Survey of sampling techniques for solids suitable for microanalysis by total-reflection X-ray fluorescence spectrometry† Reinhold Klockenka�mper* and Alex von Bohlen Institut fu�r Spektrochemie und Angewandte Spektroskopie (ISAS), Bunsen-KirchhoV-Str. 11, D-44139 Dortmund, Germany Received 5th October 1998, Accepted 16th December 1998 A survey of solid sampling techniques is given, suitable for microanalysis or even ultramicroanalysis by totalreflection X-ray fluorescence (TXRF). A sample amount of 1 ng to 100 mg is placed on a clean, flat carrier usually made of quartz glass or PlexiglasA.For quantification, a drop is added to the sample with a known amount of a single element serving as internal standard. Solid samples can be applied as small particles, fine powders, thin sections or deposits. Several techniques of solid sampling have been examined and one example of each technique is given demonstrating its suitability: direct placing of individual particles, suspension of powdered materials, collection of air dust by impaction, the touchstone technique for metals, laser ablation for local analysis, the Q-tip technique for paints and inks, freeze-cutting of organic materials and direct contamination control of wafers.Total-reflection X-ray fluorescence (TXRF) is a special variant of the sample is recorded by a spectrometer or a semiconductor detector. In comparison, TXRF uses an X-ray tube with a of energy-dispersive X-ray fluorescence (EDXRF) analysis which, today, is widely known.It was first proposed by two narrow or ‘line focus’, emitting a primary beam which is shaped like a strip of paper. The high energy part of its Japanese scientists, Y. Yoneda and T. Horiuchi, in 1971. TXRF has achieved remarkable importance in atomic spec- spectrum may be cut oV by a low-pass filter. In any case, the primary beam strikes a flat glass carrier at a very small angle trometry, and a critical evaluation of recent developments places this method in a leading position.1 Today there are and, consequently, it is totally reflected at the carrier.The sample has to be placed on the carrier in a small amount so three major and three minor suppliers of TXRF instruments, and nearly 300 pieces of equipment are in use worldwide. that total reflection is scarcely hindered. The fluorescence radiation of the sample is recorded by an energy-dispersive Several reviews have been published recently,2–5 and also one monograph6 is available, which is entirely devoted to TXRF. detector, mostly by an Si(Li) detector.The equipment in Figs. 1(a) and 1(b) mainly diVers in geometry: XRFA uses a An inherent feature of conventional X-ray fluorescence analysis (XRFA) is the investigation of any sample in its combination of about 45°/45° for incident and take-oV angle, while TXRF uses a combination of 0.1°/90° (nearly a 0°/90° original phase, i.e. of a sample ‘as is’. Solids can be analysed as solids, liquids as liquids and gases as gases.In contrast with combination). The analytical examples in this paper were carried out on most analytical methods, XRFA is a non-consumptive and, in several cases, a non-destructive method. Vaporisation or even an EXTRA II instrument (Rich. Seifert & Co, Ahrensburg, Germany), equipped with an Mo and a W tube (visible line atomisation of sample material is not necessary. TXRF is a modification of conventional XRFA, but is preferably applied to liquid samples and solutions after an easy preconcentration step: an aliquot is pipetted and dried on a sample carrier, since only the dry residue can be analysed by TXRF.Solid samples are favourably analysed after ashing and/or digestion. This preparation step is carried out for the purpose of homogenisation and possibly preconcentration before a small part of the sample is chosen for analysis. Nevertheless, TXRF can be applied directly to solid samples, which is recommended if microanalysis is needed, i.e.if only a small or minute amount of a sample is present or available. DiVerent solid sampling techniques have been developed for microanalysis by TXRF, and a survey of these techniques is given in this paper after a short review of fundamentals. Fundamentals Instrumental peculiarities of TXRF The essential diVerences between conventional XRFA and the variant TXRF are demonstrated in Figs. 1(a) and 1(b), respectively.XRFA uses an X-ray tube with a broad focus, emitting a cylindrical primary beam. The fluorescence radiation Fig. 1 Simplified instrumental set-up: (a) for conventional XRFA; (b) †Presented at the 8th Solid Sampling Spectrometry Colloquium, Budapest, Hungary, September 1–4, 1998. for the variant TXRF. J. Anal. At. Spectrom., 1999, 14, 571–576 571focus of 40 mm×8 mm), operated at 50 kV and <40 mA. For used. For direct analysis of organic samples, the lower mg g-1 range can be reached if 100 mg are applied.For inorganic energy-dispersive analysis and data processing, a Link System QX 2000 with Si(Li) detector (Oxford Instruments, High samples, e.g. oxides or metals, only traces of some 100 mg g-1 can be detected if microgram amounts are applied. Major and Wycombe, UK) was used. The total counting time for analysis was mostly set to 100 s. minor constituents can still be detected if only nanogram amounts are available for analysis.Prerequisites of TXRF Quantification by internal standard. Quantification is gener- Sample carriers. Sample carriers are mostly made of quartz ally carried out by an internal standard element added to the glass, but also silicon, glassy carbon, boron nitride and sample in a known amount. TXRF primarily determines the PlexiglasA can be used. They have to be optically flat and unknown mass of an analyte element: extremely clean at their surface. The cleanliness can be checked by a blank spectrum taken in advance of the unloaded carrier.mx= Nx/Sx Nis/Sis mis (1) If spectral peaks of any contaminants appear, the carrier has to be cleaned again. The cheap Plexiglas carriers are usually where m is the mass, N is the net intensity and S is the relative delivered in a suYciently clean condition. sensitivity, either of the analyte (x) or of the internal standard Clean carriers can be used directly for the deposition of (is). The S values are usually known since they were measured solid samples.For the deposition of liquid samples or suspen- by means of standard solutions. This kind of simple calibration sions, they have to be hydrophobic, otherwise the drops are needs to be carried out only once or very seldom. dispersed over the surface. Plexiglas, silicon, glassy carbon Relative mass fractions or concentrations can be determined and boron nitride are hydrophobic by nature, but quartz glass if the total sample mass mtot or the total sample volume Vtot is hydrophilic.It has first to be covered by a silicone solution is measurable: which, after drying, leaves a thin hydrophobic film on the surface. cx= mx mtot or mx Vtot (2) Presentation of solid samples. Conventional XRFA uses a If the total amounts cannot be determined, relative mass sample amount of 1 mg to 10 g or 0.1–50 ml, so that XRFA fractions can at least be calculated (relative in relation to the is more a macro- than a micro-method.In contrast, TXRF sum of all detected elements call): only uses a sample amount of 1 ng to 100 mg of solids or 1–100 ml of liquids yielding micrograms of a dry residue. crx= Nx/Sx SNj/Sj call (3) TXRF, consequently, is a micro- or even an ultramicromethod. This restriction is necessary since the capacity of the This sum may be 100% or simply be set to 100%. For this detector would be exceeded and the total reflection at the simple case, not even an internal standard is needed.carrier would be disturbed if a larger amount of sample This method of easy quantification was applied many times material was applied. The appropriate amount of sample mass and proved reliable. The prerequisite is a minute amount of can easily be adjusted by means of the detector, keeping its sample placed as a thin layer or a thin-film-like sample on a dead time below 50%. clean carrier so that matrix eVecte negligible. This condition Solid samples can be applied as one or several individual is usually not fulfilled for macro-samples of conventional particles, as a suspension or slurry of finely powdered mate- XRFA, but fortunately it is met for the micro-samples of rials, as a thin section of organic materials, as a thin foil of TXRF.The internal standard added via a drop should evenly metals or polymers or as a deposit of metals or non-metals. wet the sample, but it need not be homogeneously distributed In any case, the samples have to be placed in the centre of a within the small or minute sample.carrier as a thin-film-like sample with a thickness of some 0.1–10 mm. For quantification, a drop of an aqueous standard solution has to be added which usually is pipetted onto the Examples of solid sampling deposited sample. Only in the case of suspensions is it rec- General survey of techniques ommended to add a standard solution previously. This solution should contain a single element initially not present in the Numerous solid samples of diVerent kinds and shapes have sample, but afterwards serving as internal standard for all been investigated and diVerent sampling techniques have been other elements in the sample.Generally, rare elements are applied. Individual particles can easily be taken and placed chosen, e.g. Ga, Ge, Se or Y. Drying by evaporation and onto a carrier for TXRF analysis. Fine powders can be recording of the TXRF spectrum follows. suspended, and the suspensions can be pipetted onto a carrier and dried.Air dust or aerosols can be collected by adhesives Features of TXRF covering a carrier. Solid materials, such as metals, can be rubbed on a hard quartz glass carrier so that they are rubbed Detection limits. All elements with atomic numbers Z14 oV. Micro-regions of metals or non-metals can be removed by (silicon) can easily be detected by TXRF. The detection of laser ablation, and some material can be emitted and deposited lighter elements is hampered by several obstacles which are on a glass carrier.Colours or inks can be wiped oV the surface inherent to energy-dispersive XRFA, although some of them of paintings or manuscripts by means of a cotton-wool tip, can be overcome by specifically designed instrumentation.7,8 often called Q-tip, and the loaded Q-tip can be dabbed onto While detection limits of conventional XRFA are at the a carrier. Tissues or polycarbons, generally organic materials, migrogram level, those of TXRF go down to the low picogram can be freeze-cut and the sections can be placed on carriers.level. The improvement is caused by the extremely reduced Wafers can directly be applied and the contaminants on them spectral background according to the minimum portion of the analysed. One example of each technique is chosen in order impinging photon flux that penetrates into the carrier. It is to demonstrate its suitability. proportional to the low transmissivity and to the small glancing angle of incidence.Direct placing of individual particles Relative detection limits are dependent on the sample matrix and sample mass or volume being applied. For high purity Individual particles, such as grains or crumbs of several nanograms or single pieces of hair or textile fibres of several waters or acids, the ng l-1 region can be reached if 100 ml are 572 J. Anal. At. Spectrom., 1999, 14, 571–576micrograms,4,9 can simply be placed onto a carrier.A wooden or repeatability of ten determinations was 3%; the accuracy of the mean or the trueness was 4%. toothpick is recommended for this manipulation. Fig. 2 shows conglomerates of several grains of a pigment powder. The edge length is between 0.1 and 2 mm; the total mass is about Collection of air dust by impaction 1 ng. This is the lower limit of sample amount needed for Airborne particles can be fractionated according to size and TXRF analysis. collected by a cascade impactor with several stages stacked on top of each other, e.g.a Battelle- or Anderson-type impactor. Suspension of powdered materials Air is pumped through one nozzle per stage and dust particles are deposited onto an impaction plate behind the nozzle Fine powders or pulverised materials can be applied as individual particles, but also via a suspension.10,11 Natural suspen- according to their inertia or size. Larger particles are deposited behind a coarser nozzle in an upper stage, smaller particles sions, e.g.river water with a certain portion of suspended particulate matter, can be applied directly.12 For powdered behind a finer nozzle in a lower stage. Simple Plexiglas discs fitted for TXRF devices can be used as impaction plates.14–16 materials, an aliquot can be taken up by a pure solvent; usually several milligrams of the powder are suspended in 5 Particles of wet air stick to the flat discs which can be directly applied to TXRF.Particles of dry air, however, are bounced or 10 ml of water. The aqueous suspension is thoroughly homogenised with a magnetic stirrer or by ultrasonication, or blown oV.14 To avoid this eVect, the discs must first be coated, for example with a medical petrolatum,14 i.e. and a small drop of 10 or 20 ml is pipetted onto a carrier. Afterwards, the single drop is dried by evaporation leaving a VaselineA, or be sprayed, for instance with a medical plaster.17 Dust particles from a total volume of 1 m3 of ambient air dry residue of about 10 mg.A synthetic mixture of three diVerent pigments (titanium were collected in a Battelle-type impactor within 1 h. Fig. 3 shows a spectrum of particles with a diameter of 1–4 mm white, zinc white and strontium yellow) was suspended in petroleum ether with a mixing ratio of 15151. Aliquots of collected on stage no. 4. Ge (100 ng) was added as internal standard. Detection limits are about 0.1 ng m-3, so that 100 ng of the powder were used for TXRF analysis.An internal standard was not needed. First, relative mass fractions pollution during the course of a day can be observed. However, care must be taken to avoid contamination. This can be caused of the four elements Ti, Zn, Sr and Cr were determined after eqn. (3). Since oxygen could not be detected, the relative mass by erosion of the impactor walls or nozzles if made of stainless steel. For that reason, the impactor in use was reconstructed fractions of the respective oxides TiO2, ZnO and SrCrO4 were calculated according to stoichiometry.Finally, the mixing with Plexiglas; even better is an antistatic polymer.17 proportions of these oxides or pigments were determined by normalisation to 33.3%. Table 1 gives the results. The nominal Touchstone technique for metals proportion of 15151 was confirmed by TXRF. The precision When only a survey analysis is required, an old famous technique can be applied: the touchstone technique18,19 explained in Fig. 4. Usually, black jasper is used as the touchstone and the test needles as well as a piece of a precious metal are rubbed on it. The coloured strokes provide a hint of the composition of the metal. This technique was already known in antiquity. The touchstone for TXRF is a quartz glass carrier.6 Solid objects with lower hardness can be rubbed on it in a single stroke; or a quartz glass carrier can be rubbed on a fixed object. In both cases, a small amount of sample material will be smeared onto the carrier, which can be analysed by TXRF.A gold ring may serve as an analytical example: 20 ng were rubbed oV altogether and Mn (5 ng) was added as internal standard. The TXRF spectrum showed the major components Au, Cu, Ni and Zn. The quantitative results are given in Table 2. The gold value of 574 mg g-1 corresponds quite well with the hallmark 585. However, this is a favourable case; it Fig. 2 Conglomerates of several pigment grains placed on a Plexiglas carrier by a wooden toothpick and photographed by a scanning electron microscope. Numerous small, irregularly shaped grains are mixed with larger cubic grains of 2–3 mm edge length.The total mass of ca. 1 ng is suYcient for TXRF analysis. Table 1 Determination of the mixing proportions of three diVerent pigments, titanium white (TiO2), zinc white (ZnO) and strontium yellow (SrCrO4), in a suspension (after Klockenka�mper et al.13). The mean values and standard deviations are given for ten samplings of the suspension and TXRF analysis.The nominal proportion was given by 15151 Relative mass Pigment Relative mass Mixing Element fraction (%) or oxide fraction (%) proportion Fig. 3 TXRF spectrum of air dust collected from ambient air near Dortmund city. Dust particles of 1–4 mm in diameter were deposited Ti 27.0±0.8 TiO2 31.4±0.9 0.94±0.03 on stage no. 4 of a Battelle-type impactor made of Plexiglas. Ge was Zn 39.8±1.6 ZnO 34.6±1.4 1.04±0.04 added as internal standard.The spectrum gives the counts of photons Sr+Cr 33.3±1.0 SrCrO4 33.9±1.0 1.02±0.03 recorded within 100 s as a function of their respective energy. J. Anal. At. Spectrom., 1999, 14, 571–576 573of diVerent elements can be recorded. The lateral resolution may be 10 mm. Q-tip technique for paints and inks Valuable works of art, such as oil paintings or book illustrations, should be analysed only non-destructively. A very gentle method of micro-sampling has been developed for this purpose: the Q-tip method.13 By means of a dry and clean cotton-wool tip, often called a Q-tip, a minute amount of paint or ink can be removed from the surface of paintings or manuscripts.DiVerent Q-tips can be applied to diVerent spots. This technique can be regarded as non-destructive since only a microgram amount is removed. It has to be presupposed that the paintings are not covered with a varnish layer, unless this layer is to be removed for the purpose of restoration. The Q-tips can be locked in bottle caps and transported.For analysis, they are dabbed onto a glass carrier once and less Fig. 4 Woodcut of Lazarus Ercker, Prague 1574, demonstrating the touchstone technique.19 A and B are two sets of test needles; C is a than 100 ng are transferred. This technique was applied to cylindrical or a cubical touchstone made of black jasper. samples of some 20 oil paintings and book illustrations from 12 museums, in Europe, and many questions of restoration, conservation and dating have been answered.21–23 Table 2 Quantitative results for a gold ring after sampling by the One example was presented by Moens et al.23 Paintings of touchstone technique and TXRF analysis. Nearly 20 ng were rubbed the old masters often suVer from so-called ultramarine disease oV altogether; 5 ng of Mn were added as internal standard which turns parts of a glowing blue into a greyish green.Element Mass/ng Mass fraction/mg g-1 Pigments from blue spots and from green spots of the cloak of a Madonna were taken and identified. The pigments of the Ni 2.3 118 blue spots obviously contain ultramarine; those of the green Cu 4.6 234 spots, however, showed the element pattern of a totally Zn 1.5 74 diVerent pigment as demonstrated in Fig. 5. This pigment, Au 11.4 574 Sum 19.8 1000 containing K, Co and As, is called smalt; it is also a blue pigment and is based on cobalt glass.However, it is less valuable than ultramarine and changes colour over the centuries. Thus, so-called ultramarine disease is actually a smalt should be mentioned that deviations occur when preferential disease. abrasion of a soft component occurs. Freeze-cutting of organic materials Laser ablation for local analysis Organic materials, such as plant or animal tissues, and also A targeted or local analysis of solids can be carried out if a food, can be ashed or digested prior to analysis, but can also laser beam is used.The laser beam, e.g. of a Nd-YAG laser, be analysed as solid materials. As known from histology, such is focused by an objective of a long focal length onto the organic materials can be frozen and cut by a freezing micro- sample. Material is ejected from a small crater and partly tome. By means of a plastic stretcher, a thin section of deposited on a Plexiglas carrier, which can afterwards be 10–20 mm can be slid onto the scalpel of the microtome.This analysed. section can be placed on a glass carrier by gently touching it. This sampling technique was applied by Bredendiek-Ka�mper Subsequently, it has to be spiked with an internal standard, et al.20 to a ceramic superconductor Y1Ba2Cu3O6.9. A crater dried by evaporation and analysed by TXRF.24–26 of ca. 100 mm in diameter and 20 mm in depth was shot by a Table 4 shows the results for a section of lung tissue obtained single laser pulse and 22 ng of material were deposited on a from a foundry worker.Quantification was carried out after Plexiglas carrier. For quantitative analysis, Ge was used as diVerence weighing of the carrier with and without the dried internal standard. Table 3 gives the results. The small devisection on a microbalance. The total dry mass was determined ations between stoichiometry and TXRF analysis are about to be nearly 80 mg. The mass fractions of heavy metals given 5%, demonstrating a good accuracy.By successive ablations from neighbouring spots, material can be deposited side by side on a carrier and a total line scan Table 3 Quantitative results for a ceramic superconductor Y1Ba2Cu3O6.9 obtained by laser ablation and TXRF analysis of the deposit.20 Ge (5 ng) was used as internal standard. Mass fractions were normalised to the sum of the detected elements of 834 mg g-1 (besides oxygen) Mass fraction/mg g-1 Relative Element Mass/ng TXRF Stoichiometry deviation (%) Y 3.3 125 134 -6.7 Ba 11.6 435 413 +5.2 Cu 7.3 274 287 -4.5 Sum 22.2 834 834 5.5a Fig. 5 Element pattern of the blue pigment smalt showing its main aQuadratic mean. constituents besides oxygen. 574 J. Anal. At. Spectrom., 1999, 14, 571–576Table 4 Quantitative TXRF results for lung tissue obtained from a foundry worker. Sampling was carried out by a freezing microtome and a section of 14 mm in thickness with a dry mass of 80 mg was analysed. Ga (2 ng) was applied as internal standard.The values for the normal range and the upper limit of a population of 125 people have been determined by Baumgardt27 Normal range/mg g-1 Upper limit/mg g-1 Element Mass fraction/mg g-1 (95% of a population) (99.9% of a population) Ti 1400 Cr 17 0.1–10 34 Fe 5460 100–2500 4000 Ni 5.3 0.1–50 500 Cu 13 2.5–40 49 Zn 85 Up to 100 Zr 83 Pb 7.4 0.1–3.0 3.7 in mg g-1 are very high, and those of Cr, Fe and Pb are deposited in thin layers on the surface of wafers. These contaminants are called bulk type, particulate type and thin significantly above the normal range assessed for 95% of a population; Fe and Pb are above an upper limit exceeded by layer type.4,6 Thin layered structures of nanometre thickness covering a silicon substrate can also be examined.The com- only 0.1% of the population. This heavy metal contamination was probably caused by the occupational exposure of the position, thickness and density of individual layers can be determined.34–38 For this purpose, a tilting device is needed in foundry worker.order to carry out an angle scan and to record intensity profiles of the respective elements as a function of the angle Direct contamination control of wafers of incidence. The last example is taken from the field of semiconductor technology. Large but thin discs of silicon, so-called wafers, Conclusions are a basic material of semiconductors. The wafers are initially polished so that they are optically flat and they themselves Several advantages of solid sampling for TXRF can be enumercan serve as totally reflecting carriers for TXRF. ated, but some prerequisites must be observed.Only extremely Contamination at or in the surface of a wafer can be analysed small amounts (ng or mg) are needed for analysis which directly.28,29 Fig. 6 shows a spectrum of a wafer strongly therefore is nearly non-consumptive or non-destructive. contaminated by several elements. Quantitative results are Several techniques of solid sampling have been developed and usually given in numbers of atoms per cm2.For the internal can be applied to a large variety of sample materials: inorganic standard Se, 1011 atoms are equal to 13 pg. This value is still as well as biogenic solid materials. On the other hand, the abohe detection limit. taking and handling of micro-samples is not easy and represen- Even lower detection limits can be obtained after a so-called tative sampling can be a problem.Extreme cleanliness must vapour-phase decomposition.30–33 In a special reactor, a solu- be observed; clean bench working is a must. tion of HF is heated and evaporated. The vapour condenses Simultaneous multi-element detection is advantageous, and on the cold wafer and its native oxide layer of nanometre a simple and reliable quantification by an internal standard thickness is decomposed. Simultaneously, the contaminants ensures a good precision and high accuracy.On the other on the surface are dissolved. Afterwards, the film of aqueous hand, the exclusion of light elements with atomic numbers fine droplets wetting the wafer is collected over its entire area below 14 is regrettable. Light elements can be analysed only as a drop of water and this drop is placed on a carrier for with special detectors and excitation sources in a vacuum TXRF analysis. Detection limits are improved by a factor of chamber. Nevertheless, TXRF is widely applicable to the 100, so that the level of 108 atoms cm-2 is now accessible at microanalysis of solids.the expense of information on the spatial distribution of the elements. This study was financially supported by the German Further techniques have been developed for surface and Bundesministerium fu� r Bildung, Wissenschaft, Forschung und thin layer analysis. TXRF is capable of distinguishing between Technologie and the Ministerium fu� r Schule und diVerent types of contaminants: impurities distributed homo- Weiterbildung, Wissenschaft und Forschung des Landes geneously in a wafer, located in granular particles or evenly Nordrhein-Westfalen.References 1 G.To� lg and R. Klockenka�mper, Spectrochim. Acta, Part B, 1993, 48, 111. 2 R. Klockenka�mper and A. von Bohlen, J. Anal. At. Spectrom., 1992, 7, 273. 3 R. Klockenka�mper and A. von Bohlen, X-Ray Spectrom., 1996, 25, 156. 4 A. Prange and H. Schwenke, Adv. X-Ray Anal., 1992, 35, 899. 5 H. Aiginger and C. Streli, Spectrosc. Eur., 1997, 9, 16. 6 R. Klockenka�mper, Total-Reflection X-Ray Fluorescence Analysis, Wiley, New York, 1997. 7 C. Streli, H. Aiginger and P. Wobrauschek, Spectrochim. Acta, Part B, 1993, 48, 163. 8 C. Streli, P. Wobrauschek, W. Ladisich, R. Rieder and H. Aiginger, X-Ray Spectrom., 1995, 24, 137. Fig. 6 TXRF spectrum of a silicon wafer strongly contaminated by 9 A. Prange, U. Reus, H. Bo�ddeker, R. Fischer and F.-P. Adolf, Adv. X-Ray Chem. Anal.Jpn., 1995, 26s, 1. Cl, Ca, Fe, Zn and some minor pollutants. Se was added as internal standard; counting time, 500 s. All numbers are given in 1011 10 A. von Bohlen, R. Eller, R. Klockenka�mper and G. To� lg, Anal. Chem., 1987, 59, 2551. atoms cm-2. J. Anal. At. Spectrom., 1999, 14, 571–576 57511 T. Graule, A. von Bohlen, J. A. C. Broekaert, E. Grallath, R. 24 A. von Bohlen, R. Klockenka�mper, H. Otto, G. To� lg and B. Klockenka�mper, P. Tscho� pel and G. To� lg, Fresenius’ J.Anal. Wiecken, Int. Arch. Occup. Environ. Health, 1987, 59, 403. Chem., 1989, 335, 637. 25 A. von Bohlen, R. Klockenka�mper, G. To� lg and B. Wiecken, 12 U. Reus, B. Markert, C. HoVmeister, D. Spott and H. Guhr, Fresenius’ J. Anal. Chem., 1988, 331, 454. Fresenius’ J. Anal. Chem., 1993, 347, 430. 26 R. Klockenka�mper, A. von Bohlen and B. Wiecken, Spectrochim. 13 R. Klockenka�mper, A. von Bohlen, L. Moens and W. Devos, Acta, Part B, 1989, 44, 511. Spectrochim. Acta, Part B, 1993, 48, 239. 27 B. Baumgardt, PhD Thesis, Ruhr-Universita�t Bochum, 1985. 14 A. Salva`, A. von Bohlen, R. Klockenka�mper and D. Klockow, 28 W. Berneike, Spectrochim. Acta, Part B, 1993, 48, 269. Quim. Anal., 1993, 12, 57. 29 V. Penka and W. Hub, Spectrochim. Acta, Part B, 1989, 44, 483. 15 R. Klockenka�mper, H. Bayer and A. von Bohlen, Adv. X-Ray 30 C. Neumann and P. Eichinger, Spectrochim. Acta, Part B, 1991, Chem. Anal. Jpn., 1995, 26s, 41. 46, 1369. 16 B. Schneider, Spectrochim. Acta, Part B, 1989, 44, 519. 31 R. S. Hockett, Adv. X-Ray Anal., 1994, 37, 565. 17 M. Schmeling and D. Klockow, Anal. Chim. Acta, 1997, 346, 121. 32 L. Fabry, S. Pahlke and L. Kotz, Adv. X-Ray Chem. Anal. Jpn., 18 G. Agricola, De re Metallica. Libri XII, Froben-Verlag, Basel, 1996, 27, 354. 1556. 33 L. Fabry, S. Pahlke and L. Kotz, Fresenius’ J. Anal. Chem., 1996, 19 Lazarus Ercker’s Treatise on Ores and Essaying.—Facsimile 1580, 354, 266. translators A. G. Sisko and X. Y. Smith, University of Chicago 34 U. Weisbrod, R. Gutschke, J. Knoth and H. Schwenke, Fresenius’ Press, Chicago, 1951. J. Anal. Chem., 1991, 341, 341. 20 S. Bredendiek-Ka�mper, A. von Bohlen, R. Klockenka�mper, A. 35 J. Knoth, R. Bormann, R. Gutschke, C. Michaelsen and H. Quentmeier and D. Klockow, J. Anal. At. Spectrom., 1996, 11, Schwenke, Spectrochim. Acta, Part B, 1993, 48, 285. 537. 36 H. Schwenke and J. Knoth, Anal. Sci., 1995, 11, 533. 21 W. Devos, L. Moens, A. von Bohlen and R. Klockenka�mper, 37 D. K. G. de Boer and W. W. van den Hoogenhof, Spectrochim. Studies in Conservation, 1994, 40, 153. Acta, Part B, 1991, 46, 1323. 22 B. Wehling, P. Vandenabeele, L. Moens, R. Klockenka�mper, A. 38 D. K. G. de Boer, Phys. Rev. B, 1991, 44, 498. von Bohlen, G. Van Hooydonk and M. de Reu, Mikrochim. Acta, in the press. 23 L. Moens, W. Devos, R. Klockenka�mper and A. von Bohlen, J. Trace Microprobe Techniques, 1995, 13, 119. Paper 8/0769