V O L U M E 2 4 , N O . 9, S E P T E M B E R 1 9 5 2 Beadle, B. W., Kraybill, H. R., and Stricker, L.A., Oil urd Soup,
(3)
2 2 , 5 0 (1945).
(4)Booth, R. G., Kon, S. K., Dann, K.J., and Moore, T., Bzochcin. J.. 29.133-7 (1935). . . ( 5 ) Brice, A., and Swain, 11. L., J . Am. Oil Chemists’
u.
SOC.,26, 272-7 (1949).
Brice, B. A., Swain, Margaret, Schaeffer, €3. B., andilult, W.C.,
(6)
Ibid., 22,219-24 (1945). 17) CroNelI, G. K., J . Assoc. Ofic.A g r . C h e m i s t s , 27, 448-50 (1944). \ R ) Kass, J. P., Miller, E. S., Hendrickson, RI., and Burr, G. O., Abstracts of Papers, 99th Meeting, Awamc.w CHEMICAL SoCIETY, Cincinnati, Ohio, April 1940.
1399 ( 9 ) Mattirllo. .J. J., “Protective and Decorative Coatings,” Vol. IT, S e w Tc~i,k. ,Juhn TF-iley Br Sons, 1944. (10) Mitchcll. .I. l l . , Jr., and Kraybill, H. R., J . Am. C h m . S O C . , 64, 9 9 s - 9 4 11942). (11) ;\litc.hell. J. H., Jr., Kraybill, €1. R., a n d Zscheile, F. P., ISD. Esc;. (’HEM,, ASAL. ED.,15, 1-3 (1943). (12) O’Connor, R. T., Heinselman, D. C., and Dollear, F. C., Oil a n d SOUP,22,257-63 (1945). (13) Stillman, R. C., J . Am. Oil Chemists’ SOC.,26, 399-401 (1949).
RECEIVED for review October 3, 1949. Bccelited J u n e 21, 1952.
Quantitative Spectrochemical Analysis of Rare Earth Mixtures ,I. i.\OHHIS 4biD c. E:. 1’E:l’PE:K S t a b l e Isotope Research a n d I’rorlitrtiou Diri\iori. O a k Ridge \ationul I m b o r u t o r ? . 3 - 1 2 Irru. O d i Ridge, Tenri. \ porous cup spectrochemical procedure for analysis of rare earth mixtures has been developed to assist in the preparation of pure rare earth compounds for use in the electromagnetic enrichment of the isotopes of these elements. Precision of this method. *lO% at a 99.59’0 confidence level, is such that small differences can be detected, thus permitting better evaluation of the chemical separation processes. Analyses of 10 samples for 5 rare earths can be completed in approximately 1 man-day. This procedure permits the quantitative determination of individual rare earths in complex rare earth mixtures and is limited only to the elements available for preparation of standards. Use of the solution technique eliminates possible inhomogeneities in sampling. No difficulties with interelement effect, outside of direct line interference, have been noted. Data obtained give strong evidence of the importance of proper application of the spectroscopic principle of homologous lines in the selection of the internal standard element.
T
HE work of the Stable Isotope Research and Production Division has led to a study of new methods of preparation
of pure rare earth materials for use in the electromagnetic separators (calutrons). Thus far, both chemical and isotope separations have been made on samarium and neodymium, which are more readily prepared in the pure state than the other rare earths. Chemical separation methods tried have included the well-established fractional crystallization of double salts, selective precipitation with organic materials, and liquid-liquid extraction processes. The feasibility of any such purification method is evaluated principally by the concentration coefficient obtained per stage for the particular process under investigation. To establish this coefficient] a method of analysis is required that can detect differences between samples with high precision. The preparation of pure gadolinium compounds from rare earth concentrates initiated spectrochemical studies on the determination of yttrium, lanthanum, praseodymium, neodymium, samarium, europium, gadolinium, dysprosium, and ytterbium in mixtures of from 1 to 1OOc7,. Analytical values obtained by this method are limited only by the availability of pure rare earth compounds for standards ; hoxr ever, where the absolute value of concentration is not necessary, all detectable rare earths can be determined in terms of concentration ratios using one particular sample as the reference.
HISTORICAL
I n the past tht. majorit!- of rare earth analyses have been made by spectrophotometric methods. Scandium, yttrium, and Ianthanum, which frequently occur with the rare earths, and terbium and lutetium are not readily determined by this means. oeller and Brantley ( $ 3 )contains a comprehensive survey and considerable new experimental data on rare earth absorption spectra. Sumerous methods havr been reported on the emission spectrographic determination of the rare earths. Table I summarizes the salient points of niost of these techniques. O f the methods given, that of Fassel ( 6 ) is the only one applicable to the determination of major constituents in rare earth mixturefi. A method for the analysis of rare Parth fractions by molecular spectra is given by Piccardi (%j. EXPERI.MMENTAL METHOD
As the ran: earth oxides, except ceric oxide, are readily soluble in hydrochloric acid, the porous cup method (9) was considered to be easily adaptable for this analysis. Initially, a concentration of 4 mg. of rare earth oxide per ml. x-as used when the sample was exposed in the first order of a Jarrell--4sh 21-foot spectrograph; platinum was used as the internal standard. Recent work has been standardized a t a concentration of 10 mg. per mi., using the second order of the same spectrograph n i t h drontium as the internal st,andard element. The detection of 100 micrograms per nil. of impurity in a solution concentration of 10 mg. per ml. necessitated use of the spark source a t a high power level (approximately 5 to 7 radiofrequency amperes a t 12,000volts as compared to 3 radiofrequency amperes a t the sitme voltage for normal porous cup work in this laboratory). The dissipation of twice as much power resulted in undesirable boiling of the solution, an effect soon eliminated by a specia,l vater-cooled tantalum metal electrode holder (Figure 1). Tantalum metal was selected because it was the most acid-resistant material readily available. It-ith a water flow of 300 to 350 ml. per minute, t h c electrodes arc’ kept cool a n d the solutions s h o r no tendency to boil over. I’RE1’AR.ATION O F STANDARDS
Standards were made from the purest oxides commercially available (95 to 99.5%). The samarium and neodymium oxides were obtained locally (prepared by Boyd Keaver and K. A. .411en, Oak Ridge Sational Laboratory), \\-lwreas t h c gadolinium, europium, praseodymium, ytterbium, arid yttrium oxides yere obtained from Research Chenlicals, Inr., Burbank, Calif., and used as received n-ith corrections for any impurities found. I t is anticipated that subsequent ?tandai,cls \ d l be prepared from t,he purified fractions as they become available. The individual rare earth oxides \yere di;isolved in dilute hydrochloric acid to a mncentration of 20 mg. per i d . for dysprosium, gadolinium, Ianthmum, neodymium, prtneo$ymium, samarium. and yttrium: for the lees rommoa europium and
ANALYTICAL CHEMISTRY
1400
least in a general way, the qualifications for a homologous . line, Elements Sample Form. Limits in that its stage of ionization is Avthor Determined Eroitatian Reported, % PZeCiSiO" the same and the upper level Faseel (6) Y,Gd Oxide. d.o. arc 8-100 r t 2 . 5 % std. de". term lies rather close to the l'assei and Wilhelm (7, Sm in Nd. Eu in Oxide. d.c. 810 0.1-2.0 , median of the rare earths; also, 8) Sm 0.01-2.0 -4% SY. de". its t o m difference and hence its Gatterei and Junkes Eu in Sm Solution on C, d.o. 0.005-1.0 ..... wsve length are in the same (11) arc region of the spectrum. A dis4.005-5 Y ..... Flirt and Ssohtrieb (181 La. Co. Pr, Nd, Sm. Copper spsrk advantagein thenseof strontium Cd. DY is that the line has 8. tendency to Hopkins el 01. ( 1 4 ) ..... Solrltion on C . d.o. 0.1-10 m g . oxide Max.ermr *15% BL.0 per 1 mg. 7x0, self-reversal, as it involves the Lopes de Areona ( I n ) Od in Sm Oxide. d.o. 810 0.10-.3 13% ground state of the ionised atom. Although obvious reversal of the +IO% std.dev. MoCleIlsnd ( 1 7 ) ..... Oxide in iron oxidc, ... 3.e. Bra strontium line does not occur Russell (281 I.&.Y ..... ... 25% until a concentration of about Srxibzer and Mullin DY,E?, Eli. Gd O x i d e , d.e. arc 0.12-16 10% 2000 D.u.m. is reached. the line ill, density-is outside the'densitoSm in Nd. Gd in Solntion. d.e. arc 0.1-10 Visual semiquantiSel>vood (SO) metric working range a t B conS d , Le in Yt tative centration of about 20 p.p.m. yisilal aemiquanti. Short and Dutton ( $ 1 ) All Solution. d.e. &re 5-50 -i tative and the line may be affected by Smithand Wiggins (as) DY,Gd. La, Lu, Yt. O x i d o d.e. BE 0.1 Qualitative incipient self-reversaleven a t this Yb concentration. The value finally Spioer snd Ziogler ($SI La in Pr C o ~ ~ spark er 28.2-42.3 4.6%av. de". selected for the strontium internal standard concentration was 5 micrograms per ml. This law concentration of strontium is unfortunate from the contamination standpoint, 8.6 a few ytterbium a concentration of 5 mg. per ml. was miade. Required amounts of these solutions were transferred with micropipets to samples whose processing history ryas unkno,,,,, individual 10-ml. standard flasks and made up to volume with Present at this Order Of magnitude. hydrochloric acid and distilled water, and 100 microliters of a Figure 3 shows percontage deviations from the average of bath 0.0570 strontium metal solution was added. the individual line intensity and the line intensity ratios f or This method gave composite rare earth standards whose total concentration is approximately the same as the concentration of the samples 8 s run (see Table 11). Standards having varying Table 11. Composite Rare E a r t h S t a n d a r d s concentrations of the individual rare earth elements were made (VP.lues in micrograms ger ml.) in a similar manner as a check on the behavior of the composite Ele,nent set and for the separate determination of the higher percentage ag Orido 1 2 3 4 5 6 7 4600 1150 2300 230 460 60 115 values of the rare earths. Experience has shown that the range 1000 200 500 25 50 100 1t o 30% e m be covered by the composite standards against the 260 400 1000 2000 4400 100 50 Gd L& 4000 2000 1000 400 200 100 50 original a m p l e and the higher percentages determined by comNd
Table I.
L i t e r a t u r e Abstracts
~
~
2
pressures of the vanous compounds play a relatively more important part along with the energy levels in question. A general discussion of these points is given by Ahrens ( 1 ) . Marked improvement in the precision of the present method resulted when the internal standard element was changed from platinum to strontium. This change was suggested f siderations of the energy level schemes o menta concerned. As can be Seen from I the particular line of strontium chosen f i
V O L U M E 2 4 , NO. 9, S E P T E M B E R 1 9 5 2
1401
several rare earths a t diffei ent breakdown voltages of the auxiliary spark gap. Ratios of rare earth to rare earth are shown for comparison purposes Percentage deviations in the individual line intensities of more than 20% are found, whereas line intensity ratios agree to better than 3 7 over the range of voltages studied (all data obtained from the same plate). hlinimuni deviation in the ratio would indicatr analytical line criteria had been met. I
Standard error
=
1-1
4 I 96.5 5
CERIUM (24
-*
PRASEODYMIUM (26)
a -
4222 ”
\
GADOLINIUM ( 3 )
~
~ 30 20
*
TERBIUM ( 2 4 )
R a t a 99.57, confidence level
= -
s
s
14435.60 I3971.99
I
loo at a 99.5% confidence level
viliere K is the difference between the highest and lowest values of replicate determinations, and is the average of the replicate determinations. The theoretical range factor, J ’ , is a constant for any given number of replicate determinations a t a predetclrmined standard error and confidence level, and is used to determine whether the spread of values lies within the prescribed limits. This procedure involving four determinations at thta 10% stvndard error level and 99.5cc confidcnce Iewl gives an allon-ed range factor J’ of 0.15.
NEODYMIUM7 (: 4 3 ): : : $ - -
/EUROPIUM ( 2 7 )
x
where S is the standard deviation, n the number of replicates, and X the average. Iinotving a relationship betweeii standard deviation and the range of data exists, the following formula has been developed (18) for checking the precision of individual samples. For convenirnce, this is c a l l d the range factor
J LANTHANUM (12)
(3S’dn)
I: ~
~
STRONTIUM
DYSPROSIUM (15,24)-* YTTERBIUM ( 2 2 )
A GADOLINIUM/ NEODYMIUM
-369420
1-1
LUTETIUM(2I)
0
3554 4 4
10
O
a
PLATINUM (5)
- 1
2997 97 4215 5 2
1-1
CHROMIUM ( 5 ) 20,000
40,000
60,000
4550 66
80,000
100,000
120,000
c rn-’
Figure 2.
GADOLINIUM / STRONTIUM
F
4 4 0 7 7 71
STRONTIUM ( 5 )
0
f 0 GADOLINIUM
D NEODYMIUM
Xlodified Energy Level Diagram for Rare Earth 4nalysis
.
Uumbers following element refer to bibliograpb) Shaded section denotes firct ionization potential. * Indicates uncertain values
_ ~ _ _ ~ Table 111. Precision of Line Intensity Ratios for Different Internal Standards , 0 (I erape \ range factor values)
Internal Standard Element Pt I Cr I1 Sr I1
G d I1
0.22 0.12 0.08
Rare Earths S d I1 0 26 0.10 0.06
1’1
I1
0
STRONTIUM
*
NEODYMIUM / PRASEODYMIUM
A
N E O D Y M I U M / STRONTIUM
0
PRASEODYMIUM
3
30r STRONTIUM
0.13 0.07
X
PRASEODYMIUM/NEODYMIUM PRASEODYMIUM/ STRONTIUM
Table I11 compares the piecision of results for three of the rare earths studied, obtained by comparing rare earth I1 lines to platinum I, chromium 11, and strontium I1 lines. Thc average range factor, defined in the nest section, is loiwst for qtiontium in agreement with the theory of analytical line pairs (Figure 2 ) .
30 10400
11000
11600
GAP BREAKDOWN VOLTAGE
PRECISION
The expression of analytical precision varies from laboratory to laboratory, with the majority using either mean (average) deviation
or standard deviation
(VOLTS)
Figure 3. Percentage Deviation of Intensity and Intensity Ratios at Various Breakdown Voltages of -4uxiliary Spark Gap
To use this formula for comparison purposes, it is neceshaiy that the number of replicates be the same in comparing data obtained under different conditions. For example, the folloning data were obtained under two different sets of condition*. de-ignated as groups A and B:
x
where X is the individual value, the average, arid S the number of determinations. I n order to be able to follow more closely the improvements made in procedure development, a more rigorous treatment of the data is desirable. T o this end the following term for expressing precision is used:
I2200
Group A,
I/I 1.43 1.44
1.77 1.70
xR J
1.58 0.34 0.22
Group B,
I/I 0.62 0.60 0 60
0.60
0,605 0.02 0.03
~
ANALYTICAL CHEMISTRY
1402
METHOD OF A%\IALYSIS
I
I
I 25
I 1 I oca
100
I
I 1
One hundred-milligram samples of the raie earth oxides are weighed into 10-ml. volumetric flasks and 2 to 4 ml. of concentrated hydrochloric acid are added. Distilled water is added and the samples are placed on a hot plate to dissolve the oxide completely. The solutions are cooled, and made up to volume with distilled n-ater, and 100 microliters of a 0.05% strontium metal solution are added, giving a final concentration of 5 p.p.m. of strontium. The following conditions are used in the csxposure of the samples: Electrode, porous cup electrode with inch graphite counterelectrode. Spark, National Spectrogiaphic Laboratory .ource unit. Capacity 0.10 mf. Reriitance :iohms. Inductance 100 mh. Secondary gap tbreakdown voltage approximately 12,000 voltq, Radiofrequency amperes 5 to 7. Number of
