Search

Diatom frustules are an interesting tool for climatic and ecological research and they have for some time been employed in isotope research work as a paleothermometer. Diatom frustules are Said to be very stable over a long period of time, and thus the stable oxygen isotopes within the frustule's SiO$_{2}$ can under certain conditions represent a good proxy for climatic or ecological reconstructions (Juillet, 1980; Juillet & Labeyrie, 1987; Shemesh et al., 1992). Especially in those cases where carbonates are not available for climate research, diatoms seem to be a very promising alternative. This is for example the situation in the high resolution sedimentary sequences of Westeifel Maar lakes. The sedimentary sequences consist of turbidites, siderit-laminites and annually laminated diatomaceous gyttja but mostly lack any carbonates (Negendank & Brauer & Zolitschka, 1990). The problem, however, is to get highly purified diatom material for reliable oxygen isotope investigations. This implies that minerogenic impurities have to be separated since they can easily alter the isotope signal of a diatom layer and, thus, lead to wrong results. Additionally it is in many cases desirable to distinguish between diatom fractions of different sizes, i.e. to distinguish between different diatom species. It therefore was the aim of this work to develop a new method for separating quantitatively diatom frustules from lake sediments for investigations of the stable oxygen isotope composition in diatom SiO$_{2}$. This method uses a new technique for separating particles depending an their sinking velocity in a liquid, streaming through a narrow channel. The technique to be used splits particles hydrodynamically into different size classes, a procedure which is called SPLITT-fractionation (split-flow lateral-transport thin Separation cells) and was first presented by Giddings (1985). As with field-flow fractionation (FFF) techniques, thin flow channels in conjunction with a field (gravitation) applied perpendicular to the flow are used for SPLITT-fractionation. A sample suspension containing the particles to be separated is continuously introduced through one inlet into the cell while Sediment-free liquid is introduced into the cell through a second inlet. Within the cell the two flows merge smoothly, the suspension overlying the sediment-free carrier flow, resulting in a laminar flow in which the Separation of particles takes place. Within the laminar current grains migrate according to their density, shape and ABSTRACT size towards either of two outlets. As a result the sample will be divided into two fractions around a cut-off diameter, which is controlled by the ratio of the two outlet flow rates. As diatoms behave hydrodynamically different compared with mineral grains, the described method provides a tool to separate them by SPLITT-fractionation. The advantages of SPLITT-fractionation over other separating methods for diatoms are reproducibility, high throughput by continuous flow, minimum losses, fast procedure and minimum contamination of the sample. The SPLITT cell used in this study has a length of 20 cm, a breadth of 4 cm, and a height of 371 $\mu$m. The Sample concentration used was below 1 weight percent to minimise particle-particle interaction which spoils resolution. Sediment samples of different origin, age, compaction and diatom amount were tested. In a first step a method was established to suspend the material in water with H$_{2}$O$_{2}$ (30%). In this process also most of the organic matter was removed, to prevent agglomeration in the sample suspension. Afterwards the samples were sieved into fractions of > 80$\mu$m, 20 - 80$\mu$m and < 20 $\mu$m. The fractions < 80 $\mu$m were used for SPLITT-fractionation. The fraction 20 - 80 $\mu$m was fractionated in one step. Because of the small amount of grains with hydrodynamical properties similar to diatom skeletons, it was relatively easy to get a clean diatom fraction. The fraction < 20 $\mu$m was treated several times to enrich a sample fraction with diatoms. Because of the high amount of mineral grains (especially grains with equal sinking velocities) it wasn't possible to get an entirely clean sample of diatom frustules. To avoid problems related therewith an additional step was introduced by dividing the corresponding samples with 5 and 10 $\mu$m sieves. The resulting fractions do no more contain minerals with hydrodynamical properties of diatoms and therefore a separation of diatom frustules is again possible. In addition a method was developed for analysing grain distribution and amount of diatom frustules using image processing methods. For each SPLITT-fractionation subsamples from the two outlets were collected and photographed under the microscope. The pictures were then evaluated using image processing software. The photographs represented the particle size distribution of the fractions, the amount of frustules and gave a cut-off diameter at different in- and outlet flowrates.