ANALYTICAL CHEMISTRY
1014 Table XIII.
Relative Amounts of Ca Paraffins in Eaet Texan Heavy Virgin Naphtha
Table XIV. Relative Amounts of CONa hthenes in East Texas Heavy Virgin Naphtga (Volume per cent by infrared analysis)
(Volume per cent by infrared analysis)
of Paraffin-
% of Cn Psraffina'
'Nsphthepe Portion
n-Octane Meth lheptanea 2-dethylheptane 3-Methylheptane CMethylheptane
% of CI NsphthenesO Portion Trimethylcyclopentsnes 21.7 2.54 cir,tronr,cir-l,2,4-trimethylcyclopentsne 6.9 0.81 cis,cir,fronr-1,2,4-trimethylcyclopentane 0.4 0.05 cia,cir,cir-1,2,4-trimethylcyc~o entane c c cia,.tronr,cir-12 3-trimethylcyc?opentane 6.6 0.77 ci~.cir.trona-l:2~3-trimethvlcvclo~entane 1.4 0.16 cis;cir cis-1,2 BirimethylEyciopehtane 0.1 0.01 1,1,3-'r~methylcyc1opentane 4.3 0.60 1,1,2-Tnmethylcyclopentane 2.0 0.24 Methylethylc yclopentanes 9.3 1.09 cia trona-1 3-methylethylcyclopentanee 3.2 0.38 cia trona-l~2-mathulethylc~clopentanes 5.2 0.61 1,l-Methylethylcyclopentane 0.9 0.10 2.5 0.29 Propylcyolopentanes 0.5 0.06 Iaopropylcyclopentane 2.0 0.23 n-Propylcyclopentane 51.8 6.06 2.2 0.26 2.1 0.24 11.8 1.38 19.2 2.24 3.9 0.46 2.2 0.26 cis-1 4-dimethylcyclohexane 10.4 1.22 fron~-l,4-dimethylcyclohexane 14.7 1.72 Eth ylcyclohexane 100.0 11.70 Total * Cn naphthenea comprise 10.2% of total heavy naphtha (215-436 F. no inal boiling range), or 2.6% of crude. PParaffin-naphthene portion compnsea 87% of total heavy naphtha (216-436O F. nominal boiling range),,or 21.3 0 of crude. C Presence not detected by analytical metrod employed.
+ +
2,P-Dirnethj.lhexsne .Trimeth lpentanes 2 3 4-&1methylpentane 2:3:3-Trimethylpentsne 2 2 3-Mmethylpentane 2:2:4-TrimethylpBntane Ethylhexane 3-Ethylhexane Meth lethyl entanes 3-dethyl-~-ethylpentane 3-Mrthyl-3-ethylpentane
0.9
1.7 0.9 Total
-
0.6 0.3 0.0 0.0
1.7 0.9 0.0
0.11
0.20 0.10
0.07 0.04 0.00 0.00 0.20 0.10 0.00
100.0
-
(12) Kent, J. W., and Beech, J. Y., Zbid.,19,290 (1947). (13) Mair, B.J., and Forziati, A. F., J . h e a r c h Nqtl. Bur. Standards, 32, 165 (1944). (14) Pitzer, K. S.,and Beckett, C. W., J. Am. Cham. ~ o c . ,69, 977 (1947). -- ,(16) Plyler, E. K., Stair, R., and Humphreys, C. J., J. Reeeurch Natl. But. Standards, 38, 211 (1947). (16) Podbielniak, W. J., and Podbielniak, W., IND.ENO.CEEM., ANAL.ED.,13, 639 (1941). (171 . . Prosen. E. J., Johnson, W. H., and &mini, F. D..J. Reeearch Natl. Bur. Standards, 39, 173 (1947). (18) Rosenbaum, E. J., Martin, C. C., and Lauer, J. L., IND. ENQ.CHEM.,18, 731 (1946).
.--
-
(19) Roasini, F. D., Mair, B. J., and Glaagow, A. R., Jr., "Summary
of Hydrocarbone Isolated from One Representative Petroleum," A.P.I. Research Project 6, Natl. Bur. Standards, unpublished report, Feb. 28, 1949. (20) Waahburn, H. W., Wiley, H. F., and Rock, S. M., IND. ENG. CEEM.,ANAL.ED.,15, 541 (1943). RECIDIVED August 8, 1949. Presented before the Division of Petroleum Chemistry a t the 116th Meeting of the AMERICAN CHEMICAL SOCIETY. Atlantic City, N. J.
Sodium and Potassium Determination in Refractory Materials Using Flame Photometer FRANK M. BIFFEN, Johns-Manvilk Research Center, Manuille, N. J . A method has been devised by which refractory materials are sintered with calcium carbonate as in the J. Lawrence Smith method, and sodium and potassium are determined on the water extract using the flame photometer. Consideration is given to the calcium present in the extract. Results are at least as accurate, and probably more accurate, as those obtained using the J. Lawrence Smith method, and the time necessary to complete the analyds is cut in half.
U
NLESS thoroughly worked out for a specific material, the
J. Lawrence Smith method for estimation of sodium and potassium in refractory materials is tedious and time-consuming, and the optimum conditions f6r the analysis may not be a t once obtained. Various modifications of the method are given by Hdlebrand and Lundell(8). Several poaaible errors are inherent in this procedure. First, the analyst is not absolutely sure that the mixed chlorides obtained
are free from elements other than sodium and potassium (and possibly, very minor amounts of lithium). Secondly, the amount of alkali chlorides obtained from the small sample (0.5 gram) normrally used is often so very small that very slight errors in weight give noticeable errors in the results. Thirdly, sodium is usually obtained by difference; the potassium is determined, for instance, aa ohloroplatinate, on the mixed chlorides. Consequently, any error in the potacrsium determination ia reflected in
V O L U M E 2 2 , NO. 8, A U G U S T 1 9 S O the sodium figure obtained. This is particularly noticeable when much potassium and little sodium are present; indeed, the sodium figure in such a case may be 50% out. Such discrepancies have been noticed in the collaborative results given for sodium and potassium
1015 Table I. Comparison of J. Lawrence Smith and Modified Methods (Single determinations) J. Lawrence Smith Method Modified Method, Gravimetric and Flame Gravimetric and Flame Photometer Photometer GraviCalod. 1500P.P.M. 1800P.P.M. N.B.S. metric a8 total aa total CaO CaO Results chlorides, Saz0, KaO, chloridm, NatO, KtO, NarO, KrO, NarO, Kto,
with such standard samples as those Sample Cel/te 545 issued by the National Bureau of Celite 811-0-Cel Standards, All in all, the results obtained by this method, unless followed Opal glass Borosilicate glass with extreme precision and care, may well J-M firebrick be looked upon with some suspicion. If the method has been thoroughly worked out and the procedure carefully followed for specific samples, good results are obtainable. Severtheless, the method is time-consuming. I t was felt that if the alkalies in refractory materials could be obtained in solution without too much trouble, the flame photometer could be employed for rather rapid determination of the alkalies separately. Such a method has been found and very satisfactory results have been obtained. Indeed, it is believed that, in general, more consistently accurate results can be obtained in not more than half the time necessary when using the complete J. Lawrence Smith method.
%
4 7 0 %
%
6.70 2.48 5.80 20.76 7.98 4.04
3.00 0 . 7 1 0.60 0 . 5 6 .0.283.11 8.59 3.35 -1.05 0 . 1 7 0.88 1.43
6.78 2.02 5.46 21.50 7.91 3.93
%
%
2.96 0.72 0.58 0.57 0.283.11 8.82 3.43 -1.15 0 . 1 8 0 . 8 8 1:44
%
%
2.96 0 . 7 2 0.59 0.50 0.283.14 8.97 3.43 4.19 0 . 1 8 0.88 1.44
%
..
%
..
o:i83:i7 8.48 3.25 4.16 0.18
..
..
FLAME PHOTOMETER
The flame photometer used in this work.was the Perkin-Elmer instrument, Model 52A. This has a dual optical system enabling the internal standard method to be employed; this reduces the effects of gas and air pressure fluctuations, foreign ions and molecules, and viscosity differences. The effects, and they are serious ones, produced by these conditions are discussed rather fully by Berry, Chappell, and Barnes ( 1 ) and by Parks, Johnson, and Lykken ( 4 ) . The latter authors used the Perkin-Elmer Model 18,in which there is a single optical system, as did Bills, McDonald, Niedermeier, and Schaarta ( 8 ) ,who proposed t o reduce errors by using a newly designed amplification circuit and by working at lower concentration ranges than have been recommended in the past. Using the same instrument, Pratt and Larson (6) suggested the use of interference filters to reduce error due to calcium in the determination of sodium. I t is not essential to use the internal standard method, particularly when specific conditions of the above-mentioned variables can be properly duplicated in samples and standards. Under certain conditions, as when small amounts of lithium may be present in the samples tested, the direct-reading method may be more accurate. SAMPLES ANALYZED BY NEW METHOD
In order to try this method on different types of refractory materials, the materials listed in Table I were used. These included three National Bureau of Standards certified analysis samples in order to check the results obtained. The presence of small and large amounts of sodium and of otassium will be noticed. The borosilicate glass contained 12% & 0 3 . It was thought that these different samples fairly well represented the general run of refractory materials. PREPARATION OF SOLUTIONS AND COMPENSATION FOR INTERFERENCE
In the proposed method, which employs the initial steps of the J. Lawrence Smith method, calcium and chloride are present in the solutions containing the alkalies. Using a 0.5- ram original sample, the extract containing the alkalies, after a d i t i o n of sufficient lithium nitrate to act as internal standard a t a final concentration of 100 p. .m. as lithium, was bulked to 250 ml. Six refractory materiag were treated in the same manner and the calcium was determined in each and in a blank run a t the same time, after bulking to volume. Calculated as calcium oxide the
BOTM 5OCM A M 0 1OOPM.AS IWS I L A D I Y S
A L L CMlTAlM lOO?CW L l T H l U Y io
20
ao
yo-
PUTS FER w i m
50
MI i o i o i1oO smiw m WTASSIW
Figure 1
amounts varied from 1530 to 1714 p.p.m., while the blank contained 1798 p.p.m. It was, therefore, necessary to add similar amounts of calcium as chloride to each of the standard solutions used. In order to note the effect of different amounts of calcium and whether the amount to be added to the standards was critical, two sets of standard solutions were made up containing similar amounts of sodium and otassium as chlorides, but one set containin 1500 .p.m. and t\e other set 1800 p.p.m. of calcium calculateckas oxig. Results show that curves obtained with the standards coincided in the higher alkali ranges and were slightly separated, as would be expected, in the lower alkali ranges (Figure 2 ) . Results obtained on the samples indicate that it made little, if any, difference which amount of calcium was present, as long as the appropriate curve was used with the standard solutions against which the samples were run. Hence, it appears that the amount of calcium present is not critical. It is best, of course, to keep it around the amount found in solution after known conditions of treatment and extraction of the treated sample. This can quickly be determined simply by making a calcium determination. In
ANALYTICAL CHEMISTRY
1016 Table 11. Calcium in Extract Solutions (Bulked to 250 MI.) in Modified Method Caloium a8 CaO, P. P. M. 1530 1714 1638 1544 1682 1584 1798
Sample Celite 545 Sil-0-Cel Plastic clay Opal glass Borosilicate glass J-M firebrick Blank
all probability it would not be necessary to repeat this except under markedly different conditions.
given above, and National Bureau of Standards data for three of the six materials are all given in Table I. In most cases the two sets of flame photometer readings agree well among themselves and with the National Bureau of Standards results. The gravimetric chlorides figures given, although they agree with the other results fairly well, are probably somewhat less accurate for the reasons stated above, The photometer results are consistent, whereas the separate National Bureau of Standards results for sodium given on the certificate of analysis for the clay, which are averaged to give the final value, are not. In other words, the short flame photometer method may well give inherently more accurate results than the general run of good gravimetric work by the J. Lawrence Smith method. This is largely due to the fact that sodium is normally determined by difference in the latter method.
PREPARATION OF CALIBRATION CURVES
The number of standard solutions required will depend on the amount of the alkalies present in the bulked sample solutions. These are made up in the usual manner and kept in KO-Solvit (Wheaton & Company) or polythene bottles. Borosilicate glass may be used for several weeks without noticeably affecting the alkali contents even of sample solutions. To each standard, 100 p.p.m. of lithium are added as with the solutions, if the internal standard solution is used. Calcium approximately equivalent to that present in the sample solutions must now be added. A stock solution of, say 10,000 p.p.m. of calcium as oxide, is made up by dissolvin the equivaent of low-alkali C.P. calcium carbonate in just sugcient hydrochloric acid. The solution is heated gently to drive off carbon dioxide, cooled, and bulked to volume. After the requisite amounts of this stock solution have been added to the standards, it is made up to volume. Photometer readings are made, either by the direct-reading or the internal standard method (6). The frequent checking of the instrument with the standards is essential. Curves should be made as indicated, the larger the number of curves with as large a number of points as possible, the better. In practice, the number of curves necessary will be limited to actual requirements. Figures 1 and 2 show the curves made for use in this work.
IO0 95 90
65
EO 71 70 05
1:: e
60 “5
uo 38
PROCEDURE
As in the J. Lawrence Smith method, intimately mix 0.5 gram of finely powdered sample with 4 grams of low-alkali C.P. calcium carbonate and 0.5 gram of C.P. ammonium chloride. Heat in a special thimble crucibleif available, or in a 30-ml. platinum crucible with a well-fitting lid, inserted in an asbestos board in the normal manner and for the usual time. Cool the crucible, transfer the complete contents to a well-used borosilicate glass beaker, and allow the sintered mass to disintegrate completely. Filter into a 250ml. volumetric flask and wash well with hot water, using as little volume as is necessary. Depending upon the expected amount of alkali present, it may be advisable to use a 500-ml. or 1000-ml. volumetric flask, The calcium is present, in the main, as chloride with some hydroxide. To prevent precipitation of the hydroxide, neutralize with a few dro s of hydrochloric acid. Using the same amounts of calcium cargonate and ammonium chloride, run a blank in an exactly similar manner. I t is antici ated that with a final volume of 250 ml. and with careful and simiyar treatment and washing, the calcium content as oxide will ran e somewhere around 1500 to 1800 p.p.m. This work shows t i a t standards made with either amount gave very comparable results. Add sufficient lithium nitrate, if the internal standard method is used, to ive an e uivalent of 100 p.p.m. as lithium metal in the final sofution. 8ecause this is merely used as a reference standard, the actual amount added is not critical, but it is absolutely essential to add exactly similar quantities to all the samples and all the standards. Consequently, use stock solutions made so that at least 5 ml. can be measured exactly, preferably from a microburet. Bulk the Sam le solutions to volume. Read on the flame photometer and use tEe calibration curves to determine the alkali contents. RESULTS
The results of the J. Lawrence Smith method taken to the total chlorides stage, flame photometer determination of sodium and potassium on these total chloride solutions, flame photometer determination of sodium and potassium obtained by the method
30 21 20 I5
IO 5 0
‘IO‘
20
30
YO
SO
60
P W S PEll Mwol
10
80
W
100
110
120
130 1UO
SWUM IJL WTASSIUM
Figure 2
Table I1 shows the actual calcium as oxide found in the solutions and blank made up to 250 ml. DISCUSSION
The possibility that ions other than the alkalies, calcium, and chloride might be present and so produce interference in the flame photometer should be considered. This possibility is, however, rather remote, for refractory materials do not normally contain soluble materials other than the alkalies, calcium, and magnesium. The method of treatment-sintering with calcium carbonate and ammoniun chloride-renders insoluble all metals other than the alkalies, calcium, and possibly minor amounts of magnesium, amounts that would have negligible effect on the flame photometer readings. Some sulfate as calcium sulfate might be present. However, sulfate is not usually present in refractory materials, and the solubility of calcium sulfate is
1017
V O L U M E 2 2 , NO, 8, A U G U S T 1 9 5 0 small. The presence of 12% boron BaOa in the borosilicate glass analyzed did not prevent accurate results from being obtained by this method. The gas used with air in these experiments was acetylene. Substitution of propane, using an appropriate burner, might be of some advantage, aa there would then be a lower rate of excitation of any nonalkali metals present. Sodium and potsssium excite more readily than most other commonly present metals. The photometer readings were made on the instrument without previously cleaning the burner which had been in use for some time. Although it is well always to keep the burner clean, the results obtained indicate that good work may be done even when the instrument is not in optimum condition. It is important that the flow of air and of gas do not fluctuate during the readings. To this end it is advisable to insert a ma-
nometer in each line. If the air is supplied from an air pump, the presence of a large vessel in the line will help to equalize pressure. LITERATURE CITED
Berry, J. W., Chappell, D. C., and Barnes, R. B., IND.ENQ. CHEM.,ANAL.ED., 18,19-24 (1946). (2) Bills, C. E., McDonald, F. G., Niedermeier, W., and Schwarta, M. C., ANAL.CHEM.,21, 1076-80 (1949). (3) Hillebrand, W. F., and Lundell, G. E. F., “Applied Inorganio Chemistry,” pp. 788-92, New York, John W h y & Sons, 1929. (4) Parks, T. D., Johnson, H. O., and Lykken, L., ANAL.CHEM.,20,
(1)
822-5 (1948).
Perkin-Elmer Corp., “Model 52A Instruction Manual.” (6) Pratt, P. F., and Larson, W. E., ANAL.CHEM.,21, 1296 (1949). (5)
RECEIVED January 19, 1950. Presented before the Meeting-in-Miniaturea SOCIETY, Newark, N. J., January North Jersey Section, AMERICANCHEMICAL 9, 1950.
Hafnium-Zirconium and Tantalum-Columbium Systems Quantitative Analysis by X-Ray Fluorescence L. S. BIRKS A N D E. J. BROOKS U . S. Naval Research Laboratory, Washington, D . C. The x-ray fluorescence analysis method was adapted to the determination of small amounts of hafnium in zirconium and tantalum in columbium. Curves of the relative intensity of spectral lines of tantalum and columbium or hafnium and zirconium were plotted against percentage composition by using prepared standard compositions. As an example of the accuracy attainable, with a counting time of 3 to 5 minutes on a 0.5 atomic 70 specimen of tantalum in columbium, the probable error in tantalum content due to statistical fluctuations of the Geiger counter was 0.029” or 4% of the amount present.
Q
UAXTITATIVE analysis of the hafnium-zirconium and tantalum-columbium systems is difficult by standard chemical methods. It is also somewhat difficult by spectrochemical means; however, Feldman (3)has recently published very good
-
/I\
/ I\ ’
II
t t
I I
Figure 1. Principles of X-Ray Fluorescence Analysis Method
data on spectrochemical analysis of samples containing less than 1% by weight of hafnium in zirconium. Most present interest lies in the low-hafnium or -tantalum end of these systems. Hafnium usually occurs as small impurity of about 1.5 weight % in zirconium deposits and is very difficult to separate from the zirconium. X-ray analysis has been used extensively for the analysis of the hafnium-rich end of the hafnium-zirconium system since the discovery of hafnium by Coster and von Hevesy (S). Its application to the hafnium-poor end of the system is shown in this paper. Hafnium was discovered by its x-ray L spectrum when a sample of zircon was placed on the target of an x-ray tube and excited by electron bombardment. Placing the specimen on the target of 8 demountable tube is inconvenient, however, With the develop ment of the x-ray fluorescence method (4) where the specimen is outaide the tube, the procedure is simpler and faster. The principles of x-ray fluorescence are shown in Figure 1. The method has proved valuable in measuring the small concentrations of lead and bromine in gasoline and in similar problems and would seem to be directly applicable to the small concentrations of hafnium in zirconium or tantalum in columbium. Usually there is no overlapping of the x-ray spectral lines from elements with similar chemical properties because they fall in the same column of the periodic table and the difference in atomic number is such that the x-ray spectra are well separated. However, in the hafnium-zirconium and tantalum-columbium systems, the difference in atomic number is such that the K series spectrum of airconium dzracted in the second order by the crystal overlaps somewhat the L series spectrum of hafnium. The same is true of columbium and tantalum. Therefore, the method of analysis waa more complicated than with most systems. The wave lengths of