Diffraction Data Collection on Trial Gratings

Figure 5.1: Initial optical apparatus for 1−D diffraction grating.

An initial basic configuration, for diffraction analysis and for testing the image capturing possibilities (along with software and programming development), was set up using a com- mercially available diffraction grating with a period of 1.67 µm. The optical setup was designed to capture as much of the diffraction pattern as possible on the CCD detector chip in a digital camera, without unnecessary additional optics that could introduce errors to the image. To enable capture of the higher orders, and not the zeroth (corresponding simply to reflection from the surface), the laser beam was incident at an angle of 0◦ to the normal of the diffraction grating. Whilst the configuration, shown in Figure 5.1, worked well for small grating periods (<20 µm), the setup and image analysis had to be modified for larger peri- ods to capture the diffraction patterns. Due to the smaller angular separation of diffraction

orders with larger periods, the CCD camera housing obscured measurement of the first few diffraction orders when light was incident normal to the grating (see Figure 5.2).

Figure 5.2: CCD Housing limits the minimum angle of capture to the surface normal.

The experimental methodology was adapted for testing sample reflection gratings. A series of trial grating designs was fabricated on a chrome on silica glass mask plate by direct beam writing (fabrication was carried out by either InnosTM_{or RAL). The principle of operation of}

these gratings was the same as that for the biological gratings for which the setup was being developed. The contrast in reflection between the chrome lines and the silica substrate formed reflecting diffraction gratings. When the diffraction gratings were first tested, it became apparent that reflection from the surface of the silica was significant compared to the reflection by the chrome. This problem caused the captured diffraction pattern images to feature additional complexity due to multiple reflections between the top and bottom surfaces of the mask plate.

A solution to the problem of reflection of the incident laser beam from the surface of the silica, was to utilise the fact that the laser beam was polarised, and align the incident beam at the Brewster angle for glass. The Brewster angle θB, is the angle of incidence at which the reflected intensity for p-polarised light travelling from medium 1 to medium 2 is zero (see Figure 5.3), since the transmitted beam is orthogonal to the reflected beam [218]. The Brewster angle θB is related to the refractive indices of the two media (n1 and n2) and is

specified by the relationship:

5.2 Initial Diffraction Experiments 122 This means that p-polarised radiation incident on the silica surface at the Brewster angle is not reflected, and does not interfere with the diffraction pattern produced by the chrome. Light that is transmitted through the mask plate then encounters another refractive index change, leaving the silica mask plate and resuming propagation in air. Since the beam was incident on the top surface at the Brewster angle for air to silica (n1 → n2), the refracted

beam is also at the Brewster angle for the bottom surface, exiting the silica into air (n2 →n1).

Figure 5.3: When p-polarised light is incident on a surface at the Brewster angle, the reflected intensity drops to zero and all light is transmitted.

The optical apparatus was developed as shown in Figure 5.4. Using two mirrors in this configuration, it was possible to finely control the exact incident angle of the laser beam. The 632.8nmwavelength beam produced by the HeNe laser (Melles-Griot) provided around 1.0mW of power output; this was sufficient to saturate the CCD detector (Pulnix TM−1020) when reflected from the solid chrome lines. The beam intensity was attenuated by using a variable neutral density filter. The laser beam then passed through a polariser to maximise the degree of linear polarisation of the laser beam. A half-wave plate was used to rotate the polarisation of the beam to ensure thatp-polarised light was incident upon the mask plate. An adjustable mirror was used to align the angle of incidence to the Brewster angle giving the minimum reflected intensity. The diffraction pattern was subsequently captured on the camera’s CCD detector chip directly. The CCD camera was moved in an arc with a fixed observation distance by mounting an extension arm to a rotation stage centred in line with the point of incidence of the laser beam.

Alignment at the Brewster angle virtually eliminated top and bottom surface reflections for unpatterned regions of silica. However, this was not the case when the diffraction grating patterns were translated into the path of the laser beam. A secondary ‘ghost’ image was superimposed onto the original diffraction pattern as follows. Light was not reflected from the top surface of the mask (except that desired from the chrome reflection grating), but

Figure 5.4: Experimental layout developed for testing 1−D diffraction gratings. The CCD camera was moved in an arc to capture the whole diffraction pattern in multiple images.

transmitted light was also diffracted by passing through the chrome grating into the silica. Some of the transmitted light was, therefore, no longer incident on the bottom mask plate surface at the Brewster angle and was consequently reflected back. The image quality was further degraded by multiple reflections, and by the fact that light was diffracted again through the chrome grating as it leaves the silica. Also, due to the close spacing of different diffraction gratings on the mask, some of the diffracted light exited through ‘different period’ adjacent grating patterns. The net effect of all this interference was resulting diffraction images with little correlation to the expected diffraction pattern for the diffraction grating on the surface.

A method was tested to reduce light reflection from the back surface of the mask plate. Light was coupled from the mask plate to a second silica plate via refractive index matching oil. The bottom surface of the second plate had been abraded (by sand blasting) to create an uneven surface to diffusely scatter the transmitted light rather than reflect it (shown schematically in Figure 5.5). This successfully prevented the ‘ghost’ images in the observed

5.2 Initial Diffraction Experiments 124

Figure 5.5: To prevent reflections from the back surface of the silica mask plate from diffracted transmission light, a silica mask plate was abraded by sand blasting to provide a diffuse scattering surface.

diffraction image from the chrome grating, but added to the background radiation noise by scattering more light randomly in all directions. The background noise level was significantly lower compared to the high intensity of the reflected light for solid chrome lines, but was more of a problem for grating patterns with lower chrome surface area coverage.

The mask plate used was a silica substrate with a thin chrome layer (∼ 150 nm) on the top surface. In order to create a reference library of Fourier transform patterns relating to specific types of diffraction gratings, the mask plate was patterned with 46 different designs. These different patterns covered a range of periods from 1.0−20.0µm. The aim was to use the trial designs to determine the optimum grating period for the optical system. Additional grating patterns were created to analyse the impact upon the diffraction pattern of the ratio of line width to line spacing, or the effect of missing lines (1 in 3 to 1 in 10 lines missing). Some patterns were also generated that could test the effect of having fragmented lines made from dots rather than solid reflecting lines. Some sample images of these patterns (in photoresist prior to etching) are shown in Figure 5.6. The patterns were designed using a specialised mask design program, L-EditTM.

It was found experimentally that the effect of missing lines within the grating reduced the amount of interference and cancellation that is present in a normal grating. Consequently, if,

Figure 5.6: Example diffraction grating patterns created on a silica mask plate.

for example, 1 out of 3 lines was missing in a 10µmgrating, then 2 ‘sub-orders’ became visible in between the ‘normal’ orders of the original period. The periodicity of the missing feature was reducing cancellation and introducing diffraction orders corresponding to a periodicity of 30 µm in the diffraction pattern. This was the case for other ratios of missing lines too. In general, if 1 out of n lines is missing from a grating of period d, then the diffraction pattern will feature the normal grating orders corresponding to the period d, with (n−1) additional sub-orders in between main diffraction orders, relating to a period ofn×d. Figure 5.7 illustrates this effect diagrammatically, with the missing line indicated in the left hand column and the corresponding schematic diffraction pattern on the right-hand side, with sub-orders appearing in between the main diffraction orders of a regular grating.