<<

''V ersary ,

Af/NES 75TH Af/NES 7 5

Spectral Characteristics of Grinding By By W. D. Riley, B. W. Dunning, Jr., and D. M. Soboroff Sparks Used for Identification of M etals UNITED UNITED STATES DEPARTMENT OF THE INTERIOR M X 8| 9 3 2

Bureau of Mines Report of Investigations/1985 •so Report of Investigations 8932

Spectral Characteristics of Grinding Sparks Used for Identification of Scrap M etals

By W. D. Riley, B. W. Dunning, Jr., and D. M. Soboroff

UNITED STATES DEPARTMENT OF THE INTERIOR Donald Paul Hodel, Secretary

BUREAU OF MINES Robert C. Horton, Director Library of Congress Cataloging in Publication Data:

Riley, W. D, (William D.) Spectral characteristics of grinding sparks used for identification, of scrap ,

(Report of investigations ; 8932) Bibliography: p. 13. Supt. of Docs, no.; I 28.23:8932, I, Scrap s—Iden t if ¡cat ion. 2, tests, I. Dunning, B, W, II. Soboroff, D. M. {David M,). Ill, Title. IV. Series: Report of in­ vestigations (Ignited States, Bureau of Mines) ; 8932«

TN23.U43 [TS214] 822s [869.1*028] 84-600301 CONTENTS Page

Abstract...... 1 Introduction...... 2 Experimental equipment and procedure...... 3 Spectral radiance measurements...... 3 Photographing spark patterns...... 4 Sample preparation...... 5

Results and discussion...... 6 Conclusions...... 11 References...... 13

ILLUSTRATIONS

1. Types of spark patterns from metals and alloys held in contact with

moving whe e l ...... 2 2. Schematic representation of rapid-scanning spectrometer ...... system 3 3. Schematic representation of transmission filter-and- detector system...... 4 4. Photograph of spark for 1020 ...... 5 5. Spark spectra of 1020 and 316 ...... 7

6 . Effect of photodiode array exposure times on spectrum of 316 stainless steel...... 7 7. Spark spectra for several metals showing effect of increasing

content using photodiode array spectrometry...... 8

8 . Spark spectra for some nickel- alloys...... 8 9. Rapid-scanning spectrometer system...... 9 10. Spark spectra of 304 and 316 stainless steel using rapid-scanning

spectrometer system...... 1 0 11. Photographs of spark patterns showing the effect of carbide formers,

1 0 2 0 carbon steel and2 0 1 stainless steel...... 1 2

TABLES

1. Spark-identification systems...... 4 2. Alloys used in study of spark-pattern characteristics...... 5 3. Spark intensity for various alloys...... 9 4. Measurement of maximum intensity and wavelength...... 10 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT f t/mln foot per minute ms millisecond in inch nm nanometer ym s second ys microsecond wt pet weight percent mm millimeter SPECTRAL CHARACTERISTICS OF GRINDING SPARKS USED FOR IDENTIFICATION OF SCRAP METALS

By W. D. Riley, 1 B. W. Dunning, Jr.,2 and D. M. Soboroff3

ABSTRACT

The Bureau of Mines is conducting research on methods for the rapid identification and sorting of mixed scrap metals and alloys so that domestic scrap resources can be fully utilized. One method, , is the examination of the pattern of sparks that results when a metal or is ground on an abrasive wheel. This technique requires experienced sorters. Bureau researchers have shown that spec- trophotometric methods can be used in lieu of the highly skilled sorter. The spectra of a mild steel, an HSLA steel, several stainless , and four nickel-base alloys have been measured. Differences in the spectra that are sufficient for identification are apparent.

^ Chemist. ^Supervisory metallurgist. ^Supervisory research chemist. Avondale Research Center, Bureau of Mines, Avondale, MD. ?..

INTRODUCTION

Conservation of strategic metals andthe frequency of bursts is related to the minimizing imports through improved meth­carbon content, with the intensity and ods of scrap materials aresize of bursts increasing with increasing among the objectives of the Bureau of carbon content. Noncarbide-- ele­ Mines Minerals and Materials Recyclingments like , nickel, and programr Since the first step in conven­suppress spark bursts. Carbide-forming tional recycling technology for metalselements like , , and and alloys is to segregate the mixedtungsten change the color of spark pat­ scrap into groups of like materials, anterns as well as suppress spark burst. important aspect of this work has beenCobalt- and nickel-base superalloys and the evaluation and introduction of new alloys also have characteristic technologies to identify and sort scrapspark streams. Descriptions of patterns metals (j^, 1_2, 15).4 Conventional tech­ have been published for various steels, niques such as chemical spot testing, de­superalloys (2-_4> 16-17, 19), and nickel termination of magnetic properties, and(3_). Specific articles on the sorting of spark testing still play a major role carbonin and alloy steels have also been

the soring of scrap metals in many scrappublished (2 , J3, 18). Jn many of these y a r d s . articles, different descriptions are of­ fered for the same metal. The spark pat­ Spark testing requires observationtern of of 316 stainless steel has been de­

the pattern of sparks that results whenscribed a as 1 ) ( short reddish spark with metal or metal alloy is ground on an ab­few spear points (16-17); (2) orange to rasive wheel. Sparks occur because moststraw colored with no spear points (4, metals and alloys, in the finely divided9); and (3) orange to straw colored with

state, oxidize rapidly when heated to in­spear points2 0(). candescence. When metals or alloys are ground using a high-speed , Optical emission has been

the fine particles torn loose are heatedused to identify scrap materials1 _, 1 2 (, by the frictional forces of the wheel and15). There also has been a study~[5) by overcoming rupture resistance. Asthat the used optical emission spectroscopy metal or alloy reaches incandescent tem­ perature, it burns. The resulting pat­ tern of sparks can be used to identify Continuous many metals and alloys by examining the Carrier lines - spark pattern for shape of spark stream, Disjointed color, and various charcteristics that are imparted by alloying constituents. There have been attempts to standardize -Sprigs this technique by specifying the type of abrasive wheel, its speed, and the typeCarbon bursts ■ Preliminary bursts of lighting that should be employed in -Main bursts the working area2 _,( 7_, 1 1 ); however, the identification is still very dependent on Spear points the skill and experience of the tester.

Figure 1 shows some of the characteris­ Tongues tics that are found in a spark stream whose flight path Is from right to left

( 1 2 ). for example, in -base alloys;Forked tongues

FIG U R E 1 . - Types of spark patterns from ^Numbers in parentheses refer to items in the list of references at the end of metals and alloys held in contact- with mov­ this report. ing abrasive wheel, 3

to examine the ignition by sparks of ex­be used to quantify or improve this tech­ plosive atmospheres. However, identific­nique as a means of identifying metals ation of alloys from the optical spectraand alloys (14). This report describes of a spark pattern has not been previous­progress to date using several types of ly reported. Experiments were conductedoptical instruments. to see if spectroscopic techniques could

EXPERIMENTAL EQUIPMENT AND PROCEDURE

SPECTRAL RADIANCE MEASUREMENTS s. The detector and spectrometer system is controlled by a signal analyzer with The spectra of the sparks were recordedspectrometer module. Data reduction and using photodiode array optical spectros­storage is performed with a copy, rapid-scanning optical spectrom­microcomputer» etry, and transmission filters and detec­ tors. The detectors used in each systemThe rapid scanning spectrometer system were most sensitive to the visible range(fig. 2 ) used a silicon detector for of light wavelengths based on the assump­spectral detection and a spectrometer. tion that color, hue (degree of chroma- The wavelength scan rate was fixed at 10 ticity), and brightness were the most im­nm/ms. Data were displayed on an oscil­ portant factors in sample identification. loscope equipped with dual tracing unit, The ultraviolet and infrared regions areone trace displayed the spectra while the not included in this study. other displayed the wavelength- The spectrometer was also equipped with a The photodiode array optical spectrom­wavelength marker, a scan control unit, a eter has been described in detail else­sample and hold unit, and recorder out­ where (5), a general description is put.as This system could be configured to follows: The photodiode array consistedeither display an entire spectrum or of 1,024 photodiodes on a section of 25­ intensity at at fixed wavelength. Data mm silicon chip. The detector was at­reduction was accomplished on a microcom­ tached to a spectrometer. In this con­puter by digitizing the recorder output figuration, a 300-nm window can be ex­with a digitizing tablet. amined with the photodiode array. For maximum sensitivity, a 250-ym slit widthThe transmission filter and silicon de was used. Exposure time was fixed attector 0.5 system (fig. 3) consisted of three

CRT

-Detector Scanning spectrometer Recorder

F IG U R E 2. - Schematic representation of rapid-scanning spectrometer system. 4

transmission filters (transmissionft/min. cen­ The wheels were dressed prior to tered at 450, 550, 650 nm), three silicon the examination of each spectra to insure detectors, a trifrucated fiber opticthat there was no contamination from pre­ cable, and a multiplexor-amplifier digit­vious samples. Variations in intensity al readout package. This unit displayeddue to physical differences were minimiz­ the intensities at the wavelengths trans­ed by holding the speed of the grinding mitted through the filters. wheel constant and the flatplate sample size to 3- by 1.5 in and by fixing sample Table 1 lists the degree of complexity,angle of contact. The pressure of the approximate costs, and ease of use of thesample (as suggested by standard spark various sytems discussed in the paper.testing convention) was adjusted to the The most easily used system would bepoint a where the sample stopped bouncing rapid scanning spectrometer coupled withon the wheel. Therefore, hard alloy sam­ a computer to control all the functions, ples required slightly more pressure than including identification of the metal. did soft alloy samples.

The sparks were generated by holding a PHOTOGRAPHING SPARK PATTERNS sample against a grinding wheel equipped with either an8 -in aluminum oxide wheel Several attempts were made to photo­

(medium grit) or an8 -in carborundum (al- graph the spark patterns. The best oXite) rotating at approximately 7,500photographs were obtained by using 400

TABLE 1. - Spark-identification systems

Equipment Cost Equipment Ease of use

Transmission filter and silicon detector..< $ 1 , 0 0 0 Simple... Moderate. Photodiode array optical spectrometer..... 60,000 Complex. . Complex. 2,500 . . »d o. ... Moderate. Rapid-scanning spectrometer w/cornputer. ...7,000 . . .d o. ... Simple. 5

ASA color or black-and-white film in an spark pattern is easily seen. The flight autoexposure camera (shutter priority).of the sparks is from right to left. F ilm brand made l i t t l e d iffe re n c e . The room was darkened and the camera was set SAMPLE PREPARATION for a 0.30-s exposure so that time- averaged pictures were obtained. As fig­ The 3- by 1-1/2-in metal samples listed ure 4, shows, much of the detail of the In table 2 were prepared by degreasing in

TABLE 2. - Alloys used in study of spark-pattern characteristics

Al loy Major alloying elements wt pct (nominal) Other Ni Cr Mo Ti elem entsFe 201 ...... 70 5 18 NS NS Mn 301...... 73 7 17 NS NS Mn 72 9 18 NS NS Mn 3 04...... 69 9 19 NS NS Mn 316...... 72 10 16 2.5 NS Mn, Si 410...... 85 NS 13 NS NS

H a s te llo y...... 5 62 NS 28 NS

46 32 21 NS NS Mn Inconel 625...... 2.5 61 22 9.0 0.2 Cb, Al, Mn In co n el 600...... 8.6 75 16 NS NS Mn

1020 CS...... Bai. NS NS NS NS Mn,P,S,C HSLA steel...... Bai. .25 .75 .30 NS Mn,P,Si,Cu NS Not specified:

FIGURE 4. - Photograph of spark pattern for 1020 carbon steel. 6

acetone and light abrasion with a1 2 0 scrap yard, the only preparation is to grit wheel to remove anylightly abrade the sample prior to actu­ surface contamination. Typically, in theal spark testing»

RESULTS AND DISCUSSION

Each of the instruments used provided stainlessa steel (fig. 5). What appeared slightly different way of examining theto be noise In the the initial trans­ spectra of sparks. The photodiode arraymission filter and silicon detector ex­ optical spectrometer exposes the indi­periments was in reality, the fluctu­ vidual diodes to light for a predetermin­ations in the spectral intensity. ed amount of time. Data from the photo-­ diode array would not show variations Thein spark patterns apparently contain the light source. When the system isno fine structure on the time scales used

used in a "one-exposure" mode, as it wasfor collection» Figure6 shows the spec­ in this study, the spectra displayed showtrum for 316 stainless steel collected the integrated intensity at each wave­using the photodiode array spectrometer length for the measurement period. Theat relatively long (0.5 s) and short (0.1 transmission filter and silicon detectors) exposure times. system works in a similar manner. The multiplexor-amplifier is designed with an Figure 7 shows the spectrum (0.5 s) for electronics package that samples the out­a mild steel, a stainless steel, an Inco- put from the silicon detector for1 s nel, and an Incoloy. A comparison of the prior to display. This time period wasindividual spectra reveals variations In chosen because initial measurements madeintensity versus wavelength that are for some samples had rapidly varyingattributable to composition changes. As voltages on the direct-current voltmeterthe amount of alloying is increased from that appeared to be "noise." The rapid­a mild steel to stainless steel to Inco­ scanning spectrometer consists of loy an and finally to Inconel, a decrease in Ebert-type monochrometer, which uses the a maximum intensity and a shift toward continuously rotated, blazed-diffractionthe red end of the spectrum is apparent. grating as the dispersing element. The scan rate is1 0 nm/ms, and a complete Figure 8 shows the spectrum (0.5 s) for scan takes 95 ms. When unalloyed carbonIncoloy 800, Inconel 625, Hastelloy B, steels were examined using this sytem,and it Inconel 600. Conventional literature was discovered that there were veryon spark testing states that the latter strong intensity fluctuations at specificthree alloys have short red sparks. The wavelengths in the spectra. It appearsdifferences in spectral intensity versus that these changes are produced by thewavelength are evident using a spectral carbon bursts (s tar^-shaped formationsanalyzer. in figure 4), which cause an increase in ra­ diating surface due to the particle Because of the lack of spectral struc­ bursting into a number of smaller frag­ture of the spectrum, a simpler instru­ ments. Because this process is not con­ment arrangement—the transmission tinuous and the spectrometer is samplingfilter-silicon detector system—might a different spectral region of the sparkyield adequate information for alloy pattern, variations in intensity withidentification. The wavelengths of 450, time at each wavelength are present.550, and 650 nm were selected based on When the spectra of a highly alloyed, the full spectral patterns obtained pre­ low-carbon material is examined, thereviously. The measurements were made by are few, if any, carbon bursts, hence theadjusting the fiber optic probe so that curve is much smoother. The spectrum forthe light from the spark was gathered at the unalloyed carbon steel has a higherits most intense area, approximately2 to standard deviation (SD) than for 316 INTENSITY, arbitrary units INTENSITY, arbitrary units GURE E R U IG F 7 55 7 775 675 575 475 6 GURE E R U IG F - fet f htdoe ra epsr tms n pcrm of spectrum on times exposure array photodiode of Effect - . 5 - pr seta of spectra Spark - . VLNGH, nm , GTH AVELEN W nm , GTH AVELEN W 1020 abn te and steel carbon 316 ti es steel. less stain 316 ti es steel. less stain 7 INTENSITY, arbitrary units INTENSITY, arbitrary jnlts 8 htdoe ra spectrometry. array photodiode GURE E R U IG F 7 . - Spark spectra for several metals showing effect of increasing nickel content using using content nickel increasing of effect showing metals several for spectra Spark - . GURE E R U IG F 8 - pr seta o sm nce-ae alloys. nickel-base some for spectra Spark - . AEEGH nm WAVELENGTH, AEEGH nm WAVELENGTH, 9

5 in from the wheel. Table 3 shows the(fig. 1 0 ) or display the maximum, inten­ results of these experiments. The dif­sity of the spark at a particular wave­ ferences in the intensity values indicatelength (table 4). The results of these that as alloying increases, so does themeasurements show the same trends as the wavelength at which the maximum intensityprevious data. While some of the mater­ occurs. Additionally, the intensityials tested visually have different spark ratios show that there is enough of patterns,a others such as 316 and 304 spectral difference to allow the samples to be identified. Use of the ratios of TABLE 3. - Spark intensity for various the intensities at specific wavelengths alloys, in nanometers corrects for variations in overall inten­ sity observed in previous work in colorA1 loy Intensity Intensity ratio

sorting 1 ( 1 ). 450 550 650 450/650 550/650

1 0 2 0 cs. 0 . 2 0 1.4 1 . 1 0.18 1.3 Spectra were also collected using theHSLA

rapid-scanning spectrometer. This systemsteel. . .33 1 . 6 1 . 1 .30 1.5 (fig. 9), was more versatile than the316 ss. . .18 .37 .52 ,35 .71 transmission filter and silicon detectorIn cone1 because it can show the entire spectrum625____ .04 .15 .38 . 1 0 .38

FIGURE 9. - Rapid -scanning spectrometer system. 10

stainless steels have very similarcarbon pat level increased, the temperature terns. However, figure 10 shows, thereof the spark increased (1.00 pet C = are sufficient differences in the spectra1,740° C). In steels containing W, Cr, of the alloys to allow them to be sepa­and Mo, the temperature decreased as the rated using this instrumental method» amount of alloying increased» Riezler's measurements agree very well with the da­ TABLE 4. - Measurement of maximum ta generated in this study. Our work and

intensity and wavelength that of others 8 () indicate that if the alloying elements have a higher melting Maximum point than the iron matrix, then the tem­ Alloy Wavelength, intensity, perature of the sparks will be lower and

n m 1 arbitrary the color will be tinted toward red. The

units1 degree of change in color tint then de­

1 0 2 0 c s...... 646±2 97±6 pends on the amount of alloying elements. 651±5 53±2 If, however, the alloying element has a 683±3 39 ± 1 lower melting point (such as Mn), the 316 ss...... 672±3 27±1 spark pattern will be both brightened and

Inconel 625... 709±2 2 0 ± 1 shifted toward the hotter end of the

Incoloy 800... 721±3 29±1 spectrum 8 ^). ( Higher melting components ^-Numbers on the right of the symbol (±)cause larger shifts. represent the standard deviation (7 measurements). Several authors have suggested that the oxidation of carbon to carbon dioxide and Alloying changes the spark temperature.possibly plays the prin­ Riezler (13) showed that an unalloyedcipal role in the spark stream of steels carbon steel (0.10 pet C) had a spark (2_, 7_~8_, 10). A rapidly heated particle temperature of 1,640° G and that as the tends to become spherical, the surface

3,000 T

c 2,000 £• a

JO a

(/> 2 1,000

475 575 675 775 WAVELENGTH, nm

F IG U R E 10 . Spark spectra of 304 and 316 stainless steel using rapid-scanning spectrometer system. 11

begins to oxidize to , and the(0.03 pet C) stainless steels were com­ carbon is oxidized to carbon dioxide.pared. The spectra are identical because The iron oxide easily flakes off, andthe carbon was bound up as chromium and iron is rather soft and plastic at themolybdenum carbides. Because the rapid temperatures in a spark stream. Sincerate of combustion allows no time for the carbon dioxide occupies a volume onecarbon to go into solution in y iron, an order of magnitude greater than the ori­equilibrium does not exist. This is es­ ginal carbon, the gas can escape from thepecially the case in the high- and low-

pellet causing the secondary bursts2 ^, (carbon 316 stainless steels where the ]_~8) . If this is true, it has been sug­chromium and molybdenum carbides do not

gested 2 ( ) that the alloying elements go readily into solution. might change the characteristics of the oxide film in such a way as to give the This research shows that spectroscopic various spark characteristics. examination of the spark pattern provides a means of identifying scrap metals that Another explanation maybe that most ofpreviously could not be identified by the alloying elements, except nickel, co­conventional visual spark testing tech­ balt, and manganese, readily form car­niques. For example, previous public­ bides that are stable at high tempera­ations (4_, 12) indicate that the presence tures so that there is insufficient car­of a high-nickel alloy is indicated by a bon in the matrix to form iron carbide.short red spark and that additional test­ If we compare the pictures of 201 stain­ing by chemical spot testing is required less steel (approximately 0.15 pet C) tobefore the specific nickel-base alloy can

1 0 2 0 carbon steel (approximately0 . 2 pet be identified. Although there are many C) in figure 11, there are very fewhigh-nickel alloys, the preliminary bursts in the 201 stainless steel. Theindications are that use of this new

2 0 1 stainless steel contains 16 to 18 pettechnique will permit specific identifi­ Cr, which is a strong carbide former. cation A of each alloy. This technique similar observation was made when would the allow the separation of 316 stain­ spectrum of 316 (0.08 pet C) and 316 L less steel from 304 stainless steel.

CONCLUSIONS

The intensity, color, and pattern ofidentification tool in the scrap yard, grinding sparks are determined by the al­and can outperform conventional spark loying elements in the sample investi­testing, eliminating the possibility of gated. Successful separation of alloyhuman error and even identifying alloys samples by conventional spark testing inis groups that can not be separated by determined by the experience of the indi­conventional means. The results of pre­ vidual performing the operation. The in­liminary experiments show that even mate­ dividual observing the sparks, developsrials whose spark patterns appear through repetition, a memory system similarto to the eye such as 304 and 316 recognize which metal alloys are beingstainless steels, Incoloy 800, and Inco- tested. Spectrophotometric examinationnel 625 can be distinguished by this new of grinding sparks can be used as method. an 12

F IG U R E 11 .- Photographs of spark patterns showing the effect of carbide formers. A, 1020 carbon steel; B, 201 stainless steel. 13

REFERENCES

1. Brown, R. D . , Jr., W. D. Riley,Testing. Stahl Eisen., v.6 8 , No. 17, and C. A. Zieba. Rapid Identification of 18, 1948, pp. 301-303 (Brutcher Trans. Stainless Steel and Scrap.3472). BuMines RI 8858, 1984, 23 pp. 11. Maynard, A. W . , and H. S. Cald­ 2. Buzzard, R. W. The Utility ofwell, Jr. Identification and Sorting of the Spark Test as Applied to CommercialNonferrous Scrap Metals. Proceedings of Steels. J. Res. NBS, v. 11, Oct. 1933, 3d Mineral Waste Utilization Symposium pp. 527-540. (cosponsored by BuMines and ITT Res. Inst., Chicago, IL, Mar. 14-16, 1972). 3. Dehorn, J. Alloy Separation and ITT Res. Inst., Chicago, IL, 1972, pp. Verification. Proceedings of the 29th 255-264. Annual Meeting of the Investment Institute. Dallas, TX, Oct. 15-17, 1981. 12. Newell, R . , R. E. Brown, D. M. Investment Cast. Inst., Dallas TX, pp.Soboroff, and H. V. Makar. A Review of 12-16. Methods for Identifying Scrap Metals. BuMines IC 8902, 1983, 19 pp. 4. Defense Supply Agency. Defense Scrap Yark Handbook. NAVSANDA 5523, AFM 13. Riezler, W., and L. Hardt. Tem­ 68-3, MC 401.2a. June 1966, pp. IV,37- perature Measurement on Grinding Sparks.

IV,42. Z. Angew. Ph y s . , v.6 , 1954, pp. 497­ 499. 5. Driscoll, T. J., D. R. Flinn, and W. C. Lederer. Application of a Photodi­ 14. Riley, W. D . , and R. D. Brown, Jr. ode Optical Spectrometer to the Study of Device and Technique for Scrap Metal the Incidivity of Light Alloys ImpactingSpark Pattern Recognition. U.S. Pat. on Rusted Steel. BuMines RI 8684, 1982, Appl. 623753, June 26, 1984. 19 pp. 15. Riley, W. D . , R. E. Brown, and

6 . Hildorf, W. G . , and C. H. McCol- D. M. Soboroff. Rapid Identification and lam. Metal Pellets, Produced by SparkSorting of Scrap Metals. Conserv. and

Tests, Used To Identify Alloy Steels.Recycling, v.6 , No. 4, 1983, pp. 181— Iron Age, v. 126, No. 1, July 1930, pp. 192. 1-4. 16. Iron Age. Short-Cut Tests Iden­ 7. Practical Aspects of Sparktify Stainless Steels. V. 194, Sept. Testing. Iron Age, v. 124, No. 15, Oct. 1964, pp. 152-154. 1929, pp. 953-956. 17<, Metal Industry (London). Steels

8 . Hunger, J., and 0. Werner. Theand Their Sparks. V. 20, 1922, pp. 1-4. Influence of Structure on Plain Carbon and Alloy Steels Upon the Spark Stream in 18. American Society for Steel Treat­ Spark Testing. Arch. Eisenhuettenwes.,ing. Transactions of The American Soci­ v. 23, No. 8 , 1952, pp. 277-286 (Brutcher ety for Steel Treating. Aug. 1922, pp. Trans. 3198). 550-553.

9. Canadian Machinery Metal Work. 19. Tschorn, G. Spark Atlas of How Sparks Help Identify Metals. V. 73,Steels. Macmillan, 1963, 212 pp. May 1973, p. 83. 20. Weed, W. I. How To Identify 10. Janiche, W . , and K. H. Saul. How Stainless Steel. Production, v. 55, No. To Sort Plain Carbon Steels by Spark2, Feb. 1965, pp. 89-91. irU.S. C P O : 1985-505-019/20,014 INT.-BU.O F MIN ES,P GH..P A. 27896