Introduction

The possibility to determine the element composition of large samples by neutron activation analysis at research reactors was reported first in 1993 by Bode and Overwater [1]. In the following years large sample instrumental neutron activation (LS-INAA) gained growing interest worldwide and various methods were developed and validated for the characterization of homogeneous and heterogeneous samples of kilogram size [2–10] or archeological items up to several liters in volume [11, 12]. Some specific works were dedicated to corrections for neutron absorption and gamma attenuation [13–15] and to the influence of inhomogeneities on the accuracy of LS-INAA [16]. It was also demonstrated that large objects may be investigated to some extent by prompt gamma neutron activation analysis (LS-PGNAA) using cold or thermal neutron beams [17–20]. However for the analysis of very large and thick items high energy neutrons such as emitted from isotopic sources (252Cf, 241Am–Be) or DD/DT neutron generators are of better choice due to their higher penetration depth compared to cold and thermal neutrons. Therefore NAA-systems using fast neutron sources were developed for various applications like explosive detection [21], industrial on-line bulk analysis [22] or waste characterization [23, 24]. It could be mentioned here the possibility to investigate also thick samples with the fission neutron beam of the research neutron source Heinz Maier-Leibnitz (FRM II, Garching) using the fast neutron gamma spectroscopy (FaNGaS) instrument [25, 26].

In this work large sample cyclic neutron activation analysis (LS-CNAA) of 200 L steel drums filled with concrete, polyethylene or a mix of concrete and polyethylene is carried out at the multi-element determination based on instrumental neutron activation (MEDINA) facility [24]. The drums are irradiated with pulsed 14 MeV neutrons and the prompt and delayed gamma-ray spectra recorded separately between the neutrons pulses. The methodology to determine the elemental composition of the large samples is presented. It includes the evaluation of the fast and thermal neutron flux using the steel drum as monitor and thermal neutron die-away times for correction of prompt gamma background interferences. The results are compared with those obtained for irradiation and counting conditions preferably set for the acquisition of prompt gamma-ray spectra [27, 28]. The detection limits for the toxic elements cadmium, mercury and lead are evaluated.

Basic considerations

The timing diagram for the subsequent measurement of prompt and delayed gamma-rays between neutron pulses of length \( t_{\text{p}} \) is shown in Fig. 1. The prompt gamma-ray spectra are recorded over the time \( t_{\text{pr}} \) during the decrease of the thermal neutron flux and the delayed gamma-ray spectra over the time \( t_{\text{dl}} \) after disappearance of the thermal neutron flux. The repetition period is \( T = t_{\text{p}} + t_{\text{pr}} + t_{\text{dl}} \). In a previous work [29] we showed that around 97% of the thermal neutrons have already vanished after a waiting time of about 12 ms from the end of the neutron pulse. The count rate \( Z_{E\gamma } \) (s−1) of a prompt gamma-ray measured during the time \( t = n\; \times \;t_{\text{pr}} \), n being the number of neutron pulses, may be given by.

Fig. 1
figure 1

Timing diagram for the cyclic neutron activation analysis of the large samples with a pulsed 14 MeV neutron source. t p is the length of the neutron pulse, t pr is the counting time for the measurement of prompt gamma rays, t dl is the counting time for the measurement of delayed gamma rays and T is the repetition period. The dot line represents the behavior of the thermal neutron flux

Full size image
$$ Z_{E\gamma } = \frac{m}{M}N{{\sigma }}_{E\gamma } {{\varepsilon }}_{E\gamma } {{\varPhi }}_{\text{th,pr}} $$
(1)

where \( m \) (g) is the amount of element, \( M \) (g mol−1) the molar mass of the element, \( N \) the Avogadro number, \( \sigma_{E\gamma } \) (cm2) the partial cross section for gamma-ray production [30, 31], \( {{\varepsilon }}_{E\gamma } \) the gamma-ray detection efficiency and \( \varPhi_{\text{th,pr}} \) (cm−2 s−1) the thermal neutron flux averaged within the sample over time \( t_{\text{pr}} \). The contribution of epithermal neutrons is neglected here since the die-away time of thermal neutrons (~3 ms [29]) is much longer than the time required for thermalization of fast neutrons (~300 μs [32]).

For prompt gamma rays affected by background interferences a corrected count rate \( Z_{E\gamma }^{\prime} \) (s−1) is calculated taking into account the difference in thermal neutron dynamics between irradiations with and without sample as.

$$ Z_{E\gamma }^{\prime} = Z_{E\gamma } - \frac{{\varLambda_{E\gamma } }}{{\varLambda_{E\gamma , 0} }} Z_{E\gamma , 0} $$
(2)

where \( Z_{E\gamma , 0} \) (s−1) is the gamma-ray count rate measured without sample (empty drum) and \( \varLambda_{{{{E\gamma }}}} \) (s) and \( \varLambda_{E\gamma , 0} \) (s) are the thermal neutron die-away-times determined from the count rate decay of the considered prompt gamma ray in irradiations with and without sample, respectively.

For delayed gamma rays of activation products, the cumulated count \( P_{E\gamma } \) recorded may be expressed from the basic equation for cyclic activation [33] using the irradiation and counting parameters of timing diagram in Fig. 1. In the case of fast neutron activation it is given by

$$ P_{E\gamma } = \frac{m}{M}N\theta \sigma_{f} \varPhi_{f} \varepsilon_{E\gamma } I_{E\gamma } \left( {1 - {\text{e}}^{{ - \lambda t_{\text{p}} }} } \right) \left( {{\text{e}}^{{ - \lambda t_{\text{pr}} }} } \right)\left( {1 - {\text{e}}^{{ - \lambda t_{\text{dl}} }} } \right)g\left( {\lambda ,T,n} \right) $$
(3)

with

$$ g\left( {\lambda ,T,n} \right) = \frac{1}{\lambda }\left[ {\frac{n}{{\left( {1 - {\text{e}}^{ - \lambda T} } \right)}} - \frac{{{\text{e}}^{ - \lambda T} \left( {1 - {\text{e}}^{ - n\lambda T} } \right)}}{{\left( {1 - {\text{e}}^{ - \lambda T} } \right)^{2} }}} \right] $$
(4)

where \( \theta \) is the isotopic abundance, \( \sigma_{f} \) (cm2) the isotopic absorption cross section for the considered fast (14 MeV) neutron reaction [34, 35], \( \lambda \) (s−1) the decay constant of the formed activation product [36] and \( I_{E\gamma } \) the intensity of the delayed gamma ray [36]. The average fast neutron flux \( \varPhi_{f} \) (cm−2 s−1) during the neutron pulse may be expressed by [37].

$$ \varPhi_{f} = \varPhi_{f,d} {\text{e}}^{ - \varSigma d} $$
(5)

where \( \varPhi_{f,d} \) (cm−2 s−1) is the average fast neutron flux at the drum surface and \( d \) = 56 cm the diameter of the steel drum. The total macroscopic cross section for 14 MeV neutrons \( \varSigma \) is calculated from the sample composition and using the microscopic cross section data given in [34].

In the case of thermal neutron activation and considering also the buildup of the thermal and epithermal neutron flux during the fast neutron pulse \( P_{E\gamma } \) may be expressed by

$$ P_{E\gamma } = \frac{m}{M}Nh\sigma_{\text{th}} \varepsilon_{E\gamma } I_{E\gamma } \left( {1 - {\text{e}}^{{ - \lambda t_{\text{dl}} }} } \right)\left( {{\text{e}}^{{ - \lambda t_{\text{pr}} }} } \right)\left[ {\varPhi_{\text{th,pr}} \left( {1 - {\text{e}}^{{ - \lambda t_{\text{pr}} }} } \right) + \varPhi_{\text{th,p}} \left[ {1 + \left( {0.44 + I/\sigma_{\text{th}} } \right)F} \right]\left( {1 - {\text{e}}^{{ - \lambda t_{\text{p}} }} } \right)} \right]g\left( {\lambda ,T,n} \right) $$
(6)

with \( \sigma_{\text{th}} \) (cm2) the isotopic thermal neutron capture cross section [35, 38, 39], I (cm2) the isotopic resonance integral [35, 38, 39], and F = 0.18 ± 0.02 [27] the epithermal to thermal neutron flux ratio. The average thermal neutron flux \( \varPhi_{{{\text{th}},{\text{p}}}} \) (cm−2 s−1) during the neutron pulse is related to the average thermal neutron flux \( \varPhi_{\text{th,pr}} \) (cm−2 s−1) determined after the pulse as

$$ \varPhi_{{{\text{th}},{\text{p}}}} = \frac{{\varPhi_{\text{th,pr}} }}{{\left( {1 - {\text{e}}^{{ - t_{\text{pr}} /\overline{\varLambda } }} } \right)}}\left[ {\frac{{t_{\text{pr}} }}{{\overline{\varLambda } \left( {1 - {\text{e}}^{{ - t_{\text{p}} /\overline{\varLambda } }} } \right)}} - \frac{{t_{\text{pr}} }}{{t_{\text{p}} }}} \right] $$
(7)

where \( \overline{\varLambda } \) (s) is the die-away time of thermal neutrons in the steel drum used for neutron flux monitoring [27].

The numerical determination of the gamma-ray detection efficiency \( \varepsilon_{E\gamma } \) for the sample and the steel drum is well described in [27, 28]. It is based on a point source efficiency measurement and the application of a transfer function taking into account the sample and steel drum geometry, the gamma-ray attenuation and the apparent density of the sample.

Measurements

Homogeneous large samples consist of 200 L steel drums filled either with 76 concrete or polyethylene cylindrical bodies of same size (height: 20 cm, diameter 11 cm). The weights of the concrete and polyethylene blocks are 2.6 and 1.6 kg, respectively. The elemental composition of concrete is given in [24] and listed in the last column of Table 3 for the elements detected in this work. The weight concentration of hydrogen in polyethylene is 13 ± 1%. Inhomogeneous large samples consist of a mix of concrete and polyethylene blocks whose arrangements in the drums are described in detail in a previous PGNAA work [28]. Some relevant characteristics of the samples are given in Table 1.

Table 1 Thermal neutron die-away times \( \varLambda_{{1_{\text{H}} }} \), \( \varLambda_{{10_{\text{B}} }} \) and \( \varLambda_{{28_{\text{Si}} }} \) determined from the count rates of the prompt gamma rays at 2223, 477 and 3539 keV, respectively and mean thermal neutron die-away time in the steel drum \( \overline{\varLambda } \) determined from the count rates of the 56Fe prompt gamma rays at 352, 691 and 7361 keV
Full size table

LS-CNAA is performed at the MEDINA facility with a 14 MeV neutron generator operated in pulse mode at a neutron emission of 7.18 ± 0.93 × 107 s−1 for irradiation and an HPGe (104% relative efficiency) for prompt and delayed gamma rays measurement. The length of the neutron pulses is set to 2 ms and the repetition time to 40 ms. Using two multi channel analyzers the prompt gamma rays are detected after the neutron pulses for 18 ms and the delayed gamma rays for 20 ms directly after the prompt gamma-ray acquisition. Prompt and delay gamma-ray spectra are recorded for total counting time of 1620 and 1800 s, respectively. For background correction spectra of the empty drum are also recorded under the same condition as the samples. As an example, the prompt and delayed gamma-ray spectra recorded for the 200 L drum filled with concrete and the delayed gamma-ray spectrum recorded for the empty drum are shown together for various energy ranges in Figs. 2 and 3. The origin of the prompt and delayed gamma rays observed in the prompt gamma-ray spectra of the empty drum and the drum filled with concrete is well discussed in a previous work [27]. The delayed gamma rays of 71mGe (T ½ = 20.4 ms), 75mGe (T ½ = 47.7 s), 24mNa (T ½ = 20.2 ms), 16N (T ½ = 7.13 ms), 28Al (T ½ = 2.24 m), 27Mg (T ½ = 9.46 m) and 116mIn (T ½ = 53.4 m) already observed in the prompt gamma ray spectra are detected with a better counting statistics in the delayed gamma-ray spectra due to the absence of prompt gamma activity reducing thus the Compton background. Moreover the delayed gamma rays of the short and medium lived activation products 19O (T ½ = 26.9 s), 20F (T ½ = 11 s), 24Na (T ½ = 14.95 h), 49Ca (T ½ = 8.72 m), 52V (T ½ = 3.74 m) and 56Mn (T ½ = 2.56 h) are clearly identified. The isotope 19O is produced mainly through (n,p)-reaction from the interaction of fast neutrons with fluorine in the 6LiF detector shielding. The formation of this isotope through thermal neutron capture of oxygen in concrete is negligible because of the low abundance of 18O (0.205%) and the low neutron absorption cross section (\( \sigma_{\text{th}} \) = 1.6 × 10−4 b [35]). Consequently the oxygen content of concrete is determined via the fast neutron reaction 16O(n,p)16N (\( \sigma_{f} \) = 0.042 b [34]). The isotope 20F is induced by epithermal/fast neutron capture of fluorine in the 6LiF detector shielding as well as through the fast neutron reaction 23Na(n,α)20F (\( \sigma_{f} \) = 0.140 b [34]) from sodium in concrete. The isotope 24mNa and 24Na are induced through (n,α)-reaction from the interaction of fast neutrons with aluminum in the HPGe-detector housing and in concrete as well as through thermal/epithermal neutron capture of sodium in concrete. The cross section for the reaction 27Al(n,α)24mNa is \( \sigma_{f} \) = 0.065 b [35]. For the reaction 23Na(n,γ)24mNa the cross sections are \( \sigma_{\text{th}} \) = 0.40 b [35] and I = 0.312 b [38]. The isotope 49Ca is produced by thermal/epithermal neutron capture of calcium in concrete, 48Ca(n,γ)49Ca (\( \sigma_{\text{th}} \) = 1.09 b [38], I = 0.485 b [38]) The isotope 28Al is produced by thermal/epithermal neutron capture of aluminum in concrete, 27Al(n,γ)28Al (\( \sigma_{\text{th}} \) = 0.230 b [38], I = 0.123 b [38]) as well as by the reaction 28Si(n,p)28Al (\( \sigma_{f} \) = 0.279 b [34]) from interaction of fast neutron with silicon in concrete. The isotope 56Mn is generated through the reaction 56Fe(n,p)56Mn (\( \sigma_{f} \) = 0.115 b [34]) from interaction of fast neutrons with the steel drum. The isotope 52V is supposed to be produced by (n,p)-reaction from the interaction of fast neutrons with chromium contained in the steel housing of the neutron generator. Another possible source of production for 52V could be the thermal neutron capture of vanadium traces in graphite [24].

Fig. 2
figure 2

Prompt (black) and delayed (blue) gamma-ray spectra in the energy range 100–2000 keV for the 200 L drum filled with concrete and delayed gamma-ray spectrum for the empty drum (red). (Color figure online)

Full size image
Fig. 3
figure 3

Prompt (black) and delayed (blue) gamma-ray spectra in the energy range 4000–8000 keV for the 200 L drum filled with concrete and delayed gamma-ray spectrum for the empty drum (red). (Color figure online)

Full size image

Knowledge on the die-away time of thermal neutrons is required to correct the count rate of prompt gamma rays suffering from background interferences (Eq. 2) as well as to calculate the mean thermal neutron flux built during the fast neutron pulses (Eq. 7). Die-away times were determined in a previous work [29] for the empty drum and the drum filled either with concrete or polyethylene recording prompt gamma-ray spectra at various waiting times after the end of the fast neutron pulses and analyzing the decay of the count rate of selected prompt gamma-rays. The same method is applied here to ascertain the die-away times for the drums filled with mixtures of concrete and polyethylene focusing on the background interfering prompt gamma rays of 1H (2223.2 keV), 10B (477.6 keV), and 28Si (3538.9 keV) and the prompt gamma-rays of 56Fe (352.3, 691.9, 7631.5 keV) emitted from the steel drum. The resulting die-away times are presented in Table 1. The values obtained for the various samples are comparable indicating that the neutron dynamics is mainly governed by the large amount of graphite composing the irradiation chamber. However, in comparison to the values deduced for the empty drum, the presence of sample in the drum induces a reduction of the die-away time by 8–43% according to the considered gamma-ray i.e. isotope.

The steel drum is used to monitor the thermal and fast neutron flux within the sample which values are given in Table 2. The thermal neutron flux after the neutron pulses \( \varPhi_{\text{th,pr}} \) is determined from the count rates of the iron prompt gamma rays by means of relation (1). The thermal neutron flux during the neutron pulses \( \varPhi_{{{\text{th}},{\text{p}}}} \) is calculated from \( \varPhi_{\text{th,pr}} \) by means of relation (7) using the die-away time of thermal neutrons in the steel drum (\( \overline{\varLambda } \)-value given in Table 1). The fast neutron flux at the steel drum surface \( \varPhi_{f,d} \) is estimated from the 1811 keV gamma ray of 56Mn using Eq. (4). The fast neutron flux within the sample \( \varPhi_{f} \) is calculated by means of relation (5) with a \( \varSigma \)-value (Table 2) based on the known sample composition and the apparent density of the sample \( \rho_{\text{a}} \) (Table 1). As expected the fast neutron flux decreases exponentially with increasing polyethylene content i.e. hydrogen weight concentration, while the thermal neutron flux (during and between the pulses) decreases smoothly. For the later the deviation between lowest (drum filled with concrete) and highest (drum filled with polyethylene) value is 36 ± 5%. This deviation is higher than this obtained in the PGNAA of the same samples (14 ± 1%) [28], due to different irradiation and counting conditions (\( t_{\text{p}} \) = 2 ms, T = 40 ms and \( t_{\text{pr}} \) = 18 ms in this work, \( t_{\text{p}} \) = 50 μs, T = 1 ms and \( t_{\text{pr}} \) = 930 μs in [28]). As shown in Fig. 4 the thermal to fast neutron flux ratio during the neutron pulses increases asymptotically according to the polyethylene content of the sample reflecting the competition between the moderation of fast neutrons and the absorption of thermal neutrons by hydrogen.

Table 2 Average thermal neutron fluxes \( \varPhi_{\text{th,pr}} \) and \( \varPhi_{{{\text{th}},{\text{p}}}} \) within the sample after and during the neutron pulses, respectively
Full size table
Fig. 4
figure 4

Ratio thermal to fast neutron flux in the neutron pulses as a function of the polyethylene content of the sample

Full size image

Results

The elemental composition of concrete determined by LS-CNAA is given in Table 3. The weight concentrations agree well with the values obtained by PGNAA of small samples at the Budapest Research Reactor [24] and by PGNAA of the massive concrete sample [27] taking into account the respective uncertainties. The good agreement obtained for hydrogen, boron and silicon using their prompt gamma-rays confirms the validity of the corrections for background interferences using the die-away time of thermal neutrons. For aluminum, silicon and calcium, the weight concentrations deduced from the delayed gamma rays of the activation products are in accord with those obtained from the prompt gamma measurements. In the case of aluminum and silicon whose weight concentrations are determined from fast neutron reactions [(n,α) and (n,p)] the contributions of interfering thermal neutron reactions 23Na(n,γ)24mNa and 27Al(n,γ)28Al are calculated by means of Eq. (6). The sodium weight concentration is deduced from the 23Na(n,α)20F reaction and the aluminum weight concentration from the prompt gamma-rays measurement. The oxygen weight concentration agrees well with the value estimated supposing the maximal oxidation state for all elements. In the case of the drum filled with polyethylene the weight concentration of hydrogen (13 ± 2%) agrees well with the expected value (13 ± 1%).

Table 3 Elemental composition of concrete determined in this work (LS-CNAA) compared to the LS-PGNAA performed at the MEDINA facility [27] and to the PGNAA of small samples performed at the Budapest research reactor [24, 40]
Full size table

The ratios of measured to expected amounts of the considered elements in the samples composed of concrete and polyethylene are given in Table 4. For some samples no net signal for the delayed gamma rays is observed owing to a low fast neutron flux, a low concrete content and/or an unfavorable arrangement of the concrete blocks relative to the polyethylene blocks. The deviations between measured and expected values are ranging between −31 and 41% essentially due to the simplification of the quantification model where the drum is assumed to be filled with a concrete matrix of apparent density \( \rho_{\text{a}} \). The deviations are comparable than these obtained in the PGNAA of the samples (−27 to 58%) [28]. The spread of the data, including those of the homogenous samples (only concrete or polyethylene) is represented by box-and-whisker plots in Fig. 5. In the majority of cases the amount of hydrogen and oxygen is underestimated. Only 14% of all data are identified as outliers. The Gaussian fit of all data (see Fig. 5) provides a median \( \mu \) = 0.98 ± 0.01 and a standard deviation \( \sigma \) = 0.19 ± 0.01. Thus the elements amounts in the heterogeneous samples made of concrete and polyethylene may be determined with a conservative uncertainty of 39% (2\( \sigma \)), a value comparable to this achieved by PGNAA (34%) [28].

Table 4 Ratios of the measured to excepted amounts for the elements detected in the mixed samples composed of concrete and polyethylene
Full size table
Fig. 5
figure 5

Box-and-whisker plots of the ratios for each element and Gaussian distribution of all data. Continuous and dashed lines in the boxes represent the median and arithmetic mean, respectively. The close symbols represent the outliers

Full size image

A critical point related to the quantification with fast neutron reactions is the knowledge of the total macroscopic cross section \( \varSigma \) to calculate the flux of fast neutrons within the sample. As mentioned above it was determined from the known sample composition though it may be estimated after some considerations from the results obtained by thermal neutron capture. A net signal of the oxygen delayed gamma ray correlates to a high fast neutron flux indicating the sample being composed mainly by concrete (samples HC, Ra1, Ra2, Ra6, Ax1 and Ax2). If oxygen is not detected, the polyethylene content is calculated from the hydrogen prompt gamma-ray and the difference to the sample mass is associated to concrete (samples HP, Ra3, Ra4, Ra5, Ax3 and Ax4). The macroscopic cross sections \( \varSigma_{\exp } \) obtained in this way and the resulting fast neutron fluxes \( {{\varPhi }}_{{f{ \exp }}} \) agree well with the corresponding data deduced from the known sample composition (see Table 2; Fig. 4). Hence, quantification of the elements detected via fast neutron reactions based on \( {{\varPhi }}_{{f{ \exp }}} \) agree well with the results obtained from a priori knowledge of sample composition within a margin of 12 ± 9%.

The limits of detection (3\( \sigma \) of background signal) are calculated for the toxic elements cadmium, mercury and lead in the samples. For cadmium and mercury it is calculated by means of relation (1) using the prompt gamma rays at 558.3 keV (\( {{\sigma }}_{E\gamma } \) = 1860 b) and 1693.3 keV (\( {{\sigma }}_{E\gamma } \) = 56 b), respectively. The detection limit for lead is calculated by means of relation (3) using the delayed gamma ray at 1063.7 keV of 207mPb (T 1/2 = 805 ms) induced through the fast neutron reaction 208Pb(n,2n)207mPb (\( {{\sigma }}_{f} \) = 1.63 b [35]) For the concrete sample, the detection limits are 14 and 520 mg kg−1 for cadmium and mercury, respectively for an assay of 1620 s and 5.5 g kg−1 for lead for an assay of 1800 s. The values obtained for cadmium and mercury are about 3.5 higher than the detection limits determined for an assay of 1860 s by LS-PGNAA (4 and 145 mg kg−1) [27] due to a lower thermal neutron flux. Nevertheless the detection limits achieved are comparable to those estimated for a 2000 s assay of a concrete matrix in 8-gal drums with another PGNAA-system (9 and 115 mg kg−1 for cadmium and mercury and 4.4 g kg−1 for lead) [23]. As shown in Fig. 6 the detection limits increase smoothly with increasing polyethylene content of the sample. The increase of the detection limit for cadmium and mercury is mainly related to the product of the gamma-ray detection efficiency [28] and the thermal neutron flux (Table 2), the background below the considered gamma-rays being nearly the same for all samples due to the high prompt gamma background activity of the facility. In the case of lead the increase of the detection limit is essentially caused by the decrease of the fast neutron flux.

Fig. 6
figure 6

Detection limits for cadmium, mercury and lead as a function of the polyethylene content of the samples

Full size image

Conclusion

In this study LS-CNAA of 200 L steel drums filled with concrete and polyethylene was performed with a pulsed 14 MeV neutron source recording successively prompt and delayed gamma-ray spectra between the neutron pulses. Thermal neutron die-away times were measured to correct prompt gamma background interferences and to determine the thermal neutron flux in neutron pulses. We showed that the fast neutron flux may be estimated from the results obtained by thermal neutron capture with some assumptions on the sample composition according to the detected delayed gamma rays. The method proposed to evaluate the sample composition provides for the homogeneous samples a good agreement between measured and expected values within uncertainties. For the heterogeneous samples the determined elemental contents agree with the expected values within ±39%; the large deviations observed being mainly associated to local perturbations of the fast and thermal neutron flux owing different neutron moderation and absorption properties of the materials. Possible techniques to improve results accuracy for heterogeneous samples are proposed in [28]. Improvement of counting statistics and detection limits as well as a reduction of background interferences may be achieved through a better neutron/gamma shielding of the HPGe-detector [41]. Additionally a higher detection sensitivity of the fast neutron activation products may be achieved through an optimization of the length and repetition period of the fast neutron pulses relative to their half-lives.