Irradiation of Elements from Z = 3 to Z = 42 with IO-MeV Protons and Application to Activation Analysis J. L. Debrun,* J. N. Barrandon, and P. Benaben , Groupe d'Application des Reactions Nucleaires a I'Analyse Chimique CNRS, Service du Cyclotron, 45045 Orleans, Cedex, France
The sensitivities for the determination of 22 elements from 2 = 3 to 2 = 42, using 10-MeV proton activation, were caicuiated from experimentally measured yields for 47 radioisotopes obtained via (p,n) reactions. The error associated with the measurement of the activation yields ranged from 5 to 15%. For an irradiation of 1 hour at 1 PA, the calculated sensitivities are better than 10 ppb for Ca, Ti, Cr, NI, Cu, Ga, Ge, Zn, Se, Br, Rb, Y, Zr, and Mo; for LI, S, V, Fe, As, Sr, and Nb, these sensitivities are better than 100 parts per biilion. Experimental results for the simultaneous and nondestructive analysis of several trace elements in Ai, Ag, Au, Co, Dy, Ho, ir, Nb, Pr, Rh, Si, Ta, Tb and various minerals are presented.
Most workers in the field of charged particle activation analysis have concentrated their effort on the light elements. In the case of protons, however, the activation of some isolated heavier elements was also studied and recently there were several systematic studies. Barrandon, Debrun, and Kohn ( I ) studied the activation of 30 elements, from the standpoint of nondestructive multielemental analysis; relative activation yields were presented. Application to the analysis of various matrices by direct y-ray spectrometry was achieved (2). Debrun, Riddle, and Schweikert ( 3 ) ,in their study of the activation of 24 elements, limited their investigations to radioisotopes with T < 5 min. Relative activation yields and activation curves were presented; no practical applications were performed. This study was later extended by Riddle ( 4 ) . Kormali (5) studied the activation of 11 elements and gave absolute activation yields (in disintegrations/min) and activation curves. McGinley (6) studied the determination of 34 elements by proton activation followed by x-ray counting; relative activation yields and some activation curves were presented; application to the nondestructive analysis of NBS glass standards was achieved. The data presented here are part of a study on the activation of 54 elements, with application to the nondestructive analysis of a number of matrixes. The complete study includes absolute activation yields, y energy measurements, calculated a priori detection limits, experimental detection limits in 20 matrices and activation curves. The distance between the Cyclotron and the Laboratory was -6 km; for this reason, we have limited our study to radioisotopes with half-lives >15 min. Because of the large volume of this work, it was divided in several parts for publication. The activation curves will be published separately (7), and this paper is limited to elements from 2 = 3 to 2 = 42. Protons were chosen because they are of considerable interest: they are the only particles that permit the sensitive determination of many elements, without interferences and nondestructively in a number of matrices, provided that the bombarding energy is correctly chosen. At 10 MeV, high specific activities are obtained via (p,n)
+
reactions, while the ( p p ) , the (p,2n) and the (p,d) (p,pn) reactions, also energetically possible, possess low cross sections or are close to their experimental threshold and do not yield high specific activities. This means that the latter reactions cannot in general be used for trace analysis and it also means that interferences from the elements (2 2) and (2 1) are usually not important. As shown below, it is possible for all of the studied elements to find a suitable and interference free (p,n) except for Li and B. If the energy selected were lower, the sensitivities would decrease; if it were higher, interferences would become sizable and most matrices would cease being favorable for nondestructive analysis because of the radioactivity induced by the (p,d) (p,pn) reactions. Needless to say, the most interesting analytical situation is obtained when nondestructive analysis is feasible; in this respect, a proton energy of -10-11 MeV generally represents a good choice.
+
+
+
EXPERIMENTAL Irradiations were performed with the variable energy cyclotron of the French A.E.C., Service Hospitalier F. Joliot a t Orsay, as described in earlier work (8).The proton energy a t the surface of the sample was calculated to be 10.04 MeV. All Ge-Li spectra were stored on magnetic tape and automatically processed using the computer code SAMPO (9) modified in our laboratory. Our detectors were especially recalibrated in efficiency with the help of several accurately standardized radioactive sources, obtained from the Laboratoire de MBtrologie des Rayonnements Ionisants a t Saclay. The irradiated products were pure elements whenever this was possible, and well-known compounds in the other cases. For quantitation, the method of the average cross section (10) was used. The concentration x of an element is given by:
where R , and R,t are the ranges of the incident particles in the sample and in the standard, respectively. These ranges are read from the published tables (11). A, and A,t are the radioactivities of the same radioisotope in the sample and in the standard respectively, measured under the same conditions, after identical irradiations. For the standard, A,t represents here the total radioactivity for a thick target. In order to be able to perform quantitative analysis, one has to measure the quantities ASt for all the elements of interest, under standard conditions. These specific activities are then multiplied by the appropriate correction factors when a sample is not irradiated under these standard conditions. For the measurement of the specific activities, the irradiations lasted from 5 to 20 min, a t approximately 1 c(A total extracted current. Up to 12 different targets were mounted on a rotating irradiation setup (400 rpm) so that the actual average current on each target was -0.05 @A. Irradiation intensities were measured as follows: a) A 10-c( Havar foil was irradiated alone and the current was measured with a Faraday cup. This experiment, repeated several times, gave the following relations, for a given irradiation time and a t a given time after irradiation: Activity j6Co = K1I Activity jZmMn= KgI Activity j2Mn = Ksl ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
167
irradiation. For better comparison, column 7 in Table I gives the specific activities, all normalized to the range of the 10-MeV protons in aluminum, according to the Equation ( 8 )
ALUMINIUM MATRIX
A N = A , ~R-N
(2)
Rst
".
Fe. As.
Nb
Sr
.
Ca. Ti.
Zn.
Br
.
Ma.
Go. Cu. G e .
zr
Y'
se.
Rb
.
2
I 3
5
10
15
20 ATOMIC
25
30
Is
LO
NLCIBE~
Figure 1. Calculated best detection limits for 22 elements Irradiation: 1 hr, 1 fiA. IO-MeV protons
where 56C0,jZmMn,and j2Mn are obtained by (p,n) reactions on Fe and Cr contained in the Havar foil. I is the beam current, K1, Kz, and K3 are constants. b) All standards and samples were irradiated behind a Havar foil and the beam current values were obtained from the activities of 56C0, 52mMn,or j2Mn. Experience has shown that measurement ~ is in most cases quite enough. of 5 6 C only The irradiations for sample analysis lasted from 5 to 60 min (120 min for a pure silicon sample), a t 1 to 4 PA; the beam current was usually quite stable and eventual small variations had negligible effect on the results compared to other sources or uncertainty like counting statistics in the case of trace determination or variation of the mean energy of the beam from one week to another. A few irradiations for which the beam current was unstable, were discarded.
RESULTS AND DISCUSSION For each studied element, we have measured the specific activities on all radioisotopes of interest and these specific activities were measured on all main y rays; to reduce the volume of Table I, the number of y rays per radioisotope is limited to 5 maximum (the 5 major ones). The specific activity is expressed in ylmin instead of disintegrations/min; indeed the number of ylmin is what really interests the analyst. For the same reason, instead of presenting only one specific y activity (e.g. the major one) and letting the analysts calculate the others with the help of the disintegrations schemes, we felt that it was more convenient to give several other experimental specific y activities. The specific activities are obtained by measuring the experimental number of counts/min, and correcting for the efficiency of the detector. These activities are the number of y rays/min emitted by a thick target (target thicker than the range of 10-MeV protons) of the element of interest or of the indicated compound, number corrected to correspond to 1 ppm (weight) of the element. This means that the experimental number of y-rayslmin emitted by a thick target of the element of interest, was divided by lo6, since a pure element, corresponds to lo6 ppm. In addition, in the case of a compound, the experimental value was multiplied by lOO/C, where C is the percent concentration of the element of interest in the compound. The irradiation conditions are 1hr, 1 wA, 10.04 MeV, and the radioactivity corresponds to the end of the 168
*
ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
where A N is the normalized activity, R N and Rst, the ranges in aluminum and the irradiated material, respectively. The half-lives in column 4, Table I, were taken from Ref. (12); the y energies in column 5 were taken either from Ref. (12) or averaged with our experimental values. No specific activities were measured for Be, C, N, 0, Ne, F, Na, Mg, Al, Si, P, C1, K, Ar, Sc, Mn, and Co, because the radioisotopes obtained are not generally suitable for nondestructive activation analysis, at trace level. Kr may probably be determined, but was not studied. Column 8 in Table I gives the detection limits calculated from the standardized specific activities and assuming that the minimum detectable activity for a given peak is equal to 3 times the square root of the background in the corresponding energy window. The samples are supposed to be counted 1.8 X T or 60 hours maximum. The background used for the calculations is relative to the following experimental setup: Shielding: 5 cm lead, 2 mm Cd, 1 mm Cu; detector: efficiency relative to an NaI crystal under the usual conditions and for the 1332.5-KeV from 6oCo = 22%; peakto-Compton ratio: 34 for the 1332.5 KeV from 6@Co;resolution: 2.7 KeV for the 1332.5 KeV from 6oCo. The calculated detection limits are shown graphically in Figure 1 (normalized to an aluminum matrix). Column 9 shows the experimental detection limits for an ultra-pure silicon sample. The difference of the values for the detection limits in columns 8 and 9 arises from different irradiation conditions (1 hr at 1 WAfor column 8, 2 hr at 1.45 pA for column 9), but is also due to the presence of 3oP from the 30Si(p,n)3@P reaction. As discussed a little later, the calculated a priori detection limits are often quite different from the actual experimental limits. The specific activities in column 6, Table I, are believed to be accurate: they represent the mean value of 3 to 12 experiments according to the studied element. However, as indicated in Table I, the precision is rather poor (5-15%). In this error, influence of the counting statistic is negligible because the samples are sufficiently radioactive. The poor precision is attributed to the poor quality of the beam, the mean energy of which could vary from one series of experiments to another. We recently discussed this problem in the case of proton and helium-3 experiments (13). The error indicated does not correspond to a standard deviation but to the range of the values; this means that in the case of the value X f 15%,no experimental value was found outside the interval (X - 15%)-(X 15%).The average values of A,, given in Table I were used to calculate the concentrations or the limits given in Table 11. Except in the case of Nb and T a where 6 different irradiations were performed for each sample (14), all other samples were irradiated only once. For these samples, several values for the concentration of a given impurity were obtained by making use of the different radioisotopes produced, of the different y rays that belong to the same radioisotope, and of different countings. The figures in Table I1 represent the mean values and the standard deviations obtained in this way. For these calculations, the A,, are assumed to be known without error. This assumption has little influence on the calculation of the detection limits, which need not be very precise. In the case of detected elements, the actual precision of our analysis is worse than in-
+
~~
Table I. .Irradiation of 22 Elements with 10-MeV Protons: Radioisotopes, Specific Activities and Calculated Detection Limits
Element
Li
B S Ca
Ti
V Cr
Target
LiCl BN S
Radioisotope
Be 'Be 34mc1
Glass 12% CaO
44mSC
Ti
4BV
V Cr
44
sc
4*sc
' Cr 52mMn
52Mn 54Mn Fe
Fe
5 6
co
57c0
Ni
Ni
6OCU 6'cu
55c0
cu
cu
Zn
Zn
Ge
AsGa
Ge
68Ga 69Ge
"As "As 74A~ 76As
As
Se
AsGa
Se
Y Energy, Kev
1279.2 1279.2 0.53
477.6 477.6 146 1106.5 1177.5 1157 271.2 983.4 1037.4 1311.6 944.2 983.5 1311.8 320 1434.4 744.1 935.5 1434.2 834.7 846.7 1037.6 1238 1771.4 122.1 826.2 1332.5 1792 283 656.3 1185.7 931.5 669.8 962.2 1115.5 834 1039.3 1333.8 1730 1918 93.1 184.2 3 00 1076.8 553.4 574.1 872.4 1106.6 1336.6 744.9 1040.3 1114.6 630 834 595.9 634.9 559.2 657.2 121.2 136.1 264.4 279.4 400.5 559.3 520.9 665.7 554.3 619.1 698.4 776.6 827.8
3.95 58.6 43.7 386.4 667.7 0.35 134.4 7500 1855.2
6480 0.39 3.41 17.9 0.64 5851.2 9.4
67Ga
Ga
Halflife, hr
75Se
76Br 77Br aoBr 8ZBr
78.1 1.14 39
0.87 26 424.8 26.3 2880
15.9 56 0.3 35.4
Experimental No. of ylmina
54+ 5.75+ 275+ 61++ g+t+ 626+ 1.65++ 10.6++ 10.6++ 10.6++ 19++ 240++ 240++ 29 49800+ 62++ 67++ 75++ 1.5+++ 31+ 4++ 20+ 5.5++ 1.5++ 1370++ 7000+ 4211+ 44++ 43++ 16.5++ 19.5++ 4630++ 4000++ 63++
108++ 844++ 20f+ 47++ 39++ 34.5+++
18+++ 14+++ 564+ 12.5+++ 227f 244+ 678+ 79++ 847++ 3732++ 774++ 70.5++ 850f
85++ 23++ 200+ 24++ 9++ 29++ 26.5++ 13f+ 13++ 253+ 45ff 15060ff 175++ 142++ 72.5++ 251++ 76++
Number of y/min normalized t o Alb
52.7 7 257.5 57 8.4 688 1.8 11.5 11.5 11.5 15.5 196 196 23.5 40650 50.5 54.5 61. 1.2 25.5 3.3 4.5 1.2 1120 5710 3440 36 35 13.5 16 3600 3110 5.3 83 650 15.5 36 30 26.5 13.8 10.8 434.8 9. 166 180 500 58 620 2720 570 51.5 620 62 17 146 17.5 6.5 21 19.5 9.5 9.5 180 32.8 I10600 124 100 51 77 58
Calculated detection limit, ppbC
Experimental detection limit in ultrapure Si, ppbd
24 190 21.5
9.6 90 240
8.8 3190 165 175 200
6.1
9 9 43 0.4 28 31 22 1225 61
5.5 4.5 86 130 490 150 3 4 390 6
3.3
17 3.3
25
22
18
5.2
40 60 90 35
4
3.1
5 3 22
2.6
10
40 40
15
10 42 1.5 10
7.6
ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
169
Table I. (Continued)
Element
Br
Target
KBr
Radioisotope
79Kr
Halflife, hr
34.9
Rb
RbCl
85mSr
1.17
Sr
Sr-HPO,
87mSr Y
2.83 0.64 14.6
80 2568
Y Zr
Y Zr
78.4 14.6 243.8 842.4 23.35
Nb
Nb
Mo
93 m ~ 0
Mo
6.9 4.9
20 104.4
99mTc
6.05
7 Energy, Kev
217.5 261.4 306.8 397.6 606 151.3 231.7 388.5 793 974.3 1039.7 627.8 777.5 1077.4 1153.4 1920 484.9 898.2 1835.8 909.1 141.2 1129.1 934.5 765.6 460.2 569 778.7 263.1 684.8 1477.1 702.9 871.3 765.8 1074 3 14 778.2 812.6 1127 140.5
Experimental No. o f .y/mina
37+ 232+ 42+ 174+ 166+ 1635+ 10980+ 2755+ 3 I++ 24++ 16++ 81++ 55++ 232++ 77++
so++
30.5++ 29++ 29++ 1606+
939++ 1280++ 76+++
I+++
27++ 52++ 93++ 68++
143+ 162+ 124++ 144++ 5 64++ 23.5++ 2.5++ 132++ 101++ 19++ 97++
Number of r1mi.n normalized t o Ab
Calculated detection limit, ppbC
Experimental detection limit in ultrapure Si, ppbd
25.5
160 29 120 115 1260 8450 2120 28 21.5 14.5 73 50 21 0 70 72.5 27.5 26 26 1100 640 870 51.5 0.7 18 35 63 46 95
110 83 95 380 15.5 1.7 88
6
6
9 12 0.7 2 410 620 1000 13
1.9
11
44 66 2 2 3.5 34 2350 30 44 31 30 45 45 5
68
18 22
13 65
30
0.9 2
25 4.5
a Number of y-rayslmin for a thick target containing 1 ppm of the element of interest. Values at the end of an irradiation of 1 hr at 1 pA and 10 MeV. + = i5%, ++ = +lo%, + + + = *15%. b Number of y-rays for an aluminum target containing 1 ppm of the element of interest. The detection limit is calculated for a minimum detectable activity of 3 times the square root of the background, in the energy window of interest and for an aluminum matrix. d Silicon sample irradiated 2 hr at
1.45 pA.
dicated since the error on the ASt should be compounded with the indicated standard deviation. As for the accuracy of the results, it was found satisfying every time that it could be checked by another analytical method; for instance, in the case of the analysis of rhodium ( 8 ) ,of gold ( 1 5 ) ,of minerals (16),and of Cu-Pt and Cu-Pd alloys (17). If we consider the results given in Table 11, several remarks may be made: 1) There is often a big difference between the a priori calculated detection limit, which is calculated from the activity of a standard and from the background of the counting system and the actual experimental detection limit. Indeed, the radioactivity induced in the matrix itself and the radioactivity from the impurities, worsen the detection limits because of the presence of Comptons or of photoelectric peaks. It is worth noting that the matrix radioactivity is created not only by (p,n) reactions, but also by (n,y) and sometimes (n,p) or ( n p ) reactions, with the secondary neutrons produced in the sample, in the collimators, and the target holder. For instance, irradiation of an aluminium sample leads to a very short-lived radioisotope by (p,n) reaction (27Si T = 4.16 sec), but the 170
ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
analysis is perturbed by the presence of 28A1 [(from 27Al(n,y)28A1], 27Mg [27Al(n,p)27Mg] and 24Na [27Al(n,a)24Na]. In some cases, elements leading to short-lived radioisotopes only, like S or Rb, cannot be determined in matrices leading to medium-lived radionuclides like Nb or Ag, while these matrices can be analyzed for many other elements leading to long-lived radioisotopes. In short, the detection limit for a given element will vary considerably from one type of sample to another one. But, the calculated a priori detection limits (Table I, column 8) are of interest because they indicate what is the best sensitivity that can be reached in ideal conditions. 2) All y-ray interferences were carefully considered in this work, because they are a major source of inaccuracy. The standard deviations given in Table I1 are the proof that there is no major y-ray interference for the selected peaks. Indeed, we made use of several radioisotopes, several y-rays and several countings to calculate the concentration of a given element: any interference would be indicated by an anomalous value and would lead to a high standard deviation.
I
Le h
m
v
&PO
o m
vv
z o o v v
(00000
vvvvv
N ~ 0 0 0 0 00
v v v v v v v
v
I -
"? 0
N
'9
0
0 II h
3
0 3
ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
171
More detailed discussion of the y-ray interferences is given in earlier work (18). 3) As already mentioned above, the possibilities of nuclear interferences are few a t 10 MeV. It may be seen in column 3, Table I, that the only interference is between Li and B, 'Be is produced by 'Li(p,n)'Be and by 'OB(~,CI)~B~. No other interference was detected by direct y-ray spectrometry of elements irradiated at 10 MeV. With chemical separations, it would probably be possible to observe some ( p p ) reactions on the lighter elements. Anyhow, if one considers all the Q-values for the nuclear reactions and the relative positions of the stable and of the radioactive isotopes for the studied elements, one sees that for all of them it is possible to find a suitable and interference free (p,n) reaction (except for Li and B).
CONCLUSIONS 10-MeV proton activation may be used for the determination of trace concentrations of the 22 elements studied in this work; 47 radioisotopes and over 100 associated y-rays are obtained via (p,n) reactions, thus offering a great choice to perform the analysis and a great selectivity (except for Li and B). This study was complemented by an application to the analysis of trace elements in 19 different matrices. In the case of these 19 matrices, 10-MeV proton activation is a powerful1 technique because it is sensitive, selective, and also nondestructive.
LITERATURE CITED J. N. Earrandon, J. L. Debrun, and A. Kohn. J. Radioanal. Chem., 16, 617 (1973). J. L. Debrun and J. N. Earrandon, J. Radioanal. Chem., 17, 291 (1973). J. L. Debrun, D. C. Riddle, and E. A. Schweikert. Anal. Chem.. 44, 1386 (1972). D. C. Riddle, Dissertation, Texas A. 8 M. University, College Station, Texas, May 1973. S. M. Kormali, Dissertation, Texas A. 8 M. University, College Station, Texas, December 1973. J. R. McGinley, Dissertation, Texas A. 8 M. University, College Station, Texas, December 1974. J. N. Earrandon. J. L. Debrun. and A. Kohn, R. H. Spear, Nucl. lnstrum. Methods. 127, 269 (1975). J. L. Debrun, J. N. Earrandon, P. Eenaben. and Ch. Rouxel, Anal. Chem., 47, 637 (1975). J. T. Routti and S. G. Prussin, Nucl. lnstrum. Methods, 72, 125 (1969). Enzo Ri'cci and R. L. Hahn, Anal. Chem., 39, 794 (1967). C. S. Williamson, J. P. Eougeot, and J. Picard, Commissariat a I'Energie Atomique (France) Report R.3042 (1966). G. Erdtmann and W. Soyka, "Die y Linien der Radionuklide." Kernforschungsanlage Julich (Federal Republic of Germany). Report Jul-1003AC (1973). J. N. Earrandon and J. L. Debrun, Proceedings of the 7th Materials Research Symposium, Gaithersburg. Md., October 7-1 1, 1974, in press. J. N. Earrandon, P. Benaben, J. L. Debrun, and M. Valladon, Anal. Chim. Acta, 73, 39 (1974). Ch. Rouxel. J. N. Earrandon. and J. L. Debrun, unpublished results. R. Delmas, Dissertation, University of Paris, June 1975. . (17) P. Eenaben. J. N. Earrandon. and J. L. Debrun, unpublished results. (18) P. Eenaben. J. N. Earrandon, and J. L. Debrun, Anal. Chim. Acta, 78, 129 (1975).
RECEIVEDfor review April 29, 1975. Accepted September 23, 1975.
Prediction of Continuum Intensity in Energy-Dispersive X-ray Microanalysis C. E. Fiori,' R. L. Myklebust, and K. F. J. Heinrich Analytical Chemistry Division, lnstitute for Materials Research, National Bureau of Standards, Washington,D.C. 20234
Harvey Yakowitz Metallurgy Division, lnstitute for Materials Research, National Bureau of Standards, Washington,D.C. 20234
A method for background prediction in electron-excited energy-dispersive x-ray spectrometry (EDS) is described. This method yields the intensity of the continuous radiation which would be observed by the detector as a function of x-ray photon energy. The method can be incorporated into existing electron probe microanalysis data reduction programs. Tests indicate that the proposed background correction scheme is satisfactory for quantitative analysis.
In energy-dispersive x-ray spectrometry (EDS) (1), the peak-to-background ratio of the characteristic signal rarely exceeds 50 while in wavelength-dispersive spectrometry (WDS), this ratio is often higher than 500. Therefore, correction for background is far more important in EDS than in WDS. We will describe a simple and accurate method for background corrections in EDS, which is an extension and amplification of schemes proposed by Ware and Reed ( 2 ) ,and by Lifshin ( 3 ) .The expression used in this method predicts the intensity of the continuous radiation recorded in EDS as a function of x-ray photon energy. I t is based on the 172
ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
measurement of the continuum intensity at two values of energy on the target of interest. We have tested the procedure by comparison with measured spectra and application in quantitative x-ray analyses.
THEORY Prediction of Intensity of Continuous Radiation. Continuous radiation from thick targets has been investigated in detail by many authors; an extensive review has been presented by Stephenson (4).Much of the work reviewed there was directed at obtaining analytical expressions, such as Kramers' equation ( 5 ) , for the continuum intensity generated within the specimen as a function of x-ray photon energy. However, for background correction of line intensity measurements with a lithium-drifted silicon [Si(Li)] detector system, it is necessary to account for the loss through absorption by the specimen of generated continuum x-ray photons. We must also correct for absorption losses in the Si(Li) detector window and the inactive silicon layer at the detector surface (dead layer). Generation of Continuous Radiation. Kramers proposed a relation describing the intensity distribution of the
