Determination of Oxygen in Niobium Pentachloride by a Bromination Technique SIR: Niobium pentachloride is a crystalline yellow solid, n1.p. 204.7' C., which is readily hydrolyzed by moisture to NbOC13, and ultimately to hydrous NbzOs. Because of this tendency and the bad effect the resulting oxygen content has on many of its end uses, a direct method of analysis for oxygen was sought. Although methods exist for the determination of oxygen in niobium metal, including vacuum fusion ( I ) , emission spectrography (4), and inert gas fusion ( 5 ) , no reference to the determination in its salts was found in the literature. Since both NbCls and NbOC13 are volatile, an analysis based on a vapor phase reaction was thought feasible. A possible candidate was the bromination method, first described for the determination of oxygen in titanium metal (8, 3). This has been applied in somewhat modified form to the determination of oxygen in liquid Ticla (6). I n this modification, the gold boat containing graphite was replaced by filling the reactor tube with platinized graphite. The liquid sample was introduced by pipet. In the present work the main features of these methods were retained, but a special system was designed for introducing powdered samples. The overall reaction involved is believed to be stoichiometrically equivalent to : NbOC13
+ Brz + C
-t
NbBrzCL
+ CO
(1)
and
EXPERIMENTAL
Apparatus and Reagents. The complete bromination train is similar in its general features to t h a t of Codell and Norwitz (8). Most of the parts are similar in design and function to those of their apparatus, with some exceptions to be described. Sampling tubes are made from 18-mm. borosilicate glass with a standard taper 10/30 joint a t the top, a No. 6 stopcock, and a standard taper male 24/40 joint a t the bottom which fits into the reactor tube. The overall length is 10.5 inches. The reactor tube is made from 26-mm. quartz with a standard taper male 24/40 joint at the front end and a standard taper female 24/40 joint on top of the tube approximately 3 inches from the front end. The rear end has a male 18/9 ball joint. The overall length of the tube is 30 inches. Three cold traps are made from borosilicate glass with 18/9 ball joints and 24/40 standard taper joints. The traps should be small enough to fit inside a 500-ml. 1618
ANALYTICAL CHEMISTRY
n-
SAMPLE 95OOC.
Ar-+
I -b
1"J
H.
4
I -7WC. Figure 1 .
Complete bromination apparatus
Dewar flask. The three furnaces are standard 1-inch tube furnaces, 12 inches long, with powerstat control and some form of pyrometer suitable for measurements in the region of 900' C. The reagents also are similar except for 10% platinized graphite which is prepared as follows. Dissolve 30 grams of platinum metal in aqua regia. Stir in 270 grams of 8-16 mesh granules made by grinding and sieving AGR grade graphite rod. Slowly evaporate on a hot plate while stirring until dry, then ignite a t 900' C. for 1 hour in a muffle furnace. Procedure. The reactor tube is packed as follows: Loosely pack a 0.5-inch plug of Vycor wool in the main reactor tube a t the exit end. Roll a piece of platinum sheet 12 by 3.5 by 0.005 inches into a cylinder about the same sizeas the reactor tube. Force it into the reactor against the Vycor plug and seat it against the walls of the tube with a glass or metal rod. Add approximately 6 inches of the 10% platinized graphite and pack gently. Add approximately 6 inches of graphite granules (8-16 mesh, AGR rods, ground and sieved) and pack gently. Add a 1- to %inch plug of Vycor wool and pack tightly. If the packing contacts Vycor or silica a t the furnace temperature, it will react to form CO; therefore, any such contacts should be outside the furnace hot zone. Fill the bromine bottle approximately half full and then add about 0.5 inch of concentrated sulfuric acid. Purge the bottle with argon for 1 hour to expel any oxygen before assembling in the train. Assemble the apparatus as shown in Figure 1. The reagents in the tubes
are held in place with Vycor wool. ,411 stopcocks and glass joints are preferably lubricated with Kel-F grease, although other lubricants will work fairly well. Adjust the argon flow rate to 100-200 ml. per minute. Adjust all furnaces to their operating temperatures (Figure 1). I n a dry box, transfer 5-10 grams of the sample into a tared sample tube. Remove the tube from the dry box, weigh and place it in the receptacle on the reactor tube. Allow approximately 0.5 hour for a complete purge, then weigh the U-tube containing the Ascarite. Continue to weigh the U-tube a t 10-minute intervals until no further COz pickup is noted, or until the COZ pickup rate is constant. Start the bromine flow by opening the exit and entrance stopcocks on the bromine bottle and closing the by-pass stopcock. Flame the exposed portion of the reactor tube thoroughly with a Fisher burner. Continue to weigh the U-tube a t 10- to 15-minute intervals to ensure against oxygen leaks. When the weight gain of the U-tube is constant, open the stopcock on the sample tube and allow the sample to drop into the borosilicate glass boat. -4 gentle tapping or vibrating will be required to get the sample into the boat. An electric vibrator is very useful for this purpose. Ignore the material that clings to the walls of the sample tube. If the sample is large, it will be necessary to move the boat forward as it is being filled. A glass- or Teflon-covered magnetic pusher and small horseshoe magnet is used to accomplish this. Close the stopcock on the sampling tube and move the boat far forward against the Vycor wool plug with the magnet and pusher. Flame the reactor
Table 1. Comparison of Bromination Results with Neutron Activation
Per cent oxygen By activation Difference, BY wit,h 14 bromination m.e.v. neutrons 70 O , l l , 0.09 0.28, 0.28 0.10, 0 . 0 8 0.16,a . . . 0.03, 0 . 0 3 0.05, 0.05 a
0.12 & 0.02 0.31 i 0 . 0 3 0.06 f 0.02 0.25 i 0.02 0 . 0 3 + 0.02 0.06 & 0.02
-0.02 -0.03 $0.03 -0.09 0.00 -0.01 There was insufficient sample to run a
neutron activation did not possess high precision itself. As a n additional check, a sample of NbC15 was analyzed by the bromination technique and found to contain 0.06% oxygen. It was then split into four portions, each of which was exposed t o the atmosphere for a different length of time and aliowed to react with atmospheric moisture according to Equation 3 . NbCls
+ H20 * KbOCla + 2 HC1
(3)
duplicate.
tube gently, starting a t the sampling tube and working toward the boat. Increase the heat until the sample starts to boil. Xaintain a sufficient temperature until the sample has distilled slowly through the reactor tube. The NbOC13 may be seen as a white tail on the \-ellow XbC15 deposit. Keigh the U-tube after 0.5 hour and then a t 10-minute intervals until the C 0 2 pickup rate is constant (0.5 hour is usually sufficient). The blank is determined by weighing the U-tube before and after the same length of time required to run the sample. If the system has no leaks and there is no carbon touching the quartz tube, the blank should be less than 0.5 mg. RESULTS AND DISCUSSION
It was not feasible to attempt to prepare absolute standards. Therefore, checks on the precision and accuracy of the methods were made in three ways: samples were run by this method and by neutron activation, samples with known additions of oxygen were analyzed, and a standard deviation was obtained by analysis of pairs. Table I shows a comparision of results by two methods. I n general, there is good agreement, although it is to be regretted that at that time the
The amount of conversion of NbC15 to NbC13 was determined by weighing each sample before and after exposure. (It was assumed that the by-product HC1 formed was not appreciably adsorbed by the NbC15.) The samples were then reanalyzed and the results are shown in Table 11. The precision of the method was determined by running duplicate analyses on different samples having a wide range of oxygen values. These results are shown in Table 111. From these figures, the standard deviation calculated by the formula S = ( 2 ~ i ~ / 2 n )is~ ’=t0.009%. *, The method may be potentially applicable t o the determination of oxygen in other metal chlorides, although each such compound would require individual study. Some preliminary work has been done on TaC15 and XoCls, which shows promise, but has not been checked with known standards. Because the blank, under proper conditions, is low enough, the authors feel that the method could quite likely be modified to improve the precision and sensitivity by replacing the U-tube with a more sensitive detecting device such as a conductimeter. ACKNOWLEDGMENT
The authors thank V. P. Guinn and his staff a t General Dynamics, Division
Table
II.
Addttion of Known Amounts of Oxygen
Millimams of 0, Present Added Total Found 2.8 2.5 2.1 2.0
3.9 21.7 16.2 8.6
Table 111. Stauffer ’s sample no.
807 34-17-1 34-13-4 8B-24-115 P-2588-5 8B-50-406 1 2 3 5 6
6.7 24.2 18.3 10.6
7 23.7 17.6 10.4
Difference +0.3 -0.5 -0.7 -0.2
Analysis of Pairs
Or found, % 0.07, 0 . 0 6 0.13, 0.11 0 . 0 3 , 0.03 0.22. 0.22 0.18; 0.19 0.08, 0.06 0.11, 0.09 0.28, 0.28 0 . 1 0 , 0.08 0.03, 0.03 0.05, 0.05
of General Atomic Corp., San Diego, Calif., for their performance of the neutron activation measurements. LITERATURE CITED
(1) Am. Soc. Testing Mater., Kiobium
Task Force, Oxygen Subgroup, Division M, Committee E-3, 1960. (2) Codell, XI., Piorwitz, G., ANAL. CHEM.27,1083 (1955,). (3) Codell, M., Norwitz, G., Ibid., 28, 2006 (19561.
v.
A., Ibid.,
J., Banks,
C.,
G., J . Japan
JOHN C. ~ I C K A Y JOHN F. BELOW Western Research Center Stauffer Chemical Co. Richmond, Calif. 94804
VOL. 37, NO. 12, NOVEMBER 1965
1619
