behind pulse oximetry’s choice of two wavelengths on opposite sides of the crossing — and why a clinician asking a laser to “target blood” needs to say which blood state he means.
There is a second asymmetry with direct practical consequences. Above about 600 nm, oxyhemoglobin’s absorption decays faster than deoxyhemoglobin’s. The visible-light window for vascular work narrows as you move redward, and the available contrast does not just fade — it changes which hemoglobin species you are actually treating.
And then the competitor arrives. Because the vessel is deep to the epidermis, and the epidermis is full of melanin, the relevant question for a vascular laser is never “how well does hemoglobin absorb at this wavelength” but “how much better than the melanin sitting in front of it”. Move far enough into the red and hemoglobin’s advantage shrinks; move back into the green and melanin’s absorption climbs faster than hemoglobin’s. That trade-off is what produced the 585–595 nm cluster, and it is also why the same lesion in a Fitzpatrick V or VI patient is often treated with a 1064 nm laser instead. At 1064 nm hemoglobin absorbs weakly — but so does melanin, and the treatment becomes possible at all because the competitor was weakened more than the target.
That is condition two of matching, made visible: the winning wavelength is not the one where the target absorbs most. It is the one where the target absorbs most relative to everything else in the beam path.
Melanin: both a target and a creditor
Melanin absorbs broadly from the ultraviolet through the visible and decays into the red, with meaningful absorption out to roughly 1200 nm. Its two-sidedness is the point. In hair removal and pigmented lesion treatment it is the target; in every other cutaneous application it is a competing absorber that converts wanted photons into unwanted heat at the surface.
That makes skin type a wavelength variable, not a patient variable. Two patients with the same lesion and different melanin content need different wavelengths, because the ratio in condition two has changed. The clinical literature is consistent on this and it is worth stating plainly: a wavelength that is optimal for light skin can be unsafe in dark skin for reasons that have nothing to do with the lesion. The physics is the same in both cases, which is precisely why the choice has to change.
Fat has a chromophore too, and it is not water
Soft tissue is mostly water, so most infrared wavelengths are, functionally, water lasers. Adipose tissue is the exception that proves the principle. Lipid has its own absorption bands from C–H stretching overtones at roughly 920, 1040, 1210, 1730 and 1760 nm, and the contrast between them and the water bands is large: lipid absorption at 1210 nm is more than ten times its value at 930 nm, while water’s absorption at those two wavelengths stays comparatively low.
That gives lipolysis two genuinely different technical routes, and the distinction is a good test of whether someone understands chromophore matching or has merely memorized it:
- Heat the water, let the fat die indirectly. This is what 980 nm and 1470 nm diode systems do. The chromophore is water; the fat is damaged by heat conducted from the water around it. It works, and it is what the lipolysis fiber range is built around.
- Heat the fat directly. This is the 1210 nm route, aimed at the C–H overtone so that the chromophore is the target tissue. Selectivity is higher in principle; achievable depth is shallower, because the same absorption that makes it selective limits penetration.
Neither is universally better. What matters is that they are not variations on one design — they are two different answers to the question “which chromophore is doing the work”, and the fiber specification that serves one is not automatically right for the other. Fat also scatters differently from water-based tissue, and its reduced scattering coefficient falls with wavelength in the same power-law way, which is why longer-wavelength routes are sometimes chosen for reach even when their absorption contrast is poorer.
The window is a negotiation between curves, not a band you can memorize
Put water, hemoglobin and melanin on the same axes and a valley appears between roughly 600 and 1100 nm. Below it, hemoglobin and melanin close the window; above it, water closes it. The band is not defined by any single chromophore being transparent — it is defined by none of them being dominant, which is a different and more fragile kind of transparency.
Inside that valley the reachable depths are well documented, using half-attenuation depth (the depth at which half the photons have been absorbed or scattered out of the beam):
| Wavelength | Half-attenuation depth | Reaches |
| 633 nm | ~1.0 mm | Epidermis, papillary dermis |
| 660 nm | ~1.5 mm | Full dermis, hair follicle bulb |
| 810 nm | ~3–4 mm | Subcutaneous fat, peripheral nerve |
| 830 nm | ~4 mm | Muscle, joint capsule |
| 1064 nm | ~5–8 mm | Deep tendon, through skull to outer cortex |
Read that table as a negotiation rather than a menu. Scattering — dominated by collagen fibers and cell membranes — falls with wavelength across the whole range, which pushes reach upward as you go redward. Absorption by the residual chromophores pushes it back down. The much-quoted rule that longer wavelengths penetrate deeper is true inside this valley and false outside it, and the valley closes at both ends for reasons that have nothing to do with each other.
Within the valley, one further division is worth more attention than it usually gets. The valley is not uniform, and in a lumen full of fluid — a bladder, a ureter, a renal pelvis — the strong water bands at 1450 and 1940 nm will quench a signal long before it reaches anything worth seeing, so endoluminal work prefers the sub-band from roughly 1000 to 1300 nm, which is comparatively water-transparent. Set that against lithotripsy at 2100 nm, which is chosen because water absorbs it strongly. Same organ system, same fluid, opposite wavelengths — and both are correct. One treats the water band as an obstacle to route around; the other treats it as the mechanism itself. The holmium fiber sits at the second end of that logic.
Three ways a wavelength choice fails
The three conditions from the opening give a useful way to diagnose a wavelength decision that is not working, because each of them fails differently.
The light never arrives. Absorption or scattering removes it before it reaches the target depth. Blue and green light on a deep dermal lesion is the textbook case: the target may absorb beautifully, but nothing reaches it, and the energy is deposited in melanin and surface tissue instead. Increasing power does not fix this — it deepens the surface injury while the target stays dark.
The light arrives, but the target is not the biggest absorber in the room. This is the 577 nm case. Hemoglobin absorbs strongly at the peak, and so does everything in front of it, and the ratio that governs selectivity goes the wrong way. The fix is usually to move off the peak — the same move that produced the 585–595 nm cluster and the 1064 nm option for darker skin.
The light arrives and the target wins, but nothing converts it into an effect. This is the most easily missed failure because it involves no optical error at all. A perfectly matched wavelength in a photodynamic application still needs the sensitiser present and oxygen available; a photothermal wavelength still needs enough power density to cross the 10 W/cm² threshold; a photomechanical one still needs a pulse short enough to be stress-confined. Mechanism is the third condition and it can veto a wavelength that satisfied conditions one and two perfectly.
The curve that never matches
One more entry belongs in the negotiation and is usually left out: the light source’s own loss curve. It is not enough for a chromophore to be well matched and reachable. The delivery fiber has to be efficient at that wavelength too, and its loss curve is set by entirely different physics — Rayleigh scattering falling as λ⁻⁴ at the short end, hydroxyl absorption around 1383 nm in the middle, multiphonon absorption at the long end. The bare fiber range is specified across 400–2200 nm, but that is a coverage figure, not a promise of equal performance at every point in it. There is essentially no wavelength at which the tissue optimum and the fiber optimum coincide by design. Finding the point where they come acceptably close is what wavelength selection actually is, and it is a compromise engineering problem, not a spectroscopy lookup.