V O L U M E 23, NO, 9, S E P T E M B E R 1 9 5 1 content of various mixtures of zirconium and hafnium chloride solutions. The results are given in Trtble I. COMPARISON WITH OTHER METHODS
1tientical samples were analyzed for hafnia content spectrographically, by the selenite method. and by the p-bromomandelic acid method. The results given in Table I1 are averages of at, least two samples.
Table 11. Comparison of Methods
1261
Determiiiations made in the pr'eaence of aluminum, ferric, ant1 titanium I V ions gave erratic results; hence, for best results determinations should be made using a pure solution of zirconyl and hafnyl chlorides. .4 preliminary- recrystallization of the mychlorides from strong hydrochloric acid (8 to 10 31) will eliminate thes? interfering ions. In spite of these shortcomings, the results compare favoral)l~. with those by the selenite method, and require only one f o u i ~ l i the time, and the tosic hazards of selcnium are eliminated. Experiments using mandelic. acid and other glycolic acicl derivatives as reagents to determine zirconium-hafnium ratios are noiy in progress.
Per Cent Hafnium Oxide Calculated 95. BO 75 00 4 0 00
36 35
SpectroscoDic g5.9 ,4.l 38 2 LO 7 2 75
Selenite 95.4 75.3 40 7 32 3 2 2
p-Bromomandelate 95.6 75.6 40 1 21 4 3 2
ACKNOWLEDGMENT
The author wishes to thank Cyrus Feldman, Oak Ridge Sational Laboratory, for the spectrographic analyses, J. J. Klingenberg. Xavier University, for his procedure of the synthesis of p-broniomandelic acid, and Boyd Weaver, Oak Ridge Sational Laboratory. for supplying the pure zirconium and hafnium compounds usetl in this study.
DISCUSSION
I f the above procedure is followed carefully, a precision and accuracy within = ! ~ 0 . 5 7absolute ~ can be expected in samples containing more than 10% of hafnium oxide. In the lower hafnium range the precision and accuracy fall off. The method has a tendency to give slightly high results for hafnium, probably Ilerause of a very slight hydrolyPis of the tetrabromomandelate precipitate. Precipitates of theoretical composition are obtained only in hot solutions strongly acid with hydrochloric acid (about 2 :V) and in the presence of a two- to threefold excess of p-bromomandelio acid. The presence of a certain amount of free sulfuric acid aids i n ohtaining precipitates of constant composition ( 5 ) .
LITERATURE CITED
hten A. €I. W,, Nederland. Tydsch. Satuurkunde, 10, 257 (1943). Birks, L. S.,and Brooks, E. .J., ANAL.CHEM.,22, 1017 (1950). Claassen, A. A , , Z.anal. Chem., 117, 252 (1939). Feldman, C., ANAL.CHEM., 21, 1211 (1949). Gump, J. R., and Sherwood, G. R., Ibid., 22, 496 (1950). Oesper, R. E., and Klingenberg, J. J., Ibid.,21, 1509 (1949). Schumb, W. C., and Pittman, F. K., Ibid., 14, 512 (1942). (8) Van Hevesy, Georg, "Chemical Analysis by X-Ray and Its -4p plications," New York, McGraw-Hill Book Co., 1932. ( 9 ) Wernimont, G., and De Yries, T., J . B m . Chem. SOC.,57, 2386 (1) (2) (3) (4) (5) (6) (7)
(1935). RECEIVED M a y 16, 1951. Presented before the Division of Analytical CHEMICAL SOCIETY, Chemistry a t the 119th Meeting of the AUERIC.AN Boston, Mass. hrithor is on loan f r o i i i K a g n P University.
Determination of Beryllium by Photodisintegration A . M. GAUDIN AND J A l I E S H. PANNELL' Mineral Engineering Laboratory, Massachitsetts Znstiticte of Terhnology, Cant bridge, .%fnss
Analysis of beryllium in low grade beryl ores and ore treatment products was found to be difficult and tedious by ordinary chemical or spectrographic techniques. The unique nuclear properties of Be9 suggested exploitation of the y , n reaction and this was done by using, conventional electronic equipment and a portable -(-ray source. Good results were obtained in the analysis of many samples run h y nonprofessional workers. Interference by elements with very high absorption cross sections was studied and this effect of boron measured. The method is rapid, simple, and nondestructive.
A
PRELIMINARY investigation into the determination of beryllium by the y,n reaction has indicated the sensitivity and general application of the method. As i t can be applied directly to practically any beryllium-containing material, without destruction of the sample, it gives promise of supplanting conventional chemical methods under certain conditions. -4gamma source of at least 1.63 m.e.v. energy and about 1 cwie intensity is needed to irradiate the sample; SbI24is suitable for this purpose. Interaction of the photons with Be9 liberate? 1
Preaent addreae, American Cyanarnid Co., Idaho Falls, Idaho.
neutrons which are moderated and then counted i n boron t1.ifluoride-filled detectors. With reproducible sample geometr).. the neutron counting rate is a measure of the beryllium content, unless sufficient boron or cadmium be present to cause interference by absorption. For ut,most' sensitivity, large samples art' used and in this way 1 to 2 p.p.m. of beryllium can be detected. Beryllium is one of the less easily determined elements, generally because of the difficulty attending its complete separation from interfering metals, especially if tlicw are preiient in rrLlativel? lwgr concentrations. Thus, the colorimeti,ic methods (8) ofrc.11 require repetitious precipitations for the removal of iron ailti manganese, and the gravimetric method can be tedious in tho presence of aluminum. Spectrographic* methods have proved successful in certain applications ( I , 2 ) . The unique propertitv of the beryllium nucleus have suggested determination of thc, metal by photodisintegration. The smie nuclear reaction has been proposed for sorting beryl on the industrial scale ( 3 ) . As a physical method, the one described here has the advantage that dissolution of the sample is unnecessary : moreover, the sampling error, encountered in spectrography, is also minimized. The photodisintegration or y,n reaction consists of the absorption by a nucleus of a high energy photon followed by the releast, of a neutron; the threshold energy for the reaction is the binding energy of the neutron. For most elements the nuclear binding
ANALYTICAL CHEMISTRY
1262
energy is of the order of 8 m.e.v., while it is only 1.63 m.e.v. for Be* and 2.2 m.e.v. for H'. As beryllium contains the nucleus with the lowest binding energy, irradiation of any mixture of elements with photons of energy between 1.63 and 2.2 m.e.v. will release neutrons from beryllium only. For determinative work, therefore, this reaction has the ideal attribute of specificity.
result from use of DsO in place of psraffi as a moderator, but these refinements were omitted in the interests of simplicity and economy. A photograph of the detector army is shown in Figure 2. The samples were handled by remote control. The neutron detector tubes inch in diameter filled to 28 em. of mercury pressure with boron trifluaride containing enriched BO ' were obtained from the Oak Ridge National Laboratory. They were wired together and their output was fed to a single one-tube preamplifier, then to an Atomic Instrument Co. Model 204A amplifier. A positive high voltage supply of the saturable reactor type, made by the Nuclear Instrument and Chemical Cor was used and the pulses were counted io a standard scaler w i t h h e usual scale of 64. In order to test the srrawement described, low-beryllium mix-
1 5, 10, 25, and 50% beryl. Th& kixtures were tested in the photodisintegration analyzer RS well as by chemical methods by a firm of reputable analysts.
T h e chemical results were rather disappointing, as may be judged by Table I, while the photodisintegration analyses agreed more closely with what was expected, Table I shows that the chemical analyses were low for the low-grade samples and high for the high-grade samples. Table I. Figure 1. Cmss Seotion of Annular Sample Holder i n Position Amund Gamma Souroe
Beryl Added.
."
%?.
1 5
in
25
en The practicability of applying the photodisintegration reaction is dependent on overcoming the practical difficulties caused by the small cross section which the beryllium nucleus has for this reaction (4+i.e., about 1 millibar" (lo-" sq. om.) as compared with about 1 barn ( 1 0 - 2 4 sq. cm.) for most neutron absorption reactions. A consequence is the necessity for fairly intense gamma-ray source and an efficient means for detecting the neutrons emitted. The magnitude of the problem may be gaged by edculating the number of neutrons emitted by 1 gram of a substance, containing 1% beryllium, if irradiated in 1% of the directions of space (0.04 r steradians) by a source of gamma rays of 1curie. Assuming all of the photons to have the proper energy, the number of interactions will be
Chemioal and Photodisintegration Analyses of Mineral Mixtures Containing Beryllium Normalised Neutron Neutron Activity. Activity Net Counts/Mih./ CountdMin. % .. Beryl 182 182 877 175 1885 168 4223 169 ~
8427
168
Chemical Analysis, % Be0 0.07 0.53 I.? 3.1
7.3
Calcd. Corresponding
7%
B e 0 in Beryl 7.0 10.6 12.0 12.4 14.6
I n view of the unsatkfactory chemical results presented in Table I, another sample of the same stock beryl diluted with quarta was analyzed by a different commercial laboratory. This malysis, which gave 12.0% beryllium oxide for the stock mate-
>
If neutron countem can he obtained that have a detection efficiency of 2% one would thus obtain 5 counts per second under the conditions of the above example. Further improvements would depend on improving the irradiation and detection geometries. EXPERIMENTAL
For good irradiation geometr the g a m m a source was virtually surrounded by the sample; t i e source was placed on a short pedestal, and annular sample holders were slipped over it. As source, two 0.5.ineh spheres of radioantimony were used, These spheres had previously been activated in the Oak Ridge pile to give approximately 1 curie each from Sb'*4, and therefore about 0.76 curie total (6) of gammas above the 1.63 m.e.v. threshold. The details of the source arrangement are shown in Figure 1. Surrounding the sample holder was a lead cylinder 4 cm. thick, which reduced the transmitted gamma intensity to about one eighth and thereby facilitated operation of the neutron detectors without appreciably weakening the fast-neutron flux. The neutron detectors were eight in number and embedded in wax in cylindrical disposition about the lead shield. Better results could be obtained by substitution of bismuth for lead, as the farmer has a lower absorption coefficient for slow neutrons. Similarly, an improved neutron counting rate would
Figure 2.
Detector Array, W a x Block, and Lead
House
1263
V O L U M E 23, N O , 9, S E P T E M B E R 1 9 5 1 rial, agreed fairly well with the average of 11.3% beryllium oxide reported for the five samples of Table I. The value of 12.0% beryllium oxide was accepted as the correct analysis for the beryl stock. Most of the annular sample holders, consisting merely of coaxial Lucite cylinders with end caps, were designed to produce the optimum counting rate from the quantity of sample available. One of them is shown disassembled in Figure 3, together with two other types. I n order to facilitate reproducibility of sample geometry, the holders were filled to the top for each measurement. As the gamma source wm at the center of the sample, small differences in its height were unimportant, but if the holder had been but half filled, its exact height would have been critical.
which could be filled by the sample. Under the conditions in the analyzer i t may be accepted that there are no appreciable differences in absorption of gammas from one mineral sample to the next, regardless of differencesin density. Therefore, all readings made with the same holder may be divided by sample weights t o obtain counts per minute per gram. The procedure, then, was t o fill the holder, count t h e neutrons, weigh the sample, and normalise readings t o a I-cram basis. Standards were treated in the
Figure 3. Annular Sample Holder, Disassembled and Assembled, and Two O t h e r Types of Holder This is shown by Figure 4, which presents neutron counts ES a function of the filling of the sample holder. In all three instances the lack of linearity in the relationship between sampleholder filling and activity is evidence for the critical n a t u e of the extent of filling of the sample holder. It became necessary, therefore, to have sample holders of various capacities and to choose one
tt
PULSE OISCRIMINATION L E V E L S
/
15 V O L T S
Figure 5. Correlation of Neutron Activity w i t h Beryl C o n t e n t of Standard Mixtures Baokground wm generdly of the order of 10 t o 20 counts per minute and seemed to be genuine neutron counts. Typical net counting rates of standards (the artificial mixtures of quartz and beryl) are shown in Figure 5. Two detectors only were connected for thme readings, because the gamma source was a t a high intensity. Figure 5 shows that there is strict proportionality
10,000
0 20 V O L T S
,
Table 11. Comparison Between Analyses of Ore Testing Produots Per Cent Be0
B 3004 B 4005 B 6090 M 4341 11 4371 B 1071
Type of By photoProduct disinteerhtion Headsample 0.071 Head sample 0.11 Headsample 0.22 Headsample 0.25 Headsample 0.066 Head a a r n ~ l e 1.7 Head sarn~le 0.76 Concentrate 0.52
B 1081
Concentrate
0.83
M M B M
Conoentrate Conmentrate Middling Middling
5.2 8.9 0.14 1.35
ShrnPk NO.
B 1002
B 2003
I"
Figure 4.
2"
3'O
Chernioal
0.095
...
0.14
6.26 0.30
0.078
.... .... .... .... .... ....
0.18
....
b
Analytical laboratory A. AnaLytiod laborstory B.
1.7" 0.73" 0.63" 1.84b 0.86. 2.4nb 5.6'
9.4*
...
0.98"
...
....
0.04"
.... a
... ... ... ...
0.0033 0.036 0.0072 0.028 0.025 0.035
4',
HEIGHT T O WHICH HOLDER I S FILLED Effect of Sample Holder Filling on Neutron
Aotivity
4345 4346 1086 4344
Speotrogranhic
... ... ... ... ...
0.061
ANALYTICAL CHEMISTRY
1264
offenders is boron, particularly because the mineral tourmaline, between beryllium content of the sample and the observed net containing about 37" boron, is often associated with beryl in counting rate. In particular, the data indicate no appreciable pegmatites. counting loss at high counting rates. For these high rount,ing In order to determine the effect of boron, known amounts of rates and to avoid difficulties with t,he register, a scale of 512 was fused borax were mixed with beryl-quartz standards and counted, used. .Is would be expected for proportional counters arid a fast with t,he results shown in Figure 6. Small additions of boron reamplifier, the resolving time was very short. duce the counting rate, the reduction being some 10% for 1% Of the many sampla assayed 1)y thip method, check analyses boron; larger amounts are relatively less damaging. It may be were requested on only a few. Thcw are presented in Table 11. accepted, in view of the relatively small cross section of boron Some discrepancy is apparelit i n roncmtratcs 1071 and 1081, but the over-all agreemelit is eiivouragiiig, tqxcially in the IOU for fast neutrons (about 2 barns us. 600 barns), that the boras mixed n-it,h the sample absorbs the thermal neutrons in the ranges of concentration. A check o i l int,ernal consistency is prosample but that the fast neutrons art' allowed t o pass into the vided by ralrulatioris of material balanws in tht. ore dressing wax and thereafter to actuate the neutroii counters. tests. -4typical example in Table I11 shows a calculated head Further investigation of interference by absorption of iic'uof 0.Z; in comparison with a f e d assay of 0.25 beryllium oxide. trons was not expedient,, but the authors believe the effect of such loa( 1 interference can be overcome. One method for so doing might be to use a cadmium sleeve around the sample holder. Cadmiuni has a cross section of about 5 barns for fast neutrons and of several thousand barns for thermal neutrons. The use of a cadmium 996 8 k sleeve around the sample as a lining of the sample holder would do 0 5 % BERYL what horax does when mixed with the ore. .4t the price of a slight loss in sensitivity, the effect of variations in thermal neutron absorbers in the sample could be largely overcome. ,411other solution might be in the use of a small antimony-beryllium I neutron source, in place of the antimony alone, in order to obtain I values for neutron absorption which could he applied as rorrection factors.
!
ClATERIALS ANI) EQUIPMENT
86-
8
4
/
,
1
02
0 4
, , 06
08
I
LA12
IO
14
16
PER CENT B O R O N A D D E D Figure 6. Effect of Interference b y Ueutron ibsorber
'rahle I l l .
To those who know the comparative simplicity of assay by Geiger counter, a word of caution may be necessary. The art of making and using neutron detectors does not appear to be developed sufficiently, at this time, for their ready application to prccise routine measurements. While a completely untrained person might have no difficulty in operating an instrument for incasuring radioactivity, he would have considerably more difficulty in counting neutrons. The instrumentation problems are lessening, however, especially since the advent of industrial production of boron trifluoride-filled tubes and fast linear amplifiers. The equipment nocdcd nia>-be rcipreseritcd bv meane of x iilocli diagram (Figure 7 ) .
Materials Jialanre i n Concentration of Beryl Weight
Distribution, Prod ii c t Conrentrate 1 Concentrate 2 Concentrate 3 Tailing Slime fraction Caloulnted head analysis Actual head analysia
c.;
%L
Re0
G i ? 8 .1 JR.!~
0 0.5 0 Z.4 o.ii
20.0
__ 3.3
1on.o
Units Be0
Be0 rlietribution. I " r,8 4 26 1
0 . 132
n.068
0.029 !l.lio.,02
n.14
0 .Yti 0 25
"c
0 003
0 ?fil
Amplifier
10.8 3 . ;7 100 0
I
Register
- ' I -
~
IOscillograph 1
~
INTERFERENCE BY NEIJTRON ABSORBERS
Most of the neutrons emitted by the beryllium leave the sample tts fast neutrons, penetrate the lead, and enter the paraffin wax with little loss in energy or diminution in numbers. In the wax they are slowed to thermal energies Rithin a few centimeters and thereafter diffuse until absorbed. bome of the thermal neutrons will diffuse back into the sample, and some inay be produced by moderation within the sample. One may consider these neutrons to be in dynamic equilibrium with those diffusing through the wax and lead, so that if the sample were replaced by a neutron absorher, or neutron sink, it would affect the neutron population in the wax. This is precisely the mechanism by which an element with a high absorption cross section for slow neutrons may interfere by reducing the neutron counting rate. Fortunately. very few elements have such cross sections, data on this point I)eing extensive and readily available. One of the worst possible
Figure 7.
Block Diagram of Photodinintegration inal>-zer
Detectors. Boron trifluoride-filled detectors, containing BLo, have an intrinsic efficiency of several per cent for detection of incident slow neutrons. The type used in this work was made at Oak Ridge National Laboratory and wm 7/8 inch in diameter by 10 inches in length, Others, of larger diameter, available commercially, were serious1)- affected by intens? gamma radiation. iilthough the background ionization current in the Oak Ridge detectors was not negligible, this difficulty was overcome by use of an appropriate discriminator. On the other hand, the effect of irradiating the larger tubes was to reduce the neutron pulses to small size and also to introdwe a high background, so that they were not selected. Fission chambers and boron-coated tubes, either or both of which might give better performance, were not tried.
V O L U M E 23, NO. 9, S E P T E M B E R 1 9 5 1 Preamplifier. The unc~-tubepreaniplifier, or cathode follo\vei' derriihed by Jordan and Bell 15)was found satisfactory, although frequvnt replacement of electron tuhr 6.1Ii5 seemed necess -4C ' e n t ralab ceramic. c o i i i i c ~ i i w\rith a rXOO-volt rating \vas for tlt~c~oupling. Amplifier. Thr .ltoinic Inqtrunient ('o.'s .\lode1 201.l iq corisitlriwl excellent i n this application. From the various output. it was posible, simi~ltanc~ously, to count pulscs, to vie>\\theln o i i an oscjllogrdph, aiid to nionitor them with a counting rate inester. A discriminator, wliich is e.ssentia1 in nrutron munting, i. a part of this a er. It was used to determiiicl the, optimum gain or pul-r niinatioii lc~velfor obtnining r~c~pr~oducilJl(,rewlts. Scaler. A convt:nti~iiralkfigiritwthaiii waler \vas i i s d , with 11 scnlt~of 64 or 512 ac*cwtling to thc counting rate. A r)u\lont Type 248A oscillograph nith triggcwtl sweep of various duratioii, u'aq e~xcedinglyhelpful t hivughoul t lie u-ork, T i 1 oi,tler to set u p a lalxxator>- for ilrterrnining beryllium by tlii. ~)hc)todieintegratioiinit,thod, several instruments arc>needed.
In addition, one must liiivc. H high-intrnsity gamma source and nimiis lor handling anti nioiiitoring it. The source ran he had froin the Atomic Energy Coniniission (as ratlioantinion!~) at a nomirial cost, as 1 month's irradiation of ti thimble filled ivith antimony producw an activity in excess oi 1 curie. Thc :tut1ior.s' n-ork started with 2 ruriew and continued for ci nionths (tlirtvs hrtlf-lives), aftclr \vlii(*h thc3 0.25 cwie remaining continued to lx uvful. Except over v w y long periods, radioantimony is far l e n r expensive than radium, but it appears to have ariot1ic.r outstailding advantage. Ordinary radium sources emit sonic. neuti~i~ns as a result nf u,n reactions with light metal inipuritic+ contained in them. The neutron background from the I-grain ~ authors was found to reduce thr racliuni so>.ircestried L J the sensitivit,?;of the method 1 1 ) ~ :til order of magnitude. Should on(% havr :t Van de Graafl or other type of high-voltage x-ray generator available, even hetter results would be ohtained, for the beam could t)ta directed at, the sample and riot at t,he detectors. The.
1265 effrct of irdiatioii on the tulicts 1ra.4 to cause sonit: reduction i l l pular sins, a diminution in length of the voltage plateau, and ail
incroast. i i i t)ackground "noisr." all of which are hindranws t o stable opcaration. I n spite of the experimental difficulties, the authors believe this nichthod has many attractive possibilities. It may, for inatanw, be adapted to automatic- assaying by means siniilar to the automatic sample changers ( 7 ) used in routine beta counting. As ail entirc'ly instrumental nicthotl &in, almost, to the measuremelit o f y H . i t may be applicahlr to plant voiitrol. kCKNOU LE:DG\IEN'I'
Thc. authors are most appreciative of the intcwst and advice of Clark Goodman and Matthew Simds, Mmsacliusetts Institute of Technology, and of R. TV.T>odsoiiand I!. IV. I k v e n s , Jr., Cohinibia University. L I l E H . \ ' L ' l ~ K ECITEL) 1) ('holak. .J., a i d Huhhard. D. lf., I s . 4 ~ CHWM., . 20, 7 3 (194Si. (2) I h i d . , p. 9 i O . ( 3 ) (iaudin, .I.51..Dasher, .J.. Pmnell, ,J. H.. and Freyberger, \T', L., AvMining Eng., 187, 4 9 5 4 (1950). :1) Guth. E.. and IIullin, C . .J.. P h y s . Rtu., 74, 833 (1948). (*5j .Tordaii, W. H.. and Hell. P. R . , Atomir Energy Conirnissi~~ii, I
MonP-323.
( 6 ) Kei~n.13. D.. Zaffarano. I). .J.. :itid lIitchell, ..I.C . G , , P i i ~ s KIJO., . 73, 1142 (1948). ( 7 ) P e ~ ~ ~ W, c l iC,' , , r t d Good, \ V , AI., Rw. Sc.1'. I / i s t ~ t r ~ e17, ~ ~255 t~, i1946 1. 'X)
Saildell. 14:. I
