JOURNAL OFSEDIMENTARYRESEARCH,VOL. 73, NO.2,MARCH, 2003,P. 328–332Copyrightq2003, SEPM (Society for Sedimentary Geology) 1527-1404/03/073-328/$03.00RESEARCH METHODS PAPERSHOW TO OVERCOME IMAGING PROBLEMS ASSOCIATED WITH CARBONATE MINERALS ONSEM-BASED CATHODOLUMINESCENCE SYSTEMSROBERT M. REED1ANDKITTY L. MILLIKEN21Bureau of Economic Geology, John A. and Katherine G. Jackson School of Geosciences, The University of Texas at Austin, Box X, University Station,Austin, Texas 78713-8924e-mail: [email protected] of Geological Sciences, John A. and Katherine G. Jackson School of Geosciences, The University of Texas at Austin, Austin, Texas 78712ABSTRACT:On SEM-based cathodoluminescence systems (scanned CL), meth-ods are needed to overcome image-quality problems caused by persistent lu-minescence of carbonate minerals. An effective solution to the persistence prob-lem is to acquire images using only the shorter wavelengths, most easily doneby using a broadband, short-wavelength (UV-blue range) filter. The filter usedprovides 80% to 90% transmissivity in the range of 385 to 495 nm and sometransmissivity as low as 350 nm. The filter allows transmission of the relativelynonpersistent UV-violet luminescence present in most carbonates in the rangeof 350 to 425 nm, but blocks the common orange-red wavelength luminescencefound in carbonates. The lack of imaging problems in subsequent images showsthat persistent luminescence in carbonates is primarily in the orange-red wave-lengths. Cathodoluminescence images produced using a short-wavelength filterare comparable in detail to those obtained from conventional light-microscope-based cathodoluminescence systems. In almost all examples, features visible inthe orange-red wavelengths show corresponding variations in luminescence inthe shorter wavelengths.INTRODUCTIONSEM-based cathodoluminescence systems (scanned CL) afford higher magnifi-cations, better detection of weak luminescence, detection of a broader spectrum ofluminescence, and more stable operating conditions than do conventional, light-microscope-mounted CL systems. Scanned CL imaging of carbonate minerals, how-ever, presents technical challenges that are more significant than those presented bysimilar imaging of most silicate minerals. These challenges relate to two aspects ofcarbonate CL: relative luminescence intensity and persistence of luminescence.Luminescent carbonate minerals commonly emit brighter luminescence than mostpotentially associated silicate minerals (such as quartz). In a rock having a mixedsilicate–carbonate lithology, this difference can create serious imaging problems. Ifcontrast and brightness of the image are adjusted so that details are visible in silicateminerals, associated carbonate minerals may be too bright to image without loss ofinternal detail. If contrast and brightness are adjusted for observing the detail incarbonate minerals, silicate minerals commonly show little or no apparent lumines-cence. Differences in relative luminescence can also be a problem within zonedcarbonates. This problem is not unique to scanned CL imaging: it is encountered indirectly viewed, light-microscope-mounted CL systems as well.Unique to scanned CL imaging, and the primary focus of this study, is the problemcreated by persistence of luminescence (phosphorescence). Marshall (1988, p. 24)noted that ‘‘some minerals have long phosphorescence and the CL of importantcases such as calcite cannot be effectively studied on the SEM because of this.’’ Insome carbonate minerals, photoemission continues for considerable periods aftercessation of electron-beam excitation. In a scanned CL system, a typical beam rasterspeed (200ms per pixel dwell time) may be slow relative to the decay time of theCL emission, resulting in simultaneous detection of light from different areas of thespecimen. This simultaneous detection leads to bright streaking or smearing in theimage parallel to the scan direction. The effect can severely degrade image quality(Fig. 1A), and it makes it difficult to use scanned CL on most samples containingluminescent carbonate minerals.INSTRUMENTATIONScanned CL images were produced using an Oxford Instruments MonoCL2 de-tector and PA-3 amplifier attached to a Phillips XL30 SEM. This system uses aretractable parabolic mirror and a photomultiplier tube to collect and amplify lu-minescence (Kearsley and Wright 1988). For the images shown here, the CL detectorwas being run in panchromatic mode either with or without a filter. The SEM wasoperated at either 15 or 20 kV and at large sample currents. The spot size used waslarger than that typically used for secondary electron imaging on an SEM and wasmore typical of the settings recommended for EDS mapping (setting 6.3 on theXL30). Emission currents ranged between 50 and 125mA. Photomultiplier voltageson the PA3 amplifier ranged from2625 V (for dominantly silicate lithologies) to2725 V (for dominantly carbonate lithologies). Samples were positioned 1 to 1.5mm beneath the bottom of the CL mirror assembly. Samples discussed here arepolished thin sections coated with a layer of carbon 25 to 30 nm in thickness,although polished rock slabs can be examined as well. Experiments using interpixeldelay were conducted on a JEOL T330A using an Oxford Instruments PanaCLdetector and 4pi Analysis beam control and image acquisition software.APPROACHESSeveral techniques have been tested for imaging carbonate minerals using scannedCL, with varying degrees of success. Each has its own merits and problems.Dwell Time and Interpixel DelayAn approach to overcoming imaging problems associated with persistence of car-bonate luminescence in SEM systems was previously proposed by Lee (2000). Heshowed that increased dwell time (the amount of time that the beam remains on aparticular spot) during imaging can produce nondegraded images of carbonates usingscanned CL. Very long dwell times (3,200ms per pixel) produced detailed and clearimages of zoned calcite (Lee 2000).The main problem associated with an approach based solely on increased dwelltime is the increased time necessary to produce a single image. For a 1,0243760pixel image, a dwell time of 3,200ms per pixel results in an image-collection timeof approximately 46 minutes (Fig. 1B). For studies requiring large numbers of im-ages—microfracture analysis studies, for example (Gomez et al. 2001)—such animage-collection time is prohibitive. In addition, long dwell times result in