precise than those from sulfuric acid solutions. This was not considered a disadvantage, as the samples dissolve faster in hydrochloric acid, and it has been the practice to determine chloride in titanium sponge by dissolving separate samples in fluoboric acid. Manganese, if present in more than the trace amounts normally found in titanium sponge, discolors the indicator. This effect was eliminated by adding 5 ml. of 5y0 sodium sulfide to 100 ml. of the sample solution after the precipitation and decantation step. The sample was boiled long enough to evaporate more than 10 ml., cooled, and again made up to 100 ml. if aliquot portions were to be used in the titration. All water used in preparing the reagents and in the procedure was passed through a 10-inch column of cation exchange resin, such as Amberlite IRC-50. Fluoride and phosphate were found to interfere by precipitation of the magnesium. Reducing agents, or strong oxidizing agents, had adverse effects on t h e indicator. It was necessary to keep the solution
volumes used in the titration of standards and samples nearly equal in order to obtain precise recognition of the end points. Careful control of the p H of the solution during the titration was essential to reproduction of results. The p H was adjus’ted to 10 by addition of ammonium hydroxide before starting the titration. Ordinary chelating agents, such as citric acid and tartaric acid, did not interfere with the chelation of magnesium by EDTA a t p H 10, but they had a detrimental effect on the color of the indicator. The data presented in Table I show that the method is accurate to within 1% of the true value over the range of magnesium normally found in leached titanium sponge. I n concentrations as low as 0.02y0 magnesium, the error is within 5% of the true value. ACKNOWLEDGMENT
This work was done with the cooperation of Robert L. Powell, Supervisor, Process Research Division, Titanium
hIetals Corp. of America. Helpful suggestions were made by John H. Hill. LITERATURE CITED
(1) Bersworth Chemical Co., Framingham, Mass., “The Versenes,” Tech. Bull. 2 , Section 1, p. 16B, 1954. (2) Brown, E. G., Metallurgia 49, 151-5 (1954). (3) Codell, M., Cherney, A., Am. Foundryman 24, No. 2, 65 (1953). (4) Flasks, H., Purschel, R., Chemist Analyst 44, 71 (September 1955). ( 5 ) Hillebrand, W. F., Lundell, G. E. F., Bright, H. A., Hoffman, J. I., “Applied Inorganic Analysis,” 2nd ed., p. 79, Wiley, New York, 1953. (6) Latimer, W. M., Hildebrand, J. R., (‘Reference Book of Inorganlc Chemistry,” p. 398, MacmilIan, New York, 1940. (7) Potter, V. G., Armstrong, C. E., ANAL.CHEM.20, 1208 (1948). (8) Tour and Co., Sam, 44 Trinity Place, New York 6, “Research and Development of Methods of Chemical Analysfii for Titanium Metal and Alloys, Final Tech. Rept., pp. 104-6, May 12, 1950. RECEIVED for review July 8, 1957. ACcepted November 15, 1957.
Determination of Oxygen in Titanium Modified Vacuum Fusion Apparatus and Platinum Bath Technique S. J. BENNETT and L. C. COVINGTON Process Research Division, Titanium Metals Corp. of America, Henderson, Nev.
b An improved vacuum fusion apparatus incorporates an improved furnace section, a sample entry port, and an analytical system. This apparatus and the use of a platinum bath have resulted in a simplified operation, giving fast and reliable oxygen determinations in titanium. Tin, aluminum, manganese, chromium, vanadium, molybdenum, nickel, iron, and zirconium do not interfere. HE osygen content in titanium must T b e clo,~ e l ycontrolled in the production of the metal, as oxygen greatly affects such physical properties as ductility, impact resistance, and hardness of the metal ( 1 ) . The determination of oxygen is made difficult by the tremendous affinity of titanium for this element. Of the several analytical techniques reported for this determination, the vacuum fusion method is generally accepted as the most reliable. Three vacuum fusion techniques have been developed for the determination of oxygen in titanium: the iron bath method ( g ) , the dry-crucible or Walter
method (6), and the platinuni bath method ( 7 ) (not as extensively investigated). A standard platinum bath procedure has not been adopted because of insufficient data t o establish any one technique as being completely satisfactory ( 3 ) . However, the method offers the following advantages: reduced cost per determination, as more samples can be run per setup without apparatus shutdown and setup conditioning; more rapid outgassing of samples because each complete analysis requires less than 45 minutes; and elimination of tin plugs and chips which require time for preparation and manipulation. This article describes an apparatus design and platinum bath technique which have overcome many of the prevailing problems of vacuum fusion analysis and have given rapid, reliable, and accurate oxygen determination in titanium. APPARATUS
The apparatus consists essentially of a furnace section, a n evacuation system, and an analytical system.
Furnace Section. This section, shown in Figure 1, is a n air-cooled Pyrex envelope (Corning 774). A 60/50 joint a t the bottom is sealed in place with Apiezon W wax and acts as a support for a clear quartz cup. This cup positions and holds a clear quartz thimble more uniformly and easily than when the thimble is SUSpended with wires from glass hooks (6). The thimble contains a graphite crucible surrounded by sub-200-mesh graphite powder. A clear quartz sleeve lines the walls of the furnace section and can easily be replaced between runs. The furnace section is thus kept clean without introducing acid and water in the system or without removing the furnace section after every run. Access to the furnace section is made rapid and simple by placing a 102/75 mercury-seal ball joint a t the top. This joint holds a tight seal under vacuum and is self-releasing when the system is open to the atmosphere, The sleeve and thimble are easily positioned through this joint. An optical flat on top of the joint gives a straight-line view to the bottom of the crucible for temperature measurements. The optical-flat tubing extends VOL. 30, NO. 3, MARCH 1958
363
through the socket joint almost to the crucible top and guides the samples directly into the crucible. Evacuation System. An entry port which permits samples t o be introduced singly into t h e system is fabricated from a vacuum valve (Veeco 4 mercury-seal joint is R-62-S). sealed into t h e bottom of t h e valve by which the port is attached t o t h e socket joint. A capped standardtaper joint is sealed into t h e other opening of t h e valve and acts as a n open-close chamber. The chamber is evacuated through one arm of a threen-ay stopcock sealed into the standardtaper joint tubing and opened to the atmosphere through the other arm. The cap is removed for sample insertion. Induction heating attains the desired temperature range (1800" to 2400" C.). The power leads from a 6-kn.. mercury spark-gap converter (iljax-Northrup) are attached to a helix coil of 0.25-inch copper tubing which surrounds the furnace tube. h rapid withdrawal of the liberated gases from the furnace section reduces gettering and accelerates the outgassing and conditioning of the furnace setup. To accomplish this, a high-speed (80 liter per second) mercury diffusion pump with three stages (H. S. Martin M-40110) is attached to the furnace section through 41-mm. tubing. This connection is made as short as possible and is most easily made by inserting a ball joint. d diagram of the apparatus is shown in Figure 2. The main pump is backed by a second diffusion pump to maintain optimum operation and pumping capacity. An Urrp-type automatic Toepler pump carries the gas from the second diffusion pump into the analytical system. Analytical System. This absorption-type system, which gives a rapid, accurate, and direct determination for carbon monoxide ( d ) , consists of a 200-ml. collection volume, a simple McLeod gage, and a n absorption flask. The absorption flask contains Hopcalite (Mine Safety Appliance Co.) to oxidize the carbon monoxide to carbon dioxide and Ascarite to absorb the carbon dioxide. These reagents have no measurable vapors or reaction products which interfere with the determination, and the absorption is rapid and quantitative. PROCEDURE
The setup is prepared by placing a %inch layer of sub-200-mesh graphite powder (Acheson Grade 38) in the thimble and inserting a graphite crucible and funnel (United Carbon Co. C-625 and F-703) until the pon-der is just at the upper edge of the crucible. A Vycor cap is held on top of the thimble with a looped platinum wire, through which a rod is hooked to lower the setup into the supporting cup. The head is placed on the furnace section and the sample valve is lowered into position. The system is evacuated with caution to avoid blowing the carbon powder. The temperature is raised to 2400" C. while the diffusion pumps are warming (about 20 min364
ANALYTICAL CHEMISTRY
The temperature is maintained a t 1900" C. throughout the run. The samples react more rapidly when dropped into the molten bath and no time is lost in heating and cooling the crucible. Spattering and gettering are insignificant, except when powdered samples are introduced. These are introduced when the bath is at 1200" to 1400" C. At times, the bath does not properly condition and low results are obtained. K h e n this occurs, the addition of a small amount of carbon poTvder wrapped in tin foil will properly condition the bath. The gases collected in the analytical system are analyzed by measuring the initial pressure and opening the absorption bulb. After the oxidation and the absorption of the carbon monoxide (about 5 minutes) the pressure is again measured. The differential pressure permits the calculation of the oxygen content from the gas formula:
%O Figure 1. 1.
2. 3. 4. 5.
Thimble Crucible Funnel Powder Shield
PV 1600 RTW
= ~-
where TV is the sample meight. A correction is made for the carbon monoxide content of the furnace blank, (about 60% of the volume of the furnace blank). Hydrogen and nitrogen are not ordinarily determined, but they can be
Furnace setup 6. Furnace head 7. Shield window 8. Optical fiat 9. Sleeve 10. Sample port
llTO MAIN VACUUM SOURCE
1;
Figure 2. Vacuum fusion apparatus
% J%ED ,G I TVACULM 0 AUXILIARY s o, uw
i AUTOMATIC T E P L E R %MP
'- CONTROL BOX
Utes). This temperature is maintained for 60 to 90 minutes to ensure a low blank in the system. A blank of 0.02 ml. (KTP) per 30 minutes a t 1900" C. is considered satisfactory. Forty-five grams of platinum are dropped into the crucible at 1400" C. A 0.4-gram piece of titanium is also introduced to condition the bath. The temperature is raised to 1900" C. and the platinum is outgassed to the furnace blank. This completes the furnace conditioning. Samples are prepared by cutting a 0.15- to 0.25-gram piece from the parent material and leaching in approximately 10% hydrofluoric acid to remove surface oxides. Each sample is washed in water and acetone before it is weighed. To introduce a sample into the system, the loading arm is opened to the atmosphere, a sample is inserted, and the arm is evacuated. The valve is then opened until the sample falls through and into the bath.
- __
,
T3ERMCCGUPLE GAGE
measured by outgassing the hydrogen quantitatively a t 1400" C. The nitrogen is then determined as the residue after absorption of the carbon monoxide from the gases collected a t 1900" C. During the absorption of the gases, 3 second sample is introduced into the bath and the gas is collected in the Toepler pump. On completion of the analysis of the first sample, the gas from the second sample is pumped into the analytical system for analysis. The bath is replenished m-ith 45 grams of platinum after the analysis of 30 samples. This permits an additional 30 samples to be analyzed before the system is opened to the atmosphere. Samples can be analyzed until the titanium content of the bath reaches about 12% by meight. Beyond this point, Ion- results are obtained. Frequently, after prolonged use, the metal bath \\-ill seep through the crucible walls. B slight seepage neither interferes with the results nor ib serious, but large
amounts can cause powder blowing or other damage. The bath is not affected by shutting down at night. A blank is obtained the following day as soon as operating teniperature is reached.
Table 1.
% Oxygen
Sample No.
SUMMARY
The initial outgassing of the setup is complete within 1 hour a t 2400" C. The previous a x r a g e of 4 hours encountcrctl a t this laboratory has been
Sample S o 11-1
131 133 130 141 131 133 136 126 126 0.126 0.127 0.121 0,127 0.130
9- 1
0 0 0 0 0 0 0 0 0
RESULTS AND DISCUSSION
ilnalyticalresults, which are representative of the several hundred determinations made with the technique described, are sho\\-n in Table I. These determinations n-ere made on pure titanium saniplm submitted in connect'ion with an oxygcm control project. Analysis showed an oxygen content and standard deviation of 0.1307c oxygcn, 0.005 u and 0.058yG oxygen, 0.0013 u respectively. Thc~pcJssihility that the platinum bath nietllod m q - not bc equally applicable to :ill t,it:miuni alloys has been consid(wcl 13). Titanium alloys have bcen preparcd bj- mixing titanium sponge n-it11 r h i p of otlicr metals. Two buttolie \wrc p r t p r c t l of each coniposi110y~the other as an osygc~ii :ititlition containing the csact amount of titanium oxide (TiO,) to equal 0.199yc os\-gcn by w i g h t . The an:ilyticul iiictliod \\-:IS evaluatcd by coiiiparing the imount of ox!-gc'n nddcd to each alloy n-ith blie amount recovered. The analysis of the corresponding basealloy button was used as the blank. The recovery on every alloy was within thc limits of analytical precision and shoncd no erratic results nor evidence of intcrfcrtnce. The analytical results on tlic. buttons are listed in Table 11.
Unalloyed Titanium Analysis
70 O w P 0 0 0 0 0 0 0 0 0 0 0 0 0
.iverage Standard der.
II.
Alloyed Titanium Analysis
Baee 0 061 0 063 0 056
TI C.P. Ti 57c Sn T i s % AI Ti 5Yc ;\In 0 072 Ti557 Cr 0 065 T i s ? V 0.191
Ti 5yc S i 0.081 Ti 5% FP 0 075 Ti % 0 067 T155; 110 0 063
0.1997c 0 2 Addition 0 0 0 0 0 0
Diff. 194 197 196
255 260
0 0 0 0 0
252 270 261
198 186
0.206 0.198 0.197 0 201 0 197
397
0.279 0.272 0 268 0 260
reduced by inserting tubiiig of larger dianicter bore from the furnace section to the diffusion pump, making this connection as short as possible, and installing a high-speed diffusion pump. The outgassing of samples to a furnace blank takes 30 minutes or less by the platinum bath method. Results on standard samples give good rcproducibility throughout 50 sample
080
056 060 053 062
053 051 059 078 052
0.051
hverage Standard dev. 0.005
Table
056
063 051
0 0 0 0 0
055
058 052 058 0013
runs. Larger runs arc' impractical with the prescnt crucible design because of the carbon consumption from the crucible ivalls. LITERATURE CITED
(1) Bumps, E. 6 . )Kessler, H. D., Hanfien, AI,, T r a n s . Am. SOC.f o r Metals Preprint 32 (1952). ( 2 ) Derge, G., .I. Metals 1, 31 (1949), (3) Hansen, I\-. R., 1Iallett, 11. W.,
Battelle Memorial Institute, Columbus, Ohio, llemoranduni on Determination of Oxygen in Titanium (Feb. 25, 1957). (4) hladley, D. G., Stricl;lariti-ConFtnhle, R . F., Analyst 78, KO.928, 122-6 (1953). (5) Peifer, W. h., -1llegheny Ludlum Steel Corp., Brackenridge, Pa., private communication. (6) Walter, D. I., ANAL. CHEII. 2 2 , 297 (1950). (7) Wilkins, D . H., Fleischer, J. F., A n d . Chini. Acta 15, 334 (1956). RECEIYEDfor revien- August 14, 1937. Accepted November 23, 1957.
Quinoxaline-2,3-dithiol as a Colorimetric Reagent Determination of Nickel in Ammoniacal Solutions D. A. SKOOG, MING-GON LAI, and ARTHUR FURST Department of Chemisfry and Chemical Engineering, Stanford University, Sfanford, Calif. The applicability of quinoxaline-2,3dithiol to the colorimetric determination of nickel was studied. In ammoniacal solution the reagent forms a pink complex with that cation which is suitable for the determination of nickel in the concentration range of 0.03 to 3 p.p.m. The effect of such variables as ammonia concentration, reagent concentration, a g e of the reagent solution, and stability of the color was studied. The best
conditions allow the determination of nickel with a relative accuracy to 1%. Silver, copper, cobalt, and manganese interfere with the determination of nickel in the recommended procedure; however, many of the other common cations do not.
most satisfactory method for T the determination of small quantities of nickel appears to be coloriHE
a
metric one based on the red or brown color produced by addition of dimethylglyosime to an ammoniacal solution of nickel after treatment with an oxidizing agent such as bromine ( 3 ) . This procedure has been investigated by Mitchell and 11ellon ( I ) , who found it to be suitable for concentrations ranging from 0.1 to 5 p.p.m. of nickel when 1-em. cells are used. d number of the heavy VOL. 30,
NO. 3, MARCH 1958
365