Sources of Internal Scattering of Electrons in a Cylindrical Mirror Analyser (CMA)код для вставкиСкачать
SURFACE AND INTERFACE ANALYSIS, VOL. 24, 152-162 (1996) Sources of Internal Scattering of Electrons in a Cylindrical Mirror Analyser (CMA)* M. M. El Gomati and T. A. El Bakush Department of Electronics,University of York, Heslington, York YO1 5DD. UK A number of possible sources of internal scattering of electrons in a single-pass Varian cylindrical mirror analyser (CMA) have been identified and their relative contributions to the electron spectral background evaluated. These sources include: the inner surfaces of the outer cylinder; the input and output apertures of the inner cylinder; the field trimmers between the two cylinders; the walls of the exit aperture in front of the electron detector; the outer surfaces of the inner cylinder. A tbeory/experiment comparison of the above sources has verified that the two most significant of these are the walls of the exit aperture and the field trimmers. The other sources contribute only negligibly to the total signal of internally scattered electrons. Simple modifications of the spectrometer and its detector show a large reduction in the internal scattering signal. A simple experimental measurement for determining the relative contribution of the various internal scattering sources is suggested. This measurement is recommended instead of the commonly used peak-to-background ratios as a diagnostic tool for detecting and evaluating all the above-mentioned sources of internal scattering of electrons in a CMA. INTRODUCTION The characterization of electron energy analysers to produce the exact number of electrons versus their energy distribution ( N ( E ) or E . N ( E ) us. E ) in an Auger electron spectroscopy (AES) experiment has been a challenging task since the development of AES as a surface analytical technique. The identification and correction of the factors that alter the faithful reproduction of this distribution has witnessed two stages of development that accompanied signal acquisition in AES; namely the differential and direct energy distribution modes. In the early days of the AES technique and up until the early 1980s, the majority of spectra were collected in the differential mode. The spectrometer properties that were considered to affect the Auger signal in this case were the magnetic field around the sample and inside the analyser,’.’ the transmission properties (function) of the a n a l y ~ e r , ~the - ~ analyser field of view and its depth of ~ o c u s . ~As - ~the direct energy mode became more widely used, some of these factors needed revisiting, particularly the transmission properties of the analyser, the detector efficiency and its operating conditions and the electronic Internal scattering of electrons inside the analyser has become the latest factor to be taken into consideration in the characterization of spectrometers.’ 3-17 The importance of identifying these factors and removing their effects to some known precision has gathered momentum as more use of the spectral background k made in quantitative AES and in Auger mapping. It is now the practice, for example, that Auger imaging is carried out in the direct energy mode. The surface concentration of an element in this mode is estimated to be proportional to the height of its Auger peak. This height * This paper is dedicated to Professor Martin Pruttan (University of York, Department of Physics) on the occasion of his 60th birthday. CCC 0142-2421/96/030152- 11 0 1996 by John Wiley & Sons, Ltd is determined by measuring the electron count at the peak minus that due to the background at the same energy. The latter count is estimated by extrapolating the background from at least two counts on the highenergy side of the Auger peak used to map a solid surface.I8 For quantifying the resulting maps, the underlying spectral features that are used in forming it must also be quantifiable. There are also a number of studies that have shown the advantages in using the direct energy mode for quantitative analysis in preference to the differential spectra where the shape of the Auger peak in question is changed as a result of the chemical environment of the emitting atom. In addition to these, the use of the whole or part of the spectral background as an aid to the quantification of spectra and images has been suggested or demonstrated in a number of recent studies.’8-’2 These studies have led to a marked improvement in the manufacture and operation of spectrometers, as has been manifested in the various round-robin experiments that were carried out over the last 10-15 years. For example, SeahZ3compiled experimental data collected from various spectrometers in the differential mode and found that the ratio between the amplitudes of the Auger peaks of copper at 60 and 920 eV differed from the one instrument to another by nearly a factor of ten. Another study conducted by the ASTMZ4 showed that intensity ratio of the Auger signals in the differentiated mode varied from one instrument to another with a scattering factor ranging from 0.79 to 2.30. Recently, work carried out by the Japanese VAMAS groupz5also showed similar results but with a smaller scattering factor ranging from 1.03 to 1.65. In spite of these improvements, the effects of some of the above-mentioned factors were not included in these reported studies, and hence an absolute spectrum is not a product for most spectrometers available for the AES analyst or the instrument operator. One of these factors is the scattering of electrons inside the analyser, generating an unwanted contribution to the spectral backReceived 19 July 1995 Accepted 3 November 1995 INTERNAL SCATTERING OF ELECTRONS IN A CMA ground of an AES experiment. This signal, referred to as internal scattering (IS) and schematically illustrated in Fig. 1, can be a substantial percentage of the overall collected spectral background signal.' Its reduction or elimination should assist the task of quantifying AES and SAM data. Moreover, the exact electron energy distributions could be compared with Monte Carlo simulations of electrons in solids to discover the presence of systematic trends in this distribution as a function of the sample's atomic number, the incident beam energy or the experimental g e ~ m e t r y .Understanding ~~~~~ the causes of IS is therefore most useful in quantitative Auger electron spectroscopy and imaging. It is interesting to note that the effects of internal scattering of electrons inside the analyser are more pronounced in the direct mode in comparison to the differential mode. This is because the latter basically suppresses the spectral background (where this unwanted signal shows most). Consequently, addressing the internal scattering issue has accompanied the use of the direct energy mode in collecting spectra and images and has therefore come much later in the characterization of electron energy analysers than the more obvious effects like the magnetic field, the analyser transmission properties and its field of view, etc. The present study is a theory/experiment comparison of the sources of IS in one of the most widely used electron energy analysers, the cylindrical mirror analyser (CMA). Although a great deal is known about the performance and focusing properties of this analyser, an analyst still faces difficulty in obtaining the true electron intensity-energy distribution, N ( E ) [or E . N ( E ) for this type]. This has been mainly attributed to the contribution of internal scattering of electrons inside the analyser and is considered by some to be a fundamental limitation of the sensitivity of the CMA. It was with this picture in mind that we started the present investigation. The case for identifying and correcting the internal scattering of electrons in CHAs has been addressed in great detail by Seahi4 and Greenwood et and will not therefore be discussed in detail here. Seah found out that internal scattering comes mainly from the outer hemisphere, as electrons of kinetic energy a few 153 electron-volts higher than the pass energy tend to strike that part of the hemisphere near to the output position. These in turn produce secondary electrons that find their way to the electron detector. Seahi4 suggested that the effect of these electrons could be reduced to negligibly low values if the analyser is operated with pass ener~ gies of -50 eV or higher. Greenwood et ~ 1 . ' have proposed a method that corrects for the transmission properties and the effects of internal scattering of electrons in their CHA. However, such methods cannot be applied to CMAs and hence a study of the origin of IS in this analyser, its effects on the collected spectra and its elimination or minimization is required. In the present study, we have identified a number of possible sources that give rise to an IS signal in a singlepass CMA (and some will also apply to a double-pass CMA). These include; the inner surfaces of the outer cylinder; the outer surfaces of the inner cylinder; the exit slit aperture walls in front of the electron detector; the input aperture of the inner cylinder; the output aperture of the inner cylinder; the field trimmers between the two cylinders for the present type of analyser. Although the present study considers a specific analyser design (Model number 981-2707 made by Varian Associates, USA), the use of field trimmers of different shapes and position with respect to the exit aperture at the inner cylinder is an accepted practice in other CMA designs to correct for field penetration at the inner cylinder apertures. The present study and its findings may apply to such cases, particularly if these field trimmers and the output slit aperture are of the shape and/or the position as shown in Fig. 2. An important finding of the present study is the elimination of some of the above listed sources as significant contributors to internal scattering, while, on the other hand, we have identified two of these as major sources. Furthermore, a simple experimental measurement complimented with electron trajectory simulation is suggested as a diagnostic tool for the detection of, and in estimating, the relative contribution of the different IS sources. This is in preference to the more widely used peak-to-background ratio (P/B) from elemental standard spectra as an indication of the presence and amount of IS in spectrometers. (b) IS I E Figure 1. The effect of internal scattering of electrons in an AES analyser on the collected spectrum. (a) Spectra with no internal scattering showing a given P/B ratio. (b) A simple constant internal scattering (IS) contribution from the analyser internal walls. This is not necessarily the case, and a non-uniform contribution of IS is most likely. (c) The effect of IS on an N ( E ) spectrum. Note that the P/B ratio is now worse than in the case where B > B. M. M. EL GOMATI AND T. A. EL BAKUSH 154 / I VmIMFKj S e a m h y ~ T r a j ~MagneticShield Figure 2. The structure of the single-pass cylindrical mirror analyser (CMA) used in the present study. The various trajectories depicted show that electrons emitted from the sample surface with the correct energy (corresponding to the pass energy of the analyser) and angle of emission pass through and are focused at the detector (as shown in the bottom half of the diagram). Electrons emitted with higher energies than the pass energy of the analyser (shown in the top half of the diagram) strike the inner surfaces of the outer cylinder and generate secondary and backscattered electrons. Some of these may have the correct energies and angles of emission to reach the electron detector, as depicted. The theoretical investigation carried out in this study involved the use of electron trajectory simulations. Two computer programs were used: the SIMION electron and ion simulation package” and a purpose-written p r ~ g r a m . ’ The ~ latter program considers the fate of electrons originating at the outer cylinder with random energies and angles of emission, in addition to those originating at the sample. Full details of this program are given el~ewhere.’~ A new equation of motion for the former type of electrons has been derived3’ which is different from that used in the literature31 for calculating the trajectories of electrons originating at the focal point of the analyser. THE ORIGIN OF INTERNAL SCATTERING OF ELECTRONS IN A CMA Internal scattering (IS) of electrons occurs when some of the electrons emitted from the surface under analysis strike the inner surfaces of the spectrometer and generate secondary electrons that bear no direct relationship to those emitted from the sample under analysis, except in being generated by them. It has long been argued that some of these secondary electrons may be generated with the appropriate energies and emission angles to reach the detecting system and contribute to the spectral background signal, as depicted in Fig. 1. The overall contribution of these electrons depends upon the geometry of the analyser, the potential applied to its various electrodes and the degree of surface coverage of its internal surfaces. The latter arises as a result of either particles sputtered during specimen cleaning, particularly during depth profiling, or purposely coating with a low secondary electron emission material (normally used to reduce the secondary electron emission from inner analyser walls). The current perception of the AES community of the causes of IS in CMAs is based on the assumption that the secondary electrons that are generated at the outer cylinder of the CMA, as a result of energetically backscattered electrons from the surface of the solid under investigation, tend to have a direct line of sight with the detecting system or can find their way to it after a number of multiple scattering events inside the analyser. This is considered to be a limitation in the CMA performance. Another source of internal scattering of electrons in this analyser is the surfaces of the exit aperture in front of the electron d e te ~ to r .~In ’ this case it is not the highly energetic electrons emitted from the sample that are the cause of scattering but rather electrons emitted with slightly different energies and angles of emission from those being analysed, i.e. electrons of energies E AE and angles 0 f A@. Some of these electrons strike the surfaces of the exit slit aperture placed in front of the detector and generate secondary electrons which are subsequently collected by the detector and appear as an increase in the electron background signal. In addition to the above, there are four more likely sources of IS in this analyser which have not been systematically investigated before. These include the input and output apertures of the inner cylinder, which are normally covered with a high-transparency mesh to stop field penetration; the outer surfaces of the inner cylinder; and the field trimmers between the two cylinders as depicted in Fig. 2. MEASUREMENT OF INTERNAL SCATTERING IN THE CMA ~~~~~ ~~ ~~~~~~~- Two experiments have been carried out to estimate and reduce internal scattering in the present CMA. The first, used as a diagnostic tool, involves the use of a monoenergetic electron beam to represent a proportion of the electrons that leave the sample in an AES experiment. Figure 3 outlines a schematic illustration of this experiment. The spectral energy distribution E . N ( E ) is then recorded from a few electron-volts up to E , + 50 eV, where E , is the incident electron energy. Although only one electron energy is used in this experiment, it can be easily seen that it represents in an integral form (i.e. when using other incident beam energies) what happens in a real AES experiment. The second experiment involves the measurement of the P/B ratio from three elemental Cu spectra collected under different electron detector configurations. This ratio is measured in its simplest form where P = P and B = B’ in Fig. 1. The sample used was kindly provided by Dr M. Seah of NPL, UK and consists of a thin foil of pure copper measuring 10 x 10 mm. Prior to data collection, the surface of the sample was normally in situ bombarded with a beam of 3 keV Ar’ ions until no traces of surface impurities were present in the AES spectra. These were collected under the following conditions: a primary beam energy, E , , of 5 key, an electron beam current 1 PA, a dwell time of 100 ms per energy point, an energy interval of 1 eV and spectra collected in the energy range 50-2500 eV. The detection system used consists of a Gallelio analogue electron multiplier (Model 4700) followed by a purpose built signal INTERNAL SCATTERING OF ELECTRONS IN A CMA 155 I 3 I =Outer cylinder s U= Field trimmas I Energy 0 650 EP outer cylinder inner cylinder r - / Sample - _I_ Figure 3. Schematic illustrating the concept of a mono-energetic beam to detect and estimate the contribution of internal scattering of electrons in an energy analyser. (a) The spectral background in an AES experiment, with the shaded area being the total area under the spectrum in (c). (b) The experimental arrangement. (c) A schematic of the measurement in (b). digitizerz9 to produce a signal in the form of an electron count which is fed into a counter for subsequent processing. The detection configurations used were : (1) The detector operated in the normal mode, as shown in Fig. 2; (2) A pair of high-transparency metal grids placed between the exit slit aperture and the electron multiplier to act as a filter to low-energy secondary electrons; (3) as above but with the grid in front of the multiplier being biased to - 10 V. The outer cylinder The contribution of this source to the total IS signal has been addressed in an earlier theory/experiment comparative study3’ and the findings of this study suggest that this is not a significant source of IS in this analyser. This conclusion was based on trajectory simulation of electrons originating at the outer cylinder with random energies and angles of emission. The trajectory of electrons of energies up to the incident beam energy were considered, in order to take account of both secondary and backscattered electrons. A total of lo4 electrons for each incident energy in the range 1000-5000 eV in steps of 1000 eV were considered. The trajectories of these electrons are described by a new equation of motion for calculating the distance 2, that a scattered electron (off the walls of the outer cylinder) travels along the z-axis as depicted in Fig. 2. This has the form 2, = 2rz cos I ~ ( K ~ exp ) ~ .(KO ’ sin’ 0) (1) where u1 = (KOsinZ O)’.’ M. M. EL GOMATI A N D T. A. EL BAKUSH 156 q is the electron charge, E , and V stand for the energy of the scattered electron and the potential difference between the two cylinders, respectively, rI and rz are the radii of the inner and outer cylinders, respectively, and 8 is the angle that a scattered electron makes with the cylinder surface. In this simulation all the dimensions of the present CMA were fixed with the exception of the output exit slit aperture in front of the electron detector. This has a width of 290 pm. It is considered to be of paramount importance because of the rather small solid angle that it makes with the output aperture in the inner cylinder. Several other larger slit aperture widths were also considered, as well as the case of no aperture at all. Of course the change of the slit aperture size will change the energy resolution of the analyser. However, this is not important at this point, where the fate of the secondary electrons from the outer cylinder is the issue under consideration. For brevity, only the results obtained in the case of 5 keV incident electrons are shown in Table 1. In the case of the 290 pm aperture, the total contribution at all incident energies is much less than 1%, suggesting that the inner surfaces of the outer cylinder are not a major source of IS in. this CMA. This result has been experimentally confirmed by measuring the P/B of Cu and Ag spectra using two outer cylinders of different inner surface finishes under similar experimental conditions as stated above. These are the original one supplied by the manufacturer (Varian Associates, USA) with a smooth surface finish, and another purposely fabricated at York with 0.5 mm deep grooves. The grooves make an angle of 45" in one direction with respect to the axis of the cylinder, while being at right angles to the other direction and facing the incident electrons. In comparison with a smooth surface, this geometry should decrease the chances of forward scattering of both secondary and backscattered electrons off the surface of the outer cylinder towards the electron detector. It is anticipated therefore that if the surface of this electrode is a major source of IS then this will be reflected in a different P/B of the collected spectra from the two outer cylinders. The results obtained show negligible change in the P/B ratios when using either cylinder. The above result suggests that either both surface finishes contribute equally to the total IS signal and hence a distinction between them is not possible, or that this source contributes negligibly to internal scattering. If the outer cylinder is a significant contributor, then the former suggestion cannot be true due to the difference in electron scattering expected from the two surface finishes. However, it is possible that the IS contribution is only initiated at the outer cylinder by secondary and backscattered electrons off its surface to other parts of the analyser and that it is an electron cascade-like process that follows afterwards and hence the surface finish of the outer cylinder will play a less significant role in this case. This of course could be argued based on the electron trajectory simulations (Fig. 4) of secondary electrons generated at the outer cylinder being accelerated towards the inner cylinder. The second experiment, schematically depicted in Fig. 3, is considered to be more useful than the use of P/B as an aid to identifying and quantitatively estimating the contribution of the various IS sources considered here. Figure 5 shows spectra in the energy range 0-1000 eV collected with an incident electron beam energy of Table 1. The number of electrons that strike the walls of the exit slit aperture for a total of lo4 electrons incident with an energy of 5 keV Aperture size Source size 0.29 mm 0.58 mm 1.45 mm B A+B 0 0 2 2 8 0 2.90 m m 4 11 4.35 m m 7 25 mm 20.0 mm 11 17 21 34 1112 974 outor cyllndor 0 p t l c .xi* Figure 4. Electron trajectory simulation of secondary electrons generated at the outer cylinder with random energies up to those used to generate them. The case depicted is for an incident beam of 1000 eV and a pass energy of 500 eV. It is obvious that all secondary electrons are accelerated towards the inner cylinder, while for the backscattered electrons the majority strike the field trimmers but some manage to exit from the inner cylinder aperture. INTERNAL SCATTERING OF ELECTRONS I N A CMA (4 700 - Smooth surfaced cylinder II 600 - 500 - .i400- 300200100- 0 0 100 200 300 400 500 600 Energy (eV) 700 800 900 1000 157 rotated until a maximum current is obtained at the outer cylinder, which is disconnected from the voltage supply for this measurement. This ensures that the incident electrons do in fact resemble those with the same energy emitted from the sample in the usual AES experiments. In estimating the total contribution of the IS signal, the area under the curve between 50 eV and E, - AE, where E , is the incident beam energy and AE is of the order of a few electron-volts, is ratioed to the area under the peak at E,. In the present case this ratio is -0.20 for both surface finishes. Similar figures to this are obtained with different incident beam energies in the range 500-2500 eV. This suggests that the total IS signal resulting from the whole spectral energy distribution N ( E ) is therefore also of a similar magnitude. On close inspection of the spectra of Fig. 5 and by complementing them with electron trajectory simulations, one is able to draw a clearer picture of the contribution of the various parts of the analyser considered to be likely sources of internal scattering. Figure 6 is a set of trajectory simulations of fixed energy electrons incident on an outer cylinder which is being ramped in the usual manner to collect an E . N ( E ) us. E spectrum as shown in Fig. 5. Seven different analysis energies are considered. It is clear from the trajectory simulations that the 920 eV incident electrons strike the outer cylin- 100 -10 Vapplied 10 a grid c Zl00W 50Smooth surfaced cylinder 500 - - - -o 100 200 300 400 800 Energy (eV) 500 , 1 700 BM) --a 900 1m Figure 5. Spectra collected from the t w o cylinders with a 920 eV beam directed towards the inner surfaces of the outer electrode by biasing the sample with - 1 kV: (a) from the cylinder with a smooth surface; (b) from the cylinder with a sawtooth surface as shown in Fig. 2; (c) same as (b) but with - 10 V applied to a grid placed on the front of the electron multiplier. mag. x140 200 100 -920 inner beam tively eV using the two outer cylinders with different surface finishes. In this experiment the electron is directed towards the sample, which is negabiased to about -1.0 kV. The sample is first 200 300 400 500 600 700 BOO 900 1000 Figure 6. Electron trajectories of 920 eV emitted in the angular range 36.3-48.3" (a-g) The condition of the cylindrical mirror analyser (CMA) at various pass energies. It is clear from the figure that the 920 eV electrons strike the field trimmers in the energy range from -650 t o -890 eV. This is in close agreement with the experimental results shown in Fig. 5. 158 M. M. EL GOMATI AND T. A. EL BAKUSH der only when the spectrometer is set in the energy range 0-600 eV. The area under the spectrum of Fig. 5 in the corresponding energy interval is therefore the contribution of the outer cylinder to the total IS signal. For the present set-up and using either of the two outer cylinder finishes, this amounts to only about 1-1.5% of the signal under E,. In other words, the outer cylinder of this CMA is a less significant source of IS. The field trimmers The expanded portions of the spectra in Fig. 6 clearly show seven broad energy peaks in the energy region 600-950 eV. The total signal under the area of the spectrum in this energy range amounts to only 556% of that under E,. At first sight these peaks appear as plasmon loss peaks. However, they differ from normal loss peaks on three accounts. First, the incident electrons are not interacting with a sample as in the usual scattering experiments that give rise to plasmon losses and the appearance of loss peaks. Second, the energy separation of these peaks ranges between 70 and 100 eV for incident beam energies in the range 1000-2000 eV. The usual energy-loss peaks occur typically in the range 15-30 eV for most elements. Thirdly, the present peak energies and energy separation between the individual peaks are dependent on the incident beam energy. The energy separation between the individual peaks increases as the incident beam energy increases. Electron energy-loss peaks have fixed energy separation between each other, representing multiple plasmon losses, and are fixed with respect to the incident beam energy. These observations suggest that these peaks originate via a different mechanism to electron losses. A comparison between the 920 eV incident electrons and the electron trajectory simulations of Fig. 6 shows that the incident electrons at these energies strike the field trimmers placed between the two cylinders as well as small portion of the outer cylinder near the field trimmers (labelled C in Fig. 2) and the far edge of the inner cylinder. It is no coincidence that the present CMA also has five such field trimmers as shown in Fig. 2, hence making a total of seven distinct features within the analyser that the incident electrons strike in this energy range. The consequence of such an interaction is the appearance of seven peaks shown in Fig. 5(a, b). The field trimmers are made of L-shaped rings connected via a resistor chain to the outer cylinder to correct for changes to the field distribution in this region of the analyser. It is usually the practice to place these rings closer to the outermost envelope of the trajectory of the analysed electrons. This allows a shorter pair of cylinders to be used, with a ground termination between them where the field trimmers can be mounted on. The benefit of this arrangement is an easier access to the various parts of the spectrometer, particularly for the electrical connection when using a co-axial electron gun to the CMA, as is the case in the present spectrometer design. However, placing the field trimmers at this position makes a direct line of sight between any electrons that are generated at this area and the exit slit aperture and its housing. It is the aperture of the latter, shown shaded in Fig. 7, that acts as a source of ter- tiaries and other remotely related electrons that contribute to the IS signal. In order to establishlconfirm experimentally that these peaks are associated with the field trimmers, we have replaced the latter with a single electrode placed at ground potential. This was made out of an insulated disk (made out of PEEK, Polypenco Ltd) coated with a layer of gold about 200-300 nm in thickness [Plate l(b)]. Replacing a family of field trimmer electrodes with a single grounded electrode will have the effect of altering the field near the exit aperture of the inner cylinder, with a subsequent change to the spectrometer’s resolution and an energy shift of the analysed peaks. This, however, is not as important an issue for the present measurement, which is primarily concerned with establishing whether these field trimmers are the cause of the observed peaks or not and in estimating their contribution to the total IS signal. The presence of a single flat disk should, therefore, based on the above argument, replace the five peaks due to the field trimmers with two peaks. This has indeed been the result obtained, as shown in Fig. 7(c) where a total of four instead of seven peaks is obtained. This confirmed that these peaks are due to energetic electrons striking this part of the analyser and subsequently initiating an IS contribution. The energetic electrons that strike the field trimmers and the ground plate behind them give rise to both secondary and backscattered electrons. The majority of these electrons have energies in excess of a few tens of electron-volts for 1000 eV incident electrons. By placing a high-transparency metal mesh in front of the electron multiplier, biased to -10 V, it was possible to reduce the total amount of the IS signal to 1% from 8% obtained with no biasing. In addition, the peaks due to the field trimmers have also disappeared, as shown in Fig. 5(c). This implies that the IS signal from this part of the analyser is made primarily of low-energy secondary or tertiary electrons of energies < 10 eV. It is likely that most of these electrons are indeed tertiaries at the output aperture of the inner cylinder or at the slit aperture holder itself, as both the backscattered and secondary electrons that are produced at the field trimmers are accelerated under the influence of the field towards the inner cylinder, as depicted in the electron trajectory simulation of Fig. 4. There are two mechanisms/processes that could give rise to these peaks. The first is from the energetic incident electrons striking the edges of the five field trimmer metal rings. These will have a high secondary electron yield because of their position, making a near-grazing angle of incidence with the incident electrons compared with the ground electrode that carries them, which makes a near-normal angle of incidence. Figure 6 shows these peaks to appear at an energy where electrons are centred on the edge of the rings. Further, the edge sharpness of the metal rings, their thickness and material could also play a role in contributing to the appearance of these peaks. Alternatively, these peaks may arise from the physical presence of the metal rings, acting to shadow part of the produced backscattered and secondary electron signal generated at this part of the analyser. These in turn will cause a single peak from the earthed electrode to be divided into several peaks, depending on the number of rings used. In the present - - 0'1O0 2.10~ 4'10~ 100 200 300 400 500 600 Energy (eV) Field trimmers replaced with a single disk 700 800 900 1000 Plate 1. (a) A photograph of the Varian cylindrical mirror analyser (CMA) showing the position and shape of the field trimmers. (b) A single flat annular disk made out of a PEEK insulator coated with 200-300 n m gold film that is used in place of the field trimmers shown in (a). (c) Spectrum collected with the single disk shown in (b). 2 6'103 -w 8*102 1*104 INTERNAL SCATTERING OF ELECTRONS IN A CMA analyser, there are five such rings, and five peaks are also observed. It is more likely, however, that both of these processes will contribute to the appearance of these peaks. It is interesting to note that in a normal AES spectrum, and indeed for the present CMA, one does not see the peaks due to the field trimmers for two reasons. The first is fundamental and is associated with the signal-tonoise ratio obtainable in an AES experiment. This can be understood by an inspection of Fig. 5. The peak heights due to these field trimmers, or more appropriately the peak-to-trough heights, are only a very small fraction of the peak height at E , , amounting to <1/ 1000. This count (or peak-to-trough height) is much smaller than the statistical fluctuations of electron emission from the sample and the stability of the incident beam current, and would not therefore be visible in a normal spectra. The second reason is that the peak energies due to the field trimmers are a function of the incident beam energy as discussed earlier. In an AES experiment the whole spectral energy distribution, N(E), will act as a source for the production of these peaks. This will therefore have the effect of smearing off their appearance in the spectra as a result of being superim- 159 posed on top of each other. All in all, the overall contribution is to cause a smoothing effect to these peaks but to raise the value of the spectral background of the obtained spectra as shown in the diagram of Fig. 1. Finally, the overall contribution of this part of the analyser to the total IS signal is likely to be roughly constant with respect to energy, based on the above discussion, but its magnitude will greatly depend on the transmission and energy resolution properties of the spectrometer. The exit slit aperture The contribution of the final slit aperture to the IS signal has been suggested by El Gomati and El Bakush3' to be an important source of IS in this class of analysers. Figure 7 shows the trajectory of electrons that leave the sample with an energy of 500 _+ 1 eV in a solid angle of 42.3 k 6". With the outer cylinder biased to the voltage required for the analysis of 500 eV electrons, it is clear that some of these electrons strike the walls of the exit slit aperture. We have used the York trajectory program to estimate the number of electrons (c) E= 502 eV Figure 7. Trajectory simulations of electrons of energies 500 * l eV in a solid angle of 42.3 *6' (a), (b and c) for electrons with energies 498 (b) and 502 (c) and emitted at three different angles of 36.3". 42.3" and 48.3". All simulations are for a CMA pass energy of 500 eV. M. M. EL GOMATI AND T. A. EL BAKUSH 160 ~ that strike the walls of this aperture in the energy and angular range of E, 5 1 eV and 12", respectively. One hundred incident electrons of several energies (498, 499, 500, 501 and 502 eV) and an angular interval of 1" were used. Table 2 shows the number of electrons of various energies that strike the walls of the exit aperture when the pass energy of analyser is set to 500 eV. Electrons emitted with energies (499 eV or > 501 eV do not pass through this aperture, which has a width of 0.29 mm, as is evident in Fig. 8(b, c). This is an expected result in terms of the energy resolution of this analyser. The results shown in Table 2(a) were based on the assumption of ideal sharp edges. In reality, however, edges are blunt. T o assess the effects of non-sharp aperture edges, an exit slit aperture with a truncated vertex was also considered. The length of the truncated face presented to the incident electrons was chosen to be 100 and 150 pm. The number of electrons striking the edges of the aperture was found to increase as the vertex varied from a point to an increasingly broad face (Table 2). Using a monoenergetic beam of incident electrons and negatively biasing a high-transparency metal mesh in front of the electron detector to - 10 V should inhibit any of the low-energy secondary electrons from reaching the electron detector. In performing this experiment, it was possible to reduce the total electron signal at E , by about 18-20%, which is in very good agreement with the simulation results. The above experiment was repeated with the incident electron beam scanned to fill-in the input aperture of the inner cylinder. This to some extent simulates electrons that leave the sample with angles in the range 36.3-48.3". Negatively biasing the metal mesh in front of the detector to -10 V has shown a reduction of up to 25% of the signal at E , in this case. The above results have been confirmed in AES by measuring the P/B of a Cu elemental standard as described earlier. The results obtained are given in Table 2. The percentage of electrons with different energies that strike the walls of the exit aperture and produce secondary electrons: (a) for an ideal case where edges are very sharp (i.e. assumed as a point); (b) truncated edges of 100 pm, (c) truncated edges of 150 pm - Electron energy (a) Sharp edges 498 eV 499 eV 500 eV 501 eV 502 eV (b) Edges 498 eV 499 eV 500 eV 501 eV 502 eV - - 0 29 mm apeflure 0.58 mm apenure - 13 14 9 13 11 - - 14 12 100 pm ( c ) Edges - 1 50 pm 498 eV 499 eV 500 eV 501 eV 502 eV 13 18 13 - - 23 28 23 - 18 19 - 19 19 29 30 - 30 27 Table 3. Peak-to-background ratios measured from the spectra collected with different detector operating conditions Improvement with respectintoPJB Detection condition ( P P ) C" With no grid With double grid With double grid and a biased grid 0 0834 0 006 0 0927 0 006 0 0967 *O 006 * * case with no grid 11% 15% Table 3. These clearly confirm that an improvement of -15% is obtained in the P/B ratio by modifying the configuration of the electron detector. This figure of 15% should be compared with an expected total IS signal from the deflection experiment amounting to about 25-30%. Although there is a difference in the overall ratio in the two classes of experiments, amounting to nearly a factor of 2, the trends of the obtained results are consistent with the given interpretation. The reason for this discrepancy could be due to the difference in the energy and angular distribution of the emitted electrons in the two experiments. The input and output apertures of the inner cylinder The contribution of these two sources is rather difficult to measure or quantify. Their effects are likely to come from the high-transparency mesh that covers them, the purpose of which is to stop field penetration. It is unlikely that the presence of the metal mesh will act as a source of the detected IS electron signal. This is based on similar P/B ratios obtained from CMAs which either do not have a metal mesh at the inner cylinder, as reported by Frank et a1.,34or from a recent design by Flora et al.j5 which replaced the metal mesh with a family of longitudinal wires giving an 87.5% transparency. In addition, it has been shown in the trajectory simulation of Fig. 7 that electrons with energies which are greater or less by a few electron-volts than those being analysed that manage to exit from the inner cylinder tend to strike other parts of the aperture holder and hence are not detected. Any scattering off the mesh is likely to cause a larger energy loss than just 2-3 eV and hence such scattered electrons tend to have a different trajectory from those under analysis, i.e. with respect of the voltage of the outer cylinder. We have also halved the size of the input aperture but have kept the aperture at the output the same. Repeating the experiment depicted in Plate 1 resulted in similar values in terms of the internal scattering at E , . It is concluded therefore from this experiment and the preceding discussion that the input and output apertures of the inner cylinder are not a significant source of IS in this analyser. The outer surfaces of the inner cylinder Internal scattering from this surface can occur as a result of energetic electrons backscattered from the outer cylinder towards the inner cylinder. These and other generated secondary electrons could impinge on the outer surfaces of the inner cylinder and in so doing INTERNAL SCATTERING OF ELECTRONS IN A CMA initiate further secondaries or tertiaries that may later on find their way to the electron detector. It could also be due to low-energy secondary electrons from the sample, with respect to those being analysed, which are projected (i.e. turned hard on) onto the surface of the inner cylinder as they enter the analyser. This is depicted in the trajectory simulation of Fig. 8. These too could initiate another cascade of secondaries and tertiaries that may contribute to the total IS signal. We have used electron trajectory simulations to study the fate of electrons that leave the surface of this electrode at random energies and angles of emission. The results obtained suggest that none of the emitted electrons as a result of this mechanism could be detected in the spectra. The above result is confirmed by repeating the experimental arrangement shown in Fig. 3. By detecting the electron energy distribution to beyond the incident beam energy E , , it is possible to compare the obtained results with the above simulation experiment. In this case, the incident electrons have energies less than the analysis energy and hence, as schematically shown in Fig. 8, these could resemble both backscattered electrons reflected from the outer cylinder and low-energy electrons leaving the sample. Figure 5 shows the experimental results obtained. It is clear that there is no signal beyond the E , energy, which confirms that although scattering off this surface must occur, the emitted electrons are collected inside the analyser (probably flowing to ground) and hence are not detected. In an earlier experiment, the surface of this electrode was coated with a low secondary electron yield carbon dag to reduce electron emission from its surface. No visible change in the P/B ratio was obtained as a result of this treatment either. The conclusion of these experiments suggests that this electrode is not a significant source of internal scattering in the output signal from a CMA. DISCUSSION The present study has identified a number of the most likely sources of internal scattering of electrons in a single-pass CMA. Of the six different sources considered, the exit slit aperture and the field trimmers are 161 found to be the most significant contributors, while the inner surfaces of the outer cylinder were found to contribute much less than 1.5% of the total IS signal. Further, based on the findings of the present work and using simple geometrical arguments, this source is unlikely to be a major IS source for other CMAs of similar design principles. However, the only part of the outer cylinder which could act as an IS source is that part facing the exit aperture of the inner cylinder, i.e. part C on Fig. 2. The effect of this source will be similar to the field trimmers (as can be seen in Fig. 5 ) with the peak at -622 eV. It may be argued in this case that the field trimmers should be designed such that they are placed as close as possible to the outermost electron trajectory envelope as shown on the bottom part of Fig. 2. In this case the more energetic electrons will be trapped in-between the various trimmers electrodes. In addition, most generated secondary electrons from this source will not have a direct line of sight to the exit aperture, but rather they will be most likely to strike the inner cylinder and be transmitted to ground. Such an arrangement is expected, from simple geometrical arguments, to reduce the size of this signal by up to an order of magnitude in comparison with the present geometry. An alternative scheme to filter this IS signal in the case of systems employing electron counting could be achieved by raising the threshold of the electron detector to above the energy of the IS electrons. Eliminating IS from spectra is fundamental for quantitative AES and SAM. It is interesting to note that the study reported by SeahI5 for a number of different CMA designs has shown varying degrees of IS. It is likely that these variations are due to changes in the different design or constructions of the analysers studied. For example, it is very unlikely that a doublepass CMA will have an IS contribution from field trimmers. This is because the second pair of mirrors does not see any electrons of higher energies than those being analysed and hence any generated electrons from field trimmers in the first pair of mirrors, if present, will be filtered by the second pair. Also, the PHI (Perkin Elmer) single-pass CMA design (PHI Model 590 scanning Auger microscope) is unlikely to show the effect of the field trimmers. In this design a small ‘biased’ ball is placed in-between the exit aperture and the electron detector, which could be acting in this respect to deflect outer cylinder -Electrons with mergy less than the analysed ones falling on the inner cylinder ~ Secondary electrons generatedat the inner cylinder. Figure 8. Trajectories of electrons entering the cylindrical mirror analyser (CMA) with energies less than those being analysed. This resembles what happens in an AES experiment where spectra are collected sequentially in energy. 162 M. M. EL GOMATI AND T. A. EL BAKUSH the low-energy electrons produced as a result of energetic electrons striking the field trimmers. Goto et aL9 in their design used a pair of longer cylinders to avoid field changes around the exit aperture of the inner cylinder. Although this eliminates the need for field trimmers, the long cylinders are usually undesirable as they occupy space that can be best used for other purposes. The exit slit aperture is found to be the major source of IS in the present design. The effects of this source, however, are likely to be present in CMAs but with varying degrees from one instrument to another, depending on the detailed construction of the final exit slit aperture. Internal scattering from this source has been reduced in the work reported here by negatively biasing a metal mesh in front of the electron detector to - 10 V. This has the undesirable effect of losing the lowenergy part of the spectrum. A different solution under investigation in the authors’ laboratory is to modify the field trimmers so that they are placed at the outermost trajectory as shown in the lower part of Fig. 2. CONCLUSION A theory/experiment comparative study of the likely sources of internal scattering of electrons in a CMA is reported. A total of six such sources have been considered. The results obtained have eliminated one source, the outer cylinder, as being the most significant contributor of 1s in this CMA. ~ ~it has identified ~ ~ a new source, the field t r h m e r electrodes placed inbetween the two cylinders, as a significant source of IS in this analyser. Backscattered and secondary electrons off this area of the analyser are thought to impinge on the output aperture of the inner cylinder. As a result of this process, tertiary and more remotely related electrons with energies c 10 eV are produced. These have a direct line of sight to the electron detector and hence contribute to the collected spectra as background. The output exit aperture is found to be by far the largest source of IS in this class of analysers. It was found that the effect of internal scattering can be reduced by the addition of a negatively biased high-transparency metal mesh in front of the electron detector. In the present set-up this arrangement does not allow the potential of the front end of the electron detector to be altered. The drawback of this solution is the loss of that part of the spectrum up to -20 eV. However, this region does not usually have usable Auger peaks and hence this is considered to be a relatively small price to pay for the improvement of the quality of the spectra. Perhaps it would be better to place a double mesh in front of the electron detector. This will allow the potential at the front end of the electron detector to be altered to improve its gain at low voltage operation. Finally, a new experimental arrangement combined with electron trajectory simulations is suggested as a diagnostic tool to study the relative contribution of the various sources of IS in this class of analysers. Acknowledgement The authors would like to thank DI S. Bean Of York Electron Optics h ~ ~ , (YEO) of the University of York, and Dr M. Seah of NPL, UK for useful discussion, and D r R. Roberts for the critical reading of the manuscript. REFERENCES 1 . P. H. Holloway and D. M. Holloway, Surf. Sci. 66, 635 (1 977). 2. S. lngrey and W. D. Westwood, Surf. Sci. 100, 281 (1 980). 3. F. C. Eagen and E. N. Sickafus, Rev. Sci. lnstrum. 48, 1269 (1977). 4. N. E. Erickson and C. J. Powell, Surf. Interface Anal. 9, 1 1 1 (1 986). 5. D. C. Peacock, M.Prutton and R . H. Roberts, Vacuum 34,497 (1984). 6. E. N. Sickafus and P. H. Holloway, Surf. Sci. 51, 131 (1975). 7. M. P. Seah, J . Electron Specrrosc. Relat. Phenom. 48. 209 (1989). 8. M. P. Seah and G. C. Smith, Surf. lnterface Anal. 15, 751 (1 990). 9. Goto et a / . ,QSA-8. Surrey (1 994). 10. M . P. Seah and G. C. Smith, Vacuum 41,7 (1 990). 1 1 . M. P. Seah, J . Electron Spectrosc. Relat. Phenom. 50, 137 (1 990). 12. M. P. Seah and G. C. Smith, Surf. Interface Anal. 17, 855 (1991 ). 13. M. P. Seah, M. E. Jones and M. T. Anthony, Surf. Interface Anal. 6, 242 (1 984). 14. M. P. Seah, Surf. Interface Anal. 20, 865 (1 993). 15. M . P. Seah, Surf. Interface Anal. 20, 876 (1993). 16. J. C. Greenwood, M. Prutton, R. H. Roberts and Z. Liu, Surf. Interface Anal. 20, 891 (1 993). 17. T. A. El. Bakush, M. M.El Gomati and 2.Liu, Insr. Phys. Conf. Ser. 138, 213 (1992). 18. M .M. El Gomati and C. G. H. Walker,Appl. Surf. Sci. 35, 177 (1 988). 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. M. Prutton and M. M. El Gomati, Surf. Interface Anal. 9, 99 (1986). G. Ichimura, R. Shimizu, T. Sekine, K. Goto and H. Shimizu, Surf. Interface Anal. 11, 94 (1 988). R. Shimizu and 2.-J. Ding, Rep. Prog. Phys. 487 (1992). I. R. Barkshire, M. Prutton, J. C. Greenwood and M . M . El Gonati, Surf. Interface Anal. 20, 984 (1993). M . P. Seah, Surf. Interface Anal. 9, 85 (1 986). C. Powell and N. E. Erickson, J . Electron Specrrosc. 25, 87 (1 982). T. Omhura and R . Shimizu, in Analytical Electron Microscopy, edited by D. C. Joy, p. 325. San Francisco Press, CA (1987). J. A. D. Matthew, W. C. C. Ross and El Gomati, Inst. Phys. Conf. Ser. 130, 383 (1992). J. C . Greenwood, M. Prutton and R . H. Roberts, Phys. Rev. B 49,12485 (1994). D. A. Dhal, J. E. Delmore and A. D. Appelhans, Rev. Sci. lnstrum. 61, 607 (1 990). T. A. El Bakush, D. Phil thesis, University of York (1994). T. A. El Bakush and M. M. El Gomati, J . Electron Spectrosc. Relat. Phenom.. 74, 109 (1995). V. V. Zashkvara, M . I. Korrsinsku and 0. S. Komsmacher, Sov. Phys. Tech. Phys. (English Transl.) 11, 96 (1966). M. M . El Gomati and T. A. El Bakush, QSA-7, Surrey (1 992). M. P. Seah and G. C. Smith, Surf. Interface Anal. 15, 751 (1 990). P. Vasina and L. Frank, J . Phys. E 12, 744 (1 979). P. S. Flora, Huang Min, C. J. Harland and J. A. Venables, Inst. Phys. Conf. Ser. 98, 299 (1 989).