LED vs. Lasers for spinning disk/line-scan confocal?

Hi uCrew,

I’m getting quotes to add lasers to various spinning disks or line-scan systems we are considering and I’m realizing I don’t really understand why we need lasers for a spinning disk set-up (or necessarily for a line-scanner). If we have powerful polychromatic LED light sources (e.g. 500 mW/‘color’) and we only need about 100 mW of laser power, why couldn’t we use the LED in these applications (assuming coupling to an optical fiber)? I’m realizing I don’t know as much as I thought I did about the importance of coherent light, focal spots, etc. for these types of imaging, and using LEDs would potentially save 10-20k.



I hope others will chime in, but the crux is getting the LED light into the fiber.

A limited amount of light can be coupled from a LED source (usually a few mm) onto the fiber tip (a few um for single-mode fibers). Multi-mode fibers have larger core diameters and can accept more light. By shrinking the image of the LED to match the fiber size you must increase the light’s incidence angle accordingly, and the fiber has a finite acceptance angle. This is the concept of etendue.

As a practical example, look at Thorlabs fiber-coupled LEDs; notice they get single-digit mW power out of fibers with with 200um core diameters. This output power is limited not by the LED power, but by the fiber-coupling.

A useful feature of today’s laser sources (e.g. laser diodes) is that they are very bright on a per-area basis, which means they can be refocused to a very bright spot. I’m not sure how crucial this is for spinning disk confocal, but for point-scanning confocal this is essential.


Thanks, that’s exactly what I wanted to know! @JonD

I would also argue in simple terms that you cannot focus incoherent light as well as coherent light and unlike lasers, LEDs tend to not be coherent. Since the performance of your confocal/spinning disk system depends on the size of the excitation spot, having a larger spot size will degrade the performance of the system.


Sharing from discussion with @JonD some white papers have some useful comparisons between lasers and LEDs:


The portion under “monochromatic” in the second white paper is particularly relevant, besides coupling efficiency and optics.

So incoherent sources with particular spectral peaks (e.g. mercury arc lamp) were often used for spinning disks long ago. A polychromatic or “white” LED, due to it’s fabrication and operation, has much different spectral characteristics than halogen lamps often used in microscopy. Drive current is going to impact the spectral characteristics of LEDs. Furthermore, a typical bench-top diode laser usually incorporates some sort of feedback (i.e. a photodiode integrated in the laser diode package) and some sort of temperature feedback/stabilization (i.e. laser diode is mounted on a thermoelectric device). It is related to junction temperature, Boltzmann’s constant, and the various designs of semiconductor diodes.

Here is a paper about a “line” scanner using an LED. You might be able to use an LED… depending on which one of the dozens of schemes the “line” scanner implements. I’ve built various “line” scanner prototypes on optical tables, some using diode lasers (both fiber coupled and free space), some using white light lasers (super continuum), and some using LEDs https://www.spiedigitallibrary.org/conference-proceedings-of-spie/9330/933006/High-brightness-LED-in-confocal-microscopy/10.1117/12.2078191.short?SSO=1

Likewise you may be able to use an LED with a spinning disk. But probably not a “polychromatic” source. A multi-channel LED source might work. But you are going to have to work closely with the LED and scanner vendor to ensure that you’ve got the correct filters/dichroic and the system preforms according to your needs. If a third vendor such as Semrock or Chroma doesn’t have off-the-shelf filters that can perform well enough…they’ll have a minimum order quantity (MOQ) and non-recurring engineering (NRE) costs to fabricate custom filters… which is tens of thousands of dollars…negating any potential savings. Have to consider the whole system and your specific application requirements…