944 14nalyst, October, 1979, Vol. 104, $9. 944-951 Determination of Antimony and Other Trace Elements in irons and Steels by Atomic-absorption Spectrophotometry with Introduction of Solid Samples into an Induction Furnace* A. M. Aziz-Alrahman and J. B. Headridge Depavtment of Chernistvy, The Univevsity, SheBeld, S3 7HF Atomic-absorption spectrophotometry with an induction furnace has been used for the determination of 0.5-350 p g g-l of antimony in 1-20-mg samples of irons and steels dropped into the furnace. Calibration graphs of peak absorbance vevsus the mass of antimony have been constructed by using standard steels. Information is presented on the accuracy and precision of the method for 13 irons and steels. The limit of detection for antimony is Calibration graphs have also been obtained for indium in a nickel-base alloy and thallium, tin, selenium, tellurium, zinc and cadmium in steels in order to establish the conditions that will be necessary to determine these elements using a procedure similar to that employed for antimony.Factors affecting the volatility of trace elements with boiling-points below 2300 “C are discussed. 0.12 rf.g g-:. Keywovds : A ntipnony detevmination ; trace-element detevnzinations ; ivon and steel analysis ; atomic-absovption spectvophotometvy ; induction furnace The presence of antimony is known to be detrimental to certain mechanical properties of low-alloy steels. As little as 10 pg g-l produces 350 “C embrittlement and temper brittle- ness.l Low concentrations of antimony also adversely affect the hot workability of stainless steels2 There is a continuing need to determine antimony in steels within the range 1-500 pg 8-l.Many methods are available for the determination of antimony above 100 pg g-l but few methods are suitable below this level and only the most sensitive can be employed to determine concentrations of antimony below 10 pg g-l. Information about some of these methods is given in Table I. TABLE I METHODS FOR THE DETERMINATION OF ANTIMONY IN STEEL Method Molecular-absorption spectrophotometry of rhodamine B complex D.c. arc-emission spectrography after pre-concentration by precipita- . . . . . . .. . . after solvent extraction . . . . tion of Sb,S, on CuS . . . . . . . . . . . . . . Hollow-cathode emission spectrometry on solid samples .. . . Flame atomic-absorption spectrophotometry . . . . .. . . pentan-2-one after solvent extraction . . . . . . . . . . tion after dissolution of the alloy . . . . . . . . . . Flame atomic-absorption spectrophotometry of Sb(II1) in 4-methyl- Flame atomic-absorption spectrophotometry after generation of SbH, Atomic-absorption spectrophotometry with carbon-furnace atomisa- Concentration range of antimony presentlpg g-l 10-500 5-500 -0.540 20-350 2-350 1-350 Reference 3 4 5 6 7 8 9 10 Methods involving solvent extraction or other pre-concentration procedures are relatively slow and the most attractive methods so far reported are the atomic-absorption methods involving hydride generation,8 atomisation in resistively heated carbon furnacesgtio and hollow-cathode emission ~pectrometry.~ Headridge and co-workers have already reported * Presented a t the Meeting on “Research and Development Topics in Analytical Chemistry,” Heriot- Watt University, Edinburgh, July, 1979.AZIZ-ALRAHMAN AND HEADRIDGE 945 on the determination of bisrnuth,ll silver12 and lead13 in steels using atomic-absorption spectrophotometry with the introduction of solid samples into an induction furnace.In this paper, results for a similar study on antimony are reported along with data on other volatile elements that might at some time have to be determined in steels containing very low concentrations of these elements. Experimental Materials of Standards, USA, and alloys from the Institutet for Metallforskning, Sweden. that no more than three pieces need to be added to the furnace core at the same time.Standard irons and steels. Samplesfor analysis. British Chemical Standards, alloys from the National Bureau These should preferably be millings or turnings of irons or steels so Apparatus and Method for Determining Absorbances for a Series of Solid Samples The apparatus and the method were identical with those described previouslyll except that the graphite core and side-arms were made from Ultra Superior Purity Grade graphite (Ultra Carbon) and baked for 6 h under vacuum at about 1900 "C before use. Information on the hollow-cathode lamps used in this work is given in Table 11. TABLE I1 DATA ON HOLLOW-CATHODE LAMPS Element Manufacturer Line employed/nm Lamp current used/mA Antimony Indium . . Thallium Tin .. Selenium Tellurium Zinc . . Cadmium .. . . Pye Unicam . . . . Pye Unicam .. . . Perkin-Elmer .. . . Activion . . . . Activion . . .. EM1 .. . . Activion .. . . Activion 231.1 303.9 276.8 286.3 196.0 214.3 213.9 228.8 10 5 20 20 6 5 8 13 In all instances, except for selenium, a slit width of 0.2 nm was used on a Perkin-Elmer 300s atomic-absorption spectrophotometer ; the slit width used for selenium was 0.7 nm. The experimental conditions for the determination of antimony are shown in Table 111. Calibration Graphs A ntimmy For the determination of antimony in irons and steels containing less than 150 pg g1 of antimony, calibration graphs of peak absorbance versus the amount of antimony are obtained by dropping increasing amounts of 0.2% carbon steel JK2C, which contains 29 pg g1 of antimony, into the graphite core under conditions capable of producing absorbances of up to 1.0 (see Table 111).For steels containing 150-350 pg g-l of antimony, a calibration graph is prepared in a similar way using mild steel BCS 328 containing 260 pg g-l of antimony. Other elements Although series of irons and steels were not analysed for these elements, the conditions necessary for obtaining a calibration graph for each element in steel were established using one standard alloy in each instance. These conditions are given in Table IV.946 AZIZ-ALRAHMAN AND HEADRIDGE : DETERMINATION OF ANTIMONY Analyst, Vol. 104 TABLE TI1 EXPERIMENTAL CONDITIONS FOR THE DETERMINATION OF ANTIMONY Concentration Mass range of Core temperature/ Scale rangelpg g-' sample/mg "C Damping* expansion <5 10-20 2 620 1 x 5 5-150 1-10 2 550-2 640 1 x l 150-350 1-6 1810 4 x l * Damping positions 1 and 4 are for time constants of 0.2 and 10 s, respectively.Determination of Antimony in Irons and Steels Irons and steels with a silicon to antimony ratio >5: 1 When a series of irons and steels is to be analysed, suitable masses are dropped into the graphite core over a period of 2-3 h and, during the same run, various masses of JK2C or BCS 328 are also added, generally at the beginning of the run, for the purpose of constructing a calibration graph. During a run the temperature of the core should not alter by more than &lo "C. When the run is completed the calibration graph is drawn and the mass of antimony in each sample is obtained from the graph.The concentrations of antimony in the samples are then calculated. Irons and steels with a silicon to antimony ratio <5 ; 1 The procedure is similar to that above except that 5 g of silicon powder (BDH Chemicals Ltd.) are added to the cold core and well tamped down with a glass rod before the core is heated at approximately 1900 "C, as read on the optical pyrometer, for a sufficient period of time to remove all antimony from the silicon (12 h in our case); samples may then be added to the core. TABLE 1IV EXPERIMENTAL CONDITIONS FOR OBTAINING CALIBRATION GRAPHS FOR CERTAIN MORE VOLATILE ELEMENTS I N STEEL Damping position 1 (time constant 0.2 s) was used in all instances except zinc, for which damping position 5 (time constant 30 s) was employed.Element Indium . . .. Thallium . . . . Tin . . . . . . Selenium . . .. Tellurium . . . . Zinc . . .. . . Cadmium . . .. Standard alloy R3387* BCS 312 BCS 323 SRM 361 SRM 361 JKlC J K2C Content/ -10 240 CLg g-' 0.37 40: 2.0 -0.03 6: Mass range Core temperature/ of sample/mg "C 9-15 2 370 1-13 2 330 2-20 2 450 3-30 2 430 1-7 2 350 1-7 1570 8-35 2 350 Scale expansion x l x 5 x l x l x 1 x l x6 * A nickel-base alloy. t See reference 14. 1 Value for SRM 1261. Results The calibration graphs for the range 0-360ng of antimony obtained with steel JKZC passed through the origin and were slightly curved at the upper end. The mass for 1% absorption was 1.4 ng, this value being obtained from a typical calibration graph for antimony at 2630 "C. The calibration graph for the range 0-1.6 pg of antimony obtained with steel BCS 328 also passed through the origin but was appreciably curved because of the damping applied to the amplifier signals.This damping was necessary in order to keep the apparent absorbance readings below 1 .O. Samples of a series of irons and steels were dropped into the furnace in order to obtain their antimony contents using the conditions outlined in Table I11 and in the section entitled Determination of Antimony in Irons and Steels. The results are shown in Table V.October, 1979 AND OTHER TRACE ELEMENTS IN IRONS AND STEELS BY AAS TABLE V RESULTS FOR THE DETERMINATION OF ANTIMONY IN IRONS AND STEELS Calibration graph prepared using JK2C (29 pg g-' of antimony), except for results marked $, when BCS 328 (260 pg g-l of antimony) was used.Alloy BCS 260/4 336 337 456 457 459 460 SRM 361 362 363 364 365 JKlC Antimony reported/ clg g-' < 10 28, 29* 22t 110 290 70 20 42 130 20§ 3407 <0.5)1 2.111; 1, 2* Antimony Number of samples foundlpg g-1 analysed 2.2 6 31 7 29 8 116 7 270: 8 62 19 19 8 Relative standard deviation, yo 15 12 7 8 18 8 16 41 7 14 135 7 8 16 8 8 0.65 11 9 337: 7 7 1.3 12 23 947 * The figures shown were obtained by Frech.9 t Result obtained by Burke.l5 $ Si was added to the core prior to addition of the samples because Si: Sb < 5 : 1. 5 Certificate value for SRM 1263; SRM 363 is the same material in the form of chips. 7 Certificate value for SRM 1264; SRM 364 is the same material in the form of chips. (1 Single result quoted on the certificate.All other figures in the second column are certificate values. With the other elements, straight-line calibration graphs passing through the origin were obtained for indium, thallium, selenium, tellurium and cadmium. The calibration graph for tin passed through the origin but was curved towards the mass axis, particularly for masses of tin in excess of 2.5 pg. The calibration graph for zinc was extensively curved because of the high degree of damping necessary to keep the apparent absorbances below 1.0. From these graphs the mass for 1% absorption was obtained in each instance. These data are shown in Table VI. TABLE VI MASSES REQUIRED FOR 1 yo ABSORPTION FOR VARIOUS ELEMENTS Element . . . . .. . . In T1 Sn Se Te Zn Cd Mass for 1% absorptionlng .. -0.7 0.14 46 12 0.18 0.043 -0.11 Discussion Steel JK2C has not been fully standardised for the amount of antimony it contains, but contents of 32 and 27 pg g1 are reported on the certificate and Frech9 has found antimony concentrations of 27 and 29pgg-l. The average value of 29 pgg-l was selected as the antimony content of this steel. Steel BCS 328 has been fully standardised and contains 260 pg g1 of antimony. The accuracy of the results for antimony shown in Table V is considered to be good, the agreement between our results and those for standardised steels being very satisfactory. The precision of the method is acceptable, although not so good as that for a similar method for lead reported previ0us1y.l~ This is probably because, in this study, the furnace was usually operated at a temperature in the range 2550-2640 "C, which is about the maximum temperature that can be achieved with the equipment already described.ll Corrosion of the graphite surfaces of the furnace, from trace amounts of reactive gases left in the argon stream and from air not completely removed from the graphite during the heating-up procedure, is more noticeable at higher temperatures and this leads to more scatter in the results.948 AZIZ-ALRAHMAN AND HEADRIDGE : DETERMINATION OF ANTIMONY A%aL'yst, Vd.104 Lundberg and lirech16 have reported that scatter due to inhomogeneity should not be the major contribution to the relative standard deviation when milligram masses of steel are analysed for antimony by adding turnings to a furnace. For the first time since using this design of induction furnace, a matrix effect has been observed for the determination of an element in steel.Steels BCS 457 and SRM 364 pro- duced consistently high results for antimony when added to the normal graphite core. Both alloys were analysed on four different occasions using five or more samples each time and the average results were 449 pg g-l for BCS 457 and 521 pg 8-l for SRM 364. A close examination of the chemical composition of all the steels, which had been added to the core, showed that BCS 457 and SRM 364 were exceptional only in that the ratios of the concentra- tions of silicon to antimony were low at 1.6: 1 and 2.6: 1 , respectively, while in all other instances, the ratios were in excess of 6: 1. I t was felt that the presence of silicon must have reduced the rate of diffusion of antimony in molten steels, when the ratio of silicon t o antimony was in excess of 5: 1, such that the rate of release of antimony into the gas phase was diminished.I t was decided to add silicon powder to the graphite core in order toensure that the ratio of silicon to antimony in the molten metal was in excess of 5: 1 and when this was done, acceptable results were obtained for BCS 457 and SRM 364 as reported in Table V. Incidentally, steels with a silicon to antimony ratio of less than 5: 1 are unusual so the need to add silicon to a core will occur very occasionally. The atomic-absorption spectrophotometer was operated without a background corrector. To check that there was negligible molecular absorption and negligible light scattering from relatively high concentrations of the elements present in steels that are volatile at high temperatures, the instrument was peaked up at 231.1 nm using the antimony hollow-cathode lamp, which was then replaced with a deuterium lamp, and samples of Specpure manganese (Johnson Matthey & Co. Ltd.), JK2C, BCS 457 and SRM 364 were added to the furnace core at 2540 "C.In no instance was there an absorbance reading in excess of 0.01 unit. Manganese is present in almost all steels and boils at 1962 "C. The limit of detection of the method is 0.12 pg g-l of antimony and was taken as twice the standard deviation for sample SRM 365. This limit of detection is lower than any other reported in Table I. Over a period of 4 years a considerable amount of data has been accumulated on the determination of trace elements in steels using atomic-absorption spectrophotometry with the introduction of solid samples into a constant-temperature induction furnace and an assessment of the real value of the method can now be given.Detailed results have already been reported for bisrnuth,ll silver12 and lead13 in irons and steels, and have just been given for antimony in irons and steels. The method can be applied to the determination of many trace elements that are sufficiently volatile at 2600 "C or lower temperatures. These elements include indium, thallium, tin, lead, antimony, bismuth, selenium, tellurium, silver, zinc and cadmium, for which calibration graphs have been obtained. In theory the method should also be suitable for the determination of trace concentrations of magnesium, calcium, aluminium, gallium, manganese and copper in steels.However, the method is so sensitive that appreciable background absorption from magnesium, aluminium, manganese and copper continuously released from the graphite core has been obtained by us even using vacuum-degassed USP grade graphite. Suitable calibration graphs for very low masses of these elements have not yet been obtained. Gallium should present no problems with respect to background absorption. The masses required for 1% absorption for various elements have already been reported in this paper (see Results). In Table VI the sensitivity for zinc is for a damped signal and the mass for 1% absorption in the absence of damping will be considerably less than 43 pg.The results for indium and cadmium may well be significantly in error. The indium content of the nickel-base alloy R3387 used to construct the calibration graph is nominally 10 pg g-l but could be appreciably lower. The cadmium content of JK2C is given as 0.03 pg 8-l on the certificate, from a single determination, but this figure seems to be too high because the mass for 1% absorption would be expected to be similar to that for zinc. The reliable masses for 1% absorption for the elements reported here along with those for bismuth,ll silver12 and lead,13 reported earlier, have been converted into concentrations for 10-mg samples and are shown in Table VII. Also shown in this table are the concentrations of these elements in solution (pg ml-l) producing 1% absorption in an air - acetylene flame,October, 1979 AND OTHER TRACE ELEMENTS IN IRONS AND STEELS BY AAS 949 using the same resonance lines for the elements as for the furnace work,17 and these con- centrations converted into pg g-l for steels assuming that 2 g of steel are dissolved in 100 ml of solution in each instance.At the furnace temperatures employed, all these elements are monatomic in the vapour phase.18 In the air - acetylene flame with a temperature of about 2300 O C , all these elements with the exception of tin should exist mainly in the monatomic form. Tin will be present both as the monatomic form and as the monoxide.19 It will be noticed that the CIA values for thallium, lead, bismuth, tellurium and silver are all between 1300 and 1800 while those for tin, antimony and selenium are much lower.Admittedly the sensitivities reported in column A were not all obtained at the temperature of the air - acetylene flame; the sensitivity increases with increasing temperature. For example, 1 pg of lead from BCS 327 gave absorbances of 0.15 and 0.27 at 2010 "C and 2150 "C, respcctively,13 and 20 ng of tellurium from SRM 361 gave an absorbance of 0.29 at 2190 "C and of 0.59 at 2350 "C. However, if all the results in column A had been obtained at the temperature of the air - acetylene flame, then the C / A values for thallium, lead, bismuth, tellurium and silver would have been in excess of 1000 and those for tin, antimony and selenium would have been lower than those reported in Table VII.The results shown -in Table VII are very interesting. TABLE VII DATA ON CONCENTRATIONS FOR 1% ABSORPTION OBTAINED BY ADDING SOLID SAMPLES TO AN INDUCTION FURNACE AND FROM NEBULISING 2% m/V SOLUTIONS OF STEELS INTO AN AIR - ACETYLENE FLAME Concentration for 1% absorption Element Thallium . . . . Tin . . .. . . Lead . . . . . . Antimony . . . . Bismuth . . . . Selenium . . . . Tellurium . . .. Silver .. .. Furnace Resonance temperature/ line/nm "C 276.8 2 330 286.3 2 450 283.3 2 000 231.1 2 630 306.8 2 070 196.0 2 430 214.3 2 350 328.1 2 270 r A * / Pg g-' 0.014 4.6 0.017 0.14 0.015 1.2 0.018 0.001 8 Btl c:/ pgm1-l CLgg-l 0.5 25 3.5 175 0.5 25 1.1 55 0.5 25 0.5 25 0.5 25 0.06 3 CIA 1786 38 1471 393 1667 21 1389 1667 * Using induction furnace. t Following dissolution of steel in acid and atomisation in an air - acetylene flame, assuming $ After converting pg ml-l into pg g-l of steel.that the sensitivities are the same in the presence and absence of iron. These CIA values of approximately 1500 in Table VII are found for the elements that quickly diffuse out of the molten globule of steel into the gas phase within our induction furnace. These are also the elements that should be readily removed from steel on vacuum melting. Indeed, Chernov and Ageev20 reported that lead and bismuth were readily removed from iron on vacuum induction melting, tin and antimony were removed more slowly and arsenic was not removed a t all. The boiling-points of the eight elements under discussion are shown in Table VIII.On boiling-point alone, tin should be the most difficult element to remove by vacuum melting or by volatilisation in our furnace but, intuitively, one would expect these elements, which show some non-metallic properties, to be held back in molten steel because of bonding between the elements and iron atoms. These elements include tin, antimony and selenium and, indeed for them, CIA values much less than 1000 were obtained. If tin were entirely monatomic in the air - acetylene flame, the sensitivity would be less than 175 pg g-l and the CIA value would be less than 38 as given in Table VII. TABLE VIII BOILING-POINTS OF VARIOUS TRACE ELEMENTS FOUND IN STEEL Element . . . . . . . . T1 Sn Pb Sb Bi Se Te Ag Boiling-point/"C . . . . . . 1457 2270 1740 1750 1560 685 990 2212950 AZIZ-ALRAHMAN AND HEADRIDGE : DETERMINATION OF ANTIMONY Analyst, VoZ.104 In this study, attempts were made to obtain a calibration graph for arsenic in steel - arsenic sublimes at 613 "C at atmospheric pressure, but without success. Even at 2570 "C there was no evidence for the presence of arsenic atoms in the gas phase, the absorbance being zero in each instance when milligram masses of steels containing arsenic were added to the graphite core through which the resonance line for arsenic at 193.7 nm was being passed. The limits of detection for the elements that readily volatilise from steel have been estimated and are shown in Table IX. For a single-beam atomic-absorption spectrophotometer such as the Perkin-Elmer 300s and an air - acetylene flame, the relative standard deviation of a determination is often approximately 1 yo when solutions are nebulised to give absorbances between 0.2 and 0.8.Under these conditions the limit of detection is frequently one fifth to one tenth of the concentration for 1% absorption. With our induction furnace the precision is poorer with a relative standard deviation of 5-10y0, and hence one would reasonably expect a limit of detection about the same as the concentration for 1% absorption for a 10-mg sample. The estimated limits of detection in Table I X are taken as the concentrations for 1% absorption (pg g-I) that have been given in Table VII (column A ) . This was carried out as follows. TABLE KX ESTIMATED AND ACTUAL LIMITS OF DETECTION FOR TRACE ELEMENTS READILY VOLATILISED FROM STEELS Element Thallium ... . Tin . . .. . . Lead . . .. . . Antimony . . . . Bismuth .. . . Selenium .. . . Tellurium . . . . Silver . . .. . . Estimated limit of detection in steel/ Pg g-l 0.014 4.6 0.017 0.14 0.015 1.2 0.018 0.001 8 Actual limit of detection in steel/ Pg g-l - < 0.02* 0.12t 0.004S - - 0.005§ * See reference 13. t Reported in this paper. $ See reference 11. 3 See reference 12. Considering the assumptions that have been made in the method for estimating the limit of detection, the agreement between actual and estimated limit of detection in the instances where actual limits have been determined, is reasonable. I t can be said with certainty that the actual limits of detection for thallium and tellurium and also zinc and cadmium will be well below 0.1 pg g-l but that the limits of detection for tin and selenium at the tempera- tures given in Table VII will be not as good.Of course, the limits of detection for both tin and selenium may well be better at temperatures in excess of 2600 "C. The determination of trace elements in metals by atomic-absorption spectrophotometry with the introduction of solid samples into electrically heated graphite furnaces has been shown to be convenient and reliable on many o c c a ~ i o n s . ~ ~ - , 1 ~ ~ ~ ~ - ~ ~ There is evidence that introducing samples into a graphite furnace being maintained at constant temperature is to be preferred.21 The furnaces may be inductively or resistively heated but the cost of the latter is cheaper, unless a spare induction generator is available.It is to be hoped that instrument manufacturers will soon provide graphite atomisers to which metal samples can be conveniently added at controlled constant temperatures. Facilities for measuring both peak height and peak area will also be desirable. We are indebted to the BSC/BISPA Chemical Analysis Committee and the British Steel Corporation for a grant to buy the Perkin-Elmer 300s atomic-absorption spectrophotometer.October, 1979 AND OTHER TRACE ELEMENTS IN IRONS AND STEELS BY AAS References 95 1 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. Capus, J. M., Iron Steel, Lond., 1965, 38, 594. Lynch, D. W. P., Proc. Elect. Furn. Steel Conf. A m . Inst. M i n . Metall. Petrol. Engvs, 1961, 19, 220. “Methods of Chemical Analysis of Iron and Steel,” British Steel Corporation, Sheffield, 1974, p. 19. Balfour, B. E., Jukes, D., and Thornton, K., Appl. Spectrosc., 1966, 20, 168. Thornton, K., Analyst, 1969, 94, 958. Barnett, W. B., and Kerber, J . D., Atom. Absorption Newsl., 1974, 13, 56. Headridge, J. B., and Smith, D. R., Lab. Pract., 1971, 20, 312. Fleming, H. D., and Ide, R. G., Analytica Chirn. Acta, 1976, 83, 67. Frech, W., Talanta, 1974, 21, 565. - Barnett, W. B., and McLaughlin, E. A., Analytica Chzm. Acta, 1975, 80, 285. Andrews, D. G., and Headridge, J . B., Analyst, 1977, 102, 436. Aziz-Alrahman, A, M., and Headridge, J. B., Talanta, 1978, 25, 413. Andrews, D. G., Aziz-Alrahman, A. M., and Headridge, J . B., Analyst, 1978, 103, 909. Burke, K. E., Appl. Spectrosc., 1974, 28, 234. Burke, K. E., Analyst, 1972, 97, 19. Lundberg, E., and Frech, W., Analytica Chim. Acta, 1979, 104, 67. “Analytical Methods for Atomic Absorption Spectroscopy,” Perkin-Elmer Corp., Norwalk, Conn., “Encyclopedia of Science and Technology,” McGraw-Hill, New York, 197 1. Mavrodineanu, R., Editor, “Analytical Flame Spectroscopy,” Macmillan, London, 1970, p. 27. Chernov, B. G., and Ageev, P. Ya., Stal’, 1968, 28, 1003. Lundberg, E., and Frech, W., Analytica Chim. Acta, 1970, 104, 75. Lundberg, E., and Frech, W., Analytzca Chznz. Acta, in the press. Backman, S., and Karlsson, R., Analyst, in the press. Marks, J . Y . , Welcher, G. G., and Spellman, R. J., Appl. Spectrosc., 1977, 31, 9. 1976. Received, April 20th. 1979 Accepted May 18th. 1979