Application of X-Ray Fluorescence Methods to the Analysis of Zircaloy R. W. ASHLEY and R. W. JONES Chemistry and Metallurgy Division, Chalk River project, Atomic Energy of Canada lid., Chalk River, Ontario, Canada
b X-ray fluorescence methods have been investigated for the determination of tin, iron, chromium, and nickel in Zircaloy-2. In one method analyses were made an samples in the form of oxides briquetted with cellulose powder. In a second method the bulk of the zirconium in the sample was removed by extraction with thenoyltrifluoroacetone. The alloying elements were concentrated by precipitation and briquetted with cellulose powder. Both methods are precise and accurate enough to be utilized to check compliance with specifications. The direct determination of hafnium in Zircaloy has also been investigated. The sensitivity for the hafnium Lyl line, which is the only one relatively free from zirconium interference, is too low for determination of hafnium a t concentrations below 200 p.p.m. Elimination of interference of the zirconium K series with the hafnium L series by pulse height discrimination is impractical for the low concentrations of hafnium involved.
T
m use of Zircaloy-2 in nuclear reactors has resulted in a demand for large numbers of analyses to determine if material meets specifications. Spectrographic analyses have served well for determinstioh of most trace impurities, but certain impurities plus those elements incorporated as alloying constituents, have had to be determined by wet-chemical methods. These methods are time-consuming, because several samples of each batch of alloy generally have to be dissolved and treated to provide a complete anslySiS.
The four alloying constituents in Zircaloy-2 are present in amounts for which the x-ray fluorescence method appeal.ed practical. These alloying constituents are tin (1.25 to 1.65y0), iron (0.07 to 0.17%), chromium (0.06 to 0.14%), and nickel (0.03 to 0.07%). The first method explored involved simple conversion of the Zircaloy to the oxides and analysis in this form. A second method involving the separation of zirconium and concentration of the elements of interest was also investigated in an attempt to increase the sensitivity and tm study the feasibility of determining some of the impurity eie1632
ANALmcAi CHEMISTRY
were obtained with the mandelic acid precipitation but difEculty was encountered in concentrating the alloying elements by hydroxide precipitation because of the presence of excess mandelic acid in the filtrate. This scheme was discarded in favor of the solvenr extraction separation. Moore (5) described the extractiw behavior of zirconium from aqueous solutions of nitric and hydrochloric acids using TTA, but was concerned EXPERIMENTAL with only milligram amou:% of zirconium. Because the S J Z ~0; sample Instrumentation. A Philips, 3required to give sufficient intensity of specimen x-ray fluorescence spectrograph, Model 52254, with specimen fluorescent x-rays for the determination spinner-and helium path attachments of the alloying elements was of the order was used in this work. Conditions of 0.5 gram of Zircaloy, it was necessary used during excitation were: to determine conditions for extraction for these considerably larger quantities. X-ray tube Philips FA-60, W target, The zirconium extraction efficiency is operated at 50 kvp. and dependent upon the TTA concentration 50 ma. in the organic phase, the organic to Analyzing LiF ( 1 0 0 plane, 26 = 4.0276 aqueous volume ratio, and the zirconium Crystal A.) Collimators Primary, 4 inch x 1/8 inch concentration in the aqueous phase. parallel plate The practical limit for the TTA conSecondary, 4 inch X 0.005 centration was found to be about lM, inch p a d e l plate Detector Philips, Type 52245, scintilsince difficulties with phase separation lation counter (NsI crysoccur a t higher molarities. The total til, TI activated) volume of aqueous and organic was chosen for convenient handling in a An Atomic Instrument Co., Model 125-ml. separatory funnel. This, and 510, single-channel pulse height anthe limitation of a minimum sample size alyzer waa used in conjunction with of 0.5 gram of Zircaloy, governed the the scintillation counter to eliminate volume ratio and the concentration of unwanted radiation where possible. Preparation of Samples. METHOD zirconium in the aqueous phase. All 1. Samples were converted to the extractions were from 6 N hydrochloric oxides and briquetted with cellulose acid solutions to minimize the extracpowder. Samples were generally retion of iron. ceived as metal chips or turnings and The organic phase was checked for the 1 t o 2 grams were converted t o the presence of iron by backwashing with oxides by heating in a muffle furnace concentrated hydrochloric acid (6)and for about 1 hour at 1000° C. The analyzing this “backwash” solution for oxides were ground to -300 mesh iron. No iron was detected. Accordin a boron carbide, mechanically ing to Moore (6) tin, chromium, and operated mortar. A sample of the nickel are not extracted by TTA. oxide was weighed out and an equal weight of cellulose powder (Whatman Conditions for Sample Treatment. ashless cellulose powder for c h r e A p p r o b t d y 0.5 gnUn of Zirdoy-2 matography, standard grade) added to was dissolved in a mixture of sulfuric act as a binder. The mixtures RF and hydduoric acids, fumed to drybriquetted a t 60,000-pound pressure in nem, d taken up in 40 mi. of 6 N a 1-inch diameter, tungsten carbide* acid. The Solution w a s hYlined mold (ARLBriquettig Machine, t r a n e f e r z e d to a 125ml. separatory Model 4451). funnel and extracted three t i e s with 40 ml. of 1M TTA in benzene. The METHOD2. Two schemes were inTTA-bewne solution was conditioned vestigated for separating the bulk of the with 6 N hydrochloric acid prior to use zirconium from the alloying elements: in the extraction. Extraction times precipitation of the zirconium with were 10 minutes each. mandelic acid, and solvent extraction of After the finalextraction, the aqueous phaae w m drsined inia s l S r , i * the zirconium. Excellent separations
ments. Solvent extraction, using thenoyltrifluoroacetone (TTA) in benzene, was the most satisfactory for the separation. Because hafnium is diflicult to determine chemically or spectrographically in the presence of zirconium, the investigation was extended to include the possibility of using the x-ray fluorescence method.
beaker using t'mee rinses of 6N hydrochloric acid. To this were added 2.00 grams oi cellulose powder and 20 mg. of aluminum as carrier to ensure compiete precipitation of the hydroxides. The solution was then takes to the cresol red end point ( p H 7.2 to 8.8) with sodium hydroxide solution. After digesting 10 to 15 minutes, the precipitates were filtered through Whatman So. 41 paper, washed, and dried. The filtrate was tested and precipitation found to be quantitative. The dried mixture of cellulose and hydroxides was then briquetted into a pellet 1 inch in diameter at 60,000-pound pressure as in Method 1. Preparation of Standards. METHOD 1. Two-gram samples of zirconium metal wit.h a known content of iron, chromium, and nickel were dissolved in sulfuric and hydrofluoric acids. T h e solutions were evaporated t o dryness and taken u p in dilute nitric acid. Measured. additions of tin, iron, Chromium, and nickel were made. Tin, iron, and chromium standards were coprecipitated with ammonium hydroxide, filtered. washed, and ignited to the oxides. The nickel standards were precipitated with sodium hydroxide. The a i d e s were then ground and briquetted in the same manner as the samples. A sample of the starting zirconium metal was carried through the entire procedure to serve as a blank. METHOD2. Standards for the extraction method were prepared by mixing suitable aliquots of standard solutions of tin, iron, chromium, and nickel together with 20 mg. of aluminum carrier and a n arbitrary amount of zirconium (7.5 mg.). These solutions were precipitated, filtered, dried, and briquetted as the samples. The zirccnium was added to the standards, because in the sample treatment a small amount of zirconium is left in the aqueous phase after the third extraction. It was found experimen-
tally, however, that with no more than 10 mg. of zirconium present, .fariation in the amount oi zirconium in the sample briquet had little effect on the intensity of the fluorescent x-rays of the elements being determined. This amount of airconiumcorresponds to 98% extraction of the zirconium and in practice the extraction was better than 99% complete. Numerous standards were prepared with differing amounts of iron, chromium, and nickel in case of possible interelement effects. Such effects were found to be negligible. Measurement of Samples and Standards. Samples a n d standards prepared b y both methods were irradiated in t h e spectrograph and t h e K a line intensities for t h e various elements measured by t h e fixed count technique. A minimum of 64,000 total counts was accumulat-d. One or two standards were run with each batch of samples being analyzed to check on instrumental drift. I n Method 1, a blank correction was made for tin by counting the blank zirconium standard at the Sn K a peak position. For iron and chromium, Table
Analytical Method Oxide
Extraction (for 0.5-gram sample)
l.
backgrounds were determined a t posi:;ions about 1' to 1.5" iemoved from the peak positions. I n Method 2, gross counts at the iron, chromium, and nickel peaks were used, as the htckrround was found to be constant. ?'or tin it was necessary to count a background. as it varied widely from sample to sample. The pulse height anaiyzcr :V:M dsed for the iron, chromium, and nickel counts to reduce the background of scattered radiation, and was found to improve the peak to background ratio by a factor of nearly 10. I t was not used for the tin determinations, because no improvement in the peak to background ratio could be obtained. 3Iost of the energies of the scattered x-ray quanta comprising the background are too close to those of the Sn K a radiations being measured to be resolved by the pulse height analyzer. Both sides of the sample briquets were counted, and the average counts were used for calculations. Counting rates from side to side generally agreed to within 1%. I n counting the standards and samples, a helium path was used. Al-
Calibration Data Obtained from Standard Samples
Tin Concn., Net ?a c/s
Iron Conm., % c/s
Chromium Concn., % c/s
Nickel Concn., Gross
%
C/S
1.02 1.52 2.02
1495 2200 2935
0.042 0.080 0,119 0.158 0.197
620s 860 1070 1295 1465
0.080 0,119 0.158 0.197
7350 820 882 953
0.030 0.050 0.070
1425 1720 2030
1.20 1.50
12,130 14,740 17,260
0.041 0.102 0.203
1960' 3490 5820
0.041 0.103 0.145
1390b 1935 2270
0.020 0.061 0.101
2420 4690
1.80 Net per second, * Gross counta per second.
6800
Table II. Analysis of Zircaloy-2 Samples Nickel, 7c Tin, yo Iron, % Chromium, % -__ .__ X - h G x-Ray :'i-Ray X-Ray ExtracExtracExtracExtraci h d e tion Oxide tion Oxide cion Oxide tion +2hemical method method Chemical method method Chemical method method Chemical method method &I -43 1 35 1 26 0 101-0 107 0 105 0 118 0 1 0 2 4 107 0 099 0 iO8 0.045-0.048 0 055 0 057 ~
JsmDie 3Yi04FI"
SYP41Q
5Y303d
I
i 30-1 36
1.34
i 35
1 37
1 30 1.32
0 115
1 39 0 130-0 140 0 124 0 136 1 32 0 135
1 27 0.125 1.34
0 142 0.136 0.130 0 127 0.122 0.125
0.104
i)
098-0 102
0 101 0 103
0.103
0 053
u 055 0 055 0 ti56 0 057
0.054
0.108 0.106
0.107 0.101 0 099 0.098 0.103
0.051
0.056 0 054 0.054
0 052 G 048
Chemical analysis a Lathe-tuned chips from an extruded billet produced from an ingot melted by Ailegheny Ludlum Steel Cow. values represent range of resuih in routine analysis of 5 samples by thw company. Lathe-turned chips from a forged billet produced from an ingot melted by Allegheny Ludlum Steel Cop. Range of chemical resultv 3 for 3 sampies analyzed by this company. Range of results e Lath&umed chips from opposite ends of an extruded tube produced from a n Allegheny Ludlum Steel Cow. ingot. te for 3 samples analyzed by this company. %findard 3!ircdoy-2 sample from Bettis Plant, Westinghouse Atomic Power Division. Chemical values from WAPD. VOL. 31,
NO.
10, OCTOBER 1959
0
1633
though this is of real advantage only in the rase gf chromium K a radiation (65% absorbed in an air path), it was used for all elements as a matter of convenience. Determination of Hafnium. For the investigation of the hafnium determination, a group of standards was prepared from spectroscopically pure nafnium and t h e purest zirconium metal ayailable, which contained 48 p.p.ni. of hafnium as determined b y a neutron activation method. T h e method of preparation was t h e same a s for Method 1. RESULTS AND DISCUSSION
Determination of the Alloying Elements. T h e counting rates obtained on standard samples for both methods are presented in Table I. A plot of these datn for all elements is essentially linear. T h e curves for iron, chroniitini, and nickel do not intercept the ordinate a t zero intensity, because there are always small iron. c h r o n ~ i u n ~and , nickel peaks present due t o impurities in the x-ray tube. These calibrations were checked by analyzing a group of mniples prepared by both of the methods described, and the results were compared with values obtained by rhemical methods. ~
Table ill. Reproducibility of Results' Obtained by Extraction Method
Sample % Sn
' 7 Fe
% Cr
%Si
1 29 0 130 0 100 0 056 0 130 0 100 0 05.5 1 26 0 123 0 098 0 05.3 1 28 1 26 0 125 0 098 0 054 i 27 0 127 0 099 0 054 u 3 015 0 0036 0 0012 0 0010 3 Vaiues represent determinations on alicjuots from the same solution of a Lircaioy sample 1 2 3 4 97
Table IV.
2:Table I1 this comparison is shown for tin, iror,, chromium. and aickel determined by Methods 1 and 2. The results compar: very favorably n ith the chemical valtes. #'or Method 2 the first values listed for sample 5Y303 are the averages of four determinations made on d z p r e n t aliquots of the same solution. All other values were obtaine! on individual 0.5-gram samples. I n certain cases there are discrepancies between the x-ray and chemical values which cannot be attributed to the variations due to sample preparation and counting. However, a n exact comparison is difficult because of the wide range of reported chemical values on the sample. In part this is due to inhomogeneity in the gross sample. For a better estimate of the accuracy, a large sample of Zircaloy (No. 5-022) was dissolved and aliquots were taken for both chemical and x-ray analysis. Average values and calculated standard deviations for SLY determinations shon-n in Table V are in niuch Letter agreement than those obtained on separate individual samples. Tno main soclrces of error are likely to affect the precision of the results: errors in the course of sample preparation and errors associated with the countr statistical ing of the samples-i.e., variations in the counting plus any fluctuations in the electrical circuits. The precision of the sample preparation used in Method 2 n as demonstrated by the comparison of results obtained on four aliquots from the same solution of a Zircaloy sample (Table 111). The precision of the results, calculated from these data, is to about 2% of the amount present. The variance due to counting was ascertained by counting the same group of samples 0:-er a period of several days (Table IY). -4 variation in counting rate of 1 to 2% may be e.xpected for a given sample. The minimum number of counts taken :n any case was 64,000.
Variation in Counting Rates
Counts
4 >.
2 :3 i
1st day 12 246 12 169 11 966 11 810 3965 3953 3i87 3861 1833 1858 1829 1840 4169 4197 1062 41 10
Second
2nd day
IV.
O7G 056 082 02; '3953 3834 3794 3822 1830 1829 1807 IS16 -1163 4161 -io53 4122
12 226
__--1634
Der
ANALVTICAL CHEMISTRY
12 12 12 12
11 870 12 117
Std. Dev., U
109 17:3 140
that the msxiiruum statisticsi variation shnuld be 0.47,. X o s t oi :he ;wintion in the counting rates can therefcre be attributed to the instrument. Determination of Hafnium. Serera1 papers have appeared on t h e determination of hafnium in Zircaloy by x-ray fluorescence (1,4,6). Unfortunately, in all cases the lon-er limit of detection is higher than the maximum specifications for hafnium in reactor grade Zircaloy-2, 200 p.p.m. The chief difficulty in this deterniination is that the second order K series of the zirconium spectrum overlies the L series spectrum of hafnium which must be used for analysis. Ercitation of the K series of liafniuni is not possible n-ith our equipment. The nicst useful hafnium line for analysis, Lal, is completely obscured by the zirconium Kal, second-order line. The nest most useful b e , hafnium Lg,, is o;-erla.ppcd by the second-order zirconiuin k'3? line. Other hafnium La lines are obscured by the presence of the tungsten spectrum from the targPt of the s-ra.y tube. Because the hafnium Gspectrum is excited a t 12 kr.,and the zirconium K spectrum at 18 kv., the possibility of esciting the hafnium spectrum at 17 kv., was studied. However, the resulting intensities n-ere too low to be of any use. The only hafnium line which is relatively free from interference is L y I , which has an intensity of about 40y0of the Lal line. If lead is present in appreciable quantity, it nill interfere with this line. Th? specifications for Zir(day-2 call for lead to be less than 130 p.p.m., but it is possible that hafiiiuni and ie:ici may occur in similar concentrations, iznd interference might be espected. line The possilile use cf the hafnium for quantitative :ietrrm:nat,ion of k i f nium was investigated, however, iising a set of standard samples. Both :he hafnium L r l line htensity m d a suit;ti-le background xere counted on standards and rhe net line h t e n calcciatied ,\T%hleV1. The :let counts obtaincd i riile shon-ing an increase - ( i h m u c content, are n-idek : ;30
ahle errors .ssoci:tt(*(;if.31. tile - w r : t ing rates. The reiatiT-::i!- :lrpi: i'rror .-.i i 832
1852 1836
30 '23 54
is42
2i
4198
zirconium K series lines from the hafnium LQ and Lg lines. Unfortunately, this also proved to be impractical for the low concentration levels of interest. Discrimination between two energies, one of which waa of much greater intensity than the other, was required. Because the pulse height distribution curve of the higher intensity has some pulses at all amplitudes, there are more counts due to zirconium at the hafnium energy peak than to hafnium itself. This problem of separating pulses of M e r e n t energies having greatly different intensities is discussed more fully by Miller (3.
Table V. Accuracy of Chemical and X-Ray Fluorescence Results“
yoSn yo Fe
Two relatively rapid and simple methods for determining tin, iron, chromium, and nickel, in Zircaloy-2 have been developed using fluorescent x-ray spectrometry. Method 1, the “oxide” method, requires the least amount of sample preparation but has a lower sensitivity. Method 2, the “extraction” method, while necessitating somewhat more sample preparation results in at least a twofold increase in sensitivity over Method 1 and could be of possible value in the analysis of other zirconium alloys. Both methods are advantageous, in
70Xi
1 . 4 4 0 138 0 102 0 054 u 0 03 0 002 0 0 0 2 0 0 0 1 Chemical 1.422 0 13‘: 0 102 0 019 u 0.002 0.OOi 00004 OOOO3 a Averages of six individual determina-
tions for each element made on aliquots from,the =me solution of sample 517’422.
Table VI. Hf LyI Intensity Measurements on Standard Samples
HI (Lyl),
Hafnium, CONCLUSION
yo Cr
X-ray
P.P.31. 50 __
250 415 885
1620
.4v.
CIS i58 .. 797 807 865 877
Background, AV.
.4v. Set
c/s
CIS
716 725 726 748 737
42 ~~
72 81 117 140
that the four individual element dekrminations can be carried out in rapid succession on a single sample. The four elements can be determined by Method 1 in about one quarter of the time required by webchemical methods, and by Method 2 in about one half the time. The precision and accuracy of the
results of both methods are comparable nith those of existing chemical methods and satisfactory fGr routine specificntion cheeks. The determination of hafnium does not appear to be feasible a t the low-level in which it occurs in reactor grade Zircaloy-2, at least Kith ellsting equipment. Because a rapid analysis by neutroi? activation i n s been developed at tliis project for hafnium analysis (e),ic \ w s not considered worthwhile to investignti. this determination b?. x-my nicthods exhaustively. LITERATURE CITED
( 1 ) Birks, L. S., Brooks, E. J., ASAL. CHEU.22, 1017 (1950). ( 2 ) Mackintosh. W. D.. Jervis. R. E.. Zbid.. 30, llsd (1958). ’ (3j-11iiier, D. C., Sorelco Reptr. 4, 37 (1957). (4) Iloak, W. D., “Determination of Hafnium in Zirconium by X-Rav Fluorescence S ctrometrv,”-S3mposiiim on ~
X-Ray DiKaction and S-Ray Emission Spectrometry Techniques, General Electric Co., Schenectady, N. T.,1957. ( 5 ) hloore, F. L., A N A L . CHE\I. 28,
997 (1956). (6) >fortimer, D. If.,Romans, D. -4.) J. O p t . SOC.Am. 42,673 (1952).
RECEIVED for review December 15, 195s. Accepted June 26, 1959. Second Conference on Analytical Chemistry in Kuclear Reactor Technology, Gatlinburg, Tenn., September 29 to October 1, 1958.
Sealed Tube Combustions for the Determination of Carbon-14 and Total Carbon DONALD
L
BUCHANAN and BHTY J. CORCORAN
Radioisotope Service, Veterans Administration Hospital, Wed Haven, Conn., and Department
of Biochemistry, Yale University, New Haven, Conn. ,Conventional wet or dry combustions are associated with the formation of nitrogen oxides which even in minute amounts cause undesirable effects in proportional counters. Simultaneous microdeterminations of total carbon and carbon-14 may be performed by oxidation with cupric oxide in sealed Vycor tubes at 850’ C. Manganese dioxide is added to provide free oxygen and cupric chloride displaces carbonate whenever basic metols are present. The carbon determination is as precise as it is in other types of oxidation used for carbon radioassay and the carbon dioxide is very pure as judged by its counting characteristics.
vfT
rad d:. 1
m v organiL compounds that c:mtam nitrogen are ozidized for counting by n e t (14) or ‘ c.;idstion. o d e ? of nitrogen ~ ~ and r n some wili co-con- adding n minganese tlioside trnp ( 7 ) or by filling t h e heated quartz tube \rit h en!cium ozidt’, but both procedures gave sniall nirrnory effects. Sealed static oxidations of organic compounds n-ere reported early (9) irith modifications later ( 2 ) . Grosac,
Hindin, and Krshenbaum (S) tniployed isotope dilution in conjunction with static Oxidation on a semimicro scale. Sealed tube oxidations h : i ~ e recently been used for the riltranlicrodetermination of curbon hy isotope dilution (S), and a dircct ninnoinc~tric ultramicroanalysis has been d c s c r i i d (10).
IVilzbach and his coll(~ng1c.s have used scalctl tube combustions for r:itiioassay of carbon-14 (f5). They intlicati. that, t.his t:pc of Oxidation might be usoiiii ior the combined dctcrniinntioxi of carbon-14 :ind total cnr!mn. The prcwnt report givrs the tlrtnils oi i: mcthotl tle\.clopcd with thn: cznc! in vien-. EXPERIMENTA:
Vacuum Apparatus. The -,xcuuxi: system (Figure 1) is a simplification oi tiini previously described iar collectYOL. 31, NO. 10, OCTOBER 1959
@
1635
