JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 151 Microscopic X=ray Fluorescence Analysis* Invited Lecture K. Janssens L. Vincze J. Rubio and F. Adamst Department of Chemistry University of Antwerp Universiteitsplein I B-2610 Antwerp Belgium G. Bernasconi International Atomic Energy Agency Laboratories A- I I40 Seibersdorf Austria The status of microscopic X-ray fluorescence analysis with tube excitation and synchrotron radiation is reviewed in terms of the lateral resolution minimum detection limits and elemental sensitivity that can be achieved. As illustrations the utilization of two typical state-of-the-art instruments for the analysis of geological material is described; one of the instruments is based on tube excitation the other is installed at a synchrotron X-ray source.The analytical implications of the use of X-ray microprobes installed at a third generation storage ring and in particular at the European Synchrotron Radiation Facility (ESRF) are discussed. Keywords X-ray fluorescence; microscopic X-ray fluorescence; synchrotron radiation and imaging; micro- analysis; trace element mapping For more than two decades X-ray fluorescence analysis (XRF) has been a well-established and mature multi-element tech- nique capable of yielding accurate quantitative information on the elemental composition of a variety of materials in a non- destructive manner.' In the last 10 years two important variants of the bulk technique have come into existence.2 Both variants are based on the confinement of the interaction volume of the primary X-ray beam with the material being analysed.In total reflection XRF (TXRF),3 by irradiating an (optically flat) sample with a parallel X-ray beam below the angle of total reflection the in-depth penetration of the primary X-rays can be confined to a few tens of nanometers below the surface. As a result the intensity of the scatter background in XRF spectra collected in this way is significantly reduced and the (surface) sensitivity of the technique enhanced. In view of the relatively simple sample preparation required TXRF has proven itself to be an extremely useful technique for the analysis of solutions and natural waters with typical detection limits below 20 ng for 40 elements for counting times of 1000 s ~ and for the ultrasensitive surface analysis of semiconductors3 (coarse resolution 2-D mapping of impurity centres and high- resolution depth profiling).The second major variant of the XRF technique which in the last 2-3 years has received a lot of attention in the literature both with respect to the methodological develop- ments and to its applications in diverse fields is micro-XRF (p-XRF). This microanalytical variant of bulk XRF is based on the localized excitation and analysis of a microscopically small area on the surface of a larger sample providing infor- mation on the lateral distribution of major minor and trace elements in the material under study. The recent increase in the popularity of p-XRF can be attributed on the one hand to the availability of commercial instrumentation and on the other to the development of (relatively) simple devices for the focusing of X-rays.In addition the potential of all the forms of XRF mentioned above has in recent years been greatly enhanced by the increasing use of synchrotron rings as sources of highly intense X-radiation. In this paper an overview of the analytical capabilities and limitations of currently existing p-XRF spectrometers using conventional and synchrotron X-ray sources is presented. In the last part of this work the implications the use of third- * Presented at the XXVIII Colloqium Spectroscopicurn Internationale (CSI) York UK June 29-July 4 1993. t To whom correspondence should be addressed. generation synchrotron rings will have on the analytical pos- sibilities of p-XRF are briefly discussed.Experimental The laboratory-scale pXRF used for collecting some of the data presented in this paper consisted of a conventional Siemens diffraction tube with an Mo anode operated at 30 mA and 50 kV. The resulting X-ray cone was transformed into a microbeam by means of a conical glass capillary of 70 ym final inner diameter. Samples were mounted on an XYZB table with a positioning accuracy of 1 pn. The set-up operated in air the Si(Li) detector being positioned at 90" to the incoming beam and at a distance of approximately 2cm from the sample. Scanning of the sample through the beam and data collection were performed by means of a personal computer and software developed by one of the authors (G. B.) using a Canberra SlOO MCA plug-in board and associated driver software.For the micro-synchrotron radiation induced XRF (p-SRXRF) measurements the X-ray microprobe station at the bending magnet beam line X26A of the NSLS (National Synchrotron Light Source Brookhaven National Laboratories NY USA) was employed. In this instrument the white synchrotron light has a maximum flux density of about lo4 photon s-' pm-2 mA-' at 8 keV. After emerging from the storage ring ultra-high vacuum (UHV) the beam is defined by four Ta slits to a size of approximately 8 x 8 pm. Soft X-rays ( < 5 keV) are heavily absorbed in the Be end-window of the beam pipe and the air path between collimator and sample. The sample is positioned at 45" to the incoming beam mounted on an XYZB stage with 0.1 ym accuracy and can be viewed by a horizontally mounted microscope equipped with a colour TV camera.X-ray spectra are detected at 90" to the incoming beam using a well collimated Si( Li) detector. The electron probe X-ray microanalysis (EPXMA) measurements were per- formed on a Jeol JSM 6300 scanning electron microscope (SEM) system equipped with a Princeton Gamma Technology (PGT) X-ray spectrum and image collection system and using a 25 kV 1 nA electron beam. Both electron and photon induced X-ray spectra were evaluated using the AXIL p a ~ k a g e . ~ Discussion p-XRF Using Conventional X-ray Sources The simplest way of producing an X-ray microbeam is to collimate the broad cone of radiation originating from an152 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 X-ray tube by means of suitable apertures or crossed-slit systems.This concept was employed in the 1 9 6 0 ~ ~ ~ ' but did not receive a lot of attention at the time the major reason being the low count rates observed for small area samples. Also although the use of photons as an excitation source intrinsically features a number of advantages over charged particle excitation the rapid development of electron microscopy (and of electron induced X-ray micro-analysis) with its much higher lateral resolution overshadowed further developments. In the 1980s p-XRF was 're-discovered' with the first use of energy dispersive (ED) detectors for X-ray imaging applications. As the solid angle of acceptance of Si( Li) detectors can be much larger than that of crystal spectrometers fluorescent signals of much lower intensity could be used to advantage.In 1986 Nichols and Ryong equipped a rotating anode X-ray generator with a modified micro-diffractometer system to generate a microbeam with a nominal diameter in the range 10-1OOpm. The use of the diffractometer allowed manual translation of the sample through the beam and included a microscope for sample viewing. Their objective was to arrive at a relatively inexpensive system built from off-the- shelf components which would (i) provide analysis to a greater depth within the sample than the near surface analysis obtain- able using electron optical instrumen tation (ii) allow for localized analysis of large objects and (iii) be capable of performing analysis under ambient atmospheric or helium conditions rather than in vacuum.In 1987 Boehme" described the use of a p-XRF system for elemental mapping of large area (400 mm2) geological materials and pointed out that p-XRF provided information that enhances and corroborates data obtained by means of optical microscopy SEM electron probe microanalysis (EPMA) and conventional XRF analysis." Nichols et a/.12 investigated the effect of instrumental param- eters such as the position and size of the aperture and the beam profile and divergence on achievable lateral resolution and sensitivity. In 1988 Ryon et ~ 1 . ' ~ summarized the advantages of X-ray imaging as follows. (a) The penetrative character of X-rays and the complex but well understood interaction of X-rays with matter allows for the determination of layer thicknesses the analysis of sub- surface structures and homogeneity testing throughout a material.(b) High-energy photons can penetrate below the surface of opaque materials and cause the emission of characteristic radiation. (c) Unlike electron microscopies which (in most instruments) must be performed in high vacuum X-ray imaging can be done in air and on large samples requiring little or no sample preparation; also non-conducting materials can be analysed without problems. Compared with charged particle micro- scopies X-ray imaging causes low thermal loading allowing e.g. volatile components or very sensitive materials to be analysed. (d) X-ray equipment is simple in comparison with that required for scanning particle beam microscopies. Also in 1988 Wherry et presented a description of 'an automated X-ray microfluorescence materials analysis system' the first commercially available p-XRF system developed by Kevex (Valencia CA USA).This set-up consists of a low- power (50 W) X-ray tube providing a small (<250 x 250 pm) spot fitted with 10 30 and 100 pm diameter apertures placed sz 60 mm from the anode and 3 mm before the sample. Samples (in an evacuable chamber) are mounted on a motorized XYZ stage illuminated from above and below and can be viewed by means of a colour charge coupled device (CCD) video camera. Fluorescent radiation is detected using a 50 mm2 Si(Li) detector at 90" to the incident beam. As a result of the close coupled detection geometry for pure element samples and a 100pm aperture net count rates in the range 2000-7000 counts s- ' could be obtained; lateral resolutions were reported to be of the same order of magnitude as the collimator opening.Cross et all5 also have reported on the multivariate processing of multiple X-ray images for automatic phase discrimination. The use of this instrument in various disciplines has been reported including the analysis of buried layers in multi-metal multi-layer materials used in computer mainframe r n a n ~ f a c t u r e ~ ~ ~ ~ ~ the screening of toxic contami- nants and precious metals in heterogeneous wastes'' and the use of the instrumeni for miscellaneous problem solving in the fuel industry.19 Pella et a1.20.21 have reported on the analysis of coarse particles (diameters between 50 and 200 pm) using a laboratory-built instrument similar to the Kevex device and on the problems associated with the quantification of the results derived from this and other heterogeneous sample types.22 By using a micro-focus X-ray tube the photon flux loss as a result of collimation can be minimized; nevertheless in view of the large distance between anode and aperture significant losses occur and only a small fraction of the total photon flux leaving the X-ray tube arrives at the sample.An alternative approach to obtaining a more intense X-ray microbeam is to employ glass wave guides instead of collimators. As shown in Fig. l(a) through repeated total reflection off the inner walls of glass capillaries photons can be 'transported' from the tube anode to the immediate vicinity of the sample surface.In early work straight glass capillaries were used as fine collimators generating fine beams for mi~ro-diffraction~~ while in the 1970s the X-ray wave guide properties of capillaries were investigated by several g r o ~ p s . ~ ~ - ~ ~ Rindby28 described the use of straight capillaries together with conventional X-ray tubes for generat- ing X-ray beams of about 100pm in size. Carpenter and ~ o - w o r k e r s ~ ~ ~ ~ replaced the Be side-window of an HOMX 160A micro-focus X-ray tube with a capillary assembly bring- ing the end of the capillary as close as 2 mm from the 15 pm spot on the anode (see Fig. 2). In view of the small cross- section of the inner channel in the glass capillary tubing the vacuum inside the tube can be maintained without serious problem.In the Kwex instrument mentioned above the use of collimators smaller than 30 pm requires long scanning times because of the low beam flux. By the maximization of the acceptance angle of the capillary described above microbeams with cross-sections in the range 4-100 pm2 still provided acceptable flux. For these beam sizes minimum detection limits (MDLs) in the range 20-100ppm for Cu W and Mo were reported in a National Institute of Standards and Technology (NIST) Standard R.eference Material (SRM) 610 Glass Trace Elements 610 glass. Using SRMs 1832 and 1833 minimum detectable amounts of 0.01- 10 pg absolute were a~hieved.~' Applications of the microprobe described by the same workers include the identification of inclusions in a carbon structure the examination of cracking in alumina cylinders by ZrO particles and the irivestigation of the P Ca ratio in bone and teeth. In the last application the non-destructive character of p-XRF made reproducible analyses possible while in the 1 -.- 100 pm -v - 20 cm---+ 1 100 pm (b) 100 pm 10 pm Fig. 1 Principle of X-ray propagation in (a) straight and (b) tapered capillaries. Typical capillary dimensions (length entrance and exit diameters) are indicatedJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 153 Fig. 2 Close coupling of micro-focus anode and straight capillary; adapted from Carpenter and Taylor.31 T tube; H capillary housing; C capillary; E electron beam; and A anode. The system can be aligned by moving the anode up and down and by steering of the electron beam case of electron microprobe analysis changes of the P Ca ratio with time were observed as a result of beam damage.32,33 In order to increase the photon flux further it is also possible to use glass capillaries as X-ray concentrators.Conical or tapered capillaries were developed in the later 1980s by several g r o ~ p s . ~ ~ - ~ ~ As shown in Fig. l(b) the photon beam that enters the wide end of the capillary after a number of reflections off the walls of the conical tube are ‘squeezed‘ to the size of the inner diameter at the narrow end. Typical dimensions for conical capillaries used in p-XRF spectrometers are a length of 10-20cm and final inner diameters of 50-10pm. The advantage of this type of X-ray concentrators is the indepen- dence of the source.Owing to the numerous reflections inside the capillary the source size does not directly influence the cross-section of the generated microbeam; the latter is deter- mined only by the inner diameter of the capillary. The diver- gence of the beam is approximately equal to OJE) the critical angle for total reflection at the capillary walls for photons of energy E (for glass in the 10-20 keV energy range OC takes values from ~5 to 1 mrad). As a result of this low divergence which is far less than that of the original (tube generated) X-ray beam entering the capillary for very small beam diam- eters (< 10 pm) it is necessary that the sample is placed as close as possible (ie. in practice 0.5-1 mm) to the capillary tip. Instability and fluctuations in the position of the source do not affect the position and size of the focal spot.Because the capillary is a total-reflection device photons within a broad energy band are concentrated although the gain factor (k the ratio of the beam flux density entering and leaving the capillary) is energy dependent. As a result of the 1/E2 depen- dence of the gain factor Rindby et aL3’ have shown that the combined effect of the anode self-absorption and the high- energy filtering of the conical capillary can influence the original tube spectrum in such a way that a nearly monochro- matic microbeam is obtained. At present several manufacturers are developing com- mercial p-XRF instruments that are based on the use of conical capillaries. Rindby and co-workers have reported on a number of applications of this type of p-XRF instrument.Larsson et used a 19 pm capillary and a Cr X-ray tube to investigate the distribution of elements such as Ti and S in birch leaves. Engstrom et ~ 1 . ~ ~ reported on the analysis of soft biological tissues such as 8 pm thick pig heart muscle fibres and quoted detection limits in the 60-1 ppm range for the elements in the range Al-Ti. Using both Cr and Mo tubes they also analysed single hair strands along and across their longitudinal direction and could determine the elements in the Z range S-Br at trace levels varying from 1 to 1000ppm. Shakir et studied the leaching of elements from the leaves of the Kardadeh plant (Hibiscus Shadriif) by interaction with hot water (tea brewing) showing a depletion of Mn Ca and K levels.Stocklassa et aL4’ reported on the application of non- destructive trace and microanalysis by means of p-XRF in forensic science involving the analysis of small glass particles paint fragments ball point ink and single hair strands. An interesting application has been described by Rindby et a!?’ concerning the authentication of historic documents such as a Swedish Letter of Possession dated April 1499 which was suspected of having been forged in the 16th century. By scanning selected areas of the parchment old text wiped out or covered by more recent writing could be revealed by means of its trace element content. Voglis et aL4’ described investi- gations concerning the incorporation of heavy metals into bone and recorded the radial distribution of elements such as Al S Cl K Cr and Fe around Haversian Canals.Prior to the use of glass capillaries for pXRF purposes Gurker et ~ 1 . ~ ~ 9 ~ ~ suggested a method for partially eliminating the count rate limitations associated with collimator-based p- XRF s p e ~ t r o r n e t e r s . ~ ~ . ~ ~ Instead of using the very inefficient two-dimensional collimation (where less than 0.1 % of the total output power of the X-ray tube is employed) only one pair of slits was used thereby irradiating a line of points on the sample surface instead of a single point. Instead of collecting an image through point-by-point irradiation the sample is translated and rotated through the narrow band of radiation. The observed fluorescence intensity I(r,O) as a function of the translation co-ordinate Y and the rotation angle 8 is called a ‘sinogram’.By means of tomographic back-projection tech- niques the resulting series of sinograms can be converted into elemental maps. The price to be payed for the more efficient use of the total available photon flux is that artefacts and noise could be introduced in the reconstructed images as a result of the data-collection procedure and/or the back- projection algorithms employed. In general one can state that p-XRF provides new capabili- ties for the analytical chemist in that it fills the gap between bulk X-ray fluorescence and (high-resolution) electron probe X-ray microanalysis. As such a wide range of samples that have been excluded from microscopic examination because of their incompatibility with the vacuum and conductivity requirements of electron microscopy can be analysed.On the other hand in situations involving the determination of trace levels of high-2 elements (e.g. Fe-Mo) with high (1-10 pm) lateral resolution the applicability of the method is still seriously constrained by the limitations in available micro- beam flux and size. As an illustration of the advantages and limitations of p-XRF in comparison with elemental mapping by means of EPXMA Figs. 3 4 and 5 show elemental maps the electron backscattered (EB) image and X-ray spectra obtained using a p-XRF spectrometer equipped with an Mo tube and a 70 pm conical capillary and by means of scanning electron microscopy with an energy-dispersive X-ray assembly (SEM-EDX) from a heterogeneous geological sample.The sample a 60 pm thick section of igneous rock shows biotite crystals embedded in a feldspar-quartz matrix; it was scanned with a step size of 100 pm in both directions in the p-XRF spectrometer. A collection time of 40s per location was employed causing the total acquisition time to become about 14 h for the 30x 30 pixel images. During such a relatively short time only statistically significant data on the major constituents of the geological samples can be obtained. As such only meaningful maps of elements such as K Ca and Fe could be collected usingp-XRF providing more or less the same information as can be found in the EB image shown in Fig. 5. However as can be seen from the spectrum in Fig. 4(b) when XRF spectra are collected during a longer time from a particular spot on the sample information can be obtained on trace elements such as Rb and Zr which are not visible in the corresponding EPXMA data [Fig.4(a)].154 1 X 1 o 5 1 X l O d ' 1 x103 1 x 102 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 Fig. 3 Distributions of various elements obtained via laboratory scale p-XRF from a 60 pm thick section of igneous rock. See also Fig. 5 p-XRF Using Synchrotron X-ray Sources In Fig. 4(c) a spectrum collected from the sample shown in Fig. 3 and 5 when irradiated with white synchrotron light in the NSLS X-ray microprobe is shown. In comparing Fig. 4(b) and (c) some of the advantages of using synchrotron radiation (SR) are immediately clear. Synchrotron radiation is generated when light elementary particles (electrons positrons) at rela- tivistic speeds (k close to the speed of light) are forced to change their direction of motion (ie.are accelerated). This type of radiation is very intense (a factor of 1O6-10l2 more than that which can be produced in conventional X-ray tubes) extends from the infrared range into the hard X-ray region and when sampled in the orbital plane of the electron storage ring is linearly polarized. The radiation is also naturally collimated along a direction tangential to the quasi-circular movement of the electrons in the ring. The high intensity and natural collimation make SR ideally suited for the generation of X-ray microbeams; by means of simple collimation very intense X-ray beams with cross-sections in the range 5 x 5 to 10 x 20 pm2 can be generated by means of which white beam excitation p-XRF experiments can be performed with detection limits in theppm andfg As a result of the linear polarization of the radiation the intensity of the scatter background in SRXRF spectra can be significantly reduced allowing even with polychromatic forms of excitation the determination of ppm and sub-ppm levels of trace elements.This is illustrated in Fig. 6 which shows an SRXRF spectrum collected from a 100mgcm-2 sample of NIST SRM 1571 Orchard Leaves when irradiated in the XRF spectrometer of beamline L of Hasylab (DORIS I11 ring Hamburg Germany). Fig. 7 illustrates the ability of the NSLS X-ray microprobe (beamline X26A) t o provide information on the distribution of various trace elements with a lateral resolution of about 5-10 pm.In contrast to the p-XRF data shown in Fig. 3 in Fe 1- ' 1 I I 1- ' I I ' 0 6 pi- n Fe K I1 lo tl 0 5 10 15 20 E nerg ykeV Fig.4 X-ray spectra obtained from the sample shown in Figs. 3 5 and 6 using (a) EPXMA (b) laboratory scale p-XRF and (c) p-SRXRF this case also the distribution of trace constituents such as Rb Sr and Zr can be visualized. In addition to polychromatic forms of radiation the con- struction and use of several monochromatic X-ray microprobes has been r e p ~ r t e c / . ~ ~ . ~ * In contrast to X-ray tubes synchrotron sources have a low emittance (i.e. a low source size and a small divergence) making them well suited to use with de-magnifying optics such as Bragg-reflecting or totally reflecting curved mirrors.49 Conical capillaries have also been employed by various groups to concentrate SR yielding in some cases X-ray beams of sub-pm d i m e n s i o r ~ s .~ ~ . ~ ~ ~ However because the focused beam leaving the capillary is much more divergent than the original synchrotron light entering it the applicability of these devices for high- resolution p-SRXRF may be limited. As well as providing information on the concentration ofJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 155 H 100 pm Fig. 5 Electron backscattered image of the sample shown in Fig. 3. Scale bar is 100 pm Ca K n Fe J I I I I J 0 5 10 15 20 25 EnergyJkeV Fig. 6 Synchrotron radiation induced XRF spectrum of NIST SRM 1571 Orchard Leaves ( z 100 mg cm-2) collected using a 50 x 50 pm2 white beam at Hasylab beamline L (from ref.70) certain elemental species at some of the p-SRXRF stations chemical information on sample materials can also be obtained.5k56 This is done by scanning the energy of a monochromatized X-ray beam over the absorption edge of an element of interest. The shift in the half-height of the edge reveals information on the oxidation state of the element,54.57 while pre-edge peaks and the X-ray absorption fine-structure (XAFS) above the edge contain information on its chemical surrounding^.^^^^^^^^^^ A number of overview articles (e.g. refs 45 and 61) and chapters in books (refs. 57 62 and 63) have been published which provide a comprehensive overview of the technical aspects and applications of white beam and monochromatic p-SRXRF.In the earth and environmental sciences in particular p-SRXRF has proven itself a unique and very valuable technique for trace analysis of materials which are heterogeneous at the micrometer level. Comparison of the images in Figs. 5 and 7 nevertheless shows that p-SRXRF as it can be performed at currently operating stations is still seriously limited with respect to achievable lateral resolution. The latter parameter is deter- mined by two factors on the one hand the penetrative character Fig.7 X-ray maps of various major minor and trace elements obtained using p-SRXRF (NSLS microprobe) from the central (dark) crystal shown in Fig. 5 of hard X-rays (e.g. Fe Ka X-rays can emerge from a depth of several tens of micrometers out of geological material without appreciable attenuation) and on the other hand by the relatively large photon beam sizes used (5-10 pm).Even at such second- generation sources as the NSLS smaller beam sizes can only be obtained at the expense of a considerable reduction of total beam flux and thus of analytical sensitivity. A number of the limitations of current-day p-SRXRF men- tioned above will with high probability soon be eliminated by the combined use of third-generation synchrotron X-ray sources and of X-ray optics of increasing sophistication. Around the world several large storage rings are being built which will provide synchrotron light of unprecedented intensity and brilliance. The European Synchrotron Radiation Facility (ESRF Grenoble France) is already operating and synchro- tron light will probably be available to external users from the end of 1994 onwards. Other facilities such as the APS (Advanced Photon Source Argonne IL USA) and Spring-8 (Harima Japan) will become operational in 1996 and 1998 respectively. The ALS (Advanced Light Source Berkely CA USA) began operation in October 1993.At all the third- generation rings mentioned above X-ray fluorescence micro- probes are planned or under development.6k67 For the APS and ESRF it is highly likely that undulator beamlines will be dedicated for p-SRXRF and related applications. In contrast to bending magnet sources which provide a continuous energy spectrum the output spectrum of an undulator source features sharp maxima called harmonics. By means of broad-band monochromators one of these harmonics can be isolated and used after appropriate focusing for monochromatic excitation in a p-SRXRF spectrometer.Various optical configurations can be employed for generating monochromatic microbeams from undulator sources. An overview of the various approaches investigated by the scientific groups which are active in this field can be found in ref. 68. As an example Fig. 8 shows a combination of a channel cut monochromator and two ellip- soidal mirrors mounted in a Kirkpatrick-Baez geometry. In Fig. 9(a) the attainable spot size as predicted by means of the ray-tracing code SHADOW is shown for the case of 20 keV156 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 w .- ‘ E .- - 1.000 0 al .- 4- w $ 0.100 f .- 0.010 .- 2 M2 - - - S F L MI Fig.8 Ellipsoidal mirrors in Kirkpatrick-Baez geometry focus the synchrotron beam in horizontal and vertical directions. S X-ray source; C crystal monochromator; and M1 and M2 ellipsoidal mirrors 0 10 Horizontal distance/pm 100.000 I (b’\ 1 10.000 1 ,\ layer. When using this type of advanced optics the advantages of obtaining a sub-pm beam must be weighed against the disadvantages caused by the loss of flux which results from these drastic beam. de-magnifications. At the ESRF p-XRF beamline with high probability both combinations (ie. either high flux plus a 5-10 pm spot or a lower flux plus a 1 pm spot) will be realized in two separate set-ups installed at the same beam port. Both are expected to be available for analyt- ical purposes in the near future. When in operation these spectrometers will combine the quantitative reliability and accuracy of XRF with the sensitivity and to some extent the lateral resolution of destructive microanalytical techniques such as secondary ion mass spectrometry (SIMS).Conclusions In this work an overview of recent developments in the field of microscopical X-ray fluorescence analysis is presented. Using state-of-the-art apparatus p-XRF can be performed using either conventional X-ray tubes or synchrotron storage rings as X-ray sources. Iin the former case the development of glass capillaries of different shapes as a simple and inexpensive means of focusing X-rays is opening interesting prospects for laboratory-scale p-XRF as is attested to by published appli- cations of this method in art and archeology industrial research the geosciences and clinical chemistry and biology.The use of SR provided by second-generation sources allows for trace element mapping at theppm level of detection with a lateral resolution better than 10 pm. In the near future by means of radiation originating from undulators of third- generation storage rings the sensitivity and lateral resolution of p-SRXRF are expected to reach the 10-100 ppb and 1 pm level respectively. 0.001 b 15 20 25 30 35 40 Atomic number 1 We express our gratitude to a number of people for the use of their equipment and beam time and for assistance with per- forming some of the measurements K. Jones S. Sutton M. Rivers and S. Bajt at the NSLS F. Lechtenberg S. Garbe G. Gaul and A. Knochel at Hasylab and N.Hasselberger A. Markowics and V. Valcovic of the IAEA Labs Seibersdorf. K. J. is a fellow of the Belgian National Science Fund NFWO (Brussels); this research was sponsored by FKFO (Brussels) Grant No. 2009201N. Fig. 9 (a) Dimensions of the microspot obtained by ray-tracing results using the SHADOW program for the configuration shown in Fig. 8.” (b) Predicted MDLs achievable at an ESRF low-p undulator beamline using the optical configuration shown in Fig. 8 in comparison with MDLs currently achievable at the NSLS X26A SRXRF station operated in broad beam monochromatic mode A NSLS 2 x 0.5 mm2 t = 300 s (FNsLs=4 x lo9 ph s-’ at 180 mA); and B ESRF 3 x 9 pm2 t=300s(F,,,,=5~lO~~phs-’at 100mA) photons originating from a standard ESRF low-/? undulator. Using this optical configuration a monochromatic flux density of the order of 1010photons-1pm-2 per 100mA will be attainable. As shown in Fig.9(b) the expected detection limits for a p-SRXRF spectrometer equipped with this optical con-. figuration will be situated in the 10-100 ppb region. By means of other optical elements such as Bragg-Frensel lenses or capillaries with ellipsoidal shape smaller spot sizes can be obtained but probably at the expense of a certain loss in flux. Using a Bragg-Frensel lens and an ESRF low-/? undulatoir source Kuznetsov et ~ 1 . ~ ~ obtained a lateral resolution of 0.8 pm when scanning a narrow strip of Cr; at the same beam line a tapered glass capillary was used for a micro-diffraction experiment on a Zr alIoy at 8 keV. An approximately 2 pni beam was obtained to sample a 10 pm thick ZrOz corrosion 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 References Bertin E.P. Principles and Practice of X-ray Spectrometric Analysis 2nd edn. Plenum New York 1975. Janssens K. EL and Adams F. C. J. Anal. At. Spectrorn. 1989 4 123. Spectrochim. Acta Part B 1991 46 1313 (Special Issue). Schwenke H. and Knoth J. Handbook of X-ray Spectrometry eds. Van Grieken R. E. and Markowicz A. A. Marcel Dekker New York 1993. Van Espen P. Janssens K. and Nobels J. Chemom. Intell. Lab. Syst. 1986 1 109. Adler I. Axelrod J. and Branco J. J. R. Adv. X-ray Anal. 1992 2 OOO. Heinrich K. F . J. Adv. X-ray Anal. 1992 35 15. Rose H. J. Christian R. P. Lindsay J. R. and Larson R. R. Geol. Surv. Pmf. Pap. (US) 1969 650-B B128. Nichols M. C.and Ryon R. W. Adv. X-ray Anal. 1986 29 423. Boehme D. R. Ado. X-ray Anal. 1987 30 39. Boehme D. R. and Yang N. Y. C. Microbeam Anal. 1993,2 S86. Nichols M. C. Boehme D. R. Ryon R. W. Wherry D. Cross D. and Aden G. Adv. X-ray Anal. 1987 30 45. Ryon R. Martz H. E. Hermandez J. M. Cross B. and Wherry D. A.dv. X-ray Anal. 1988 31 35. Wherry D. C. Cross B. J. and Briggs T. H. Adv. X-ray Anal. 1988 31 93. Cross B. J. Lamb R. D. Ma S. and Paque J. M. Adw. X-ray Anal. 1992 35 1255. Zaits M. A. Proc. 27th Microbeam Analysis Society Meeting,JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 157 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 Boston August 16-21 1992 ed. Bailey G. W. Bentley J. and Small J. A. San Francisco Press San Francisco 1992 p.1756. Zaits M. A. paper presented at the 42nd Denver Conference on Applications of X-ray Analysis Denver CO USA August 2-6 1993 book of abstracts p. 150. Gissot R. G. and Boehme D. R. paper presented at the 42nd Denver Conference on Applications of X-ray Analysis Denver CO USA August 2-6 1993 book of abstracts p. 152. Havrilla G. J. paper presented at the 42nd Denver Conference on Applications of X-ray Analysis Denver CO USA August 2-6 1993 book of abstracts p. 153. Pella P. A. Lankosz M. and Holynska B. Proc. 27th Microbeam Analysis Society Meeting Boston August 16-21 1992 ed. Bailey G. W. Bentley J. and Small J. A. San Francisco Press San Francisco CA USA 1992 p. 1754. Pella P. A. Lankosz M. and Holynska B. paper presented at the 42nd Denver Conference on Applications of X-ray Analysis Denver CO USA August 2-6 1993 book of abstracts p.151. Lankosz M. Pella P. paper presented at the 42nd Denver Conference on Applications of X-ray Analysis Denver CO USA August 2-6 1993 book of abstracts p. 154. Hirch P. B. in X-ray Diffraction and Polycrystalline Materials eds. Peiser H. S. Roocksby H. P. and Wilson A. J. C. Plenum Press London 1955. Mosher D. and Stephanskis S. J. Appl. Phys. Lett. 1976 29 105. Spiller E. and Segmuller A. Appl. Phys. Lett. 1974 27 101. Chung P. S. and Pantell R. H. Electron. Lett. 1977 13 527. Pantell R. H. and Chung P. S. IEEE J . Quantum Electron. 1978 14 694. Rindby A. Nucl. Instrum. Methods 1986 A249 536. Carpenter D. A. Taylor M. A. and Holcombe C. E. Adu. X-ray Anal.1989 32 115. Carpenter D. A. and Taylor M. A. Adu. X-ray Anal. 1991 34 217. Carpenter D. A. and Taylor M. A. Proc. 27th Microbeam Analysis Society Meeting Boston August 16-21 1992 ed. Bailey G. W. Bentley J. and Small J. A. San Francisco Press San Francisco CA USA 1992 pp. 1758-1759. Carpenter D. A. and Taylor M. A. Microbeam Anal. 1993,2 S84. Carpenter D. A. and Taylor M. A. paper presented at the 42nd Denver Conference on Applications of X-ray Analysis Denver CO USA August 2-6 1993 book of abstracts p. 148. Stern E. A. Kalman Z. Lewis A. and Lieberman K. Appl. Opt. 1988 27 5135. Rindby A. Engstrom P. Larsson S. and Stocklassa B. X-ray Spectrom. 1989 28 109. Thiel D. J. Hoffman S. A. and Bildenback D. Physica B+C 1989 158 314. Larsson S. Engstrom P. Rindby A. and Stocklassa B.Adu. X-ray Anal. 1990 33 623. Engstrom P. Larsson S. Rindby A. and Stocklassa B. Nucl. Instrum. Methods 1989 B36 222. Shakir N. Larsson S. Engstrom P. and Rindby A. Nucl. Instrum. Methods 1990 B52 194. Stocklassa B. Nillson G. and Paulsson N. Proceedings of the European Conference on EDXRF Myconos Greece May 30- June 6 1992 p. 61 (book of abstracts). Rindby A. Voglis P. Nilsson G. and Stocklassa B. Adu. X-ruy Anal. 1992 35 1247. Voglis P. Attaelmanan A. Engstrom P. Larsson S. Rindby A. Bostrom K. and Helander C. G. X-ray Spectrom. 1993 22 229. Gurker N. X-ray Spectrom. 1979 8 149. Bavdaz M. and Gurker N. X-ray Spectrom. 1993 22 65. Jones K. W. and Gordon B. M. Anal. Chem. 1989 61 341A. Bavdas M. Knochel A. Ketelsen P. Petersen W. Gurker N. Salehi M. H.and Dietrich T. Nucl. Instrum. Methods 1988 A266 308. 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 Van Langevelde F. Bowen D. K. Tros G. H. J. Vis R. D. Huizing A. and de Boer D. K. G. Nucl. Instrum. Methods 1990 A292 719. Goshi Y. Aoki S. Hayakawa S. Yamaij H. and Sakurai K. Jpn. J. Appl. Phys. 1987 26 L1260. Van Langevelde F. Janssens K. Adams F. C. and Vis R. D. Nucl. Instrum. Methods 1992 A317 383. Engstrom P. Larsson S. Rindby A. and Stocklassa B. Proc. 2nd European Conference on Progress in X-ray Synchrotron Research ed. Balerna A. Bernieri E. and Mobilio S. SIF Bologna 1990 pp. 283-286. Engstrom P. Larsson S. Rindby A. Buttkewitz A. Garbe S. Gaul G. Knochel A. and Lechtenberg F. Nucl. Instrum. Methods 1991 A302 547.Thiel D. J. Bilderback D. H. and Lewis A. Nucl. Instrum. Methods 1992 A317 597. Bilderback D. H. Thiel D. H. and Hoffman S. A. Microbeam Anal. 1993 2 S73. Sutton S. R. Jones K. W. Gordon B. M. Rivers M. L. Bajt S. and Smith J. V. Geochim. Cosmochim. Acta 1993 57,461. Janssens K. Vincze L. Adams F. and Jones K. W. Anal. Chim. Acta 1993 283 98 Jones K. W. and Gordon B. M. in Microscopic and Spectroscopic Imaging of the Chemical State ed. Morris D. M. Marcel Dekker New York 1993 ch. 9. Iida A. and Gohshi Y. Trace Element Analysis by X-ray Fluorescence Handbook of Synchrotron Radiation eds. Ebashi S. Koch M. and Rubenstein E. North-Holland Amsterdam 1991 Waychunas G. A. Am. Mineral. 1987 12 89. Behrens P. Felshe J. Vetter S. and Schulz-Ekloff E. K. J. Chem. Soc. Chem.Commun. 1991,61 678. Bertch P. M. Hunter D. B. Sutton S. R. Bajt S. and Rivers M. L. NSLS Annual Report 1992 eds. Hulbert S. L. and Lazarz N. M. Upton NY 1993 p. 385. Janssens K. Adams F. Rivers M. L. and Jones K. W. Proc. 11th Pfefferkorn Conf. Amherst MA August 7-14 1992 Scan. Microsc. Suppl. 1994 in the press. Vis R. D. in Applications of Synchrotron Radiation eds. Catlow C. R. A. and Greaves G. N. Blackie Glasgow 1990 ch. 13 Jones K. W. in Handbook of X-ray Spectrometry eds. Van Grieken R. E. and Markowicz A. A. Marcel Dekker New York 1993. Smith J. V. Conf. Ser. Inst. Phys. 1992 160 605. Vincze L. Janssens K. and Adams F. C. Con$ Ser. Inst. Phys. 1992 160 613. Gohshi Y. personal communication. Chapman K. L. Thompson A. L. and Underwood J. H. Proc. 42nd Denver Conference on Applications of X-ray Analysis Denver CO USA August 2-6 1993 book of abstracts p. 119. Proceedings of a Workshop on X-ray Focusing Techniques and Applications NSLS Upton NY USA May 18 1993 ed. Rothman E. Lazarz N. and McWhan D. NSLS Newsl. July 1993 p. 11. Kuznetsov S. Snigirev A. Snigireva I. Engstrom P. and Riekel C. Appl. Phys. Lett. submitted for publication. Janssens K. Vincze L. Lechtenberg F. Garbe S. Gaul G. Knochel A. and Adams F. X-ray Spectrom. 1994 submitted for publication. Lai B. Chapman K. and Cerrina F. Nucl. Instrum. Methods 1988 A266 544. VO~. 4 ch. 9 pp. 307-348. pp. 514-550. Paper 3/05717H Received September 22 I993 Accepted November 18 1993