It is crucial for a rotation at the back focal plane to translate into a translation at the sample plane, such that the optical trap loses as little power as possible. This is princi- pally accomplished by fixing the position of the microscope objective such that it does not translate, but instead having the sample stage move in x, y, and z (here, z is defined as the optical axis), which is contrary to commercial microscopes, which translate the sample stage in x and y but fix the stage in z. Of course, if one fixes the stage in z and assumes that the sample used will always be mounted at the same plane, then these two different strategies are the same in practice, as an optical trapping objective has at most a depth of focus of 100 µm (for expensive high numerical aperture water immersion objectives with IR wavelength correction), which translates to a beam deviation off axis for the largest deflections (1.85o as stated above) of 3.23 µm, which is around the expected precision of translation stages that are used to position lenses, and hence the accuracy of determination of the precise location of the back focal plane of the objective. However, in my case, the setup was also used for imaging embryos that require mounting at different sample planes, at millimeter distances inz, which results in beam deviations of >30µm. Thus, I fixed the objective in z.
along a straight path (which is defined by two irises), just as two points are required to define a line. Naturally, if the beam is to be expanded first, then those set of lenses are put into place. A mirror is placed on the mount which carries the objective, and reflects the laser back to its origin. In this manner, we can test that the laser is normal to the objective mount. On the objective mount is placed an iris, so we can make sure that the incoming laser is also centered, in addition to being normal. At this point, the imaging camera is placed at its appropriate location; if the laser mirror which reflects the laser into the objective is at 45owith respect to the objective, then the camera position is well defined. Since there is at least one dichroic in the optical path, we can also see the laser reflecting into the camera. Once the laser is properly normal and centered, we adjust the camera such that the laser appears in the center of the detector element. The tube lens is placed in front of the camera at the proper distance such that the beam is still in the center of the detector.
All that is left to do is to place any telescopes in the beam path. There will be at least one telescope if one chooses to use an AOD or set of mirrors to deflect the optical trap from its centered position. A telescope is therefore required to image the deflection onto the back focal plane of the objective. The lens closest to the objective must be at a distance of one focal length away from the back focal plane. This is achieved by placing the lens in the path with the objective in place. Since the beam is collimated going into the lens, and is focused into the back focal plane, the beam will appear collimated coming out of the objective. One should look at a spot far away from the objective to make sure that the laser is indeed collimated, and parallel to the optical axis. The second lens in the telescope is now placed at its proper location after the collimation and parallel requirements are met. The location is determined by removing the objective and once again putting a mirror on top of the objective mount. The light should be reflected back along the correct path, and when removing the mirror, the beam should be collimated at infinity. In this or similar manner, the lenses are put into place. When adding telescopes to the optical system (in general), I use a set of irises to define the optical path, and check to make sure that the beam stays a constant width (i.e., its collimated) after a long distance, after the beam has been expanded to its desired size. This is accomplished either through translation stages
(expensive) which have micron-level precision or through the use of clamps (cheap), which allow the optical elements to be placed arbitrarily on the grid defined by the optical table. Additionally, it is crucial that lenses are normal to the optical axis. Thus it is wise to define the height of the optical path with respect to the optical table as an initial condition. The normal condition is achieved by looking at the back reflection of the laser off the surfaces of the lens. There are usually two to three visible reflections: a reflection off the incident surface of the lens, and a reflection off the exit surface of the lens, and other higher order reflections, also depending on the whether the lens is a singlet, doublet, or triplet. When all the back reflections are aligned with the incident beam, then one can be sure that the lens is properly normal and centered about the laser.
There are a host of other mostly mechanical issues with how to mount the various components, what is the best sample stage, how to make filter wheels on the cheap, etc. But these issues are best learnt by trial and error (in addition to being too numerous to discuss), as each part has varying advantages and disadvantages depending on its contextual use, and more importantly, the manufacturer. I hope the above discussion is sufficient for the ingenue to begin trapping experiments. Small changes to the laser alignment produces total- internal reflection fluorescence (TIRF) microscopes, dark-field illumination, stroboscopic illumination, etc. Small changes to the detection side results in fluorescence correlation spectroscopy, and even confocal microscopy. These are all variations on a theme. An even larger issue is working within the constraint of a limited supply of funds, as optical components, to put it bluntly, are expensive. It is therefore imperative that the experimental physicist to develop machining skills in order to custom fabricate parts or to subtly modify existing ones to suit his or her needs at his or her whim. The experimental physicist should also possess basic electronics and programming skills to quickly whip up control schemes for instruments without spending the huge sums of money often required to purchase software or custom hardware that someone else built, often for profit - and which, moreover, are unsuitable to the exact needs of the experiment1. The acquisition of such skills and judicious
deployment of them provides a level of satisfaction that is absent from the exclusive use of manufactured products and instills a certain confidence in the scientist that carries through
1
For instance, we built a field inversion gel electrophoresis (FIGE) apparatus out of spare parts, the total cost of which was not more than $20. A commercially purchased FIGE apparatus would be in excess of $1000. We also built an automated filter wheel for multi-color microscopy out of servo motors and spare parts for a cost of $200; the cost for a commercial automated filter wheel would also be in excess of $1000. Of course, the polish of our home-built devices is, perhaps, lacking. And they might break more easily.
produce consistent and reproducible results. Science demands reproducibility because there
is an underlying logic to things. Because experiments are reproducible, it is simply up to the scientist to uncover what that logic is, for it is certainly there.