tlie h i c k nall, but hon- niuch a n d what W I S the particle size of the CaSOl that \vas added to the cement which x i s used to make the mortar d i i c h was used to lay the h i c k s thnt make u p the \v:111, ; Ifacetious suggestion wis once made thitt because an alloy \vas no stronger than the grain boundary niatc~rial it contained, why not make alloys of the coniposition of the grain boundary material. This suggestion niay Fame day have merit as the p1ij.sic.s of the solid state are twtter understood and the reasons for the iiiovenient and pinning of dislocations, for example, are known. 1'hcn the analyst, will be asked to determine if there is enough of this or bhat compound presout to produce the desired effect, Hc will be askcd not only the aniount of the desired compound but how i t is distributed. For example, if it were possible to detwniine a cloud of several thousand atonis, some understanding could be obtained of the structures present in molten metals. The microprobe work a t the Kava1 Research Laboratory ( 2 ) and a t Irsitl (4) in Fr:ince is a step in this direction. l'hf- have learned mwli about the variation of the composition of alloys> cvcw though they have to look at the rather large area of a micron cube. So far, no one has even nttrnipted to look a t molten metals \vith such tools, although there is a gmving belief that the seeds of structurc,s s ; ( ~ ~ 1in1 solid metal may exist in thc nioltcn nictal and determine prolwrtic~s as surc.1)- as the genes p r ( w n t in the sc~cclsof plants tleterminc tho color of thc flo\\.ers. C'cxrtainly the growth and control of dmired structures in engineering materials \vi11 receive incrcming attention. The analytical chtmid will rcccivcs niore requests to tlrt(~niincthe small aniounts of nuclcating agent that shapcd the gronth of the structures. L\g:tin we will have to learn niorc ahout
separating ant1 estiniat,ing metallic conipounds. as well as elements. Metals of higher purity will also receive more of the anal!.st's attention in the futurc. For the opt'iniuni electronic and other properties. parts per billion and lrss of impuritivs become important. I n general. such analyses require new techniques, since concentration of large samples frequently is not possible because reagc,nt blanks, even on the excellent reagcnts available, become too large for comfort,. .\lso, the dirt elements-silica, alumina. antl lime-appear in ail!. manipulations that are carried out in a n or dinar^laboratory. Activation analysis has great promise in this area of parts per billion. Xumerous papers are beginning to appear and as reactors or other devices for activating samples become more common, the pioneering work that has been done under the auspices of the Atomic Energy Commission ( 5 ) will be extended considerably. Where this technique is not applicable, there is a need for more spnsitive reagents and,'or methods. The relativc :ibuntlance and cheapness of beta emitters from reactors could lead to a large usage of instruments containing them to provide cheap sources of x-rays for specific analyses. The work t h a t Muller described a t the 1959 LSL-Symposium sliould be studied closely by all analytical chemists interested in rapid methods ( 7 ) . I n recent years. gas chromatography has made radical changes in the analysis of gaseous and liquid mat'erials. The poor man's mass spectrometer has made the detection and estimation of very sniall quantities of many gases both rapid antl precise. The three common gases, oxygen, nitrogen, and hydrogen, beconic incrcasingl~.important in metals. Vscuuni fusion gas analyses have movctl from a research to a routine control tool in the last 10
years! hut Iwtter apparatus and procedurcs are iic~tled. Liken-isc. mass spectronwters are becoming easicr to operat(,. lcss expensive, antl morc versatik. H ( w the need is for rcwarch on different \rays of introducing solids into the instrunieiit to make it more nidcly applicable to the analysis of nietals. The principles inherent in it may offer solutions to some of tlie above problems. CONCLUSION
When the .\ward winner first became associated trith analytical chemistry, the tools of such chemists were a balance, a fen beakers, platinum crucibles and specially selected burets, and filtering funnels. If the present trend continues for four decades in the future, an analytical chemist who knows hon- to fold a filter paper to fit a funnel will be a n exception, but analyses of higher precision will be produced rapidly and in greater volume than we ha\-e ever known. However, there is a great need for specialized tools and the future of analytical chemistry lies in developing such tools. The analytical chemist of the past has n-orked in all branches of chemistry. To meet the needs of tomorroa., it looks as if he will have to work in all branches of physical science. LITERATURE CITED
H.F., =\sir,.CHEX. 2 4 , 1095-100 (1952)). 12) Birks. I,. d.. B r o o k .' IVitcleonz'cs8, S o . 3 , 62-65 (1951 1. (6) Lundell, G . E. F., Hoflman, J. I., J . Znd. Eng. Chetri. 13, 540-3 (1931). ( 7 ) Muller, R . H., LSV Symposiiim, 1950. I~ECEII-ED for rrvien .\ugllst 35, l!E9. .\rcepted .4ugllst 2 5 , 1959.
END OF SYMPOSIUM
Neutron Activation Analysis of Silicon Carbide LESTER
F. LOWE, HARRlET
D. THOMPSON, and J. PAUL CALI
Air Research and Development Command, Air Force Cambridge Reseorch Center, Bedford, Mass.
F A method has been developed for the analysis of high purity silicon carbide by neutron activation. After irradiation with thermal neutrons, the silicon carbide is decomposed with chlorine and oxygen a t 1250" C. The induced activities are collected and analyzed to give the impurity levels.
A
activation method for the determination of trace impurities in silicon carbide has been developed. The need for this method has come with the advent of silicon carbide as a semiconductor material whose impurity levels must be knon-n in the parts per billion range. NEUTROX
The genchral principl(bs involvcd in activation analysis have bwn reportctl (1, 2).
Because 5ilicon carbide hns a lolver microscopic absorption cross section for neutrons than silicon (0.09 barn for silicon carhide and 0.13 barn for silicon), and as it is k n o w i 16) that 10 g r x n s of VOL. 31, NO. 1 2 , DECEMBER I959
1951
silicon cause a flux depression of less than 1%, there is no reason in principle why silicon carbide cannot be analyzed by neutron activation analysis. The major difficulty with this compound is, however, that it is not readily dissolved or decomposed.
Element
EXPERIMENTAL
Preparation of Sample for Irradiation. Depending on t h e purity and t h e sensitivity required, a 100- t o 1000-mg. sample is used. It is washed with a 1 t o 1 mixture of hydrofluoric a n d nitric acids t o remove surface contamination and then rinsed \\ith distilled watcr. Grinding t h e sample must be avoided, as this lends t o serious contamination (Table I). Irradiation. T h e sample was irradiated a t the Brookhaven National Reactor a t an average thermal neutron flux of 7 x lo1*neutrons per second per sq. cm. The flux value was determined at Brookhaven by foil irradiation and has an uncertainty of f 10%. The nuclrar properties of the isotopes produced ( 5 , 7 , 8 ) are shown in Table 11. These elements mere chosen chiefly because their impurity levels were in the spcctrograpliic range, and hence a n independent check was available on the :ictivation results. Decomposition of Irradiated Silicon Carbide. T h e search for a niethod t o decompose silicon carbide was narlowed doirn to a catbonate fusion technique or chlorination a t elevated temperatures. T h e former a a s not used bccause the silicon carbide had t o be ground to 2 t o 300 mesh before fusion was practicable, and this resulted in serious contamination (Tsble I), and impurities wcre lost during fusion. For example, when a szmple of silicon carbide was trc:rtcd with a 90% sodium carbonatelO% sodium nitrate mix a t 900" C. for 2 hours, 8670 of the antimony-124 tracer added and 98% of the indium-114m were lost. The chlorinations were carried out a t 1200" to 1250" C. using a high frequency induction hcatcr n i t h a dccomposition train as shonn in Figure 1. Below these tempcrnturcs, no measurable decomposition observed The trentmt,nt of an irradiated sample of silicon carbide is as folious: The .ample is spread out on a quartz combustion boat and placed in the inner quartz combustion tube, which is placed within the graphite heater. The trap is filled with 15 to 20 ml. of 2 N hydrochloric acid to which have been added microgram amounts of stable isotopes of the elements to be determined. All connections are made and the system is flushed with chlorine for 5 minutes a t 2 to 3 ml. per minute. The graphite heater is protected with an inert gas. The induction unit is activated and the temperature of the graphite brought to 1200" to 1250" C. 1952
ANALYTICAL CHEMISTRY
combustion area. After combustion of the carbon residue is complete, the induction unit is shut down, and, after cooling, the decomposition apparatus is disassembled.
Table 1. Contamination of Silicon Carbide by Grinding in a Boron Nitride Mortar and Pestle ( e u l t s shown were obtained by neutron
iii
Mn
cu
activation .) Before Grinding, P.P.hI. 0.3 0.02
1.8 7.3 0 4
0 1
CO
Fe
0.02
780 120
Cross
Fraction Isotopic Abundance
1.4
Table
Target Xucleus A11166
Fe58 Si64
Section, Barns 13.3 0.98 1 6
Collection of Induced Activities. Several experiments were carried o u t t o determine the distribution of induced activities throughout the apparatus. I n one, 80 to 85% of t h e antimony-124 activity rcmained in t h e reaction tube and only 15 t o 20% &-as found in the trap. This experiment also established the fact t h a t
After Grinding, P.P.M.
II.
Nuclear Data
Isotope Formed by n, y Mn" Fe5@
1.00
0.0033 0.01
B
Figure 1 .
D.
E. F.
+ ZCI~
+02
4sic14
The chlorination is continued until the reaction is complete according to the following equation: 1200" c.
+ 2 C1, ---+
SiC1,
+
con f
C
1
0 2 -.,
Inlet to inner quartz member far Clz and Inlet to outer quartz tube for argon Transite plug Outer quartz tube Inner quartz tube High frequency induction coil, lepel Model T2.5-1
Sic
2 1: 1 0 : O . f
Decomposition apparatus for silicon carbide
c
C.
2 6h
LTax. Energy of B - , 1l.e.v. 2.8; 1.0;0.65 0.26; 0.46
F
Sic
A. B.
T*12 2.5h 47d
+C
The time required for complete reaction must be determined empirically for a given sample size and shape. A few typical reaction times are,0.75 hour for 100 nig. of pondered matcrial and 2 hours for a 15o-mg. solid picre. After chlorination, the unit is turned off and the chlorine is swept gently out of the system with nitrogen or argon. Xext, air or oxygen is blown over the carbon residue in the boat and the temperature is brought to a dull red heat. The carbon residue is converted to carbon dioxide in 3 to 5 minutes. Care must be taken as too fast a flow of oxygen or too high a temperature results in an explosive reaction between the carbon and oxygen which sprays the unreacted carbon outside of the
G.
H. J. K.
I.
Carbon heater Quartz boat containing silicon carbide 318 / 9 ball-and-socket joint Trap Exhaust to fume hood
volatile chlorides such as antimony(111) chloride or antimony(V) chloride can be complctely recovered. I n another, an irrndintcd szmple of silicon carbide containing mostly cobalt-60 was decomposed and thcn thp apparatus monitored at various points. Of the total activity GOOo remainrd within 2 inches of the rcaction boat and 81% remsinrd in the renction tube. On the basis of these euprriments the following proccdurc is Cscd. All parts of the conibustion apparatus, including the boat i r e thoroughly etched 11 ith concentratcd hydrofluoric acid. These etchings, with the trap solution, are put in a platinum dish and fumed several times to dryness with hydrofluoric acid to remove the silicon activity as silicon tetrafluoride. The residue is taken into solution with 6 N hydrochloric acid and transferred to a 100-ml. volumetric flask from which suitable aliquots are taken for the various analyses.
Radiochemical Separation and Measurement of Induced Activities. Standard radiochemical procedures (3, 6, 9) were employed and the final precipitate was counted for beta activity on counting equipment which had been previously calibrated with standard beta emitters. This calibration took into consideration coincidence losses, energy dependence, and the counting geometry; this made it possible to convert the raw count rate into disintegrations per second.
Table 111.
Element Ni Mn
cu
Zn
Sb hl0
,4 35
Results
Type Analysis Spectrographic” B C N.D. 3
0.1
