A confocal microscope is a scanning microscope – like a scanning electron microscope it scans the sample with a focussed beam and builds up an image point by point. The crucial difference – the fact that makes it confocal – is that in front of the detector (a photomultiplier tube or PMT) is a small pinhole.
The light source is a laser – or more than one to give a range of wavelengths. Nothing else can pack enough light into a small spot. The laser beam is focussed by the objective to a diffraction-limited spot – an Airy disk – on the specimen. The fluorescence (or reflection) from this spot is imaged by the same objective and brought to a focus at the pinhole.
These ray-paths are shown in blue. Outside the plane of focus we still excite fluorescence, so in a conventional microscope we would see blurred objects or just an overall haze spoiling the image quality. In the confocal microscope, light from outside the focal plane (red dotted lines) is smeared out over a wide area by the time it reaches the pinhole, so very little will go through. The confocal image therefore only contains in-focus information. By collecting a series of images, changing the focus between each, we can collect a full three-dimensional representation of our specimen. 3D rendering software can then give us a range of different views of our sample.
The lateral (XY) resolution is usually the same as in conventional microscopy. By making the pinhole very small (so that only a small part of the image Airy disk can pass) we can get slightly improved resolution – in principle about 150nm with an NA 1.4 objective – but this wastes so much light that it is impractical in fluorescence, which is where most cell biologists work. Therefore the pinhole is normally set to be the same diameter as the magnified Airy disk. The axial (Z) resolution is always worse, and unlike the lateral resolution (which relates directly to the NA) the axial resolution depends on the square of the NA. An oil lens (NA 1.4) will give a Z resolution of about 500nm, while a dry NA 0.7 objective will be 4 times worse, at 2 µ
Since we are collecting our image point by point we have to consider whether we have enough pixels to actually capture this resolution. The usual rule – the Nyquist criterion – is that to obtain Rayleigh resolution we need a minimum of 2.3 pixels within our expected resolution. So for 250nm resolution we need our pixel size to be ~110nm. To give a margin of error we might go to 3 pixels (~80nm) – but going any further is pointless. There is nothing to be gained, and we will just increase the bleaching of the sample.
A practical confocal microscope
Here we see a schematic of a simple confocal microscope. The laser beam enters (usually from an optical fibre) and is deflected down into the microscope by the primary beamsplitter – usually a dichroic mirror. Since a simple dichroic will need to be changed for different laser wavelengths, some makers use a double or triple dichroic which reflects two or three wavelengths while passing the rest of the spectrum. Another alternative is a polarizing beam-splitter which reflects polarized light (the laser beam) while passing non-polarized light (the fluorescence).
From there the beam passes via other mirrors and maybe lenses to the scanning mirrors which scan the beam across the sample. These have to be at a point conjugate with the back focal plane of the objective. (This means that a change in angle of the beam at the mirror will accurately translate into a change in position on the specimen).
Then the beam enters the microscope itself via a ‘photo eyepiece’ just like the one used for taking photographs from the microscope. The returning fluorescence is de-scanned – returned to a stationary spot – by the scan mirrors, passes through the dichroic and arrives at the pinhole. This must be accurately located at the image plane so that all the in-focus light will be at one spot and will pass the pinhole, while out of focus light is blocked. It follows that any chromatic or spherical aberration will seriously affect performance.
Beyond the pinhole the different wavelengths of the signal are sent to the appropriate detectors. In this schematic dichroic mirrors are used, and commercial systems offer anything from one to five detectors in such an arrangement. Barrier filters in front of the PMTs provide additional blocking of any stray laser light, and allow us to be more selective about which wavelengths we detect – we don’t have to take everything reflected by the dichroic mirror.
A more versatile alternative, which is now becoming popular, is to disperse the returning light into a spectrum with a prism or diffraction grating and direct the region or regions we are interested in to particular detectors. We are then no longer limited by what filters are installed, and we have the added bonus that we can actually measure the spectrum of the fluorescence and then decide how to detect it most effectively.
PMTs have a very wide dynamic range, but just as in any method of image recording we need to get the exposure right, and the microscope will have a false-colour palette to assist with this. Always reduce the laser power to the minimum needed to get a good image before making the final adjustments to the PMT gain. Acquiring a confocal 3D stack will always be harder on your specimen than recording a single wide-field image, but taking time to optimise imaging conditions will go a long way towards minimising damage.