MARCH 1979 Vol. 104 No. 1236 The Analyst Energy-dispersive X-ray Emission Analysis A Review W. C. Campbell Imperial Chemical Industries Limited Petrochemicals Division Reseavch and Development Department, P.O. Box 90 Wilton Middlesbrough Cleveland TS6 8 J E Summary of Contents Introduction Instrumentation Excitation X-ray tubes Radioisotopes Electrons and protons Detectors Electronics Pre-amplifier Amplifier Multi-channel analyser Dead time Data processing Spectral features and interpretation Sum peaks Escape peaks Diffraction peaks Anomalous silicon gold and argon peaks Scatter peaks Spectral background Comparison of wavelength- and energy-dispersive systems Applications Atmospheric particulates Waters Clinical and biochemical Rocks ores and cement Metals and alloys Coal and petroleum On-stream analysis Others Future developments Keywords ; Review ; energy-dispersive X-ray emission awalysis Introduction In their review of X-ray fluorescence analysis in 1970 Carr-Brion and Paynel stated that resolutions of the order of 190 eV at 6 keV had been reported for semiconductor detectors and that better values were to be expected with improvement in detectors and associated electronics.Today detectors are routinely produced with resolutions in the 150-160-eV range and values as low as 140 eV can be attained. Energy dispersion using a lithium-drifted silicon Si(Li) detector was first introduced as a practical tool for X-ray spectrometry in the mid-l960~.~-~ Initially the major impact of the new technology was as an accessory on electron beam microprobes and scanning electron microscopes.6 It was not long however before dedicated X-ray spectrometers were being built around the new detectors.7-l2 Today energy-dispersive attachments are almost standard on electron microscopes and a number of manufacturers offer dedicated X-ray fluorescence equipment for qualitative and quantitative analysis.17 178 CAMPBELL ENERGY-‘DISPERSIVE X-RAY Analyst Vol. 104 It is intended to limit this review to those techniques which utilise the energy-dispersive properties of semiconductor detectors specifically the Si( Li) detector. There is some confusion over the terminology applied to this field of analytical chemistry. Detectors, such as the scintillation and flow proportional counters used in wavelength-dispersive X-ray fluorescence analysis are capable of limited energy resolution.It is therefore possible to consider these as energy-dispersive detectors and indeed use is made of this property in the application of “pulse-height analysis.” However the inability of these detectors to provide sufficient spectral resolution leads to their use in combination with diffraction crystals. Terms such as semiconductor solid-state non-dispersive and energy-dispersive have appeared associated with the Si(Li) detector. It is intended here to follow the guidelines given in the I UPA C Information BuZZetin on X-ray emission spectroscopy,13 for terminology and symbols. It has become the convention to express X-ray wavelengths in Angstroms when considering wavelength-dispersive systems and X-ray energy in kiloelectronvolts when considering energy-dispersive systems.The principal difference between energy-dispersive electron microscope attachments and energy-dispersive X-ray fluorescence analysers is in the mode of excitation used. The techniques share much of the data collection and data processing technology. It is therefore difficult to differentiate between the two systems in an absolute manner and some reference will be made to electron-excitation systems. For those new to the technique a number of books and descriptive articles are available with respect to both X-ray fluorescence analysis in genera11v14-20 and energy-dispersive X-ray fluorescence analysis in p a r t i c ~ l a r . 7 ~ ~ ~ ~ ~ ~ - ~ 0 Instrumentation Fig.1 shows the components of a typical energy-dispersive X-ray fluorescence analyser. The source of excitation shown is the X-ray tube but of course excitation of secondary X-rays can be achieved by using a variety of sources including electrons protons and other charged particles. The building blocks of the system are a source of excitation the sample compartment the solid-state Si(Li) detector the electronic package including pre-amplifier, amplifier and multi-channel analyser (M.C.A.) and the data processing package generally including computer with relevant software to convert the raw data into meaningful results. The over-all system efficiency of an energy-dispersive X-ray fluorescence instrument is a Sample changer Cont ro I assem b I y unit Fig.1. Typical energy-dispersive X-ray fluorescence spectrometer March 1979 EMISSION ANALYSIS. A REVIEW 179 function of a number of parameters including geometry mode of excitation excitation cross-section fluorescence yield and detector efficiency. Cothern et aL31 investigated the system efficiency for the situation where the source of excitation was broad-band X-rays. Excitation As in conventional wavelength-dispersive X-ray fluorescence analysis it is necessary to remove core electrons from the atoms of interest in order to produce the secondary fluorescent X-rays which are characteristic of the elements present in the sample. Normally X-rays, from an X-ray tube are used to fulfil this function and this is still true of most commercial energy-dispersive X-ray fluorescence spectrometers.The alternatives to the X-ray tube are electrons protons and radioisotopes all of which are capable of ejecting core electrons and all of which have various advantages and limitations. Jaklevics2 has compared the use of electrons charged particles (protons or alpha particles) and Xxays (continuous or mono-energetic) as excitation sources for energy-dispersive X-ray analysis. Electron excitation was shown to have much poorer limits of detection. The high continuous background found in electron-induced spectra increases the difficulty of determining low concentrations. Middleman and GelleF have demonstrated the improved peak to background ratios that can be achieved in the X-ray spectra from an electron microscope using X-ray excitation instead of the usual electron beam.Reldy et aE.34 have used muons to excite secondary X-rays. This produces a muonic X-ray spectrum in which the characteristic X-rays are raised in energy hence allowing the light elements to be determined more easily. X-ray Tube X-ray tubes on conventional wavelength-dispersive spectrometers use up to 3 kW of power and produce a high characteristic X-ray flux from the sample In energy-dispersive X-ray fluorescence systems all X-rays are incident simultaneously on the detector and because there is a finite counting capacity it is necessary to modify the X-ray tube output to reduce the secondary X-ray flux. There are basically two approaches to this problem. One is to reduce the power of the X-ray tube to around 10 W and the other to use a secondary target system.Fig. 2 shows the configurations typical for primary or direct excitation and secondary excitation. Specimen Specimen Optional I/’ \\ / / \ \ ,’ filter ; /X\ X’ 1 ;<; ’\ I ‘\ ‘\ \ n Secondary target X-ray tube Detector Detector (a) (b) Fig. 2. X-ray excitation (a) primary or direct; ( b ) secondary. The low-power tube used for primary excitation emits a broad band of X-ray energies from just below the tube potential. Used directly this has the advantage of exciting a wide range of elements but unfortunately produces a high spectral background due to scattering of the radiation by the sample. The use of a filter placed between the tube and the sample attenuates the X-rays and can be used to moderate the primary broad-band radiation to approximate to a mono-energetic source.A proper choice of filter will increase the sensitivity for particular elements while reducing the spectral background over a given region 180 CAMPBELL ENERGY-DISPERSIVE X-RAY Analyst Vol. 104 In the secondary-excitation geometry X-rays from a high-power tube impinge on a chosen target material. The target material itself is’ induced to emit characteristic and mainly monochromatic X-rays which are made to fall upon the sample. This produces a well monochromated beam that gives rise to a lower spectral background contribution. Gedcke et aZ.35 compared the detection limits that could be obtained with the two geometries for a variety of elements and concluded that over the 5-30-keV range there was little to choose between the two techniques.However in a similar study Artz and Short36 found that the best sensitivity was achieved in the atomic number range 35-60 with direct radiation from a tungsten X-ray tube but that secondary excitation with properly chosen targets produced a higher sensitivity for other elements. In a study of the detection limits for light elements in a “synthetic rock mixture” and NBS orchard leaves Anselmo37 found that secondary excitation was generally superior. Investigations of this nature are complicated by a dependence on system geometry. Both primary- and secondary-excitation systems are available on commercial spectro-meters. The primary-excitation system with the optional filter appears to offer a more flexible approach in terms of exciting a large number of elements simultaneously or effici-ently exciting just a few.The secondary target geometry probably gives somewhat better sensitivity over narrow ranges necessitating the use of a variety of targets to cover the X-ray spectrum of interest. Counting times on energy-dispersive X-ray fluorescence instru-ments are typically 100-1000 s owing to the low count-rate capability of the detection and amplification system. This places a high stability requirement both long and short term, on the output from the X-ray tube. Low-power generators are more stable than high-power generators and this favours the use of primary excitation. Van Espen and ad am^^^ have described a technique to compensate for the fluctuating X-ray intensity from a high-power system by making use of a reference signal from a thin metal wire.Detection limits are a function of counting time excitation conditions spectrometer geometry matrix and atomic number. For X-ray excitation t’hey are typically less than 1 p.p.m. for the transition elements and about 0.1% for sodium. The use of pulsed X-ray beams is advocated as a means of increasing the counting-rate capacity of energy-dispersive ~ y s t e m s . ~ ~ ~ * ~ Basically the system operates by turning off the X-ray tube as soon as an event is detected by the spectrometer. Pulse pile-up events are virtually eliminated and counting rates are considerably increased. Polarised X-rays can be obtained by scattering from a suitable material such as boron carbide or from a synchrotron source.Pola,rised radiation is not scattered isotropically. Therefore if a detector is placed in the plane of polarisation and a t right-angles to the incident beam the scatter signal reaching the detector is considerably reduced. This results in a lowering of the spectral background and therefore an improvement in detection l i m i t ~ . ~ l - ~ ~ When using the X-ray tube and polariser it is necessary to use high power to compensate for energy loss on polarisation. R y ~ n ~ ~ found improvements in detection limits of up to 4.5 times for the elements from potassium to strontium in NBS orchard leaves when compared with direct excitation. Radioisotopes Most radioisotopes decay with the emission of X-rays y-rays or a-particles or a combina-tion of these and they are therefore suitable for use as excitation sources.Typical radio-isotope sources used in energy-dispersive X-ray fluorescence work include iron-55 cadmium-109 americium-241 cobalt-57 and gadoliniuni-153 for photon excitation and polonium-210 for a-particle excitation. It is possible to produce broad-band excitation sources utilising radioisotopes. p-Emitters such as promethium-147 or tritium when mixed with a suitable target material produce an X-ray continuum. However as the conversion efficiency is low high activity levels must be employed to obtain good counting rates. A description of the available photon-emitting radioisotopes and a discussion on the design and construction of new sources were given by Leonowich et al 47 A major constraint on the use of radioisotopes is that the emission should be as simple as possible.Where a large number of photon energies are emitted elastic and inelastic scatter from the sample will give rise to a complicated X-ray spectrum. Where high-energy y-rays are emitted high backgrounds can occur owing to detector Compton escape. Hence, to cover the spectral region of interest several simple emitters are preferable March 1979 EMISSION ANALYSIS. A REVIEW 181 The use of an or-emitter such as polonium-210 together with a windowless detector has been shown to allow the detection of oxygen and fluorine by energy-dispersive X-ray Curium-244 a photon and particle source has been used for the excitation of the light elements in rocks by Fran~grote.~g In a study of the limits of detection that can be attained using radioisotope excitation Spatz and Lieserso found that given the correct choice of isotope values were almost as good as those found using X-ray tube excitation.Electrons and Protons The use of electrons to excite characteristic sample X-rays is the basis of electron-microprobe analysis and it was in this field that energy-dispersive X-ray emission first found appli~ation.~~ The topic has been extensively dealt with by a number of worker^.^^-^^ The major disadvantage in the use of electron excitation is the high spectral background produced by electron deceleration. The use of protons to excite characteristic X-rays is attractive from a number of view-points. The ionisation cross-section shows a marked increase with decreasing atomic number thus increasing the sensitivity in this region.Protons (and other charged particles) do not produce the intense continuum found with electrons and the X-ray spectral back-ground is therefore small. However proton-induced X-ray emission or PIXE when applied to conventional “thick” samples can in fact exhibit a high spectral background. This is caused by the production of energetic photoelectrons in the sample that give rise to an X-ray continuum and for this reason PIXE has found most application in “thin” sample analysis. The need to obtain the use of a charged-particle accelerator is an obvious draw-back limiting the production of standard equipment. The principles of the technique have been described57258 and two review papers have a~peared599~O along with application studies in water analysis,57~61 biological material57262 and airborne parti~ulates.~3 A description of the design requirements of the experimental apparatus and a thorough evaluation of the qualitative and quantitative nature of the technique have been given by Johansson et aP4 Reuter and Lurlo65 investigated the application of proton excitation to “thick” samples and compared it with the use of electron excitation.Proton excitation produced sensi-tivities higher by factors of 2 or 3 when applied to low-alloy steels. It has been demon-strated66 that the sample depth probed by protons is smaller and more consistent across the elemental range than it is with X-ray excitation. Ahlberg and AdamsB7 compared the use of proton with X-ray excitation in the analysis of air particulate matter.PIXE was shown to be more sensitive for most elements. However the inhomogeneities found on air particu-late filters caused problems due to the small sampling area in the PIXE technique. Detectors Most X-ray detectors have some capacity to resolve photons in terms of their energy but none perform this task so well as the solid-state detectors. It was chiefly the advent of the solid-state detector with its associated pulse-processing electronics which gave rise to the technique of energy-dispersive X-ray fluorescence. Fig. 3 shows a diagrammatic repre-sentation of a solid-state Si(Li) detector. The detector is a diode consisting of a cylindrical piece of p-type silicon doped with lithium to increase the electrical resistivity. A Schottky barrier contact at the front of the detector produces the p - i - n diode and the application of a reverse bias voltage typically 1000 V depletes the diode of free charge carriers.The dimensions of the detector vary with the application but are typically 4-16 mm in diameter and 3-5mm thick. An X-ray photon entering the diode causes ionisation of the silicon and produces a number of “electron-hole pairs,” the number of which is proportional to the energy of the incident X-ray photon. The applied voltage sweeps the charge from the diode and it is collected at a charge sensitive pre-amplifier. The use of lithium-drifted silicon arises from two factors. The ionisation energy or band gap is small enough at 1.1 eV to produce sufficient charge carriers for good statistical definition of the pulse size but is high enough to prevent thermal-electron excitation being significant.The detector is normally contained in a vacuum behind a thin beryllium window. Both the detector and pre-amplifier are held at liquid-nitrogen temperature to reduce lithium migration and minimise electronic noise. A useful review of the design of solid-state detectors and thei 182 CAMPBELL ENERGY-DISPERSIVE X-RAY Analyst Vol. 104 application to X-ray spectrometry has been given by Heath,68 while Keith and L o ~ r n i s ~ ~ have critically examined the use of the detector with respect to energy calibration detector efficiency and detector phenomena such as escape and sum peaks. Gold - _-_-I Schottky barrier contact - n-type silicon Intrinsic or active region Gold -+, Fig.3. Silicon(lit1iiuin) detector. The efficiency of a solid-state detector in recording an event is a function of such para-meters as area and thickness of the active region dead layer and contact material and thickness of the entrance window. At low X-ray energies (low atomic number) much of the intensity is attenuated by the beryllium window. At high X-ray energies the efficiency is dependent on the detector thickness. Between 4 and 20 keV the detector efficiency exceeds 90% but falls off above and below this energy range.e9 The fundamental limitation to the resolving power of the Si(Li) detector lies in the statistical fluctuation in the number of electron-hole pairs produced by a given X-ray energy. It is customary to express the resolution of an energy-dispersive X-ray system as the full width at half maximum (FWHM) of a peak in the energy spectrum normally that of manganese KM.The FWHM contains contributions from noise sources other than the detector and in particular from the pre-amplifier. At low energies electronic noise is greater than that associated with statistical fluctuations. The detector resolution is normally expressed as FWHM (eV) = 2.3552/FE~ where E is the energy of the X-ray F the Fano factor and E the average energy to create an electron-hole pair. The Fano factor is related to the fractional amount of total energy absorbed resulting in the production of electron-hole pairs and is normally assumed to be about 0.15.68 The total system resolution including the contribution from electronic noise is expressed -as Typically the values of FWHM at 5.9 keV (manganese Ka) are in the range 150-180 eV.The Si(Li) solid-state detector finds application in other fields including X-ray diff rac-tion70~71 and astronomical spectroscopy.72 At photon energies above about 50 keV it is usual to employ another solid-state detector namely the lithium-drifted germanium diode. Electronics The electronic package used to process the 'output from the detector consists of the pre-amplifier amplifier pile-up rejector and multi-channel analyser. It is beyond the scope of this review to describe these components in. detail but they will be discussed briefly for completeness. A more detailed explanation of the design and operation of these components can be found elsewhere.l2P2 March 1979 EMISSION ANALYSIS.A REVIEW 183 Pre - amplifier The function of the pre-amplifier is to convert the charge pulse from the detector into a voltage signal while still retaining the proportionality to the incident X-ray photon energy and adding as little electronic noise as possible. The operation generally involves a form of current integration utilising a cooled field-eff ect transistor (FET) and electronic band-pass filters controlled by electronic shaping time constants. There are a number of pre-amplifier types each using a different technique to minimise the noise contribution. These include continuous optical feedba~k,‘~ drain feedback,7*975 modified-resistive feedback,76 pulsed-optical feedback77 and dynamic-charge re~toration.~~ Amplifier The function of the amplifiers175 is to convert and amplify the signals from the pre-amplifier in such a manner as to make them suitable for presentation to the multi-channel analyser.This function is achieved by “pulse shaping” techniques to attempt to obtain the optimum in energy resolution and counting-rate performance. A large shaping time constant produces optimum energy resolution at the expense of counting-rate performance and a small constant produces the reverse effect. T k important characteristics of a good amplifier are sophisticated pulse shaping good correlation between pulse height and energy and stability of gain and base line to changes in tempera-ture and counting rate. The use of long shaping time constants to obtain optimum energy resolution increases the probability of pulse overlap or “pulse pile-up.” This gives rise to two undesirable spectral features.Pulses are lost from the full energy peaks and a pulse pile-up continuum extends from just above the full energy peak for all peaks in the spectrum giving rise to complicated spectra. A further complication is that the overlap increases with counting rate and the pile-up loss is non-linear with counting rate. These problems can be almost completely overcome by the use of a “pulse pile-up rejector” system.79 This operates by inspecting the time interval between successive pulses from the pre-amplifier and denies entry to those signals where overlap is detected so that the system has a dead time associated with it. The pile-up rejector eliminates all but those sum peaks that are within the pulse width of the pre-amplifier and the intrinsic charge collection time of the detector.The use of a pulsed-excitation source will markedly reduce the problems of pulse overlap.39 Mu1 ti -channel Analy ser The MCA performs the function of sorting the pulses from the amplifier in terms of pulse amplitude and placing these in a “memory” composed of voltage windows or channels. The first stage of the MCA the analogue to digital converter (ADC) allows the incoming pulse to charge a capacitor which is then discharged at a constant current. The time of discharge is proportional to the pulse amplitude and this is used to gate on a constant frequency oscillator to produce a number of pulses. This “number” can then be related to a specific address or channel number.During this process the system will exhibit a dead time during which pulses cannot be accepted and this must be accounted for. There must be sufficient channels available in the MCA memory to span the energy range of interest with good coverage. Normally 1024 channels are used each with the capacity to store lo6 counts. This is important in order to allow good statistical precision when analysing trace amounts in the presence of major components. The information stored in the MCA memory is therefore a histogram of number of counts veisus channel number or after proper Cali-bration energy. The important features of the MCA12 are channel number varying linearly with energy uniform channel widths low dead time live-time clock adequately compensating for dead time and sufficient channels to cover the required energy range.Clearly these must be weighed against each other. Dead Time It is imperative to employ adequate dead-time compensation for quantitative analy~is.’~9~~ Normally the correction procedure employs the concept of “live time.” A clock measures the actual data accumulation time and ignores those intervals when the system is dead. This generally gives adequate compensation of up to about lo4 counts s-l and above this level pulse pile-up effects can cause problems 184 CAMPBELL ENERGY-1)ISPERSIVE X-RAY Analyst Vol. 104 From the foregoing discussion it is obvious that each parameter must be considered in the light of the requirements and limitations of the others.Typically counting rates of about lo3 counts s-1 can be achieved with amplifier shaping constants of between 4 and lops, giving rise to little peak shift or resolution degradation. Data Processing The raw data accumulated by the MCA must be processed to obtain useful information. For simple qualitative analysis the spectral peaks must be identified. More important to make use of a spectral feature for quantitative analysis the peak area spectral background, overlap effects and inter-element absorption and enhancement effects must be considered. I t has become customary to interface the MCA to a mini-computer or micro-processor. Data stored by the MCA are passed to the computer processed according to the user’s program and the final data printed out. An obvious progression from this situation is the use of the computer to control spectrometer parameters such as X-ray tube current and voltage filter or secondary target material and sample changer.The coupling of a computer to an energy-dispersive X-ray fluorescence system and the design of suitable software has been discussed by Keenan.8l Most commercial manufacturers offer systems with either partial or full computer control. Making use of a computer to help identify the elements giving rise to an energy-dispersive X-ray spectrum is a relatively simple matter. By storing data such as spectral line energies and relative intensities of lines from the same element the computer can be used to produce K L and M line markers with which spectral identification can be made. Positive identifi-cation is made by the presence of two X-ray lines in the correct intensity ratio for a particular element.In quantitative analysis there are certain features in the energy-dispersive spectra that complicate quantitative analysis when compared with wavelength-dispersive X-ray analysis. The poorer energy resolution increases the number of spectral overlaps and the peak to background ratio is normally poorer making the operation of background subtraction much more important. Fortunately the phenomena (of higher order reflections from the diffraction crystal can be ignored. The need to correct for inter-element effects both absorption and enhancement exists with all X-ray fluorescence systems. The problems of background subtraction and peak overlap effects are inter-related and a number of papers have dealt with these t ~ p i ~ s .~ ~ ~ ~ ~ - ~ ~ There are principally four methods used to calculate the background under a peak of interest shape fitting interpolation blank subtraction and the regressed constant method. Peak stripping utilising library spectra, overlap factors generated peak shapes or a combination of these is generally used to over-come overlap effects. RusP gives an excellent review of the methods applicable to back-ground subtraction and peak overlaps and lists a number of typical computer programs in BASIC that can be applied to the interpretation of energy-dispersive X-ray spectra. The high backgrounds encountered in electron-excited spectra make it imperative to use effective background calculation techniques.*8~S9 In a n attempt to remove the dependence on efficient background subtraction Nielsons7 has described a method of direct peak analysis based on a method proposed originally by C o ~ e l l .~ ~ No background subtraction is attempted and only a partial peak area is used. The precision is claimed to be acceptable for many routine applications. Statham919g2 has proposed a procedure for deconvolution and back-ground subtraction by suppressing the background using a digital filter and then proceeding to a conventional least-squares fit. The procedures developed for conventional X-ray analysis are equally applicable to energy-dispersive X-ray spectra for the correction of inter-element effects. There are basically two approaches the empirical and the theoretical.The empirical approach proposed by Lucas-Tooth and C O - W O ~ ~ ~ ~ S ~ ~ ~ ~ * - ~ ~ employs a number of standards to determine influence coefficients. The theoretical approach or “ fu ndamental-parameters” met hodgs-loo ut ilises known values of absorption coefficients and fluorescence yields as well as instrumental parameters. The specific application of these correction procedures after modification to energy-dispersive X-ray fluorescence analysis has been discussed by a number of workers.101,102 Neilsonlo3 has described a matrix correction program based on the measurement of the scatte March 1979 EMISSION ANALYSIS. A REVIEW 185 of the characteristic primary radiation. Shen and Russlo4 have produced a modified version of Stephenson’sgg fundamental-parameter model for application to energy-dispersive X-ray spectra.The results obtained indicate that when used carefully this approach can produce acceptable results for many applications without resort to conventional standardisation but that the method should not be used for work where high accuracy and precision are required. Spectral Features and Interpretation In the X-ray spectra obtained with the energy-dispersive X-ray fluorescence spectro-meter the major features are the characteristic X-ray lines from the elements present in the sample being analysed. Unfortunately these are not the only spectral features encountered. A number of other phenomena can give rise to features that must be recognised so as to avoid confusion with the characteristic X-ray lines or “full-energy” peaks as they are known.Sum Peaks Pulse pile-up effects were discussed above and it is inevitable with conventional systems that a certain amount of pulse pile-up will occur. Pulse pile-up increases with increasing counting rate and its effect is to produce sum peaks in the spectra which are essentially the sum of two X-rays being detected simultaneously. Normally sum peaks will appear only for the most intense features of a spectrum. For example in the X-ray spectrum of a steel sample it is probable that sum peaks will occur for the intense iron Ka and KP lines. These will appear in the spectrum at energies corresponding to 2Ka 2K/3 and Ka + KP. Elimination of sum peaks can be accomplished by either reducing the counting rate or using a filter between the sample and detector to remove a portion of the spectrum.Escape Peaks The escape peak phenomenon is caused by the escape of a silicon Ka X-ray from the intrinsic region of the detector. Generally an X-ray entering the detector transfers all of its energy to ionisation in this region. This results in the production of silicon Ka X-rays, which are normally contained within the active region producing further ionisation. How-ever where the probability of silicon Ka X-ray production is high less than 10 keV incoming X-ray energy a significant number of these photons can escape from the intrinsic region. The energy deposited in the detector will therefore be the energy of the incoming photon less the energy of the escaping X-ray 1.74 keV. This will give rise to a spectral feature at full energy minus 1.74 keV.The ratio of the escape peak to the full energy peak decreases with increasing atomic number and is normally about 1 200 for chlorine and 1 1000 for iron. This feature is characteristic of the detector independent of the sample and cannot be eliminated from the spectrum. Diffraction Peaks A sample of a highly ordered or crystalline nature can give rise to a peak in the spectrum by diffraction of the primary X-ray beam. A diffraction peak will appear when the incident X-ray energy and the spectrometer geometry are such that Bragg’s law is satisfied. The diffraction peak can be identified by altering the sample to detector or X-ray tube distance, thus changing the angle and causing the peak to appear at a different energy.This pheno-menon is normally seen where a continuum source of X-rays is used. The probability of satisfying the Bragg condition with mono-energetic excitation is small. The diffraction peak is generally broad and irregular in shape. Anomalous Silicon Gold and Argon Peaks The anomalous silicon and gold X-ray peaks are produced by the interaction of secondary X-rays with the detector dead layer and the gold Schottky barrier. X-rays interacting in these regions produce silicon and gold X-rays which have a probability of reaching the intrinsic region of the detector and being recorded as discrete pulses. Instances of anomalous silicon and gold peaks are fortunately rare; however where a high secondary X-ray flux is incident on the detector with energy between 2 and 3 keV the anomalous silicon peak ma 186 CAMPBELL ENERGY-DISPERSIVE X-RAY Analyst Vol.104 appear. In electron microprobe analysis the appearance of gold absorption edges in the spectral background can complicate background calculations.105 If an air path exists between the X-ray tube sample and detector then argon present in the air can be excited giving rise to secondary argon Ka X-rays at 2.96 keV. The occurrence of anomalous argon peaks is easily avoided by the use of either vacuum or helium paths in the spectrometer. Scatter Peaks There are two types of X-ray scatter coherent (Rayleigh) and incoherent (Compton). Peaks in the X-ray spectrum are generated by scatter of the characteristic lines in the primary-excitation radiation by the sample.Coherent scatter occurs without loss of energy and therefore produces a peak at the energy of the characteristic primary radiation and with a similar band width. Incoherent scatter occurs with energy loss and therefore, produces a spectral peak at a lower energy than that of the primary-excitation radiation and with a fairly broad band width. The ratio of coherent to incoherent scatter peak intensities is a direct function of atomic number. Spectral Background The spectral background continuum is determined by the nature of the excitation system. With direct or primary X-ray tube excitation the spectral background is fairly high and is due mainly to scattering of the primary bremsstrahlung. Other contributions to the back-ground are incomplete charge collection in the detector and sample-generated bremsstrahlung.With a near monochromatic secondary target or filtered X-ray excitation the background can be significantly reduced and is mainly associated with incomplete detector charge collection. With charged-particle excitation the background is due mainly to sample-generated bremsstrahlung. Cooperlos has discussed the features to be found in the energy-dispersive X-ray spectrum and proposed means of identification. Keith and loo mi^^^ have described the origin of some of the spectral features and proposed computer programs to correct the X-ray spectra for such phenomena as escape and sum peaks. The use of computer-generated K L and M line markers enables many of the above-men tioned spectral features to be distinguished easily from full-energy peaks.Comparison of Wavelength- and Energy-dispersive Systems A number of parameters must be considered when comparing energy-dispersive and wavelength-dispersive X-ray fluorescence equipment. These include resolution counting-rate capacity spectral interference peak to background ratio and the time cost and con-venience when carrying out particular analytical procedures. A comparison between energy-dispersive and wavelength-dispersive X-ray spectrometers for electron microscopes has been made by Malissa et aZ.107 but this is not completely relevant to the present discussion. Walingalos has discussed the advantages and limitations of the energy-dispersive X-ray technique in both fluorescence and diffraction in comparison with conventional or wave-length-dispersive systems.He concluded that energy-dispersive X-ray fluorescence analysis would find its major applications in qualitative and semi-quantitative analysis. However, this viewpoint seems dated when compared with more recent publication^^^ and the large number of papers appearing on quantitative analytical applications. Techniques for the analysis of samples of air pollutants have been. compared by Gilfrich et aZ.lo9 They investi-gated the use of various excitation sources such as X-rays a-particles radioisotopes and protons when used on energy-dispersive X-ray equipment and compared detection limits with those found using a wavelength-dispersive spectrometer. Detection limits were found to be comparable when the X-ray tube excited-energy dispersive system was used.Dewolfs et a1 .l10 compared a high-powered secondary-target energy-dispersive X-ray spectrometer with a commercial wavelength-dispersive X-ra.y spectrometer in terms of energy resolution, spectral interferences intensity and peak to background ratio sensitivity reproducibility and precision. The eventual conclusion was that the energy-dispersive technique was advantageous when applied to an analytical sj tuation requiring multi-element analysis with limited precision. In a study of energy- and wavelength-dispersive systems on the electro March 1979 EMISSION ANALYSIS. A REVIEW 187 micro-probe Dunham and Wilkinsonlll found that the techniques compared well in terms of accuracy and precision but that the energy-dispersive system produced poorer limits of detection while being faster and more convenient to use.The resolution of energy-dispersive X-ray systems is poor in comparison with wavelength-dispersive systems with the exception of the K lines of the heavy elements with atomic number greater than 45. This produces more spectral overlaps often requires the use of mathematical fitting procedures or forces the use of less sensitive lines. The counting-rate capacity is poorer for the solid-state Si(Li) detector and the associated electronics and this means that the sensitivity is poorer when only the number of counts per second measured is considered. This is at least partially balanced by the fact that the multi-element capacity of energy-dispersive systems allows much longer counting times to be tolerated.Peak to background ratios are generally poorer for energy-dispersive systems placing more emphasis on efficient background subtraction. Qualitative analysis is considerably simplified and less time consuming using the energy-dispersive system especially when computer-generated markers are used. It is difficult to emphasise sufficiently the benefits of accumulating data simultaneously. For quantitative analysis the energy-dispersive technique can produce good multi-element data with reasonable precision ; however the counting statistics usually produce poorer limits of detection than is the case with wavelength-dispersive spectrometers. Inter-element absorption and enhancement effects should be similar given similar spectro-meter geometries and excitation conditions.The time taken for a particular analysis favours wavelength-dispersive systems when only a few elements are being quantified but favours energy-dispersive systems for multi-element analysis. Energy-dispersive X-ray spectro-meters are somewhat cheaper than sequential wavelength-dispersive spectrometers and much cheaper than simultaneous wavelength-dispersive spectrometers. The energy-dispersive spectrometer is usually smaller and less bulky (especially true of low-power generator systems) and can be initially simpler to use for the non-X-ray spectroscopist. The newer energy-dispersive X-ray fluorescence technique can be seen as a competitor to the wavelength-dispersive spectrometer or as a complement to it depending on the particular analytical requirements of a given situation.For many applications the ultimate in precision is not required and the energy-dispersive system can provide an adequate versatile and relatively inexpensive solution. When fast qualitative analysis is important then the energy-dispersive system must be carefully considered. When very high precision work is required the simultaneous or sequential wavelength-dispersive spectrometer must still be the instrument of choice. When multi-element analysis is required on a limited budget then the energy-dispersive system may appear advantageous when compared with a simultaneous wavelength-dispersive spectrometer. In some analytical situations the choice will be complicated by the need to satisfy a number of conflicting criteria and here the energy-dispersive system can be seen as a complement to the wavelength-dispersive system.Applications The last few years have seen a large increase in the number of publications dealing with the application of the energy-dispersive X-ray emission technique to practical analytical problems. For convenience these will be detailed here in terms of the various fields of interest with which the papers have dealt. All of the applications have the solid-state Si(Li) detector in common but excitation of the characteristic sample X-rays may be by a variety of means as already detailed. Atmospheric Particulates Energy-dispersive X-ray fluorescence analysis has attracted considerable interest for the analysis of airborne particulate matter.l12 There are a number of reasons for this popularity, including its potential multi-element capability non-destructive nature and the advantages gained when working with thin samples.A number of papers have dealt with the problems to be overcome before acceptable multi-element analyses can be 0btained,ll~-~~8 while others describe the results found in various locations and their implication^.^^^-^^^ Air particulates are normally collected by passing a known volume of air through a filter. This results in “thin-specimen,’ samples and these possess a number of advantages over conventional “infinitely-thick” samples. Inter-element effects are eliminated or at worst 188 CAMPBELL ENERGY-:DISPERSIVE X-RAY Analyst Vol. 104 reduced considerably peak to background ratio is normally increased particle size effects are simplified and linear calibrations are generally found over wide ranges.The filter must be as thin as possible to reduce absorption and enhancement effects and contain no contami-nants heavier than sodium. Cellulose fibre or membrane filters are popular although weighing can be a problem owing to hygroscopicity. Using radioisotope excitation Rhodes113 and Rhodes et al.l14 have investigated particle-size distribution sampling and standardisation with respect to thin-filter samples. Adams and co-workers,116-118 using secondary-target X-ray tube excitation have considered a number of potential problems including sample homogeneity and position background subtraction and spectral overlap effects. Generator instability was compensated for by placing a thin zirconium wire below the sample surface and normalising all spectral features to the zirconium Ka X-radiation.Adams and Van Grieken115 have also demonstrated that absorption effects cannot be ignored for elements of low atomic number and correction procedures are necessary for both the air particulate matrix and the filter material. The procedure normally adopted to correct for filter attenuation involves the measurement of front to back intensity ratios of secondary X-rays from the material deposited on the filters.122 Ahlberg and ad am^^^ demonstrated the extra sensitivity that can be obtained using proton excitation as compared with X-ray excitation, especially for the lighter elements. The use of 5-MeV protons to excite secondary X-rays from filter samples has been described by Pilotte et aZ.121 A number of elements including sulphur chlorine and potassium were determined with sufficient sensitivity to allow observa-tion of time variations in elemental concentrations.The variation of elemental composition across the particle size range is of interest with respect to the determination of the source of the particulate matter. Jaklevic and co-worker~~~3J~~ have described a “dichotomous” sampler which is designed to separate airborne particulates above and below 2.4 pm particle diameter. The two particle-size ranges were collected separately and subjected to energy-dispersive X-ray fluorescence analysis using photon excitation. The use of energy-dispersive X-ray analysis on a scanning electron microscope for the characterisation of atmospheric particulate matter was described by Butler et al.lZ5 Specific particles can be chosen for investigation thus allowing a correlation to be made between particle size and elemental composition.The limitation will of course be the lower sensitivity that is available. Calibration standards prepared by vacuum deposition on thin films are commercially available from Micro Matter Co. Seattle and are most useful as an aid to the quantification of energy-dispersive X-ray fluorescence data of filter samples. Waters Information can be sought on one or both of these phases or a total figure may be sufficient. For particulate matter filtration produces thin samples similar to those obtained with atmospheric particu-lates and much of the above discussion applies equally.In most instances dissolved solids are present at concentrations below those directly observable using the energy-dispersive X-ray technique and some form of pre-concent ration is usually employed. In a study of the analysis of sediments and particulate matter in sea water Vanderstappen and Van Grieken126 concluded that a 0.4-,urn Nuclepore polycarbonate filter was optimum for filtration. Samples from the North Sea arid Mediterranean were analysed for a number of elements at the parts per billion (lo9) level with acceptable accuracy and precision. The use of trace-element precipitation by complexing with the chelating agent ammonium tet ramet h ylenedithiocarbamate (ammonium-1 - p yrrolidine dit hiocarbamate APDC) followed by filtration is recommended for the determination of dissolved solids in natural waters.1279128 APDC forms insoluble complexes with about 30 transition metals but does not complex the alkali metals or the alkaline earth metals.Where only the dissolved solids are of interest, particulate matter must first be removed by filtration or centrifugation. Elder et al.128 reported the analysis of a number of elements at the parts per billion level by APDC precipita-tion and filtration through a membrane filter while Pradzynski et ~ z . 1 ~ ~ determined uranium, molybdenum and thorium at the 1 p.p.b. level. Electro-deposition has been advocated as a pre-concentration technique to determine reducible metals in aqueous media by wavelength-dispersive X-ray fluore~cence.~~~ Boslett et a2.130 have proposed a similar technique for use with energy-dispersive apparatus.Zinc copper and nickel were determined at the 2-100 Waters can contain both dissolved solids and particulate matter March 1979 EMISSION ANALYSIS. A REVIEW 189 p.p.b. level using potentiostatic electro-deposition on to a graphite rod producing a thin metallic film. Special cylindrical monochromators were necessary to reduce scatter from the graphite rod and fairly long deposition times were required. Carlton and Russl3l described the use of ion-exchange resin loaded filter-papers to pre-concentrate trace elements. An automated sample collection and preparation modple was presented and sensitivities obtained were in the parts per billion range. Van Grieken et ~ 1 . l ~ ~ have shown that Chelex-100 ion-exchange membranes can be used to pre-concentrate trace elements prior to energy dispersive X-ray fluorescence analysis provided that the water samples contain only modest concentrations of alkali and alkaline earth metals.The use of chelating ion exchangers based on cellulose was described by Burba et ~ 1 . l ~ ~ and illustrated by the determination of uranium in natural waters down to 0.3 p.p.b. A simple sample-spotting procedure has been proposed by Smits and Van Grieken.134 About 1.5 ml of the sample is placed on a cellulose filter and held in position with a wax ring. The water is evaporated using an unheated air stream from below. The choice of the pre-concentration technique depends on the elements of interest the analysis time restrictions and sensitivity requirements.The sample-spotting procedure is probably the simplest and most widely applicable but could give problems at high concentra-tions where crystallisation can occur. Birks and G i l f r i ~ h l ~ ~ have evaluated seven typical energy-dispersive X-ray fluorescence instruments with respect to the determination of trace elements in polluted waters. All were considered capable of measuring elemental concentra-tions at levels appropriate to the problem. Claimed precisions are 15-20% at the 50-100 p.p.b. level. Clinical and Biochemical In many clinical and biochemical studies it is necessary to determine a number of elements in small samples of blood urine tissue etc. When only one element is determined at a time problems can be caused in terms of sample size and analysis time.The multi-element capacity of energy-dispersive X-ray fluorescence is therefore attractive in this field. How-ever its relatively poor sensitivity compared with say atomic-absorption spectrometry, requires that more effort be devoted to sample preparation. The application of the energy-dispersive X-ray technique in this area has been discussed by a number of w o r k e r ~ . ~ ~ ~ - - 1 ~ ~ The determination of trace elements in whole blood plasma and serum has attracted considerable interest. Bearse et ~ 1 . l ~ ~ using a plasma-ashing preparation technique followed by proton-induced X-ray emission determined iron copper zinc selenium and rubidium in 0.1-ml samples of blood. Detection limits for elements with atomic numbers 2545 were shown to be between 0.1 and 1 p.p.m.which should be useful for many applications. Holynska and Marko~iczl~~ also determined selenium in whole blood (and tissue) but used wet ashing and coprecipitation followed by low-powered X-ray tube excitation. Levels as low as 100 p.p.b. were determined. The use of proton excitation allowed the determination of selenium in blood serum at the 10 p.p.b. le~el.l4~ Freeze-drying and pelletising have been used145 to prepare samples of whole blood and plasma for the determination of copper zinc, bromine and rubidium by energy-dispersive X-ray fluorescence using secondary target X-ray excitation with detection limits in the 100-400 p.p.b. range. Knoth et aZ.146 suggested the use of total X-ray reflectionf4’ on a sample support as a means of reducing the spectral background in the determination of copper and iron in blood serum.A 10-pl sample was dried on a support of optically flat silica glass and the incident X-ray beam adjusted to strike the support a t a very low angle such that it is “totally reflected.” Agarwal et ~ 1 . l ~ ~ applied energy-dispersive X-ray fluorescence with secondary target X-ray excitation to the analysis of copper zinc and lead in urine. They suggested the use of an ion-exchange chelating resin as a pre-concentration step and also used yttrium as an internal standard. The method is limited to those elements chelated by the resin. Inductively-coupled plasma-emission spectroscopy has a similar multi-element capacity to energy-dispersive X-ray fluorescence. An evaluation of both techniques for biological work has been carried out by Irons et aZ.149 with reference to sensitivity precision and accuracy.They conclude that the choice of method depends on the sample type plasma emission having advantages for fluids and X-ray fluorescence for solids. The need to apply inter-element corrections in X-ray work was emphasised hence increasing the computer size requirement relative to plasma-emission spectroscopy 190 CAMPBELL ENERGY-DISPERSIVE X-RAY Analyst Vol. 104 Rocks Ores and Cement Cooper and Sch10fkel~~ have described the application of energy-dispersive X-ray fluorescence with direct X-ray tube excitation to rock and ore analysis. Samples of this type normally receive minimal preparation prior to analysis. They are generally ground and pressed into disc form.StanddYdisation is achieved by using similar standard reference materials. Corrections for inter-element effects and peak overlaps must be applied to obtain acceptable results. The general utility of energy-dispersive X-ray fluorescence analysis to geochemical specimens has been demonstrated by Giauque et al.15l Twenty-six trace and two major elements were determined in reference materials use being made of the relationship between the intensity of the incoherent scattered radiation specimen mass absorption coefficient and spectral background intensity to correct for inter-element effects. Other workers have undertaken the analysis of platinum152 and copper153 in ores. Hebert and Bowman154 described a special spectrometer for the analysis of the light elements in rocks which possessed a sensitivity of 5 p.p.m.for sodium. In the analysis of cement-type materials both major and minor elements are determined, although the most important are aluminium silicon calcium and iron. Energy-dispersive X-ray fluorescence with primary X-ray tube excitation has been used to determine these four and eight other elements in Portland cement by Cooper and ~o-workers.15~J56 Once again inter-element correction procedures were shown to improve the quality of the analytical results. Carr-Brion et a1.l5’ have described an on-stream energy-dispersive X-ray fluorescence analyser for the determination of the four main elements of interest in cement raw meal. Results appear to be sufficient for raw meal feed control requirements despite the need to analyse the raw meal “unground.” Metals and Alloys The use of energy-dispersive X-ray emission as an alloy-sorting technique has been des~ribed.l58J~~ Alloys can be quickly characterised by their spectral features in a non-destructive fashion using a portable analyser with minimum or no sample preparation.Janssens et aZ.fsO have proposed a high-precision method for the determination of manganese in ferromanganese. Radioisotope excitation (using cadmium-109) with careful control of the instrumental conditions and standardisation produced a relative standard deviation of 0.274,. The application of the energy-dispersive X-ray technique to multi-element analysis of nickel-alloys has been investigated by Verbeke et al.,lal who found that acceptable results coald be obtained only after applying inter-element corrections using a fundamental-parameters approach.The analysis of thin nickel- gold films has been examined by Franken,162 who found that energy-dispersive X-ray fluorescence produced better data than conventional X-ray diffraction because it was less sensitive to structural effects. Coal and Petroleum The multi-element analysis of coal coke and fly-ash materials by energy-dispersive X-ray fluorescence has been shown to provide acceptable data after application of an inter-element correction routine using multiple standards.163 Lloyd and Francis164 compared the results obtained for the determination of sulphur in coal by conventional ASTM procedures with energy-dispersive X-ray fluorescence results.They concluded that the instrumental tech-nique with proper matrix corrections was close to meeting the ASTM standards of accuracy and precision but was far superior in terms of speed of analysis and convenience. A different approach to the analysis of coal has been taken by De Kalb and Fassel,ls5 who minimised the inter-element effects by converting the powdered coal into a thin film using the Chungl66 technique and obtained good results without inter-element correction procedures. The application of PIXE to coal analysis has been described by Cronch et aZ.167 Yousif and Al-ShahristanP8 applied energy-dispersive X-ray emission with radio-isotope excitation (using iron-55) to the determination of sulphur and vanadium in crude oils. At the concentration levels found in crude oil it was necessary to correct the vanadium data for the presence of sulphur.Vanadium was determined down to 2 p.p.m. and sulphur to 0.03%. Tellerlsg has described an immersion probe designed to determine sulphur in fuel oils or lead in refinery products by the simultaneous measurement of scattered and transmitted low-energy X-rays March 1979 EMISSION ANALYSIS. A REVIEW 19L On-stream Analysis Energy-dispersive X-ray fluorescence possesses a number of attractive features for on-stream analysis when compared with wavelength-dispersive equipment. The use of high take-off angles and broad-beam geometry helps lessen the effects of surface roughness and a well designed system is relatively insensitive to temperature variations. Very short path lengths can be achieved thus increasing light-element sensitivity.The ability to use a variety of excitation systems increases the over-all system flexibility. The advantages and limitations of energy-dispersive X-ray emission and diffraction equipment in process control situations have been discussed by Carr-Brion170 and Kawatra and Da1t0n.l~~ On-stream systems have been described for the process control of cement raw-meal powder157 and finished ~ i n t e r . l ~ ~ Others Applications of the energy-dispersive technique have been both numerous and varied. Cobalt and molybdenum have been determined in hydrodesulphurisation c a t a l y ~ t s l ~ ~ 9174 noble metals in automotive exhaust catalysts,175 silver in photographic materials,176 thorium in optical uranium in aqueous media,17* technetium in nuclear fuel processing waste,l79 copper and silver in Roman silver coins180 and in paper additives.181 Other uses include the analysis of sedimentary pollutantsls2 and agricultural wastes,ls3 ion-exchange s t ~ d i e s l ~ ~ ~ ~ ~ and process control.ls6 Hansonls7 proposed the use of energy-dispersive X-ray fluorescence as a technique to determine the authenticity of art objects.The use of the technique to determine particle sizes has been described by Tominaga et aZ.ls8 while Kawamoto et aZ.lss have designed a milli-analyser to investigate small particles of the order of 1 mm. Future Developments It is unlikely that any dramatic improvement in detector resolution will be made in the immediate future. Small resolution improvements although welcome will not significantly affect the over-all system performance.Improvements in the quality of the detector can, however produce a lowering of the spectral background which would be of considerable benefit. Improvements to the stability of the X-ray generator especially for high-powered systems, would be welcome. The use of pulsed X-ray sources now becoming commercially available, should help to improve the counting-rate capability of the system but it is not clear whether the pulse-processing electronics are sufficiently sophisticated to allow full benefit to be gained. With many commercial systems the determination of the heavy elements can present problems. The typical X-ray generator is designed to produce a maximum of 50 kV and this does not efficiently excite the K lines of elements with 2 m 50-60,91 while the L lines are subject to a variety of possible spectral overlaps.A move toward generators capable of producing 60-80 kV would be welcome. The most significant advances in the near future will lie in the area of computer software. The power of the dedicated mini-computer especially when coupled to interactive disc drives has still to be fully utilised. The identification of spectral features is primarily the work of the instrument operator though generally aided by the computer with such devices as K L and M markers. The extraction of quantitative data from energy-dispersive X-ray spectra using the fundamental parameters approach will improve to a high degree of sophistication.190 The techniques of deconvolution and inter-element correction though now well documented require further development.Finally although a large number of publications have appeared dealing with applications of the technique many more studies are necessary to define the best fields of application, especially with respect to conventional infinitely thick samples. There appears to be no successor to the solid-state Si(Li) detector on the horizon. This task will become more and more computer dependent. References 1. 2. 3. Carr-Brion K. G. and Payne K. W. Analyst 1970 95 977. Elad E. Nucl. Instrum. Meth. 1965 37 327. Elad E. and Nakamura M. Nucl. Instrum. Meth. 1966 41 161 192 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 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. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. CAMPBELL ENERGY-DISPERSIVE X-RAY Analyst Vol. 104 Bowman H. R. Hyde E. K. Thompson S. G. and Jared R. C. Science N . Y . 1966 151 562. Elad E. and Nakamura M. I.E.E.E. Trans. Nucl. Sci. 1967. 14 523. “Energy Dispersive X-Ray Analysis X-Ray and Electron Probe Analysis,” Special Publication Heinrich K. F. J. Barrett C. S. Newkirk J. B. and Ruud C. O. Editors Advances in X-Ray Heath R. L. Adv. X-Ray Analysis 1972 15 1. Giauque R. D. and Jaklevic J. M. Adv. X-Ruy Analysis 1972 15 164. Dyer G. R. Gedcke D A. and Harris T. R. Adv. X-Ray Analysis 1972 15 228.Jaklevic J. M. Giauque R. D. Malone D. F. and Searles W. L. Adv. X-Ray Analysis 1972 15, Gedcke D A. X-Ray Spectrom. 1972 1 129. “Appendices on Provisional Nomenclature Symbols Terminology and Conventions-Number 54,” Bertin E. P. “Principles and Practice of X-Ray Spectrometric Analysis,” Plenum Press New Liebhafsky H. A. Pfeiffer H. G. Winslow E. H. and Zemany P. D. “X-Rays Electrons and Jenkins R. and De Vries J. L. “Practical X-Ray Spectrometry,” Macmillan London 1975. Birks I,. S. and Gilfrich J. V. Analyt. Chem 1976 48 273R. De Vries J . L. “X-Ray Fluorescence Spectrometry-Review of Literature,” Seventh Edition, Jenkins R. “An Introduction to X-Ray Spectrometry,” Heyden London 1974. Bertin E. P. “Introduction to X-Ray Spectrometric Analysis,” Plenum New York 1978.Woldseth R. “X-Ray Energy Spectrometry,” Kevex Corp. Burlinghame Calif. 1973. Schunemann D. Miner. Slov. 1973 5 569. Friant A. Onde Elect. 1976 56 69. Russ J. C. X-Ray Spectrom. 1972 1 119. Friant A. Gras R. Lorin A. and Saliou C. Bull. Instrumn Nucl. 1971 42 34. Porter D. E. and Woldseth R. Analyt. Chem. 1973 45 604A. Lister D. B. “Application of Energy-Dispersive X-Ray Fluorescence,” AID 75437 Instrument Gedcke D. A. and Elad E. “Proceedings of the 6th International Conference on X-Ray Optics Goulding F. S . and Jaklevic J . M. A . Rev. N’ucl. Sci. 1973 23 45. Russ J. C. Shen R. B. and Jenkins R. “Energy Dispersive X-Ray Analysis of Materials Principles Cothern C. R. Manuel H. L. and Millette R. J. X-Ray Spectrom. 1974 3 53. Jaklevic J. M. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich.1976,” Middleman L. M. and Geller J. D. “Proceedings of the 9th Annual Scanning Electron Microscope Reldy J. J. Hutson R. L. Daniel H. and Springer K. Analyt. Chem. 1978 50 40. Gedcke D. A. Elad E. and Denee P. B. X-Ray Spectrom. 1977 6 21. Artz B. E. and Short M. A. “Proceedings of :ERDA X- and Gamma-Ray Symposium Ann Arbor, Anselmo V. C. Report Lawrence Livermore Laboratory UCID 17317 1976. Van Espen P. J. and Adams F. C. Analyt. Chem. 1976 48 1823. Jakievic J. M. Goulding F. S. and Landis D. A. I.E.E.E. Trans. Nucl. Sci. 1972 NS19 392. Jaklevic J. M. Landis D. A. and Goulding F. S. Report Lawrence Berkeley Laboratory LBL Howell R. H. Pickles W. L. and Cate J. L. Adv. X-Ray Analysis 1974 18 265. Kaufman L.and Camp D. C. Adv. X-Ray Analysis 1974 18 247. Dzubay T. G. Jarret. B. V. and Jaklevic J. M. Nucl. Instrum. Meth. 1974 115 297. Howell R. H. and Pickles W. L. Nucl. Instmm. Meth. 1974 120 187. Ryon R. W. Report Lawrence Livermore Laboratory UCRL 78063 1976. Aiginger H. Wobrauschek P. and Brauner C. Nucl. Instrum. Meth. 1974 120 541. Leonowich J. Pandian S. and Preiss I. L. J . Radioanalyt. Chem. 1977 40 175. Robertson R. Nucl. Instrum. Meth. 1977 142 121. Franzgrote E. Adv. X-Ray Analysis 1972 15 388. Spatz R. and Lieser K. H. 2. Analyt. Chem. 1977 288 267. Fitzgerald R. Keil K. and Heinrich K. F. J Science N . Y . 1968 159 528. Gedcke D. A. in Holt D. B. Muir M. D. Grant P. R. and Boswarva I. M. Editors “Quantita-Birks L. S. “Electron Microprobe Analysis,” Interscience New York 1963.Waldl E. Wolfermann H. Rusovic N. and Warlimont H. Analyt. Chem. 1975 47 1017. Desborough G. A. and Heidel R. H. APPl. Spectrosc. 1973 27 456. Reed S. J. B. and Ware N. G. X-Ray Spectvom. 1973 2 69. Walter R. L. Willis R. D. Gutknecht W. F. and Joyce J. M. Analyt. Chem. 1974 46 843. Johansson G. Akselsson R. Bohgard M. Carlsson L. E. Hansson H. C. Lannefors H. and Johansson S. A. E. and Johansson T. B. Nwcl. Instrum. Meth. 1976 137 473. STP 485 American Society for Testing and Materials Philadelphia Pa. 1971. Analysis 1972 Volume 15. 266. I UPAC Information Bulletin 1976. York 1970. Analytical Chemistry,” Wiley New York 19’72. Philips Eindhoven 1976. Society of America 1975. and Microanalysis Osaka Japan 1971,” University of Tokyo Press Toyko 1972.and Experiments,” Edax International Prairie View Ill. 1978. NTIS Springfield Va. 1977 p. 1. Symposium Chicago 1976,” Scanning Electron Microsc. 1976 9 Part 1 171. Mich. 1976,” NTIS Springfield Va. 1977 p. 7. 4248 1978. tive Scanning Electron Microscopy,” Academic Press New York and London 1974 Ch. 12. Malmqvist K. Proc. Analyt. Div. Chem. SOL 1978 15 24 March 1979 EMISSION ANALYSIS. A REVIEW 193 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115.116. 117. 118. 119. Folkmann F. J . Phys. E 1975 8 429. Lochmiiller C. H. Galbraith J. W. and Walter R. L. Analyt. Chem. 1974 46 440. Mangelson N. F. Hill M. W. Nielson K. K. and Ryder J . F. Nucl. Instrum. Meth. 1977 142, Malmqvist K. Akselsson R. A. and Johansson G. Proc. Analyt. Div. Chem. Soc. 1978 15 13. Johansson T. B. Van Grieken R. E. Nelson J . W. and Winchester J . W. Analyt. Chem. 1975, Reuter W. and Lurlo A. Analyt. Chem. 1977 49 1737. hhlberg M. S. Nucl. Instrum. Meth. 1977 146 465. Ahlberg M. S. and Adams F. C. X-Ray Spectrom. 1978 7 73. Heath R. L. Adv. X-Ray Analysis 1972 15 1. Keith H. D. and Loomis T. C. X-Ray Spectrom. 1976 5 93. Martin G. W. and Klein A. S. Adv. X-Pay Analysis 1972 15 254. Laine E. Lahteenmaki I. and Kantola M. X-Ray Spectrom.1972 1 93. Holt S. S. Space Sci. Instrumn 1976 2 205. Elad E. “Energy Dispersive X-Ray Analysis X-Ray and Electron Probe Analysis,” Special Publication STP 485 American Society for Testing and Materials Philadelphia Pa. 1971 p. 57. Goulding F. S. Walton J. T. and Pehl R. H. Report Lawrence Berkeley Laboratory UCRL 19377 1977. Walter F. J. “Energy Dispersive X-Ray Analysis X-Ray and Electron Probe Analysis,” Special Publication STP 485 American Society for Testing and Materials Philadelphia Pa. 1971 p. 82. Elad E. I.E.E.E. Trans. Nucl. Sci. 1972 19 403. Landis D. A. Goulding I;. S. Pehl R. H. and Walton J. T. I.E.E.E. Trans. Nucl. Sci. 1971, Gedcke D. A. Elad E. and Dyer G. R. “Proceedings of the 6th National Conference on Electron Reed S. J. B.J . Phys. E 1972 5 997. Reed S. J. B. J . Phys. E 1972 5 994. Keenan J. A, Int. Lab. 1975 MaylJune 49. Fiori C. E. Myklebust R. L. Heinrich K. F. J. and Yabowitz H. Analyt. Chem. 1976 48 172. Connelly A. L. and Black W. W. Nucl. Instrum. Meth. 1970 82 141. Smith D. G. W. Gold C. M. and Tomlinson D. A. X-Ray Spectrom. 1975 4 149. Geiss R. H. and Hwang T. C. X-Ray Spectrom. 1975 4 196. Russ J. C. X-Ray Spectrom. 1977 6 37. Nielson K. K. X-Ray Spectrom. 1978 7 15. Comins N. R. and Thirlwall. J. T. X-Ray Spectrom. 1978 7 92. Bauer R. and Rick R. X-Ray Spectrom. 1978 7 63. Covell D. F. Analyt. Chem. 1959 31 1785. Statham P. J. Analyt. Chem. 1977 49 2149. Statham P. J. X-Ray Spectrom. 1978 7 132. Lucas-Tooth H. J. and Price B. J. Metallurgia 1961 64 149. Lucas-Tooth H.J. and Pyne C. Adv. X-Ray Analysis 1964 7 523. Traill K. J. and Lachance G. R. Geol. Surv. Can. Pap. 1965 No. 64. Traill R. J. and Lachance G. R. Can. Spectrosc. 1966 11 (3) 63. Rasberry S. D. and Heinrich K. F. J. Analyt. Chem. 1974 46 81. Criss J. W. and Birks L. S. Analyt. Chem. 1968 40 1080. Stephenson A. Analyt. Chem. 1971 43 1761. de Jongh W. K. X-Ray Spectrom. 1973 2 151, Russ J. C. Sandborg A. O. Barnhart M. W. Soderquist C. E. Lichtinger R. W. and Walsh, C. J. Adv. X-Ray Analysis 1973 16 284. Ong P. S. Cheng E. L. and Sroka G. Adv. X-Ray Analysis 1974 17 269. Nielson K. K. Analyt. Chem. 1977 49 641. Shen R. B. and RUSS J . C. X-Ray Spectrom. 1977 6 56. Brombach J. D. X-Ray Spectrom. 1978 7 81. Cooper J . A. Am. Lab. 1976 Nov. 35. Malissa H.Grasserbauer M. and Hoke E. Microchim. Acta 1974 5 465. Walinga J . .‘:Proceedings of the 8th Philips X-Ray Analytical Conference Birmingham England, Gilfrich J. V. Burkhalter P. G. and Birks L. S. Analyt. Chem. 1973 45 2002. Dewolfs K. De Neve R. and Adams F. Analytica Chim. Acta 1976 75 47. Dunham A. C. and Wilkinson F. C. F. X-Ray Spectrom. 1978 7 50. Dzubay T. G. Editor “X-Ray Fluorescence Analysis of Environmental Samples,” Ann Arbor Rhodes J. R. Am. Lab. 1973 July 57. Khodes. J . R. Pradzynski A. H. and Sieberg R. D. Air Qual. Instrumn 1974 2 1. Adams F. C. and Van Grieken R. E. Analyt. Chem. 1975 47 1767. Adams F. C. and \’an Espen P. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Van Espen P. and Adams F. C. Analytica Chim. Acta 1974 75 61. Van Espen Y.Nullens H. and Adams F. C. 2. Analyt. Chem. 1977 285 215. Rhodes J. R. Pradzynski A. H. Hunter C. B. Payne J. S. and Lindgren J. L. Envir. Sci. 133. 47 855. 18 115. Probe Analysis,” Electron Probe Analysis Society of America New York 1971 p. 5.4. 1972,” Philips Eindhoven 1973. Science Publishers Ann Arbor Mich. 1977. Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 182. Technol. 1972 6 922 194 CAMPBELL ENERGU-DISPERSIVE X-RAY Analyst VoE. 104 120. 121. 122. 123. 124. Hammerle R. H. and Pierson W. R. Envir. Sci. Technol. 1975 9 1058. Pilotte J . O. Nelson J. W. and Winchester, J. W. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 161. Davis D. W. Reynolds R. L. Tsou G. C. and Zafonte L.Analyt. Chem. 1977 49 1990. Jaklevic J. M. Loo B. W. and Goulding F. S. Report Lawrence Berkeley Laboratory LBL 4834, 1976. Loo B. W. Jaklevic J . M. and Goulding F. S. in Liu B. Y. H. Editor “Fine Particles Aerosol Generation Measurement Sampling and Analysis,” Academic Press New York London 1976, p. 311. Butler J. D. MacMurdo S. D. and Stewart C. J. Int. J . Environ. Stud. 1976 9 93. Vanderstappen M. and Van Grieken R. 2. d4nalyt. Chem. 1976 282 25. Pradzynski A. H. Henry R. E. and Draper E. L. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 175. Elder J. F. Perry S. K. and Brady F. P. Envir. Sci. Technol. 1975 9 1039. Vassos B. H. Hirsch R. F. and Letterman W. Analyt. Chem. 1973 45 792. Boslett J. A.Towns R. L. R. Megargle R. G. Pearson K. H. and Furnas T. C. Jr. Analyt. Carlton D. T. and Russ J. C. X-Ray Spectrom. 1976 5 172. Van Grieken R. E. Bresseleers C. M. and Vanderborght B. M. Analyt. Chem. 1977 49 1326. Burba P. Gleitsmann B. and Lieser K. H. 2. Analyt. Chem. 1978 289 28. Smits J. and Van Grieken R. Analytica Chtim Acta 1977 88 97. Birks L. S. and Gilfrich J. V. Appl. Spectrosc. 1978 32 204. Cox H. L. Jr. and Ong P. S. Med. Phys. 11977 4 99. Kaufman L. Nelson J. Price D. Shames D. and Wilson C. J. I.E.E.E. Trans. Nucl. SCi., 1973 20 402. Mangelson N. F. Allison G. E. Christensen J. J. Eatough D. J. Hill M. W. Izatt R. M. and Nielson K. K. “Proceedings of the 2nd International Symposium on Trace Element Metabolism in Animals 1973,” University Park Press Baltimore Md.1974 p. 439. Reuter F. W. and Reynolds W. L. Adv. Exp. Med. Biol. 1974 48 621. Kaufman L. Deconinck F. Camp D. C. Voegele A. L. Friesen R. D. and Nelson J. A., “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich. 1976,” NTIS, Springfield Va. 1977 p. 118. Zombola R. R. Kitos P. A. and Bearse R. C. Analyt. Chem. 1977 49 2203. Bearse R. C. Close D. A. Malanify J. J. and Umbarger C. J. Analyt. Chem. 1974 46 499. Holynska B. and Markowicz A. Radiochem. Radioanalyt. Lett. 1977 31 165. Berti M. Buso G. Colautti P. Moschini G. Stievano B. M. and Tregnaghi C. Analyt. Chem., Stump I. G. Carruthers J. D’Auria J . M. Applegarth D. A. and Davidson A. G. F. Clin. Knoth J. Schwenke H. Marten R. and Glauer J. J . Clin. Chem. Clin. Biochem. 1977 15 537.Wobrauschek P. and Aiginger H. Analyt. Chem. 1975 47 852. Agarwal M. Bennett R. B. Stump I. G. and D’Auria J. M. Analyt. Chem. 1975 47 924. Irons R. D. Schenk E. A. and Giauque R. D. Clin. Chem. 1976 22 2018. Cooper J. A. and Schlofke D. B. Skillings .Min. Rev. 1976 65 1. Giauque R. D. Garrett R. B. and Goda L. Y. Analyt. Chem. 1977 49 62. Coombes R. J. Chow A. and Flint R. W. Analytica Chim. Acta 1977 91 273. Shenberg C. Ben Haim A. and Amiel S. Analyt. Chem. 1973 45 1804. Hebert A. J. and Bowman H. R. Report Lawrence Berkeley Laboratory LBL 3268 1975. Cooper J. A. Wheeler B. D. Bartell D. M. and Gedcke D. A. Adv. X-Ray Analysis 1976,19, Cooper J. A. Wheeler B. D. and Bartell D. M. Cem. Technol. 1976 March/April 68. Carr-Brion K. G. Kipping P. J. New R. Nutter J.C. and Thomlinson F. I. Wld Cem. Technol., Pradzynski A. H. and Draper E. L. Jr. “Proceedings of the 9th Symposium N.D.E. San Antonio, Harrison P. E. and Kenna B. T. Talanta 1972 19 810. Janssens R. Maenhaut W. and Hoste J. Analytica Chim. Acta 1975 76 37. Verbeke P. Nullens H. and Adams F. Proc. Analyt. Div. Chem. Soc. 1978 15 18. Franken P. E. C. Thin Solid Films 1976 31 337. Cooper J. A. Wheeler B. D. Wolfe G. J. Bartell D. M. and Schlafke D. B. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977, p. 169. Lloyd W. G. and Francis H. E. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 166. De Kalb E. L. and Fassel V. A. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich.1976,” NTIS Springfield Va. 1977 p. 209. Chung F. H. Adv. X-Ray Analysis 1976 19 181. Cronch S. M. Ehmann W. D. Laumer H. W. and Cabbard F. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 190. Yousif A. N. and Al-Shahristani H. Int. J . Appl. Radiat. Isotopes 1977 28 759. Teller S. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich. 1976,” 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. Chem. 1977 49 1734. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 166. 157. 158. 159. 160. 161. 162. 163. 1977 49 1313. Biochem. 1977 10 127. 213. 1977 8 123.1973,” Southwest Research Institute San Antonio Texas 1973 p. 47. 164. 165. 166. 167. 168. 169. NTIS Springfield Va. 1977 p. 194 March 1979 EMISSION ANALYSIS. A REVIEW 195 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. Carr-Brion K. G. X-Ray Spectrom. 1973 2 63. Kawatra S. K. and Dalton J. L. Canmet Rep, 1977 77 21. Grobstuck P. Gumprich M. Ihlefeldt J. Kopineck H. J. and Tappe W. Stahl Eisen 1977 97, Labrecque J. J. J. Radioanalyt. Ckem. 1977 41 127. Labrecque J. J. Preiss I. L. and Pandian S. “Proceedings of ERDA X- and Gamma-Rap Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 173. Elgart M. F. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich.1976,” NTIS Springfield Va. 1977 p. 227. Ehn E. X-Ray Spectrom. 1973 2 27. Keenan J. A. Appl. Spectrosc. 1975 29 63. Bertrand C. C. and Linn T. A. Jr. Analyt. Chem. 1972 44 383. Metcalf S. G. Analytica Chim. Acta 1977 93 297. Klockenkamper R. and Hasler K. 2. Analyt. Chem. 1978 289 346. Buchnea A. McNelles L. A. Sinclair A. H. and Hewitt J. S. “Proceedings of ERDA X- and Wogman N. A, Rieck H. G. and Kosorok J . R. Nucl. Instrum. Meth. 1975 128 561. Reuter F. W. Hautala E. Randall J. M. Friedman M. and Masri M. S. “Proceedings of the 2nd International Conference on Nuclear Methods of Environmental Research 1974,” NTIS, Springfield Va. 1974 p. 168. 1106. Gamma-Ray Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 242. Nucci J. F. and Stearns R. L. Soil SGZ’. 1977 123 264. Mucci J. F. and Stearns R. L. J. Chem. Educ. 1975 52 750. Delmastro A. M. “Proceedings of ERDA X- and Gamma-Ray Symposium Ann Arbor Mich., 1976,” NTIS Springfield Va. 1977 p. 215. Hanson V. F. AppZ. Spectrosc. 1973 27 309. Tominaga H. Enomoto S. Senoo M. and Tachikawa N. “Proceedings of ERDA X- and Gamma-Kawamoto A. Hirao 0, Kashiwakura J . and Gohshi Y. “Proceedings of ERDA X- and Gamma-Hurley R. G. and Goss R. L. X-Ray Spectrom. 1978 7 70. Ray Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 211. Ray Symposium Ann Arbor Mich. 1976,” NTIS Springfield Va. 1977 p. 263. Received July 6th 1978 Accepted October 5th 197