sity. If a large number of Y203 samples of uncertain purity are to be analyzed, it is definitely advantageous to use Er as the internal reference element provided a check on the Y z 0 3 itself indicates the absence of Er. The presence of Er in the YzOasamples can be readily detected by monitoring the line at 5636 A.
fluorescence is generally observed to increase in intensity with increase in Y203purity and is quenched by the presence of large amounts of impurities, as is observed here with Fe and Ca. The enhancement observed with A1 is unusual since this element is not known to be a good sensitizer. This enhancement is observed for Y P 0 4 to a far lesser extent and not at all in YVO,. In Yzo3 the enhanced fluorescence on the addition of A1 results i? a narrowing of the band fluorescence with a peak at 4100 A. The effect of impurities on Dy line intensity is substantially different in the three hosts. In Y203,all have a suppressing effect. In Y P 0 4 ,both A1 and Fe enhanced while Ca suppressed the Dy line emission. NO significant changes were observed in YVO,. The significance of the internal reference element is however clearly demonstrated since the intensity ratio of Dy/Er remained substantially constant in spite of wide variations in the Dy line inten-
ACKNOWLEDGMENTS We are grateful to R. Barba of Pechiney-St. Gobain, Paris, France, for supplying the high purity Y 2 0 a sample and R. Conzemius of the Ames Laboratory mass spectroscopy group for the SSMS data. RECEIVED for review June 19, 1972. Accepted November 1, 1972.
Instrumental Activation Analysis of Iron and Iron Ores with Short Lived Isotopes and Accurate Dead-Time Correction Christiaan D e Wispelaere,’ Jan O p de Beeck, and Julien Hoste
Institute for Nuclcar Sciences, Ghent Unicersity, Ghent, Bdgium
A method is presented for an instrumental nondestructive determination of trace impurities in iron and iron ores by means of thermal neutron activation analysis using short-lived isotopes. A dead time correction method for the general case of a mixture of short and long lived isotopes is described. The elements Mo, Cu, Co, V, Ni, AI, and Mn were determined in high purity iron with concentrations of respectively 70, 40, 80, 2.5, 1000, 0.5, and 10 ppm. The elements V, AI, and M n were determined in seven different ores. A digital PDP-9 computer was used for all data processing and interpretation.
cal solution for this problem. Wiernik (22) has published a review of several methods proposed for correction of dead time losses. Schonfeld (23) derived the general equation that allows the calculation of the true counting rate of an isotope at the beginning of the measurement, taking dead time losses into account. This equation is valid for any mixture of isotopes, short as well as long lived. In this paper an application of this equation to the analysis of high purity iron and iron ores is proposed. PRINCIPLE OF THE DEAD TIME CORRECTION METHOD
SHORTLIVED ISOTOPES are often used in activation analysis. Counting errors may be introduced due to the dead time of the equipment. In particular the automatic dead time correction available on most multichannel analyzers fails in the case of appreciable decay during the counting period. Several authors have proposed correction methods by means of additional electronic circuits in both single- and multichannel analyzers (1-9). Others (10-21) have proposed a mathematiResearch associate of the Interuniversitair Instituut voor Kernwetenschappen. (1) F. Rau and G. H. Wolf, Nucl. Instrunt. & Methods, 27, 321 ( 1963). (2) H. Seufert, ibid., 44,335 (1966). (3) J. Sida, ibid., 59, 179 (1968). (4) C. B. Nelson, J. M. Hardin, and G. I. Coats, ibid., 97, 309 ( I97 1). (5) J. Harms, ibid., 53, 192(1967). (6) H. H. Bolotin, M. G. Strauss, and D. A. McClure, ibid., 83, 1 (1970). (7) W. Gorner and G. Hohnel, ibid., 88,193 (1970). (8) W. Corner, D. Peters, and J. Zschau, ibid., 98, 371 (1972). (9) J. BartosEk, F. Adams, and J. Hoste, ibid., 103,45 (1972). (10) E. J. Axton and T. B. Ryves, In(. J . Appl. Radiat. Isotopes, 14, 159 (1963). (1 1 ) K . Low, Nucl. Instrum. & Methods, 26, 216 (1964). (12) E. Junod, R e p . CEA-R2980, France, 1966.
If r*,(r) represents the true time counting rate of the isotope r and DT(r) is the fractional dead time of the multichannel analyzer at the same moment, the number of counts collected in a time interval dt is given by
s at time
dN,
=
r*,(t)[l
- DT(t)]dt
(11
Since r*S,t= r*,,oe-’“, Equation 1 can beintegrated to result in
N,T=
s,’
r*,,o[l
- DT(r)]e-Xatdt
(2)
(13) J. P. Op de Beeck, Institute Nuclear Sciences, University of Ghent, Internal Report, 1966. (14) P. Paatero and K. Eskola, Nucl. Instrum. & Methods, 44, 357 (1966). (15) A. Gavron, ibid., 67,245 (1969). (16) F. Rossitto and M. Terrani. Int. J . ADD^. Radiat. IsotoDes, . 19, ’ 871 (1968). (17) M. Wiernik and S. Amiel. J . Radioanal. Chern.. 3.245 (1969). (18j 0. K. Nikolaenko, A. V. Sidorov, and A. S. Shtan, “Radiation Technology,” Ed. 5 Atomizdat, 1970. (19) H. A. Das and J. Zonderhuis, Nucl. Technol.,10,328 (1971). (20) K. G. A. Porges and A. De Volpi, I n r . J . Appl. Radiar. Isotopes, 22,581 (1971). (21) W. Filippone and F. J. Muno, Nucl. Sci. Eng., 47,150 (1972). (22) M. Wiernik, Nucl. Instrum. & Methods, 95, 13 (1971). (23) E. Schonfeld, ibid., 42, 213 (1966).
__
ANALYTICAL CHEMISTRY, VOL. 45, NO. 3, MARCH 1973
547
'I. 45
) € A D TIME
\i
OT.O.L54 - 0 2 5 0 . 1 @ - 3 . T* 0 . 2 5 1 ~ 1 0 ~ 6 ~ 1 2 - 0 .1l 5 C1- 9 , T 3 + 0 395.10-13.TL
LL
43 L2
o
-
%.
LI
experimental points L?-horderpolynomlal
flt
ISOTOPE
LO
39
38 37 36 35 3L
33 32
I
100
200
300
500
LOO
600
700
I
I
800
900
T
P 1000
1100
1200
SEC
Figure 1. Decay of analyzer dead time during the counting period
10 kHz CLOCK
M U LT IC H A N N E L ANALYZER
MEMORY
OSCl L LATOR
CORE
In
1
-
LIFE TIME PROCESSOR,
PULSES
MEMORY
*'c 0M P EN s A T ED
*
G A T E ,,
Enable
I
CONTROL
out
A D C. 1
'r'
DECIMAL SCALER
AMPLIFIER
I
PROGRAMMER 1
TELETYPE printer + puncher
Figure 2. Attachment of additional equipment to the multichannel analyzer for the measurement of instantaneousfractional dead time 548
ANALYTICAL CHEMISTRY, VOL. 45, NO. 3, MARCH 1973
with T = total counting time, Ns the total number of counts of the isotope s during T, and As the disintegration constant. Equation 2 can be solved to yield r*s,o, the true counting rate at the beginning of the measurement, as a function of the instantaneous fractional dead time
Table I. Concentrations of Impurities in Iron Ores. Values Are Averages of Three Determinations Aluminum Vanadium Manganese ~ _ _ ppm
=
At4
+ Bt3 + C t 2 + D t + E
(4)
The integral in Equation 3 can be rewritten as
__ e--XaT A s 6B]
+ [CAT2 - + Y-T + 2C] s2
S
e-XeT
Ax3
8094
Ouenza
1680 5173 5553 6617 5606 5171
Hamersley Tubaraogravel Miferma Marquesado MBR
In order to calculate the integral in Equation 3, it is necessary to know the function DT(t) during the measurement period. Schonfeld (23) has suggested to make successive readings from the dead time meter on the multichannel analyzer panel. Not only is the accuracy of such an indicator questionable but the whole procedure becomes unpractical when isotopes with a half-life of the order of a minute are involved. In order to sample the dead time with a sufficient high frequency, a special electronic circuit and additional counting equipment have to be used, as will be described in the following. Furthermore, the dead time function DT(t) can be well approximated numerically by a polynomial. Experiments have shown that a fourth order polynomial is sufficient. The dead time function can then be represented by : DT(r)
Krivoi Rog
+ [DT -+
z
SD 1.2 0.42 0.56 2.5 2.2 2.5 0.68
pprn 18.9 10.8
17.1 92.6 25.2 14.8 24.7
z
SD 12 30 7.0 3.0 20 12 5.7
z
ppm
SD
337 21,930 224 275 312 17,470 183
1.5 1.9 13 3.6 8.7 3.4 3.8
measure the dead time exactly at any moment. The signal is used to trigger a gate that controls the passage of clock pulses to a scaler. The number of pulses recorded per unit time divided by the clock frequency gives the fractional dead time during this recording period. The time intervals during which clock pulses are sampled have to be very small compared to the shortest half life among the isotopes being measured. In Figure 2, a block scheme is presented showing the practical set-up which was used by the authors. The gate was modified according to Bartosik’s method (24) to compensate for the finite width of the clock pulses. An electronic programmer was used to start, stop, and reset the scaler periodically and to initiate the scaler output onto a Teletype printer and puncher. In a practical case, for isotopes having a half-life of 2 minutes and higher, the dead time sampling period was 10 seconds. For isotopes with a shorter half-life, the scaler and programmer were replaced by a 400 channel analyzer operating in multiscaler mode. Sampling periods of 5 seconds down to a few milliseconds can be used if the clock frequency is proportionally increased. EXPERIMENTAL
AS
DEAD TIME MEASUREMENT
Preparation of Standards and Samples. High purity iron samples were obtained from a n iron bar of 90 kg with a length of 6.80 meters which was part of a batch of 200 kg high purity iron, prepared for the Verein Deutscher Eisenhiittenleute. Four disks of 43.5-mm diameter and 30-mm thickness, cut from that bar at different positions, were available. For the nondestructive analysis, small cylindrical samples of 6-mm diameter and 6-mm length were machined out of these disks. The weight of the samples was approximately 1.45 grams. Seven different powdered iron ores, as listed in Table I, were obtained from a steel producing plant in the vicinity of Ghent,; Belgium. Samples, weighing approximately 0.3 gram of the powdered ore, were mixed with 70% of their weight of high purity graphite. This mixture was pressed into pellets with good mechanical properties and with the same dimensions as the high purity iron samples. Standards were prepared by adding accurately known amounts of the elements to be determined, either in solution or as powdered oxides, to carbonyl iron powder. Commercially available “carbonyl iron” powder was used, obtained by decomposition of iron pentacarbonyl. From the mixture, pellets were pressed with the same dimensions as the samples and with excellent mechanical properties. The density was approximately 80% of that of wrought iron. The detailed method to prepare these standards and to test their homogeneity and reproducibility will be published elsewhere (25).
Modern multichannel analyzers are equipped with an output giving a signal which is high or low during the instantaneous dead time periods. This signal can be used to
(24) J. Bartoszk, G. Windels, and J. Hoste, ibid., 103,43 (1972). (25) C. De Wispelaere, J. P. Op de Beeck, and J. Hoste, A17al. Clzim. Actu, in press.
In order to obtain the values of the parameters A, B, C, D, E for every experiment, the sampled dead time values were fitted to a fourth order polynomial using a FORTRAN program developed for a D.E.C. PDP-9 computer. An example of the polynomial obtained from a set of instantaneous dead time measurements, during the measurement of a complex mixture of short-lived isotopes, is given in Figure 1. The same program also calculates the correction factors Fs
=
L
T
[l
- DT(t)]e-X’tdt
(6)
for every isotope present in the mixture. If long lived isotopes are present, the product X,T is much smaller than unity. In order to avoid precision problems due to truncation, the expression e - x a T has to be expanded in a Taylor series,
ANALYTICAL CHEMISTRY, VOL. 45, NO. 3, MARCH 1973
549
Channcl
e
Number
t-
t
t
Figure 3. Gamma spectrum obtained from a thermal neutron irradiated high purity iron sample with a 40 cm3 Ge-Li detector
Table 11. Nuclear Properties of the Isotopes Used Parent Specific Parent cross Peak activity, abunsection, energy, dis. gg-1 Isotope Half-life dance, barn keV sec-1 ZaAl 2.24 min 100 0.24 1778.9 422 62V 3.75 rnin 99.76 4.9 1434.4 3480 66CU 5.10 rnin 30.83 1.9 1039.0 274 6 o m C ~ 10.47 rnin 100 19 1332.4 5470 lolMo 14.6 rnin 9.63 0.20 2.55 ('O'Tc) (14,2 min) 306.8 (lolTc) (0,282) G5Ni 2.55 h 1.16 1.7 1481.7 0.277 100 66Mn 2.582 h 13.3 846.8 322
Thus standards were obtained that contained a mixture of the elements to be determined in a form similar to the samples, so that irradiation and counting conditions were almost equal for samples and standards. Thermal Neutron Irradiation. All irradiations were carried out in reactor Thetis of the University of Ghent. This is a swimming pool type reactor operating at a power level of 150 kW. Short irradiations of iron and iron ore samples with thermal neutrons produce very little matrix activity. The most important activity is produced by the threshold reaction j6Fe(n,p)b6Mn. The importance of this can be greatly diminished by selecting a n irradiation site with a high thermal to fast neutron flux ratio. In Thetis, ratios between 50 and 100 are readily available, corresponding to thermal neutron fluxes to n cm-2 sec-I. of 3 X Irradiations of exactly 5 minutes were carried out for all samples and standards, a t the same irradiation location with a thermal neutron flux of 101ln cm-2 sec-l and a thermal to fast flux ratio of 100. Because of the short irradiation periods and the stability of Thetis, samples and standards could be irradiated in sequence without the use of flux monitors. The isotopes, together with the energy of the peaks used for the activation analysis, the half-life, parent abundance, and specific activity after a 5-minute irradiation with a flux of 10lln cm-2 sec-l are given in Table 11. Specific activities are given as disintegrations Mg-l sec-I. Data Collection and Processing. After the irradiation, a decay time of exactly two minutes and a counting time of exactly 10 minutes was used for both standards and samples. All spectre were recorded with the aid of a 40 cm3 Ge-Li de550
ANALYTICAL CHEMISTRY, VOL. 45, NO. 3, MARCH 1973
tector (Canberra) with a resolution of 2.3 keV and a peak to Compton ratio of 25, for the 1332 keV gamma ray of 6oCo. The detector was connected to an Intertechnique DIDAC 4000 channel analyzer. The multichannel analyzer was online connected to a PDP-9 computer, such that spectra could be immediately transferred and stored on magnetic tape before being processed. Apart from the programs mentioned above, for the correction of dead time changes during the counting period, several FORTRAN programs are available for the c,trwplete data processing of Ge-Li spectra. Figure 3 shows the plot of a typical spectrum obtained from a high purity iron sample during an actual analysis. The plot was made from the original data by a CALCOMP incremental plotter connected to the PDP-9. Data reduction routines give the complete quantitative analysis of each spectrum : photo peak location and energy, total peak area with standard deviation and detection limit, peak widths and boundaries for all peaks present in the spectra. Complete activation analysis programs provide final analysis results, such as concentrations and upper limits for selected elements. A detailed description of the counting equipment, the data storage facilities, the connections with the computer hardware, and the software developed has been published elsewhere (26). RESULTS AND DISCUSSION
The results for the analysis of the high purity iron are shown in Table 111. The results for the analysis of the iron ores are shown in Table I. Table I11 lists the results of 9 iron samples irradiated successively. Averages are given, including their standard deviation. The errors listed with the individual results are based on counting statistics. The spectrum obtained from one of these samples is given in Figure 3. Since A,Ni, and Mo-Tc have only small photo peaks, one would expect a lower precision for these elements as is shown in Table 111. A very small photo peak of 76As resulting from arsenic present in the iron can be seen in Figure 3. This peak was too small however to allow the determination o f the arsenic concentration with sufficient precision. Furthermore j6As with its 26.4-hour half-life cannot be considered as a short lived isotope. Arsenic can be determined (together with cobalt, gallium, manganese, tungsten, and copper) with much better (26) J. P. Op de Beeck, J . Radioatzal. Cliem., 11,283 (1972).
A1
Table 111. Concentrations of Trace Iinpurities in High Purity Iron Mo co V cu
_____ ppm % S D
Sample No. 1 0.92 18 0.60 22 2 0.63 26 3 4 0.61 33 5 0.45 26 6 0.63 23 7 0.63 24 0.68 17 8 0.60 22 9 0.64 f 0.04 Averagea a Errors are standard deviations in
P P ~ %SD
2.54 0.88 2.36 0.89 2.39 0.91 2.39 0.93 2.56 0.88 2.54 0.87 2.56 0.87 2.51 0.86 2.80 0.88 2.52 i 0.04 ppm.
PPm % S D 43.8 2.5 39.8 2.8 43.8 2.7 42.0 2.7 42.0 2.5 44.7 2.4 43.1 2.5 40.8 2.5 45.8 2.5 42.9 3= 0 . 6
precision after a longer irradiation (25). Although 56Mn and 65Ni, both with a half-life of approximately 2.5 hours, are not short lived isotopes in a strict sense, the results obtained were precise enough to be considered meaningful. For nickel, this was mainly due to its large concentration in this particular iron metal. The determination of molybdenum presented a special problem. The highest photo peak, resulting from the presence of molybdenum in the iron, stems from 10ITc,a daughter of IOlMo. At no time after irradiation equilibrium is reached for this mother-daughter relation, because of the similarity in half-life. Therefore irradiation, decay, and counting times for both standards and samples had to be standardized rather accurately. The interference of the reaction 6BFe(n,p)56Mnwas decreased as much as possible by irradiating at a reactor location with a high thermal to fast neutron flux ratio. To correct for this possible source of error, the virtual concentration due to this interference was determined experimentally using the well known method of irradiating samples and standards with a n d without a cadmium thermal neutron absorber (27). A value of 0.70 ppm was obtained and used to subtract from the total Mn concentration. The values in Table I11 have all been corrected for this effect. In Table I, the Mn concentrations listed are so high that the correction was completely negligible. For the determination of cobalt, the 1332-keV peak of 6 0 m Cwas ~ preferred over the 59-keV peak, although the latter is much more intense. Figure 3 shows, however, that the shape of such low energy peaks does not tolerate an easy subtraction of the underlying continuum. Furthermore the low energy photons give difficulties due to a difference of absorption in samples and standards, that cannot be accounted for. Since 6 0 m C decays ~ to W o , and, as this isotope gives rise to a peak at 1332 keV, exactly the same energy as the peak of 6 0 m Cthat ~ is used for the analysis, a special kind of mother-daughter relationship has to be considered. Because of the absence in the spectrum, shown in Figure 3, of the peak at 1173 keV from 6oCo,which has the same intensity as the 1332-keV peak, it can be safely accepted that the contribution of W o to the 1332-keV peak is negligible. However, as in the case of Mo-Tc, the standardization of irradiation, decay, and counting time takes care of any systematic error due to this effect, Table I shows only results for the elements AI, V, and Mn. No important photo peaks other than those resulting from (27) R. De Neve, D. De Soete, and J. Hoste, Radioclzim. Acta, 5, 188 (1966).
%SD 77.6 3.3 80.6 3.3 76.2 3.3 78.2 3.3 79.5 3.3 77.6 3.3 77.9 3.3 16.0 3.3 81.5 3.3 78.3 i 0 . 6 PPm
PPm % S D 65.8 7.7 65.2 7.2 57.2 8.0 67.6 8.3 74.5 7.3 71.6 7.2 66.5 7.2 66.8 7.6 61.1 7.7 66.3 3= 1.7
Mn
Ni %SD 845 15 1106 12 951 13 931 14 983 12 824 16 15 901 722 15 940 15 912 f 36
ppm
ZSD 11.1 1.2 10.7 1.2 10.3 1.3 10.6 1.2 11.0 1.1 10.7 1.2 10.5 1.2 10.9 1.3 10.8 1.2 10.7 i U . 1
PPm
Table IV. Concentration of Trace Impurities in Carbonyl Iron. Errors Are Standard Deviations in ppm Concentration, ppm Element Mo 55.9 i 0 . 4 W 1.78 i 0.02 cu 0.75 i 0.03 co
