INTRODUCTION

Despite advances in recent years, microfossils, or small details on larger fossils, can be difficult to image. Advances include: equipment that allows routine scanning electron microscopy without the need to coat specimens with conductive materials; techniques such as scanning light microscopy (Scott et al. 2000) and extended focus image montage methods (e.g., Holbourn and Henderson 2002; Knappertsbusch 2002) that are able to overcome the depth of field limitations of optical photomicrography; high resolution colour laser scanning (Lyons et al. 2000) and computed tomography (Rowe 1996; Brochu 2000). These techniques and methods are increasingly applied to palaeontological problems, yet there are still instances when availability of equipment, cost considerations, or the nature of the material under investigation mean that direct imaging using these techniques is impractical or impossible. I describe here a series of simple and inexpensive methods for specimen preparation and replication that allow for high resolution imaging of microscopic details in such instances. I also describe a simple and rapid process for producing anaglyph stereo images, and provide a set of Adobe Photoshop® actions for partial automation of this process. Except in some details, none of these techniques is new. The purpose of this paper is to highlight their usefulness in palaeontological research and communication, especially in dealing with microscopic subject matter, to provide specific details of methods and widely available materials that have been found to work well, and to illustrate with examples from conodonts and other jawless vertebrates how they can reveal important detail that is otherwise unobtainable. It is worth emphasizing that before applying any of these methods to borrowed material, full permission should be obtained.

Silicon Rubber Moulding

For most palaeontological applications, the choice of moulding material is between latex-and silicon-based compounds. Latex rubber solution is still in common use by palaeontologists and is suitable for many purposes, but it has the major disadvantage of shrinkage (Goodwin and Chaney 1994). Silicon-based materials, especially room temperature vulcanising (RTV) silicon rubbers, on the other hand, have lower levels of shrinkage and also offer other advantages over latex, including strength, flexibility, (Goodwin and Chaney 1994), and, once cured, good shelf-life stability (although some are susceptible to degradation by ozone and UK light, so storage in a dark, anoxic environment will maximise shelf life). Silicon rubbers also have the capability of capturing details at the submicrometre level (Rose 1983). Latex moulds have poor resistance to epoxy and polyester resins, and a parting agent must be used to protect the mould (Goodwin and Chaney 1994). Consequently, methods using latex moulds with resin cannot replicate microscopic details.

The use of room temperature vulcanizing silicon rubbers (and epoxy casting - see below) in palaeontology was pioneered by Waters and Savage (1971) and is now widespread. Waters and Savage (1971) and more recent publications (e.g., Reser 1981; Chaney 1989; Goodwin and Chaney 1994) cover the general methodology of mould making and contain details of precautions that will ensure that use of RTV silicon rubbers causes no damage to either the fossil or the humans involved. The focus of this paper, however, is moulding and casting microscopic detail, and although Waters and Savage (1971), for example, replicated micromammal remains down to a few millimetres in size, the use of silicon rubber to mould smaller fossils, including microfossils, has been limited. Chaney (1989) attempted to mould forams, but had only partial success, whereas Siveter (1984), obtained good results using RTV rubber to cast the fine details of ostracods preserved as natural moulds. Some of the methods presented here have developed from those of Siveter (1984).

The basic steps of the process as I have applied it are as follows. Prior to moulding, the surface of the specimen is coated with a thin layer of separator (or consolidant) or a release agent. This layer serves several purposes: it prevents loose parts of the specimen from being pulled off when the cured mould is removed; it prevents silicon fluid from penetrating the surface of the specimen and leaving a dark stain (particularly important if the specimen preserves traces of soft tissue remains); and it improves the chances of removing the cured mould cleanly from the specimen without leaving torn-off fragments of rubber in undercuts and the deeper recesses of the specimen. A solution of pvb [a terpolymer of poly(vinylbutyral), poly(vinyl alcohol), and poly(vinyl acetate)] in methanol has been found to be an effective consolidant and separator, which has the advantages of being safe, stable, and reversible (Elder et al. 1997). The solution is easily prepared to the right consistency to allow good impregnation of the surface without overglazing and obscuring morphological details. If applied too thickly, excess consolidant can be removed or redistributed by brushing methanol over the surface. Determining how much separator is required to protect the specimen without obscuring detail is a matter of judgement and experience, but if patches of the dried separator on the surface of the specimen appear smooth and shiny, it is probably too thick. Chaney (1989) suggested that use of a separator prevents the mould from picking up microscopic detail, but I have found that if the coating is thin enough, the mould can pick up details down to a few tens of micrometres or less (see examples below, especially Example 3). Nevertheless, if the specimen will allow it, maximum detail will be obtained if no separator is used. It is worth noting here that in selecting a separator, reversibility is a particular important property. Materials with questionable or limited reversibility should be avoided as they may be difficult to remove from the specimen.

A release agent may be used instead of, or in addition to, a consolidant or separator. In itself this release agent will do little to stabilise the surface of the specimen, but it will reduce or prevent penetration of silicon fluid and facilitate clean removal of the cured mould. A 50% solution of domestic dish-washing detergent (washing-up liquid) in water makes an effective release agent that does not obscure surface details. A thin coat can be painted onto the specimen with a soft brush and left to dry before moulding. 

In most cases, the area of the specimen to be moulded will need to be surrounded by a wall to prevent silicon rubber from flowing away over the surface of the specimen. A variety of materials are suitable for this purpose. Non-curing modelling clay, such as plasticene, is widely available and easy to use, although dark colours may stain the surface of pale specimens, and silcone rubber may adhere to some clays (see Goodwin and Chaney 1994). For walls surrounding small areas, rapid curing two part vinyl polysiloxane rubbers designed for dental applications, such as Colténe® President, can be very effective in forming mould walls although they tend to be expensive.

As for the moulding material itself, many different RTV rubbers are available from a number of manufacturers and suppliers, and the choice of rubber will depend to some extent on the nature of the specimen and the mould required. For moulding microscopic details of fossils, however, the uncured rubber will need to be of low viscosity, and the cured rubber should have a high elongation at break and high tear strength. These attributes will maximize the chances of removing the mould cleanly from the fossil without leaving torn-off fragments stuck in the deeper recesses and undercuts (which can be very difficult to remove without causing damage). When dealing with fragile material, however, high tear strength rubbers can increase the possibility of damage to the specimen because the specimen may break before the rubber tears, so careful consideration of the properties of the cured rubber is important. Having tried a number of different rubbers, I have had good results with a product sold as Ambersil RTV913 in the UK (manufactured by Ambersil Silicones) or as QM113 in North America (Quantum Silicones). Distributors in other parts of the world are listed on the UK manufacturers website  (The Amber Chemical Company). This rubber is a two component room temperature condensation curing silicone compound. The cured rubber is very flexible and has good shelf-life stability. Full details of the physical properties are available from Ambersil, but the important attributes are as follows (manufacturers data): viscosity of base compound approximately 15,000 MPa.s (MilliPascal seconds; the viscosity of the compound with catalyst added is slightly less); linear shrinkage < 0.5%, tensile strength of cured rubber, 3.1 Mpa; elongation at break, 650%; tear strength 22 kN/m. The potlife of the catalysed rubber (i.e., the working time for mould pouring) is approximately 45 minutes.

The rubber is mixed according to the manufacturers instructions, taking appropriate precautions for the safe handling of the materials (see health and safety information below). Optical imaging of the mould may be desirable, in which case black pigment (e.g. Aniline Black or Lamp Black) can be added to the rubber at the same time as the catalyst. After mixing, placing the rubber under vacuum for a few minutes can reduce the number of bubbles, but when applying rubber as outlined here I have not found this step to be necessary. All the examples presented below have used Ambersil RTV913.

The quality of the final mould depends to a large extent on the way in which the catalysed liquid rubber is applied to the specimen. The best results are achieved by trickling a small amount of rubber down the side of the wall or onto the specimen, and allowing it to flow slowly over the surface, so that the rubber creeps over the specimen, and is pulled into details and recesses by surface tension and capillary action. If the advancing edge of the rubber is allowed to bulge outwards and "roll" over the surface of the specimen there is a high probability that air bubbles will be trapped in fine details and recessed areas. Using a mounted needle or something similar to gently pull back the advancing edge of the rubber, taking care not to contact the surface of the specimen, can slow the rate of advance and prevent the edge from rolling over. The rate of flow is also much easier to control if liquid rubber is added in small increments. Cured RTV913 is very flexible and lacks rigidity. For small specimens or moulds this flexibility does not present a problem, but larger moulds may require a rigid supporting jacket (or mother mould), or addition of a layer of stiffer RTV around the mould.

The next stage of preparation will depend on the nature of the material, and whether it is the mould of the specimen that is of interest or a replica of the original. In cases where rubber is being used to cast a natural mould or a mould prepared by acid preparation of a specimen (see Example 1), it will be the rubber cast itself that is of interest. Once cured this rubber cast can be mounted and coated for scanning electron microscopy using standard methods (although because of the poor conductivity of the rubber, longer coating times may be required than for most specimens). If a replica of the original specimen is required (see Example 2 and Example 3), an epoxy resin cast should be prepared.

Epoxy Resin Casting

In making high resolution casts, the choice of casting medium depends on a number of factors, but for detailed replication of microscopic details on small specimens, epoxy resins are ideal. Not only are epoxy casts rigid and durable, their fidelity to the mould is very high, better than any other material (Chaney 1989), and they are stable over long periods. They are also easily mounted and coated for scanning electron microscopy. One factor that should also be borne in mind is that not all resins and moulding materials are compatible, and reactions between the mould and the casting medium can significantly reduce the quality of the cast. For reviews of the problems relating to compatibility between epoxies and silicon-based moulding compounds in the context of reproducing tooth wear facets and details of microwear see Gordon (1984) and Teaford and Oyen (1989), or Bromage (1985) for more general comments on replicas in scanning electron microscopy. After the issue of compatibility, probably the most important factor in determining the quality of reproduction of microscopic detail is the viscosity of the resin. Low viscosity resins flow more easily into small recesses and are less likely to trap air bubbles.

I have had good results with Araldite 2020, a widely available two-component low-viscosity, water-white epoxy adhesive which is also suitable for casting (see Example 2 and Example 3). Not only is the viscosity low enough for epoxy to flow easily into details of the mould, it is compatible with RTV 913, yielding good quality casts with high levels of detail (including features only a few micrometres or tens of micrometres in size [see Example 3]). According to the manufacturers technical support department, casts in Araldite, if stored away from UV light, should last at least 50 years, although projection beyond that time is difficult (Noel Moss, Vantico Ltd., personal commun., 2003).

The process of pouring an epoxy cast is similar to that for preparing a rubber mould outlined above. The epoxy resin and catalyst are mixed according to the manufacturers instructions, taking appropriate precautions for the safe handling of the materials (see health and safety information below). The usable life of the catalysed epoxy is about 45 minutes at 23°C (but note that the exothermic reaction between catalyst and resin can cause a rapid rise in temperature if large amounts of resin are mixed, and this temperature increase will accelerate curing). Small batches of epoxy weighing only a few grams can be prepared if the proportions of catalyst and resin are measured by weight using scales of appropriate precision. The catalysed epoxy is then trickled slowly into the mould, or added drop by drop, and allowed to flow slowly over the surface. The use of hand-driven centrifuges is advocated in some discussions of epoxy casting (e.g., Waters and Savage 1971), but I have not found this step necessary when pouring one-part moulds using Araldite 2020. The epoxy is water clear, so if optical examination or imaging will be required, pigment should be added to the resin at the same time the catalyst is mixed in. Addition and thorough mixing-in of Lamp Black or Aniline Black pigment produces opaque black casts. Other pigments suitable for epoxy may also be used. It is worth noting that if the epoxy is coloured using a solution of eosin stain in ethanol, the method of Jernvall and Selänne (1999) can be used to acquire high-resolution digital representations of fossil shape through laser confocal microscopy (although the effects of eosin solution on the cure properties Araldite 2020 have not been tested). For Scanning Electron Microscopy, it makes no difference whether the cast is pigmented or clear.

Health and Safety and RTV Rubber and Epoxy

Care must be taken, when using RTV rubbers and epoxy resins, to follow manufacturer instructions regarding safe handling and storage. Some of the components of RTV rubbers, epoxy resins and their catalysts are toxic and/or irritants. Material Safety Data Sheets (MSDS) for Ambersil RTV913 and other Ambersil products are available via the Ambersil and Quantum Silicones websites. Instructions on how to obtain MSDS for Araldite 2000 series epoxy resins are available from the Huntsman distributors website.

Stereo Anaglyph Images

Techniques for obtaining and publishing stereo images of fossils, and especially small fossils, are widely known in the palaeontological community, and their greater use has been advocated by a number of authors over the years (Evitt 1949; Sylvester Bradley 1971). Conodonts were among the first fossils to which stereophotographic methods were applied (Branson and Mehl 1933; Evitt 1949), yet with the exception of ostracods (Siveter 1984) stereo images are seldom employed to illustrate microfossils, or the microscopic details of larger specimens. The standard approach is to obtain two images of the fossil which differ in their effective viewing angle by 8-10° (although 5° may give better results when the depth of a specimen, in the orientation being photographed, exceeds the average width; Sylvester Bradley 1971). Optical or digital stereo photomicrographs can be obtained using simple tilting or sliding stages (Branson and Mehl 1933; Feldman 1989); scanning electron micrograph pairs are easily obtained by tilting the specimen stage. Once obtained, the images are reproduced side by side, with the axis of image rotation aligned vertically between them, and viewed using stereo viewers (see Feldman 1989 for details). This method can be very effective in communicating three-dimensional (3D) geometries of fossils, but the size of the images is limited, and unless large format stereoscopic viewers are used, the maximum width at which images can be reproduced is only 5 or 6 centimetres (see Sylvester Bradley 1971 for discussion).

What is perhaps less well known among palaeontologists is that red-green or red-blue anaglyph images offer an alternative that with the widespread availability of powerful image editing software such as Adobe Photoshop® and Corel Photopaint® are quick and simple to produce. Images can be reproduced at any size (for printing, viewing on screen, or even for projection in presentations) and viewed in stereo with readily available and inexpensive cardboard and plastic viewers (with one red and one green lens, or one red and one blue, depending on the chosen anaglyph format; viewers are available from scientific equipment suppliers, or from the numerous potential vendors that are revealed by a www search for "3D glasses"). Stereo images produced in this way can be extremely useful as research tools, for communication via the Internet and email, and also for publication. Reproduction of colour plates and figures is now more widespread and cheaper than in the past, and is even made available at no additional cost by some journals. 

Knappertsbusch (2002) recently outlined a method combining digital stereo imaging through an optical microscope with extended focus montaging and QuickTime VR? authoring to produce animated anaglyph stereo images of microfossils. He highlighted some of the advantages of anaglyph stereo images and considered aspects of stereo image acquisition in some detail, but the paper was focussed on a particular method, using relatively expensive hardware. My goal here is to provide a simple guide to the production of anaglyph images using widely available equipment and software.

The exact method of producing anaglyph stereo images varies according to the software package used, but Appendix 2 to this paper includes step-by-step guides to producing images using Adobe Photoshop®. A set of actions that will automate much of the process in Photoshop® is also available for download (Anaglyph_RG.atn for Photoshop® v. 4 or newer). Producing anaglyph images using software packages that do not support image layers is less intuitive, but a www search for "analglyph" brings up instructions for many common digital imaging or photo editing packages that should be readily applicable to palaeontological images.