Assessment of the surface quality of the etched flat

The effectiveness of the etched cell design hinged on three factors- creating an accurate channel, retaining the surface finish of both the etched and non-etched parts of the flat and finally being able to form an accurate cell from the quartz components.

We employed a variety of techniques to assay the quality of the etched cuvette.

Channel sides (non-etched part of the flat): A Lapmaster monochromatic light

source (using a yellow helium light source) was in conjunction with a Zerodur® 1/20th of a waveband optical flat used to check the surface finish of the channel sides (figure 3.7). This was confirmed to be 1/20th of a waveband or an

approximate surface roughness of 30 nm over the region of the flat covered by the interference pattern.

Figure 3.7: Interference pattern of an etched flat. Note the band pattern on both sides of the channel, where optical flatness has been retained.

Channel base (etched part of the flat): A WYKO NT-2000 non contact optical

profiler, was used to generate images that show the etched surface of the channel.

Figure 3.8: WYKO image of a 6.5 µm etched flat channel / plain surface. The surface finish of both the etched and non-etched components is of exceptional quality, in terms of surface roughness and sharpness of interface. There is very little evidence of undercutting or smoothing at the interface, proving the efficacy of the acid mask.

The data shown in figure 3.8 are of excellent quality when compared to equivalent surfaces such as silicon or even polished quartz. The degree of surface quality on the initial substrate has been maintained throughout the etching process, creating an accurate sample channel. Of particular note is the interface between the channel and the non-etched face of the quartz flat, where there is very little evidence of undercutting at the etched/non etched interface. Typically undercutting occurs at the sides of the etched area, where as well as etching downwards, HF also attacks the sides of the channel as it is formed, leading to damage of the unetched quartz face and undermining the interface.

We also confirmed the data quality generated qualitatively by obtaining an interference spectrum of the etched flat (figure 3.9).

Figure 3.9: Assembled etched cuvette inside the sample cell holder. The interference pattern of the two flats can be clearly seen, identifying the surfaces of the two pieces of quartz as being optically flat.

As can be seen in figure 3.9, the interference pattern showed that the quartz cuvette created was indeed flat on both faces, hence the straight band interference pattern observed.

The WYKO data shown in figure 3.8 (one of the most accurate means of path length determination available) put the depth of the channel at 6.58 µm (± 0.10 µm), as given by the Rt value that represents the difference between the highest point on the surface and the lowest point on the surface. As the other flat is

depth of the channel is the effective path length of the cell. The depth of channel recorded by the WYKO was consistent with a measurement carried out using a Taylor Hobson Form Talysurf (data not shown), which gave a depth measurement of 6.63 µm (± 0.10 µm).

The interference fringe method7 gave 14 full fringes (figure 3.10) between 599 nm and 366 nm. From equation 3.1 it follows that the path-length is 6.586 µm (± 0.015 µm). Reassembling and repeating the measurements leads to ~1.5% variation (compared to 10–20% for a standard demountable cell) so the path- length of this etched cell is 6.6 µm (± 0.10 µm).

Figure 3.10: UV / Visible spectrum absorbance spectrum of the etched cell. The spectrum was acquired under the following conditions: Bandwidth 1 nm, data pitch 0.2 nm, scan speed 100 nm/min.

We also determined the path-length of the etched cell using a 0.2 M solution of potassium chromate (stabilised with a pellet of KOH). The absorbance of the solution was tested using a V-660 Jasco UV / Vis spectrophotometer.

-0.38 -0.26 -0.35 -0.3 350 400 500 600 650 Abs Wavelength / nm

Figure 3.11: UV/Vis K2CrO4 path-length determination. A 0.2 M solution was used to test the micron path-length cuvettes, whilst a 0.002 M solution was used for the 1 mm path-length cuvette. Parameters for the UV/Vis analysis were: bandwidth 1 nm, data pitch 0.5 nm, speed 100 nm/min.

Using the Beer-Lambert law with the absorbance reading at 372 nm from figure 3.11, and the absorption co-efficient of potassium chromate at that wavelength of 4830 mol-1dm3cm-1 it is possible to calculate the path-length of each cuvette. These are as follows:

1 mm cuvette = 1.041 mm (approximately 4% path-length error)

6.6 micron etched cuvette = 6.77 micron (approximately 3% path-length error) 10 micron Starna demountable cuvette = 4.104 micron (approximately 59% path-length error)

From this figure it can be shown that the error in the expected path-length for the etched cuvette is of a similar order to that of the 1 mm cuvette, whilst the 10 micron Starna demountable cuvette is very inaccurate.

3.3.5

Sample loading

The uniformity of the etched cell made loading the sample reproducibly even more challenging that with existing rectangular or cylindrical demountable cells available from e.g. Starna and Hellma. We solved this problem by grinding

0 0.5 1 1.5 2 2.5 200 250 300 350 400 450 500 550

Starna nominally 10 micron cuvette Starna 1 mm cuvette

Etched cuvette (6.6 micron)

A b s o rb a n c e Wavelength / nm

(using an Elliot 618 surface grinder) a 30º recess into the side of the optical flat as illustrated in Figure 3.7. This ground area of the flat facilitates sample loading and removal using a gel-loading tip piggy-backed onto a P20 Gilson pipette tip.