A procurement engineer sent us a specification last month. Three lines, cleanly typed:
Core 400 µm. Cladding 440 µm. Coating 470 µm.
Every number, as far as I could tell, was aspirational. The fiber he had been buying for two years was a 400/730/800 part. He had copied the first column of a competitor’s datasheet, replaced the middle two numbers with what he assumed they “should” be — core plus a bit, then a bit more — and sent it out with a request for quotation. Four suppliers quoted it. Not one of them wrote back to ask why a 400 µm core suddenly needed a 40 µm cladding.
That is the whole problem in one email. Core, cladding and coating diameter look like three graduations of the same thing. They are three unrelated specifications that happen to be measured with the same instrument. They are set by different physics, they fail in different ways, and — this is the part that costs money — only one of them has anything to do with how much power you can put through the fiber.
The three diameters in the specification above, incidentally, describe a fiber that does not exist. The real range is on the bare fiber page, and the numbers there do not move in step.
The core diameter does exactly one job: it divides
If you want one sentence to carry away from this article, it is that the core diameter is the only one of the three that touches power density. Irradiance is power over area, and area goes as the square of diameter, so halving the core quadruples the loading. The bare fiber range runs 200 to 1000 µm; at the page’s 30 W ceiling, a modest CW figure for a device of this class, the three diameters that matter most inside that range land as follows.
| Core | Emitting area | Irradiance at 30 W |
| 200 µm | 3.14 × 10⁻⁴ cm² | 95.5 kW/cm² |
| 400 µm | 1.26 × 10⁻³ cm² | 23.9 kW/cm² |
| 600 µm | 2.83 × 10⁻³ cm² | 10.6 kW/cm² |
| 1000 µm | 7.85 × 10⁻³ cm² | 3.8 kW/cm² |
The published practical figure for continuous-wave operation on a fused silica end face is about 250 kW/cm². Run the numbers and the conclusion is uncomfortable for anyone who thinks of the 200 µm fiber as a small version of the 600 µm one: 30 W on a 200 µm core consumes 38 % of the damage budget before the fiber has touched anything. Contamination, a chip, a slightly angled cleave, and it is gone. On a 1000 µm core, the same laser is at 1.5 %, and the fiber will survive conditions that would end a 200 µm in one case.
Notice what does not appear in that table. Not the cladding. Not the coating. Not the NA, the length, or the wavelength. Power density is a function of core area and delivered watts, and nothing else changes it. If a case is burning through fibers, the first question is not “which cable is better” but “what is the core diameter, and do we actually need this one.”
The core diameter also sets how many modes travel. A 400 µm core at NA 0.22 carries thousands of them, which is why the output of a large-core medical fiber is a flat-topped, spatially incoherent patch rather than a beam. That is a feature for tissue work — the spot is more uniform than a Gaussian — and a nuisance for anyone expecting to focus it. But be careful with the deeper spec-sheet implication: mode count depends on both core and NA, and NA is not a property of the core.
The cladding diameter is not the core plus a margin
Here is the assumption that produced “440 µm” in that email: that the cladding is a fixed jacket wrapped around the core at some standard ratio. In telecommunications it very nearly is. IEC 60793-2-10 defines multimode fibers as 50/125, 62.5/125 and 100/140 — the cladding is 125 or 140 µm in every case, regardless of core, and the cladding diameter is not something the buyer specifies at all.
Large-core medical fiber sits outside that system entirely. Nothing in IEC 60793 covers a 200–1000 µm core, so the cladding diameter is not a standard dimension — it is a design choice made by the fiber manufacturer, and the ratio is all over the place. The clearest evidence is on our own radial fiber page, which lists two families side by side: 400/800 and 600/950 in one, 400/730 and 600/750 in the other. Same 400 µm core, two cladding diameters 70 µm apart, different end-face processing, different product. The cladding is not the core plus a bit. It is its own number, and it is doing its own work.
What work? Two things, and they pull in opposite directions.
The first is light guidance. The cladding’s refractive index relative to the core is what makes total internal reflection possible at all; a cladding thickness of a few wavelengths is enough optically, so everything above that exists for other reasons. The second is mechanical. A thicker cladding makes the fiber stiffer and more resistant to buckling and crush — good for a fiber you are pushing through a needle. But the bending strain on the outer surface of the glass scales with cladding diameter over bend radius, so at any given coil radius, a 800 µm-clad fiber is under roughly 10 % more surface strain than a 730 µm one. That is a fatigue life difference, not a cosmetic one, and it is why the minimum bend radius on the datasheet is tied to the cladding figure rather than the core.
There is a third consequence that has nothing to do with the glass. The cladding diameter is what has to fit inside a ferrule or a connector bore. That is where this specification stops being an optics question and becomes an assembly question, and it is the reason a fiber you liked on one probe can turn out to be unavailable in another — the customization side of the catalog is where those constraints get resolved.
The coating diameter is a thermal spec that everyone reads as a mechanical one
The coating is the polymer on the outside — acrylate for the general-purpose fibers, polyimide where temperature or chemical resistance matters. Its diameter is the number most likely to appear in a purchase order and the one most often misread, because the coating’s job is not to protect the glass from bending. It is to protect the glass from you — from the clamp, the tray, the friction of being pulled out of a coil, and from moisture-assisted crack growth at the surface.
Two things make the coating diameter a specification rather than a manufacturing detail.
It sets the temperature ceiling, and the ceiling is much lower than people assume. Standard UV-cured acrylate is conventionally qualified with an upper accelerated-aging temperature of 85 °C; IEC 60793-1-51 uses exactly that figure, and it is not a coincidence — it is roughly where the coating material itself stops being a reliable design element. Push a coated fiber’s end region past that and the failure is not gradual. The coating carbonises, the blackened polymer absorbs light far better than silica does, absorbed light makes heat, and heat makes more blackening. From there the damage can travel. The industry term is a fiber fuse, and it propagates back toward the source at metres per second — which is why the safety conversation about high-power fibers is about the laser’s case, not the tissue.
Polyimide moves that ceiling by hundreds of degrees, which is why it appears on fibers for contact work where the tip runs hot, and why it costs more. Which material you need is a question about the tip temperature in your protocol, not about the fiber’s price.
It sets the strip force, and strip force is a real dimension. The coating comes off the cladding under a controlled force, and that force is standardized: IEC 60793-1-32 is the test method, and the numbers are narrow. Average strip force for a standard coated fiber sits between 1.0 and 5.0 N; peak force may not exceed 8.9 N. Both bounds matter, and they fail in opposite directions. Too low and the coating slips or debonds in service. Too high and the person preparing the fiber — or the machine doing it — pulls harder than the glass can take, or the blade digs into the cladding to find purchase.
That is the entire reason the fiber stripper exists as a product rather than a suggestion to use a razor. Its stated design principle is No-Force Coating Removal: heat softens the coating so the blade separates it without cutting into the cladding. A scratch in the cladding is not a cosmetic defect. It is a scattering site, and light scattered out of the core lands in the coating — where, per the paragraph above, absorption leads to carbonisation leads to heat. Every coating removal step is a decision about whether you are creating that site.
And the coating diameter is the number that decides whether the finished fiber fits the needle, catheter or handpiece the customer is actually holding. We have been over that arithmetic elsewhere; the point here is only that it lives on the coating line, not the core line.
The number that is in none of the three
There is a fourth dimension in every fiber and almost no procurement document mentions it: concentricity — how far the core is off-centre inside the cladding, and the coating off-centre around the cladding.
For telecommunications fiber it is specified tightly (IEC 60793-2-10 caps coating-to-cladding concentricity error at 12.5 µm) because a splice or a connector depends on where the core actually is. For large-core medical fiber it is usually unstated, and it is the dimension that quietly decides how a fiber behaves at the tip. Grind a flat or radial end face onto a fiber whose core is off-centre and you have not made a symmetrical emitter — you have made a slightly eccentric one, with one wall of the tip closer to the core than the other. In a bare contact tip nobody notices. In a radial or side-firing tip, where the geometry is the product, off-centre geometry shows up as uneven heating around the ring.
If you take nothing else from this article, take the ordering rule: specify the core for the dose, the cladding for the mechanical and optical envelope, and the coating for the thermal route — and add a concentricity callout on any fiber whose tip geometry carries the clinical function.
What a usable specification looks like
The email at the top of this article failed not because the numbers were wrong but because it named three diameters without saying which of them was free. On the bare fiber page, the fibers are organised by core diameter with the coating material, NA and connector as separate choices — which is the correct structure, because it exposes where the real trade-offs sit.
A specification that will survive contact with a fiber manufacturer reads more like this:
- Core: the dose-critical dimension. Name it, then check it against your delivered watts using the table above before you negotiate anything else.
- Cladding: name the target and ask what it is tied to. If the supplier offers more than one cladding diameter on the same core — as the holmium fiber range does with its paired 272/420, 365/550 and 550/750 part numbers — that is a sign someone has actually thought about the tip, not just the glass.
- Coating: name the material and the diameter. The material sets your temperature ceiling; the diameter sets what the fiber fits into.
- Concentricity: at least ask. Silence on this line is the default, and default is not the same as good.
- Strip force and end-face condition: if you strip and cleave in-house, these are your yield; if your supplier does it, they are your incoming inspection.
Three diameters, three different failure modes, three different owners. A spec sheet that treats them as one column is a spec sheet nobody can build to.