- Email: [email protected]
An X-ray filter assembly for fluorescence EXAFS measurements
Nuclear Instruments and Methods 172 (1980) 397-399 © North-Holland Publishing Company
AN X-RAY FILTER ASSEMBLY FOR FLUORESCENCE EXAFS MEASUREMENTS Edward A. STERN and Steve M. HEALD
Physics Department, Universityof Washington,Seattle, WA 98195, U.S.A. Fluorescence detection, in principle, permits the detection of the extended X-ray absorption fine structure (EXAFS) of more dilute atoms than can be obtained in absorption. To take advantage of this it is necessary, in practice, to eliminate the background that normally accompanies the fluorescence signal. We describe an X-ray filter assembly that accomplishes this purpose. The unique characteristic of the assembly is a slit system that minimizes the fluorescence background from the filter. The theory of the slit assembly is presented and is found to agree with measurements made on the Fe EXAFS of a dilute sample. The filter assembly has a better effective counting rate in this case than that of a crystal monochromator design. 1. Introduction
scatttered radiation. Let the signal consist of N0 fluorescent and Nb background counts per second. The signal to noise ratio S/N is given by
One of the directions presently being pushed to take full advantage of the measurements being opened up by synchrotron radiation sources is to detect the presence of dilute amounts of one type of atom imbedded among other atoms. Examples of such cases are measuring the extended X-ray absorption fine structure (EXAFS) of the metal atom in metalloproteins and measuring the EXAFS of dilute impurities in solid and liquid solutions. A solution proposed to this detection problem is to observe the fluorescence radiation from the dilute atom [1,2]. The fluorescence emission is characteristic of the atom of interest and by tuning the detector to this characteristic energy it is, in principle, possible to eliminate all background. In detecting a dilute signal it is necessary to eliminate as much as possible the large background [1] and to have as large an efficiency as possible in detecting the desired signal. We describe a simple solution to eliminate the background before reaching the detector, employing X-ray filters [3]. The use of X-ray filters has been suggested previously [2] b u t did not give significant improvement for the application at hand. By a suitable design improvement on the X-ray filters we have been able to measure the fluorescence EXAFS of a dilute sample with a better signal to noise ratio than has been done before, including by a crystal monochromator [4].
S/N= ( No'c 11/2 \1 +A] '
(1)
where A =Nb/N o and 7- is the time of measurement. Note that the SIN has the same value as if no background were present and the signal No is reduced by the factor (1 + A ) -1 . This effective signal Ne is the measure we will use to evaluate the effectiveness of a given detector system. Thus, eliminating the background has the same effect as increasing the signal by a factor (1 + A ) . In the example discussed here A 20 and one can get 21-fold increase in effective signal N e by background elimination. Placing an appropriate filter [3] between the sample and the detector can selectively decrease the
Table 1 The effective counting rate Ne of the filter assembly under various filter conditions when incoming background to signal ratio A = 20 and the ratio of the attenuation coefficient for signal to that of the background isf = 1/8. ~ is the fraction of the incoming background radiation that induces a background fluorescent signal from the filter and (/~X)l is the optimum product of the attenuation coefficient and thickness at the background energy for that
2. Design considerations When a sample is irradiated by X-rays its fluorescent signal is mixed with elastically and inelastically 397
Case
n
(gx)t
Ne
1. No filter 2. Perfect filter 3. Real filter 4. Real filter + Soiler slits
0 0.15 0.019
5.47 3.56 4.87
N1 9.1N 1 2.6N 1 6.4N 1
VII. DETECTORS AND OTHER SUBJECTS
E.A. Stern, S.M. Heald / X-ray filter assembly
398
background relative to the signal. Since it is important to maximize No by bringing the detector as close as possible to the sample to maximize its subtended solid angle, the filter is required to be in close proximity to the detector. The scattered background will, in general, be energetic enough to excite fluorescence radiation in the filter. Such fluorescence radiation will be essentially as weakly attenuated as the desired fluorescent signal and thus will not be discriminated against by the filter. For a filter of thickness x with an absorption coefficient/J for the background and ft~ for the signal, where f < 1, and including the contribution of the fluorescent photons from the filter, it can be shown that the effective counting rate becomes Ne ~
No
sible to the filter fluorescent radiation can be significantly decreased, correspondingly decreasing the background fluorescence. The slit assembly can accommodate both line and point sources as obtained in the non,-focused and focused beam lines at the Stanford Synchrotron Radiation Source (SSRL), respectively. In the design we describe below, 77 = 0.019, and its effectiveness is tabulated as case 4 in the table. As can be noted, the effective counting rate is significantly improved from 2.6 without the slits to 6.4 with the slits.
3. Filter assembly Fig. l(b) shows a filter assembly for a line focused X-ray beam built along the lines described above. The solid angle that can be subtended by a detector is 1/8 of 47r measured from the center of the sample. The slit consists of 15 sections of planes, each of which would, if extended, pass through the line of intersection of the X-ray beam and the sample surface. The sample is oriented at a 45 ° angle to the beam direction. Each section of the slit has a projected width on the horizontal plane of 1/2" and the sections are evenly spaced in angle about the pivot line. The filter assembly was used with a scintillation detector to measure the fluorescence of the Fe
EA(l+(f ~1))e-#X(1-2"D+ (1 (fA~77i-~)e[#X 1' (2)
where 77 is the fraction of the elastically scattered background that is converted to a background fluorescence signal from the filter. The optimum Are is obtained at the value
1 ( b G ) l - (1 _ f ) l n
f(1-A
f ~ 1)
(3)
In table 1 we have tabulated several cases appropriate for the conditions A = 20 and f = 1/8. In case 1 no filter is used and the effective counting rate is N~ =No/21. In case 2 we present the result for the ideal filter with 77=0, no fluorescent background from the filter. In case 3 we give the result for a real filter made of Mn metal where the 7/= 0.3/2 = 0.15. We note that the filter fluorescence causes a serious deterioration in the effective counting rate, decreasing the improvement factor from 9.1 to only 2.6. Fortunately we are able to minimize the deteriorating influence of the fluorescent background by taking advantage of its non-directional character and placing a SoUer-type slit assembly between the filter and the detector. To understand how this works consider fig. l(a) which shows a schematic cross section of the X-ray paths. The slit assembly consists of X-ray opaque thin strips whose planes intercept at the signal source. Such a configuration does not appreciably attenuate the signal but the detector solid angle acces-
~
SLITS fFILTER
• DETECTOR M'~--,-@ ~
...
~
"~
(a)
Fig. l(a). The geometry of the slit system to selectively decrease the background fluorescence signal from the filter. The fluorescent X-rays from the source are not attenuated by the slits but the fluorescent X-rays from the filter are paxtially cut off by them. Only the X-rays between the solid lines eminating from the filter can reach the detector. The 4~r distribution of the filter fluorescence is schematically indicated by the circle centered on the filter, and that radiation not accessible to the detector is indicated by the radial short dashed lines.
E.A. Stern, S.M. Heald / X-ray filter assembly X - R A Y S FROM SYNCHROTRON RABIATION
SHIELD~4 FI LTER SLITS-.......
SAMPLE FACE
/
OF SAM PL.E .,,,,,-__ i" ..__~
{b) Fig. l(b). The dimensions of the filter assembly built to test its feasibility. The opening in the shield is a square 1.50" on a side. K-edge EXAFS in a bacterial photosynthetic reaction center where the Fe concentration is about 1 per 104 low Z atoms. Under the conditions used at SSRL and after passing through the filter assembly the counting rate was No = 20 000/s and A = 1, equivalent to an effectively N e = 10 000/s. The filter had a px = 4.2 and the background consisted o f about equal amounts o f fluorescence from the filter and scattered radiation that passed through the filter. The slits decrease the filter fluorescence about a factor of 8 from its intensity in their absence. Without any filter before the detector the counting rate would be No = 34 000/s and A = 20, equivalent to an effective Ne = 1600/s. The improvement in Ne agrees with the theoretical prediction of case 4 in the table. However, the improvement by the filter assembly is even greater because the detector is limited to a total counting rate of 10S/s before non-linearities due to
399
dead time set in. In the case at h a n d , w i t h o u t a filter the detector would have to be moved away from the sample to decrease the subtended solid angle and total counting rate to below l 0 s , decreasing the effective Are to 230. Thus the filter assembly, in practice, increases the effective counting rate in a scintillation detector from 230 to 10 000/s, an increase o f a factor of over 40! Under the same operating conditions, a crystal m o n o c h r o m a t o r [4] subtending 1.5% o f 47r with a 50% diffraction efficiency would produce a counting rate No = Are = 2400/s with no background. We note that the filter assembly's ATe is about 4 times greater. The filter assembly more than compensates for its poorer background discrimination b y the larger solid angle that it subtends. The slit assembly can be made even more effective if the source is a point focus. Then the slit assembly can consist o f two sets o f perpendicular intersecting planes, each set focusing to one of two perpendicular lines intersecting at the point focus. One o f us (EAS) is indebted to a stimulating conversation with Dr. David F. Anderson which sparked the line o f thought leading to the design presented here. Our thanks are also due to Dave Garcia and Don Russell for machining the slit assembly. SSRL is supported by the National Science Foundation in cooperation with the Department o f Energy. This work was supported in part b y a research grant from the National Science Foundation.
References [1] J. Jaklevic, J.A. Kirby, M.P. Klein, A.S. Robertson, G.S. Brown and P. Eisenberger, Solid State Commun. 23 (1977) 679. [2] F.S. Goulding, J.M. Jaklevic and A.C. Thompson, Workshop on X-ray instrumentation for synchrotron radiation research, SSRL Report No. 78/04, May 1978, edited by H. Winick and G. Brown. [31 See, for example, A. Taylor, X-ray metallography (Wiley, New York, 1961) pp. 28-29. [4 ] J.B. Hastings, M.L. Perlman, P. Oversluizen, P. Eisenberger and J. Brown, 5th Annual SSRL Users Group Meeting, Oct. 1978, SSRL Report No. 78-09.
VlI. DETECTORS AND OTHER SUBJECTS
AN X-RAY FILTER ASSEMBLY FOR FLUORESCENCE EXAFS MEASUREMENTS Edward A. STERN and Steve M. HEALD
Physics Department, Universityof Washington,Seattle, WA 98195, U.S.A. Fluorescence detection, in principle, permits the detection of the extended X-ray absorption fine structure (EXAFS) of more dilute atoms than can be obtained in absorption. To take advantage of this it is necessary, in practice, to eliminate the background that normally accompanies the fluorescence signal. We describe an X-ray filter assembly that accomplishes this purpose. The unique characteristic of the assembly is a slit system that minimizes the fluorescence background from the filter. The theory of the slit assembly is presented and is found to agree with measurements made on the Fe EXAFS of a dilute sample. The filter assembly has a better effective counting rate in this case than that of a crystal monochromator design. 1. Introduction
scatttered radiation. Let the signal consist of N0 fluorescent and Nb background counts per second. The signal to noise ratio S/N is given by
One of the directions presently being pushed to take full advantage of the measurements being opened up by synchrotron radiation sources is to detect the presence of dilute amounts of one type of atom imbedded among other atoms. Examples of such cases are measuring the extended X-ray absorption fine structure (EXAFS) of the metal atom in metalloproteins and measuring the EXAFS of dilute impurities in solid and liquid solutions. A solution proposed to this detection problem is to observe the fluorescence radiation from the dilute atom [1,2]. The fluorescence emission is characteristic of the atom of interest and by tuning the detector to this characteristic energy it is, in principle, possible to eliminate all background. In detecting a dilute signal it is necessary to eliminate as much as possible the large background [1] and to have as large an efficiency as possible in detecting the desired signal. We describe a simple solution to eliminate the background before reaching the detector, employing X-ray filters [3]. The use of X-ray filters has been suggested previously [2] b u t did not give significant improvement for the application at hand. By a suitable design improvement on the X-ray filters we have been able to measure the fluorescence EXAFS of a dilute sample with a better signal to noise ratio than has been done before, including by a crystal monochromator [4].
S/N= ( No'c 11/2 \1 +A] '
(1)
where A =Nb/N o and 7- is the time of measurement. Note that the SIN has the same value as if no background were present and the signal No is reduced by the factor (1 + A ) -1 . This effective signal Ne is the measure we will use to evaluate the effectiveness of a given detector system. Thus, eliminating the background has the same effect as increasing the signal by a factor (1 + A ) . In the example discussed here A 20 and one can get 21-fold increase in effective signal N e by background elimination. Placing an appropriate filter [3] between the sample and the detector can selectively decrease the
Table 1 The effective counting rate Ne of the filter assembly under various filter conditions when incoming background to signal ratio A = 20 and the ratio of the attenuation coefficient for signal to that of the background isf = 1/8. ~ is the fraction of the incoming background radiation that induces a background fluorescent signal from the filter and (/~X)l is the optimum product of the attenuation coefficient and thickness at the background energy for that
2. Design considerations When a sample is irradiated by X-rays its fluorescent signal is mixed with elastically and inelastically 397
Case
n
(gx)t
Ne
1. No filter 2. Perfect filter 3. Real filter 4. Real filter + Soiler slits
0 0.15 0.019
5.47 3.56 4.87
N1 9.1N 1 2.6N 1 6.4N 1
VII. DETECTORS AND OTHER SUBJECTS
E.A. Stern, S.M. Heald / X-ray filter assembly
398
background relative to the signal. Since it is important to maximize No by bringing the detector as close as possible to the sample to maximize its subtended solid angle, the filter is required to be in close proximity to the detector. The scattered background will, in general, be energetic enough to excite fluorescence radiation in the filter. Such fluorescence radiation will be essentially as weakly attenuated as the desired fluorescent signal and thus will not be discriminated against by the filter. For a filter of thickness x with an absorption coefficient/J for the background and ft~ for the signal, where f < 1, and including the contribution of the fluorescent photons from the filter, it can be shown that the effective counting rate becomes Ne ~
No
sible to the filter fluorescent radiation can be significantly decreased, correspondingly decreasing the background fluorescence. The slit assembly can accommodate both line and point sources as obtained in the non,-focused and focused beam lines at the Stanford Synchrotron Radiation Source (SSRL), respectively. In the design we describe below, 77 = 0.019, and its effectiveness is tabulated as case 4 in the table. As can be noted, the effective counting rate is significantly improved from 2.6 without the slits to 6.4 with the slits.
3. Filter assembly Fig. l(b) shows a filter assembly for a line focused X-ray beam built along the lines described above. The solid angle that can be subtended by a detector is 1/8 of 47r measured from the center of the sample. The slit consists of 15 sections of planes, each of which would, if extended, pass through the line of intersection of the X-ray beam and the sample surface. The sample is oriented at a 45 ° angle to the beam direction. Each section of the slit has a projected width on the horizontal plane of 1/2" and the sections are evenly spaced in angle about the pivot line. The filter assembly was used with a scintillation detector to measure the fluorescence of the Fe
EA(l+(f ~1))e-#X(1-2"D+ (1 (fA~77i-~)e[#X 1' (2)
where 77 is the fraction of the elastically scattered background that is converted to a background fluorescence signal from the filter. The optimum Are is obtained at the value
1 ( b G ) l - (1 _ f ) l n
f(1-A
f ~ 1)
(3)
In table 1 we have tabulated several cases appropriate for the conditions A = 20 and f = 1/8. In case 1 no filter is used and the effective counting rate is N~ =No/21. In case 2 we present the result for the ideal filter with 77=0, no fluorescent background from the filter. In case 3 we give the result for a real filter made of Mn metal where the 7/= 0.3/2 = 0.15. We note that the filter fluorescence causes a serious deterioration in the effective counting rate, decreasing the improvement factor from 9.1 to only 2.6. Fortunately we are able to minimize the deteriorating influence of the fluorescent background by taking advantage of its non-directional character and placing a SoUer-type slit assembly between the filter and the detector. To understand how this works consider fig. l(a) which shows a schematic cross section of the X-ray paths. The slit assembly consists of X-ray opaque thin strips whose planes intercept at the signal source. Such a configuration does not appreciably attenuate the signal but the detector solid angle acces-
~
SLITS fFILTER
• DETECTOR M'~--,-@ ~
...
~
"~
(a)
Fig. l(a). The geometry of the slit system to selectively decrease the background fluorescence signal from the filter. The fluorescent X-rays from the source are not attenuated by the slits but the fluorescent X-rays from the filter are paxtially cut off by them. Only the X-rays between the solid lines eminating from the filter can reach the detector. The 4~r distribution of the filter fluorescence is schematically indicated by the circle centered on the filter, and that radiation not accessible to the detector is indicated by the radial short dashed lines.
E.A. Stern, S.M. Heald / X-ray filter assembly X - R A Y S FROM SYNCHROTRON RABIATION
SHIELD~4 FI LTER SLITS-.......
SAMPLE FACE
/
OF SAM PL.E .,,,,,-__ i" ..__~
{b) Fig. l(b). The dimensions of the filter assembly built to test its feasibility. The opening in the shield is a square 1.50" on a side. K-edge EXAFS in a bacterial photosynthetic reaction center where the Fe concentration is about 1 per 104 low Z atoms. Under the conditions used at SSRL and after passing through the filter assembly the counting rate was No = 20 000/s and A = 1, equivalent to an effectively N e = 10 000/s. The filter had a px = 4.2 and the background consisted o f about equal amounts o f fluorescence from the filter and scattered radiation that passed through the filter. The slits decrease the filter fluorescence about a factor of 8 from its intensity in their absence. Without any filter before the detector the counting rate would be No = 34 000/s and A = 20, equivalent to an effective Ne = 1600/s. The improvement in Ne agrees with the theoretical prediction of case 4 in the table. However, the improvement by the filter assembly is even greater because the detector is limited to a total counting rate of 10S/s before non-linearities due to
399
dead time set in. In the case at h a n d , w i t h o u t a filter the detector would have to be moved away from the sample to decrease the subtended solid angle and total counting rate to below l 0 s , decreasing the effective Are to 230. Thus the filter assembly, in practice, increases the effective counting rate in a scintillation detector from 230 to 10 000/s, an increase o f a factor of over 40! Under the same operating conditions, a crystal m o n o c h r o m a t o r [4] subtending 1.5% o f 47r with a 50% diffraction efficiency would produce a counting rate No = Are = 2400/s with no background. We note that the filter assembly's ATe is about 4 times greater. The filter assembly more than compensates for its poorer background discrimination b y the larger solid angle that it subtends. The slit assembly can be made even more effective if the source is a point focus. Then the slit assembly can consist o f two sets o f perpendicular intersecting planes, each set focusing to one of two perpendicular lines intersecting at the point focus. One o f us (EAS) is indebted to a stimulating conversation with Dr. David F. Anderson which sparked the line o f thought leading to the design presented here. Our thanks are also due to Dave Garcia and Don Russell for machining the slit assembly. SSRL is supported by the National Science Foundation in cooperation with the Department o f Energy. This work was supported in part b y a research grant from the National Science Foundation.
References [1] J. Jaklevic, J.A. Kirby, M.P. Klein, A.S. Robertson, G.S. Brown and P. Eisenberger, Solid State Commun. 23 (1977) 679. [2] F.S. Goulding, J.M. Jaklevic and A.C. Thompson, Workshop on X-ray instrumentation for synchrotron radiation research, SSRL Report No. 78/04, May 1978, edited by H. Winick and G. Brown. [31 See, for example, A. Taylor, X-ray metallography (Wiley, New York, 1961) pp. 28-29. [4 ] J.B. Hastings, M.L. Perlman, P. Oversluizen, P. Eisenberger and J. Brown, 5th Annual SSRL Users Group Meeting, Oct. 1978, SSRL Report No. 78-09.
VlI. DETECTORS AND OTHER SUBJECTS