THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING A STUDY OF TEE FUNDAMENTALS OF THE EFFECT OF DEOXIDATION ON THE CREEP CHARACTERISTICS OF PLAIN CARBON STEEL Clayton Dale Dickinson This thesis was submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the University of Michigan. June, 1956 IP-166

ACKNOWLEDGEiMENT We wish to express our appreciation to the author for permission to distribute this thesis under the Industry Program of the College of Engineering. i

AUTHOR'S ACKNOWLEDGEMENT The author wishes to express his appreciation for both the inspiration and helpful guidance he received from Dr. JO Wo Freeman during this study. He also wishes to thank the other members of his doctoral committee, Professor L, O. Brockway, Professor R. Ao Flinn, Associate Professor J. J. Martin, Professor C. A. Siebert, Associate Professor L. H. Van Vlack, and other members of the staff at the University of Michigan who, in many ways, have contributed immeasurably to this studyo The suggestions and help of members of the Research Laboratories of the Allis-Chalmers Manufacturing Company are also deeply appreciated by the author. The author is grateful to the Allis-Chalmers Manufacturing Company for both the moral and financial support they have given this project. The author is keenly aware that this work could not have been completed without the help of many other individuals and wishes to thank each one. ii

TABLE OF CONTENTS Page LIST OF TABLES. o o o o...o o o o iv LIST OF ILLUSTRATIONS.......... v Section I INTRODUCTION.. C e... o e. o o 1 IIo REVIEW OF THE LITERATURE....... 6 Effect of Deoxidation and HeatTreatment Effect of Other Manufacturing Variables Description of Strain Aging Phenomena Effect of Strain Aging on Creep Phenomena III. EXPERIMENTAL DESIGN, MATERIALS AND PROCEDURE................ 26 Experimental Design Materials Procedure IV. RESULTS.......... o e 40 Effect of Deoxidation and HeatTreatment Effect of Other Manufacturing Variables Effect of Other Variables Mechanism Tests V. DISCUSSION OF RESULTS........ 55 Relation of Dissolved Nitrogen and Creep Strength Discussion of the Mechanism VI, SUMMARY AND CONCLUSIONS........ 76 TABLESo. o o.o o o o 0 o. 80 FIGURES a................. 94 APPENDIX......... 118 LIST OF REFERENCES................ 129 iii

LIST OF TABLES Table Page I. Chemical Analyses of Commercial Steels.. 81 II. Chemical Analyses of Vacuum and Air Melted Steels 0 0 o o o 0 0. 82 IIIo Austenitic Grain Size of Steels at Various Austenitizing Temperatures. o.. 83 IV. Creep Rate and Dissolved Nitrogen for Deoxidized Steels. 0 0 84 Vo Analysis of the Effect of Heat-Treatment on Aluminum and Aluminum Oxide in Aluminum Killed Steels 0 o. o.. 0 85 VI. Creep Rates and Total Nitrogen for Rimmed Steels o 0 0 0 o 0 o 0 0 0 0 0 0 0 0 86 VII. Creep Rate and Dissolved Nitrogen for Steels in the As-Rolled Condition' o. 87 VII i Creep Rate and Dissolved Nitrogen for Spheroidized and Unspheroidized Conditions of Steels "C" and "F" 0 0 0 0 O o a o 88 IX, Creep Rate and Dissolved Nitrogen for Vacuum Melted and Vacuum Extracted Steels o 89 X. Hardness of Deoxidized Steels in Basic Correlation of Figure 2.. o... 90 XI Hardness of Rimmed and Other Steels Tested 91 Tested 0 0 0 0 o 0 0 o0 0 o 0 o 0 0 0 o 0 0 91 XII. Measurement of Aluminum Nitride Before and After Testing at 8500 F. for Times up to 1660 Hours o0 o o o...0 92 XIII. Summary of Activation Energies for Creep Tests on Plain Carbon Steel o..0 a o o 0 93 iv

LIST OF ILLUSTRATIONS Figure Page 1, Design of Basic Experiment for Commercial Materials Showing Nominal and Actual Aluminum and Silicon Contents. o 95 2, Correlation of Creep Rate With Active Nitrogen for Deoxidized Steels.... 96 3. Effect of Nitrogen on the Creep Rate of Rimmed Steels o o.0 o o o o o 97 4, Effect of Active Nitrogen on the Creep Strength of As-Rolled or Stress Relieved Plain Carbon Steel............ 98 5. Effect of Spheroidization on Creep Strength of Fine and Coarse Grained Steels... 99 6. Relation Between Dissolved Nitrogen and Creep Rate Before and After Spheroidization... 100 7o Microstructure of Vacuum Extracted Sample of Silicon Killed Steel "C"........ 101 8. Effect of Nitrogen Extraction on the Creep Rate of Silicon Deoxidized Steel "C". o. 102 9o Effect of Nitrogen on the Deoxidation-Heat Treatment Relation for Steels with Low and High Nitrogen..,......... o o 103 10o Manganese Effect for Steels with 0001 to 0,002 Percent Nitrogen...... 104 11. Isostrain Curves for Steels "tC" and "F" in Fine and Coarsened Condition.. o o 105 12. Creep Rate - Time Plot for Steel "C" and "F" in the Stress Relieved Condition at 10000 Fo 106 13. Creep Test with Variable Stress at 8500 F.. 107 14, Creep Tests at Constant Stress, Variable Temperature............... o o 108 v

(continued) LIST OF ILLUSTRATIONS Figure Page 15. Stress-Creep Rate Curves at 8500 F. and 1000~ F. for Coarse Grained (High Nitrogen and Fine Grained (Low Nitrogen) Steels o. 109 16. Microstructure of Silicon Killed Steel "C" and Silicon Aluminum Killed Steel "F", Airand Furnace Cooled from 21500 F, o o 110 17. Steels "C" and "F" After Air and Furnace Cooling from 16500 F, 0o e. 0 0 0 0 0 111 18. Typical Microstructures of High and Low Range of Carbon Content. o e..... 112 19o Microstructure of Materials Tested in the As-Rolled Condition e o o o 113 20. Microstructures of Steels "C" and "F" in Spheroidized Condition....... 114 21. Relation Between Creep Strength and Active Nitrogen for Vacuum Treated and Special Air Melted Steels 0 0 0 0 0 0 0 0 115 22. Heat 1019, Vacuum Melted Steel.. 0 0 0 116 23o Microstructure of Rimmed Steels.... 117 vi

A STUDY OF THE FUNDAMENTALS OF THE EFFECT OF DEOXIDATION ON THE CREEPI CHARACTERISTICS OF PLAIN CARBON STEEL by Clayton Dale DiRki^onn The relationship between the cteep strength and dissolved nitrogen in pla:i carbon steels was studied in order to explain the wide vari.ation in the reported creep strength for plain carbon steels with different deoxidation practices and heat tre astments, This relationship between nitrogen and creep rate was studied for a group of comirerci.l p laia carbon steels which were representative of the range of rimmed, ssilicon and aluminum deoxidation practices for plain carbon steelo A specially prepared group of vacuu.. melted and air melted steels were also tested, The creep rates of the steel s were determined for the 500 to 600 hour interi.. at 8500 F, ~d 15 000 psio The dissolved nitrogen was determined by the difference between the values.for thPe total, ni;trogein,and the nitrogen in the form of nitrides itr the stee o The atmount of dissolved nitroge n in the s.tels #a, changed. by varyi'ng the heat treatment or procedssing prior to the creep tests, o For the heat treated deoxidi zed steel the logarithm cf the creep r:.te s?.db,,oswed a very signki.:.icant linear decrease with an inciea.se na the amount of nitrogen in solid solution, The deotxda1tion pr'actice and hea.t treatment had no real effte:t cn the creep rate except for the effect of these pr:e:ep i".the nitrogen in solid solution in the steel, The logarithm of1 the t:'ie. rate of rimmed steels was found to decrease linea:rlty wilth an iecrease in ttal nitrogen; however9 the rate of increase in strength from nitrogen is not as great If"::.),r rimmed steel s as for Odeoxidized steelso Hot rolling9 stress relijeving and spheroidizing processes were found to cause a change in the amount of nitrogen in solid solution i. the stel., The creep rate of both silicon and aluminum kille.ed steels was related directly to the amouxt ofn nitrogen retained in solid solution after these treatments o

Removal of nitrogen by vacuum melting or vacuum annealing caused the creep rate:c. increase to the same degree as the removal of nitrogen by the precipitation of nitrides No consistent relation was found which related creep rate with austenitic grain si.ze, microstructure or hardness. The level of creep rate increased with manganese in the range of 0o00 to 0,8. percent for a, narrow range of low dissolved nitrogen o However9 increased dissolved nitrogen further decreased the creep rate at alli manga.nesee levels. The relatson between creep rate and dissolved nitrogen consistently explained the effect of deoxidation and prior processing on the creep strength, Silicon deoxidized steels have high creep streng.th in most conditions of heat-treatment because the nitrogen is dissolved and retained in solid solution by t:hese treatments, The nitrogen in fully deoxidized aluminum killed steels is precipitated as aluminum nitride for most heat treatments,, Heat treatm.ents which improve the tcreep strength of aluminum killed steels also increase the dissolved nitrogen. Rimmed steels are strengthened by nitrogen, but the absence of silicon or aluminum in the steel to interfere with the movement of the nitrogen atoms, reduces the strengthening effect of nitrogen in rimmed steels at temperatures above about 600~ F, Nitrogen increased the strength of steel only under conditions which involve plastic deformation, A modification of Cottrell's strain aging mechn...ism offers the best explanation for this behar ior,

I INTRODUCT ION A major source of variation in creep strength exists in plain carbon steels and low alloy steels as a result of variation in the degree and type of deoxidation and heattreatment. However, the mechanism for the effect of. deoxidation and heat-treatment is not fully understood. This study was conducted to test the proposal that the change in creep strength with deoxidation and heat-treatment resulted largely from the effect of these processes on the nitrogen in solid solution in the steel. oIt was further proposed that there would be a correlation between the dissolved nitrogen and the creep strength which would explain the gross changes in creep strength in spite' of minor variations which might result from differences in prior treatment. The object of this study is to test the extent of the correlation between nitrogen and the creep strength and to propose a mechanism to explain this correlationo The magnitude of the effect of the variation in high temperature strength with deoxidation varies considerably with stress and testing temperature, but has a very significant effect in the range of temperature and stress at which plain carbon steel is most widely used for high I

'Il.LU M R CLzo 5L

2 temperature service. For plain carbon steels in the normal range of composition, stresses from 6,000 psi to 18,000 psi have been reported for a creep rate of 1 x 10-5 percent per hour at 8000 Fo(1, 2). These figures are only typical values and the variation in strength is proportionally as great at other temperatures. Deoxidation and heat-treatment are not the only factors which control the creep strength, and even the type of melting furnace has been shown to influence the high temperature strength. The variation in strength from the interrelated factors of deoxidation and heat-treatment is much greater than the effects of other manufacturing variableso However, there is no one heattreatment or deoxidant which will universally yield high or low creep strength, and it will later be shown that neither factor is the fundamental cause of the large variation in creep strengtho Creep strength is only one of many variables or properties of steels which are affected by deoxidation practice and heat-treatment. The type and degree of deoxidation are known to cause a variation in the austenite grain size of most steels. Differences in the response to heat-treatment and in the final properties of the steels in many cases seem to parallel the grain size variation. Rimmed and silicon-killed steels have low coarsening temperatures and, as a result, exhibit fairly large austenitic grain sizes when heated to normal austenitizing temperatures. Fully killed aluminum-deoxidized steels have

3 high coarsening temperatures and a fine austenitic grain size is obtained with normal heat-treatment operations In addition, the microstructure and particularly the carbide distribution differ for steels with various deoxidation and heat-treatment histories. The simultaneous variation of carbide distribution, austenitic grain size, and creep strength has led to the belief that differences in the creep strength were caused by these factors Later studies determined that the variations between grain size and creep strength were only manifestations of the same basic changes which resulted from variations in deoxidation practice and heat-treatment (3) As a result of this work, a mechanism was proposed based on the solution and precipitation of an unknown product of deoxidation, presumably oxides However, this mechanism has not been verified experimentally and is, in fact, in conflict with the thermodynamics involved in the deoxidation process Beeghly (4) has developed a method of chemical analysis for aluminum nitride which has been applied very successfully in the study of mechanisms for the effect of deoxidation on many properties such as austenitic grain growth, graphitization, strain aging at room temperature, strain embrittlement and strain induced recrystallizationo Beeghly's method of analysis permits a new approach to the problem of the effect of deoxidation on creep strength of plain carbon steels and should provide information which

4 will lead to a better understanding of the mechanism for these effects. As a matter of fact, as this study was being completed, Bardgette (5) reported that there was a correlation between the creep strength of plain carbon steel and the aluminum precipitated as aluminum nitride, as shown by Beeghly's method. Since the total nitrogen remains constant for a given steel, it is apparent that there is also a correlation between the creep strength and the dissolved nitrogen as proposed in this study. Therefore, although Beeghly's method of analysis offers a key to the effect of deoxidation on the creep strength, the meaning of the correlation between the nitrogen or aluminum nitride and the creep strength must be studied before the basic mechanism can be evolved. A thorough study of the investigation of the mechanism for the effect of deoxidation on the creep strength of plain carbon steel is justified because the basic mechanism should have much broader application, At the present time, this factor of major importance in the creep and rupture strength of plain carbon steels, is not adequately explainedo It is quite probable that a factor of such importance in the high temperature strength of plain carbon steel could also be involved in the strengthening of many other high temperature materials. Equally significant is the fact that such gross effects from a

5 source which is not thoroughly understood could mask some of the effects of alloying elements, such as chromium, molybdenum, vanadium, carbon, etc. Therefore, before the effect of various alloying elements can be evaluated exactly, this source of uncontrolled variation must be understood. The purpose of this investigation is to determine whether there is a correlation between the creep strength and the change in the amount of precipitated or dissolved nitrogen, as shown by Beeghly's method. In addition, the processes which could influence the creep strength will be examined in order to explain the large variation in creep strength now attributed to deoxidationo

II REVIEW OF THE LITERATURE Effect of Deoxidation and Heat-Treatment The effect of manufacturing variables and particularly deoxidation variables on the creep strength of plain carbon steels is fairly complex and has received much attentiono The creep strength of mild steel is sensitive to prior history, and even the type of melting furnace has been found to influence the creep strength. Steels which have been produced by the Bessemer process or melted in an electric arc furnace generally exhibit higher creep resistance than steels made by either the acid or the basic open hearth methodo The magnitude of this effect, however, is dwarfed by the range of creep rates reported for steels with varying deoxidation practice, and this variable has received even more attention. The deoxidation variable is further complicated by the effect of subsequent heat-treatment and, in particular, austenitizing temperature. The variation of creep strength with heat-treatment, furthermore, has been shown to be dependent on the type and degree of deoxidation so that the important factor is believed to be deoxidation, Investigations into these effects have been quite thorough so that certain generalizations are possible 6

7 (3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) Silicon killed steels, for example, have high creep strength in almost all conditions of heat-treatment regardless of austenitizing temperatures or cooling rate (3, 8, 9) At temperatures above 8000 F., the strength of rimmed steel is much lower than that of silicon-killed steels, and again the strength is relatively independent of prior heat-treatment (7)o Also, the creep strength of the core of a rimmed steel is stronger than the rim material (8). Aluminum-killed steels which have been deoxidized with minor amounts of aluminum (up to about lo5 pounds per ton) generally possess creep strengths similar to that of silicon-killed steels. The amount of aluminum added per ton of steel is a poor criterion for determining the limit for high strength, and it has later been shown that there is some relation between the creep strength and the amount of soluble or metallic aluminum remaining in the steel after deoxidation (5, 6) Steels, which have been deoxidized with aluminum in amounts greater than two pounds per ton or with high residual aluminum, are generally considered to have low creep strength in most conditions of heat-treatment. It has been shown, however9 that these steels have creep strengths as high as that of silicon-killed steels after they have been austenitized above the coarsening temperature and air cooled in reasonably small sections. The literature also indicates that these steels possess high

8 creep strength in the as-rolled condition regardless of deoxidation practice (12, 13). Air cooling aluminum-killed steels from a fine grain austenite or furnace cooling from either the coarse or fine grain austenite results in a steel with a considerably reduced creep strength. The above generalities are not universal, however, as shown by the results obtained from a non-aging steel by Cross (9). The creep strength of this steel was not improved by coarsening and air cooling, and it was necessary to water quench from above the coarsening temperature in order to improve the creep strength of this very high aluminum —deoxidized steel. Although most of these studies were conducted on steels with 0,10 to 0o20 percent carbon, the effect of deoxidation is not limited to this carbon range or to plain carbon steels. Steels containing as high as 0,35 percent carbon also exhibit the same variation with deoxidation and heat-treatment (3) Deoxidation has also been shown to cause changes in creep strength of specially prepared steels with carbon in the range of 0o006 percent. Steels with up to 0.5 percent molybdenum also exhibit wide variations in creep strength as the result of deoxidation and heattreatment, and there is every indication that the effect of deoxidation is important in other low alloy steels (15, 16, 17, 18, 19). The reasons leading to the attempts to develop a mechanism based on the relation between austenitic grain

9 sizes and the resulting creep strength can easily be seen upon the examination of the effects of aluminum deoxidation on the grain size at various levels of added aluminum. Silicon-killed steels to which no aluminum has been added generally have a very coarse austenitic grain size at 1700~ Fo (20, 21, 22, 23, 24, 25) o Aluminum additions up to about lo5 pounds per ton cause little or no change in the grain size and the resulting creep strengths are as high as those of the silicon-killed steels Additions of 1,5 to 2 pounds per ton of aluminum as a deoxidant can result in either fine or coarse grained austenite, and steels with this range of added aluminum may even exhibit a mixed grain size in which both coarse and fine grains exist side by sideo Both high and low creep strength materials have been reported for steels with this range of aluminum addition. Steels deoxidized with greater than two pounds of aluminum per ton exhibit a fine grain size at 17000 F,, and the creep strength of these steels is generally low in most conditions of heat-treatment, These same steels, air cooled from temperatures which result in a coarse austenitic grain size, exhibit creep strength as high as the inherently coarse gra.ined materials, Grain size has been shown to affect the creep strength of many nonferrous materials, and the phenomenological conclusion is that the austenitic grain size affects the creep strength of the steel o A mechanism for the effect

10 of austenitic grain size on the creep strength of the steel could be evolved if it were not for the fact that the material is tested in the ferritic condition. To further complicate such a mechanism, the ferritic grain size is relatively independent of austenitic grain size and more dependent on cooling rate w:ith the slower cooling rate or high transformation temperatur es resulting in larger ferritic grain sizeso However) Cross (3, 8, 9, 10) concluded that neither grain size nor microstructlure were involved in the effect of deoxidation from the fol1owing results' A silicon killed steel whict was heated to either a low or high austenitizing temperature ard air r furnace cooled had high creep strengths which were the same within the accuracy which can be expected between two creep tests. An aluminumkilled steel with the added aluminum in the range of two pounds per ton had high creep strength only when cooled fairly rapidly from above the coarsening temperature Furnace cooling the coarse austenitic grains resulted in creep strengths as low as found for the fine grained condition of this material, T.e austenitic grain size of all four coarsened samples was iun:)formly large and, therefore, all of the steels should be strenigthened by the effect of austenitic grain size if such effect existed. The slowly cooled steels would have the largest ferritic grain size and would be expected to be strengthened if the ferritic

11 grain size was a factor This, of course, was not the case and, therefore, the effect of deoxidation could not be directly connected with the grain sizeo Of course, there are differences in microstructures between the air and furnace cooled samples. The coarsened air cooled samples had a Widmanstatten structure and the microstructure of the furnace cooled samples consisted of coarse lamellar pearlite, However, the creep strength of the silicon-killed steel with the coarse lamellar pearlite is as high as that of the same steel in other conditions of heat-treatment so that microstructure and/or carbide distribution could not be the major factor involved in the effect of deoxidation. This led to the conclusion that the similarities between austenitic grain size variations and creep strength are not related directly but are manifestations of some more fundamental mechanism involving the deoxidants or products of deoxidation, This conclusion was supported by comprehensive study of the effect of deoxidation and heat-treatment on the creep strength of plain carbon steels~ As a result of these studies, a mechanism was proposed based on the Dorn and Harder mechanism for grain growth of carbon steels which had developed at that time (23), According to Dorn's mechanism, steels which had been deoxidized with excess aluminum did not exhibit grain growth at normal austenitizing temperatures because of the formation of a film or

12 network of some product of deoxidation which inhibited grain growth. Increasing the austenitizing temperature resulted in the solution of this network in some regions which resulted in a duplex grain size at temperatures just below the coarsening temperature (24), The same nonuniform solution of the grain boundary network could be used to explain the mixed grain sizes found by the McQuaid-Ehn test for some steels which had been deoxidized with aluminum in the range of 1o5 to 2 pounds per ton, It has since been shown that continuous envelopes are not required for grain growth inhibition and grain growth can be prevented by fine precipitates distributed in the grain boundaries, With these modifications, the generally accepted mechanism for grain growth inhibition in aluminum-killed steels has been evolved from the basic concepts set forth by Dorn and Harder. An explanation of the effect of heat-treatment and deoxidation on the creep strength was developed from this hypothetical solubility with the assumption that the unknown, "X", was aluminum oxide (8), Cross proposed that the silica in silicon-killed steels was dissolved in the austenite at almost any austenitizing temperature and, on cooling, remain in supersaturated solution to precipitate as submicroscopic particles, These particles interfered with slip and greatly reduced the creep rateo Rimmed steels were weak according to this mechanism because they contained no silica to precipitate on coolingo The high creep strength of steels

13 which had been deoxidized with silicon and minor additions of aluminum was explained on the basis that there was still sufficient silica to produce high creep strength in most conditions of heat-treatment The low creep strength of aluminum-killed steels, which had a fine McQuaid-Ehn grain size, was attributed to the formation of "X" (aluminum oxide or aluminum silicates in silicon-aluminum-killed steels) in these steels, According to this mechanism, the aluminum oxides precipitated at the grain boundary of the newly formed austenite as the steel passed through the critical region With normal austenitizing procedure carried out below the coarsening temperature of the steel, the aluminum oxide remained in the grain boundaries and agglomerated into large precipitates which could not interfere with the creep process. As the austenitizing temperature of the steel was increased above the coarsening temperature, the aluminum oxide was dissolved in the austenite and, with fairly rapid cooling rates, remained in solid solution and precipitated as submicroscopic inclusions similar to the silicon-killed steels. In silicon-aluminum-killed steels, the oxygen was assumed to be present in the form of aluminum oxide or complex aluminum silicates which have the same solubility as the oxides in the aluminum-killed steels Thus, the change in strength with heat-treatment of these steels is explained on the same basis that was proposed for the

14 aluminum-killed steelso In rimmed steels, no oxides of deoxidants are present and the iron oxide in the steel was precipitated by any condition of heat-treatment in a form which does not improve the creep strength. Cross' comprehensive study of the effect of deoxidation did much to define the interrelated effects of heattreatment and deoxidation on the creep strength of ferritic materials, However, the mechanism presented to explain these effects was necessarily hypothetical, based on the solution and precipitation of oxides of aluminum or silicono Solution and precipitation of aluminum oxide in the solid state is hardly possible since this material is only slightly soluble in the liquid. In addition, oxygen in iron has been shown to have a solubility of less than 0.003 percent over the entire austenitic region (26) More recently, it has been shown that aluminum nitride is dissolved above the coarsening temperature to permit austenitic grain growth in aluminum-killed steels Therefore, aluminum nitride is the "X" of the Dorn-Harder mechanismo Since it is possible to measure accurately the solution and precipitation of this element with heat-treatment, it should now be possible to understand the more basic factors involved in the effect of deoxidation on the creep strength. Many recent investigations have confirmed Cross' experimental results and have added much to the knowledge of the effect of deoxidation However, the variation in

15 the microstructure and austenitic grain size is still considered to be an important factor by some authorso Effect of Other Manufacturing Variables Steels produced in an electric furnace were shown to have higher creep strength than basic open hearth steels in one of the first investigations on the effect of manufacturing variables on creep strength (7), More recently, several comprehensive studies were conducted on the combined effects of melting furnace and deoxidation variables (12, 149 19)o The effect of melting furnace variation, including acid open hearth, basic open hearth, electric furnace, and the Bessemer process, was studied for steels with varying degrees of deoxidation by relatively shorttime creep tests (12), The steels were tested in the asrolled condition and after normalizing at about 16500 F. Little significant difference was found in the creep strength of the steels in the as-rolled condition. The creep strength of the normalized steels, however9 depended very markedly on the type and degree of deoxidationo The majority of the tests in this study were run at 17,900 psi at 4500 CO (8420 F,. However, additional tests conducted at 4000 C. and 5507 C, (7520 F, and 10220 F,, respectively) indicated the same order and magnitude of creep strength differences existed at these test temperatures, Smith and Dulis (14) studied the effect of the type of melting furnace and deoxidation on the strength of

16 carbon steelso The steels were tested by creep and rupture tests at 8500 F. after normalizing at 16500 F, Bessemer steel was shown to be superior to open hearth steel in the rimmed or capped condition but no difference between melting furnaces was found for the creep strength of fully killed aluminum deoxidized steelso Fully killed aluminum. deoxidized steels and rimmed steels were found to have inferior creep and rupture strength at 8500 F, Steels which had been deoxidized with silicon and with minor additions of aluminum or titanium had superior strength at this temperature It was concluded that the reported wide variation in the creep strength of carbon steels was related to variation in deoxidation practice and other differences in prior history. Although no mechanism was presented to explain these effects, it was also concluded that, once the basic cause of these variations was recognized, it would be possible to produce materials which had more uniform high temperature strength, The effect of deoxidation on the creep strength of plain carbon plate and pipe steels in the as rolled or as rolled and stress relieved conditions was reported by Ro F, Miller (13). The results of this investigation were very interesting, particularly so since two of the steels tested in the present study were obtained from these commercial heats, No difference between the rupture strength of these steels was found at 10000 F even for

17 rupture times of 9,600 hours, and the pipe steels had similar creep strengths for the stresses considered, regardless of deoxidation, However, the plate steels had been stress relieved as is generally the practice for this grade of material, and the silicon-killed steels had superior creep strength at lower stress levelso The strain aging peak at 4000 Fo was not observed in the stressrelieved aluminum-killed steels but was very pronounced for the tensile tests of the as rolled and stress relieved silicon-killed steelso Beeghly's method of analysis indicated that stress relieving had caused precipitation of some of the aluminum nitride, and this precipitation was apparently responsible for the reduction in strain aging tendencies o In the discussion (27), it was pointed out that, in the range of stress of from 2,000 to 4,000 psi, both the silicon and aluminum-kill.ed st'eels showed erratic behavior in creep which resulted in alternately high and low creep strengths as the test progressed, In the final stages of the present investigation, a relation between the creep strength and the soluble aluminum or the aluminum precipitated as aluminum nitride was reported by Bardgette (5) The results indicated that steels containing more than 0,015 percent aluminum as metallic aluminum could be expected to exhibit poor creep strength. In addition, a more precise correlation was

18 developed between the aluminum present as aluminum nitride and the creep strength variation. It is now generally accepted that aluminum nitride is the grain growth inhibitor and therefore determines the austenitic grain size of the steel It was concluded that the loss in creep strength from aluminum deoxidation was due to the precipitation of aluminum nitride in those steels, which inhibited grain growth and caused a very fine austenitic grain size. It was further shown that, as a fine grained steel was heated above the coarsening temperature, the aluminum nitride was dissolved and the creep strength increased Although the solution and precipitation of nitrogen (and as a correlate aluminum nitride) is believed to be the basic factor involved in the variation of creep strength with deoxidation practice, a mechanism which depends on the growth of austenite for improved creep strength is not in agreement with the relations first developed by Cross' investigation, The status of the effect of deoxidation on creep strength is thus very complex today. Several popular theories have been and are still being presented which relate the change in creep strength by deoxidation to such factors as microstructural and austenitic grain size differences, In addition, a mechanism for strengthening, involving the critical dispersion of submicroscopic precipitates, has been proposed It has also been shown that the

19 tendency towards spheroidization is increased by strong deoxidation with aluminum (28), and the effect of this process on the creep strength should be examined. Another mechanism which has not been proposed to date could be developed based on the effect of deoxidation on the dissolved or soluble nitrogen and the subsequent effect of this state on the strain aging tendencies of the steels. This basic concept was examined carefully so that its effect on both the room and elevated temperature properties of the steel was understood, and a brief synopsis of Cottrell's theory for strain aging under creep conditions follows. Description of Strain Aging Phenomena A large number of diverse phenomena are peculiar to plain carbon steels and a few other commercial materials. Among these are the yield point, strain aging, serrated stress-strain curve, and the increase in ultimate tensile strength to a maximum value in the range of 4000 Fo These seemingly diverse phenomena are all closely related in that they occur primarily in materials which contain interstitial elements in solid solution, Cottrell has developed a theory which explains the function of the interstitial solid solution elements in these phenomena (29), and recently, he extended this mechanism to include the effect of strain aging on the creep strength at low and high temperatures (30)

20 The effect of interstitial elements on the room and slightly elevated temperature properties has been verified, in many cases, by experimental evidence. For steel, it has been found that nitrogen is, perhaps, the more important element involved in strain aging phenomena, and that the removal of nitrogen from solid solution, either by precipitation or by complete elimination, results in the disappearance or decrease of many strain aging phenomena, The following discussion of the mechanics of Cottrell's theory. is presented in terms of the special case of nitrogen dissolved in iron, Nitrogen dissolved in interstitial solid solution causes a very large strain in the crystal lattice. The lattice contains imperfections such as grain boundaries and dislocations and the nitrogen in solution around these imperfections not only results in less distortion of the crystal lattice but effectively reduces the strain of misfit at the edge of the dislocations The overall internal energy is substantially reduced if the nitrogen atoms are grouped as "atmospheres" around the dislocations in the metal o Since a dislocation which has an atmosphere of dissolved nitrogen surrounding it is at a lower energy state, the strain energy necessary to move the dislocation is increasedo The stress9 therefore, to cause plastic deformation is increased due to the increased energy required for plastic deformationo

21 At room temperature, playtic deformation of plain carbon steels takes place catastrophically, which gives rise to the upper agd lower yield point frequently observed for carbon steels. Steel behaves'elastically for all practical purposes up to the upper yield point after which there is a sudden drop in load. Subsequent plastic deformation proceeds in a diseontinuous manner at a lower and almost constant stresso This phenomenon is explained by the theory that the disloa.tioans which have been surrounded by atmospheres of nitrogen atoms, cannot be moved until a critical stress is reached, Once this stress has been reached9 a few dislocations break free from the atmospheres of nitrogen and move through the crystal lattice~ As these dislocations pass other "trapped" dislocations, they impart sufficient energy to remove the "trapped" dislocations from the atmospheres of nitrogen, This resaLlta in a chain reaction which permits catastrophic plastic deformation at a much lower stress than originally required, Another phenomenon involved in strain aging is the return of the yield point, A steel which has been deformed past the yield point and tested immediately will show a simple stress-strain curve with no yield pointo The same steel, which has been aged for several days at room temperature or for a shorter time at a slightly higher temperature, will again exhibit a yield point, but at a higher level than previously observed The time and

22 temperature involved for the return of the yield point is related to the diffusion rate of the nitrogen atoms and is explained as follows. A steel which has been plastically deformed beyond its yield point contains a great number of dislocations which are no longer surrounded by atmospheres of nitrogen The nitrogen diffuses slowly at room temperature and more rapidly at slightly elevated temperatures and the dislocations again accumulate atmospheres of nitrogen by diffusion. The steel is then in its original state except that the dislocations are now interacting due to plastic deformation, and strengthening has occurredo Therefore, a yield point can again be observed, but at a higher critical stress level. Cottrell's theory has been extended to include the strain aging peak which is frequently observed for plain carbon steels. As the test temperatures increased for tensile tests, the ultimate tensile strength increases to a maximum at a temperature of about 400~ F, Above this temperature, the tensile strength again decreases. Tensile tests which are conducted slightly below the temperature for the maximum ultimate tensile strength exhibit a serrated stress-strain curve. At this temperature, plastic deformation and strain aging occur at approximately the same rate, The serrated stress-strain curve is a result of alternate cycles of yielding and strain aging which occur progressively at higher and higher stresses due to the

23 plastic deformation which has occurred. At temperatures above the strain aging peak, no serrations in the stressstrain curve are noted. Therefore, it has been proposed that, at the strain aging peak, the atmospheres of nitrogen are reformed at the same rate that dislocations are able to escape from an existing atmosphere. A dislocation which has attained sufficient energy to break away from its atmosphere moves slower and slower as deformation proceeds and gathers another atmosphere of nitrogen atoms until it is again trapped by a combination of its interaction with other dislocations and the nitrogen atmosphere which it has collected. Effect of Strain Aging on Creep Phenomena At slightly elevated temperatures and stresses at which creep becomes an important phenomenon, the effect of interstitial elements in solid solution is slightly more complex. The effect of the interstitial elements can be best explained by dividing the creep region roughly into two temperature ranges, a low temperature range and a high temperature range. At low temperatures, the dislocations are surrounded by atmospheres of nitrogen, but at these temperatures a lower energy is necessary for the movement of dislocations and many dislocations have sufficient energy to move. At low stress levels, the rate of deformation is slow enough so that deformation can occur by movements of slow dislocations. These dislocations retain their atmospheres

24 of nitrogen which diffuse with the dislocation at a fairly slow rate. As the stress level is increased, more and more dislocations have sufficient energy to break away from the atmospheres of nitrogen and become active and move more rapidly through the lattice. These "fast" dislocations eventually slow down due to interaction with other dislocations and again collect an atmosphere of nitrogen atoms. However, at rapid strain rates, little effect would be expected since a majority of the deformation would take place by the movement of fast dislocations. At more elevated temperatures, the picture is complicated by recovery which occurs during the creep tests; however, the relationship between slow and fast dislocations and the stress level should still exist, Due to the increase in temperature, the rate of movement of the slow dislocations is much more rapid. Interacting dislocations can be rendered ineffective by recovery at higher temperatures so that an additional condition is necessary to increase the effect of the interstitial elements. High temperature creep phenomena require an additional element which either causes precipitation of a second phase in minor quantities or creates the potential for precipitation of the interstitial element. Elements such as silicon or aluminum, which have a definite thermodynamic preference for the nitrogen atoms, serve in this capacity. Thus, a dislocation collects an

25 atmosphere not only of nitrogen but of aluminum or silicon and must move at a slower rate, roughly dependent on the diffusion rate of the nitride forming element The dislocations are further hindered by the small precipitates in the lattice. Additional energy is necessary for recovery due to the other odd sized elements in solution so that interactions between dislocations are effective at the higher temperatures There is good experimental evidence to show that strain aging at room temperature and the effect of strain aging for tensile tests at slightly elevated temperatures is directly related to nitrogen in solid solution. Previous investigations have shown that the effect of deoxidation and heat-treatment on phenomena normally connected with strain aging are directly related to the effect of the heattreatment on the dissolved nitrogen (31) o There remains, however, the problem of checking the validity of the strain aging mechanism at the temperatures and stresses involved in the present investigation in order to logically explain the relationship between the dissolved nitrogen and creep rate o

III EXPERIMENTAL DESIGN, MATERIALS AND PROCEDURE Experimental Design It has been proposed that there is a correlation between the creep strength of plain carbon steel and the amount of nitrogen in solid solution in the steel. To show that this correlation exists, a group of commercially and specially produced steels representing various deoxidation practices were obtained and the dissolved nitrogen and the creep rates were determined for these materials. The basic experimental design evolves into four groups of tests with the following objectives: 1. To show that the creep strength of commercially prepared plain carbon steels varies with deoxidation and heat-treatment as a result of the amount of nitrogen retained in solid solution by these treatmentso 2, To show that the effect of deoxidation and heat-treatment on the creep strength is not significant for steels produced with low nitrogen contents. 3. To examine the previously proposed mechanisms for the effect of deoxidation on the creep strength of steel in light of the correlation between the nitrogen and the creep strength and accept or discard these mechanisms on the basis of critically designed experiments. 4. To propose and test a mechanism which will explain the relations between nitrogen and creep strength. 26

27 With these objectives in mind, a group of commercial steels which represented rimmed, silicon killed, siliconaluminum killed, and aluminum killed deoxidation practices, were assembled for testing. The steels were austenitized at 16500 Fo and 21500 Fo The higher temperature was chosen in order to dissolve the aluminum nitride for the aluminum deoxidized steels. The steels were then furnace cooled and air cooled to vary the amount of nitrogen retained in solid solution at room temperatureo A group of steels, which contained low nitrogen, were prepared by vacuum melting or vacuum extraction. These materials were tested to show that steels, which contained low nitrogen, would have low creep strength regardless of deoxidation practice or heat-treatment, The commercial materials were subjected to several other processes to investigate the effects of these processes on the creep strength A silicon and an aluminum killed steel were tested in the a-rolled condition, and the aluminum killed steel was also tested in the stress relieved condition. Two samples of a silicon-aluminum killed steel were rolled from 17500 F, and 21500 F, respectively in order to obtain materials in the as-rolled condition which contained high and low dissolved nitrogen. In addition, a silicon and an aluminum killed steel were subjected to a spheroidizing treatment in order to determine whether the effect of spheroidization on the creep strength was related to the dissolved nitrogen~

28 Creep tests were run for all these materials at 8500 F. and 15 000 psi. The total and insoluble nitrogen were determined for the materials in order to obtain a value for the dissolved nitrogen by difference. Additional creep tests and high temperature tensile tests were conducted on some of the materials to better evaluate the mechanism for the effect of nitrogen on the creep strength. Both the materials and testing procedures are discussed in greater detail in the following section, Since statistical analysis was used as a tool for evaluating the significance of the correlations developed, the techniques used are also discussed briefly Materials Steels The effect of deoxidation had been found to be significant at many temperatures for commercial materials. Since the aim of this study was to explain this variation in strength, commercial plain carbon steels were chosen to develop the basic correlation between nitrogen and creep strength. The results of previous investigators in the field of deoxidation effects have shown that the major effects of aluminum deoxidation occur in steels which have been deoxidized with approximately two pounds of aluminum per ton of steel, and that steels deoxidized with one pound of aluminum per ton normally behave as

29 coarse-grained silicon-killed steels. For this reason, a group of commercially prepared steels were assembled with two levels of silicon deoxidation and three levels of aluminum deoxidation as shown in Figure 1. Fortunately, because of the great amount of research being conducted into the effects of deoxidation, it was possible to assemble such a group of steels for which the deoxidation practice and rolling practice had been carefully recorded. The chemical compositions of these steels are shown in Table I. Steels "A", "D" "E", and "J" were obtained from the Research Laboratories of the United States Steel Corporation through the efforts of R. Lo Rickett and bear the same designation as that used in an investigation of the effects of deoxidation and heat-treatment on strain aging (31)o Steels "C" and "F" were obtained from the Chemical and Petroleum Panel of The ASTM-ASME Joint Committee on the Effect of Temperature on the Properties of Metals with the help of R, F. Miller of that committee. These steels were in the form of one inch plate in the hot-rolled condition and were among the plate steels tested for Project No. 8 of the Chemical and Petroleum Panel. Steels "C" and "F" were designated CG2 and FG1, respectively, in this previous study (13) Steel "H" was received in the form of 1-1/4 inch plate from the Allis-Chalmers Manufacturing Company and is a rimmed steel with very high nitrogen content, The details

30 of manufacturing process for this steel are not known; however, the nitrogen content of this steel is as high as that normally found for steel produced by the Bessemer process. Steel "B" was obtained from the United States Steel Corporation in the form of 2-1/4 inch diameter bar stock. This bar stock was subsequently hot-rolled in a closed pass rolling mill into 7/8 inch square bar stock, The rolling temperature was approximately 2150~ F., and at least two reheats were necessary between rolling passes. Following the final reheat, the bar was rolled from 1-1/4 inch square to 7/8 inch square. Following the basic premise of this study, it was proposed that the primary reason for the reported high creep strength of all steels in the as-rolled condition, regardless of deoxidation, was that these steels were actually rolled above their coarsening temperature and cooled rapidly enough to produce high dissolved nitrogen regardless of deoxidation. In order to test this hypothesis, one bar of steel "B" was heated only to 17500 F. during the final pass in order to produce a steel in the as-rolled condition with low dissolved nitrogen. Since 17500 F, was below the coarsening temperature of the steel, it would be expected that aluminum nitride would be precipitated at the rolling temperature and the steel would contain low dissolved nitrogen in the as-rolled condition,

31 Several special heats of steel were produced by vacuum melting through the courtesy of the Ford Motor Company and the Engineering Research Institute of the University of Michigan. The chemical analyses of these heats are given in Table Ilo The heats were melted in vacuum to remove both the nitrogen and the oxygen, and carbon dioxide was introduced into the chamber to reoxidize the steel without increasing the nitrogen content. In this way, it was hoped that the composition of the steels would be comparable to those of the commercially produced heats with the exception of the low nitrogen. For aluminumkilled steels of this type, only minor changes of strength from heat-treatment would be expected if nitrogen were primarily responsible for the change in strength. Steel 1019 was produced at the Metallurgical Research Laboratories of, the Ford Motor Company and had a very low manganese content evidently due to high loss from vaporization Heats 1016 and 1018 were produced at the University of Michigan, and similar difficulty with high vaporization loss produced very low manganese in both heats. The difficulties encountered in controlling the composition of relatively volatile elements in vacuum made it evident that it would be very difficult to exactly duplicate the analyses of the commercial heats by vacuum meltingo In addition, it was found that some residual aluminum was picked up from either the raw materials or the zirconia crucibleo This

32 made the composition even more difficult to control and made it impossible to produce a straight silicon-killed steel in vacuum without considerable experimentation Theref ore, it was decided that the analyses of the vacuum melted heats would be matched by air melted heats produced from similar starting materials. These two heats were melted in a fourteen pound induction furnace at the Allis-Chalmers Research Laboratories, and the chemical analyses are also shown in Table II for comparisono The steels were poured in a cast iron mold with a copper base plate and the assembly held approximately 1. pounds of steel including hot top Low nitrogen samples for silicon-killed steel "C" and rimmed steel "A" were produced by vacuum annealing, The samples were held for 100 hours at 2000~ Fo in a vacuum of less than one micron, The calculation for the removal of nitrogen by diffusion is shown in Calculation 19 in the Appendix, and indicates that all of the nitrogen should be removed by this treatment o The samples were then annealed in vacuum prior to final heat-treatment for testing in order to refine the grain size and minimize the effect of the prior treatment The vacuum melted heats were rolled in the closed pass rolling mill with a procedure similar to that used for steel "B"o The air melted heats were forged from 21500 Fo by a hammer forge into approximately one inch bar stock,

33 Heat-Treatment The fine grain heat-treatment temperature of 1650~ F. was chosen because it was below the coarsening temperature of those steels which were fine grained as shown by the McQuaid-Ehn grain size. Steels austenitized at this temperature should, therefore, have a variation in dissolved nitrogen depending on deoxidation. The higher austenitizing temperature of 21500 Fo was chosen because it was above the coarsening or solution temperature for practically all of the steels tested. Therefore, it would be expected that the dissolved nitrogen in this condition of heat-treatment wouldvary with both deoxidation and cooling rate. The two temperatures should give the extreme conditions of dissolved nitrogen and creep strength as well as intermediate nitrogen contents depending on cooling rate. The cooling rates were arbitrarily limited tc air and furnace cooling since it has been shown by previous studies that the creep rates of the fine grained steels can be changed significantly by these two cooling rates. The bars were heat-treated in sizes no larger than one inch square so that air-cooling was fairly rapid. The cooling rate for furnace cooling varied between approximately 300 degrees per hour to 400 degrees per hour depending on the size of the furnace,

34 Procedure Creep Tests Creep tests were run both at the University of Michigan and at the Allis-Chalmers Research Laboratories and check tests indicate that there is good agreement between the tests run at the two laboratories. The creep tests at the University of Michigan were conducted on several units of slightly differing design, however, both temperature and load were carefully controlled in all cases. The creep rates were measured with the modified Martin's type extensometer with a sensitivity of 4 x 10-6 inches per incho The creep units at the Allis-Chalmers Manufacturing Company are of the overhead beam type with a 15 to 1 beam factor. The furnaces are wound so that each of the five zones in the furnace can be shunted separately in order to obtain good temperature distribution. The creep rates were measured by measuring the relative movement between Knoop hardness indentations in the platinum strip and channel which are fastened to the shoulders of the specimen. These distances are measured.by a 100 power microscope with a filiar eyepiece, and the sensitivity of this method was about 2 x 10-5 inches per inch. In both laboratories, the temperature of the bar was controlled within two degrees of the control temperature and the temperature of the bottom and top of the bar was within four degrees of the temperature of the center couple

35 The majority of the tests were conducted at 8500 F. and at a stress of 15,000 psi. A temperature of 8500 F. was chosen since many very comprehensive studies had been conducted at this temperature, and it was reasoned that the previously reported values might be of some aid in understanding the effect of varying stress on the creep strength. The stress of 15,000 psi was chosen because it resulted in a reasonable range of creep rates for the large variation in strength anticipated In some of the first tests, a stress of 10,000 psi was chosen; however, this stress was found to be too low to result in a measurable creep rate for some materials. The stress for these tests was increased to 15,000 psi after the tests had been in progress for approximately 700 hourso This may have resulted in a slight shift in the level of the creep rate, but the results of these tests are still of value. The creep tests were continued for at least 600 hours, and the creep rates were measured for the time interval from 500 to 600 hourso Since some of the tests were continued for times up to 1,000 hours, it was necessary to arbitrarily choose this shorter time in order to permit a comparative evaluation of the various materials For the majority of the tests, the creep rates so measured were reasonably close to the eventual second-stage creep rate,

36 High Temperature Tensile Tests The stress-strain and tensile tests were conducted on a Baldwin-Southwark hydraulic tensile machine equipped with a furnace for elevated temperature tests. An O,S. Peters Model TS-M Dual extensometer modified for elevated temperature use was connected to the stress-strain recorder and automatically produced a stress-strain curve at all temperatures. The speed of deformation was controlled by hand and was kept constant by means of a pace setter which indicated changes in the rate of head movement. The stress-strain curves were continued slightly beyond the ultimate tensile strength of the material, and the final breaking stress based on the area of the broken section was also recorded. The results of the tensile tests were then used to calculate the true stress and true strain curves from the following equations (32): Ao 1 et = log - = 2 log -.. 0o......000..0. OIII A I1 in which: et is the true strain A and Ao are the true area and original area, respectively 1 and Io are the true length and original length, respectively From these equations, the true strain and the area at any point can be calculated if the change in length of the

37 sample is known. The true stress at any point can be calculated by dividing the load by the true area of the specimen McGregor and others (32, 33) have shown that the plot of true stress and true strain between the ultimate tensile strength and the breaking strength of a specimen is approximately a straight line and, therefore, the complete true stress-true strain diagram can be constructed by drawing a straight line between these two pointso A detailed calculation of one true stress-true strain test is included in Calculation 2 in the Appendix, The true stress-true strain method results in a better indication of changes in a material which affect both the strength and ductility. An isostrain diagram, in which the stress for a given strain at the various test temperatures is plotted as one curve, was then constructed for several values of true strain to give a fairly complete picture of the effect of temperature on the stress-strain properties of the steels with various heat-treatments o Nitrogen and Alamijnum Nitride Analyses The method of analysis for nitrogen and for nitrogen combined as aluminum nitride or other nitrides in the steel has been developed largely through the efforts of H. F. Beeghly (4, 34). In determination of the total nitrogen, digestion in a one to one sulphuric acid solution containing

38 five percent phosphoric acid was found necessary for complete recovery of nitrogen, The solution and separation of the nitrides was accomplished by th.e methyl acetate-bromine method described by Beeghlyo The nitrogen was distilled as ammonia in a micro-Kjeldahl apparatus and collected with 50 cc of distillate in a Nessler tube. The color developed by the Nessler's reagent was measured on a Beckman spectrophotometer at 410 wave length with a slit width of 0.15 mmo The Nessler's solution was calibrated with a known solution of ammonium chloride. Duplicate analyses were conducted on all aluminum nitride analyses f-or basic correlationo The samples were coded to remove the possibility of operator bias and were randomly arranged so that duplicate samples were not necessarily run on the same day, Since the separations were run in groups of four this served as a cross check between days, The accuracy of the nitrogen determinations was +0:0005 percent nitrogen. Statistical Methods Statistical methods were used to determine the significance of the correlation between the logarithm of creep rate and dissolved nitrogen and also used to determine the limits of accuracy of the analytical method. The Method of Least Square's described by Ezekiel (35) was used to determine the regression equation and the significance of this equation. The accuracy of the aluminum nitride

39 determinations was calculated using the difference between the duplicate samples and the "t" or difference method of statistical analysis (36) The analysis of variance method was used to calculate the significance of heat-treatment on the soluble or metallic aluminum in three aluminum deoxidized steels

IV RESULTS Effect of Deoxidation and Heat-Treatment Basic Correlation A group of steels were obtained with the zero and 0.20 percent silicon levels and three levels of aluminum deoxidation as previously mentioned.. A graph showing how these materials fit into the basic experimental design and the letter designation for both the deoxidized and rimmed steels is shown in Figure 1o The metallic aluminum and silicon contents of each of the steels are given in this figure. The total nitrogen content of these heats of steel was between 0~004 and 0.005 percent, The two austenitizing temperatures and cooling rates included in the basic correlation experiment, along with the coding used throughout to designate these treatments, are also shown in Figure 1, The austenitic grain size was determined at 16500 F. and 21500 F, for all of the test materials, and these results are listed in Table IIIo The dissolved nitrogen and creep rate at 500 to 600 hours at 8500 F, and 15,000 psi were determined for the four conditions of heat-treatment for all of the commercial heats, 40

41 Deoxidized Steels~ —The results for the deoxidized steels are listed in Table IV. The logarithm of the creep rate has been plotted against the amount of nitrogen in solid solution in Figure 2, There is a linear decrease in the logarithm of the creep rate with increased nitrogen, The correlation was examined statistically by the method of least squares, and the following relation was determined (Appendix, Calculation 3) log e — 2 083 -593 x % Nitrogen..........IV-1 The corrected correlation coefficient for this equation is 0.923, The metallic aluminum and insoluble alumina were determined for the fou.r conditions of heat-treatment of steels "B", "E" and "F" which had been heavily deoxidized with aluminum. The results are shown in Table V, and the results of an analysis of variance determination of the significance of the difference between heat-treatment is also shown in this tableo There is no significant variation in aluminum or alumina with heat-treatment for these aluminum killed steelso Rimmed Steels — Previous authors have shown that the creep behavior of rimmed steels differs from that of deoxidized steels, and the results for the rimmed steels have been analyzed separately, The pertinent details and coding for the rimmed steel experiment are also given in Figure 1o Rimmed steels with two levels of nitrogen were

42 obtained from commercial practice and are designated as steels "A" and "H", An additional sample of steel "A" was heated in vacuum at 20006 F for approximately 100 hours to remove the nitrogen by diffusion This sample was subsequently normalized twice to minimize the effects of prior treatment. Since the literature' indicates that all of the nitrogen should be soluble at testing temperature in rimmed steels, only total nitrogen was determined for the two commercial rimmed steels and for steel "A" after the vacuum annealing treatment. The nitride nitrogen was determined for the fine air cooled sample of steel "A" after testing at 8500 Fo and only 0,0004 percent nitrogen was found in this formo The total nitrogen and the results of the creep test for the rimmed materials are listed in Table VI, The logarithm of the creep rates for these tests are plotted against the total nitrogen in Figure 3, and rimmed steels also exhibit a decrease in creep rate with increased nitrogen The best straight line for the logarithm of the creep rate versus percent nitrogen was also determined for the rimmed steels with the assumption that the maximum effect occurred at 0.0063 percent nitrogen (Appendix, Calculation 4). This assumption will be verified later. The equation for this relation to 0,0063 percent nitrogen iso log 6 = -2,175 -333 x % nitrogen,._0_.....IV-2

43 The correlation coefficient for this straight line is 0~976. The correlation curve for the deoxidized steels from Figure 2 is shown in Figure 3 for comparison, The variation in creep strength with nitrogen is not as great for the rimmed steels as for the deoxidized steels; however, the curves apparently intersect at zero nitrogen. Effect of Other Manufacturing Variables Effect of Rolling The creep strength of deoxidized steels in the asrolled condition has previously been reported to be high regardless of deoxidation, It is proposed that the reason for this lack of variation in strength with deoxidation was the high temperature of rolling, which resulted in high dissolved nitrogen, regardless of deoxidation practice. Steels "C" and "F" were tested in the as-rolled condition, and steel "F" was tested after a stress relief of 11500 F. for two hours Samples from steel "B" were also tested in the as-rolled condition with rolling temperatures of 17500 F. and 2150~ F. Creep tests were started at 10,000 psi at 8500 F., and after approximately 700 hours at this stress the bars were retested at 15,000 psi. The results of the creep tests and analyses for dissolved nitrogen are shown in Table VII for the steels in the as-rolled condition. Stress relieving of steel "F"

44 caused a decrease in the dissolved nitrogen and a corresponding increase in creep rate. A decrease in creep rate with increased active nitrogen was also found for steels in the as-rolled condition, The relationship between the logarithm of the creep rate and the percent dissolved nitrogen for these steels in the as-rolled condition is plotted in Figure 4. The correlation curve for the heat-treated deoxidixed steels is also shown in this figure. The results for steel "B" are displaced from the correlation curve by an amount that is greater than can be explained by scatter. However9 these results were obtained after prior testing at 10,000 psi and 8500 F. and it is probable that this displacement was caused by the prior testing. Effect of Spheroidization The creep strength in low carbon9 plain carbon steels was correlated primarily with the dissolved nitrogen and was not significantly related to the microstructure. This led to the speculation that the loss of strength due to spheroidization might also be caused by the precipitation of the nitrogen in the steelo Therefore, the fine grained air cooled condition of the silicon-killed steel "C" and the coarsened and fine grained air cooled conditions for steel "F" were spheroidized by holding for 100 hours at 12500 F This treatment was sufficient to spheroidize both materials.

45 The dissolved nitrogen content for the three samples before and after spheroidization are given in Table VIII along with the resulting creep rates. The variation in strength between the spheroidized and as-treated materials is shown in Figure 5o The spheroidization treatment greatly increased the creep rate of the steels which originally contained high dissolved nitrogeno The fine grained condition of the aluminum killed steel exhibited little change in creep rateo Since the nitrogen was already precipitated by the original heat-treatment for the fine grained material, the change in dissolved nitrogen with spheroidization was also very small. The creep rates are plotted against the dissolved nitrogen for both the as heat-treated and spheroidized condition in Figure 60 The correlation curve for the heat-treated deoxidized steels is also given in this figure for comparison. The creep rates for the spheroidized steels are very close to the rate predicted by the active nitrogen in these materials. Effect of Other Variables Low Nitrogen Steels As it became evident that the amount of nitrogen in solid solution could be correlated with the creep strength of plain carbon steels, a question arose. What would be expected from a group of steels which had low nitrogen as a result of vacuum melting or some similar process? It

46 was proposed that silicon-killed steels with low nitrogen should have reduced creep strength due to the removal of the nitrogen. Also, the variation in creep strength with austenitizing temperature for aluminum-killed steels should be greatly reduced or possibly eliminated for vacuum melted steels. The samples were prepared both by vacuum melting and vacuum annealing as previously discussed. The nitrogen was removed from two samples of steel "C" by annealing in vacuum for 100 hours at 2000~ F. As in the case of the vacuum annealed, rimmed steel, a double normalizing treatment wa given to minimize the effects of long time at high temperature. The microstructures were of the Widmanstatten type normally expected for an air cooled silicon-killed steel, Figure 7. The resulting creep strength and total nitrogen for the steels are listed in Table IX, and the logarithm of the creep rate is plotted against the total nitrogen in Figure 8. The results for steel "C" in the high nitrogen condition and the correlation curve for the deoxidized steels are also shown in this figure. The creep rate for the vacuum extracted sample of steel "C" is in the range predicted by the nitrogen content, The vacuum melted steels and the corresponding air melted steels with similar analyses were also tested at 15,000 psi at 8500 F. Because of the extreme-weakness ofthese vacuum melted steels, some of the tests ruptured in times as short as 100 to 500 hours, and in these cases the

47 creep rates, Table IX, are the minimum rates observed for the test. Since all of these steels are effectively aluminum killed because of the high residual metallic aluminum, these steels were tested in the fine and coarsened air cooled condition. The dissolved nitrogen and the creep rates for these steels are given in Table IX. Steel 1019 is compared with its equivalent low manganese air melted steel 1379 in Figure 9 for both the coarsened and fine grained air cooled condition. Steels 1018 and 1442 are similarly compared in this figure and represent a higher manganese level, The creep rates for the vacuum melted samples were high regardless of austenitizing temperature. The creep rates of the fine grained condition of the air melted steels were equally high. However, the coarsened condition for comparable air melted steels exhibited much lower creep rates as a result of the increase in dissolved nitrogen. Manganese Effect The results for the vacuum melted steels made it evident that some additional variable was responsible for a change in the level of the creep strengtho However, this variable is not directly connected with the deoxidation effect as shown by the lack of variation between creep strength and heat-treatment for the low nitrogen aluminumbearing steels. The major change in composition for the vacuum melted steels was the low manganese content.

48 The creep strength for most of the steels had been determined in some condition of heat treatment which yielded a low dissolved nitrogen; therefore, the effect of manganese on the creep strength was checked by plotting the creep rate versus the manganese content for those steels which contained between 0o001 and 0.002 percent dissolved nitrogen, Figure 10 The results indicate that manganese has an important influence by increasing the level of strength of carbon steel. In order to test the significance of this relation, the correlation between manganese and creep rate for these steels was determined by the least squares method. The details of this calculation are given in Calculation 5 in the Appendix and the results are as follows~ log 6 = -0.772 -33o8 x % Manganese oo..o....IV-3 The corrected correlation coefficient was 0.901 which is very significant Effect of Other Variables on the Basic Correlation Curve Since there was a co relation between the manganese content and the level of strength for steels at a given nitrogen content, it was thought that possibly some of the variations from the correlation curve could be accounted for by manganese content, deoxidation or heat-treatment variables. The effects of these variables were tested by an analysis of variance of the deviation from the correlation curve for the heat-treated deoxidized steels. These

49 calculations are summarized in Calculations 6 and 7 in the Appendix, and the results are as follows, The various heat-treatments employed caused no significant variation. There was no significant variation from the curve for any one individual steel or any group of steelso There was, however, an indication that those steels, which were air cooled from either the fine or coarse grained austenite were on the high strength side of the correlation curve, and those which were furnace cooled were on the low strength side. This relationship was not statistically significant The as-received hardnesses for the heat-treated deoxidized steels and for the steels used in the other tests are listed in Tables X and XI, respectively, Hardnesses were obtained on the as-treated materials and on the reduced section and shoulder section after testing. The hardness in the reduced section represents the hardness after testing for the strained material and those in the shoulder section represent the unstrained9 heated conditiono The deviations from the creep rate correlation curve of Figure 2 were tested statistically for possible effects of hardness9 Appendix, Calculation 8, and the effect of hardness was not significant The lack of variation between steels indicated that manganese was not responsible for the scatter in the correlation curve for the heat-treated deoxidized materials.

50 The manganese content of these steels varied from 042 to 0.82 percent which is within the expected range for commercial plain carbon steels. The effect of manganese determined for steels with 0o005 to 0~82 percent manganese at low active nitrogen levels was significant, Perhaps this paradox for the effect of manganese could be clarified by a study of steels with a wider range of manganese variation at several levels of dissolved nitrogen. It is also possible that cooling rate and hardness influence the creep strength over a wider range of variation at a given nitrogen level. Mechanism Tests A group of tests were devised to check the possible mechanisms for the effect of nitrogen on the creep strength. Cross (9) has previously shown that neither grain size nor microstructure could be related to the change in creep strength, and this conclusion is substantiated by the present study. A precipitation strengthening mechanism based on the solution and precipitation of nitrogen with varying deoxidation and heat-treatment should be considered, Cross (9) proposed such a mechanism based on the hypothetical solution and precipitation of "X"o It is now obvious that "X" is aluminum nitride. In addition, a strain aging mechanism was also examined since nitrogen is known to be important to strain aging of plain carbon steel In order to determine if a precipitation strengthening phenomenon was responsible for the increase in creep

51 strength, the combined nitrogen, after creep testing9 was determined for several of the steels tested, Table XIIo The combined nitrogen, determined by Beeghly's method for the as-heat-treated material9 and the change in combined nitrogen during test are also listed in this table. A slight increase in the amount of aluminum nitride has occurred for the deoxidized steels. Isostrain curves were determined for samples of steels "C" and "F" which had been air cooled from both 21500 Fo and 16500 F. These tests were conducted by the method outlined by MacGregor (32)9 and the results for the individual steels in the two conditions of heat-treatment are shown for true strains of 0,05 0o10, 0.509 and 1o00 inches per inch9 Figure 11, The results for all four materials have been compiled in Figure 11 for true strains of 0,05 and 0o5 inches per inch. For the materials which contained high dissolved nitrogen there is an increase in strength with increasing temperature which reaches a maximum at 4000 F, This behavior gives rise to a "strain aging" peak at about 4000 Fo Steel "F" in the fine grained air cooled condition exhibited no "strain aging" peak, As previously mentioned, steels tested at 10000 F, at very low stresses showed creep rates which were alternately high and low as the tests progressed. These tests were conducted for the stress relieved condition of steels "F" and "C" for times up to 8,000 hours The resulting

52 creep rates are plotted against time for these materials in Figure 12o Steel "F" at two stress levels showed a very low initial creep rate and a subsequent increase in creep rate after about 1,000 to 19500 hours of testing. The creep rates reached a maximum and again decreased. Steel "F" at 2,000 psi went through another maximum at about 5,500 hours, Steel "C" which was a silicon-killed material, showed a normal creep rate-time curve out to about 2 500 hours after which a very rapid increase in creep rate was experienced. The creep rate for this test reached a maximum at about 4,000 hours and again decreased for longer times, Although the tests were not run beyond 8,000 hours, steel "C" apparently would have gone through a series of alternately high and low creep rates and eventually reach a lower creep rate, as did the other two samples, A series of tests were run at constant temperature with varying stress, Figure 13, and varying temperature with constant stress9 Figure 14, in order to better understand the effect of stress and temperature on the mechanism involved, Stel "C" air cooled from 16500 F0, and two samples of steel "F", air and furnace cooled from 21500 F. were tested at 15,000 psi, with varying temperature, The resulting creep curves are shown in Figure 14, These tests were conducted at a constant stress in an attempt to determine whether there was a difference in activation energy for steels with varying dissolved nitrogen At 9000 F 9

53 a phenomenon similar to that observed at 10000 F, was foundo The creep rate at the start of the 9000 Fo test was reasonable for the steel at that temperature; however, the creep rate increased after several hundred hours of testing had taken place, This phenomenon was observed, however, only for steels "FCA" and "CFA" which originally had high dissolved nitrogen, Steel "FCF", which had low dissolved nitrogen9 showed a normal creep curve at 9000 Fo The resulting creep curves were not entirely satisfactory for an accurate determination of the activation energy, The results do indicate that the oscillating creep rate behavior observed at 1000~ F, does not occur for steels with low dissolved nitrogen A creep curve for steel "F" in the coarsened air coole cooecondition was obtained with varying stress at 8500 F, and is shown in Figure 13, This steel was tested in a stress range of from 17,000 to 22,500 psi, As the stress was increased from 20 000 psi to 21,875 psi, little change in creep rate was observed, However9 at 22,250 psi. the rate increased very rapidly, The reduction of stress to 20,375 psi resulted in a large change in creep rate to the range of that originally observed, An additional decrease in stress to 18,625 psi caused no additional change in creep rates however, the final stress of 17,000 psi caused the creep rate to decrease further. These tests indicate that there is a critical stress above which the creep rate increases very rapidly,

54 Additional creep data were obtained from the literature (3, 8, 9, 10, 13, 14) for steels which should contain high and low dissolved nitrogen, and the creep rate-stress curve is shown in Figure 15 for 8500 F. and 10000 Fo At both temperatures, a maximum effect is apparent for the difference in strength between low and high dissolved nitrogen At very high creep rates' little or no effect is observed for varying nitrogen, and at low creep rates the curves at 8500 Fo again seem to converge. These results are in good agreement with the rupture tests at 1000~ Fo for these materials, which showed no difference in rupture strength between steels with low and high dissolved nitrogen at high stresses at 1000~ F. (13). The creep curve at 8500 F, in Figure 15 for the coarsened or high nitrogen condition shows a change in slope at about 22,000 psi with stresses above this value causing a much larger increase in creep rate. This confirms the observations for the behavior of the test with varying stress at 850~ Fo There apparently is a critical stress at each test temperature above which nitrogen is less effective in interfering with the creep process.

V DISCUSSION OF THE RESULTS Relation of Dissolved Nitrogen and Creep Strength Heat-Treated Deoxidized Steels A correlation was found between the creep strength of plain carbon steel and the dissolved active nitrogen in the steel for varying deoxidation practices and heattreatments. An increase in the active nitrogen resulted in a large decrease in creep rate for the heat-treated deoxidized steel as shown in Figure 2, The correlation was examined statistically and the high correlation coefficient is valid proof that the correlation exists. This relation was developed for commercially manufactured steels, and the extremes in deoxidation practice and significant variations in heat-treatment were included in the test program. The large variation in creep rate of the deoxidized steels was related to the nitrogen retained in solid solution by prior processing to a degree which far overshadowed any other differences which existed between the steels tested. The microstructures of the steels tested included the extremes in ferritic grain size and carbide distribution for mild plain carbon steels. The lack of dependence 55

56 of creep rate on microstructure is perhaps more markedly shown by a series of microstructures for steels "C" and "F", Figure 16. The microstructures of the coarsened air cooled conditions of the two steels are very similar, and the creep rates in this condition are also of the same order of magnitude. The microstructure of these two steels in the coarsened furnace cooled condition are also similar. Both steels have been cooled from the coarsened austenite, However, the creep rates of these two steels vary by a factor of more than 30. The microstructure of the two steels in the fine, air cooled and fine, furnace cooled conditions show no significant differences in carbide distribution or ferritic grain size between the two steelso However, the creep rates again vary significantly between the aluminum and silicon killed steels9 Figure 17. The microstructures in Figure 18 show that varying carbon content and its influence on the microstructure does not cause a significant variation in the level of cteep strength, For example, steel "D" contains 0.10 percent carbon, and the microstructure consists largely of ferrite. The carbides in this material are distributed uniformly throughout the microstructure in both the fine and coarse air cooled conditionso The maximum creep strength of steel "D" did not vary significantly from that of steel "C" which contained a higher percentage of well distributed carbides.

57 The complete lack of correlation of creep rate with microstructure grain size or carbide distribution confirms the results obtained by Cross (9) The good correlation of the creep rate with the dissolved nitrogen strongly indicates that the important factor involved in the variation of creep strength of steels with deoxidation and heattreatment is the nitrogen contained in solid solution as a result of these treatmentso Effect of Other Processing Variables Rolling and Stress Relievingo —Steels "C" and "F' in the as-rolled condition contained high dissolved nitrogen and had high creep strength. Stress relieving steel "F" caused precipitation of some of the nitrogen and greatly reduced the creep strength. The creep rates for all three materials were in the range predicted by the dissolved nitrogen and the correlation curve for the heat-treated deoxidized steels, Figure 4o The creep rate for steel "B" rolled at 1650~ F. and 2150~ F. also varied with the dissolved nitrogen resulting from these two treatments9 Figure 4; however, the creep rates for this steel have been displaced from the correlation curve because of prior testing. The microstructures of the steels in the as-rolled condition are given in Figure 19o For steels "C" and "F" there is no apparent variation in microstructure, either between deoxidation or as a result of the stress relieving

58 operation for these steels. The structure of the steels consists of a moderately fine ferritic structure with well distributed carbides tending somewhat toward a Widmanstatten type structure. Stress relieving of steel "F" resulted in no apparent change in the pearlite or in the ferrite matrix. A similar situation exists for the low and high temperature rolling treatments for steel "B"o The microstructures of steel "B" in these two conditions, shown in Figure 19, are very similar and are typical of a low carbon aluminum killed steel in the fine grained air cooled conditiono There is no difference in microstructure or grain size between the two conditions of rolling although the creep rates varied significantlyo Spheroidizationo — Spheroidization did not significantly change the creep strength of the fine grained air cooled aluminum killed steels which originally contained low dissolved nitrogen. However, spheroidization significantly reduced the strength of both aluminum and silicon killed steels with high dissolved nitrogen, as shown in Figure 5, because the spheroidization also caused precipitation of the nitrogen. The microstructures for the three spheroidized steels are given in Figure 20, The coarsened condition of both steel "F" and steel "C" still show indications of theWidmanstatten structure, typical for this condition. On the other hand, the fine grained air cooled condition of

59 steel "F" shows a marked change in the carbide structure. The grain size of the fine air cooled condition is somewhat finer than that shown in the microstructure of the other two spheroidized steels However, there is little difference between the creep rates of these three steels in the spheroidized condition, It is significant that the gross changes in carbide distribution caused by spheroidization did not result in a significant deviation from the correlation curve for heat-treated deoxidized steels as shown in Figure 6. Spheroidization of the fine grained aluminum killed steel progressed more rapidly and would have caused a greater change in the strength for the steel in the fine grained condition if microstructure were the important factoro However, since there was very little change in the amount of dissolved nitrogen, the creep strength did not change significantlyo On the other hand, the Widmanstatten structures of steel "C" and steel "F" in the coarsened air cooled conditions are more resistant to spheroidizationo However, a major change in the amount of dissolved nitrogen occurred during spheroidization for these materials, and the creep strength in the spheroidized condition was reduced by an amount predicted by the correlation curve. The effect of hot rolling, stress relieving and spheroidization on the creep strength is clearly related to the dissolved nitrogen in the steel. The commercial

60 plain carbon steels in the as-rolled condition contained high dissolved nitrogen because of the high temperature of rolling and therefore possessed high creep strength regardless of deoxidation practice. Any variation in rolling practice or a subsequent stress relieving treatment, which reduced the nitrogen in solution, caused a simultaneous loss of creep strength. The creep strength of the high dissolved nitrogen conditions of both silicon and aluminum killed steels was reduced by the precipitation of aluminum or silicon nitride during the spheroidization treatment. It would therefore seem likely that any prior treatment which causes the precipitation of nitrogen in an inactive form could be expected to reduce the creep strength. Low Nitrogen Steels The results for the creep tests for the vacuum melted and vacuum annealed steels are further proof that nitrogen is the main element involved in the deoxidation effect on the creep strength. For example, the removal of the nitrogen by vacuum extraction from the silicon killed steel "C" resulted in a higher creep rate for both conditions of heat-treatment. This reduction in creep strength was related by the correlation curve to the decrease in dissolved or active nitrogen as shown in Figure 8o A similar reduction in strength resulted for the rimmed steel "A" due to the removal of nitrogen. The vacuum extracted sample of steel "C" had a coarse Widmanstatten structure as shown in Figure 7, and

61 the austenitic grain size of this material was large The two conditions are generally indicative of high creep strength, but the lower nitrogen content for these samples resulted in a reduction in creep strength The removal of nitrogen by vacuum melting for the strongly deoxidized aluminum killed steels caused a very significant change in the effect of heat-treatment on the creep rate. The creep rates for steel 1019, which had only 0.0017 percent nitrogen9 in the coarsened and fine grained air cooled conditions were very similar, Figure 9o An air melted steel of similar composition, but with 0,0094 percent nitrogen showed a change in creep rate of about three cycles on a logarithmic scale between the fine and coarse grained, air cooled condition. Similar results are seen for the comparison of steels 1018 and 1442. These results also indicate that the method of removal of the nitrogen is not important. The creep rates are practically identical for both conditions of steel 1019 with low total nitrogen and for the fine grained condition of steel 1379 with low dissolved nitrogeno The nitrogen in steel 1379 had been removed by precipitation of aluminum nitride due to the low austenitizing temperature but this effectively reduced the strength to that of the vacuum melted steels. A comparison of steels 1018 and 1442 leads to a similar conclusion, although the level of strength for these heats differs from that of steels

62 1019 and 1379. Steels 1018 and 1442 contained about 0.30 percent manganese and the level of strength would be expected to be higher because of the significant increase in the level of strength due to manganese. A plot of the creep strength versus active nitrogen for the vacuum melted and vacuum extracted steels is shown in Figure 21 along with the results for the comparable air melted steels It was assumed that the manganese affects the level of strength uniformly over the entire range of nitrogen, and parallel lines have been drawn for the different manganese levels A dotted line has also been drawn to include the high active nitrogen condition of steel 1379, and a break in the curve occurs approximately at 0o006 percent nitrogen. The microstructure of steel 1019 is shown in Figure 22 in the fine grained air cooled condition and is similar to other steels of this carbon content o The relation of nitrogen to creep strength explains some of the results previously reported for silicon killed steels. The creep strength of British steel "E", tested by Cross (9), was abnormally low, in spite of the fact that it had been deoxidized only with silicon, Further checks on the chemical analysis showed that the nitrogen content of the steel was 0.002 percent, which is abnormally low for this type material (10) However, the low nitrogen could be expected to result in lower creep strength for

63 this materialo The reported results of reference 10 for the 0,5 percent silicon steels can also be explained on the basis of change in the amount of dissolved nitrogeno Beeghly (34) has shown that, at about 0,5 percent silicon, silicon nitride is precipitated after austenitizing from temperatures just above the critical region. The precipitation of nitrogen as silicon nitride would remove the nitrogen from solid solution and would result in low creep strength as was experienced for these steels Rimmed Steels The relation between nitrogen and creep rate for the rimmed steels is significant in several respectso The microstructures of steels "A" and "H" in the coarsened air cooled and coarsened furnace cooled conditions are shown in Figure 23, There is a significant variation between the ferritic grain size and cooling rate which is expected for ferritic materials However, there is no significant variation between the creep strengths for the-two conditions of heat-treatment. There is a difference between the creep rates of the two steels and this difference is related to the total nitrogen in the steels. The intercept of the correlation curve for rimmed steel at zero percent nitrogen is, for all practical purposes, the same as the zero intercept for the deoxidized materialso This fact indicates that, in the complete absence of nitrogen, there is no significant difference

64 in the creep strength of rimmed and deoxidized steels at 8500 F. The variation in creep strength for rimmed steels is also a result of the dissolved nitrogen, although the strengthening effect is less pronounced than for deoxidized steelso Results in the literature (6) indicate that, at temperatures of about 6000 F,, the rimmed steels are as strong as the silicon deoxidized steel, and therefore the relationship for rimmed steels at 8500 Fo cannot be extended to all temperature ranges. As a matter of fact, it would be expected that the variation between creep strength and nitrogen for rimmed and deoxidized steels at 6000 F, would be very similar. At temperatures higher than 8500 Fo, it would be expected that nitrogen would have a less pronounced effect on the creep strength of rimmed steels. The nitrogen correlation explains the test results for rimmed steels in reference 8 and 9o Cross found that the core material had greater creep strength than that of the rim material for rimmed steels Results in the literature indicate that the coro material has a higher nitrogen level than the rimmed material (37), and this difference is statistically significant, Appendix, Calculation 9o Therefore, it would be expected that, for rimmed steels in general, the higher nitrogen in the core material would result in a higher creep strength for this material o

65 Causes of Scatter The scatter around the correlation curve in Figure 2 is about one percent of the total variation in creep rate with active nitrogen over the range of nitrogen contents of the steels tested. The effect of manganese, deoxidation practices heat-treatment, hardness and microstructure on the scatter were examined for the commercial material None of these variables had a definite influence on the level of strength for the commercial materials exclusive of the effect on the dissolved nitrogen in the steel This is not surprising however if it is considered that the normal variation in measurements of creep rate and nitrogen contents could easily represent a sizable part of the total scattero Manganese had a definite effect on the level of creep strength at low nitrogen contents, Figure 10o From Figure 21, it appears as if increased manganese in the range of 0o0 to 0,40 percent causes a decrease in creep rate at all levels of nitrogen. No definite conclusions can be given for such sparce data and these trends have only been indicated here, These results do show that after the importance of the effect of nitrogen is understood it is possible to better determine the effects of other alloying elements on the high temperature strength.

66 Discussion of the Mechanism The large change in creep strength of plain carbon steels with deoxidation and heat-treatment is directly related to the nitrogen in solid solution in the steel after processing No acceptable mechanism can be based on changes in carbide structures or grain size because of the lack of consistent correlation of creep strength with these two variables. The solution and precipitation of aluminum oxide with austenitizing temperature and cooling rate does not occur as shown by the lack of significant variation between heat-treatments for the aluminum oxide, Table V There was no real change in either alumina or aluminum with heattreatment for the three steels testedo The solution behavior of nitrides with deoxidation, austenitizing temperature and cooling rate is similar to that which Cross (9) proposed for the oxides~ therefore, the lack of variation of oxides does not exclude a precipitation mechanismo The important fact to be explained is how a minute increase in the amount of nitrogen in solution at the start of creep testing can cause a large increase in creep strength. There are two mechanisms which could possibly explain this relationshipo The precipitation of finely divided submicroscopic particles proposed by Cross must be considered. A mechanism for the effect of strain aging on materials under creep conditions has been

67 outlined by Cottrell and should also be considered, The precipitation mechanism proposed by Cross has been previously discussed in Section II. A brief summary of Cottrell's theory for strain aging phenomena is also included in Section II o A mechanism involving the precipitation of coherent submicroscopic particles and a mechanism based on Cottrell's strain aging theories are at least in partial agreement with the results. Both mechanisms require the solution of nitrogen to explain the effect of active nitrogen on the creep strength' However, the tests of these two mechanisms indicate that neither can explain all of the results o The isostrain tensile test results in Figure 11 indicate that steels with high dissolved nitrogen undergo an increase in strength with temperature up to 4000 Fo Over this temperature range, steels with low dissolved nitrogen do not exhibit increased tensile strength. The difference in strength between the two conditions of nitrogen reaches a maximum at about 4000 F, and above this temperature the strength of the two materials gradually approaches a common value, The increase in strength for the high nitrogen materials is not associated with a change in room temperature hardness, Bars of steel "C"' which contained high dissolved nitrogen, were heated to 4000 F, for one hour and showed no change in hardness

68 at room temperature. The hardnesses of the shoulder sections of the creep bars showed a negligible increase or decrease in hardness after testing and this minor change in hardness bore no relationship to the creep rates of the materials 0 These results cannot be reconciled with a precipitation mechanismo A precipitation mechanism would be expected to cause an increase in hardness for temperature conditions which resulted in increased strength. This is not the case in any of, the phenomena involving dissolved nitrogen. The maximum difference in strength between high and low nitrogen steels occurred at about 400G Fo for'the isostrain tensile results, An apparent maximum effect of nitrogen on the strength occurs at much higher temperatures for tests with lower strain rates, Figure 15, For example, nitrogen has no apparent influence on the strength of steels at 1000 Fo for stresses above 109000 psi. but causes a major change in creep rate in the range of 49000 psio Similarly, the creep rate curves at 8500 F, show the largest effect for differences in nitrogen at about 22 000 psi, Figure 15. These observations clearly indicate that the mechanism is a dynamic one involving plastic deformation rather than a'simple precipitation reactiono Isostrain peaks observed by Glenn (33) for various alloy additions in steel

69 also occurred at temperatures below the temperature range in which these elements still have an important influence on the creep properties Beeghly's method of analysis showed an increase in the amount of nitride nitrogen in the steels after creep testing9 Table XIIo These results indicate that the amount of active nitrogen has decreased during the test even at 8500 Fo However9 if the actual active nitrogen in the steel had decreased, then an increase in creep rate would be expected. This was not the case. The precipitation of all of the nitrogen during test would be expected if precipitation of the nitrogen were responsible for the change in strengtho The correlation exists between the nitrogen in solution in the steel prior to testing and the creep strengths In order to explain this correlation on the basis of precipitation, all of the nitrogen should have precipitated in a form which interfered with creep, The results in Table XII show that only part of the nitrogen has precipitated and the amount precipitated has no direct relation to the strength, However, these same results pose a problem in accepting a strain aging mechanism because the nitrogen must be free to move to dislocations in order to continually strengthen the material, No increase in creep rate was observed during any of the creep tests at 850~ F, although the results of Table XII indicate a decrease in dissolved

70 nitrogen. The creep tests at 1G000 Fo in Figure 12 showed a rapid increase in rate early in the test; however, the creep rates of these steels decreased to the original rate for longer times. It is possible that Beeghly's method of analysis does have some limitations in that nitrogen or nitrides, which are still active in terms of interfering with the creep process, are collected by Beeghl.y's separation Internal friction studies indicate that the nitrogen in iron alloyed with manganese, chromium, molybdenum and others form a type of pre-precipitated nitride in solid solution (38, 39). It is claimed that these compound groupings interfere with the complete precipitation of nitrides as measured by changes in internal friction (38)o It is possible that a pre-precipitated phase, which is still in a state which interferes with the creep process, was detected by Beeghly's method of analysis This phase could be either nitrides of aluminum or silicon, If the existence of these pre-precipitated phases could be confirmed for tested creep bars, then a modified strain aging mechanism could be accepted without reservationo Further study of the mechanism was based on the assumption that some modification of Cottrell's strain aging mechanism was responsible for the decrease in creep rate with increased dissolved nitrogen~ It was further assumed that the increased strength was associated with

71 an increased activation energy for the movement of "trapped" dislocations o The activation energy was calculated by Dorn's method (40) from the tests at a constant stress and varying temperature, and also from the maximum and minimum nitrogen for the deoxidized steels and the special air melted steels. Additional results were obtained from the literature (7, 40) A summary of the calculated activation energies and those obtained from the literature are given in Table XIII. Cottrell's strain aging mechanism was derived to explain the strengthening behavior caused by interstitial solution elements, If we examine the results in light of this mechanism9 certain observations can be explained. Cottrell assumed that the nitrogen atmosphere increased the energy for movement of dislocations. For the calculations for constant temperature and variable nitrogen, there is a difference in activation energy associated with a difference in nitrogen for both the rimmed and deoxidized steels. This increment in activation energy is in the range of the activation energy for the diffusion of nitrogen in steel The results for the tests at constant stress and variable temperature for low and high nitrogen steels also show an increase in activation energy with increased nitrogen. The creep curves for these calculations, shown in Figure 14, show

72 unusual behavior at 9000 F o therefore, the activation energy calculations in this case are not reliable, Dorn (40) calculated the activation energy from rupture test results for plain carbon and low alloy steels and found that the energy varied from 90 000 to 105,000 alo/moleo The indications of the activation energy data are that increased nitrogen in steel results in an increase in the activation energy for the creep process, Since the activation energies shown in Table XIII were not accurately determined, the actual values are not reliable enough to permit any conclusions concerning the magnitude of the change in activation energy with nitrogen. From the results for the rimmed steels shown in Figure 3, and for the very low manganese air melted steels in Figure 21, it appears that additional dissolved nitrogen above about 0,006 percent did not cause an additional decrease in creep rate. A strain aging mechanism would predict that additional nitrogen in excess of the amount necessary to form atmospheres around each dislocation should not cause further strengthening, To test this hypothesis, the number of nitrogen atoms for a given surface one atom layer thick was calculated for steel containing 0,0063 percent nitrogen, Appendix, Calculation 10. There are 2 x 1011 atoms of nitrogen per square centimeter at the 0.0063 percent nitrogen level, Estimates of the number of dislocations in the same area vary from 108

73 for annealed materials to 1012 for severely cold worked materials (41) o The number of atoms of nitrogen for the formation of an "atmosphere" is not known, but assuming an average 0 dislocation length of about 500 A about 200 atoms would be necessary for each plane through the dislocationo This gives a value' of 1 x 109 atmospheres for about 1 x 108 dislocationso The agreement between these two values is reasonable. No stronger conclusion is possible Both the rimmed steels and the air melted steels show a maximum effective nitrogen, and this is further support for a Cottrell type mechanism. The weight of evidence from the results of the mechanism tests indicated that a strain aging mechanism is responsible for the change in strength. There was no evidence of an increase in room temperature hardness for steels heated to 4000 F, (where the isostrain "peaks" were observed) or for tested creep bars. There was no consistent relation between the amount of nitrogen precipitated during test and the creep strength of the materials. The increase in strength with increased nitrogen is controlled by stress rather than strain rate, The dissolved nitrogen caused the greatest increase in tensile strength at.00~ F, for the strain rate of the tensile tests. At 8500 F. and 10000 Fo the effect of nitrogen on the creep strength is reduced at high stress levels and

74 occurs primarily at slow strain rates. If strain rate were the controlling factor, then it would be expected that nitrogen would be effective only at a more rapid strain rate at higher temperatures since the rate of diffusion of nitrogen increases with temperature, This was not the case, If stress were the controlling factor then the stress for activation of "trapped" dislocations could be expected to decrease with temperature since the increased thermal energy of the atoms would supply part of the necessary activation energy. This is in agreement with the observations from the effect of nitrogen on the isostrain tests above 4000 F. and the lower "critical" stresses for the effect of nitrogen at 8500 F, and 10000 F, There is every indication that the mechanism involved is a dynamic phenomenon involving plastic deformation as well as the solution of nitrogen, However, Beeghly's method of analysis indicated that precipitation of part of the nitrogen occurred during testing at 8500 Fo without the increase in creep rate which presumably should occur when an increase in precipitated nitrogen is detected. At 10000 Fo. a.reduction in strength occurred after several hundred hours, but the creep rate eventually decreased again and, in one case, went through several cycles of high and low creep rate. These results would not be expected if the nitrogen were precipitated in an inactive form, There is the possibility that as

75 deformation proceeds during the creep test, groups of interacting dislocations with their atmospheres of nitrogen are so closely bonded with iron and the minor alloys that they effectively form a pre-precipitate phase during testing which is detected by Beeghly's method. If such groupings of nitrogen and other elements in the steel were closely bound to the dislocations and, therefore, still active in preventing further deformation, Cottrell's mechanism would be acceptable, On the basis of these tests, Cottrell's mechanism can not be accepted without reservation. However the approach for further fruitful research into this problem is apparent" The mechanism involved apparently results in a strengthening. effect for nitrogen below a critical stress level and the stress decreases with temperature, A study could be made of the precipitates obtained by Beeghly's method by X-ray or electron diffraction studies and electron microscope studies to determine whether the precipitates, which form during testing, are aluminum or silicon nitrides or some pre-precipitation phaseo For the present, it can be concluded that the mechanism involved is stress sensitive9 that is, probably a dynamic rather than a static strengthening phenomenon. The true mechanism follows very closely with the predictions of Cottrell's theory in that its effect is detected only under conditions which involve plastic deformation.

VI SUMMARY AND CONCLUSIONS The correlation between the creep strength of plain carbon steels and the nitrogen in solid solution in the steels has been studied in order to explain the fundamentals involved in the variation of creep strength with deoxidation, heat-treatment and other prior processing variables which affect the creep strength at elevated temperatures. Steels from commercial practice and specially prepared vacuum melted and air melted steels were studied with the following results. lo For deoxidized steels, the effect of deoxidation and heat-treatment on the creep strength of plain carbon steels can be directly related to the nitrogen in solid solution in these steels as a result of these variables, At 8500 F, and 15,000 psi there is a linear decrease in the logarithm of the creep rate with an increase in dissolved nitrogen in the steel which is given by the following equation: log e 2 -2,083 - 593 x % dissolved nitrogen 20 The total nitrogen and creep strength are similarly correlated for rimmed steels by the following equations log e = -2.175 - 333 x % nitrogen The change in creep strength is less drastic for rimmed steels, but the creep rate for zero percent nitrogen is about the same as that of deoxidized steels 76

77 3S The results of other metallurgical processes on the creep strength of steel are as follows~ Both silicon and aluminum killed steels in the hot rolled condition possess high creep strength because the dissolved nitrogen is high, Any subsequent treatment which causes the precipitation of aluminum nitride and lowers the dissolved nitrogen, substantially reduces the creep strength. Spheroidization causes drastic changes in the creep strength of silicon and coarsened aluminum killed steels due to the precipitation of nitrogen but causes little change in the creep strength of the fine grained aluminum killed steels in which nitrogen is already precipitated. In all cases the creep strength, which resulted from a change in processing history, was related to the resulting dissolved nitrogen, 4. For steels with low nitrogen obtained by vacuum melting or vacuum extraction, the creep strength is low regardless of deoxidation or heat treatment, The level of creep strength for steels with low nitrogen is the same as that of similar steels in which the nitrogen was removed by precipitation of aluminum nitride, Therefore, steels with low dissolved nitrogen have low creep strength regardless of the method of nitrogen removal, 5, Deoxidation, heat treatment, carbide distribution, and grain size caused no deviation from the correlation curve for the commercial deoxidized steels which could be related to these factors, The manganese content and the hardness may cause minor variations from this curves however9 for air and furnace cooled mild steels in the range of commercial compositions only the effect of dissolved nitrugen was found to be significant Manganese over a wider range of 0,00 to 0,82 percent manganese caused a variation in the level of the creep strength for steels with low nitrogen, The strong relation between creep strength and dissolved nitrogen still existed at all manganese levels o The creep strength of steel is related to the dissolved nitrogen in the steel prior to creep testing, Silicon killed steels generally have high strength because silicon

78 nitride does not precipitate with normal heat treatment Spheroidization or any other process which caused precipitation of silicon nitride would decrease the strength of silicon killed steels. The nitrogen in aluminum killed steels is precipitated as aluminum nitride for most conditions of heat treatment and these steels have poor creep strength. Heating an aluminum killed steel above the coarsening temperature dissolves the aluminum nitride and fairly rapid cooling retains the nitrogen in solid solution. The creep strength in the high nitrogen condition is higho The nitrogen in rimmed steels is in solid solution regardless of heat treatment At 600~ Fo the creep strength of rimmed steels is as high as that of silicon killed steels. At higher temperatures, rimmed steels lose creep strength more rapidly, Apparently the silicon and aluminum in deoxidized steels serve to immobilize the nitrogeno In rimmed steels, no "stabilizers" are present and nitrogen is less effective in reducing the strength The effect of nitrogen on the creep strength is related along parallel lines to the effect of this element on the strain aging tendencies at room and slightly elevated temperatures o However, both precipitation aging and strain aging phenomena were examined and neither of these mechanisms could be accepted without reservation, The results of these tests do indicate that the nitrogen is only effective in increasing the strength under conditions involving

79 plastic deformation and that this change in strength is not accompanied by a change in hardness. A change in activation energy for creep corresponding to the change in dissolved nitrogen was noted, and the activation energy changed roughly by the same amount as the activation energy for the diffusion of nitrogen in iron. In light of a strain aging mechanism, this would indicate that the dislocations required more energy for the start of plastic flow, A maximum effect of active nitrogens was indicated at about 0,006 percent nitrogen and was roughly related to the number of dislocations in the steel o The true meohanism is apparently stress-ccntrolled, and this stress is, in all probability, related to the Increase in energy necessary to move the dislocAtions surrounded by nitrogen atmospheres. The energy for activsation is probably the elastic strain energy correspondirng to a "critical" stresso It is also probable that the nitrogen precipitated during test, as shown by Beeghly's method) is still in a form which interferes with the creep process since no change in creep rate was noted during the testing at 8500 F Considerably more research wil' be necessary before a clear-cut proposal of a mechanism is possible, and an approach to this study has been outlinedo For the present, the mechanism involved is very closely allied to a strain aging mechanism, as outlined by Cottrell, but this mechanism cannot be fully accepted without reservation or further testing.

TABLES

TABLE I CHEMICAL ANALYSES OF COMMERCIAL STEELS C Mn P S Si A] A1203 Cr* Mo* Ni* Cu* N Steel _ _O Steel | % % % % % % | % |% % A 0.10 0.41 0.011 0.024 0.01 0.001 0.001 0.03 0.04 0.01 0.06 0.0040 B 0.15 0.43 0.016 0.031 0.18 0.024 0.004 0.05 0.01 0.03 0.8 0.0047 C 0.20 0.68 0.028 0.034 0.27 0.015 0.003 0.02 <0.05 0.02 0.10 0.0048 D 0.10 0.42 0.011 0.025 0.01 0.013 0.017 0.03 0.03 0.01 0.05 0.0038 E 0.10 0.43 0.010 0.025 0.01 0.047 0.019 0.03 0.03 0.01 0.06 0.0036 F 0.19 0.68 0.026 0.036 0.24 0.053 0.004 0.05 0.006 0.07 0.09 0.0046 H 0.15 0.50 0.011 0.035 0.01 0.002 0.005 0.01 0.001 0.00 0050.0125 J 0.19 0.82 0.010 0.019 0.16 0.029 0.013 0.03 0.003 0.00 0.05 0.0047 * Spectrographic Analysis

TABLE II CHEMICAL ANALYSES OF VACUUM AND AIR MELTED STEELS'Cr* Ni* Cu* Zr SCt Mn P S Si Al ^O3 Cr* Mo* Ni* Cu Zr Steel % % % % % ^ % % % %% Air Melted 1 1 1 0.0 1442 10.09 0.29 0.023 0.0301 0,04 0.070 0,008 0.03 0.002 <0.01 0,09... 0.0098 1379 0.10 0.005 0.004 0.006 0,15 0_033 0.058 0 o060.05 000 00.00... 0.0089.. ~5 oo 0,0 0 Vacuum Melted 1016 0.0900 0.016 0.060,021 0o,06 0.01,., 0.03 0O01 0.07 0.17 0.004 0.0014 1018 0,09 0.25 0.004 0,008 0,00 0.,08..... 0003 0.005 0.03 0.00 0,002 0.0014 1019 0,09 0,05 0.004 0.012_ 0044 0.06 _._.. 0 1.0 0.005 0. 0.00 0.008 0.0017 * Spectrographic Analysis

TABLE III AUSTENITIC GRAIN SIZE OF STEELS AT VARIOUS AUSTENITIZING TEMPERATURES* Grain Size _____Steel__________ at A B C D E F J 18 19 79 42 16500 F., 1 hr. 2-4 6-7 2-4 7-8 5-7 7-9 6-7(8) 5-7 1-3(5) 5-8(7) 6-8 17000 F., 10 hrs. 3-4(1) 7-8 1-2 3-4(6) 7-8 6-8 6-7 5-6(4) 1-2 6-8 6-8 18000 F., 1 hr. 0,o oo aoo e O o ~ e o ~~ aooO eooooo 0oo0 ooo 4'5(7) 3-4 1-2(6) 7-8(9) 20000 F, 00 1 h r. oo o ooo 0 0,,,,*,.O, ~( o,oa.es Q 0o ~ ~ OaO 3-4 1 1-2(6) 3-5 3-5(6-7) 21500 F., 1 hr. 0-2 0-2 0-1 0-1 2-4(1) 1-2(4) 3-5 1-2(0) 1-2(0) 2-4(1) 2-4(6) * Units: ASTM grain size determined for carburized specimens

84 TABLE IV CREEP RATE AND DISSOLVED NITROGEN FOR DEOXIDIZED STEELSa Austenitizing Temperature Cooling Creep Rate Dissolved Steel Code oF Rate %/hr. Nitrogen % BFA 1650 Ab 0 0026 0.0008 BFF 1650 FC 0.0057 0.0009 BCA 2150 A 0,000039 0O0032 BCF 2150 F 0o.0040 0.0012 CFA 1650 A 0 000050 0,000029 0.0042 CFF 1650 F 0.000015 0.0042 CCA 2150 A 0,000025 0, 0042 CCF 2150 F 0.000033 0, 0042 0o000025 0,0046 DFA 1650 A 0.000066 0o0029 DFF 1650 F 0o00275 0.0010 DCA 2150 A 0.000048 0 0035 DCF 2150 F 0,00021 0o0034 EFA 1650 A 0 0046 0 0001 EFF 1650 F 0o0039 0 0009 ECA 2150 A 0,000050 0.0031 ECF 2150 F 0o 00295 0o0019 FFA 1650 A 0o 0143 0 00080 0,0029 FFF 1650 F 0o 00235 o 00008 FCA 2150 A 0.00005 0 0033 FCF 2150 F 0,00105 0,0025 JFA 1650 A 0 0035 0,0003 JFF 1650 F 0 000325 0 o0002 JCA 2150 A 0o000069 0o0035 JCF 2150 F 0o00031 0 0018 aCreep tests conducted at 8500 Fo and 15,000 psi bAir cooled CFurnace cooled

TABLE V ANALYSIS OF THE EFFECT OF HEAT-TREATMENT ON ALUMINUM AND ALUMINUM OXIDE IN ALUMINUM KILLED STEELS Steel B - E ----. B__ I__ f F_____ Heat- All 1l203 Al A1203 Al A1203 Treatment % % % % % | % 1650 AC. 0,022 0o 002 0o054 0o003 0.048 0.001 1650 FoCo, 0020 0,002 0o054 0.003 0.052 0,001 2150 A.C, 0.022 0,002 0.054 0,003 0.045 0o002 2150 F.C. 0,022 0,001 0.056 0o003 0o052 0.001 Analysis of Variance Aluminum Aluminum Oxide Degrees Degrees Between Steels 1o91 3 <0 01 13,1 3 <0o01 Between Heat-Treatment 1,04 2 Not 2 12 2 Not Residual 1,00 6 oooo 1,00 6 o oooo

86 TABLE VI CREEP FATES AND TOTAL NITROGEN FOR RIMMED STEELSa Austenitizing Temperature Cooling Creep Rate Total Steel Code OF Rate %/hro Nitrogen % AFA 1650 Ab 0o00021 0.0040 AFF 1650 Fc 0.00036 0 0040 ACA 2150 A 0.00022 0.0040 ACF 2150 F 0 00046 0.0040 A Vacuum Annealed 2150 A 0,00236 0.00135 HFA 1650 A. 0.000094 0.0125 HCA 2150 A 0.000031 0.0125 HCF 2150 F 0,000049.0.0125 aCreep tests conducted at 8500 F. and 15,000 psi bAir cooled cFurnace cooled

87 TABLE VII CREEP RATE AND DISSOLVED NITROGEN FOR STEELS IN THE AS-ROLLED CONDITIONa Creep Rate Dissolved Steel Rolling Conditions %/hr, Nitrogen % Bb Hot rolled from 21500 F, 0,00025 0,0047 Bb Hot rolled from 17500 F, 0 0050 0 0022 C As Hot Rolled-Mill 0 000025 0.0046 F As Hot Rolled-Mill 0,00011 0.0038 F Hot Rolled-Mill and Stress Relieved 2 hrs, at 1150~ F, 0.00085 0.0024 aCreep tests at 8500 F, and 15,000 psi bRetest after prior test at 10,000 psi and 8500 F.

88 TABLE VIII CREEP RATE AND DISSOLVED NITROGEN FOR SPHEROIDIZED AND UNSPHEROIDIZED CONDITIONS OF STEELS "C" AND "F"* Sample Spheroidization Creep Rate Dissolved Code Treatment %/hr. Nitrogen % FFA None 0.00143 0.0029 \o0 00080 FFAS 1250 F., 100 hrs, 0,00104 0.0012 FCA None 0000050 0.0033 FCAS 1250 F,, 100 hrs. 0.00207 0.0010 CCA None 0,000025 0.0042 CFAS 1250~ F, 100 hrs. 0.00455 0,0006 *Creep tests were conducted at 8500 F. and 15,000 psi

89 TABLE IX CREEP RATE AND DISSOLVED NITROGEN FOR VACUUM MELTED AND VACUUM EXTRACTED STEELSa Condit-ion Austenitized Cooled Nitrogen Creep Rate SteelF F% _ %/hro Vacuum Extracted C 1650 AO 1 1 00011.060 C 2150 A 0.0016 0. 0037 A 2150 A 000135 0o00236 Vacuum Melted _...,..... 1016 1650 A 0.00135 0 076 1018 1650 A 0 0014 0042 0.0011 1018 2150 A j o'oo~ 0o00632 1019 1650 A 0.0017 0.400 0.0017 1019 2150 ~ A 21000 0210 Air Melted 1379 1650 A 0.0012 0,222 1379 2150 A 0.0080 0,0004 1442 1650 A 0.0015 0,0156 0 00048 1442 2150 A.0.0036 0 00000756 aCreep tests were conducted at 8500 F. and 15,000 psi bAir cooled

90 TABLE X HARDNESS OF DEOXIDIZED STEELS IN BASIC CORRELATION OF FIGURE 2 Hardness After Testing Steel and Original Rockwell B Log Deviation Heat-Treatment Hardness Reduced From Correlation Code Rockwell B Section Shoulder Curve x 10 BFA 65.0 66 62 - 30 BFF 5705 61 56 +271 BCA 63.5 o0 -460 BCF 52 0 6 53 +395 CFA 75.0 72 70 + 54 CFF 6800 67 65 -252 CCA 75,5 73 75 - 21 CCF 69,0 73 70 + 99 DFA 55.0 48 50 -760 DFF 4105 48 39 +143 DCA 57.5 65 71 + 6 DCF 38.0 40 37 +450 EFA 57o0 53 50 -167 EFF 53,0 53 52 - 24 ECA 5800 50 46 -352 ECF 40o0 69 45 +559 FFA 7505 74 74 +734 FFF 74,0 70 70 - 74 FCA 74~5 72 71 -233 FCF 68,0 70 69 +586 JFA 75,0 66 67 -281 JFF 7100 63 66 -289 JCA 77,5 45 43: - 4 JCE 720o 57'60 -330

91 TABLE XI HARDNESS OF RIMMED AND OTHER STEELS TESTED Steel and Hardness After Test Heat-Treatment Original Hardness Reduced Code Rockwell B Section. Shoulder Rim Core AFA 52 65 61 58 AFF 38 48 43 51' ACA 51 64. o ACF 38 50. A, Extracted 53.0 HFA 65 66 66 HCA 64 63 63 HCF o 56 54 C, As-Rolled o 72 73 CFA, Extracted 70 o o CCA, Extracted 73 0o0 0 F, As-Rolled 72 o. F, Stress Relieved 69. 0 FFA, Spheroidized o 67 64 FCA, Spheroidized 69 62 63 CFA, Spheroidized 60 67 63 1016 FA 49 58 39 1018 CA 53 58 51 1018 FA 58 61 48 1019 CA 60 67 54 1019 FA 64 70 59 1379 CA 54 53 45 1379 FA 52 57 46 1442 CA 59 o. 1442 FA 62 00

92 TABLE XII MEASUREMENT OF ALUMINUM NITRIDE BEFORE AND AFTER TESTING AT 8500 F. FOR TIMES UP TO 1660 HOURS Percent Nitrogen as Aluminum Nitride After As Heat- Test at Change in Steel and Time of Test Treated Times 8500 F. AlN, % F Stress Relieved, 763 hrs. 0.0022 0,0031 +0 0009 F As-Rolled, 1661 hrs. 0.0008 0.0022 +0,0014 FCA, 1292 hr$. 0.0013 0.0028 +0o0015 BCA, 1240 hrs. 0,0016 0.0023 +0.0007 FFA, 858 hrs. 0.0019 0.0032 +0.0013 AFA, 710 hrs....... 0.0004 0o0o00 DFA, 688 hrs. 0,0009 0.0014 +0.0005 ECA, 600 hrs. 0,0007 none.oo ____ __ detected

TABLE XIII SUMMARY OF ACTIVATION ENERGIES FOR CREEP TESTS ON PLAIN CARBON STEEL Tempo ^H Act. OF Cal/Mole Type Steel Method Significance 850 13,900 Rimmed Steels Dorno (40) T Const. Represents H Intercept to Max. N2 Difference Due to Nitrogen 850 16 500 All Deoxidized Dorno T Const. Represents aH Commercial Steels Intercept to Max, N2 Difference Due to Nitrogen 850 21,000 Steel-1379 DornO T Consto Represents^H o Intercept to Max. N2 Difference Due to Nitrogen 800 - 900 112 000 High Nitrogen Dorn: T Variable, Activation Steels % N Constant Energy Iron + Nitrogen 800 - 900 82,000 Low Nitrogen Dorn: T Variable9 Activation Steels % N Constant Energy Iron 1000-1400 90 000 0o15% C. (40) Dorn: T Variable, Activation (Rupture) Energy Iron 900 -1200 137,500 (High Nitrogen)(7) Dorn: T Variable Activation Coarse Grained Energy Iron + Nitrogen

J1 UUKUS

DEOXIDIZED STEELS: 0.004-0.005 % NITROGEN NOMINAL | _____ALUMINUM ADDED SILICON SILICON 0 AL I#AL T. 2 AL T. STEEL D STEEL E SEE RIMMED 0 Si S RME 0.013 AL 0.047 AL 0.01 Si 0.0 Si STEEL C STEEL J STEEL F 0.25 Si 0.015 AL 0.029 AL 0.053 AL 0.27 Si 0.16 Si 0,24 Si STEEL B 0.024 AL 0.18 Si RIMMED STEELS: STEEL % NITROGEN A 0.001 AL 0.004 0.01 Si H 0.001 AL 0.0125 0.01 Si A VAC ANN 0.0014 HEAT TREATMENT DESIGN AND CODE: AUSTENITIZING TEMPERATURE COOLING RATE 1650~F IHR. 2150~F I HR. AIR COOL - FA CA FURNACE COOL * FF CF STEEL DESIGNATION DESIGN QO BASIC EXPERIMENT FOR COMMERCIAL MATERIALS SHOWING NOMINAL AND ACTUAL ALUMINUM AND SILICON CONTENTS FIGURE I

10-2 10-2 Kll|HEAT TREATMENT 1650 F A.C. 0 \. BE) 16500F F.C. L G. E 2150~F A.C. A 0, 0 \ E XB 21500F F.C. X,83'O X E 0 BjAJ' 0 L <i0|3 X F_ _ _ O1- \ xbJ J a. XD w U 10-4 GD AJ, ~ D....___________A B XC STEEL TYPE \ C B Si -AL E AL. 2 -/T. C A C_ Si F_ Si-AL. 2 /T. D AL I /T. J Si-AL I /T. 0.001.002.003.004 ACTIVE NITROGEN -% CORRELATION OF CREEP RATE WITH ACTIVE NITROGEN FOR DEOXIDIZED STEELS. FIGURE 2

10-2 HEAT TREATMENT CODE 1650 F, A.C. 0 16500F, F.C. 0 5 0 \\ 21500F, A.C. 0 o \\ I I I I E 22150 F, F.C. $n \ 1 A-VACUUM I \ I \l I I I 1 1 Q-\ I \ -2% o. 10-3 I A A a. DEOXIDIZED STEELS \ 0 H I *H -5 OH -2 0.002.004.006.008.010.012.014 NITROGEN -% EFFECT OF NITROGEN ON THE CREEP STRENGTH OF RIMMED STEELS FIGURE 3

10O2 I i0o^~~2 r l l [ ~DEOXIDATION STEEL Si -AL* B \^B ^Si-AL 2' T. F "\ Si C 0 0 a. 2 w HEAT TREATED 10-34.001 TREATMENT CODE \ \ HOT ROLLED-MILL X HOT ROLLED AND C X STRESS RELIEVED D _ HOT ROLLED 21500F 0 * AFTER PRIOR TEST AT 10,000 PSI, 850OF HOT ROLLED 17500F * ACTIVE NITROGEN -% EFFECT OF ACTIVE NITROGEN ON THE CREEP STRENTH OF AS ROLLED OR STRESS RELIEVED PLAIN CARBON STEEL FIGURE 4

V/\ TR^ I~L SPHEROIDIZED - I\AS TREATED U_ 1250F, 100 HR. 102 Si STEEL C b- 16500F,A.C. 5 0 0 Xo 2. AL/T. STEEL F n 16 50~ A.C. I~ 2#AL/T. A - STEEL F 21500F A.C. I- 10- 3.C. 0^~ ~~5// CL 2 / /2 EFFECT OF SPHEROIDIZATION ON CREEP STRENGTH OF FINE AND COARSE GRAINED STEELS FIGURE 5

10-2 1 0 0 co 2 |-1 < 10-3 ~~L I- N HEAT TREATED < AN A SDEOXIDIZED FU 6 STEELS LaJ 10-4.001.002 A_ H__AA 5 SPHEROIDIZED AS HEAT TREATMENT TREATED m* 0El~ F 1650 A.C. 0 *O__~ e__0 C 1650 A.C. 2 A A F 2150 A.C. CORRELATION ro5 __________ CURVE ro- I 0.003.0034 ACTIVE NITROGEN-% RELATION BETWEEN DISSOLVED NITROGEN AND CREEP RATE BEFORE AND AFTER SPHEROIDIZATION. FIGURE 6

FIGURE 7: MICROSTRUCTURE OF VACUUM EXTRACTED SAMPLE OF SILICON -KILLED STEEL C VACUUM EXTRACTION 100 HR. AT 2000 F. HEAT TREATMENT 1650~F AIR COOL ITHEN, 2150OF AIR COOL. MAGNIFICATION 100 X ETCHANT NITAL CREEP RATE 0.00037 %/./HR.

10-2 AS VACUUM \ ____TREATED EXTRATED 1650 A.C. 0 5 o^ \ 1650 F.C. B 3 2150 A.C. A A 2150 F. C. X o 2 < 0-3A ~~ HEAT TREATED DEOX IDIZED z~if ^ ~ \I~ STEELS FIGURE 8 STEL C.-2 JE 8 bJ 10-~4_ 0.001.002.003.004 ACTIVE NITROGEN, % EFFECT OF NITROGEN EXTRACTION ON THE CREEP RATE OF SILICON DEOXIDIZED STEEL C. FIGURE 8

CREEP RATE %/HR. AT 15,00 PSI 850*F o o o, o 0 0 -q O 00 1018 VACUUM MELTED STEEL 0.0014 NITROGEN 1442 AIR MELTED STEEL 0.0086 NITROGEN | [ | P-1 "r -1 l\o ru - \" m: 0 1442 AIR MELTED STEEL 0.0086 NITROGEN (J) ~j z 0 __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ _ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ _ __ _ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ __ __ _ __ _ _ _ __ __ _ __ _ __ _ __ __ _ __ _ __ _ _'

01 3.fnou1'N300t8lN o'00'0 01. 100'0 HJMII S133.LS Oi0 L'J.03_,l-3 3S3NYVONV'1 %/ 3S3NV ON V 0'1 9' 9' 0 I _-01 _, I0 0.\~~~~~~ l^~~(n r11 \ oo 0* ('I 0 A^ \on~~~~ I(n I~~1-01 01 \~ g,_~~~~~~~~

ALUMINUM KILLED STEEL SILICON KILLED STEEL COMPARISON OF COARSE AND FINE GRAINED STEEL F. 2150 AC ~ STEEL C 2150 AC ~ STEEL FOR.05 AND 0.5 TRUE STRAIN 1650 AC X 1650AC X CFA FFA X CCA O FCA O 16C 160Cui 160 * I o t 1.05 t, I 140 140 140 ~____..< D6C X/ Ex =OC: \ _ __ __ 20 20 ~ 20 ~ ~ -= TRUE STRAIN C I_ O0 200 400 600 800 1000 0 200 400 600 800 1000 0 200 400 600 800 1000 TEMPERATUREOF ISOSTRAIN CURVES FOR STEELS C AND F IN FINE AND COARSENED CONDITIONS. FIGURE II

0.4 TESTS AT 1000I F 0.3 __ 0 0 0^~ J:4~ ~ \ 000 PSI.:Si -AL uw 0.2 a \ PSI.:Si 0.1!' A:=4. F00 PSI..:Si-AL 0 1000 2000 3000 4000 5000 6000 7000 8000 TIME -HOURS CREEP RATE-TIME PLOT FOR STEEL "C"AND"F" IN THE STRESS RELIEVED CONDITION AT IOO0~F FIGURE 12

0.03 o FCA LO CHANGING 0.02 LOAD 0.0~'~ UJ Or0 _ __ 0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800 TIME - HOURS CREEP TEST WITH VARIABLE STRESS AT 850 F FIGURE 13

0.08 0F 00_F____ ______ _ ~ 0.07 0.06 0.05.. 0.04 FCF:Si-AL 0.03 0~~ 0.02 0.01 800~ F F 0. oF 0.014 1~ ~~ ~~~~~~~ C I. ___ 0.012 z o 0.008 - FCA: Si-AL 0 0.006 z 1500 1600 1700 1800 1900 2000 2100 220 230 0.002 800F _ 0.012 0.012 E 8500F RESET _ _ 0.010 0.010 0 0.008 0.008 ~ ~ 1500 1600 1700 1800 1900 2000 2100 2200 2300 O 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 TIME - HOURS CREEP TESTS AT CONSTANT STRESSVARIABLE TEMPERATURE FIGURE 14

0-50 r INU 10 6 ~^ - 10-~ ~~ 1 0~2 x - I0-5 STRESS -CREEPRATECURVES _AT 850 0F AND 1000 F FIGURE 95 10-6 10-5 10-4 10-3 10-2 10-1 CREEP RATE O/o/HOUR STRESS-CREEP RATE CURVES AT 850=F AND IOO10F FOR COARSE GRAINED (HIGH NITROGEN) AND FINE GRAINED (LOW NITROGEN} STEELS (3,8,9,1041,14 FIGURE 15

....................i~il A500 X 500 X 500 X 500 X i. —. ii _ ":-i'Aii 76( 97:i-: siii~~~~~~~~~~~~~~~~~~~~~~~~~~~~i A XV t 5 4 57f ____:A CCA: SILICON - KILLED STEEL: FCA SILICON-ALUMINUM KILLED STEEL, CCFr SILICON -KILLED STEEL FCF:SILICON-ALUMINUM KILLED SEL CREEP RATE 0000025%/HR. 2T/T: CREEP RATE 0.00005 %/HR. CREEP RATE 0.000033%/HR. 20/T' CREEP RATE 0.00105 9/a/H FIGURE 16 MICROSTRUCTURES OF SILICON KILLED STEEL SILICON ALUMINUM KILLE SELFARNDUNCEOLDFOM 2150F, NITAL ETCH.

STEEL C: Si STEEL F: Si -Al CFA: CREEP RATE 0.000029 %/HR. FFA: CREEP RATE 0.00080 %/HR. CFF: CREEP RATE 0.000015 %/HR. FFF: CREEP RATE 0.00235 %/HR. FIGURE 17: STEELS"C AND F AFTER AIR AND FURNACE COOLING FROM 1650~F, NITAL ETCH, MAGNIFICATION 500 X. ~~B~~r I~~~i~~~sr~~~~:-...........................-ii~i~iiiiii~iiii i~ I i'i~~~~l-.''':~l~i'~~'~'~~~:'6 ~~:-::~:~;''` l ~~~l ~:'::'~'s;~...........'i.;~ iii-i i~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ii~~...... iii-iii-i"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...... CFF: CREEP RATE 0.000015 %/HR. FFF: CREEP RATE 0.00235 %/HR.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~::::~:::::::::::::::::::::::::::::::-::

STEEL D STEEL J O.10 % CARBON 0.19% CARBON AL I#/T Si-AL I./T..........N A. g..''...''...EW s...|.......i i.....-.E: ~~~~~~~~~~~~~~~~~~:E:E-E::i:-E::E::::I::::-fE::::j'-::::::::::::::::E:: -:S:::g:: ~~~~~~~~~~~~~~~~~..0.:'.B..'.... l'. tsl.:::t.0ti.: sti:: 0: ttgg <:; 9;....................................'t. Ei:. A:C EE:i,:E 90~ -:::::::-JCF:..CREEP RATE 0.00031, s/HR............'..........JFA.: CREEP RATE 0.0035.o/HR........ F'IGURE 18:TYPICAL M ICRO ST RUCTURES OF' HIGH AND LOW RANGE. OF' CARBON CONTENT, NITAL ETCH ~MAGNIFICATION.500 X Bg a B'.'...,..................:;....-.'lg,,g g- t ~ i....................:. 1E:E':. E.9''.''...'9::. t.:: -: Sf::::::,:: f.::.. i: -.:....-...S:......:9....s.....i...E.:.....:...>............. =.................B..:. s -.::::: B E..-.a...sB-S u:B,; e. B:. E D.->BE:) >.: E > a E E:::::::":: z B ~.....>...:B:::: DCA CREEP RATE 0.000048 %/HR. JCF: CREEP RAJE 0............ | iRE;ik EE~i~f~i:S~g.#R i~~iR. Wi~i i: g:-ii-gi E: gig~i i-igi'Si g ig~gE-E-i~EE;'iiE~~iiB~.;g.......................... 2 is g8 rs..........E5| iv:.~:p..::~ia~g..a::v::~:i:2:~i iv:-~g~i<..................... va ia:,<-g 88ase~g#:@;ga ~:l<..;< 1>-R;R@ eX.........;............R ":~::1:l> B 8:.a"::::1~ g6; -s.N.-.-, —........................................................ a.:.>s"l... DFA: CREEP RATE 0.000066 %//HR. JCFA CREEP RATE 0.00031 %/HR.. FIGURE 18 TYPICAL MICROSTRUCTURES 0F HIGH AND LOW RANCE OF~...............................................L-TC sMA NFIATON 50

STEEL F-Si-AL STEEL F-Si-AL STEEL C-Si STEEL B-Si-AL 500 X 500 X 500 X HOT ROLLED 17500F, 500 X CREEP RATE 0.0050%/HR 110F2HR.CREEPRATE0085HR.. R....00. FIGUE 1: MCROTRUTURE OFMATRIAS TSTEDIN S RLLE CODITN, NITAL, ETCH. Al- 4~ TW~ 1OX IOOX I oo x HOT ROLLED 215O'F SOO X AS ROLLED-MILL,STRESS RELIEVED AS ROLLED -MILL AS ROLLED-MILL CREEP RATE 0.000~5 %/HR 11500F 2HR.)CREEP RATE 0.00085%/HR. CREEP RATE 0 00011%/HR. CREEP RATE Q000025 %/HR. FIGURE 19. MICROSTRUCTURES OF'MATERIALS TESTED IN AS ROLLED CONDITION, NITAL ETCH.

STEEL C: Si 2150~F, A.C. STEEL F:Si-AL,2150~F,A.C. THEN 12500F, 100 HRS. THEN 12500~F, 100 HRS. CREEP RATE 0.0045 %/HR. CREEP RATE 0.0010 %/HR. STEEL F:Si-AL, 1650~F, AC. THEN 1250~F, 100 HRS..NITAL ETCH, MAGNIFICATION 500 X..::::::.::::::::.............. i.... i i * FIGUR 20 IRSRCUE F TESCA NSPEODZDCNIIN NITAL ETCH, MAGNIF~~~~~~~~ICAIO 500X.

10 1650~F 2150o1 TREATMENT CODE 1650OF 2150 ~ F TREAT M ENT CODE AIR COOLEDAIR COOLED AIR MELTED ^ 19 0 VACUUM MELTED O VAC. EXTRACTED EQ 19 79 -. 10-I 0 0 o \ 1 \oie OI)l~ tA 2 co-C2 I 18 w \1 (0.05 Mn 3 \ \0.3 Mn C, —- -~ -- 2. 10-4.3Mn 0 0.002 0.004 M006 0.008 0.010 ACTIVE NITROGEN - % RELATION BETWEEN CREEP STRENGTH AND ACTIVE 10-4 ~ ^ ~~~-I NITROGEN FOR VACUUM TREATED AND SPECIAL AIR MELTED STEELS FIGURE 21

'H1L3 91VIN'UH/% OV'O 31V~ d3303'l331S 0a31~3 InnO VA'6101 I V3H:H Z3~nOlJi X 001oo ____________0____w___$_, /~Nb'i~i~: ~~::~:i-~1~:i4:; Vil'i~il'i~isiiiiii~i ifiii ~~~~~ ~~~~~~~~~~~~~~' -;:~i'ii~:ii.::::::::::-$: 4::::::: -::::: ~p:i-i:iiiiii;:h:::: ~l:: —-:-::::-:1::::: - -iii~l l~ii 1V~~~b,~~ 00

STEEL A STEEL H 0.004 % NITROGEN 0.012% NITROGEN ACA CREEP RATE 0.00022 %/HR. HCA: CREEP RATE 0.00031 /HR. ACF: CREEP RATE 000046HR. HC CREEP RATE 0.00049 %/HR. FIGURE23: MICROSTRUCTURES OF RIMMED TEELS, NITAL ETCH EPAG NIFICATION 500 X. FIGURE23: MICROSTRUCTURES OF RIMMED STEELS, NITAL ETCH, MAGNIFICATION 500 X.

APPENDIX

119 Calculation 1o -- Time required for removal of nitrogen from 0.252 inch bars by vacuum diffusion. For diffusion, the variation of concentration with time is correlated by a ratio Dt Where: D = the diffusion rate t = time in hours L = thickness of the bar The values for nitrogen extraction are: t = 100 hrso C = 40 x 10-8 cmo/seCo at 2000~ F L = 1/4 inch Dt 40 x 10-8 x 100 x 3600 _= 0593 L -0.252 x 25.4 for Dt Cm-Co for Dt = 0.593; -= 0.85 for a round bar (42). Where: Cm = mean final concentration Co = mean original concentration = 0.004 Cs = surface concentration = 0.000 Cm - 0.004 I 0004 = 85 Cm = 0.0007% (calculated) Cm = 0.0013 to 0.0016% (measured)

120 Calculation 2, -- True stress-true strain calculation for steel FCA at 4000 F Formula L do etr In L- 2 in d-... 0 0 C2) where: etr is the true strain L and Lo are the instantaneous and original measured length d and do are the original and instantaneous diameters True True Strain Are. Load Stress L In L/Lo etr in /in. in. lbs. psi 2.995 0 0.1995 50 250 2.9953 0.00010 0.00010.1995 400 2,010 2.9955 0.00018 0.00018.1995 1,000 5,020 2.9959 0.00030 0.00030 o1995 1,500 7,530 2.9962 0.00040 0.00040.1995 2,000 10,050 2.9964 0.00048 0.00048,1995 2,500 12,550 2,9968 0,00058 0.00058.1994 3,000 15,050 2.9984 0.00112 0.00112.1993 3,500 17,550 2.9985 0.00115 0.00115.1993 4,000 20,100 2.9988 0.00127 0.00127.1993 4,500 22,550 2.9990 0.00134 0.00134.1992 5,000 25,100 2.9993 0.00144 0.00144.1992 5,500 27,600 2.9998 0.00159 0.00159 o1992 6,000 30,100 3.0089 0.00464 0.00464.1986 6,900 34,700 3.0315 0.01219 0.01211.1971 9,000 45,700 3.0404 0.010.01516 0.01506 o1965 10,310 52,500 3.0667 0.02394 0.02366.1948 11,690 60,100 3.0931 0.03275 0.03222.1932 12,710 65,800 3.1066 0,03726 0.03658.1923 13,090 68,100 3.1205 0.04190 0.04027.1915 13,420 70,200 3.1487 0.05130 0.05005,1898 13,880 73,200 3.1766 0,06063 0.05886.1881 14,210 75,600 3.2059 0.07042 0.06805,1864 14,420 77,400 3.2394 0,08160 0.07844.1844 14,550 79,100 3.2683 0,09125 0.08732.1829 14,610 79,900 3.2881 0.09786 0.09336.1817 14,590 80,500 Max. 3.3054 0.10364 0.09861.1808 14,530 80,500 3.3254 0.11032 0.10465.1797 14,430 80,500 3.3415 0,11569 0.10947.1788 14,280 80,000 3.3597 0.12177 0.11491.1778 14,075 79,300 3.3784 0.12801 0.12054.1768 13,820 78,200 o.0.00o..0o.. 0,19492.0901 11,980 133,100

Calculation 3. —Correlation (35) of dissolved nitrogen with logarithm of the creep rate for deoxidized steels. Dissolved y Y Nitroge^ Log Code % x 10 Creep Rate |x. - n(Mx)2 |x Exy - nMxMy ry My |y2 _ n(My)2 BFA 8,0 -2.587 17356.50 551 -2177.445 1-82.678 -3,445 304,387 BFF 9.0 -2,346 -12650o00 1898',149 -285.156 BCA 31,5 -4.411 4706o50 I- 279.296 19.3 BCF 12.0 -2.400 -~.... CFA 42.0 -4o529 xy- nMxMy -279,296 CFF 42o0 -4,826 b = x2 b; b05 93 CCA 420O -4.604 n 4706.50 CCF 42.0 -4,484 - DFA 29 0 -4.563 a = My -bMx a = -3 445 + 1 362 =-2 083 DFF 10.5 -2 563 DCA 35.5 -4,182 log creep rate = -20083 - 593 x % N DCF 34,5 -3.679 EFA 1.5 -2.339 - nM.My r 279.296 EFF 8,75 -2 611 rr L x2 - n(Mx)x [y2 - n(My2; r 0 192317 ECA 37,5 -4, 303 n(M6 5 x 1 ECF 17o0 -2,532 FFA 29.5 -3,099,2 1 - 141 x 2 = 0,852 FFF 8.0 -2 631 0 FCA 33.5 -4.303 FCF 25,0 -2,980 r' = 0.923; significant JFA 3.0 -2 542 JFF 2,0 -2 o491 JCA 35,0 -4.163 JCF 18,5 -3,510

122 Calculation 4, —Correlation (35) of nitrogen with the logarithm of the creep rate for rimmed steels. (Assumption 0.0063 maximum effecto) x y % N x 104 log creep rate 40.0.0 0000000000 O 0000 -3o678 4000..ooo..ooooooo..... -3444 4000 o....oooo.oooooooo -30658 400,oo..... o...,,. -3.337 63,0 00ooo.000000000000000 -4509 63,0,,... -40310 13,5 o0oo0..,0000 0,,,,, -2,627.. ~ ~ ~ ~ ~' ~ ~ ~ ~ ~ ~ ~ ~ ~ - ~ ~ ~ ~~.,, _'.' x2 - n(Mx)j2 x |xy - nMxMy Ey My Vy2 _ n (My)2 18489.25 362,5 -1409,443 -29059 -30699 111,930 -16425.78 1340.797 -109,446 ~2063,47 ~68.646 2.0482 xy -Y nMxMy -68 646 b y b -0.0333 x2 _ n(Mx)2 b 67 -0333 a = My- bMx, a = -3,699 + 1,524 = -2,175 log creep rate = -2,175 - 333 x % N xy - nMxM 8646 09588 r 7 -2 n(=/S 1y2 _ n( )2 /2063 47 x2 o482 r2 = 1 - 0,041i x = 0,952 r' 0,976, significant

Calculation 5. —Correlation of manganese with creep rate for steels with 0,001 to 0,002 percent nitrogen. y x Log Code Mb % x 102 Creep Rate x 2- n(Mx) 2 x xy - nMxM My yy - -n(MY) BCF 43 -2.400 32825 573 -1551.871 -30.941 -2.063 78.3233 DFF 42 -2.563 -21889 1182.290 -63.8602 ECF 43 -2.532 1096 369.581 14.4631 JCF 82 -3.510 FCS 68 -2.346 FFS 68 -2.688 _xy- nMMy -369581 -033795 CFX 68 -3226 b -; b - 19 CCF 68 -3.432 x -n(MX)2 936 19FA 5 -0.389 19CA 5 -0.679 a = My- bMx; a -2.063 + 1,291- -0,772 79FA 1 -0 658 18FA 25 -1.377 log creep rate = -0,772 - 33,8 x% Mn 18CA 25 -2,199 42FA 29 -1.827 rxy - n MVMX 369,581 16CA 1 -1:119 r _ r^, ^-^ ~l6CA^ 1 -x11 - n(Mx) y2 - n(My)2] /_(10936) x 14,463 ='2 1 i -0,137 x 14 = 08125 r' 0.901, significant

124 Calculation 6o —Difference between actual and calculated creep rates. log % log E Actual - Calco Code Calculated Actual Difference x 103 BFA -2.557 -2.587 - 30 BFF -2,617 -2o346 271 BCA -3 951 -4 411 -460 BCF -2.995 -2o400 395 CFA -4 583 -4 529 54 CFF -4.583 -4.826 -243 CCA -4.583 -4.604 - 21 CCF -4,583 -4, 484 99 DFA -3,803 -4.563 -760 DFF -2.706 -2.563 143 DCA -4 188 -4,182 6 DCF -4.129 -3.679 450 EFA -2,172 -2,339 -167 EFF -2,587 -2.611 - 24 ECA -3,951 -4,303 -352 ECF -3,091 -2,532 559 FFA -3.832 -3, 099 734 FFF -2.557 -2.631 - 74 FCA -4.070 -4.303 -233 FCF -3,566 -2, 980 586 JFA -2, 261 -2,542 -281 JFF -2 202 -2 491 -289 JCA -4.159 -4,163 - 4 JCF -3.180 -3o510 -330

125 Calculation 7o —Analysis of variance (36) of the effect of heat-treatment and steel on the variation from the correlation curves for steels "B" "D", and "E" (0o43 Mn) and all deoxidized steels, Sum of Degrees Mean F Item Squares Freedom Square Ratio ISignificance 0o43 Manganese Steels Between Heat-. Treatment 410,155 3 136,718 0.611 None Between i Deoxidation 569383 2 289191 0,13 l None Residual 1,342,637 6 223,773 Total j1,709,175 II 1..1 All Deoxidized Steels Between HeatTreatment 497,696 3 165,899 105 None Between Steels 1199480 i 5 239896 0015 None Residual 2,374,661 15 158,311,.W_ __,, r_ _.. Total 2,994,737' 23 ii'' Conclusiono Neither heat-treatment nor type of deoxidation causes any significant variation from the correlation curve

Calculation 8. —Correlation (35) of the deviation of the individual testi from correlation curve in Figure 2 with hardness. Hardness y Rockwell B Log E |x2 - n(Mx)2 |x Exy- nMxM y | y y2 n(My)2 65.0 -0.030 98704.00 1 1513,0 -23.5060 +0.018 +0.0075 2.431 57.5 +0.271 -95382.04 1 - 11347 63.5 -0.462 3321.96 _ -24.6408 52.0 +0.395 75.0 +0.054 Mxy - nMxMy -24.6408 68.0 -0.252 b = y b =3i -0.0074175 75.5 -0.021 x2 - n(Mx)2 3321.96 69.0 +0.099 55.0 -0.760 a = - b Mx; a = 0.4676 + 0.0075 = 0.4761 ~41.5 +0.143 57.5 +0.006 38.0 +0.450 log creep rate = 0.476 -0.007 x hdns. 38-.0 +0.450 57.0 -0.167 53.0 -0.024 -24.6408 58.0 -0.352 r? x2..y r- =0.2535 40.0 +0.559 - n(Mx) y - n(My)5 75.5 +0.734 74.0 -0.074 74.5 -0.23 p =>0.10, not significant 74. 5 I-0.233 68.0 +0.586 75.0 -0.281 71.0 -0.289 77.5 -0.004 72.0 -0.330

Calculation 9, —Segregation of nitrogen in rimmed steels (37) and test for difference between rim and core, t = x y / (n + m- 2) degrees freedom S Vn + m 2 2 2 2 S xi - (xi )/n + Yi - (yi) /n (n + m - 2) (33.74) S 25 + 8.74 0.636 (1.22) t 2.74 - 1.52 x 27 5.67 d.f. 53 6.36 55 for the 0.001 level, t = 4.523 therefore p = <0.001 Conclusion: There is significantly more nitrogen in the core material than the rim for a rimmed low carbon steel ingot.

128 Calculation 10. —Calculation of maximum nitrogen in terms of nitrogen associated with dislocations. Assume 0.0063 nitrogen maximum effect (from test data) 0.0063 gm N/14 0.00045.025 Atomic 100 gm Fe/58 1.7921 5 Atom 2.5 N 2 atoms N' 10,000 atoms Fe 8 x 103 atoms Fe cm2 metal x( 1 atom 1 atom Fe! cm2 metal x 2.861 x 10 c 1.23 x 101 cm2 Take a section 1 cm2 by 1 atom 1 2 2 1.23 x 11 - 2 2.03 x 1011 atoms N/cm2 1.23 x 10-1 8 x 103 There are from 108 to 1012 dislocations per cm2. (41) This iS of the same order as the number of atoms of nitrogen/cm2 assuming 1 atom of nitrogen per dislocation. If the dislocations' average length is 500 A (the width of a slip plane) and it is assumed that nitrogen must form an atmosphere along the complete length: 2.03 x 1011 2.200 = l 1 x 109 atmospheres/cm2 which is still of the same order as the dislocations.

LIST OF REFERENCES

130 LIST OF REFERENCES 1. ASTM-ASME Joint Committee. Compilation of Available High Temperature Characteristics of Metals and Alloys ASTM-ASME, Philadelphia, Pa., 1938, 2. Miller, R. Fo and J, J Hegero Report on the Strength of Wrought Steels at Elevated Temperatures Special Technical Publication No. 100, American Society for Testing Materials, Philadelphia 3, Pa., 1950, pg. 90 3. Cross, H, Co and J. Go Lowther, "Study of the Effects of Variables on the Creep Resistance of Steels," ASTM Proceedings, 1938, pgo 490 4. Beeghly, H, Fo "Determination of Aluminum Nitride Nitrogen in Steel," Analytical Chemistry, v, 21, 1949, pgo 1513, 5. Bardgette, W. Eo and M, Go Gemmillo "Causes of Variable Creep Strength in Basic Open Hearth Carbon Steel," Journal of the Iron and Steel Institute, v, 179, 1955, pgo 211o 6. Bardgette, Wo Eo Correspondence on "Abnormal Creep in Carbon Steels" by J. Glen, Journal of the Iron and Steel Institute, v. 157, 1947, pg. 579. 7. Clark, Co L, and A, E White, "Influence of Grain Size on High Temperature Characteristics of Ferrous and Nonferrous Metals," ASM Transactions, 1934, pg. 1069 8, Cross, Ho Co and J. Go Lowther, Second Progress Report on "The Study of Variables on the Creep Rate of Steels," ASTM Proceedings, v, 40, 1940, pg. 125, 9. Cross, H. C. and W Simmons, Third Progress Report on "The Study of Effects of Variables on the Creep Resistance of Steels," ASTM Proceedings, v, 44, 1944 10. Cross. H, C, and J. A. Van Echo, "Study of Effect of Variables on Creep Resistance of Steel," ASTM Proceedings, 1951, pg. 223. 11-. Glen, J. "Abnormal Creep in Carbon Steels," Journal of the Iron and Steel Institute, v, 155, pt. 4, April 1947, pgo 501.

131 12. Jenkins, C. H, M. and H. J, Tapsello "Factors Influencing the Creep Resistance of Wrought Steels," Journal of the Iron and Steel Institute, v 171, 1952, pg. 359. 13. Miller, R, F "The Strength of Carbon Steels for Elevated Temperature Applications," ASTM Proceedings, v. 54, 1954, pgo 964. 14. Smith, G, J. and E, J, Duliso "Effect of Manufacturing Practice on the Creep and Creep Rupture Strength of Low Carbon Steel," ASTM Preprint, 1949. 15. Glen, J. "The Creep Properties of Molybdenum, Chromium-molybdenum and Molybdenum-vanadium Steels," Journal of Iron and Steel Institute, Jan. 1948, po 37, 16. Miller, R. F. and N, J. Kearny. "Effect of Deoxidation Practice on the Creep Strength of C-Mo Steel at 850 and 1000~ F,," ASME Transactions, 1943, pgo 309. 17. Weaver, So Ho "The Effect of Carbide Spheroidization Upon the Rupture Strength and Elongation of Carbon Molybdenum Steel," ASTM Proceedings, 1946, v, 46, pgo 856. 18. Wilder, A. B, and 0. Light. "Long Time Elevated Temperature Tests of Cr-Mo Steels," ASM Transactions, pgo 323, 1951. 19. Wright, Eo C, "The Manufacture and Properties of Killed Bessemer Steel," Metals Technical Pub,, No. 1692, 1944. 20. Austin, C. R, and M. C. Fetzer, "Cementite Stability and its Relation to Grain Size, Abnormality, and Hardenability," ASM Transactions, 1941, pg, 339, 21. Davenport, E. So and E, C, Bain, "General Relations Between Grain Size and Hardenability, and the Normality of Steels," ASM Transactions, 1934, pg. 879. 22. Derge, G., A, R, Kommel, and R. F, Mehlo "Some Factors Influencing the Austenitic Grain Size in High Purity Steels," ASM Transactions, v. 26, 1938, pg. 153. 23. Dorn, J So and C, E. Harder. "Relation of Pretreatment of Steels to Austenitic Grain Growth," ASM Transactions, 1938, v,o 28, pgo 105-126.

132 24 Miller, Oo 0, "The Influence of Austenitizing Temperature and Time on the Grain Size," ASM Transactions, 1951, pgo 260o 25, Schuyter, J, "The Effect of Chemical Composition on the Coarsening of Austenite Grains," ASM Transactions, 1952, pgo 853. 26. Kitchener, J. A,, Bockris, J. O'Mo, Gleiser, Molly, and J. W, Evans, "The Solubility of Oxygen in Gamma Iron," ACTA Metallurgica, v, 1, Jan0 1953, p. 93. 27. Freeman, J,W. Discussion on "The Strength of Carbon Steels for Elevated Temperature Applications," ASTM Proceedings, v. 54, 1954, pg0 989. 28. Smith, G. Vo, R. Fo Miller, and Co 0, Tarr, "Structural Changes in Carbon and Molybdenum Steels During Prolonged Heating at 906 - 11000 Fo as Effected by Deoxidation Practice," ASTM Proceedings, v. 45, 1945, pg. 486o 29. Cottrell, A. H, "Theory of Dislocations", Chapter 2, Progress in Metal Physics, VI, Bruce Chalmer's Interscience Publishers, Inc,o New York, New York, 1949o 30. Cottrell, A. H, "Creep and Aging Effects in Solid Solutions," Symposium on Creep and Fracture of Metals at High Temperatures, Paper No, 8, National Physical Laboratory, 1954. 31. Leslie, W, C, and Ro L, Rickett, "Influence of Aluminum and Silicon Deoxidation on the Strain Aging of Low-Carbon Steels," AIMME Transactions, 1953, pg, 1021. 32. MacGregor, Co W. "True Stress Strain Tensile Test," Franklin Institute Journal, v. 238, 1944, pgo 111 330 Glen, J, "An Experimental Study of the Strength and Ductility of Steel at Elevated Temperatures," Symposium on Strength and Ductility of Metals at Elevated Temperatures, Special Technical Publication 128, AoS.ToM, Philadelphia, Pa., 1952, pg, 184. 34. Beeghly, H. F, "Behavior of Nitrogen and Some of its Compounds in Steel," Analytical Chemistry, v. 24, 1952, pg. 1095. 35. Ezekiel, M. Methods of Correlation Analysis, Sec. Ed., John Wiley & Sons, Inc, New York, 1947, pg. 67.

133 36, Freeman, Ho A. Industrial Statistics, John Wiley & Sons, Inc,, New York, 1949, pg. 47, 58-62. 37. Hayes, Ao and J. Chipmano "Mechanism of Solidification and Segregation in a Low-Carbon Rimming Steel Ingot," AIMME Transactions, v. 135, 1939, pgo 85-132. 38, Dijkstra, Lo J. and R. J. Slabecko "Effect of Alloying Elements on the Behavior of Nitrogen in Alpha Iron," AIMME Transactions, Jan. 1953, pg. 69. 39. Fast, J, D, "Strain Aging in Iron and Steel," Phillips Technical Review, v. 14, No, 2, 1952, pg. 60. 40. Orr, Ro Lo, Oo E. Sherby, and J. Eo Dorn. "Correlations of Rupture Data for Metals at Elevated Temperatures," ASM Transactions, v. 46, 1954, pg. 113 41. Barret, Charles So Structure of Metals, McGraw-Hill Book Company, Inc,, New York, N.Y,, 1952, pg. 402-403. 42. Darken, L, S, "Formal Basis of Diffusion Theory," Atom Movements, American Society for Metals, Cleveland, Ohio, 1951.~ 43o Bain, E, Co "Hardenability of Steel," A.SSST Transactions, 1932, pg, 385. 44, Bayertz, Mo "Effects of Initial Structure on Austenite Grain Formation and Coarsening," ASM Transactions, 1942, pgo 458, 45. Born, V K and W, Koch, "The Influence of Aluminum upon the Properties of Mild Unalloyed Steel," Stahl und Eisen, v. 72, No, 21, Oct, 1952, pg. 1268, 46, Cash, G. V., To M, Merril, and R. Io Stephensono "The Effect of Deoxidation on Hardenability," ASM Transactions, 1941, pgo 75, 47o Darken, Lo S,, R, P. Smith and Eo W, Filer. "Solubility of Gaseous Nitrogen in Gamma Iron and the Effect of Alloying Constituents - Aluminum Nitride Precipitation," AIMME Transactions, v, 191, 1951, pg, 1174, 48. Dennis, W. F. "Hetrogeneous Neucleation for Graphite in Hypereutectoid Steels," Journal of Iron and Steel Institute, v, 171, May 1952, pg. 59. 49, Dulis, E J. and G, V. Smith, "Roles of Aluminum and Nitrogen in Graphitization," ASM Transactions, v. 46, 1954, pgo 1318,0

134 50. Fast, J, D. "Aging Phenomena in Iron and Steel after Rapid Cooling," Phillips Technical Review, vo 11, 1949, pg. 211. 51. Holden, A. N, "Dislocation Collision and the Yield Point in Iron," Journal of Metals, 1952. 52. Kanter, J. E. and E, A, Sticha, "Studies on Susceptibility of Casting Steels to Graphitization," ASTM Proceedings, v. 46, 1946, pg. 182, 53. Leslie, W, C., R, L. Ricket, C. L, Dotson, and Co.S Walton, "Solution and Precipitation of Aluminum Nitride in Relation to the Structure of Low Carbon Steels," ASM Transactions, v, 46, 1954, pgo 1470. 54. Low, J, R. and M. Gensamero "Aging and Yield Point in Steel," AIMME Transactions, vo 158, 1944, pg. 207. 55. Mehl, Ro F. "The Structure and Formation of Pearlite," ASM Transactions, 1941, pgo 813. 56. Smith. G. V0, J. Ao MacMillan and E, J. Dulis. "Some Aspects of Graphitization in Steels," ASM Transactions 1951, pg, 692o 57. Wilder, A. B. "Graphitization of Steel at Elevated Temperatures," ASME Transactions, 1948, v. 40, pgo, 233-262. 58. Work, H, K and Enzian, G. H. "Effect of Deoxiation on Strain Sensitivity of Low Carbon Steels," AIMME Transactions,, v. 162, 1945, pgo 723.

UNIVERSITY OF MICHIGAN 3 9015 02539 7111