microscope slide 2mm
200µm
3mm 8mm
PMMA platelet cover slip
#1.5 ( 1.7µm) step 1
step 2
Figure 5.2: A new version of the sandwich cell, schematic view (left) and one of the samples (right).
The 3mm bore hole in the center piece is not only the filling entrance, it also holds the air bubble in place during measurements. Note that filling is done upside down compared to the sketch, directly after step 1. In the microscope the cell is held upside down compared to these pictures.
In order to hold the air bubble in a place where it cannot disturb the measurement a new type of cell was invented with a template made of acrylic glass (PMMA). A milling groove with a diameter of 8mm and a thickness of 200-500µm serves as the sample volume (see Figure5.2). A 3mm bore hole close to the border of the sample volume has a twofold function: It is the entrance for the suspension and it is there to hold the air bubble in place. Physically the air bubble wants to be in the bore hole as this position minimizes its surface.
After cleaning the cover slip is glued onto the PMMA platelet with the UV curing glue (step 1 in Fig.5.2).
After that the cell is turned over so that one can easily fill in the suspension via the borehole, usually using a Pasteur pipette made of glass. As opposed to the old one does not need a syringe, where there is always a certain amount of lost suspension that sticks in the tube and the needle. This is a big advantage if only small quantities of the sample are available. Since the entry channel to the sample volume points upwards, it is easy to seal the cell with the microscope slide in a way that the uncured glue does not come into contact with the suspension. In order to facilitate the production the sample volume used here was a thin cylinder. However, one could think of other shapes, for instance a long box with a bigger volume.
Another idea is to make the template of glass. In this form, only using glass parts, the sample cell has become used with great satisfaction in the chemistry group AG Zumbusch. Nevertheless, in combination with PMMA particle suspensions, no problems occurred with the centrepiece being made of PMMA.
For quick checks of samples it was very convenient to use rectangular glass tubes glued on a microscope slide (VitroTubes, VitroCom). The ones that came in use here had an inner thickness of 200µm, a width of 4mm and a length of 5cm. Since their wall thickness (also 200µm) does not fit perfectly to the microscope objectives, they were not an option for precise measurements. The image quality was a lot worse with those compared to measurements with sandwich cells.
5.2 Confocal Microscope
For all the microscopy experiments a disc scanning confocal microscope was used, which is by design tuned to allow high image refresh rates. Another advantage compared to the more common laser scanning confocal microscopes (LSCM) is the greatly reduced bleaching of the fluorescent dyes: Since an LSCM scans one point after the other, a high excitation intensity is necessary in order to obtain enough light from the sample in a short time. With the disc many points are illuminated at once, which allows for a longer effective scanning time per point for both excitation and light collection.
Figure 5.3:Schematic picture of the confocal scanning unit (adapted from [90]).
5.2.1 System parts
The main parts of the system can be summed up very briefly. A laser serves as the light source. The light is guided to a confocal scanning unit which is attached on one side to a light microscope and on the other side to a camera. Image recording and operation of the system is done with a computer using a commercial software. In the following paragraphs all the parts are described in detail.
Laser
Light source is an argon/krypton mixed-gas laser (Coherent Innova 70C) with three main lines covering the colours red (647nm), green (514nm) and blue (488nm). While the red line is ideal for particles with the NIR dye, the green line suits well to rhodamine and the blue line to fluorescine. All wavelengths together yield a power output of at most 4.5W. Single lines are in the range of 1-2W at the maximum output. Since the light is guided through a single mode optical fiber (cut-off wavelength 430nm) the maximum power input into the confocal scanner is reduced to about 600mW. Maximum power was only required for the NIR particles, for all others an input of about 50-100mW was sufficient.
Nipkow disc scanning system
At the entrance to the scanning unit (Perkin Elmer/Yokogawa CSU-100) light coming from the optical fiber is simply collimated by guiding it via mirrors and tubes through the excitation filter towards the scanning discs (see Figure5.3). As there are no lenses for collimation, only the relatively flat tip of the Gaussian intensity profile reaches the discs, which leads to a more uniform illumination. However, for the NIR particles the usage of an adjustable lens collimator at the end of the fiber was required in order to get more intensity. It was possible to tune the illumination to be as uniform as without the collimator.
Passing through the first scanning disc, the light gets focused by helically arranged microlenses. Each microlens has its own pinhole on the second disc, and effectively each pinhole acts as a point light source.
The objective lens focuses the light into the sample where it excites fluorescent light. Of this emitted light, only the part that takes the same way back as the exciting light can pass through the pinhole. That’s why only light emitted from the focus contributes to the recorded image. Moving the focal plane is done simply by moving the microscope objective. Since only the light from the focal plane contributes, the images are 2D slices through a 3D sample. Depending on the working distance of the objective, slices from a depth of up to100µm inside the sample can be recorded.
5.2 Confocal Microscope The dichroic mirror between the discs is designed to let through all three possible excitation wavelengths from the laser, while most of the wavelengths emitted from the fluorescent samples are deflected to the camera. Additionally, there are excitation and emission filters that restrict the illumination and the detection wavelengths further. This avoids unnecessary bleaching of the dye or overexposure of the CCD chip. The discs rotates at a speed of 1800rpm. Lenses and pinholes are arranged in a way that each rotation illuminates the field of view 12 times, yielding a maximum possible frame rate of 360fps.
Microscope
The scanner unit is connected to a commercially available light microscope (Axiovert 200, Zeiss), with an objective that is movable in the𝑧-direction using a piezo nano-positioning system (P-721.10, PI) con-trolled by the computer. The minimum allowed step size of the piezo system is 0.1µm, due to restrictions of the software. The direction of view is upwards and the sample cell is fixed in a holder sitting on two manual micro positioning stages allowing a precise adjustment of the𝑥and𝑦position of the sample.
A 100× oil objective with a numerical aperture of 1.45 and a 63× water objective with a numerical aperture of 1.2 were used in the experiments. Both objectives require an immersion oil between sample cell and the outer objective lens. Using the water objective for samples with refractive index of about 1.49 required the𝑧values from the piezo software to be divided by a factor of 1.19. This number was determined from fitting the 2D structure factor of an isotropic sample in𝑥-𝑦direction to the one in𝑥-𝑧 direction.
CCD camera
Image capturing was done with a 12-bit CCD camera (ORCA-ERG, Hamamatsu). It allows a maximum resolution of 1344×1024 pixels at a frame rate of 8.8fps. One can choose2 × 2binning or4 × 4binning which means that the readout of 4 or 16 pixels is combined. This allows higher frame rates of 16fps or 27fps, respectively. Thermal noise is reduced with a Peltier element that cools down the CCD chip.
The latter has a good quantum efficiency of about 70% for visible light. Both, noise reduction and high efficiency are important for long time measurements. They allow for a relatively low excitation intensity, which makes the fluorescent dies less prone to bleaching. The exposure time (actually the integration time) of the CCD chip must be carefully adapted to the rotation of the Nipkow discs. Otherwise the image quality suffers from blinking stripes that originate from the moving pinholes and microlenses.
5.2.2 Recording images
For measurements with smaller particles (∼ 1µm) the 2×2 binning mode of the CCD chip was used.
Together with the 100×oil objective this yields a pixel size ofΔ𝑥 = Δ𝑦 = 0.1245µm. Medium sized particles (∼ 1.5 − 2.0) were recorded with the same objective using 4×4 binning, which doubles the pixel size. For the biggest particles (≳ 2µm) the resolution of the 63×objective with 4×4 binning was enough, corresponding toΔ𝑥= Δ𝑦 = 0.4008µm. With the 100×objective the field of view was about 50µm×60µm, with the 63×one it was about 80µm×100µm.
For 2×2 binning the exposure time was always adjusted to be61ms, which allows the maximum frame rate of the camera in that mode (16fps). Faster recording was possible by using 4×4 binning. With an exposure time of36ms the maximum possible frame rate was 27fps. For the recording of 3D images one has to keep in mind that the movement of the objective also takes its time. Usually, one frame recorded by the camera has to be skipped while the piezo is moving to the next𝑧 position. In most of the 3D measurements stacks of 30-120 images were recorded with a distanceΔ𝑧 = 0.3µm between the slices
(corresponding to ∼ 10-40µm in depth). The recording time for one stack is 2-10 seconds. To avoid any effects from the flat surface of the cover slip all measurements were done at a minimum depth of 𝑧 = 15µm. A particle with a diameter of 1.5µm occurs in about 6 slices which means that the particle usually moves during the recording. This results in the finite exposure time problem which is described in full detail in the particle detection chapter (see Section 3.4). In practice, for most of the physical quantities, only 𝑥and𝑦coordinates of the particles are used. They can be determined from single 2D slices that have a much shorter exposure time. However, for the𝑧coordinate one always needs to rely on the combination of several slices.
With the 100×objective the accuracy in the determination of the particle positions is about 15nm in𝑥 and𝑦direction and 30nm in𝑧. For the 63×objective the errors are larger with∼ 34nm and∼ 110nm.
This is discussed in Section3.3of the chapter on particle detection.
Figure 5.4:Top panels: Raw 3D image from a binary suspension (1µm and 2µm particles), slices in the 𝑥-𝑦(left) and𝑦-𝑧(right) plane. Bottom panels: The same slices after flattening the intensity profile by adjusting the intensity so that all 2D slices in all three directions have the same average intensity. Pixel sizes areΔ𝑥= Δ𝑦= 0.1245µm andΔ𝑧= 0.3µm.
5.3 Experimental issues and challenges