Emission Spectrometric Determination of Oxygen in Niobium Metal

Effect of the column to particle diameter ratio on the dispersion of unsorbed solutes in chromatography. John H. Knox and Jon F. Parcher. Analytical C...
0 downloads 0 Views 3MB Size
Emission Spectrometric Determination of Oxygen in Niobium Metal F. MONTE EVENS and VELMER A. FASSEL Institute for Atomic Research and Departmenf of Chemistry, lowa State University, Ames, lowa

b The emission spectrometric technique for the determination of oxygen in metals has been extended to the determination of oxygen in niobium. A d.c. carbon-arc discharge is used to extract the oxygen content of the sample as carbon monoxide into a static argon atmosphere. Special electrode assemblies which support a platinum-niobium alloy or provide a critical graphite-niobium rehtionship are employed. The d.c. carbon arc dissociates the evolved carbon monoxide and the emission spectrum of atomic oxygen is excited. The intensity ratios of 07712A./Ar,891A. and 0777LA./Ar7891A. are related to the oxygen content of the samples. The concentration range from 0.004 to 0.60 weight can be covered. The preparation and evaluation of synthetic oxygen standards are discussed.

70

simplifications into the gas-handling system. The revised system is shown schematically in Figure 1, together with equipment specifications. If purification of the argon is required, the purification system previously described (7) may be inserted between the tank and excitation chamber. Two paths for exhausting the excitation chamber are provided in the simplified system shown in Figure 1. To protect the diffusion pump oil during initial evacuation of the excitation chamber gases, the bypass path through valve 6 is employed. When the pressure is reduced to the safe range of the diffusion pump, valve 6 is closed and valves 7 and 10 are opened. Valve 10 is c1os.d only during the initial evacuation. General refinements have also been made in the excitation chamber. A combination demountable port as-

A

the presence of small amounts of residual oxygen in niobium tends to increase the strength of the metal, larger concentrations have a degrading effect upon the ductility, creep strength, corrosion resistance, and other desirable properties (11, 16). Precise knowledge of the total oxygen content is, therefore, required to evaluate properly the effect of oxygen on the physical properties of the metal and to provide analytical control for commercial metal. Various modifications of the vacuum fusion (9, 12, I S , 17) and inert gas fusion (10, 14) techniques and vacuum extraction (8) have been employed to obtain this analytical information. As part of a continuing study on the scope of application of the emission spectrometric techniques developed in our laboratories (8-7), experimental conditions were established for the determination of oxygen in niobium metal. LTHOUQH

APPARATUS

Detailed descriptions of the experimental facilities and operating techniques have been presented (2-7). Several modifications worthy of note have been made in the high vacuum system and excitation chamber. The commercial availability of higher purity argon has made it possible to introduce 1056

ANALYTICAL CHEMISTRY

0-

Figure 1. High vacuum and handling system

gas-

1. Compressed gas cylinder 2, 6. Vacuum valves (Hoke, Inc. series 4 5 0 ) 3 . Excitation chamber 4. Absolute pressure g a g e (Arthur F. Smith Co. 0-760 mm.) 5. Alphatron ionization g a g e (National Research Corp. No. 5 2 0 ) 7. High vacuum gate valve (Vacuum Research Co. No.VG-1 N 5 ) 8. Stainless steel liquid nitrogen trap 9 . O i l diffusion pump (National Research Corp. No. H.4-P) 10. Shut-off valve (Imperial Brass Co. No. 493-CS) 1 1 . Phillips g a g e tube (Consolidated Vacuum Corp. No. PHG-06) 12. Auxiliary outlet to mercury manometer

sembly and counter-electrode holder has been added to the upper plate of the chamber as shown in Figure 2. The stainless steel bellows support for the counter-electrode holder provides flexibility in the vertical adjustment of the upper electrode. The closure assembly makes it possible to load individual samples into the supporting electrodes without exposing the outgassed electrodes and chamber to atmospheric gases. Recent studies have shown that the electrode and chamber outgassing operations can also be simplified. The outgassing and loading procedure now employed is as follows : The desired number of supporting electrodes are placed in the positions on the rotary table and the loading port is sealed. With one of the supporting electrodes in position opposite th? counter-electrode and with 400-mm. pressure of argon, a 30-ampere arc discharge ia initiated. The argon pressure is then reduced quick1 to approximately 25 to 50 mm., a t wKich pressure the arc is continued for 15 seconds. Without terminating the arc discharge, the next electrode is moved into position and arced for 15 seconds under the initial 25- to 50-mm. argon pressure. The arc discharge, on its own accord, readily transfers from one electrode to another as the electrodes are moved in and out of position. After each electrode has been arced in this manner, the chamber is evacuated, a fresh charge of argon is admitted, and the entire procedure is repeated. During these operations, only the O-ring cavities ( A , Figure 2) are water-cooled. After the outgassing operations are completed, the port closure assembly is opened while the chamber is flushed with argon a t flow rates of 2 to 3 liters per minute. To facilitate loading of samples into the electrode receptacles, a glass tube which fits snugly around the supporting electrode is lowered through the port opening. The samples are then dropped down the tube into the electrode. After all the samples are loaded, the port closure is sealed and the chamber is evacuated to a pressure of a t least lo-’ mm. Hg. The samples are then excited in the manner previously described (8-7). Another minor modification has been made on the location of the rotary table which supports the sample electrodes. The central pin is positioned off-center toward the front of the excita-

BMLER PLATE

0

TRANSE STAINLESS STEEL (316Ll STAINLESS STEEL BELLWSIROBERT N L X N CMrmOLS W l BRASS OR PLATED. BRASS (MOWFIED CENW VACUUM CCUPLING.NO. 8 4 w - 4 1

k

NEOPRENE

0 RING

PYREX CYLINDER ROTA&?

NEOPRENE 0 RING

-

ELE

HaDEl (13m:

Figure 2.

Excitation chamber

tion chamber. The resultircg increase in distance from chamber wall to d.c. arc discharge at the rear of the chamher has reduced glass ring breakage by more than 50% PRELlMiNARY EXPERIMENTS

The emission spectrometric technique for the determination of oxygen in metals employs a d.c. carhon-arc discharge in pure argon to melt the sample and to extract the oxygen content as carbon monoxide. The same arc discharge dissociates the extracted carbon monoxide and excites the emission spectrum of atomic oxygen. The in-

k - 5 . 2 1

tensity ratio of selected oxygen-argon line pairs is related to the oxygen content of the sample. The rate and dcgree of evolution of the oxygen content of the samples are critically dependent on the environmental conditions in the graphite electrode (9). For some metals, such as steels (4) and vanadium (Z), rapid quantitative extraction can he achieved if specially designed graphite supporting electrodes are employed. For these supporting electrodes, it is essential that the molten globule completely dissolve the retaining wall of the cavity. The anode spot of the arc discharge then rests directly on the globule. For other metals, such as titanium, zirconium, and yttrium, an electrode assembly is used which provides a molten platinum bath after the arc is initiated (3). In essence, the platinum hath provides a reaction me-

dinm from which the carbon monoxide evolution proceeds rapidly, reproducibly, and quantitatively. Optimal evolution of the oxygen content of niobium has been achieved with both types of supporting electrodes. As shown below, concordant analytical curves were obtained under both sets of experimental conditions. For niobium metal, a sample weight of 0.3 gram provides an acceptable compromise between concentration sensitivity and reasonable size of sample for accommodation by the supporting electrodes. When niobium metal samples are arced in graphite supporting electrodes of optimal dimensions, their behavior is similar to that observed for steels (7) and vanadium (9). Appreciable carbon dissolution from the receptacle wall occur8 rapidly, thus supplying an abundance of carbon to the melt, and allowing the anode spot to rest directly on the globule. Observations on the effect of progressive variation in electrode dimensions, similar to the studies reported for vanadium (9),showed that rapid, reproducible, and quantitative evolution of the oxygen content could he achieved at %-ampere arcing current with the electrode geometry shown in Figure 3. The electrodes before and after arcing are shown in Figure 4. The results of moving plate observation during the arcing oycle are shown in Figure 5,A. The slightly higher intensity ratio observed during the first 30 seconds reflects the higher concentration of carbon monoxide in the analytical gap during the evolution reactions. The constant intensity observed after the first 30 seconds indicates that steady state conditions are attained in this time period. Consequently, the spectrographic exposures are made during the 30- to 60-second period. The dimensions of the supporting electrode for thc platinum bath tpchnique are identical to those employed for titanium and other metals (S, 4). Figure 6 sham that for 0.3-gram samples of niobium only 0.2 gram of platinum

._.

Figure 3. Graphite electrode geometry Dimensions in millimeters

I

.

Figure 4. Graphite electrode beforc and after sample arcing

Figure 5. Intensity ratio a s a function of time A.

B.

Graphite electrode Platinum bath

VOL. 33, NO. 8 , JULY 1961

-

1057

02

0

‘)IEI(MT OF

0.

OB

OB

PLATIHW IGR4YSl

Figure 6. Intensity ratio as a function of weight of platinum

is required for maximum evolution of the oxygen content. However, for these weights of platinum, it is difficult t o prepare wafers that provide a well behaved reaction during the initial phases of the arcing cycle. Platinum wafers formed from 0.6 gram of platinum not only provided an easy-to-manipulate support for the sample during the preparative phase, but also gave maximum precision in the observed analytical intensity ratios. These electrode assemblies are arced under the same twostage arcing cycle previously described (8, 4). The alloying reaction between the niobium and platinum occura in a manner similar to the titanium-platinum reaction, although significant ,amounts of the oxygen content are Bvolved during the reaction flash (3). AB shown in Figure 5,B, a steady-state intensity ratio is not obtained until the 60- to 90-second arcing period. Consequently, the speotrographic exposures are made during this time interval. CALIBRATING STANDARDS

The standards employed for establishing quantitative calibration were derived from three sources: synthetic standards prepared by the dissolution of niobium pentoxide in niobium metal of low oxygen content, synthetic standards prepared by reacting molecular oxygen with niobium metal of low oxy-

Table 1.

gen content, and samples in which the total oxygen content was determined by vacuum fusion methods (9, 18, 19, 17). The dissolution of niobium pentoxide in niobium metal was accomplished by conventional arc melting of a 20- to 25-gram metal button after a preliminary packing and sealing of high purity oxide into several drilled cavities. Preliminary a d d s were made to seal each of the surface plugs in order to avoid loss of oxide through spattering. Homogeneous dispersion of the dissolved oxide was obtained by eight succcssivc meltings with intervening turning of the button. Loss of weight during thcse operations was negligible. The preparation of synthetic standards by the reaction of molecular oxygen with niobium metal involved a procedure similar to that described by Walter (16). The temperature necessary for obtaining complete reaction 25” C. Each prepared was 650” standard represented a sample size equivalent to that consumed during a single spectrometric determination.

*

CALIBRATION EXPERIMENTS

The calibration experiments were conducted with both versions of supporting electrodes. Pertinent experimental details for both are summarized in Table

I. Examination of the data obtained from the three groups of calibrating standards revealed that the standards prepared by the addition of molecular oxygen to niobium metal and vacuum fusion-analyzed samples produced concordant data. However, an appreciable number of the standards prepared by the arc-melting dissolution of niobium oxide in niobium metal yielded analytical intensity ratio data which were significantly low. This suggests that some of the oxygen was lost during the preparative procedure, perhaps

Experimental Conditions for Spectrographic Evaluation of Prepared Standards

Graphite Electrode Line pairs

0mi.m and

T e of electrodes Tathode

Anode

Argon pressure Anal tical gap Emdsion Wave length Filter slit Exposure time Development Emuleion calibration Background corrections

1058

Platinum Bath A.

A. 0 . 3 gram Nb 0.3 gram Nb 0.6 gram Pt National Carbon Co. Spectro-Tech Grade inch diameter graphite rod, 1”s i n c b long Undercut electrode as Undercut electrode as shown in Fig. 3 shown in (3) 640 mm. reproduced to 6 mm. 6 mm. Eastman 1-N 7300-8500 A., 1st order Corning 2-63 0.05 mm. 30-second prearc 60-second prearc 30-second exposure 30-second exposure 4 minutes a t 21” C. in Eastman Kodak D-19 with continuous agitation Two-step sector, preliminary curve method None required for normal exposure Ar

Weight of sample

Oms.rr

ANALYTICAL CHEMISTRY

1891.076

e,{

0

4o

A PLATINUM BATH

GRAPHITE ELECTRODE

‘*y//

OXYGEN CONCEWTRATION I W 1 . x )

Figure

7. Analytical curves

through the formation of a volatile lower oxide (NbO) of niobium. Figure 7 ahows that concordant analytical curvea were obtained from the two version8 of supporting electrodes. The major portion of the curvature at lower concentratione was produced by the slight amount of background accompanying individual sample exposures. This background contribution to the measured intensity ratio was significant only at ratios below 0.4. The magnitude of the “oxygen blank” present during the arcing of consecutive samples waa less than 0.001%. For the analysis of samplea, the graphite supporting electrode version is preferred, eince sample preparation is simplified and a 5- to 7-p.p.m. oxygen blank contribution from the platinum bath is avoided. Certain niobium alloya, notably thoee containing up to 1.0% titanium, do not appear to liberate their total oxygen content when the simple graphite supporting electrodes are used. In theae instances, the platinum bath version is indicated. PRECISION

Precision data were obtained from single exposures of samples on indi-

Table II.

Precision Data for Graphite Electrode Procedure

Av.

Relative Standard Deviation

%

(1)

Oxygen Concn.,

0.180 0.092

3.96

0.020

0.60 0.35 0.27

0.045

0.009

0.005

2.28

1 .os

8 11 11 9

7 7

3.0

3.3 8.7

3 .O 10.0

20.0

O m c a . / A r t a ~ i ~ .

0.420 0.180 0.092 0.046 0.020

4.16 1.85

1.os

0.58 0.34

5 8 11 11 9

3.6 3.9

3.3 8.7

10.0

vidual photographic plates over a period of several weeks. Table I1 summarizes the data obtained for the graphite electrode procedure. Relative standard deviation is reported on the basis of oxygen concentration. The platinum bath procedure yielded similar data. ACKNOWLEDGMENT

The authors are grateful to the Fanstcel Metallurgical Gorp. and to E. I. du Pont de Nemours & Go., Inc., for supplying several niobium samples used in this investigation. LITERATURE CITED

(1) ANAL.CHEM.33, 480 (1961).

(2) Fassel, V. A., Altpeter, L. L., Spedtocham. Acta 16,443 (1960).

(3) Fasscl, V. A., Gordon, W. A., ANAL. CHEM.30, 179 (1958). (4) Fassel V. A., Gordon, W. A. Jasinski, R. J., “hoc. 2nd Intern. Conference on Peaceful Uses of Atomic Energy,” Vol. 28 p. 583-92, 1958. (5) Fwse V. A., Gordon, W. A., Jminski, R. J., Evens, F. M., Rea. unw. nines 15, 278 (1959). (6) FasscI, V. A. Gordon, W. A., Taheling, It. W., A S T M Spec. Publ. 222, 41-60 (1958). (7) Faasel, V. A., Tabeling, R. W., Speclrochin. Acta 8, 201 (1956). (8) Hansen, W.R., Mallet, M. W., ANAL. CHEM.29, 1868 (1957). (9).Harris, W. F., “Technology of Columbium (Niobium),” p. 57-9, Electrochemical Society, %ley, New York, 1958. (10) Laboratory Equi ment Cor St. Jose h, Mich., “Con&ctimetric &ygen for Ferrous and Nonferrous Annkaer . _. Met&and Alloys,” 1959. (11) McIntosh, A. B., J . Zml. Metals 85, 367 (1957).

’P,

(12) Niobium Task Force, Oxygrn Srth Group, Division M, Committee E-3, ASTM, “Recommended Method for the Determination of Oxygen in Com-

mercial Grade Niobium, Vaciium, Fusion Platinum Bath Technique,” 1960. (13) Parker, A., “Determination of Case3 in Metals,” pp. 64-74 Iron and Steel Institute, Percy Lund, Humphries & Co. London, 1960. (14) Qmiley,W. G.,ANAL.CHEM.27,1098 (1955). (15) Tottle, C. It.,J. Inst. Metals 85,375 (1957). (16) Walter. D. I., ANAL. CHEM.22. 297 ’ (i950). ’ (17) Wilkens, D. H., Fleischer, J. F., Anal. Chim. Acfa 15,334 (1956). RECEIVED for review January 3, 1961. Accepted March 29, 1961. Contribution No. 977. Work performed in the AmLaboratory of the U. 5. Atomic Energy Commission.

A Critical Review of Colorimetric and Spectrographic Methods for Gold F. E. BEAMISH Departmenf o f Chemistry, University of Toronto, Toronto 5, Ontario, Canada

b This review deals with the colorimetric and spectrographic methods for gold recorded up to August 1960. In view of the wide industrial and scientific uses for gold and its alloys one must hope for the development of a more useful choice of well defined instrumental methods. Among those currently available the author ‘recommends the bromaurate, tin(ll), and rhodanine spectrophotometric methods, and internal standard methods for spectrographic determinations of solutions. For the direct analysis of gold alloys no spectrographic procedure was found which gave adequate attention to homogeneity.

I

T IS evident that the most acceptable spectrographic methods are of Europcan origin, which situntion may be due to the reluctance of the industrial organizations and consulting institutes which possess modern spectrographic equipment, t o make their methods of analysis generally available. This secrecy is regrettable, particularly since these organizations are privileged to use the benefits of a century of scientific literature. Furthcrmore, from a long range point of view, one may doubt that the economic advantagcs of a policy of secrecy would equal the benefits derived from a free interchange of analytical information. I n the case of spectrophotometric methods for noble metals,

i t is unlikely that private organizations retain any significant amount of analytical knowledge unavailable to the analyst. The history of the development of colorimetric and related methods for gold emphasizes the general approach of the researcher to the development of new analytical procedures. Long established color reagents are forced into the role of quantitative adaptations and relatively little effort is made to tap the great volume of organic compounds, from which source one can cxpect sensitive and specific reagents. While investigations directed toward the establishment of structure-reagent relationships and the subsequent syntheses of suitable compounds have provided few new analytical procedures, one cannot thus dismiss the responsibility of the researcher to produce a choice of more effcctivc spectrophotometric methods for metals such as gold, whose applications are a concern of so many fields of scicntific activity. An empirical approach, however distasteful to the inflexible mind, will permit the examination, in a relatively short time, of large numbers of reagents and conditions of application. With a little ingenuity in the construction of suitable equipment one may obtain a volume of data which could provide many new analytical reagents as well as encourage more fruitful generalizations concerning structure-reagent relationships.

While it is generally accepted that very dilute solutions of gold salts rapidly decrease in gold content, there has been little recognition of the action of light on these solutions. Svedberg (63.4)has discussed the finding that alkaline gold solutions yield both more numerous and smaller particles if exposed to ultraviolet light prior to the addition of a reductant. Since many of the proposed colorimetric methods require a very low acid or a basic medium, one may expect that, prior to the addition of the reduct8ant,lengthy exposure to light would be detrimental to precision and accuracy for both colloidally dispersed and dissolved colored constituents. An examination of this characteristic could be of considerable value. The instability of dilute solutions of gold and other dissolved constituents was also discussed by Leutwein (90A), who ascribed the loss of strength to base exchange reactions with the glass container and to adsorption. Solutions containing 0.001% of gold showed 0.1 to 0.3% of the original strength after storage of 230 days in Jena glass, whereas in quartz flasks the loss was very slight. With one exception ( S A , @ A ) , the colorimetric methods for gold involve absorptivity measurements on colloidal gold solutions, on the colored oxidized product of the organic precipitant, or on an organic extract of the colloidal suspension. No colored goldVOL. 33, NO. 8, JULY 1961

1059