Optical fiber selection for uniform TIRF illumination

[Kyle Michael Douglass]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hi all,
In the past my lab has struggled to obtain uniform TIRF illumination
(i.e. free of fringes or speckle) from a certain commercial module.
Because of this, I would like to implement through-the-objective TIRF on
one of our custom-built microscopes using a fiber launch.

My main question is: what guidelines should I follow for the selection
of the optical fiber? My first instinct is to use a multi-mode fiber
because they have a higher power coupling efficiency and because mode
scrambling should prevent interference fringes from ruining the
illumination on the sample. The downside is that there will still be
speckle in the illumination profile, so I will need to mechanically
shake/twist the fiber to achieve good mode scrambling. We will be
imaging between 0.1 and 1 kHz.

What other factors am I not considering? Is there an optimum core diameter?

Thanks as always everyone,
Kyle

--
Kyle M. Douglass, PhD
Post-doctoral researcher
The Laboratory of Experimental Biophysics
EPFL, Lausanne, Switzerland
http://kmdouglass.github.io
http://leb.epfl.ch

[John Oreopoulos]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Kyle,

There's a good reason why TIRF systems of this type do not use multi-mode optical fibres. Though it is true the multi-mode fibre would offer a higher throughput of light from the laser source and it can be made fringe-free and speckle-free if shaken sufficiently, there is fundamental property of light that will prevent you from focusing down the light from this type of fibre to a small enough size at the back-focal plane of the objective lens (etendue - the optical equivalent of conservation of energy). You need to consider not only the core size, but also the NA of the fibre. The product of the fibre NA and the cross-sectional area of the fibre core is a constant.

For a well defined TIRF penetration depth, you need to focus the light down onto the back-focal plane to as small a size as possible. The reasoning behind this is as follows: Having a well-defined penetration depth means that all of the illumination light must approach the water-glass interface at a well defined angle. That angle is in turn defined by the radial position of the light focused at the back-focal plane, and hence the size/area of the spot of light located there. A larger spot focused at the back of the objective will cause the light to approach the interface at many different angles of incidence, and thus the sectioning ability of the evanescent wave will be compromised.

You can work through the mathematics associated with this scenario and you'll find that having an error of less than 10 nm associated with the penetration depth requires that the spot size at the back of the objective be on the order of 5-15 um. You cannot achieve that spot size with a multi-mode fibre because of etendue (believe me, I've tried). You can, get to sizes of that order with a single-mode fibre (which has a core size of about 5 um). Again, the importance of this is that if your beam is well defined in terms of angle of incidence (ie: a very collimated beam at the front focal plane of objective), you will be able to achieve very thin sectioning at the lowest penetration depths offered by the objective lens (50-90 nm depending on the NA of the objective).

Now I say all that, and then present to you the following paper which appeared fairly recently (open-access):

http://onlinelibrary.wiley.com/doi/10.1002/jbio.201500324/abstract

This paper presents an TIRF illumination method that does utilize multi-mode fibres, but if you read the fine print, you'll see that they do sacrifice some of the sectioning ability that could be achieved with a single-mode fibre. The main focus of this article is to use the multi-mode illumination method for localization microscopy where thin sectioning is not always or not strictly needed, and so in that sense there is some innovation here. But it's difficult from the images presented to say how this really would compare to TIRF implemented in the traditional sense using a single-mode fibre. The raw images are not available, and I'm not familiar with the structures being imaged in the cell samples they use. My instinct says that this just won't be as good as what can be achieved the "normal" way, but it may be worth revisiting in the lab. It's also questionable whether or not they really get the benefit of the higher throughput again because of etendue arguments. They have to dump most of the light outside of the objective aperture to get a TIRF image without any non-evanscent (widefield / sub-critical) light. And who knows where all that unused scattered light goes. If any of the authors of that paper are present here on the listserver, perhaps they can offer their perspective on this, I'd be interested to know.

Well, hope that sets on the right path (whatever that is). Do let us know how you get on if you end up trying this.

Cheers,

John Oreopoulos
Staff Scientist
Spectral Applied Research
A Division of Andor Technologies
www.spectral.ca


On 2016-06-01, at 6:01 AM, Kyle Douglass wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Hi all,
> In the past my lab has struggled to obtain uniform TIRF illumination (i.e. free of fringes or speckle) from a certain commercial module. Because of this, I would like to implement through-the-objective TIRF on one of our custom-built microscopes using a fiber launch.
>
> My main question is: what guidelines should I follow for the selection of the optical fiber? My first instinct is to use a multi-mode fiber because they have a higher power coupling efficiency and because mode scrambling should prevent interference fringes from ruining the illumination on the sample. The downside is that there will still be speckle in the illumination profile, so I will need to mechanically shake/twist the fiber to achieve good mode scrambling. We will be imaging between 0.1 and 1 kHz.
>
> What other factors am I not considering? Is there an optimum core diameter?
>
> Thanks as always everyone,
> Kyle
>
> --
> Kyle M. Douglass, PhD
> Post-doctoral researcher
> The Laboratory of Experimental Biophysics
> EPFL, Lausanne, Switzerland
> http://kmdouglass.github.io
> http://leb.epfl.ch

[Kwasi Kwakwa]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hi Kyle,

I spent a bunch of time playing with multi-mode fibers for TIRF illumination
(shameless plug, paper here
http://onlinelibrary.wiley.com/doi/10.1002/jbio.201500324/abstract) and so I
can confirm that your general idea is correct. You can get more power out of an
mm fibre and you don't get the fringes, but you do get speckle.

The size of the speckle depends on the core diameter of the fiber, so the larger
you go, the less of an issue it becomes. Also, the easier it becomes to scramble
those modes and achieve fairly even illumination. The problem is that larger
cores also mean a larger spot to image on to the back of your objective, which
makes TIRF a bit more complicated.

So my recommendation would be to use the largest core fiber you can safely
acheve TIRF with and then start thinking about how to twist/vibrate it fast
enough to even out the speckle.

Best,
Kwasi A. Kwakwa
Research Associate,
Photonics Group, Physics Department,
Imperial College London

[Kwasi Kwakwa]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hi John,

I'm glad some is actually reading that paper! I will have to agree with a lot of
what you wrote. Using a single-mode fiber you can more accurately position the
focused spot can more accurately define the penetration depth of your
evanescent field and so end up with better sectioning. The larger spot size and
NA means that while you do end up with TIRF, it is probably a lot harder to
accurately define your sectioning depth. Practically, for TIRF STORM, it works,
but for really accurate TIRF measurements I don't think I am in a position to
make extravagant claims.

As far as dumping some of the power of the laser outside the objective aperture,
the TIRF setup I used was originally designed for a single-mode fiber and so
magnified the MM spot excessively. It should be possible, with a redesigned
illumination setup, to image the spot from the smaller core MM fibers fully into
the annular region of the microscope objective necessary to achieve TIRF. It is
on my list of things to try out, but I am waiting to borrow a microscope frame to
try it out.

Best,
Kwasi A. Kwakwa
Research Associate,
Photonics Group, Physics Department,
Imperial College London