no one of these anions was found to produce any interference. One hundred milligrams of fluoride, chloride, iodide, monohydrogen phosphate, perchlorate, oxalate, nitrite, nitrate, and sulfate ions were added separately to a standard solution containing 0.210 mg. of boron, buffered at p H 8.75; in no case did the differential peak height differ from that of the pure boron solution of the same concentration. Cations which are known to react with chromotropic acid and its derivatives, such as ferric ion. aluminum ion, and chromium ion, precipitated as the hydrous oxides a t the p H a t ivhich absorbance measurements were made, and therefore, a separation of these substances was required before a boron analysis was attempted. This procedure m-as used in the analysis of several rare earth borides for boron content (2). After the boride had been brought into solution with nitric acid, the p H of the solution was adjusted to about 8.5 with a solution of sodium hydroside, and the rare earth hydrous oxides were then allowed to settle. An
aliquot of the supernatant liquid was taken for boron analysis with Victoria Violet using the Bausch &. Lomb Spectronic 20. Strong oxidizing agents such as permanganate, ceric ion, and dichromate cannot be present, because of their effect upon the dye itself.
tion in the sample beam, and zero the instrument a t 800 mfi. Scan downscale to 450 mp and measure the height of the differential absorbance peak. Compare with a calibration curve previously prepared by running standard solutions of boric acid in exactly the same manner. ACKNOWLEDGMENT
RECOMMENDED PROCEDURE
An aqueous aliquot containing from 0.02 to 0.60 mg. of boron (0.1 to 3.5 mg. of boric acid) should be adjusted to p H 8.5 to 9.0. The volume of the solution a t this point should be 10 to 15 ml. T o this solution add 2 ml. of a 6.5 X 10-3M solution of Victoria Violet and dilute to 25 ml. with the stock bicarbonate-carbonate buffer solution. The p H of the final solution 0.05 units. Prepare should be 8.75 a blank solution a t the same time b y pipetting 2 ml. of the 6.5 X 10-3M solution of Victoria Violet into a 25 ml. volumetric flask and diluting to the mark with the same buffer. After 30 minutes of development time, place the sample in the reference beam of the Beckman DK-1 and the blank solu-
*
The author acknowledges the support of the United States Atomic Energy Commission for this n-ork. LITERATURE CITED
(1) Cogbill, E. C., Yoe, J. H., Anal.
Chim. Acta 12, 455-63 (1955). (2) Eick, H. A., Gilles, P. W., Abstracts, 134th Meeting, ACS, p. lS, September
1958. (3) Grob, R. L., Yoe, J. H., Anal. Chim. Acta 14, 253 (1956). (4) Kuemmel, D. P., Mellon, M. S., ANAL. C H E ~ 29,378-82 I. (1957). (5) Snell, J. D., Snell, C. T., “Colorimetric Methods of Bnalysis,” Vol. 2, pp. 703-4, Van Nostrand, New York, 1949. RECEIVEDfor review October 31, 1958. Accepted January 26, 1959.
Direct Determination of Oxygen and Nitrogen in Titanium and Titanium Alloys Use of Bromine Trifluoride WILLIAM A. DUPRAW’ and HUGH J. O’NEILL Armour Research Foundafion o f lllinois lnsfitufe of Technology, Technology Center, Chicago 7 6,
b This investigation was undertaken to expand the usefulness of the bromine trifluoride technique. Low levels of the gaseous impurities, oxygen and nitrogen, in titanium and titanium alloys can b e determined. The relative deviation for oxygen over the entire range of oxygen values (0.05 to 1.0%) averages 4.8%. The presence of iron or aluminum in the alloy reduces the total nitrogen recovery but does not effect oxygen recovery. The method offers a new approach for determining oxygen and nitrogen levels in other metals which would b e almost impossible to analyze by more common procedures.
T
accurate determination of oxygen, nitrogen, and hydrogen in titanium metal and alloys is essential to the metallurgical evaluation of the metal in terms of physical performance. Loss of ductility, increased hardness, increased notch sensitivity, and emHE
1 Present address, Columbia-National Corp., P. 0. Box 1247, Pensacola, Fla.
1 104 *
ANALYTICAL CHEMISTRY
brittlement are associated Fvith these impurities. Current analytical methods are based on techniques of vacuum fusion (1, 3, 6, 13, 14, 17), chlorination (2, 6), bromination (4), and, more recently, emission spectroscopy (9). The vacuum fusion and bromination techniques h a r e been used more extensively. The use of bromine trifluoride as a reagent was first suggested by Emelbus and Woolf (8, 15), who reported its action on 28 metal oxides. Hoekstra and Katz (IO) studied its application to determination of combined oxygen in many metal osides. The literature does not cite any investigation on the applicability of this reagent to the determination of low amounts of oxygen. Bromine trifluoride liberates all combined oxygen, as molecular gas, from those metal oxides which form volatile fluorides or fluorides TT hich are soluble in bromine trifluoride. The reaction for titanium is quantitative a t approximately 75’ C., and for oxides analogous to titanium dioside it is:
3MOr
Ill.
+ 4BrF3
-.f
3MF4
+ 2Br2 + 302
A similar reaction, resulting in quantitative liberation of molecular nitrogen, has been assumed for metal nitrides. Metals forming reactive oxides are indicated by Emelkus and Woolf (8, 15) and further elucidated by Hoekstra and Katz (10). However, when oxygen and nitrogen are present in residual amounts in titanium, the gases are not in combination as oxides and nitrides. Titanium has a single phase with oxygen and nitrogen dissolved in solid solution and dispersed in the hexagonal lattice of the titanium metal. This single phase exists up to approximately 14 weight % oxygen and 3 weight yonitrogen (16). Titanium reacts with bromine trifluoride as follows: 3Ti 4BrF3 43TiF4 2Br2 It was assumed that this reaction could provide the basis of a method for determining residual levels of oxygen and nitrogen in titanium metal and alloys. The so-called interstitial gases would be released as molecular oxygen and nitro-
+
+
gen, hydrogen would react t o form the fluoride, and, b y appropriate freezing techniques to remove the halide reaction products, the total volume of oxygen and nitrogen could be measured.
Table 1.
Standard Deviation between Labs 0 013 0 014 0 032 0 017 0 017 0 017
Average Olygen," yo 0 132 0 131 0 340 0 035
Alloying Metals, 70 4 AI, 4 hfn (RC-130 B) ITA-10 Unalloyed \VA-12 2 7 Cr, 1 . 3 Fe (Ti-150 A ) II-A-49 0 023 C, 0.01 \V \TA-50 0 166 0 023 C, 0 003 W \VAA-51 0 280 0 027 C, 0 01 IT ITA-52 0 557 0 041 0 027 C, 0 01 W IV.1-53 1 11 0 105 0 013 C, 0 01 IT ITA-58 0 lob 8 >In (RC-130 A ) a Values for oxygen reported by Thsk Force on O q g e n by Vacuum Fusion Analysis (number of cooperating laboratories averaged about 12). * Nominal value. Sample ITA-9
APPARATUS
The apparatus (Figure 1) is essentially t h a t described by Hoekstra and Katz (IO). It consists of two major portions: nickel for tlie reaction and retention of the corrosive halogen products and glass for collecting and measuring evolved gases. The reaction system is connected to the glass-measuring system with inch Teflon tubing. These Teflon seals were satisfactory and required no attention. The glass portion of the apparatus contained, in addition to the measuring manometers and Toepler pump, a copper furnace used to remove oxygen from the total gas sample. The copper furnace tube was prepared by filling a 10-cm. borosilicate glass tube with copper turnings. The furnace tube is covered by an insulated heating coil connected to a variable rheostat (Variac). For the quantitative removal of oxygen, the tube is heated to 400" C. The metal reaction and glass-nieasurin8 systems are connected a t both cuds and in the center to a high-vacuum manifold which contains a LIcLeod gage and a diffusion pump. The entire apparatus was pumped down by a Duoseal high-vacuum Welch pump,
tributed to nienibers of tlic Task Force on Oxygen by Vacuum Fusion Analysis (Table I). PROCEDURE
Bromine trifluoride was freed from its lighter components immediately before use. The volatile impurities n ere distilled from the reaction flask a t room temperature into the adjacent Fluorothene trap cooled t o liquid nitrogen temperature. Approximately 5 nil. of reagent is used for each determination. To measure the evolved gases, tlie reaction tube is cooled to liquid nitrogen temperature and the evolved gases are pumped through the cooled Fluorothene traps, by the Toepler pump, into the calibrated receiver until a constant rolume is maintained. When the readings of the total gas volume are completed, the mercury in the calibrated manometer is lowered and the gas passes through the copper furnace. The gas is then cycled for approximately 10 minutes. The mercury is again introduced into the calibrated manometer to trap the cycling gas until a constant volume is obtained.
REAGENT AND SAMPLES
The bromine trifluoride (HarshaIY Chemical Co.) !vas 90 to 95% pure, the major impurities being bromine and bromine pentafluoride. Other possible impurities are bromine monofluoride, hydrogen fluoride, and nonvolatile fluorides. The titanium alloys were furnished by the Watertown Arsenal anti dis-
I
Standard Titanium Samples
-
C
R
Y
Np
Table II. Average Oxygen Content of Titanium-Base Alloys Determined by Bromine Trifluoride Method
-1verage Sample Oxygen, DeviaNo. % tion K.1-9 0 112 0 008 II-A-10 0 161 0 011 W.1-12 0 324 0 011 W'A-40 0 051 0 003 IvAA-50 n
wa-5i o
Maximum D!viaS o . of tion Analyses 0 015 0
0 022 0 017 0 007 174 0 008 0 015
307
o
010
o
019
\7-.1-52 0 608 0 012 0 019 \VA1-53 0 985 0 047 0 100 ITA-58 0 145 0.008 0.015
-
G 4 ci 4 5 4
The mercury is raised into the calibrated arm a t the top of the Toepler pump, and a second set of readings is obtained. Thc volume of nitrogen is determined from these second readings and by subtracting this value from the total gas volume, the amount of oxygen is calculated. Imniediately after the libcratcd gases are puniped from the reaction tube to the Calibrated cell the reaction tube is m-armcd to room temperature. The volatile reaction products are then distilled into the adjacent trap and a second sample is introduced after recooling the bromine trifluoride. This method is used for three consecutive samples. Care is taken not to mix alloys, lest certain nonvolatile fluorides remain in the reaction tube and offer some surface inhibition. This method permits the entry of a second saniple into the reaction tube while the previous sample is being analyzed in the glass portion of the apparatus. RESULTS
1
u Figure 1 .
Diagram of apparatus
NICKEL ?EA&TION
The values for oxygen and nitrogen obtained by the bromine trifluoride method are presented in Tables I1 and 111. All values obtained for each saniplc are reported, except for a fen- dctcrminations n hich u-ere completed even though contamination seemed evident. The normal pumping time for the systcm VOL. 31, NO. 6, JUNE 1959
1105
when in constant operation is approximately 20 minutes. When contamination (water vapor, etc.) is present, this pumping time is prolonged-sometimes as many as 2 hours. When this occurs, extremely high results are obtained, no matter how long the pumping time. These results were omitted. The ovygen values determined by the bromine trifluoride system agree reasonably well with those obtained by the vacuum fusion technique. The maximum deviations associated with both systems cover the entire range of oxygen values and fall within the same limits. The average deviations of both techniques also agree. The relative deviation for oxygen over the entire range of oxygen levels by the bromine trifluoride system averages 4.8y0, although the range includes values from 0.05 to 1.0% oxygen. The large deviation associated with Sample K.4-53 can be explained as follows: By increasing the ratio of gas volume to sample size many variables appear more significant than in smaller ratio samples. For instance, it is difficult to cut the material without excessive heating. Also, in preparing the surface of the metal, a small sample is almost impossible to polish with the Table 111.
same precision as a larger one. Homogeneity also becomes more important with decreasing size. The nitrogen values determined by the bromine trifluoride method also agree reasonably well with the Kjeldahl nitrogen values, except for samples WA-9, -10, and -12. Sample WA-10, which has the largest deviation in both the oxygen (except WA-53) and nitrogen values, is the unalloyed titanium. The high value obtained cannot be explained. The nitrogen values obtained for samples WA-9 and WA-12 deserve special mention. I n a previous investigation (7) two similar types of alloys a t various nitrogen levels were analyzed. The results from the previous investigations, with the results of the above samples, are listed in Table IV. I n both investigations the nianganesealuminum alloy exhibited nitrogen retention. Also, the fact that sample WA-58 (8% manganese) did not show this effect indicates that the aluminum present may be responsible for the low recovery. Similarly, the nitrogen retention of the iron-vanadium and iron-chromium alloys might be attributed to the iron. This holdup of
Average Nitrogen Content of Titanium-Base Alloys
Kieldahl Method, Nitrogen, c/c Armour Watertown Research Arsenal Foundation 0 025 0.024 0 010 0.010 0.044 0 042 0 011 0 014 0 OOi 0 009 0 034 0 038 0 035 0 031 0 030 0 022 0 030 ~
Sample No. WA-9
W.4-10
Wil-12
WA-49
K.4-50
Wh-51 WA-52 WA-53 W.4-58
Bromine Trifluoride Method Xitrogen, Maximum S o . of % Av. dev. dev. analyses 0.006 0.001 0.002 6 0.037 0,009 0.020 7 0 016 0.001 0.003 4 0 008 0 001 0 003 5 0 003 0 001 0 003 4 0 045 0 005 0 017 i 0 028 0 005 0 009 4 0 020 0 002 0 004 5 0 027 0 008 0 017 4
Table IV.
Nitrogen Values Obtained with Different Types of Alloys
Type of Alloy 4bIn 4A1 (IT--4-9) 514n 5-41 1.3Fe 2.7Fe (ITA-12)
Total Nitrogen, % Bromine Kjeldahl trifluoride method method 0.025 0.006 0.073 0 026 0.103 0 037 0 195 0.124 0 043 0 016
5Fe 51'
Table V.
0.076
0.135 0.207
0.046 0.129
36
34 62
Oxygen and Nitrogen Values Obtained for Titanium Alloy (10 Mo)
Bromine Trifluoride Method Oxygen, % Nitrogen, % 0,030
0.038
0.031
Av. 0.033 1106
0 027
Recovery, 7 24 36 36 64 3i
0,019
0.022 0.024 0.022
ANALYTICAL CHEMISTRY
Vacuum Fusion Method Oxygen, % Nitrogen, % 0.099 0.021
Kieldahl Method Nitrogen, % 0.017
nitrogen is difficult to explain. If the nitrogen were present in these alloys as the nitride of aluminum or iron, it would be reasonable t o obtain low nitrogen values because both are unreactive in this system. However, the presence of these nitrides is unlikely because of the more readily formed titanium nitride. Therefore, it must be assumed that the aluminum and iron form a reaction product capable of removing nitrogen from the reaction medium. Nevertheless, consistently low nitrogen values were found for these samples, and nitrogen recovery is almost proportional to total nitrogen content. One way to explain this would be to identify the reaction products. During this investigation a suitable procedure for the hydrolysis of bromine trifluoride was not known, and its extreme reactivity towards water and organic solvents discouraged this approach. Hon ever, a hydrolysis procedure (reaction with a ferrous sulfate solution in a polyethylene jar) (II), is now available and future work will certainly include a reaction product identification. The technique described by Sheft, Martin, and Kat2 (22) might also aid in the elucidation of the mechanism involved in the nitrogen retention of certain alloys. Table V illustrates that certain metals are capable of removing gases from the reaction medium. The oxygen and nitrogen values obtained for a titanium alloy (10 110) are given. This alloy was prepared by the hletals Research Department of Armour Research Foundation. K'itrogcn values by vacuum fusion are listed in Table V because they agree well with the Kjeldah1 values. HoLvever, in some instances the nitrogen values by vacuum fusion arc unreliable. The lo^ oxygen recovery for sample 10 110 illustrates the formation of a stable intermediate. In this case the stable intermediate is the oxyfluoride (PIIoOF4) (IO), and the oxygen is retained in the reaction medium. A similar mechanism may possibly explain the low nitrogen values in the iron and alruninum alloys. CONCLUSIONS
The data in Tables I1 and I11 represent a reasonable complete surrey of the bromine trifluoride method with titanium and titanium alloys. Oxygen values obtained by this technique agree within the experimental range of the vacuum fusion method. The nitrogen values, in most cases, compare with those obtained by the Kjeldahl method. Hon-ever, the presence of aluminum or iron in the titanium alloy tends to reduce the total nitrogen recovery. The overall results indicate that the bromine
trifluoride method is comparable to the vacuum fusion method for oxygen analysis of titanium and titanium alloys, and can also be used t o determine the nitrogen content. Several features of this system offer excellent possibilities for further study. The use of a zero blank eliminat>es any error due to varying blank values during the course of an analysis. Also, the apparatus can be modified for any range of gas values by changing the calibrated receiver. The bromine trifluoride system also offers possibilities for the determination of gases in metals 11hich is difficult by other methods. For example, the sulfur group, sulfur, selenium, and tellurium, cannot be tolerated in the vacuum fusion procedure, because these elements poison the cupric oxide-ceric oxide catalyst in this apparatus. Also, a limited number of silicon samples have bcrn analyzed by this method ivith good agreement with vacuum fusion results. This system could be useful in the analysis of lead and uranium-bismuth alloys. Hoekstra and Katz ( I O ) have suggested possible modifications to increase the number of compounds which can be analyzed. Further work u-ith othw metals and alloy systems should be important to any analytical laboratory involved in metals research and evaluation. I n future work performed by this
technique two important modifications are possible. Other metals such as Monel or copper should be considered t o replace the all-nickel reaction system. These metals are more economical to use and are much more easily tooled. The measuring system would be more accurate if the calibrated manometer (Figure 1) were replaced by a Kesslor 3-stage circulating pump with a modified McLeod gage, such as the one on the Sational Research Corp.'s vacuum fusion apparatus. The circulatory pump with a Toepler pump would be more efficient in quantitatively transferring the evolved gases, and the use of the McLeod gage for volume and pressure measurements would reduce the reading error by one half. ACKNOWLEDGMENT
The authors are indebted to the Waterton-n Arsenal and the Chicago Ordnance District for their sponsorship of this investigation, and to Samuel Vigo of the Watertown Arsenal for encouragement and guidance. They thank Harold Combs and Myron Hillmer of the Armour Research Foundation for the Kjeldahl nitrogen and vacuum fusion data, and H. R. Hoekstra and J. J. Katz of the Argonnr Sational Laboratory for their generous assistance throughout.
LITERATURE CITED
(1) Albrccht, W. M., Mallett, 31. \i7., ANAL.CHEM.26,401 (1954). (2) Aldrich, J. J., Chase Brass & Copper Co., Inc., Xaterbury 20, Conn., personal
communication. (3) Bennett, S. J., Covington, L. C., *%NSL. CHEV. 30,363 (1958). (4) Codell, Maurice, Norwitz, George, I b d , 27, 1083 (1955). (5) Corbett, J. A., ilnalyst 76, 652-7 (1951). (6) Derge, Gerhard, J . Metals 1,31(1949). (7) Dupraw, W. A., O'Neill, H. . J., Symposium on Analysis of Titanium and Titanium Alloys, 128th Meeting, ACS, llinneapolis, hlinn., 1955. (8) Emelbus, H. J., Woolf, A. A., J . Chem. SOC.1950, 164. (9) Fassel, V. A,, Gordon, W. A , A x . 4 ~ . CHEM.30, 179 (1958). (10) Hoekstra. H. R.. Katz. J. J., Ibid.. ' 25, 1608 (1953). ' (11) Schnizlein, J. G., Argonne Kational Laboratories, Lemont, Ill., personal communication. (12) Sheft, Irving, Martin, A. F., Iiatz, J. J . . J .