Journal of Radioanalytical Chemistry, Vol. 15 (1973) 173--179

I D E N T I F I C A T I O N O F S A F E - P A C K I N G M A T E R I A L S

D. J. R. EVANS,* P. L. KENNEDY,* A. K. BERGH**

*Atomic Energy of Canada Limited, Commercial Products, Research Division, Ottawa, Ontario (Canada)

**Royal Canadian Mounted Police, Ottawa, Ontario (Canada)

Neutron activation analyses of ground samples of safe-packing insulation have shown that dust from different sources may be differentiated by trace dement content. Between 10 and 20 dements were identified in each of 54 samples, and comparison of the activation "fingerprints" offers a good prospect for positively or negatively matching two or more samples.

Introduction

In recent years neutron activation analysis has been used to separate similar sam- pies from different origins by means of their characteristic e lemental concentration pattern. This fingerprinting approach has been used in such diverse areas as the iden- tification of oil samples, 1 narcotics, 2 archeological artefacts 3 and soils. 4

In general the technique is applicable to samples with bulk homogeneity, but does not provide positive identification. However, the more elements considered, and the better the intersample correlation the more probable is the correct assign- ment of identicality or lack thereof. The method does not require absolute measure- ments of concentration or indeed even positive assignment of gamma-l ines to iso- topes, and because it is comparative within individual samples for different elements and within a suite of samples for the same element some of the usual neutron act i - vation problems, such as sample detector geometry, are of less cri t ical importance. Furthermore, the absence of a given element (or gamma-ray line) may be as im- portant in the final analysis as its presence.

When a safe or vault is exposively entered the packing material around the in- stallation, which generally serves as fire-proofing, may be reduced to dust and scattered in the immediate vicinity. Some of this dust is often recover- able from the clothing or possessions of a suspect. However, the small quan- tities available for analysis (no more than a few mg) make conventional analysis difficult. The mul t i -e lement potential of NAA, together with the possibility of non- destructive assay thus permitting retention of the exhibit, suggests its likely useful- ness in this area.

J. RadioanaL Chem. 15 (1973) 173

D.J.R. EVANS et a l . : IDENTIFICATION OF MATERIALS

Experimental methods and results

Fifty-five samples of safe insulation, taken either as the raw materials or as east around an installation, were provided by the Ottawa Laboratories of the RCMP. Many of these samples were historical exhibits from actual cases.

In most cases the samples were supplied as chunks of definitely inhomogeneous, plaster-like material . These were reduced to dust by grinding for 4 rain in a CRC miU. In aU cases this produced a fine powder. Analysis of N 5 mg aliquots of this material showed it to be generaUy homogeneous.

For the "fingerprinting" experiments 5 tug of each sample was accurately weighed into a 0.2 ml polyethylene vial which was localized in the bottom of a SLOWPOKE reactor pneumatic transfer vial. Samples were each irradiated at 20 kW (thermal flux = 1012 n �9 cm "2 �9 sec-1) for 10 min in the Tunney 's Pasture SLOWPOKE Re- actor, allowed to decay until the 28A/( the major activity at reactor removal) had decayed to such an extent that the counting system deadtime was approximately 1~o. The counting system used consisted of a 35 cm 3 Ge(Li) detector, pre-amp and ampli - tier connected directly to a PDP-9 computer system. The Spectanal spectral anal- ysis programme used has been described by T h o m p s o n. 5

Following a 10 rain count the samples were reirradiated for 8 hrs at a flux of 1012 n �9 cm "2 �9 sec -1, allowed to decay for approximately 72 h a and counted for 25 rain on the above system.

The basis for comparison of samples in order to demonstrate the possibilities of the technique are merely a tabulation of saturation activities in counts �9 mg "1 �9 sec-1 for the prominent spectral peaks. This data is given in Table 1. The samples are list- ed in order of increasing aluminium content. This was an arbitrary device chosen because aluminiurn occurred in all samples and had a wide (2000-fold) range over the entire suite of materials (c. f. Ca with approximately 10-fold range). It can be seen that A1, Mn, Ca, Na, K, V, Sin, La, Sc and Cu were observed in aU or n~.arly all samples. Mg, Sr, Ba and Ti were observed in many samples and would have been even more universally observed if counting equipment problems had not required high settings of the lower energy discriminator cut-off for some samples (marked x in Table 1).

A gamma-ray line was observed in nearly every long irradiation sample between 555 and 562 keV. This could be attributed to 76As, 82Br or 122Sb or combinations of these elements. In most cases no significant peak was observed at 777 keV (82B 0 or 1. 691 MeV (124Sb which should accompany ]22Sb), and the observed ~eak gener-

aUy had an energy listed as 559 keV. Consequently this was attributed t o ' 6 A s , though the tabulated figures may be somewhat in error due to small contributions form 82Br or 122Sb.

140La was observed in aU samples (except No. 8) and the 328 keV gamma- l ine from this isotope was not resolved by the analysis programme in some cases (marked * in Table 1) from the 321 keV 51Cr line.

174 I. RadioanaL Chem. 15 (1973)

D.J.R. EVANS c t a l . : IDENTIFICATION OF MATERIALS

Co pjper values were obtained from the annih i la t ion radiat ion ac t iv i ty corrected for a ~'~'Na contr ibut ion. Another possible contributor to the 0. 511 MeV peak is 65Zn

but this was e l imina ted since in aU cases the observed signal at 1.12 MeV could be at t r ibuted to 46Sc.

Despite the foregoing it should be emphasized that the purpose of the exper iment was to show a character is t ic g a m m a - m y profi le or fingerprint. It might therefore be argued that assignment of g a m m a - r a y energies to par t icular isotopes is not necessary

provided a l l measurements are carried out under s imi lar i r radiat ion, decay and count- ing histories.

Also l isted in Table 1 are the manufacturers of the various instal la t ions studied,

and i t can readi ly be seen that in some cases these tend to be grouped based upon a luminium content e . g . Mosler and Security Safe. However, if samples were rank- ed by another e l emen t (Na or Mn, Sm) this trend would not be evident .

Tab le 2 is a l isting of the concentrat ion ranges observed for a number of the gen-

eral ly occurring e lements . It can be seen that the measured range of most of the e lements exceeds a factor of 50. Furthermore, there is a trend (refer Table 1) for high values of different e lements to occur in the same samples e .g . , although Table 1 is ordered by A1 content , the highest l e v d s of Mn, Na, Ba, Sm, Cu and K tend to occur in the last few samples l is ted.

As a test of the method six of the samples were s d e c t e d and prepared for analysis by an independent and hitherto uninvolved worker. The samples were subjected to the short i r radia t ion history assay only. Four of the six were correct ly ident i f ied by comparison with Table t , one was ten ta t ive ly assigned correct ly and one was assign- ed to two possible sources, one of which was correct . It should be noted these last two were both from the same manufacturer. The results obtained are shown in Table 3. Unknown E, assigned as No. 6 or No. 7, was in fact No. 7. It can be seen that in

nearly a l l cases there was one or more d e m e n t which showed considerable var ia t ion, and this can be at t r ibuted ei ther to inhomogenei ty or to the r d a t i v d y low count-ra tes (and therefore high standard deviat ion) for some measurements. However, in genera l the technique permits posit ive ident i f ica t ion of a common source for two samples

with a reasonable degree of probabi l i ty . Two samples of room dust were also studied and their results are also shown in

Table 1. Sample I was taken from dust accumula t ion underneath a computer output dev ice and sample 2 was sweepings from a basement concre te floor. C-enerally the same elements a re observed as in the safe-packing mater ia ls but again the e l emen ta l profi le tends to be unique.

Conclusions

The results presented show that neutron ac t iva t ion analysis can identify 10 - 20 e lements in character is t ic safe-packing mater ia ls , and comparison of the ac t iva t ion "fingerprints" of samples offers a very good prospect for posit ively or negat ive ly matching two or more samples.

J. RadioanaL Chem. 15 (1973) 175

I

0

,g

"N

o~ "0

U

t~

O0

N

' ~ ~ . . . . . . ~ ~ " " ~ o 0 0 0 0 0 0 ~ ~ n O t ~

~ I O I I I I I I H I I N ~ I ~ I ~ ~ I ~ N I

d d A ~ d 4 4 d 4 4 ~ N ~

I

~ ~ ~ ~ N ~ ' ' ~ ' ~ ' ' ~ ' ~ ~

~ N N N ~ ~

176 J. Radloa~L Chem. 15 (1973)

~ o ~ ~ o ~ o ~ ~ ~ o ~ o o ~ o ~ ~

g ~ N a

~ ~ ~ �9 ~ ~

J. Radioanal. Chem. 15 (1973) 1 7 7

1 2

Table 2

Elemental concentrat ion ratio ranges*

Element Saturation act ivi ty range counts , mg "1 �9 sec-1 Not detected

Aluminium

Manganese

Ca lc ium

Sodium

18 -2040

2 - 390

2 - 17

10 -1880 Vanadium

Samarium

Potassium

Lanthanum

Copper Scandium

Arsenic

Barium

Titanium

0 .4 - 40

1.7 - 89.4

1 - 32

0.1 - 6.8

1 - 138

0.7 - 57

0.3 - 6.4

1 - 17

0.3 - 25

0/53 0/53 1/53 0/53 3/53 o153 3/53 0/53 4/53

3/53

3/53

7/43

5/43

*Sample No. 1 has been excluded tram this table since it was purely organic filler and the very low elemental concentrations measured tend to falsely increase the apparent ranges.

Table 3

Identif ication of unknowns

Unknown

A

24

C 17

D 19

20

40

12

~ssigned as A1 Mn

33 592 4 4 580 48

220

234

57.6

64

193

198

40.0

38.9

39.0

250

189

128 10

143

4

8

39 8.4

40 8.0

2 9.2

2 10

3 12

210 9 .2

166 8.0

Ca Na M~

6.4 130 2.

6.4 143 3.

40 60 x 9.1 35

9.2 8 8.

9.2 10 7.

170 15.

197 16

- 3

12 i 4

10 3

10 i6

2. x

St

4

4

x

x

22

24

3

~0

I-4

~0

m

x

U Ba

~9 68

~8 7 . 6

7.2 x

6.8 x

0.8 4

2 6

3 10

4 14

0.8 3

0 .4 4 .4

0.8 --

7 .6 7 .6

6 x

Ti

9.6

4.8

0.8

0.4

0.8

1

0.3 m

2

X

178 J. RadloanaL Chem. 15 (1973)

D.J.R. EVANS et al . : IDENTIFICATION OF MATERIALS

In actual forensic work more detailed study of individual samples would probably

be required and in this case the irradiation decay and counting histories chosen could be optimised for the particular samples under study.

There is some doubt that the dust and debris caused by explosive entry to a safe

installation would have the same composition as the powder obtained when the insula- tion material is ground in the laboratory.

References

1. V. p. G u i n n , S, C. B e l l a n c a , Proc. Intern. Conf. on Modern Trends in Activation Analysis, Gaithetsburg, Maryland, 1968, p. 93.

2. A. K, P e r k o n s , R.E. J e r v i s , proc. Intern. Conf. on Modem Trends in Activation Analysis, Gaithetsbutg, Maryland, 1968, p. 256.

3. D . I . R . Ev ans , R. W i l m e t h , AECL Report, CPSR-314, 1971; Abstract of PapersANAL 0"/6, 161st ACS National

4. C . M . H o f f m a n , R . L . B r u n e l l e , K . B . S n o w , M . J . P r o , Intern. Conf. onModem Trends in Activation Analysis, Gaithetsburg, Maryland, 1968, p. 251.

5. C. J. T h o m p s o n , Nucl. AppL, 6 (1969) 559.

J. RadioanaL Chem. 15 (1973)

lZ*

179