Analysing Ice Samples From Antarctica (Part 1)
Antarctic Sea Ice Research in Cape Town
"In the winter and spring of 2019, a group of researchers and students from the Marine and Antarctic Research Centre for Innovation and Sustainability (MARiS) at the University of Cape Town travelled to the Antarctic Marginal Ice Zone on board the SA Agulhas ll to collect sea ice samples for study. This area of the Southern Ocean around the Antarctic continent has not been explored and studied until recently and the sea ice in the area has yet to be characterised. One such study on sea ice is the investigation of the structure of sea ice using CT scanning techniques, which has several logistical difficulties, epecially as ice melts at room temperature. With the design of a cooling chamber that keeps the ice samples at a constant sub-zero temperature, the CT scanning has been largely successful in visualising the ice and salty brine internal structure of the sea ice. These images, once published, will be the first of its kind in the sea ice community with high resolution images that are able to clearly distinguish the different features of the samples. This exciting work has been possible thanks to the great help from Paul and Colm at X-Sight, and we look forward to continuing this ground-breaking science."
Technical Challenges of Scanning Polar Ice
To prevent the ice from melting during the scan, a small freezer chamber was constructed using a glass jacket of ethylene glycol circulating at -10 degrees. The ice sample was then placed inside the chamber while scanning.
It was found that drag imposed by the glycol circulating pipes during rotation caused the freezing chamber to move during the scan. To overcome this, the glass chamber was bonded directly to a manipulator centre mount. This required a bonding agent compatible with extreme cold and wet conditions. After a few "trials and errors", Pratley Quickset (rear-view mirror glue) worked well.
Circulating pipes had to be managed during the scan rotation to prevent mechanical fouling and kinking which would prevent circulation and cause the ice to melt. This was achieved by suspending the circulating pipes from above with enough slack and degree of "pre-twist" allowing the pipes to untwist during the scan rotation. This along with close monitoring saved any ice from melting.
In order to image the ice, x-rays would first need to pass through the glycol and freezing chamber glass walls. It would be necessary to remove these additional densities from the scan data leaving as much dynamic range for the ice samples as possible. By imaging the chamber with no ice sample inside, a white reference image depicting no attenuation of x-rays due to the ice samples was created. This image was used to calibrate the scan image to give attenuation due to the ice sample only. A new calibration image was captured for each scan. It was found that this image was being corrupted by the formation of frost on the chamber walls introducing artefacts into the CT data. To prevent this, a second glycol filled chamber was used to capture the calibration image at room temperature free from frost. As much frost as possible was scraped from the chamber before each scan.
During the initial configuration the glycol circulating and cooling pump was located inside the x-ray enclosure. During a scan the overall heat inside the x-ray enclosure increased due to the additional heat pumped out by the cooling system and circulating pumps - this increase in temperature had a direct effect on image quality. By reconfiguring the equipment with the cooling and circulating pump placed outside the x-ray enclosure and using fans to assist in cooling the detector, better image quality was obtained.
Given the aggresive frost accumulation during the scan, many towels were needed to absorb melting frost and prevent this from ingressing into sensitive system components.
On top of all of this, of course, was the ever-present challenge of load-shedding. We had to allow an hour and a half to pump the set-up down to -10 degrees; each scan took 66 minutes; and then we had to capture a new calibration image between each scan. So it took about two and a half hours to do the first scan each day - plenty of time to be sabotaged by Eskom!
Our next newsletter will contain some examples of the data we managed to achieve.