Marine sediment cores are archived in segments of 1 m in length for convenient handling and cut longitudinally in two halves for further processing. (This is not required for box and kasten cores, where samples are taken from the outer side.) After photography and a visual lithological description of the sediment sequence (structure, texture, colour), the “work” half is sampled for various analyses, with the preparation of X-ray slabs being the first step of the sampling workflow (Grobe, 1986; Fig. 1). It was common practice first to prepare sediment slices for radiographs and use the exposure as a guide for further sampling, in particular across strongly bioturbated sections. The remaining “archive” half is sealed in airtight D-tubes for future investigation. As a common practice in geological repositories worldwide, the core segments are archived in repositories at +4 ∘C.

To ensure optimum quality of the X-radiograph, sediment slices have to be prepared with the utmost care so as not to destroy the original sediment structures and to avoid artefacts. The surface of the longitudinal core half is smoothened with a wet glass plate, eliminating the grooves and furrows on the sediment surface usually caused by larger sediment particles. A hard Plexiglas^® lid measuring 25 cm × 10 cm × 1 cm is pushed gently into the level surface of the working half to provide support and stability for the sediment slab (Fig. 2). Core label, depth interval and an arrow pointing downwards in the direction of penetration of the coring device are then marked on the lid (in reverse chronological order, in contrast to the convention at other research institutes and programmes, such as the International Ocean Discovery Program). A kasten core typically prepared for the sampling of X-ray slabs is shown in Fig. 3. The slab within the lid is separated from the remaining core by pulling a nylon line up-core behind the plastic lid, thereby using the two sides of the lid orientated parallel to the core axis as guides. The slab is carefully lifted with a cheese knife and removed from the core. The slit may be moistened if the slab tends to adhere to the main sediment core. The slices in the lids are vacuum sealed in polyethylene lay-flat tubing after evacuation of the enclosed air. The equipment used for the preparation is shown in Fig. 4. It is important to note that the X-radiograph technique is non-destructive and that – after X-raying – the sample material could be used for further analyses.

X-ray film is coated on both sides and thus has higher sensitivity and contrast compared to film used in light photography. For exposure the film type Structurix D 4 manufactured by Agfa-Gevaert was chosen. The film is not sensitive to red light and thus can easily be handled in a darkroom under low-light conditions. The film is cut into 25 × 10 cm strips and stored in black film covers. Each 25 cm long slab is exposed to the X-ray beam with the slab surface not covered by the lid facing the X-ray source (Fig. 5).

The characteristics of the X-radiation determine the quality of the sediment images, with the wavelength being the most important factor. High energy will produce “harder” radiation with shorter wavelength while lower energy will result in radiation with longer wavelength. Due to the soft composition of unconsolidated sediments, a spectrum of longer wavelengths and thus “weaker” radiation is preferred to produce images with a moderate contrast (Werner, 1975).

Exposure times depend on sediment type and are mostly controlled by grain size and compaction, the thickness of the slab and the strength of the radiation. A 1 cm thick slice has to be exposed for a time span from 3 (soft clay) to 20 min (high sand content) at a voltage of 30–35 kV and an electrical current of 3 mA. Identification of the core ID and the depth interval on the negative is assured by putting corresponding lead letters and numbers on the film during exposure. The film is developed in a darkroom for 3 min by using the developer G124 and the fixer G335 (Agfa), washed for 20 min in distilled water and dried. A detailed description of the procedure can be found in Werner (1975).

Over 50 years of marine geological research at GIK, a suite of 1355 sediment cores with a total length of 3547 m were investigated (Fig. 6). More than 18 500 sediment slices were prepared for X-ray imaging and exposure. Between 2010 and 2014 the images were transferred to the PANGAEA department at the Alfred Wegener Institute for Polar and Marine Research, Bremerhaven (AWI). They were digitized at a resolution of 600 dpi by a transmitting light scanner (Microtek ScanMaker 9800XL). The X-radiographs, photos and descriptions were supplemented with the metadata of the corresponding cores including position, water depth, sampling device, expedition, date/time, ship, etc. The resulting collection of 1355 datasets was combined into a single parent dataset, which is available at 10.1594/PANGAEA.854841.

Besides the documentation by light photography, X-radiography became a standard imaging technique in marine geology to complement and support the visual description of sediment profiles, which comprises the detailed logging of the lithological composition of the sediment, its texture and sedimentary structures (Bouma, 1969). In particular, radiographs reveal details of structures, such as bioturbation and graded bedding, diagenetic modifications and large internal components (e.g. fossils or dropstones) which are not discernible in normal light photography.

The dominant control on beam attenuation is bulk sediment density (Holyer et al., 1996), which in turn is affected by grain size, mineralogical composition, abundance of biogenic components and physical parameters such as water content, porosity and compaction. Thus radiographs can be used to determine a whole range of various sediment properties. During examination of the images, some specific points need to be considered. These are described in brief in the following examples of typical sedimentary textures as shown in Fig. 7.

Structures that may be unrecognizable to the naked eye include boundaries of strata, non-conformities, fine lamination, graded bedding and most importantly – bioturbation (Werner, 1968; Winn, 1974; Wetzel, 1979). Lebensspuren are the most common structures in marine sediments which allow the identification of the species and the reconstruction of paleoecological and paleoenvironmental conditions (Löwemark, 2001). The resulting taxonomy of palichnology is the basis for its identification and classification (Bromley, 1999; Seilacher, 2007). Sedimentological sequences formed by distinct processes (e.g. deposition by turbidity or contour currents) and their evolution over time are also part of this structural group. Features such as base and top boundaries, type, thickness, frequency, rhythms and cycles indicate facies differentiation and changes.

Physical properties can be identified by the brightness of the negatives, as well as by internal structures such as layering (e.g. ash layers), lamination, bedding planes, cross-bedding, current ripples or sorting. By using a magnifying glass while investigating the X-ray image, individual grains with a size of > 1 mm can be classified in terms of grain shape and composition. Support of a high-resolution sedimentology is given in the millimetre to centimetre scale, including large components such as mud clasts or gravel grains, e.g. as ice-rafted debris (Grobe, 1987; Principato, 2004). Any gravel fraction reveals itself by the distinct appearance of each individual grain.

Extracted high-resolution greyscale curves were correlated with physical parameters (colour, gamma-ray density, magnetic susceptibility, grain size; St-Onge et al., 2007; Kowalczk, 2005). There is a clear relationship between grain size distribution and brightness/blackening of the film. Sandy to clay-like sediments are composed of just a few minerals (quartz, feldspar, clay minerals, carbonates), all of which have a similar specific grain density of 2.6 to 2.8 g cm-3. Diatomaceous oozes may result in nearly black X-radiograph negatives because of the low density of opal. Thus, differences in brightness result mostly from changes in grain size rather than from a heterogeneous distribution of minerals. Water content, density and porosity are the major factors governing greyscale values; porosity increases from coarse-grained to fine-grained sediments. The sediment density and thus the brightness of the image negative increases with core depth because the compaction results in reduced pore space and water content.

Minerals may result from diagenesis including authigenic pyrite, zeolite, or the rarely formed porcelanite (Gerland et al., 1997). Heavy minerals such as pyrite and other iron sulfide as well as iron oxide minerals can easily be identified by their high brightness/X-ray attenuation and their specific grain shapes and internal structures. However, dark grey features visible in negatives can be areas with an extremely high water content, plant fossils, wood or even small voids. In cases where the samples were stored for a longer period, new minerals may have formed through chemical processes in the sediment (e.g. as per Fig. 6g).

Artefacts reflecting the post-depositional disturbance of the original sedimentary structure must be identified within the core. These effects can have various causes which should always be considered while investigating and interpreting the images (Skinner and McCave, 2003). During the coring process and recovery, especially with a gravity, kasten, piston or vibrocorer, the mostly soft and often “soupy” sediments from near the seafloor surface (i.e. at the core top) may flow, resulting in a loss of the original structure. In some instances it is not even possible to prepare a sediment slab suitable for X-radiography from the upper decimetres of the core. Coring disturbance caused by the piston or gravity coring process may result in “pseudo-tectonic” features (e.g. faults, fractures, sediment mixing, “flow-in”) which are predominantly observed at the bases of longer piston and gravity cores. In particular, gravity coring can cause micro-faulting within the sediment and result in an artificial shortening of the sediment column (Fig. 6c). Especially in clay-like sequences, even pseudo-hiatuses can occur when parts of a sediment section succession allow the core barrel to pass but are squeezed out and thus not recovered. In addition, the outer edges of a core segment may show downward bending of layers in close proximity to the core liner, which results from the friction between the liner and the sediment when the core barrel penetrates the seabed.

Further effects visible in radiography might include the following: if the sediment slice is not properly sealed in lay-flat tubing it may dry out and produce drying cracks, which can, however, be clearly identified. If a sediment slice has a variable thickness, the brightness of the X-radiograph will vary throughout the sample. If the sediment contains larger particles, the marginal areas around the particle may be disturbed during the preparation. Regular stripes and patterns observed also on an X-radiograph are usually the result of insufficient smoothening of the slab surface and reflect tracks from the cutting wire. The projection also has to be taken into account during analysis: structural elements of a three-dimensional slice are projected in two dimensions onto the film. A 1 cm thick slab sample for X-raying does not work well for gravel- and pebble-rich sediments (such as diamictons), and thus X-raying of half cores works better for those sediments. However, X-raying of half cores has its own problems, especially overlaps and projection issues.

More than 18 500 exposures of X-radiographs were digitized using two A3-format scanners, model Microtek ScanMaker 9800XL, at a resolution of 600 dpi and stored in JPEG format with moderate compression to generate file sizes for convenient Internet download times. Not all images were post-processed and thus some may still be underexposed. Brightness, lucidity and contrast can be corrected as required for investigation with any image processing software. The full information is in each image due to the high-resolution scan of the fine-grained film. Images were uploaded to PANGAEA and stored in a database. One dataset includes all images of one core. Metadata and additional documentary files including core descriptions and photos were added, if available. The metadata cover core ID, latitude and longitude, water depth, recovery, coring device and date/time when the core was taken. The label of the expedition linking to the cruise report (if published) is also provided. Each dataset starts with a “Citation” tagged line, consisting of the name of the principal investigator(s), a standard title “Documentation of sediment core GIKxxxxx-x”, year of electronic storage and thus public availability, and the source institute (in this case always set to GIK). The DOI as a persistent link to the dataset is a mandatory part of any modern citation. If the images were already used in publications, the corresponding references can be found under the “Related to” field. Selected examples of images from this collection are presented in Fig. 7.