tinuing interest and support of C. F. Metz are gratefully appreciated. LITERATURE CITED
(1) Bricker, C. E., Waterbury, G. R. ANAL.CHEM.29, 558 (1957). ( 2 ) BuLermak, J., Lew, M., Matlack, G., lbid., 30, 1759 (1958). (3) Hahn, R. B., Ibid., 23, 1259 (1951). (4) Hahn, R. B., Baginski, E. S., Anal. Chim.Acta 14, 45 (1956). (5) Hahn, R. B., Skonieczny, R. F.,
A’ucleonics 14, No. 4, 56 (19%). (6) Hume, D. N., “Radiochemical Studies: The Fission Products,” C. D. Coryell and N. Sugarman, eds., National Xuclear Energy Series, Div. IV, Vol. 9, p. 1499, McGraw-Hill, New lg51. (7) Klingenberg, J. J., “Organic Syntheses,” T. L. Cairns, ed., Vol. 35, p. 11, Wiley, New York, 1955. (8) Klingenberg, J. J.1 PaPucci, R. A., ANAL.CHEM.24, 1861 (1952). (9) Kumins, c. A., Ibid., 19, 376 (1947). (10) hforris, D. F. C., Scargill, D., Anal.
Cham. Acia 14, 57-61 (1956). (11) Oesper, R. E., Klingenberg, J. J., ANAL.CHEM.21, 1509 (1949). (12) Stanley, C. W., Ford, G. P., Lang, E. J., U. S. Atomic Energy Commission Rept. LA-1721 (Rev.), J. Kleinberg, ed.,
December 1956, unclassified. (13) Youden, W,. J., “Statistical Methods for Chemists, p. 12, Wiley, New York, 1951. RECEIVED for review June 6, 1960. Accepted July 18, 1960. Work done under the auspices of the Atomic Energy Commlsslon.
Determination of Oxygen in Zirconium and ZircaIoy by the Inert Gas Fusion Method Pi- INEAS ELBLING and G. W. GOWARD Bettis Atomic Power laboratory, Westinghouse Electric C o p , Pittsburgh, Pa. ,The important effects of oxygen content on the physical properties of zirconium and Zircaloy necessitated a rapid and accurate method for determining oxygen in these metals. The inert gas fusion method has been found satisfactory for this determination. Oxygen is released from the metal as carbon monoxide by reaction with carbon in the presence of molten platinum in an induction-heated graphite crucible. The carbon monoxide is swept from the reaction furnace with argon, oxidized to carbon dioxide, and determined conductometrically. A minimum weight ratio of platinum to sample of 7 to 1 is necessary for complete removal of oxygen from samples. The coefficient of variation of the method is about +6% at the 1000-p.p.m. level of oxygen. Analysis of samples containing known amounts of oxygen in the range of 900 to 3700 p.p.rn. has shown that the method is quantitative within the limits of precision.
widespread use among producers and users of zirconium. Codell and Norwitz (a) have applied the bromination method to the determination of oxygen in zirconium. This procedure involves the reaction of the metal and its contained oxygen with bromine and carbon a t 825” C. in the .presence of an inert carrier gas to produce carbon monoxide. The carbon monoxide is swept from the reaction tube, oxidized to carbon dioxide, absorbed on Ascarite, and weighed or determined conductometrically (3). The inert gas fusion technique was
first described by Singer (7) and later improved by Smiley (9). The method is analogous to vacuum fusion with the vacuum replaced by an inert sweep gas, thus eliminating the difficulties associated with the production and maintenance of high vacuum. As in the vacuum fusion method, carbon reacts with the oxygen in the sample to form carbon monoxide, usually in the presence of a molten metal, such as platinum (9), as the reaction medium. The resulting carbon monoxide is swept from the furnace tube and oxidized to carbon dioxide by means of heated copper oxide (7) or iodine pentoxide
S
EVERAL ?~IETHODS have been used,
with varying degrees of success, for the analysis of zirconium for oxygen content. The hydrogen chloride method, as developed by Read and Zopatti (6), has been used until recently in this laboratory for acceptance and engineering testing. The adaptation of vacuum fusion techniques for the analysis of oxygen in zirconium has been reported by several authors (6, IO). Sloman and Harvey (8) discussed the fundamental reactions in the vacuum fusion method and described its application t o the determination of oxygen in zirconium. Because of the delicate and expensive nature of the apparatus required and the need for highly skilled operators, the vacuum fusion method has not found
1610
ANALYTICAL CHEMISTRY
TO L E C O CONDUCTOMETRIC UNIT
Figure 1 . 1.
Inert gas fusion apparatus
Two-stage regulator Third stage regulator 3. Purifying train A. Sulfuric acid tower 8. Ascarite and magnesium perchlorate C. Flowmeter A. Needle valve 5. Neoprene tubing 6. Argon purification tube A. Nickel tube 8. Glars wool C. Zirconium chips D. Split-tube furnace, 700’ to 800’ C.
2.
7. 8. 9. 10. 11. 12. 13. 14.
Hose clamp Vycor furnace tube Radiofrequency coil Crucible thimble assembly Cooling blowers Sample storage arm Sample loading port Oxidizer furnace A. Iodine pentoxide 8. Glass wool C. Electrical heating t a p e D. Sodium thiosulfate
1530 1500-
1400 1300
2
e
-
1200-
e
e
e
e
e
e
e
m
e
1000
1000-
800
e
e e
e
900 BOO
L
e
1400
e
2 1100 -
E
e +e
e
e
e
700
700 600 600
I
I
Figure 2. left.
I
I
I
I
I
I
I
Effects of platinum-to-zirconium ratio for identical samples wrapped in platinum Analyzed in dry crucible
(9). The carbon dioxide is then measured gravimetrically after absorption on Ascarite (?'), manometrically with a capillary trap (9), or conductometrically. An instrument, based on the latter principle of measurement, has been designed and marketed by the Laboratory Equipment Corp. of St. Joseph, blich. Considerable unpublished development work on the application of the inert gas fusion method to a variety of metals has been done by E. L. Bennet and associates ( I ) . This paper describes the application of the inert gas fusion method to the determination of oxygen in zirconium and its principal alloy, Zircaloy 2. Optimum fusion conditions necessary for obtaining consistent results have been established. The absolute nature of the method as applied to zirconium has been established by the preparation and analysis of samples containing knoi5n amounts of oxygen.
EXPERIMENTAL
Apparatus and Reagents. T h e main components of the apparstus constructed in this laboratory, and assembled as shown in Figure 1, are as follo.ivs: ARGOKPURIFICATION TRAIN. Sulfuric acid bubbler, magnesium perchlorate-Ascarite absorber, and hot zirconium chips (800' C.1. POWER SOURCE.Lepel 2.5-kw., 450kc. radiofrequency generator. F u ~ r ; a c s ASSEMRLYAND SAMPLE LOADING-STORAGE DEVICE. The furnace tube is fabricated of Vycor and uses a 65/40 ball joint' for c!osure of the thimble entry end of the tube. The qi.!artz thimblc, containing the graphite crucible a!?.d graphite insulation, is supporttd inside the furnace tube by a quartz tube ns s h o v in ~ Figure 1. The sampk loading-storage arm connects to the top of the furnace tuhe ky a 35,/25 ball joint and is of sufficient length to hold 5 simples. CARBOX hfONoxIi)x OXIDIZER, iodine j,entoxicie (Labomtory Eqi~ipmect
Right.
Corp. NO.501-96) is contaiiied in a 1 X 24 em. borosilicate glass tube maintained a t 130' to 150' C. by means of electrical heating tape. The iodine pentoxide is followed by a packing of sodium thiosulfate to absorb the evolved iodine. CARBON DIOXIDE ANALYZF,R. Laboratory Equipment Corp. h-0. 515 conductometric carbon determinator. GRAPHITECRUCIBLEAND GRAPHITE FUNNEL.United Carbon Products KO. PE-5-20-58 and No. PE4-21-58, respectively. CAHN ELECTROBALAKCE. This balance has been found satisfactory for weighing oxides used for calibration. SODIUMHYDROXIDE SOLUTION.Dissolve 4 grams of sodium hydroxide and 2 grams of white gelatin in 500 ml. of carbon dioxide-free water. Filter this solution into 18 liters of carLon dioxidefree water. Add 3 nil. of 2-ethy1-1hexanol, st,opper, and mix thoroughly. PLATINUM. One- to three-mi! foil is convenient for wrapping samples. Chopped l/s-inch thick sheet is used for addition of platinum to the bath, h;lother chemicals used are of reaeent g~iiil: quality. Prxedure. Place a graphite crucible, fitted with a funnel as shown in Figure 1, in the quartz thimble. Surround the crucible, to the top, with lightly packed 200-mesh spectrographic grade graphite. Place the thimb!e in its holder and assemble the furnace as shown using Dow Corning silicone high-vacuum grease on the lower ball joint. Admit purified argon into the furnace assembly over the crucible and out' the sample loading port a t a monitored rate of 0.35 liter per minute. Outgas the crucible by heating a t 2500' to 2600' C. for 30 minutes. Lower the temperature to 21C)OO C. and close the sample loading port., thereby shunting the gas stream through the iodine pentoxide and into the sodium hydroxide solution of the Leco conductometric apparatus. With a fresh change of solution in the absorption ce!!, Ilieasiire the resistance change of the solution for a timed period, iimally 10 mingtes, to determinp the blank rate. If ihe biank rate is greater than 0.1 ohm ;,CY minute (abour 2.0 pg. of oxygen u
Analyzed in platinum bath
per minute), continue the outgassing procedure until the desired rate is obtaiued, Prepare platinum (0.1 to 0.15 gram) foil packages containing 0.0 (blank), 1.00, 2.00, and 3.00 mg. of uranium oxide (U308) (0, 133, 306, 468 pg. of oxygen). Alternatively, appropriate amounts of silvei. oxide, zirconium dioxide, or niobium pentoxide may be used. Load the foil packages into the sample storage arm, manipulating the sample loading port so that atmospheric gases are prevented from entering the apparatus. Fill the absorption cell with fresh solution and balance the conductance bridge with argon flowing through the cell. With a magnetic pusher, manipulate a foil package into the hot crucible and allow the oxide to react while sweeping with argon for 10 minutes. Balance the bridge and take t'lie reading of the resistance duodial. Repeat for each of t,he t the resistance read! P standards for the blank, and plot thc corrected readings cs. known oxygen content of the standards (in microgracis) on linear coordinate paper. The graph is a straight line up to 700 pg" of oxygm. Prepare a new curve for each bottle of sodium hydroxide soliution. Check the curve daily with rtt ieast two calibration standards. Cut a 0.1- to 0.3-gram solid sample from the specimen to be analj.zed. abrade OR any heavy oxide coating. Degrease ti.? sample with 1,1,2-t,richloroethylcne followed hy a rinse in acetone. Allow to dry and then weigh. ;'ease, in the same piece of 1- or 3-mil If the weight of the foil is not 7 h r s that of the sample, make i i p the diffcrmce to the required platinum ratio wit:! pieces of '/s-inch thick pheci drc3i.,!~d into the hot crucible. C ~ : x i u l l ywrap the zirconium sampk with the platinum sheet. taking care thfJt ihi: samplr and the platinum do not becoine contaminated during this stcp. Load the nrapprrl samplrs into t'he storage a r n thro:~giit h e sample lyze the samples in the oxide standards, ci:cible trmperature
f6
TO VACUUM
Apparatus for charging zirconium with oxygen
Figure 3.
1-4. Platinum-wrapped zirconium 5. Silver oxide decomposition tube 6. Silver oxide packages 7. Zirconium getter
at 2100' to 2200' C. Calculate the oxygen content of the samples by reference to the previously determined calibration curve. Experimental Investigation of PlatiAlthough num-to-Sample Ratio. many investigators have reported vacuum fusion work carried out by means of a platinum bath, none have systematically investigated the platinum-to-sample ratio required. To determine t h e minimum ratio necessary, samples from a strip of Zircaloy 2 were analyzed under conditions of various platinum-to-sample ratios. The ratio was vaned by using different wrapping weights and dropping the wrapped samples into a dry crucible or by wrapping the samples in small constant weight of platinum and varying the ratio by the addition of platinum to the crucible prior to analysis. I n agreement with the experience of Hansen and coworkers (4) on titanium, analysis of unwrapped samples dropped into a bath is generally unsatisfactory; erratic results are obtained. The results of the above experiments are presented graphically in Figure 2. These results indicate that the minimum platinum-zirconium ratio for a
Table I.
Sample No. 7-2 -3 -4 -5 -6 -9 -10 -11 -7 8-2 -4 -5 -6 -7 -8
1612
Sample
Wt., G. 0 . i76 0.171 0.141 0.142 0.152 0.143 0.316 0.412 0.257 0.194 0.228 0.120 0.137 0.144 0.201
8. 9. 10. 11.
Movable furnaces Quartz heating mantle Thermocouple vacuum g a g e Liquid nitrogen freeze trap
successful analysis is in the range of 7 to 1. It is of no consequence whether the sample is wrapped with this ratio of platinum and subsequently analyzed in a dry crucible or sparsely wrapped and analyzed in a platinum bath. Precision of Method. The precision of calibration of the method has been established by the analysis of weighed amounts of uranium oxide. Analysis of 60 samples of this oxide over a 2-month period indicated a coefficient of variation of 4.5yc in the range of 150 t o 500 pg. of oxygen Within experimental error, analyses of stoichiometric oxides of silver, zirconium, and niobium yield calibration curves of the same slope as that obtained with uranium oxide. Analysis of 43 samples from a strip of Zircaloy 2 over a period of 2 months indicated a coefficient of variation of 6.1%) not statistically significantly different from the precision of calibration or analysis of pure oxides. The average oxygen content of this material was determined to be 1310 p.p.m. Preparation and Analysis of Standard Samples. T o determine the accuracy of the method, samples containing known amounts of oxygen
Analysis of Standard Samples Oxveen. ue.
Amt. in starting crystal bar zirconium 13 1R 11 11 11 11 21 31
w
15 18 9
IO 11 1,5
ANALYTICAL CHEMISTRY
Added 274 347 363 492 552 3 16 297 347 288 169 274 338 407 49 1 556
Corrected Recovery, added Recovered %
Oxygen,
105 98 104 100 99 101 97 111 102 107 89 108 103 98 104 Average 102 Coeff. of variation 5.3
1690 2070 2770 3540 3680 2310 980 1020 1270 1010 1130 3110 3110 3400 "50
287 360 374 503 563 327 321 378 308 184 292 347 417 502 571
297 354 390 502 560 330 310 420 315 197 258 375 429 490 ,592
were prepared and analyzed by the described procedure. The standard samples w r e prepared by the addition, in vacuum, of known amounts of oxygen, obtained from the thermal decomposition of silver oxide, to high-purity crystal bar zirconium, initially containing about 7 5 p.p.m. of oxygen. The apparatus used for the addition of oxygen is shown schematically in Figure 3. Small (0.15 to 0.3 gram) solid pieces of the low-oxygen zirconium were first manually abraded to remove surface oxide, weighed, and then sparsely wrapped with 1-mil platinum foil. The silver oxide was dried a t 110" C. and then weighed on a Cahn Electrobalance, accurate to 1.0.4%. The samples to be charged with oxygen and the silver oxide packages were then loaded into the charging apparatus (Figure 3). A vacuum of about 20 microns was obtained by means of a rotary oil pump. The system was isolated from the pump by means of a stopcock and was then gettered free of residual atmospheric contamination with hot zirconium. A liquid nitrogen trap assisted in obtaining a vacuum of less than 1 micron. The getter material was then cooled, and tubes holding the silver oxide and a sample to be charged were heated to 500" C. and 1000" to 1200" C., respectively. The oxygen was usually absorbed by the sample in less than 15 minutes, as indicated by the return of the system to less than 1micron pressure. The sample was then annealed for 2 l / 2 hours at 1000" to 1200" C. to diffuse the ox!-gen into the metal. The process was repeated for each sample in the system. .halysis of the depleted silver oxide packages indicated that all of bhe oxygen had been driven from the oxide. The individual samples were then analyzed, still in their original platinum wrapping, by the described procedure. The results of these analyses (Table I) indicate an average recovery of 1027c oxygen with a coefficient of variation of 5.3%, which is not st'atistically different from the previously e.stablished precision of the method.
P.P.M. DISCUSSION
The method as described provides a rapid and reasonably precise and accurate measure of the oxygen content of zirconium and Zircaloy. h single operator can perform from 20 to 25 sample determinations along with calibration and st'andard checks in one 8-hour shift. The crucible askembly is usually satisfactorily blanked out by a 20- t o 30-minute hmting period. The normal blank cw-responds to about IO fig. of oxygcn, n.hich is qiiite antkfactory for the :inaij in cornmeri>i;ii zirconilim
purchased to a specification of 1000 p. p .m . minimum- 1400 p .p.m. maximum. The main source of the blank is thought to be reaction of the graphite insulation bith the quartz thimble. Smiley (9) reports much lower blanks, using an uninsulated crucible heated by a rather high-powered radiofrequency generator. The choice of conductometric measurement of carbon dioxide in this laboratory was dictated mainly by the availability of the equipment and satisfactory sensitivity in the range of interest. Gravimetric measurement is quite feasible for materials of high oxygen content, while the capillary trap method described by Smiley (9) is an excellent method for measure-
ment of much smaller quantities of carbon dioxide. ACKNOWLEDGMENT
The technical assistance of P. E. Bauer in obtaining many of the reported data is acknowledged. The authors are also indebted to B. N. Nelson for statistical analysis of data.
( 6 ) Read, E. B., Zopatti, L. P., U. S.
Atomic Energy Comm., Rept. AECD2798 (February 1950). (7) Singer, L., IND.ENQ.CHEM.,ANAL. ED.12, 127 (1940). (8) Sloman, H. A., Harvey, C. A., J. Inst. Metals 80,391 (1951-1952). (9) Smiley, W. G., ANAL.CHEM.27, 1098 (1955). (10) Stanley, J. K., Von Hoene, J., Wiener, G., Ibid., 23, 377 (1951).
LITERATURE CITED
( 1 ) Bennet, E. L., et al., Laboratory
Equipment Corp., St. Joseph, Mich., unpublished data. (2) Codell, M., Norwite, G., ANAL.CHEM.
28, 2006 (1956). ( 3 ) Ibzd., 30, 524 (1958). ( 4 ) Hansen, W. R., Mallet, M. W., Trzeciak, RI. J.,Zbid., 31, 1237 (1959). ( 5 ) McDonald, R. S., Fagel, J. E., Jr., Balk, E. W., Zbid., 27, 1632 (1955).
RECEIVEDfor review March 14, 1960. Accepted June 27, 1960. Division of Analytical Chemistry, 138th Meeting, ACS, Sew York, N. Y., September 1960. Work supported by the United States Atomic Energy Commission under Contract AT-11-1-GEN-14 with the Westinghouse Electric Corp., Bettis Atomic Power Laboratory.
Determination of Oxygen in Yttrium and Yttrium Fluoride by the Inert Gas Fusion Method CHARLES V. BANKS, JEROME W. O'LAUGHLIN, and GEORGE J. KAMIN Institute for Atomic Research and Department o f Chemistry, lowa Stafe University, Ames, lowa ,The inert gas fusion technique has been applied to the determination of oxygen in yttrium metal and yttrium fluoride. The problem arising from the interaction of the fluoride with the glass apparatus was solved by the addition of a magnesium oxide reagent in the train. The method was shown to be simple, rapid, and precise. The results on yttrium metal were checked against the vacuum fusion and emission spectroscopic techniques; results on yttrium fluoride, against the vacuum distillation and KBrFl techniques.
I
1940, Singer (9) developed the inert gas fusion method and SUCCPSSfully applied it to the determination of oxygen in steel. Smiley (IO) significantly modified the technique to achieve sufficient sensitivity to allow successful analysis of samples of the order of 0.1 gram. A number of workers have since applied the technique to the determination of oxygen in steels (1, 7 , 8). Elbling and Goward (3) have reported the determination of oxygen in zirconium and Zircaloy. In the work described here, yttrium and yttrium salts have been analyzed using the commercially available Leco oxygen analyzer. A magnesium oxide reagent has been used to trap out silicon tetrafluoride formcc' during the yttrium fluoride analysis accl a platinum flux has been used instead of B plqtinum bath. N
EXPERIMENTAL
Apparatus and Reagents. T h e basic
instrument used n a s the I m o oxygen analyzer KO.534-300, consisting of a n induction furnace and conductometric analyzer. A schematic diagram of the apparatus is shown in Figure 1. The argon was led from the tank, 1, through a conventional valve, 2, and then through a Matheson low pressure Pancake regulator No. 70, 3, which afforded a careful control of the pressure a t 1.25 p s i . The argon entered the fused silica reaction chamber, 4, from below and swept past the graphite crucible, 5 , which was heated by the indli:!,ion coil, 6. The carbon moncvidc liberated from the melt was swept by the argon through Ascarite, 9, to remove any acidic gases, and through iodine pentoxide, 10, which was heated to 160" C. to convert the carbon monoxide to carbon dioxide 5CO
+
160' 1206
__*
5COn
+
12
The iodine liberated during this conversion was trapped by sodium thiosulfate, 11. A flowmeter, 12, was placed in the train, so that a flow rate of 250 ml. per minute could be maintained by means of the stopcock, 13, a t the base of the conductivity cell, 14. The carbon dioxide was measured by the change in conductance of a barium hydroxide solution. When samples of yttrium fluoride were analyzed, it vias necessary to place a magnesium oxide reagent, 8, in the liii.. Samples were introduced into the systcm b y means of the entrance port, 7 . Comniercially available argon was of sufficient purity so that a t oxygen levels of about 1000 p.p.m. no further purification ( I , S9 10) was necessary. This
is in agreement with the findings of Peterson et al. ( 7 ) . However, uranium turnings, heated to 300" C., can be used effectively to reduce the blank due to impurities in the argon (4) in cases where lower oxygen levels are being considered. The line from the tank to the reaction chamber was made of l/4inch soft copper tubing to minimize the probability of oxygen eaterlng the system. The argon flow rate was carefully maintained, primarily to prevent excessive frothing in the conductivity cell. The reaction crucible was held in a fused silica cun packed with carbon black which acted as an insulator. The crucible itself consisted of a high purity, graphite rod, l/g inch in diameter, about 21j4 inches long, with a center hole 6/* inch in diameter and 16/8 inches deep. This crucible was longer than that usually used with the instrument. The added length not only allowed more analyses to be performed without changing the crucible, but also prevented excessive drifting of the carbon black insulation into the crucible. Such drifting of the carbon black may cause erratic w u l t s . The small amount of carbon black which was swept out of the retaining cup by the flow of argon and drifted back into the crucible was minimized by the use of commercially available cai t 3n black, specially prepared for use in inert gas fusion instruments. This material was in the form of small gianules, rather than flocculent particles, and was not as easily blown about by the argon. The iodine pentoxide used was that furnishdby the manufacturer of the instrument. The Smiley modification of Schutzc' reagent (IO)waq also tried VOL. 32, NO, i 2, NOVEMBER 1960
1613
