- Email: [email protected]
Medium-voltage EDX and EELS
302
Ultramicroscopy 28 (1989) 302-307 North-Holland, Amsterdam
M E D I U M - V O L T A G E EDX AND EELS M.H. L O R E T T O Department of Metallurgy and Materials, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK Received at Editorial Office 3 November 1988; presented at Workshop March 1988
Experiments have been carried out in which the influence of beam broadening on the spatial resolution of X-ray analytical data has been assessed at 200 and at 400 kV. It has been found that any improvement at the higher voltage becomes significant only at sample thicknesses which axe greater than those that would be used for microanalysis, i.e. at thicknesses at which absorption corrections become dominant. Analyses of particles of TiC have been carried out using electron energy-loss spectroscopy at 400 kV and using a thin window energy-dispersive X-ray analysis facility, at an accelerating voltage of 200 kV. It has been found that the electron-energy-loss data can be quantified without any deconvolution, for reasonably thick samples; and changes in stoichiometry, identified using electron diffraction, are measurable from the electron-energy-loss data. At this stage the thin window energy-dispersive data has been found to be useful only for qualitative analysis in view of the undetermined, and presumed large, influence of absorption.
1. Introduction The potential advantages of medium-voltage analytical electron microscopy have been well catalogued but there is as yet a distinct lack of data which can be used to back up the various claims which have been made by manufacturers and users alike. Indeed there appears to be some disagreement between various workers, with some claiming significant experimental improvements in the spatial resolution of X-ray data from very thin samples [1] and others claiming that the anticipated advantage in spatial resolution is not relevant for practical situations [2]. Recent work using fidd emission scanning transmission electron microscopy has allowed high spatial resolution analytical data to be obtained from suitably thin samples [3] and it is not obvious therefore whether an increased accelerating voltage or field emission is the better choice - if both are not available simultaneously. There appears to be agreement that the increased voltage available allows electron-energyloss spectroscopy to be used for reasonably thick samples, and work on Ti-based metal matrix corn-
posites is presented in this paper which confirms this expectation. X-ray data is presented which is taken to mean that any increase in spatial resolution is not useful.
2. Experiment Samples of silicon have been used to determine the influence of accelerating voltage and sample thickness on beam spreading. The technique used has been described at the earlier Cornell conference and elsewhere [4,5] and consists simply of using gold bars on the bottom of Si specimens, of measured thickness, to detect where electrons exit from the bottom surface of the Si. The technique is illustrated in fig. 1 and it is clear that the intensity of the signal from the gold bars, when the probe is positioned in the gap between the bars, is a measure of the beam spreading, so that measurements taken from the same areas can be used directly to compare the beam spreading as a function of thickness at different voltages. The relative thickness of the areas which have been examined has been determined from the
0304-3991/89/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
M.H. Loretto / Medium-voltage EDX and EELS
electron beam 1-----~ beamscan Si sample I I
~_~Au
303
3. Results 3.1. Beam spreading
bars ~--
Aox'F 7 \ / count [
~
Distance Fig. 1. Schematic diagram illustrating the principle of the technique used to compare beam spreading at 200 and at 400 kV. The Si sample has Au bars on the electron exit surface, and the electron probe is scanned over the top surface. The rate of change of the intensity of the Au X-ray signal as the probe is scanned across the edge of the Au, shown schematically above, will be different for the two voltages only if the extent of beam spreading is significantly different. Traces obtained from regions of different thicknesses at the two voltages therefore allow the influence of beam spreading on the spatial resolution to be determined as a function of voltage and of thickness.
Fig. 2 shows X - r a y line traces t a k e n across an a r e a o f thickness 3200 A, o b t a i n e d at 200 a n d 400 kV. This figure shows the A u X - r a y c o u n t p l o t t e d a g a i n s t d i s t a n c e across the s a m p l e a n d it is clear that w i t h i n the e x p e r i m e n t a l a c c u r a c y t h e r e is n o m e a s u r a b l e difference in the a m o u n t o f b e a m spreading. T h e d a t a for the two voltages s u p e r i m p o s e at this thickness a n d a n y difference w h i c h w o u l d b e e x p e c t e d f r o m the r e d u c e d h i g h - a n g l e elastic scattering at 400 k V a r e lost in t h e e x p e r i m e n t a l errors. C o u n t i n g times o f four m i n u t e s were u s e d for each p o i n t a n d although, in p r i n c i p l e , it is p o s s i b l e to use the d r i f t c o r r e c t i o n p r o g r a m m e a v a i l a b l e w i t h the E D X s y s t e m (so that f r e q u e n t checks to allow for a n y d r i f t w o u l d n o l o n g e r b e necessary), a v e r y large i n c r e a s e in c o u n t i n g accur a c y w o u l d b e r e q u i r e d to d e t e c t a n y difference
3000
n Q
r a t i o s o f the Si X - r a y c o u n t rates o b t a i n e d w h e n using the s a m e e l e c t r o n - o p t i c a l c o n d i t i o n s . T h e a b s o l u t e thickness o f the thin r e g i o n o f the s a m p l e h a s then b e e n o b t a i n e d using c o n v e r g e n t - b e a m e l e c t r o n d i f f r a c t i o n [6] to c a l i b r a t e the X - r a y m e a surements. T h e E D X w o r k has b e e n c a r r i e d o u t in S T E M u s i n g a J E O L 4 0 0 0 F X i n t e r f a c e d to a L I N K A N 1 0 0 0 0 a n a l y t i c a l system. T w o L I N K E D X detectors are a v a i l a b l e o n this m i c r o s c o p e , o n e o f w h i c h is a thin w i n d o w d e t e c t o r a n d the o t h e r a c o n v e n t i o n a l B e - w i n d o w e d detector. T h e energyloss s p e c t r o m e t e r h a s an energy r e s o l u t i o n of a b o u t 2 eV at 400 kV. T h e usefulness o f electron-energy-loss spect r o s c o p y for l i g h t - e l e m e n t analysis o f r e a s o n a b l y thick s a m p l e s has b e e n d e m o n s t r a t e d b y a n a l y s i n g the c a r b o n c o n t e n t across particles o f T i C in a m e t a l m a t r i x c o m p o s i t e where, as is s h o w n in the n e x t section, e l e c t r o n d i f f r a c t i o n d a t a suggested t h a t there was a c h a n g e in s t o i c h i o m e t r y across the i n d i v i d u a l particles.
U
[]
2000
Aucount 1000[" 0 • -1000
lira • ~ _ll il
u
0 Distancefrominter/ace(,~)
e
1000
Fig. 2. Traces of the integrated Au X-ray count obtained at 200 kV (open symbols) and 400 kV (filled symbols) from a 3200 thick sample of Si, as the electron probe w a s moved towards the edge of a Au bar which was on the bottom surface of the Si specimen. The error in the positioning of the probe was about 5 nm and the number of counts define the statistical accuracy of the measurements. Within the experimental scatter there is no difference in the shape of the two traces obtained at the two voltages.
304
M.H. Loretto / Medium-voltage EDX and EELS
~ii:~iiii~ii!ii!iil
Fig. 3. Transmission electron micrograph taken at 400 kV of a particle of TiC in a matrix of Ti6AI4V. The sample has been heat-treated at 9 0 0 ° C for 5h before being thinned for exarnination.
between these two sets of data. Similar data obtained from a region of the sample which was 9000 A thick also showed no measurable differences between the traces obtained at 200 and 400 kV. It should be emphasised that at this thickness the degradation of the image of the gold,
because it is on the bottom of the foil, causes experimental problems. On the basis of the above observations it would be reasonable to infer that the beam spreading observed in silicon is detectably smaller at 400 kV than at 200 kV only for samples of thickness of, say, I gm. At such a thickness, and equivalent thicknesses of other materials, it would not generally be possible to correct X-ray analytical data, in view of the uncertainty in absorption data. Thus, although there is no doubt that an increase in voltage from 200 to 400 kV leads to reduced high-angle scattering, and therefore to reduced beam spreading, the thickness of sample at which this reduction becomes experimentally significant is greater than that at which EDX would be carried out, if the data are required to be quantified. In the earlier work [4,5], where the beam spreading in silicon was compared at 40 and 100 kV,'there were detectable differences as small as 1500 A, where absorption correction would not be dominant, so that there is a useful improvement in the spatial resolution of X-ray data between 40 and 100 kV. On the basis of the work presented in the present paper there is little practical advantage in increasing the accelerating voltage much above 100 kV for EDX work if the aim of the experiment is to obtain quantitative analytical data at the highest possible spatial resolution. It therefore follows that if such data is required then it is essential to use thin samples, and in order to improve the statisti-
Fig. 4. Transmission electron diffraction patterns taken either side of the dislocation array in the TiC particle shown in fig. 3. The diffraction pattern from the annulus shown in (a) has extra diffraction maxima whereas the pattern from the centre of the particle, shown in (b), is typical of those obtained from stoichiometric TiC.
M.H. Loretto / Medium-voltage EDX and EELS
cal accuracy, it is obviously advantageous to use a bright electron source such as a field emission gun [3]. (The obvious disadvantage of this approach is the increase in the influence of any surface films when using very thin foils.) The choice of voltage is then of secondary importance, and factors such as the sensitivity of the sample to ionisation damage, as in the case of silver bromide, or sensitivity to knock-on damage should govern t h e choice. It should perhaps be made clear that this discussion does not imply that E D X facilities should not be interfaced to medium-voltage analytical electron microscopes, since in a wide range of analytical work neither the highest spatial resolution nor accurate quantitative analysis is required, and the ability to obtain semi-quantitative E D X data from the thick samples which can be examined at medium voltage is vital in m a n y applications.
3.2. Electron-energy-loss spectroscopy and thin window E D X analysis As part of a programme on m e t a l - m a t r i x composites, work has been carried out to determine the nature of any reactions between various particulates and fibres with titanium alloy matrices [7]. In the work presented below some of the results obtained from a T i 6 A 1 4 V - T i C composite will be used to illustrate the value of medium-voltage EELS for analysis of difficult-tothin materials.
305
Fig. 3 shows a particle of TiC in a Ti 6A14V-TiC composite imaged under two-beam conditions which brings an array of dislocations inside the particle into contrast. Diffraction patterns, taken either side of the dislocation array, are shown in fig. 4, and it is clear from these that the structure either side of the dislocation array is different, extra diffraction maxima are seen in (a) taken from the annulus. Analysis of convergent-beam electron diffraction patterns taken from these two regions shows that the centre has a space group of F m 3 m and the annulus has a space group of F d 3 m [7]. It thus appears that the annulus is non-stoichiometric TiC, referred to as Ti2C in the literature [8] and that the extra diffraction maxima are due to the ordering of the carbon vacancies. This is consistent with the fact that exposure to the b e a m removes these maxima as the vacancies are randomised [7]. Measurements of the lattice parameters of the (disordered) TiC in the annulus and in the centre confirm the conclusion that there is a smaller C content in the annulus, since the parameters are 4.27(8) ,~ and 4.32(3) .~, respectively, which shows that the C contents of the annulus and the centre are about 33 at% and between 43 and 48 at% [9]. Electron-energy-loss spectra taken either side of the dislocation array are shown in fig. 5, and analysis of these using the L I N K software shows that the C content is indeed significantly higher in the centre. The absolute values obtained have been nor-
Fig. 5. Electron energy-loss spectra taken from either side of the dislocation array in the TiC particle shown in fig. 3. These spectra were obtained at 400 kV.
306
M.H. Loretto / Medium-voltage EDX and EELS
malised to give a value of 33 at%C in the annulus which then gives a value of about 44 at% in the centre - within the range suggested by the electron diffraction data. N o deconvolution has been carried out on this and similar data and it m a y well be that deconvolution and adjustment of the cross-sections used would remove the necessity of normalising the data. Thin window E D X data have also been obtained from various TiC particles, and in all cases it has been found that the relative heights of the Ti and C peaks are consistent with the view that the annulus is C-deficient. An example of two such spectra is shown in fig. 6. N o serious attempt has been made to quantify this data, since as mentioned above the role of absorptiort would be very significant and the qualitative indication is sufficient. The obvious advantage of thin window E D X over EELS, despite the difficulty in quantifying the data, lies in the fact that the difference in the Ti and C content in the annulus and in the centres of TiC particles could be detected for all particles whatever their thickness. In the case of the EELS data the presence of C could no longer be detected in thicker samples, and in some particles even the Ti edge was not detectable in the raw data.
4. Discussion
The E D X results presented in this paper are in agreement with the earlier work [2] in which
spherical particles of A 1 / G e were used to assess the extent of b e a m broadening at 100 and at 300 kV. There is, of course, no doubt that the extent of high-angle elastic scattering decreases with increase of accelerating voltage and therefore that the extent of b e a m broadening will be decreased at higher voltages, but as has been shown in earlier work [10] the change in any composition profile, associated either with increase in thickness or with decrease of voltage, is to be found in the tail of the profile, and it is here that counting statistics are at their worst. In the present technique it is essential that the gold layer be thin, since it is acting effectively as an electron detector, so that the counting statistics will be unfavourable for the assessment of the extent of any tail in the profile. It is not clear what is the origin of the disagreement between the present work and that of Bando et al, [1], who report a significant improvement through their published data. Although the EELS work has not been undertaken simply to demonstrate the advantages of medium-voltage EELS, the results show very clearly that meaningful analyses can be carried out on realistically thick samples at 400 kV. The E D X work using the thin window detector makes it manifest that these two techniques can complement each other when samples of various thicknesses must be examined, but of course there is no obvious advantage of any sort in using 400 kV for thin window EDX. In fact, with the current state of development of medium-voltage E D X we have found that the spectra obtained using the thin
Fig. 6. EDX data obtained using athin window detector (a) from the annulus and (b) from the centre of a particle of TiC. Obtained at 200 kV.
M.H. Loretto / Medium-voltage EDX and EELS
window detector, when the microscope is operated at 400 kV, are more influenced by artefacts (which lead, for example, to large changes in background heights either side of a characteristic peak) than are the spectra obtained at 200 kV. The advantage, and indeed the necessity, of computer-controlled changes of accelerating voltage are therefore obvious if the optimum use of the microscope for the various modes requires frequent changes of voltage.
Acknowledgements I would like to acknowledge SERC for partial funding of the analytical electron microscope, Professor R.E. Smallman FRS for continued support, Dr. D.G. Konitzer for allowing me to quote from joint work which is as yet unpublished, and Mr. A.J. Burbery for technical assistance.
307
References [1] Y. Bando, Y. Matsui, Y. Kitami and Y. Inomata, in: Electron Microscopy and Analysis 1986, Inst. Phys. Conf. Ser. 78, Ed. G.J. Tatleck (Inst. Phys., London-Bristol, 1986) p, 575. [2] J.A. Eades, H.L. Fraser and M.H. Loretto, Norelco Reporter (1986). [3] J. Titchmarsh and I.A. Vatter, in: Prec. Berkeley Symp. on Radiation-Induced Sensitization of Stainless Steels, Ed. D.I.R. Norris (CEGB, 1986) p. 74. [4] R. Hutchings, I.P. Jones, M.H. Loretto and R.E. Smallman, Ultramicroscopy 3 (1979) 401. [5] T. Stephenson, M.H. Loretto and I.P. Jones, Quantitative Analysis with High Spatial Resolution (Metals Society, London, 1981). [6] P.M. Kelly, A. Jostsons, R.G. Blake and J.G. Napier, Phys. Status Solidi (a) 31 (1975) 771. [7] D.G. Konitzer and M.H. Loretto, Acta Met., in press. [8] H. Goretzki, Phys. Status Solidi 20 (1967) K141. [9] W.B. Pearson, Handbook of Lattice Spacings of Metals and Alloys, Vol. 2 (Pergamon, Oxford, 1967). [10] M. Twigg, M.H. Loretto and H.L. Fraser, Phil. Mag. 43 (1981) 1587.
Ultramicroscopy 28 (1989) 302-307 North-Holland, Amsterdam
M E D I U M - V O L T A G E EDX AND EELS M.H. L O R E T T O Department of Metallurgy and Materials, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK Received at Editorial Office 3 November 1988; presented at Workshop March 1988
Experiments have been carried out in which the influence of beam broadening on the spatial resolution of X-ray analytical data has been assessed at 200 and at 400 kV. It has been found that any improvement at the higher voltage becomes significant only at sample thicknesses which axe greater than those that would be used for microanalysis, i.e. at thicknesses at which absorption corrections become dominant. Analyses of particles of TiC have been carried out using electron energy-loss spectroscopy at 400 kV and using a thin window energy-dispersive X-ray analysis facility, at an accelerating voltage of 200 kV. It has been found that the electron-energy-loss data can be quantified without any deconvolution, for reasonably thick samples; and changes in stoichiometry, identified using electron diffraction, are measurable from the electron-energy-loss data. At this stage the thin window energy-dispersive data has been found to be useful only for qualitative analysis in view of the undetermined, and presumed large, influence of absorption.
1. Introduction The potential advantages of medium-voltage analytical electron microscopy have been well catalogued but there is as yet a distinct lack of data which can be used to back up the various claims which have been made by manufacturers and users alike. Indeed there appears to be some disagreement between various workers, with some claiming significant experimental improvements in the spatial resolution of X-ray data from very thin samples [1] and others claiming that the anticipated advantage in spatial resolution is not relevant for practical situations [2]. Recent work using fidd emission scanning transmission electron microscopy has allowed high spatial resolution analytical data to be obtained from suitably thin samples [3] and it is not obvious therefore whether an increased accelerating voltage or field emission is the better choice - if both are not available simultaneously. There appears to be agreement that the increased voltage available allows electron-energyloss spectroscopy to be used for reasonably thick samples, and work on Ti-based metal matrix corn-
posites is presented in this paper which confirms this expectation. X-ray data is presented which is taken to mean that any increase in spatial resolution is not useful.
2. Experiment Samples of silicon have been used to determine the influence of accelerating voltage and sample thickness on beam spreading. The technique used has been described at the earlier Cornell conference and elsewhere [4,5] and consists simply of using gold bars on the bottom of Si specimens, of measured thickness, to detect where electrons exit from the bottom surface of the Si. The technique is illustrated in fig. 1 and it is clear that the intensity of the signal from the gold bars, when the probe is positioned in the gap between the bars, is a measure of the beam spreading, so that measurements taken from the same areas can be used directly to compare the beam spreading as a function of thickness at different voltages. The relative thickness of the areas which have been examined has been determined from the
0304-3991/89/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
M.H. Loretto / Medium-voltage EDX and EELS
electron beam 1-----~ beamscan Si sample I I
~_~Au
303
3. Results 3.1. Beam spreading
bars ~--
Aox'F 7 \ / count [
~
Distance Fig. 1. Schematic diagram illustrating the principle of the technique used to compare beam spreading at 200 and at 400 kV. The Si sample has Au bars on the electron exit surface, and the electron probe is scanned over the top surface. The rate of change of the intensity of the Au X-ray signal as the probe is scanned across the edge of the Au, shown schematically above, will be different for the two voltages only if the extent of beam spreading is significantly different. Traces obtained from regions of different thicknesses at the two voltages therefore allow the influence of beam spreading on the spatial resolution to be determined as a function of voltage and of thickness.
Fig. 2 shows X - r a y line traces t a k e n across an a r e a o f thickness 3200 A, o b t a i n e d at 200 a n d 400 kV. This figure shows the A u X - r a y c o u n t p l o t t e d a g a i n s t d i s t a n c e across the s a m p l e a n d it is clear that w i t h i n the e x p e r i m e n t a l a c c u r a c y t h e r e is n o m e a s u r a b l e difference in the a m o u n t o f b e a m spreading. T h e d a t a for the two voltages s u p e r i m p o s e at this thickness a n d a n y difference w h i c h w o u l d b e e x p e c t e d f r o m the r e d u c e d h i g h - a n g l e elastic scattering at 400 k V a r e lost in t h e e x p e r i m e n t a l errors. C o u n t i n g times o f four m i n u t e s were u s e d for each p o i n t a n d although, in p r i n c i p l e , it is p o s s i b l e to use the d r i f t c o r r e c t i o n p r o g r a m m e a v a i l a b l e w i t h the E D X s y s t e m (so that f r e q u e n t checks to allow for a n y d r i f t w o u l d n o l o n g e r b e necessary), a v e r y large i n c r e a s e in c o u n t i n g accur a c y w o u l d b e r e q u i r e d to d e t e c t a n y difference
3000
n Q
r a t i o s o f the Si X - r a y c o u n t rates o b t a i n e d w h e n using the s a m e e l e c t r o n - o p t i c a l c o n d i t i o n s . T h e a b s o l u t e thickness o f the thin r e g i o n o f the s a m p l e h a s then b e e n o b t a i n e d using c o n v e r g e n t - b e a m e l e c t r o n d i f f r a c t i o n [6] to c a l i b r a t e the X - r a y m e a surements. T h e E D X w o r k has b e e n c a r r i e d o u t in S T E M u s i n g a J E O L 4 0 0 0 F X i n t e r f a c e d to a L I N K A N 1 0 0 0 0 a n a l y t i c a l system. T w o L I N K E D X detectors are a v a i l a b l e o n this m i c r o s c o p e , o n e o f w h i c h is a thin w i n d o w d e t e c t o r a n d the o t h e r a c o n v e n t i o n a l B e - w i n d o w e d detector. T h e energyloss s p e c t r o m e t e r h a s an energy r e s o l u t i o n of a b o u t 2 eV at 400 kV. T h e usefulness o f electron-energy-loss spect r o s c o p y for l i g h t - e l e m e n t analysis o f r e a s o n a b l y thick s a m p l e s has b e e n d e m o n s t r a t e d b y a n a l y s i n g the c a r b o n c o n t e n t across particles o f T i C in a m e t a l m a t r i x c o m p o s i t e where, as is s h o w n in the n e x t section, e l e c t r o n d i f f r a c t i o n d a t a suggested t h a t there was a c h a n g e in s t o i c h i o m e t r y across the i n d i v i d u a l particles.
U
[]
2000
Aucount 1000[" 0 • -1000
lira • ~ _ll il
u
0 Distancefrominter/ace(,~)
e
1000
Fig. 2. Traces of the integrated Au X-ray count obtained at 200 kV (open symbols) and 400 kV (filled symbols) from a 3200 thick sample of Si, as the electron probe w a s moved towards the edge of a Au bar which was on the bottom surface of the Si specimen. The error in the positioning of the probe was about 5 nm and the number of counts define the statistical accuracy of the measurements. Within the experimental scatter there is no difference in the shape of the two traces obtained at the two voltages.
304
M.H. Loretto / Medium-voltage EDX and EELS
~ii:~iiii~ii!ii!iil
Fig. 3. Transmission electron micrograph taken at 400 kV of a particle of TiC in a matrix of Ti6AI4V. The sample has been heat-treated at 9 0 0 ° C for 5h before being thinned for exarnination.
between these two sets of data. Similar data obtained from a region of the sample which was 9000 A thick also showed no measurable differences between the traces obtained at 200 and 400 kV. It should be emphasised that at this thickness the degradation of the image of the gold,
because it is on the bottom of the foil, causes experimental problems. On the basis of the above observations it would be reasonable to infer that the beam spreading observed in silicon is detectably smaller at 400 kV than at 200 kV only for samples of thickness of, say, I gm. At such a thickness, and equivalent thicknesses of other materials, it would not generally be possible to correct X-ray analytical data, in view of the uncertainty in absorption data. Thus, although there is no doubt that an increase in voltage from 200 to 400 kV leads to reduced high-angle scattering, and therefore to reduced beam spreading, the thickness of sample at which this reduction becomes experimentally significant is greater than that at which EDX would be carried out, if the data are required to be quantified. In the earlier work [4,5], where the beam spreading in silicon was compared at 40 and 100 kV,'there were detectable differences as small as 1500 A, where absorption correction would not be dominant, so that there is a useful improvement in the spatial resolution of X-ray data between 40 and 100 kV. On the basis of the work presented in the present paper there is little practical advantage in increasing the accelerating voltage much above 100 kV for EDX work if the aim of the experiment is to obtain quantitative analytical data at the highest possible spatial resolution. It therefore follows that if such data is required then it is essential to use thin samples, and in order to improve the statisti-
Fig. 4. Transmission electron diffraction patterns taken either side of the dislocation array in the TiC particle shown in fig. 3. The diffraction pattern from the annulus shown in (a) has extra diffraction maxima whereas the pattern from the centre of the particle, shown in (b), is typical of those obtained from stoichiometric TiC.
M.H. Loretto / Medium-voltage EDX and EELS
cal accuracy, it is obviously advantageous to use a bright electron source such as a field emission gun [3]. (The obvious disadvantage of this approach is the increase in the influence of any surface films when using very thin foils.) The choice of voltage is then of secondary importance, and factors such as the sensitivity of the sample to ionisation damage, as in the case of silver bromide, or sensitivity to knock-on damage should govern t h e choice. It should perhaps be made clear that this discussion does not imply that E D X facilities should not be interfaced to medium-voltage analytical electron microscopes, since in a wide range of analytical work neither the highest spatial resolution nor accurate quantitative analysis is required, and the ability to obtain semi-quantitative E D X data from the thick samples which can be examined at medium voltage is vital in m a n y applications.
3.2. Electron-energy-loss spectroscopy and thin window E D X analysis As part of a programme on m e t a l - m a t r i x composites, work has been carried out to determine the nature of any reactions between various particulates and fibres with titanium alloy matrices [7]. In the work presented below some of the results obtained from a T i 6 A 1 4 V - T i C composite will be used to illustrate the value of medium-voltage EELS for analysis of difficult-tothin materials.
305
Fig. 3 shows a particle of TiC in a Ti 6A14V-TiC composite imaged under two-beam conditions which brings an array of dislocations inside the particle into contrast. Diffraction patterns, taken either side of the dislocation array, are shown in fig. 4, and it is clear from these that the structure either side of the dislocation array is different, extra diffraction maxima are seen in (a) taken from the annulus. Analysis of convergent-beam electron diffraction patterns taken from these two regions shows that the centre has a space group of F m 3 m and the annulus has a space group of F d 3 m [7]. It thus appears that the annulus is non-stoichiometric TiC, referred to as Ti2C in the literature [8] and that the extra diffraction maxima are due to the ordering of the carbon vacancies. This is consistent with the fact that exposure to the b e a m removes these maxima as the vacancies are randomised [7]. Measurements of the lattice parameters of the (disordered) TiC in the annulus and in the centre confirm the conclusion that there is a smaller C content in the annulus, since the parameters are 4.27(8) ,~ and 4.32(3) .~, respectively, which shows that the C contents of the annulus and the centre are about 33 at% and between 43 and 48 at% [9]. Electron-energy-loss spectra taken either side of the dislocation array are shown in fig. 5, and analysis of these using the L I N K software shows that the C content is indeed significantly higher in the centre. The absolute values obtained have been nor-
Fig. 5. Electron energy-loss spectra taken from either side of the dislocation array in the TiC particle shown in fig. 3. These spectra were obtained at 400 kV.
306
M.H. Loretto / Medium-voltage EDX and EELS
malised to give a value of 33 at%C in the annulus which then gives a value of about 44 at% in the centre - within the range suggested by the electron diffraction data. N o deconvolution has been carried out on this and similar data and it m a y well be that deconvolution and adjustment of the cross-sections used would remove the necessity of normalising the data. Thin window E D X data have also been obtained from various TiC particles, and in all cases it has been found that the relative heights of the Ti and C peaks are consistent with the view that the annulus is C-deficient. An example of two such spectra is shown in fig. 6. N o serious attempt has been made to quantify this data, since as mentioned above the role of absorptiort would be very significant and the qualitative indication is sufficient. The obvious advantage of thin window E D X over EELS, despite the difficulty in quantifying the data, lies in the fact that the difference in the Ti and C content in the annulus and in the centres of TiC particles could be detected for all particles whatever their thickness. In the case of the EELS data the presence of C could no longer be detected in thicker samples, and in some particles even the Ti edge was not detectable in the raw data.
4. Discussion
The E D X results presented in this paper are in agreement with the earlier work [2] in which
spherical particles of A 1 / G e were used to assess the extent of b e a m broadening at 100 and at 300 kV. There is, of course, no doubt that the extent of high-angle elastic scattering decreases with increase of accelerating voltage and therefore that the extent of b e a m broadening will be decreased at higher voltages, but as has been shown in earlier work [10] the change in any composition profile, associated either with increase in thickness or with decrease of voltage, is to be found in the tail of the profile, and it is here that counting statistics are at their worst. In the present technique it is essential that the gold layer be thin, since it is acting effectively as an electron detector, so that the counting statistics will be unfavourable for the assessment of the extent of any tail in the profile. It is not clear what is the origin of the disagreement between the present work and that of Bando et al, [1], who report a significant improvement through their published data. Although the EELS work has not been undertaken simply to demonstrate the advantages of medium-voltage EELS, the results show very clearly that meaningful analyses can be carried out on realistically thick samples at 400 kV. The E D X work using the thin window detector makes it manifest that these two techniques can complement each other when samples of various thicknesses must be examined, but of course there is no obvious advantage of any sort in using 400 kV for thin window EDX. In fact, with the current state of development of medium-voltage E D X we have found that the spectra obtained using the thin
Fig. 6. EDX data obtained using athin window detector (a) from the annulus and (b) from the centre of a particle of TiC. Obtained at 200 kV.
M.H. Loretto / Medium-voltage EDX and EELS
window detector, when the microscope is operated at 400 kV, are more influenced by artefacts (which lead, for example, to large changes in background heights either side of a characteristic peak) than are the spectra obtained at 200 kV. The advantage, and indeed the necessity, of computer-controlled changes of accelerating voltage are therefore obvious if the optimum use of the microscope for the various modes requires frequent changes of voltage.
Acknowledgements I would like to acknowledge SERC for partial funding of the analytical electron microscope, Professor R.E. Smallman FRS for continued support, Dr. D.G. Konitzer for allowing me to quote from joint work which is as yet unpublished, and Mr. A.J. Burbery for technical assistance.
307
References [1] Y. Bando, Y. Matsui, Y. Kitami and Y. Inomata, in: Electron Microscopy and Analysis 1986, Inst. Phys. Conf. Ser. 78, Ed. G.J. Tatleck (Inst. Phys., London-Bristol, 1986) p, 575. [2] J.A. Eades, H.L. Fraser and M.H. Loretto, Norelco Reporter (1986). [3] J. Titchmarsh and I.A. Vatter, in: Prec. Berkeley Symp. on Radiation-Induced Sensitization of Stainless Steels, Ed. D.I.R. Norris (CEGB, 1986) p. 74. [4] R. Hutchings, I.P. Jones, M.H. Loretto and R.E. Smallman, Ultramicroscopy 3 (1979) 401. [5] T. Stephenson, M.H. Loretto and I.P. Jones, Quantitative Analysis with High Spatial Resolution (Metals Society, London, 1981). [6] P.M. Kelly, A. Jostsons, R.G. Blake and J.G. Napier, Phys. Status Solidi (a) 31 (1975) 771. [7] D.G. Konitzer and M.H. Loretto, Acta Met., in press. [8] H. Goretzki, Phys. Status Solidi 20 (1967) K141. [9] W.B. Pearson, Handbook of Lattice Spacings of Metals and Alloys, Vol. 2 (Pergamon, Oxford, 1967). [10] M. Twigg, M.H. Loretto and H.L. Fraser, Phil. Mag. 43 (1981) 1587.