refractory metals with the W. If solution difficulties are encountered in spite of precautions taken to avoid loss of too much HF during the dissolution steps it may be advisable, in the analysis of samples containing 0.1 to 5% of W,to modify the procedure as follows:
W was then extracted as the thiocyanate
After the destruction of the "OS, pour, and wash the solution to a platinum dish. Add 2 ml. of HF, heat to dissolve the WO3, cool, add 50 ml. of water, pour, and wash back into the beaker and thence into the 100-ml. volumetric flask. Then proceed as directed.
T o evaluate the proposed method for the analysis of steels, 0.1-gram samples of various XBS steels plus varioui ahquots of standard ITT solution (1 mg. or 0.1 mg. of W per ml.) were transferred to 100-ml. beakerb and the composite samples were then analyzed for W by the proposed method. I n some instances the analysis was made on the whole sample after removal of the bulk of the Fe by extraction with ketone (analysis of Iow-JY steels); in other instances, on an aliquot of the sample solution (analysis of low-alloy steels containing 0.1 to 5Y0 IT). Correction was made for any W found in the steel itself.
RESULTS
T o investigate the removal of interference of various met hls in the determination of JT by the proposed method, 1 ml. of standard W solution (0.1 mg. of If' per ml.) plus 2 rrl. of HF(1 9) plus 1 ml. of HC1 plus 1-ml. aliquots of standard chloride or fluoride solutions of various metals was added, in the order indicated, to a 75-ml. separatory funnel. The solution was diluted to 15 ml. and, unless otherwise stated, a single cupferron separation was made. The solution was evaporated and the
+
and measured spectrophotometrically. The results obtained are shown in Tables I to 111. The ketone extract of the JY was slightly brown in the analysis of the samples containing 10 mg. of N o .
Of those steel samples tested, the only one found to contain It' was KBS sample 101-e. This sample n a s found to contain 0.0274 W. The results obtained are shonn in Table IV. Several W-bearing XBS samples of
steel were analyzed by the proposed method. Samples 158, 153, and 134-a were dissolved in HSOa-HF (9 1) and samples 123-a and 123-b were dishdved in HCI. The, latter samples were analyzed on aliquots containing 10 mg. of sample and also on 50-mg. portions of the samples after removal of most of the Fe by ketone extraction. I n the latter instances the samples were dissolved, oxidized, and treated with formic acid as directed in the procedure for corrosion-resisting steels and then treated as directed in the procedure for steels containing less than 0.174 W. Chromic acid was reduced before the cupferron extraction.
+
The results obtained are shown in Table V. LITERATURE CITED
(1) Kakita, Y., Goto, H., S a . R r p t . Res. Znst. 7'ohoku Cnzo. 15 ( A ) , N o I , 1 (1963 1 ( 2 j Luke, C. I,., AXAL.CHEM.2 8 , 1275 (1956). (3) Zbid., 33, 1365 (1961). ( 4 ) Zbid., p. 1964.
RECEIVED for review February 3, 1064. Accepted Ifarch 30, 1964.
A Combus tio n, T herma I -Conduc tivity An a Iyze r for Carbon in Steel L. L. LEWIS
and
M. J. NARDOZZI
Applied Research laboratory, United States Steel Corp., Monroeville, Po.
b For carbon determinations, steel samples are burned in an induction furnace in which carbon is converted to carbon dioxide in a sti,eam of oxygen. The gaseous mixture i:, swept from the furnace and through a molecular-sieve trap, which collects the carbon dioxide a t ambient temperature. After all of the carbon dioxide ha 5 been collected, the trap is placed in art oxygen carriergas stream of a gas chromatograph and heated to desortl the carbon dioxide, The desorbed carbon dioxide then passes through a thermal-conductivity detector that gives an electrical signal, which indicates the carbon content of the sample. The analysis is complete 30 seconds Zfter the end of the flush and combustion period, which requires 95 seconds. Data obtained on a group of different steels containing from 0.01 to 0.9%, carbon showed a relative standard deviation of 2.570, the maximum being 4.3%. The relative standard deviation a t the 0.001 level was 10%.
70
F
A N D MORE SENSITIVE control analyzers for carbon are required to keep pace with the new developments in steelmaking operations. These
ASTI:,K
analyzers must be rugged and simple for reliable routine use, and must be accurate and sensitive enough to analyze steels containing as little as 0.001% carbon. Chemical methods for the determination of carbon invariably involve combusting the sample in osygen t'o convert the carbon to carbon dioside (1, 4, '7). The carbon dioxide content can be determined in about 2 minutes with inexpensive and sturdy equipment. The lower sensitivity limit for rapid routine carbon determinations, however, is only about, 0.037, ( 2 ) . The measurable carbon range has been extended below 0.03y0 by absorbing the carbon dioside in a solution of barium or sodium hydroxide and relating the change in the electrical conductivit'y of the solution to the carbon content of the sample. The time required for one determination is relatively long, about 5 minutes, because the osygen flow rate must be limited to about 400 cc. per minute for complete absorption of the carbon dioxide in the solution (a!6). Recently developed carbon-control analyzers are based on the thermal conductivity of a mixture of carbon dioxide and oxygen ( 3 , 9). The carbon
dioxide-oxygen combustion mixtures usually contain some argon and nitrogen. The argon and nitrogen contents n-ill vary during measurements in accordance with the nitrogen content of the metal sample, the purity of the oxygen used for the combustion, and the completeness with which air is purged from the measuring a1)paratus before the determination. Argon and perhaps nitrogen may often be pres:rnt in quantities large enough to affect thermal-conductivity measurements and thus limit the precision and accuracy of carbon determinations, es1)rcially a t carbon concent'rations be lo^ 0.057,. When argon and nitrogen are not separated from the carbon dioxideoxygen mixture, the carbon detcrininations will be susceptible to the described accuracy limitatioiis. 1he trchniyue of using a molecular r 7
sieve to concentrate carhon dioxidc rapidly and q u a n t i t a t i v e l ~ for ~ detrrmining carbon or oxygen content in metals has been mentioned in the literature ( 11 ) . With this techriiq~e. argon and nitrogen do not interftw with Present address, I.B.31. Itesearch Center, Yorktonn Heights, S . Y VOL. 36, NO. 7, JUNE 1964
1329
the determination because they are not adsorbed on the trap from the osygen stream. Since work on thi- technique was completed, inqtruments similar in principle to the one developed have been desciil~ed(8, 13).
MTECTOR OXlGCN S T R U M
TO A W O S P H E R L
REFERWE SIDE
YEASURlNe
EXPERIMENTAL
Combustion Apparatus. Regular t a n k osygen was prepurified by passing it through a copper oxide furnace, a sulfuric acid scrubber, and a column containing magnesium perchlorate and Ascarite. The purified osygen entered the induction furnace (Leco S o . IH16) through the top of a quartz jet tube (Leco Xo. 550-122) that contained a crucible (Leco Y o . 528-11) holding the sample and 1 gram of copper or tin to accelerate the combustion. (Ah induction furnace was used instead of a rcsistance furnace t)o obtain a high temperature for a short combustion period.) The gas misture resulting from the combustion was swept by the osygen through a dust trap (Leco KO. 501-10) to remove metal osides, over manganese dioside to remove sulfur oxides, through a heated tube containing copper o d e to osidize a n y carbon monoxide, t,o carbon dioside, and over niagnesium perchlorate to remove any water. Adsorption Trap. T h e carbon dioxide was separated from the oxygen stream by adsorption in a trap rontaining a molecular sieve. The trap consisted of a quartz V-tube (6-mm. o.d., 4-mm. i d . , and 54 cm. long) t h a t contained 4 grams of l I 16-inchmolecularsieve pellets (Linde Co., No. 5:i). After studies were conducted with traps of different dimensions and various amounts of molecular-sieve pellets, the trap described was chosen because with it, carbon dioside could be collected reproducibly from osygen streams flowing a t 1100 cc. per minute. The pe11(+ filled each side of the U-tube to within 2 cm. of the top. Tufts of glass wool were placed above the pellets to prevent them from blowing abaut nhcn the Imssure changed. Glass-to-metal
Figure 1. Solenoid-operated handling system
connectors (Veeco Corp ) joined the U-tube to the gas-handling bystem T o deborb the caibon dio\ide from the molecular-.ieve trap. the trap u a s heated psith resistance ribbon (DrirerHarrir C o ) The ribbon was wound tightly aroiind the part of the trap that held the molecular cieve. The temperaturc of the molecular sic\ e during desorption could not he meahured holrever, the resistance heater became a dull orange in about 10 seeonds, and thc carbon dioxide was completely de,orl)ed within 30 secondI., ANAL. CHEM.34, 868 (1962). (12) blooney. J. B., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 1962. (13) Walker, J. M.,Kuo, c . w., AKAL. CHEM.35, 2017 (1963).
RECEIVED for review December 16, 1963. Accepted March 23, 1964.
VOL. 36, NO. 7, JUNE 1964
1331
