Freezing Nucleation Apparatus (FNA-Basel), an instrument for immersion freezing assays

Lukas Zimmermann and Franz Conen

Departement of Environmental Sciences, University of Basel, Switzerland

Principle of operation

Transfer of light through water is reduced upon freezing because light gets scattered by inclusions in ice, such as air bubbles and brine pockets (Perovich, 2003). In immersion freezing assays the phase change from liquid to ice can therefore be detected visually when the sample changes from translucent to ‘milky’. Radiative transfer of visible light is increasingly reduced, the longer its wavelength. This is why large pieces of ice sometimes appear blue. Recording the intensity of a red light beam passing through a sample in a translucent vial allows reliable detection of the phase change from liquid to ice.

Application in an apparatus

Here, we present the example of an apparatus we built around this principle. It consists of an array of 8 x 7 red LEDs submersed in a cold bath and pointing upwards through 0.5 ml Eppendorf tubes containing between 0.1 and 0.4 ml sample liquid. Four tubes have a Pt1000 temperature sensor cast in. A camera in a black hood placed above the sample array looks down onto the lids of the tubes, which are illuminated from below. Images are recorded every five seconds. Light intensity in the area of each tube lid is extracted from each image and written into a file together with the temperature at the time the image was taken.

We have built two such apparatus. One is in use since June 2012 in Basel, one since September 2012 in Avignon.
A detailed description may be found here.

Freezing Nucleation Apparatus Overview

Critical steps


The LED array is submersed in liquid and exposed to a large temperature amplitude. Care has to be taken to avoid condensation on conducting parts. Our approach was to cast the circuit board in polyurethane in a polycarbonate case, with a silicone insulated power supply cable (total power supply 2.5 V DC, 0.20 A).

To avoid temperature gradients in the cooling liquid across the array of tubes, the cooling liquid should be stirred. The built-in stirring of a Lauda RC 6 cold bath is sufficient in our case, as long as the external circulation of the cooling liquid is short-cut and fully opened. We also left a horizontal distance between the LED array and the side-walls of the cold bath of at least 1 cm to allow vertical circulation. This way, the maximum measured temperature difference between tubes within the array is reduced to 0.5 °C at a cooling rate of 0.4 °C min-1.

Camera optics / geometry?


Pure water freezes without brine pockets and with few air bubbles, so no change in light transmission upon freezing is noticeable. Hence, it is essential that samples contain a small amount of salt, such as a buffer (e.g. 0.1 % (w/w) NaCl or more), which is also recommendable for reasons of sample stability.

Camera settings have to be such as to avoid overexposure. Otherwise, no noticeable change in light intensity may be observed, when overexposure is still rife after tubes have frozen. Equally, underexposure may result in too little change in light intensity upon freezing. Since samples may vary in their volume and light transmitting property, it may be necessary to readjust the camera settings (shutter opening time) accordingly.

Data analysis

The output file of the image analysis program (ucview) contains a time series (5 s time steps) of light intensities for each tube and of the four temperature readings. Freezing temperatures can be read off a graphical image produced from these data (Figure 2). It can also be extracted from the data file (csv) when copied into a spreadsheet (e.g. Excel or Openofficw Calc). One way to identify the freezing temperature is by calculating for each time step the relative difference between (a) the average of the last ten light intensities preceding it and (b) the three light intensities immediately following it.

The warmest temperature at which b < xa usually coincides with the onset of freezing. Depending on the sample type and volume, the factor x can vary between 0.87 and 0.93. However, for a specific kind of sample (e.g. 0.2 ml snow water buffered with NaCl (0.8 % w/w)) the value for x is pretty stable. Although convenient, this kind of analysis needs to be verified regularly by comparing the output for random samples with the graphical image, especially when new types of samples are analysed, because of different changes in light transmission upon freezing with different kinds of samples. Eventually, the factor x needs to be readjusted.

Example of a measurement (provided by Emiliano Stopelli, Uni Basel)

Pseudomonas syringae strain cc0094, grown for 3 days on agar plates with NB maximum media, then scraped and stored at 4 °C in 1 ml TP buffer (1L H2O + 8.75 g K2HPO4 + 6.75 g KH2PO4). Of this solution 100 µl into 10 ml TP buffer, final concentration around 107 cells/ml. Analysis of 52 samples à 100 µl.

Figure 1: Images recorded when all samples were liquid (left) and when all were frozen (right).
The corner positions are occupied by the temperature sensors.

Figure 2: (coloured lines) time course of light intensity as recorded for the six sample tubes in the upper row of the images shown in Figure 1 and temperature (black line).


Perovich, D. K. (2003) Complex yet translucent: the optical properties of sea ice. Physica B 338: 107-114.

last modified: 2012-12-19 Copyright 2012 by Franz Conen and Lukas Zimmermann W3C Validator