pentaborane, for example tetraethylpentaborane, were steam volatile and remained unchanged. None of the compounds examined resisted oxidation by Van Slyke combustion fluid, but the method is limited to the analysis of compounds with a low vapor pressure a t room temperature. A more complicated apparatus would be required t o prevent loss of volatile material during preliminary evacuation of the system.
ACKNOWLEDGMENT
The authors are indebted to R. L. Williams and N. J. Blay for assistance
in obtaming samples of organoboron compounds. LITERATURE CITED
(1) Arthur, P., Annino, R., Donahoo, W. P., Ibid., 28, 1852 (1957). (2) Atteberry, R. W., J . Phgs. Chem. 62, 1458 (1958); Fauth, M. I., McNerney, C. F., ANAL.CHEM. 32, 91 (1960); Griffiths. J. V.. Williams. R. L.. Chem. & Ind. ( b n d & ) 1959, 655. (3) Blay, N. J., Dunstan, I., Williams, R . L., J . Chem. SOC. 1960, 430, 5006, 5012, 5016; Blay, N. J., Williams, J., Williams, R. L., Ibid., p. 424. (4) Bush, G. H., Higgs, D. G., Analyst 76, 684 (1951). (5) DeFord, D. D., Lucchesi, C. A., Thoburn, J. M., “Analytical Chemistry of Boron,” Northwestern University,
Evaneton, Ill.. June, 1953, Gchechter, W.H., Jackson, C. B., Adams, R. hl., “Boron Hydrides and Related Compounds,” 2nd ed., Callery Chemical Co., Callery, Pa., May 1954. (6) Etherington, T. L., McCarty, L. V., Arch. industr. Hyg. 5 , 447 (1952); Hill, D. L., Gipson, E. I., Heacock, J. F., ANAL. CHEM. 28, 133 (1956); Hill, W. H., Johnston, M. S.,Ibzd., 27 , 1300 (1955); Kuhns, L. J., Forsyth, R. H., Masi, J. F., Ibid., 28, 1750 (1956). (7) McReady, R. M., Hassid, W. Z., ISD. EKG CHEM., ANAL. ED. 14, 525 (1942). (8) Schaeffer, R., J . .4m. Chem. SOC.79, 1006 (1957). (9) Strahm, R. D., Hawthorne, M. F., ANAL.CHEM.32, 530 (1960). RECEIVED for review March 16, 1961. Accepted June 22, 1961.
Phase Separation and Analysis of Sintered Titanium Carbide-Nickel Cermets Using Alcoholic Bromine EDWARD J. VIOLANTE Scientific Laboratory, Ford Motor
Co., Dearborn, Mich.
b The physical characteristics of sintered titanium carbide-nickel cermets can be correlated to the distribution of titanium between the carbide and metallic phases. In the phase analysis of this material it is necessary to prevent any dissolution of the carbide phase which would falsely indicate a higher percentage of titdnium in the metallic binder. The method which has been developed uses alcoholic bromine at -20” C. which readily dissolves the metallic phase, but is essentially without effect on the carbide. A complete separation can be made in less than 8 hours with equipment which is common in most laboratories.
I
a study of the properties of fjtanium carbide - based cermets cemented with nickel, the problem arose of determining quantitatively the composition of the metallic phase in a n effort to explain the wide variations in Curie point and electrical resistivity. These cermets were assumed to be composed of titanium carbide grains dispersed in a metallic nickel matrix, and, therefore, the Curie point of these cermets should have been the same as that of pure nickel. N
1600
ANALYTICAL CHEMISTRY
The assumption is often made that the final composition of each of the phases in two-phase materials, such as cemented carbides, changes only slightly with high temperature sintering. This is not true of the system which was studied. It was determined that during the high temperature sintering the nickel dissolves a small amount of titanium carbide. Upon cooling, titanium and carbon atoms (nonstoichiometric titanium carbide) are rejected by the nickel, dependent upon the solubility limit of the atoms in the nickel. Carbon being only slightly soluble a t room temperature, is almost completely rejected. However, titanium being soluble over a wide range of compositions forms a solid solution with the nickel binder. The properties of the binder then are no longer those of pure nickel, but rather those of a nickel-titanium alloy. The problem was to determine the amount of titanium present in the metallic nickel binder To have an accurate phase analysis it is essential that nonmetallics be practically unattacked during the separation. If there is any dissolution of the nonmetallic phase, the results of a chemical analysis of the phases would be in error.
A chemical method has been developed using alcoholic bromine at a low temperature to dissolve the metallic binder, without affecting the titanium carbide present. This method also is rapid and requires a minimum of elaborate equipment. METHODS OF SEPARATION
The most popular method of effecting a phase separation has been the classical electrochemical method. where the specimen is made the anode in a weak hydrochloric acid electrolyte. Thus, by anodic oxidation the metallic phase is separated from the nonmetallics which settle to the bottom of the electrolyte. This conventional method is not only time consuming, but also is unreliable in the large number of cases wherein the nonmetallic phase is appreciably soluble in the electrolyte. Solvent extraction methods have been gaining in favor for the separation of nonmetallics, such as oxides and carbides, from metallic matrices. These methods generally employ a solution of a halogen dissolved in an organic solvent which will dissolve the metallic phase, but x-hich will not appreciably dissolve the nonmetallic phase. For the estimation of oude inclusions in steel,
20
5
30;
z a
5 I-
s
1.0
* -40
Figure 1.
-
1
-30
-
1
-PO -IO 0 TEMPERATURE
Titanium found in
5%
+IO
+20
5
0
bromine-methanol
After 6-hour treatment of 1 '/2-micron titanium carbide
IO
TIME
('C 1
Figure 2.
5
23
(HOURS1
Titanium found in 5% bromine-methanol and per cent carbide recovered After treatment at + 1 8 " C.
Rooney and Stapleton (4) used a solution of iodine in anhydrous methanol to dissolve the matrix and then recovered the oxides by filtration. Vernon, Wormwell, and Nurse (6), in a study of oxide film thickness on iron, also used iodine in methanol to attack the oxide-metal interface, thereby freeing the oxide film from the base metal. The success of this method depended on the exclusion of all traces of moisture from the apparatus and the reagents with the result that the equipment and procedure became complex. Mahla and Nielsen (1-3) used a solution of bromine in anhydrous methanol to remove oxide films from stainless steel for electron microscope studies, and also to isolate chromium carbides in stainless steel. They stated that when stripping the oxide film from stainless steel, absolute moisture elimination was of no consequence as it is when iodine is used. However, when isolating the carbides in stainless steel, anhydrous methanol was used with the implication that all traces of moisture must be eliminated from the apparatus and reagents. This, again, requires complex equipment and procedures to ensure absolute moisture exclusion. EXPERIMENTAL
The experimental work in developing this method was undertaken in three parts: It was necessary to determine the conditions under which titanium carbide showed a negligible rate of solution in alcoholic bromine; because there were no standards available against which to check the proposed method, it was necessary to devise an independent means of determining the amount and composition of the phases; a phase separation was effected by the proposec nietiior, and the results compared with the inaermdently determined compos:tion
To determine the conditions under which titanium carbide showed a negligible rate of solution in alcoholic bromine, a number of 0.5-gram samples of titanium carbide powder were treated with 100 ml. of a 5% solution of bromine in reagent grade absolute methanol. The titanium carbide used was a commercial grade material with an average grain size of 11/2 microns. The reaction vessel was a three-necked 250-ml. flask fitted with a reflux condenser, a constant speed stirrer operating a t 400 r.p.m., and an inlet tube for dry nitrogen which supplied an inert atmosphere. The inert atmosphere was necessary to prevent acid hydrolysis due to atmospheric moisture. The samples were allowed to react for 6 hours a t temperatures ranging from +18" to -40" C. The temperatures above 0" C. were controlled in a constant temperature water bath, while the temperatures below 0" C. were maintained by dry ice additions to methanol in a Dewar flask. After the 'elapsed time, each sample was filtered through a fine fritted-glass crucible and the residue washed with cold absolute methanol. The filtrate was evaporated to dryness and cooled. To oxidize any organics which formed, 20 ml. of 1: 1 sulfuric acid and 5 ml. of concentrated nitric acid were added. The solution was then evaporated to fumes and additional concentrated nitric acid added dropwise to ensure complete oxidation of any remaining organics. The beaker was cooled, and 50 ml. of water added to dissolve the salts. The solution was then made up to 100 ml. in a volumetric flask, and the per cent titanium was determined colorimetrically using hydrogen peroxide. At temperatures below -20" C., the solubility of the titanium carbide is negligible, but increases abruptly a t the hgher temperatures, as shown in Yigurt 1 . Tht amount of titamum carbide which is dissolved in 6 hours a t temperatures helow -20" C. is equivalent t n onh 0 Oi h% titarnun:
During the high temperature sintering of a titanium carbide-nickel cermet, there is a considerable amount of carbide grain growth wherein the initial ll/r micron titanium carbide particles grow to approximately 10 microns. Considering the effect of particle size on the rate of solution, it is safe to assume that the amount of titanium carbide dissolved from a sintered sample would be considerably less than that exhibited by the l*/rmicron material used for these data. The portion of the curve below -20" C. was redetermined in the same manner as before, but without an inert atmosphere, by allowing the titanium carbide powder to react with the bromine solution in a 250-ml. Erlenmeyer flask which was left open to the atmosphere. The results were identical to those obtained with the dry inert atmosphere.
It was next necessary to determine the amount and composition of the phases present in a series of sintered titanium carbide cermets cemented with varied amounts of nickel to serve as a check for the proposed method. This was accomplished by treating a 0.5gram sample of the crushed cermet in 100 ml. of the 5y0 alcoholic bromine solution a t +18" C. for times varying from 5 to 20 hours. As before, an inert nitrogen atmosphere was supplied and also constant speed stirring. To standardize the effect of the surface area on solubility, only the crushed material which passed through a 40-mesh screen and was retained on a 60-mesh screen was used. After treating the samples for the required time, they were filtered through weighed fritted-glass crucibles and the residue washed with cold absolute methanol. The crucibles were dried a t 110" C. for 15 minutes, cooled, and reweighed. The per cent residue (or carbide) was then calculated. The filtratt. was evaporated and treated as VOL, 33, NO. 11, OCTOBER 1961
1601
before, and the per cent titanium determined colorimetrically. The analytical data are plotted in Figure 2, and show the increase in titanium in the bromine solution with time a t f18" C. and the corresponding decrease in the amount of carbide recovered. During approximately the first 4 hours a t +18" C., the bromine solution dissolves the metallic binder and also partially dissolves the titanium carbide which is being exposed. The per cent titanium which is found in the bromine solution is then the titanium present in the nickel binder plus the titanium due to the dissolved titanium carbide. By extrapolating the data to the zero time axis, which in effect is subtracting the amount of titanium due to the dissolved titanium carbide, a value is obtained for the per cent titanium present in the nickel binder. The per cent carbide in the cermet is also estimated by extrapolation. Finally, a phase separation of the sintered cermets was effected by the proposed method as follows: 100 ml. of the 5% solution of bromine in reagent grade absolute methanol, in a 250-nd. Erlenmeyer flask, is cooled to -20" C. using alcohol and dry ice as the cooling medium. A 0.5-gram sample of the crushed cermet is added and allowed to react for 6 hours without an inert atmosphere and without constant speed stirring. Agitation is accomplished by stirring every half hour with a thermometer. The temperature is maintained a t -20" C. by occasional dry ice additions to the cooling bath. The solution is then filtered through a weighed fritted-glass crucible and washed with cold absolute methanol.
Table I.
Comparison of Results
ExtraDolated Resdts, % Ti Carbide
Ni,
5%
10 15 20 25
0.20 0.35 0.42
0.57
89.3 84.1 78.5 73.9
-2OO
c.
Method Results, % Ti Carbide 0.21 0.31 0.43 0.60
89.3 84.2 78.7 73.8
The crucible is dried a t 110" C. for 15 minutes, cooled, and reweighed, and the per cent carbide calculated. The filtrate is evaporated and treated as before and the per cent titanium determined. These results are tabulated in Table I and compare favorably with the results which were extrapolated from Figure 2. A material balance, made by summing the per cent carbide and the per cent metallic binder, approximated 100% only when the amount of titanium in the nickel was considered.
can be accomplished by performing the separation a t -20" C. which eliminates the need for anhydrous reagents and atmospheres. From the standpoint of technique, the method described in this paper eliminates the necessity for preparing and transferring anhydrous reagents, which is always tedious, and eliminates the uncertainty that the last traces of moisture have been removed from the apparatus. Maintaining the temperature a t -20" C. presents no problem. I t has been the practice in this laboratory to cool the solution to approximately -25" C. and allow it to react with the sample until the temperature approaches -20" C., and then to cool again to approximately -25" C. If the cooling bath is contained in a Dewar flask, cooling is required only about every hour. Because the separation is performed a t a low temperature, the rate of solution of the metallic phase is decreased somewhat. However, this is more than compensated for by the time saved in not having to prepare special reagents.
DISCUSSION
When alcoholic bromine is used to separate titanium carbide for a phase analysis of titanium carbide-nickel cermets, it is absolutely necessary to prevent the hydrolysis of the solution. If the solution hydrolyzes, the acid products which are formed will react with the carbide with the result that the phase analysis will be in error. Preventing the hydrolysis by using moisture-free reagents and equipment requires extreme care in preparing the reagents and equipment prior to performing the actual separation. This
LITERATURE CITED
(1) Mahla, E. M., Nielsen, N. A., J . A p p l . Phy. 19, 378 (1948).
(2) Mahla, E. M., Nielsen, K. A,, J. Electrochem. SOC.93, l ( l 9 4 8 ) . (3) Mahla, E. M., Nielsen, N. .4., Trans. Am. SOC.Metals 43,290 (1951). (4) Rooney, T. E., Stapleton, A . G., J. Iron Steel Inst. (London) 131, 249 (1935). (5) Vernon, W. H. J., Wormwell, F., Nurse, T. J., J. C h m . SOC.Pt. I, 1939, p. 621.
RECEIVED for review March 20, 1961. Accepted July 13, 1961. Eighth Detroit Association of Analytical Chemists (Anachem) Conference, October 1960.
Rapid Flame Photometric Determination of Alkalies in Glasses and Silicates P. BRUCE ADAMS Research and Development Division, Corning Glass Works, Corning, N. Y. A flame photometric method has been developed to permit the rapid determination of the alkali elements in glasses and other silicates. Analyses are accomplished in approximately '/zhour by decomposition of the sample with a hydrofluoric-hydrochloric acid mixture and direct dilution for the flame determination. The procedure can be speeded by the use of a rotating multiple sample holder and a strip-chart recorder. Results are reproducible to approximately f 1 relative %, and they check wet chemical and conventional flame analyses to approximately f t 2 relative
70.
1602
ANALYTICAL CHEMISTRY
A
numerous procedures have been reported for the determination of alkalies in glasses (1, 6, 6) and other silicates (2, S),most of them require 2 to 4 hours. The most timeconsuming step is generally the digestion with hydrofluoric acid and subsequent fuming to remove it. This was the logical step to shorten or eliminate. Most glasses and many silicates can be decomposed for flame spectrophotometric analysis in 5 minutes with a hydrofluoric-hydrochloric acid mixture without externally applied heat. After dilution with water, a conventional manual flame photometer analysis (6) LTHOUGH
can be made or a rotating multiple sample holder and a strip-chart recorder can be utilized. Recent work by Konopicky and Schmidt (4) has illustrated the usefulness of this approach to the analysis of minerals. REAGENTS AND APPARATUS
Reagents. Hydrofluoric-hydrochloric acid was prepared by mixing 3 volumes of 48% hydrofluoric acid with 1 volume of 37y0 hydrochloric acid (both reagent grade). Stock standard solutions of 500 to 1000 p.p.m. were prepared by addition