Use of Multiple Irradiations and Reference Materials as Comparators in Quality Control of Neutron Activation Analysis Data of Biological Samples

We propose that several different reactor irradiation times followed by assaying of activity for differential counting periods may be employed for quality control (QC) of neutron activation analysis (NAA) data of biological samples. It is also recommended that three to four reference materials (RMs) of similar matrix but from different agencies such as National Institute of Standards and Technology (NIST, USA), International Atomic Energy Agency (IAEA, Vienna), Institute of Nuclear Chemistry and Technology (INCT, Poland), and National Institute of Environmental Studies (NIES, Japan) including a synthetic multielemental primary standard should always be analyzed simultaneously along with the samples. Finally, the mean ± σ values so obtained may be considered as more reliable after statistical analysis. Our analytical data for Na and K in three RMs (SRM 1572, CRM H-9, and Bowen’s Kale) using different irradiation periods of 15 m, 2 h, 6 h, and 1 day were comparable with the certified values within error range of + 0.2 to − 2.7%. We report our data for 20 elements in two candidate RMs corn flour (INCT-CF-3) and soya bean flour (INCT-SBF-4) from the INCT, Poland, where Z-score values for most elements are in reasonable range of certified values.


Introduction
Quality control (QC) of analytical data comprise of a set of experimental and statistical procedures designed to test, systematically and continually, if the measurement process is in full control during all the steps of analysis and can produce reliable data with confidence. International Atomic Energy Agency (IAEA) have issued a TECDOC 1350 dealing with development of Reference Materials (RM) as quality control materials and for quality assurance (QA) in neutron activation analysis (NAA) [1]. Bode [2] summarized his experiences gained at the Delft University of Technology in quality management on the basis of which his laboratory for NAA at Interfaculty Reactor Institute was accredited for its quality systems for compliance with the accreditation criteria following ISO-25 guide. Emons et al. [3] have outlined the role of reference materials (RMs) in analytical QA with special emphasis on trace element analysis of foodstuff and discussed major influences on the uncertainty of certified values. Byrne [4] suggested the need of different modes of NAA (Instrumental NAA and Radiochemical NAA) to provide independent data sets in the same laboratory thus allowing internal validation or crosschecking.
However, it is an established fact that RNAA, though more cumbersome and time consuming, gives more reliable data for a single element only after suppression of interferences. Hence, in our view, this approach does not seem to be practical and hence not being used so frequently. Further, Einax and Reichenbacher [5] have discussed some statistical solutions such as homogeneity of the variance and simplification of the linear regression function including validation of the mathematical model to meet QA challenge. Recently, Dybczynski [6] demonstrated unique role of NAA for analytical QA and its contribution to the certification of candidate RMs during the last few decades. It has been emphasized that NAA plays vital role in determining the accuracy of analytical results in the trace element analysis (TEA) of inorganic elements in a variety of matrices. The quality of analytical data also depends on the minimization of errors at each stage of NAA methodology where one should have full knowledge and control of contamination, for discovering ways to minimize the run-to-run variation in analytical practice, and for improving the sample-to-blank ratio.
Instrumental neutron activation analysis (INAA) is being extensively used for the determination of essential, trace, and toxic elements in medicinal herbs and herbal formulations [7][8][9], commonly used spices [10][11][12][13][14] where elemental data are used for dietary intake resulting in possible health risk assessment [12] and nuclear forensic studies [14]. Therefore, it is all the more important to validate analytical data so that it may be used for dietary intake determination [8] and toxicological studies [9]. Standard/certified RMs are considered as the best tools for harmonizing the results and providing QC of any analytical methodology. RMs have distinct advantage of wide acceptance and form the basis for intercomparison of measurement systems for testing data produced under diverse conditions and by various laboratories around the world using similar or different techniques [15]. It is also recommended that one may prepare his own synthetic multielemental standard, also called primary standard, in the absence of a suitable in-house comparator standard. Similarly, statistical analyses of data such as box plot, Z-score, or U-score have also been considered as tools for checking reliability of an analytical measurement as these express divergence of the experimental results and indicate how far and in what direction results deviate from the mean of its distribution [15].
In the present work, we report our data on a duplicate set of synthetic primary multielemental standards containing As, Co, Cr, Fe, Mn, Hg, Se, and Zn prepared in our laboratory. We have determined Na and K in three biological RMs: citrus leaves SRM 1572 (NIST, USA), mixed human diet H-9 (IAEA, Vienna), and Bowen's Kale (UK) by using different irradiation periods of 15 m, 2 h, 6 h, and 1 day and compared our data with the certified values. We also present our data for 20 elements (Al, Br, Ca, Cl, Co, Cr, Cs, Eu, Fe, K, Mg, Mn, Na, P, Rb, Sb, Sc, Sn, Th, and Zn) for two candidate RMs: corn flour (CF-3) and soya bean flour (SBF-4) developed at INCT, Poland, where we participated.

Sample Collection and Preparation
During the past few decades, we have analyzed a large number of standard/certified reference materials (S/CRMs) of biological, environmental, and geological matrices acquired from different agencies as a part of our ongoing NAA studies. Some of the biological S/CRMs of plant origin but derived from the National Institute of Standards and Technology (USA), International Atomic Energy Agency (Vienna), Institute of Nuclear Chemistry and Technology (Poland), and National Institute of Environmental Studies (NIES), Japan, are listed in Table 1. Bowen's Kale, the first biological CRM developed by Prof. HJM Bowen, University of Reading (UK), was sent to us as a gift. Some RMs were gifted by Prof. R S Dybczynski INCT, Poland. All the samples were kept in oven at 100 °C for 2 h before weighing. For short irradiation of up to few hours, ~ 50 mg sample was weighed and packed in Alkathene. For longer duration of irradiation of ≥ 1 day, however, ~ 100 mg sample was packed in aluminum foil. For 1-2 m irradiation in pneumatic carrier facility (PCF) of Dhruva, high-density polypropylene Composition of a typical synthetic standard is given in Table 2.

Irradiation and Counting
The samples along with comparator RMs were irradiated at high Cd ratio position (high thermal to fast neutron ratio) in APSARA/CIRUS/Dhruva reactors at the BARC, Mumbai. Thermal neutron flux in APSARA, CIRUS, and Dhruva reactors were ~ 10 10 , 10 12 , and 10 13 n cm −2 s −1 , respectively. Since all the samples along with RMs were irradiated in a small packet (3 cm long × 1.5 cm diameter), vertical or horizontal flux variations were likely to be minimal or insignificant. Care was taken so that in each packet, at least 3 RMs and one primary synthetic multielemental standard were taken as comparator standards. Dhruva reactor was used for 1 or 2 m irradiations only using its PCF allowing short transfer time (~ 6 s) and convenient access to high flux irradiation position. It enables minimal loss of activity thereby ensuring better reproducibility. Besides PCF of Dhruva, 5 m irradiation in APSARA was carried out followed by counting using high-resolution gamma spectrometry (80 cm 3 coaxial EG & G ORTEC HPGe detector with a peak to Compton ratio of 40:1 and 4 k multichannel analyzer) at the reactor site and later at the Radiochemistry Division of BARC. For long-lived radionuclides, 1 day irradiations were carried out in CIRUS/APSARA reactor followed by counting using HPGe detector with 8 k MCA and GENIE-2000 software (Canberra, USA) at our laboratories at IIT Roorkee. A schedule of reactor irradiations, delay, and counting times including radionuclides identified in each case is listed in Table 3. Optimum nuclear properties of half-lives and Eγ (keV) of the radionuclides identified were same as reported in literature [12]. Overall samples were counted for up to maximum delay after 2 months to eliminate interferences due to short-lived radionuclides. Counting of 32 P (β − ) activity was performed using an end window GM counter (Nucleonix, Hyderabad) and an Al absorber of 27 mg cm −2 as per procedure developed earlier in our laboratory [16]. Soon after irradiation and suitable cooling, the container is cut open and samples unpacked. Surface of the Alkathene/aluminum packed samples was decontaminated by swiping with acetone-soaked cotton/tissue paper. These were then mounted on a Perspex sheet with suitable geometry. After initial counting, samples were also unwrapped and transferred (recovery 80-95%) in butter paper packets (1 cm × 1 cm) in a glove box to avoid contamination of radionuclide impurities in Al-foil. All the samples were then  65 Zn, 134 Cs, 60 Co, 152 Eu counted using high-resolution gamma spectrometry in our laboratory at IIT Roorkee as described in our earlier publications [10,17].

Statistical Calculations
All the samples were analyzed in triplicate and the results reported are means ± σ. In cases where multiple irradiations/counting were performed, these were included in calculating means. For the analysis of proposed CRMs, Z-score analysis was performed [15]. Some workers have used U-score, instead, to test the validity of their data [12]. Use of statistical approach has been widely used for data validation in neutron activation analysis of essential and trace elements in medicinal plants as reported in literature [18].

Results and Discussion
During the past years, we have analyzed a large number of S/CRMs of different matrices derived from several agencies such as NIST (USA), IAEA (Vienna), INCT (Poland), and NIES (Japan) including synthetic multielemental standard along with other samples of medicinal herbs [7,10], herbal formulations [8], and dust particulates from major Indian metropolitan cities [19]. In this study, we present our results on the analysis of two synthetic multielemental standards along with our results for Na and K in 3 RMs of biological origin using four different irradiation periods in APSARA reactor. Further, we present our data on 20 elements for two proposed candidate RMs CF-3 and SBF-4 obtained from INCT, Poland, as part of our intercomparison study.

Synthetic Multielemental Standard
In order to check the accuracy and precision in our NAA measurements, synthetic multielemental primary standards for selected elements were prepared, irradiated, and their elemental concentrations were calculated using any two RMs from NIST (USA) and IAEA (Vienna) as comparator standards or using a duplicate standard. We had prepared two such standards I and II as mentioned above. These were irradiated in CIRUS reactor for 1 day followed by counting at different intervals up to 2 weeks as mentioned in Table 3. Analytical results for the same are listed in Table 2. Also included in this table are errors calculated for both the standards. It is observed that most values are comparable within < ± 10% of the actual amounts taken except for Hg in standard I and As in standard II where observed concentrations are higher by + 12.7% and + 14.0%, respectively. Though these two data are beyond acceptable limit but may be still accepted with in reasonably good agreement. However, further care was taken while preparing these standards in cases of Hg and As for further studies. Hence, we can assume that our preparation methodology of primary standards is reasonably reliable. Based on these observations, it is presumed that our further measurements should also be in good agreement and reliable.

Different Irradiation Periods
In order to cross check on our INAA procedures and measurements, we had selected two elements, Na and K, that are of nutritional importance in dietary samples and these yield short-lived radionuclides 24 Na (t ½ = 15 h) and 42 K (t ½ = 12.4 h) and have sufficiently large cross section of 0.5 b and 1.3 b, respectively. Three RMs selected were of biological importance, citrus leaves-SRM 1572 [20], mixed human diet, H-9 [21], and Bowen's Kale [22,23], derived from different agencies of different countries NIST (USA), IAEA (Vienna), and UK, respectively. These were irradiated for different irradiation periods of 15 m, 2 h, 6 h, and 1 day in APSARA reactor at ~ 10 10 n cm −2 s −1 . Activities of 24 Na (Eγ = 1368.6 keV) and 42 K (Eγ = 1524.7 keV) in all three RMs along with comparator RMs were measured by high-resolution gamma spectrometry at the Radiochemistry Division of BARC, Mumbai. 24 Na is also formed by 24 Mg (n, p) and 27 Al (n, α) reactions if higher energy neutrons are present. Nuclear reaction cross section of these reactions is much small (of the order of ~ 100 mb) for fast neutrons. Moreover, care was taken that thermal to fast ratio at the irradiation position was very high. Normal concentration levels of Mg and Al in biological samples are of the order of mg/g and ng/g level, respectively. Therefore, any nuclear interference due to (n, p) and (n, α) reactions is likely to be much smaller or insignificant compared to high concentration of Na or even K in biological samples which are > 100 mg/g.
Mean elemental concentrations ± σ so calculated on the basis of triplicate measurements are listed in columns 3, 4, 5, and 6 corresponding to different irradiation periods of 15 m, 2 h, 6 h, and 1 day respectively in Table 4. Further, average of all the four concentrations ± σ are reported in column 7 of the same table. A perusal of our data in Table 4 shows that for both Na and K concentrations as obtained for four different irradiation periods are comparable with the certified values. Not only that but its σ values calculated for all the four cases are also much < 10% except in case of Na in SRM 1572 where its certified Na concentration is 160 ± 20 µg/g. Further, if we compare average Na and K concentrations obtained from four irradiation measurements, then these are much closer to certified values. Also ± σ for average is also improved with RSD much < 5% and also less than those for certified values. Thus, average concentrations are more accurate and precise compared to certified values. In fact, the errors listed in last column are in close range of + 0.2 to − 2.7% only.
Therefore, it is concluded that this approach of multiple irradiations yields more accurate and precise data. Hence, it could be followed for quality control of a few selected major or minor constituent (not trace) elements so as to get more reliable and accurate analytical data of the elements in a sample.

Intercomparison Studies of CF-3 and SBF-4
Based on the quality of our INAA data during the past few decades, we were invited to participate in intercomparison studies of two candidate RMs corn flour (INCT-CF-3) and soya bean flour (INCT-SBF-4) developed by Prof. R S Dybczynski at INCT, Poland. Moisture contents of the two RMs were reported to be 8.77 ± 0.22% and 5.44 ± 0.07%, respectively [24]. CF-3 was prepared from corn grown in Poland according to Polish standard PN-A-74205. It was sieved through 250 µm nylon sieves. SBF-4 was prepared from soya bean grown in India and sieved through 150 µm nylon sieves. Approximately 50 kg each of both flour samples were stored in the polyethylene bags and examined by optical spectroscopy whereby Martin's diameter was found to be 25 µm and 50 µm, respectively [24]. Further, both the flour samples were homogenized by rotating in three directions and sterilized by electron beam irradiation for longer shelf life. In this case, four CRMs of rice flour-SRM 1568a [25], and wheat flour-SRM 1567 [26] from NIST (USA), Whey Powder (CRM-155) [27] from IAEA (Vienna), and rice flour no 10a from NIES, Japan [28], along with a primary synthetic multielemental standard (containing As, Cr, Fe, Co, Zn, Hg, and Se) were used as comparator standards. These were irradiated in Dhruva reactor for short irradiation of 2 m. For longer irradiation of 1 day, CIRUS reactor was used. It may be noted that though we have used up to four RMs as comparators but it is not essential. However, use of a minimum of two RMs is recommended. Elemental concentrations so calculated were considered only if these were satisfactory (< ± 10% of certified value) and finally mean ± σ were calculated for 20 elements on dry weight basis and these are listed in Table 5. In the meantime, two candidate RMs have become certified and their certified/information values [24] are also listed in Table 5. Two RMs were certified on the basis of worldwide interlaboratory comparison study where 92 laboratories from 19 countries had participated. Our elemental data for Eu and Sn may be considered for information only. A perusal of data shows that for some elements such as Cl, Cr, Fe, Mg, Mn, Na, Sc, and Zn, high errors (> 10%) are observed for CF-3 or SBF-4 or both.

Statistical Analyses
As a part of statistical analysis, Z-score values [15] were calculated for those elements (14 for SBF-4 and 10 for CF- 3) where certified values were available as listed in Table 5. A plot of Z-score values in both the candidate RMs is shown in Fig. 1. No Z-scores were calculated for information values. It is observed that for SBF-4, Z-scores of 8 elements out of 14 are within ± 2.0 and for CF-3, only four out of 10 are within ± 2.0. It is further noted that Z-scores of 4 out of 14 elements in case of SBF-4 and 4 out of 10 in case of CF-3 are within the range of ± 5.0. Therefore, we can say that 12 out of 14 elements in SBF-4 and 8 out of 10 in CF-3 are within Z-score values of ± 5.0. Overall, Fig. 1 shows that concentrations for most elements besides Fe, Mg, Mg, Mn, and Zn for SBF-4 and Br, P, Rb, and Zn lie between ± 3.0 suggesting that our data to be within 99.7% confidence limit. Also, our data in Table 5 show that %RSD for all the elements in both the candidate RMs is < 10% to indicate high precision of our measurements. However, concentrations of Cr, Na, and Sc in SBF-4 are higher by ~ 25%. This could be perhaps because these are reported as information values. For CF-3, our data for Cl, Cs, K, Sb, and Sc are within + 15%. In general, our data are on the higher side of the reported certified values as it is clear from Z-score plots. A cursory look of Fig. 1 shows that most of our data suggest positive bias and only a few points are on the negative side. Surprisingly, Zn concentration is way off in both cases; for SBF-4, Z-score is − 5.54 with error − 13.8% and for CF-3, Z-score is + 6.13 with error + 25.4% suggesting some error in the measurements. This may perhaps be due to the fact that multiple irradiations could not be performed due to constraints of reactor time availability. However, elemental data within Z-score of ± 3.0 are certainly acceptable within reasonable accuracy. On the other hand, our data with Z > 5 such as that of Fe, Mn, and Zn in SBF-4 and Br, P, and Zn in CF-3 are not acceptable which may perhaps be due to some interferences. Some elements such as Eu, Sb, and Sn have no certified or even informative value, and these are reported for comparison only. During the last three decades, we have extensively employed INAA for the determination of minor, trace, and toxic elements in a large number of biological samples of plant origin such as Ayurvedic Indian medicinal herbs [15] and herbal formulations [8]; Indian vegetarian dietary constituents of cereals, vegetables, and spices [10]; Indian chewing tobacco products [29]; Arjuna bark, an Indian herbal heart tonic [30]; and turmeric powder-an Indian spice for forensic studies [14] where multiple RM comparators have been used for data validation and found it quite useful. Our method is simple and less cumbersome that can be used very easily.

Conclusion
For a long time, INAA has been recognized as a suitable analytical technique for data validation of minor and trace elements in RMs of biological origin. However, it is essential to control various sources of errors in each step of INAA including sampling, packing, irradiation, and counting of activity before we calculate elemental concentrations. In order to get quality data, it is proposed that at least two RMs of similar matrix but derived from different agencies along with a primary synthetic multielemental standard be analyzed simultaneously along with the sample. Also, for some elements such as Na, K, and P, sample should be irradiated for multiple time periods and then mean ± σ be calculated to have better accuracy and precision. Our data for INCT-CF-3 and INCT-SBF-4, two candidate RMs from INCT, Poland, were found in good agreement with the certified/informative values for most elements. Also, Z-score plots may be considered as means of statistical analysis for NAA data validation. It is suggested that concept of multiple RMs may be also used in other analytical methodologies such as atomic absorption spectrophotometry using comparator method.