of our method, and our realistic limiting sensitivity is probably not much better than 5 X lO-*A' iodine. The break in the titration curve of Figure 1 is considerably smaller than we would calculate from the redox POtentials of the As(III)-As(V) and I-IB- couples. Apparently the system prior to the end point does not establish an potentia' at the platinum electrode.
LITERATURE CITED
( 1 ) Bard, A. J., Lingane, J. J., Anal. Chzm. Acta 20, 463 (1959). ( 2 ) Grover, E. W., Reilley, C. N., ANAL. 26, 1750 (1954). (3) Knowles, G., Lowden, G. F., Analyst 78* 159 (1953). (4) Kolthoff, I. M., Belcher, R., "Volumetric ~ ~ ~ yol. l ~ 111, ~ pp, i 205-210, ~ , " Interscience, New York, 1957. (5) LiebhafskY, A., J . PhYs. C'hem. 35, 1648 (1931). ( 6 ) hlalmstadt, H. V., Pardue, H. L., ANAL.CHEM.32, 1034 (1960).
cHEM,
( 7 ) Megregian, S., U . S. Public. Health Reports 63, 137 (1948). ( 8 ) Roebuck, J. R., J . Phys. Chem. 6 , 365 (1902). ( 9 ) Washburn, E. W., J . Am. Chem. SOC. 30, 32 (1908).
FRANCI L. ANDERSON EVANH. APPELMAN Argonne National Laboratory Argonne, Ill. WORK erformed under the auspices of the 8 s .Atomic Energy Commission.
Carbon Determination in Hyper-Pure Elemental Boron Utilizing Gas Chromatography SIR: I n an earlier work ( I ) , we developed a method for oxidizing a boron sample and determining the carbon content by conversion to carbon dioxide. The procedure consisted of firing a powdered boron sample in an oxygen atmosphere with tin accelerator under carefully controlled conditions in a high frequency induction field. At that time the electroconductometric method was chosen for the carbon dioxide detection, and a sensitivity of approximately 20 p.p.m. carbon was suggested. I n subsequent papers, we were able to show a maximum detector sensitivity of approximately 5 p.p.m, carbon under ideal parameter conditions using gas chromatography (2) and combustion-gas chromatography ( 3 ) . I n the work presented here, the gas chromatographic method was applied to carbon analysis in elemental boron and the lower limit of sensitivity imposed by the electroconductometric detection system has been extended below the suggested 20 p.p.m. carbon. This method employs a programmed temperature gas chromatograph utilizing a molecular sieve column to trap the evolved carbon dioxide, a thermal conductivity detector, and a recorder equipped with a Disc chart integrator. EXPERIMENTAL
Preparation of Standard Curve. A standard curve was obtained by using different weights of NBS 170a, 55e, a n d 16d steel a n d observing t h e number of counts recorded for known weights of carbon. Linear results were obtained over the range of 700 p.p.m. down to less t h a n 20 p.p.m. T h e curve passed slightly above the origin when extrapolated t o zero carbon content. The number of counts above the origin corresponded to what was obtained by firing a Leco tin blank and was of the order of 5 p.p.m. This is the basis for the 5 p.p.m. sensitivity suggested above. At the 1% carbon level, the change in the slope of the
standard curve was less than 5%. The standard deviation in the range of 104 p.p.m. to 600 p.p.m. was +1.5oJ, or less. Procedure. T h e helium two-stage regulator was turned on a n d a helium flow rate was established. T h e column furnace a n d fan were turned on and the column temperature was set at 100" C. T h e bridge current was turned on and set at 130 amperes and t h e system was allowed t o a t t a i n equilibrium overnight. T h e helium flow rate was set at 150 ml./minute (12 p.s.i.), while the oxygen regulator was set a 11.5 p.s.i. which corresponded to a n oxygen flow rate of 100 m1Jminute. The prefire valve was opened and the crucible loaded with 0.600 + 0.0009 gram of 170a steel and 1.00 f 0.001 gram of Leco tin accelerator. The combustion tube was purged, the furnace was turned on, and the stop watch was started. The furnace was turned off after 15 minutes and the crucible was removed and loaded with sample after 20 minutes. While the combustion tube was purged for 1 minute, the attenuator was set on X512 and the O2 valve was opened and the helium valve was closed simultaneously. With 0 2 passing through the column, the sample was combusted for 7 minutes and swept for 1 minute. The helium valve was opened and the oxygen valve was closed simultaneously. After 2 minutes the attenuation was set properly (depending on C content and size of sample) , the temperature program started at 42" C./minute and the chart drive started. The C 0 2 was eluted a t 240-60' C. but the column temperature was allowed to reach 500" C., after CO, elution to maintain constant column parameters. The chart drive and oven were turned off and the oven door was removed. The oven was allowed to cool and stabilize a t 100" C. T o shorten analysis time per run, a crucible may be prefired while the C 0 2 from the previous samples is being eluted. The periods of prefiring, sample combustion, and column oven cooling afforded ample time for weighing of materials and recording of results. The time of analysis per run, including the prefiring of a crucible was less than 30 minutes.
The value of various parameters for each analysis were: Helium flow rate at the sample side: 150 ml./minute at 100' C. column temperature. Helium flow rate at reference side: 50 ml./minute a t room temperature. Oxygen flow rate at the sample side: 100 ml./minute at 100" C. column temperature. Program temperature rate: 42" C. per minute. Block temperature: 200' C. Bridge setting: 130 ma. Temperature limit setting: 500" C. Helium pressure: 12 p.s.i. Oxygen pressure: 11.5 p.s.i. Size of boron samples: 100 mesh. Weight of boron samples: 0.250 gram. Tin/boron ratio: 4/1. RESULTS AND DISCUSSION
The results of the boron metal analysis are shown in Table I. It was extremely important that parameters such as sample size, sample weight, tin/boron ratio, etc., be carefully controlled. Optimal conditions as found by Kuo, Bender, and Walker (1) gave maximum precision and accuracy in this work. I t was verified for the range of 9 to 16 p.s.i. that oxygen pressures have no effect on the combustion of boron samples. The oxygen pressure was carefully controlled, however, to maintain a constant flow rate. The length of the column did not effect the peak width so long as the flow rates remained constant. C 0 2 peaks obtained from 8-inch and 12-inch
Table I. Boron Metal Analysis Gas Chromatographic Carbon Sample No. rune Carbon, p.p.m, M 6312AN 4 420 f 8 M 6312 BG 4 1032 f 21 M 6401 A 0 4 448 i 27 M 6401 AR 4 760 f 34 M 6405 CP 5 644 f 24 M 6405 CJ 3 364 f 6 M 6405 CII 5 2416 i 87 .M 6312 BF 5 645 i 25 3 245 i 29 M 6404 A P M 6406 AQ 5 22,900 i. 400
VOL. 37, NO. 2, FEBRUARY 1965
299
columns were nearly identical. Column packing played an important role in the determination of other parameters as more pressure was required for a given flow rate on the 8-inch column than on the 12-inch. I n most cases, the COz was eluted in the range of 240' to 260' C. The elution temperatures of the 12inch column were slightly higher. The boron samples analyzed had carbon contents ranging from 245 to 22,900 p.p.m. At maximum sensitivity, 5 p.p.m. carbon gave an integration of 120 counts. A major problem in achieving desired precision and accuracy was the difficulty in attaining complete combustion of the boron samples. This seemed to be
especially true of the samples with higher carbon content. The precision was improved by mechanically mixing the boron and tin with a stirring rod. This gave better combustion than did mixing by agitation. The nonuniformity of sampling of the available boron also affected the precision. Even with seemingly complete combustion, the precision attained was not equal to that attained using NBS steel. Prefiring the crucibles in a bomb (under constant oxygen pressure, but with system closed off), instead of a steady stream of oxygen, gave erratic results, probably due to the adsorption of COz on the crucibles. This technique furnishes a rapid,
relatively simple, and precise method of carbon determination in elemental boron. LITERATURE CITED
(1) Kuo, C. W., Bender, G. T., Walker, Joe M., ANAL.CHEM.35, 1505-9 (1963). (2) Stuckey, W. K., Walker, Joe M., Ibid., pp. 2015-17 (1963). (3)Walker, Joe M., Kuo, C. W., Ibid., pp. 2017-19 (1963). JOEM. WALKER JAMES SPIGARELLI
GARYBENDER Department of Chemistry Kansas State College of Pittsburg Pittsburg, Kan. Presented in part 2nd International Conference of Hyper-Pure Elemental Boron, Paris, France, July 17-18, 1964.
Determination of Aluminum Nitride in Steel by Spectrophotometric Analysis for Aluminum SIR: Beeghly has published several papers ( I , 2, 3) on the determination of nonmetallic inclusions in steel. Reference (2) deals specifically with aluminum nitride which he isolates from the steel matrix by refluxing the sample in a solution of bromine in methyl acetate and filtration of the residue containing carbides and insoluble nitrides. The amount of aluminum nitride recovered can then be calculated from the micro-Kjeldahl nitrogen analysis of the halogen-ester insoluble residue. For highly complex ferrous alloys, this procedure was found to be time consuming, requiring 3'/2 to 4 hours for sample solution alone. Furthermore, it was found that other nitrides could be retained in the residue. Therefore, a different solvent for the matrix was sought. A mixture of bromine in methyl alcohol was found to be considerably more reactive and, in most cases tried, required no external source
of heat. Because the reaction itself is exothermic, enough heat is supplied internally to maintain the reaction. Most steels tested were dissolved in about one hour. X-ray diffraction analyses of filtered residues showed that the only form of aluminum present was the nitride. A sample of relatively pure aluminum nitride was purchased from the Kern Chemical Corp., 261 1 Exposition Blvd., Los Angeles 18, Calif., for investigation. Analysis of the as received material showed it to be approximately 80%
Table 11.
Sample size, mg. 2 2 1 1
aluminum nitride. X-ray diffraction analysis revealed the material to be aluminum nitride with a few unidentified reflections, none of which could be traced to aluminum compounds. Wet chemical analyses of the purchased material and portions of bromine-methanol insoluble residues were performed to determine the effectiveness of the separation a t various levels of the nonmetallic. The samples were analyzed for aluminum spectrophotometrically by a modification of the method reported by Hynek and
Kjeldahl Nitrogen on Aluminum Nitride
History As received BrZ-CH30H insoluble As received Brz-CH30H insoluble
Nitrogen recovered, g.
Xitrogen calcd.,. g.
0.000558 0,000558 0.000286 0.000286
0.000550 0,000550 0,000265 0.000265
Calculated from grams A1 recovered. Table 1. Aluminum Recoveries (Untreated and Treated AIN) Aluminum recovered. e .
Sample size, mg. 2
1
0.5
0.1
300
Untreated A1N 0.00106 0.00107 0.00106 O.OO106 0.00052 0.00051 0.00025 0,00026 0.00006 0.00004
Brominemethanol insoluble 0.00104 0.00106 0.00106 0.00107 0,00051 0.00052 0.00025 0.00025 0.00005 0.00005
ANALYTICAL CHEMISTRY
Table 111.
Sample Purchased AlN Residue No. 1 Residue No. 2 Residue No. 3 Residue No.4 Steel 169 A175 G417 G874 B15697 36671
Recovery of Aluminum Nitride
A1N Recovered, yo
Av.
Std. dev.
0.161,0.163,0.161,0.161 0.158,0.161,0.161,0.163,0.161 0.077,0.077,0.077,0.079,0.079 0.038,0.036,0.038,0.039,0.038 0 . 0 0 8 , 0.008, 0.009,0 . 0 0 8 , 0.006
0.162 0.161 0.078 0.038 0.008
0.002 0.001 0.001 0.001
0.0014,0.0014,0.0017,0.0011 0.039,0.036,0.038,0.038,0.044 0.065,0.059,0.061,0.065,0.064,0,053 0.082,0.074,0.082, 0.082,0.082,0.068 0.052,0.052,0.041,0.052,0.053 0 074, 0 071, 0 071, 0 067
0.0014 0.039 0.061 0.080 0.050 0 071
0.0002 0.003 0.005 0.006 0.005 0 003
0.001