1. Giới thiệuPhân tích kích hoạt gamma Prompt (PGAA) là mộtphương pháp phát triển nhanh chóng trong phân tích nguyên tố(Molnar và cộng sự, 1993, Moln Ar và Lindstrom, 1998). Nólà không phá hủy và đòi hỏi một lượng nhỏChuẩn bị mẫu. PGAA có thể được thực hiện tối ưutrên một chùm neutron có hướng dẫn cung cấp tuyệt vời, thấpđiều kiện nền cho các thí nghiệm trong chùm. Trongmặc dù điều này, bản chất của kết quả phản ứng (n. g)một nhiễu quang phổ phức tạp, và một tích lũynền liên tục do Compton tán xạ trongmáy dò. Do đó phổ PGAA cho các mẫu thựcphức tạp hơn nhiều so với quang phổ của InstruPhân tích kích hoạt neutron thần kinh (Ehmann vàVance, 1991a, b) (INAA). Các detec được sử dụng rộng rãi nhấttion cho PGAA là một Compton bị đàn ápquang phổ (Belgya, 1996), làm giảm đáng kểkích thước của liên tục và các đỉnh núi thoát hiểm.Phát hiện coincident của tia gamma (g2g coincidence) là một kỹ thuật được thành lập trong lĩnh vực hạt nhânnghiên cứu cấu trúc (Wapstra, 1979). Cũng được biết rằngcác phương pháp trùng hợp ngẫu nhiên có thể làm giảm đáng kể giao thoaence, và do đó cũng là sự phức tạp của quang phổ. Cácích lợi của kỹ thuật trùng hợp g2g trong các nguyên tốphân tích (Ehmann và Vance, 1991) đã đượcđã chứng minh trong trường hợp INAA của một sốcác tác giả (Meyer, 1987, Meyer và cộng sự, 1993, Jakubek et al,1998; Koeberl và Huber, 2000). Họ tìm thấy nóđặc biệt hữu ích trong việc xác định Ir và Se về địa chấtmẫu ở mức ppb. Mục tiêu chính của họ là để cải thiệnđộ nhạy của phương pháp bằng cách ngăn chặnnền liên tục và giảm liên tục quang phổference.Phương pháp thông thường để thực hiện sự trùng hợp ngẫu nhiên g2glà yêu cầu một mối quan hệ ngẫu nhiên với một lựa chọn đầy đủđỉnh cao năng lượng; ở đây chúng ta gọi đây là đỉnhtrùng hợp ngẫu nhiênphương pháp. Hạn chế này làm giảm phổ tớitín hiệu của các photon gamma nằm trong một thácmối quan hệ với đỉnh cao đó. Nó làm giảm nềnđáng kể, tăng tỷ lệ đỉnh điểm lên nền.Tốc độ đếm số liệu đỉnh cũng giảm, vì nó là
Trang 1Improvement of the capabilities of PGAA by coincidence
techniques
P.P Ember*, T Belgya, G.L Moln !ar
Institute of Isotope and Surface Chemistry, Chemical Research Centre, Hungarian Academy of Sciences, POB 77, H-1525 Budapest,
Hungary Received 1 July 2001; received in revised form 16 August 2001; accepted 28 August 2001
Abstract
Applicability of the g2g regional coincidence method to prompt gamma activation analysis has been examined in a series of model experiments It is shown that the requirement of coincidence with a spectral range instead of a single peak greatly improves the signal to background ratio, yet preserves the linear relationship between concentration and analytical signal The method is especially suitable for trace analysis of solutions containing hydrogen, and of matrices containing boron or other strong single g-ray emitters r 2002 Elsevier Science Ltd All rights reserved
PACS: 25.40.Lw; 81.70.Jb; 82.80.Ej; 82.80.Jp
Keywords: (n,g); E n =cold; PGAA; coincidence; spectral interference
1 Introduction
Prompt gamma activation analysis (PGAA) is a
rapidly developing method in elemental analysis
(Moln!ar et al., 1993; Moln!ar and Lindstrom, 1998) It
is non-destructive and requires a minimal amount of
sample preparation PGAA can be optimally performed
on a guided neutron beam providing excellent,
low-background conditions for in-beam experiments In
spite of this, the nature of the (n; g) reaction results in
a complicated spectral interference, and a cumulated
background continuum due to Compton scattering in
the detector Thus the PGAA spectra for real samples
are much more complex than the spectra of
Instru-mental Neutron Activation Analysis (Ehmann and
Vance, 1991a, b) (INAA) The most widely used
detec-tion instrument for PGAA is a Compton suppressed
spectrometer (Belgya, 1996), which greatly reduces the
size of the continuum and of the escape peaks
Coincident detection of gamma rays (g2g coinci-dence) is a well-established technique in nuclear structure studies (Wapstra, 1979) It is well known that coincidence methods can substantially reduce interfer-ence, and hence also the complexity of spectra The usefulness of the g2g coincidence technique in elemental analysis (Ehmann and Vance, 1991) has already been demonstrated in the case of the INAA by several authors (Meyer, 1987; Meyer et al., 1993; Jakubek et al., 1998; Koeberl and Huber, 2000) They found it especially useful in determining Ir and Se in geological samples at a ppb level Their main goal was to improve the sensitivity of the method by suppressing the continuous background and reducing spectral inter-ference
The conventional method for doing g2g coincidence
is to require a coincidence relation with a selected full-energy peak; here we call this the peak-coincidence method This constraint reduces the spectrum to the signals of those gamma photons which are in a cascade relation with that peak It lowers the background substantially, increasing the peak to background ratio The peak count rate is also reduced, since it is
*Corresponding author Fax:+36-1-392-2222/3153.
E-mail address: ember@alpha0.iki.kfki.hu (P.P Ember).
0969-8043/02/$ - see front matter r 2002 Elsevier Science Ltd All rights reserved.
PII: S 0 9 6 9 - 8 0 4 3 ( 0 1 ) 0 0 1 5 3 - 1
Trang 2proportional to the product of two full-energy peak
efficiencies A significant disadvantage is that this
method is extremely time consuming due to the low
coincidence count rate
Instead of using the peak-coincidence method, we
propose to define a coincidence relation not with a single
peak, but with a selected part of the spectrum containing
several peaks and a part of their Compton continuum to
increase the coincidence efficiency; here we call this the
g2g regional coincidence method (Ember et al., 2001)
In PGAA, a frequent problem is the presence of a few
strong g-rays emitted by the matrix Their Compton
continuum may mask the lower energy transitions of
trace elements Two of the most ‘‘problematic’’ elements
are hydrogen and boron Hydrogen is dominant in
biological samples and solutions Boron is used in heat
resistant glasses, and is also part of the neutron shielding
around the neutron guide Sometimes reactive samples
arrive, and they must be irradiated in sealed borated
glass containers, greatly increasing the background
Both elements have a monoenergetic g-spectrum with
no coincident cascades, so they may appear in
coin-cidence spectra only due to random coincoin-cidence
Recently, the possibility to eliminate the hydrogen
prompt g-ray background while increasing the signal/
noise ratio in PGAA was demonstrated (Gardner et al.,
2000) in a qualitative way
Here we compare the g2g regional coincidence
method for PGAA with the traditional singles mode
and Compton suppressed spectroscopies, demonstrating
the simplification of spectra and the increase of peak to
background ratio We also show that the coincidence
intensity is linearly related to the concentration, so the
method is quantitative and is suitable for
non-destruc-tive elemental analysis
2 Experimental work
In the measurements reported here we have studied
five different samples in three different experimental
arrangements using the same HPGe detector, at the
Budapest Research Reactor guided neutron PGAA
facility The detector had an energy resolution of
1.8 keV, and a relative efficiency of 25% Each of the
samples were measured for 13,700 s with the three
setups The neutron beam was collimated to
approxi-mately 2 cm 2 cm with an essentially constant
incom-ing neutron flux of 2 106cm 2s 1
The first arrangement is our standard setup with
Compton suppression4 In this geometry the HPGe
detector is surrounded by a BGO annulus which rejects
the unwanted Compton events, and also serves as an
active shield against background and scattered g-rays
approaching the detector The BGO shield is protected
by a massive lead passive shielding, to decrease the
number of g-rays that would hit the large volume of BGO and thereby overload the system Due to this requirement 23.5 cm is the minimum distance of the front surface of the HPGe detector from the sample The second setup is the singles mode, when the HPGe detector with no BGO annulus is placed 2 cm from the sample, thus maximising the solid angle while still maintaining some passive shielding For accumulation
of spectra in these two cases we have used a CANBERRA 8713 ADC and an S100 MCA card, with
16 K channels conversion gain
The third arrangement is a coincidence setup with two HPGe detectors Its overall performance is determined
by three factors These are the geometry, the settings of the timing electronics, and the data reduction procedure
We digitised the g-ray energy signals in 16 K channels, and the time signals in 4 K, and accumulated the events
in list mode The coincidence analysis was carried out off line This enabled us to determine the best gate conditions to improve the sensitivity The details of the electronics and the software will be described elsewhere (Ember et al., 2001)
2.1 Test samples
For the demonstration experiments we prepared a series of test samples with different concentrations The ingredients were chosen to simulate some of the difficult situations for PGAA These difficulties are usually associated with very complex spectral interference, extremely large capture cross section of a component,
or just a large amount of the matrix in the total sample
We have chosen a solution of CoCl2 and H3BO3 This sample contained two of the most problematic matrix elements for PGAA: hydrogen and boron As noted above, the main problem with this sample is the high Compton background due to the g-rays of boron and hydrogen All the intense cobalt lines and some of the low-energy chlorine peaks are situated on this back-ground Moreover, the back-scattering of hydrogen also interferes with the most intense peak of cobalt at
229 keV
All samples (S0, S1, S2, S3, S4) contained boron and hydrogen with approximately the same concentration Samples S0 and S2 had similar CoCl2mass concentra-tions, they differed only in the total mass; S0 contained
1 ml, the other four samples contained approximately 2.5 ml of solution, all in sealed Teflon containers The exact concentrations are shown in Table 1 in mass percent for each sample The last row of the table gives the sample mass in mg
In the coincidence measurement with S0, the detectors were placed at 2 cm from the sample This was the closest geometry we could achieve, thus it was used for comparison with the two conventional methods The other four samples were accurately positioned in a target
Trang 3chamber to assure the reproducibility of measurement
geometry Since this geometric precision for these four
measurements were adequate for the determination of
the count rate as a function of concentration (calibration
curve) Due to the size of the target chamber the closest
possible position for the detectors was 4.5 cm from the
sample
2.2 Experimental arrangement forg2g coincidence
As shown in Fig 1, two detectors were used for the
coincidence setup HPGe-I was mentioned above The
second detector (HPGe-II) had an energy resolution of
1.8 keV, and a relative efficiency of 15% They were
placed opposite each other and perpendicular to the neutron beam A 2 mm layer of6LiF loaded polyethy-lene protected both detectors from scattered neutrons A
1 mm thick lead plate was used to cut down the number
of X-rays, back-scattered and other low-energy gamma photons, which would only increase the dead time of the electronics The geometrical setup for the experiments with a target chamber was very similar In that case, a thin-windowed aluminium chamber was situated be-tween the detectors, surrounding the sample Lead bricks were placed around the sample and the detectors for biological and detector shielding, as shown in Fig 1 The detector signals were processed using standard electronics We will publish the details later (Ember
Table 1
Composition of the solutions, and total masses of the samples
Sample concentration (m%)
Fig 1 Experimental g2g coincidence setup with no target chamber.
Trang 4et al., 2001) The range of the time-to-amplitude
converter (TAC) was set to 200 ns, and the coincidence
resolving time was around 11.5 ns full width at half
maximum (see Fig 2)
2.3 Correction procedures
2.3.1 Random coincidence correction
Fig 2 shows a time peak with all collected events
summed up (full projection) The events of the hatched
area are considered as true coincidence events, although
there are random coincidence events among them The areas ‘‘Bkg1’’ and ‘‘Bkg2’’ selected for background, cover the same number of channels, but they consist only of random coincidence events They can be used to estimate the number of random coincidence events in the energy spectra
In this particular case, the above mentioned correc-tion does not improve the signal/noise ratio significantly Furthermore, if the coincidence experiment is performed using hardware gates instead of software gates, this correction is difficult to obtain Thus we did not apply this correction in our off-line analysis
2.3.2 Neutron flux correction The hydrogen peaks from the spectra collected
in singles mode with the second detector shown in Fig 1, were used as a neutron flux monitor The peak areas in the coincidence spectra were multiplied by the ratio of the corresponding hydrogen peak, setting the correction to unity for S1 We applied this correction only to samples S1–S4, used for obtaining the calibra-tion curve
3 Results and discussion
The spectra of sample S0, accumulated in the three different setups, are shown in Fig 3 The HPGe-I detector was at 23.5 cm from the sample in the
Fig 2 Typical time spectrum.
Fig 3 Spectra of sample S0 Thick line: ungated singles/9, medium-thick line: Compton suppressed, thin line: coincidence SE and DE denote single escape and double escape peaks, respectively.
Trang 5Compton-suppressed case but only at 2 cm in the other
two cases For easier comparison of the shapes of the
three spectra, the counts of the ungated singles spectrum
were divided by 9; thus the height of the hydrogen peak
became the same as in the Compton-suppressed
spectrum The third spectrum in the figure is a
coincidence spectrum, created by defining a coincidence
relation with every event having an energy higher than
2230 keV (just above the hydrogen peak) This region
mainly contains signals from primary and secondary
capture g-rays from the elements of interest Comparing
the singles and the Compton-suppressed spectra, we can
make the following observations
The Compton suppression reduces the hydrogen
single-escape peak to full-energy peak ratio by a factor
of 10, and reduces the double-escape peak to
unidentifi-able Due to the longer distance from the sample, the
hydrogen peak is lowered by a factor of 9 Furthermore,
the continuous Compton background is lowered by a
factor ofB100 at g-ray energies between 1 and 2 MeV However, the boron peak and its Compton background still dominate the low-energy part of the spectrum Peaks below the boron peak can hardly be seen on this background The Compton background due to boron is reduced only by a factor of 20–40, showing that the Compton suppression is very energy dependent
In comparison, the coincidence method completely removes the hydrogen peak, its Compton edge, and both
of its escape peaks The boron peak and its Compton continuum are also reduced substantially, making the low-energy g-rays much more visible
In Fig 4, the low-energy parts (under 550 keV) of the same spectra are enlarged Table 2 shows the total peak areas and their average background counts (for one channel) in cps for the two most intense cobalt peaks (229, 277 keV) and two intense chlorine peaks (518,
1164 keV) measured in the three setups for sample S1 The table also shows the signal/noise ratio calculated by
Fig 4 Spectra of sample S0 Dashed line: ungated singles/9, thin line: Compton suppressed, thick line: coincidence.
Table 2
Signal-to-noise ratio (=total peak area/average background) for the three setups with sample S1 Total peak areas and average backgrounds/channel are also given
E (keV) Peak (cps) Bkg (cps/ch) S/N Peak (cps) Bkg (cps/ch) S/N Peak (cps) Bkg (cps/ch) S/N
Trang 6dividing the peak area with the average background As
can be seen, Compton suppression gives a 1.5–2 times
better peak-to-background ratio than the ungated
singles, while the global g2g coincidence method gives
a 25–40 times better value at low energies With the
present conditions the signal/noise ratio for the highest
energy (1164 keV) is best with the Compton suppression
Fig 4 also shows that, the cobalt peak at 158 keV can
be identified only in the coincidence spectrum, where as
the background peak at 198 keV has been eliminated
completely
Fig 5 shows the calibration curve for the two
most intense peaks of cobalt at 229 keV and 277 keV
(solid curves), and for a chlorine doublet at
1164 keV (dashed curve) for comparison Data from
samples S1–S4 were used As can be seen in the figure,
the count rates increase linearly with the concentration
The calibration curves for 277 and 1164 keV peaks
go (within the 1s uncertainty) through the origin,
but the 229 keV calibration curve is only within 3s of
the origin This is because, as described earlier, the
maximum of the hydrogen back-scatter peak is
just under the 229 keV Co peak Back-scattering gives
a true coincidence between the two detectors, and
the lead shielding layers could not eliminate it
completely Changing the geometric positions of the
detectors might help with this problem For instance,
placing the detectors perpendicular to each other
and the neutron beam) instead of opposite to each
other should reduce the number of back-scattered
events The more intense 277-keV peak, which is not
affected by this problem, is therefore more suitable for
analytical work
4 Conclusions
We propose the regional g2g coincidence method
to improve the conditions of prompt gamma activat-ion analysis It improves the peak to background ratio substantially, yet the obtained calibration curve is linear, which means that the method can be used for analytical purposes over a wide range of concentra-tions This wide, dynamic range is an important advantage in those cases where both the major constituents and the trace element are to be studied The efficiency of the coincidence method can also
be improved over the efficiency of the Compton-suppression method by improving the efficiency of the second detector, thus capturing more coincidence events A trivial possibility is to use several ‘‘second’’ detectors with a higher total efficiency, in order to keep the individual detector count rate within the capability
of the electronics
We believe, it is possible to create a very accurate elemental analysis method with our coincidence ar-rangement It is clear however, that absolute measure-ments are not easy with this method Thus, it is simpler
to perform relative experiments based on standardisa-tion Also, the energy gates should be set differently for every sample composition, but once the setting is determined, then it is possible to set the energy gates electronically, simplifying the analysis by skipping the offline processing
A newly invented digital coincidence counting unit was recently introduced for 4pb–g counting (Butcher
et al., 2000), useful e.g for determining isotope activity Its use could greatly simplify the proposed regional
Fig 5 Calibration curve for the 229 keV (thick line ), 277 keV (thin line) cobalt peaks, and 1164 keV (dashed line) chlorine peak.
Trang 7coincidence method, and will enable much higher event
rates, facilitating the use of more efficient detectors
Acknowledgements
We are thankful for the help of Jesse Weil and the
continuous interest of P!al T!et!enyi during our work
References
Belgya, T., R !evay, Zs., Fazekas, B., H!ejja, I., Dabolczi, L.,
Moln !ar, G., Kis, Z., Ost.or, J., Kasz!as, Gy., 1996.
Proceedings of the Ninth International Symposium on
Capture Gamma-Ray Spectroscopy and Related Topics,
Vol 2, Budapest, Hungary, pp 826–837.
Butcher, K.S.A., Watt, G.C., Alexiev, D., van der Gast, H.,
Davies, J., Mo, Li., Wyllie, H.A., Keightley, J.D., Smith,
D., Woods, M.J., 2000 Digital coincidence counting
Fini-tial results Nucl Instrum Methods A 450, 30–34.
Ehmann, W.D., Vance, D.E., 1991 Radiochemistry and
Nuclear Methods of Analysis Wiley-Interscience, New
York, pp 302–303 (Chapter 9.6.4).
Ehmann, W.D., Vance, D.E., 1991 Radiochemistry and
Nuclear Methods of Analysis Wiley-Interscience, New
York, pp 271–291 (Chapter 9.2).
Ember, P.P., Belgya, T., Molnar, G.L., 2001 Nucl Instrum.
Methods A, submitted for publication.
Gardner, R.P., Mayo, C.W., El-Sayyed, E.S., Metwally, W.A., Zheng, Y., Poezart, M., 2000 A feasibility study of a coincidence counting approach for PGNAA applications Appl Radiat Isot 53, 515–526.
Jakubek, J., Nuiten, P., Pluha $r, J., Pop!ı&sil, S., $Sinor, M., $Stekl, I., Timorack !y, S., Vobeck!y, M., 1998 Coindence gamma-gamma spectroscopy system for instrumental neutron activation analysis Nucl Instrum Methods A 414, 261–264.
Koeberl, Ch., Huber, H., 2000 Optimalization of the multi-parameter g–g coincidence spectrometry for the determina-tion of iridium geological materials J Radioanal Nucl Chem 244, 655–660.
Meyer, G.J., 1987 Multiparameter coincidence spectrometry applied to the instrumental activation analysis of rocks and minerals Radioanal Nucl Chem 114, 223–230.
Meyer, G., Piccot, S., Rocchia, R., Toutain, J.P.J., 1993 Simultaneous determination of Ir and Se in K-T boundary clays and volcanic sublimates Radioanal Nucl Chem 168, 125–131.
Moln !ar, G., Lindstrom, R.M., 1998 Nuclear Methods in Mineralogy and Geology, Plenum Press, New York, London, pp 145–164 (Chapter 3).
Moln!ar, G., R!evay, Zs., Veres, !A., Simonits, A., Rausch, H.,
1993 Cold neutron facility for prompt gamma neutron activation analysis J Radioanal Nucl Chem 167, 133– 137.
Wapstra, A.H., 1979 Alpha-, Beta- and Gamma-ray Spectro-scopy, 5th Edition, North-Holland Publishing Company, Amsterdam, New York, Oxford, Vol I, pp 539–555 (Chapter VIII/C).