TIRF objective for routine imaging
Locked
12
12
 
|
James Pawley |
Re: TIRF objective for routine imaging
|
Hello Jim, Thanks for filling me in on the subject, it was as clear and concise as I could have asked for. Could you give an example of the loss in spacial resolution the 40x would have compared to the 60x or 100x assuming the N.A (which being 1.3 theoretically should have a optical resolution of 0.2115 um at 0.550 um) is the same. Thank again! Pete Hi Pete,
It all depends on the optics.
Are you sitting comfortably? Then I'll begin...
Most confocals use the end of a single-mode fiber as the virtual
point illumination source. The light leaving the tip of such a fiber
forms a cone of expanding spherical waves, centered on the tip. The
cone angle is determined by the NA of the fiber (Yes, fibers have them
too. Depends on their size and the RI of the core and cladding.) and
the wavelength of the light (longer wavelength, wider angle).
At any plane perpendicular to the axis of this cone, the light
pattern resembles an Airy Pattern with a bright central maximum and
fainter rings. If you place a lens along this axis, positioned so that
it is one focal length from the end of the fiber, at that location the
cone will be converted into a parallel beam that also looks like an
Airy Disk in cross section. If it is a good Airy disk (nicely polished
fiber, no dust etc.), about 80% of the total light will be in the
central spot and the intensity distribution of the light in this
central spot will have close to Gaussian in profile. As the beam is
parallel, it will progress through the various mirrors, dichroics and
filters without much change, eventually arriving at the BPF of the
objective (OK, not quite. There is a tube lens in there somewhere, so
what I really mean is that the effect of the tube lens plus the
parallelizing lens that I described as being "one focal length
from the fiber tip," will together provide a parallel beam in the
objective BFP.)
An aperture may be used at this point to remove the 20% of the
light in the outer Airy rings. The size of this aperture and of the
more-or-less Gaussian central maximum will be proportional to the
focal length of the parallelizing/tube lens combination because, with
a longer focal length, the cone will have expanded more before it is
made parallel.
If you set things up to make the aperture/central-beam-diameter
the same as the entrance pupil of the objective (basically the
entrance pupil is the diameter of the objective's BFP), then the
objective will be "filled" with a quasi-Gausian beam. As
such a beam has lower intensity for the high-NA rays than would be the
case if it were illuminated with a "uniform parallel plane wave"
(the magical wave we see in the optics textbooks), the focus in the
specimen will be closer to a Gaussian than to an Airy Disk: no rings
but with the diameter of the central spot slightly larger (Any perfect
lens converts the light pattern at its BFP into the Fourier Transform
of this pattern in the focal (image) plane, and the maths say that the
Fourier Transform on any Gaussian is another Gaussian).
On the other hand, if the central spot of the parallel light beam
is say, 10x larger than the entrance pupil of the objective, then the
intensity will be approximately constant across the entire BFP and you
will get an Airy disk of the size and shape predicted by the Abbe
Equation. The only problem is that about 85% of the laser light will
be "wasted" (i.e.,"not in the final beam" implying
the need for a somewhat larger laser, and in addition the excess light
may produce a stray light problem.)
From this you see that merely saying that a beam "fills the
BFP" is really not very precise: Do you mean "just fills
with a quasi Gaussian" or "overfills" and if so by how
much? (and all this assumes that everything is properly aligned and
the center of the Gaussian spot really does coincide with the optical
axis etc.)
From all this, you can see that, if you want to change the size
of this beam at the BFP, you need to change both the position and
focal length of the parallelizing/tube lens system (i.e., you need
a zoom lens). Furthermore, you want to do this in a way that works for
all the laser wavelengths that you plan to use (i.e., you need a zoom
lens with no aberrations). Not a simple task. As Guy mentions, Leica
does this, at least to some extent. Perhaps he will be willing to tell
us how they do it, and the the BFP-beam diameter and wavelength range
over which it operates. Maybe even what compromises are
employed?
(Over to you Guy!)
As far as "How much will this sort of problem reduce
resolution?" Well, a shortish answer might start by splitting it
into x-y resolution and z-resolution.
x-y resolution,
Because the light signal comes from a volume of the specimen that
"has been imaged twice," the x-y resolution of a "truly
confocal" microscope yields an effective Airy spot only about
0.707 as wide as one would get using the same optics for wide-field
imaging (i.e, 1.4x "better" resolution). However, this only
occurs if you use a VERY small pinhole (say 0.1 Airy units), and, as
such a small pinhole implies throwing away about 98% of the signal in
the central spot, in biology we can virtually never take advantage of
this possible resolution increase. As a result, the x-y resolution is
essentially the same as that with widefield. In other words, as long
as the "signal-collection" pathway is operating at full NA
and using a pinhole smaller than one Airy Unit, the collection path
will determine the x-y resolution, (almost) regardless of how big the
illuminating beam might be because of an underfilled illumination
pupil.
On the other hand, there is still a price to pay: you will be
illuminating a large spot but only collecting signal from a smaller
spot in the center of the illumination, as constrained by the
"signal-collection" optics. i.e. you will be creating a lot
more damage than you need to. Of course, you could open the pinhole
wider so that signal from the entire, larger illuminated spot passes
through the pinhole. Then you would collect all the signal you were
exciting but the x-y resolution will now be determined by the
large pinhole aperture and will be roughly what you would expect had
you used an objective with an NA the same as that which you actually
"filled" (i.e., BAD!).
Z-resolution
As Z-resolution is inversely proportional to (NA)*2, it will be
even more seriously affected by having an underfilled objective.
Z-resolution =
where
Suppose we compare a 40x NA 1.2 to a 100x NA 1.2. Assuming that
the illumination optics just fills the pupil of the latter with a
Gaussian beam, then for the 100x, NA1 equal NA2 and we get "full
z-resolution" for this optical set up (other variables hidden in
"k" include wavelength, pinhole size, freedom from
aberration etc).
For the 40x, however, NA1 is 2.5x smaller than NA2 and therefore,
we would expect that the z-resolution would be 2.5 worse
(larger).
Were you to open the collection pinhole to collect "all"
of the in-focus" signal from the larger illuminated spot (as
described above), the z-resolution would be about 6.25x worse.
Although this math would not "cut the mustard" at
OPTIK, I think it gives a useful approximation of the likely
results.
Cheers,
Jim P.
*********************************************************************************
Prof. James B. Pawley, Room 223, Zoology Research Building, 1117 Johnson Ave., Madison, WI, 53706 3D Microscopy of Living Cells Course, June 12-24, 2010, UBC, Vancouver Canada Info: http://www.3dcourse.ubc.ca/ On Mar 23, 2010, at 19:37 PM, James Pawley wrote:
-- |
|
|
Paul Rigby-2 |
Re: TIRF objective for routine imaging
|
|
In reply to this post by James Pawley
Hi All,
A fascinating and extremely informative discussion, yet again. I will only add one small bit of practical information that I have observed from direct measurement. In recent testing of objectives, we measured reflections from a semi-silvered mirror (0.17mm coverslip etc) for two objectives (Nikon 60x, 1.49NA TIRF and Nikon 20x PlanApo VC, 0.7NA?) across four wavlengths (405nm, 488nm, 561nm and 638nm). Pinhole set to 1.0 Airy. The 20x VC objective showed almost no spherical abberation with barely discernable misalignment of the reflected images in the Z direction. However, the 60x TIRF objective (not a VC design) did show good coalignment of the 488 and 561nm reflections, but both the 405 and 638nm reflections were displaced in Z either above and below the 488 and 561nm images. I suppose this was not unexpected and I will say that the abberation correction collar of the TIRF objective may not have been fully optimised. Would we expect both the 405 and 638nm lines to correctly align in z if the collar was correctly adjusted or would we likely only bring one wavelength into correct alignment? Do people think purchasing the Nikon 60x 1.4NA Planapo VC objective is worth the extra cost? Thanks again to everyone for the enlightened discussions. Paul Rigby Dr Paul Rigby Associate Professor Centre for Microscopy, Characterisation & Analysis (M510) The University of Western Australia 35 Stirling Highway Crawley WA 6009 Phone (61 8) 9346 2819 |
|
|
James Pawley |
Re: TIRF objective for routine imaging
|
|
>Hi All,
>A fascinating and extremely informative discussion, yet again. > >I will only add one small bit of practical information that I have >observed from direct measurement. In recent testing of objectives, >we measured reflections from a semi-silvered mirror (0.17mm >coverslip etc) for two objectives (Nikon 60x, 1.49NA TIRF and Nikon >20x PlanApo VC, 0.7NA?) across four wavlengths (405nm, 488nm, 561nm >and 638nm). Pinhole set to 1.0 Airy. > >The 20x VC objective showed almost no spherical abberation with >barely discernable misalignment of the reflected images in the Z >direction. However, the 60x TIRF objective (not a VC design) did >show good coalignment of the 488 and 561nm reflections, but both the >405 and 638nm reflections were displaced in Z either above and below >the 488 and 561nm images. I suppose this was not unexpected and I >will say that the abberation correction collar of the TIRF objective >may not have been fully optimised. > >Would we expect both the 405 and 638nm lines to correctly align in z >if the collar was correctly adjusted or would we likely only bring >one wavelength into correct alignment? Do people think purchasing >the Nikon 60x 1.4NA Planapo VC objective is worth the extra cost? > >Thanks again to everyone for the enlightened discussions. >Paul Rigby > >Dr Paul Rigby >Associate Professor >Centre for Microscopy, Characterisation & Analysis (M510) >The University of Western Australia >35 Stirling Highway >Crawley WA 6009 >Phone (61 8) 9346 2819 > Dear Paul, Congratulations on taking the trouble to actually make some (very interesting) measurements! And also congratulations for using an actual metal mirror rather than just a water-glass interface for your reflective surface (The latter preferentially reflects high-NA rays, giving an anomalously good z-resolution. For others who may try this, don't forget to use a very low laser power and a low PMT voltage.) However, may I suggest another interpretation of your result: what you saw seems more likely to be chromatic aberration, not spherical. The latter can be corrected by using the collar (the adjustment of which which will also cause a Z-shift). The former should not be affected. Of course, the two do interact somewhat because SA is caused when the RI between the objective and the focus plane not being exactly as specified by the manufacturer (the temperature as well as the material present can be important), while CA obviously is about how RIs of the lens elements and the oil etc change with wavelength. In other words SA is always corrected more completely at some wavelengths than at others. However, we usually think of a focal shift, such as the one you have detected, as being primarily indicative of CA. Any remaining SA in the z-PSFs that you recorded will produce asymmetry in the z-direction (The way that the reflected light signal decays above, vs below the maximum will look different when focus is above vs below the peak. Usually SA produces a longer rippled tail in one direction). In practice, the focal shift caused by chromatic aberration can also reduce z-resolution. Imagine a dye that is excited at 561nm and emits at 633nm. The total confocal PSF will be a product (actually a convolution) of these two PSFs and it will spread out in the z direction much more than would the PSF recorded at either wavelength. Did you happen to measure the focal shift in microns? Or their extent (full-width at half maximum in microns)? The latter should be in the range 500-700 nm depending on wavelength and pinhole setting. But if the (very large) pupil of the 1.49 is not fully illuminated, the performance will be somewhat less as discussed previously. Not owning this lens, I have no specifications other than what I can find on the WWW, but this seems to indicate that both the 1.49 TIRF objectives, as listed at http://www2.nikonusa.com/fileuploads/pdfs/TIRF%20WTIRF%20Brochure%202006.pdf are Apochromats. This means that they are supposed to be corrected for at least three wavelengths. Although it doesn't mention them being a plan lenses, I expect that they cannot have much field curvature because the TIRF object is extremely planar. Plan-APOs are often corrected for 4 wavelengths. So something seems wrong. On the other hand, what do we mean by "corrected for"? When we were viewing by eye or using color film, it meant that one was able to view the image of a thin, black & white pattern in transmitted light without the edges of the features having red or blue fringes. In practice, this meant that the wavelength induced focal shift was less than the depth of field. However, in confocal microscopy, we can LOCATE THE POSITION of the peak reflection far more accurately than we can RESOLVE two features separated in z (i.e., Z-position can be determined to a precision much smaller than the z-resolution.). I am assuming that this is what you have measured. Quantitative focal-shift vs wavelength curves can be found in Figure 7:15 in The Handbook. Alternative explanations might involve CA in the other optical components between the end of the fiber and the back of the objective. You can see how this sort of spectral displacement would normally be less trouble in a TIRF lens that is mainly used to image light from a single fluorophor at any one time and where the z-position of the excitation is set not by the focus position so much as by the location of the coverslip/water interface. In any case, if you do need the best possible TIRF x-y resolution from two fluorophors at the same time on the system you have measured, I suggest that you choose fluorphors emitting near near 488 and 561nm, otherwise one or the other may be slightly out of focus (depending on the magnitude of the focal shift) Happy TIRF! Jim P. -- ********************************************************************************* Prof. James B. Pawley, Ph. 608-263-3147 Room 223, Zoology Research Building, FAX 608-265-5315 1117 Johnson Ave., Madison, WI, 53706 [hidden email] 3D Microscopy of Living Cells Course, June 12-24, 2010, UBC, Vancouver Canada Info: http://www.3dcourse.ubc.ca/ Applications still being accepted "If it ain't diffraction, it must be statistics." Anon. |
«
Return to Confocal Microscopy List
|
1 view|%1 views
| Free forum by Nabble | Edit this page |