Search Documents

By E. Breinan, M. Salkind

Al3Ni whiskers were chemically extracted from unidirectionally solidified Al-A13Ni eutectic ingots, bent into loops, and heated for 0.1 to 10 hr at 320°, 415", and 510°C. The initial strains ranged from 0.003 to 0.055. In all cases, permanent plastic deformation was noted after heat treatment. The deformation consisted of relatively uniform bending at low stresses and temperatures and short times and kinking followed by fracture at high stresses and temperatures and long times. After kinking, the whisker segments adjacent to the kinks were found to have straightened, which is evidence of a dislocation condensation mechanism. The range of temperatures and strains at which time dependent plastic deformation was found indicates that creep of whiskers probably plays a role in the creep of A13Ni whisker-reinforced aluminum. WHISKERS may be defined as nearly perfect single crystals which exhibit high strength. Because they can support high stresses at relatively low strains, they have been successfully employed in reinforcing metals at both ambient and elevated temperatures. In studying the creep behavior of A13Ni whisker-reinforced aluminum at elevated temperatures,1,2 it was noted that the composites exhibited measurable creep deformation. This investigation of the creep relaxation of individual A13Ni whiskel, extracted chemically from the composite was initiated to determine if creep of whiskers could con. "bute to the overall creep of the composite material. Many observations of plastic deformation of metal and halide whiskers have been made. Brenner3-8 noted that copper, silver, and iron whiskers exhibited heterogeneous plastic deformation at room temperature when strained beyond their yield points. Gyulai9 and Gordon10 observed plastic deformation of relatively large (>3 µ) NaCl and KC1 whiskers, although the smallest, most perfect whiskers were completely elastic. Eisner" noted plastic deformation and microcreep of iron and silicon whiskers at room temperature after straining beyond the yield point. Whiskers reported to exhibit creep at stresses below the yield point were zinc1'-" and Silicon.15 Cabrera and price" observed some zinc whiskers which crept at room temperature after a short incubation period but then stopped creeping after a short time. Because some of their specimens exhibited no creep, they concluded that those whiskers that crept were relatively imperfect. Pearson, Reed, and Feldman15 observed similar creep behavior of silicon whiskers at 800°C. They also concluded that creep of the whiskers was a result of imperfections in their crystals. Brenner16 observed delayed failure of A12O3 whiskers at elevated temperatures but found no evidence of plastic deformation up to 2030°C (99 pct of E.EREINAN and M.SALKIND,JuniorMembers AIME,are Research Scientist and Chief, respectively, Advanced Metallurgy Section, United Aircraft Research Laboratories, East Hartford, Conn. Maunscript submitted April 5, 1968. IMD the melting temperature). Brenner also noted7 that some copper and iron whiskers exhibited delayed kinking above 350°C while others did not. One can conclude from these observations that small relatively perfect whiskers could exhibit completely elastic behavior during sustained elevated-temperature loading of composites. Since A13Ni whiskers tested in both bending and tension were found to exhibit no evidence of plastic deformation at room temperature'7'18 this study was initiated to determine whether or not creep of A13Ni whiskers occurred at the elevated temperatures at which creep in the composites was observed. Whiskers were chemically extracted from ingots of unidirectionally solidified A1-A13Ni eutectic, constrained in bending to various elastic strains and heat-treated. The bending constraints were removed after heat treatment and the amount of permanent set was taken as a measure of the time-dependent plastic deformation. EXPERIMENTAL PROCEDURES Ingots of eutectic Al-A13Ni containing nominally 6.2 wt pet Ni were unidirectionally solidified at approximately 11 cm per hr using a process described elsewhere.19,20 The starting materials were 99.99 pct pure. Cylindrical sections cut from the center of each ingot were placed in a 3 pct aqueous solution of hydrochloric acid and the whiskers were extracted as described previously.17 The whiskers nearest the surface were blackened somewhat due to overexposure to the acid while the center of the ingot was being dissolved These partially attacked whiskers were discarded. An intermediate zone of silver-gray-colored whiskers was collected and stored in methanol for use in relaxation experiments. Individual long pieces of A13Ni whiskers were placed on Fisher Precleaned Microscope Slides. These normally straight whiskers were bent elastically into arcs or loops of varying radii by manipulating their ends with a slender probe. The mass attraction between the whisker and the probe was sufficient to cause the whisker to follow the probe. The whiskers were retained in the elastic bend by the surface tension of a fine residual film on the slides. By using long whiskers, the action of the surface tension on the unlooped ends of the whisker allowed high elastic strains to be maintained in the loops. After each whisker was bent, a photomicrograph was taken for use in measuring the bending strain. The range of strains studied was 0.003 to 0.055. The bent whiskers were then encapsulated in Pyrex tubes at pressures between 10"6 and 5 x 10"6 mm of mercury and heat-treated at 320°, 415°, and 510°C (respectively 53, 61, and 70 pct of the peritectic decomposition temperature). After each heat treatment, the liquid film on the slides was found to have dried, but the whiskers were held in their original shapes by a residue on the slide. The minimum radius of curvature of each bent whisker was measured before and

Jan 1, 1969

By R. E. Smallman

R. E. Smallman (University of Birmingham, England)—Recently, Tedmon and Ferriss11 have determined the yield stress parameters oi and ky for tantalum by measuring the lower yield stress as a function of grain size 2d and fitting the results to a relationship of the form They report that although ky , which is taken to be a measure of the dislocation locking strength, is small (- 2 to 4 x 106 cgs units) a substantial yield drop is nevertheless observed in a normal tensile test. Niobium gives a similar result,12-14 as pointed out in the original work by Adams et a1.,12 and in order to check this apparent anomaly the yield-stress parameters of electron beam-melted niobium have recently been reanalyzed15 by the Luders strain technique. In this method the strain hardening part of the stress-strain curve is extrapolated to zero plastic strain; the intercept on the preyield portion of the curve is taken to give oi, whilst the difference between oi and the lower yield stress gives kyd-1/2. The results indicate that ky increases with increasing grain size and hence, a plot of vs d-112 yields an apparent ky, which is lower than the true value. A similar effect could account for the small ky found in the relatively pure tantalum used by Tedmon and Ferriss. The variation of ky with grain size shows that dislocations are more strongly locked in coarse-grained specimens than in fine-grained samples. In niobium, this may be attributed to the fact that the dislocation density in the fine-grained material is higher than that found in the coarse-grained samples which are given a sufficiently prolonged anneal to remove any residual substructure and, since the metal contains only a small amount of interstitual impurity, a variation in locking occurs. By contrast, application of both the grain size analysis and the Luders strain method to yield-stress data from commercially pure vanadium containing a large amount of interstitial impurity gives consistent values of oi and ky, with ky independent of grain size and temperature. Electron microscope observations show minor variations in dislocation density from grain size to grain size, but in any case in this material the dislocations are heavily locked with precipitate. On yielding new dislocations are generated and, as a consequence, the importance of any differences in dislocation density between the various specimens of different grain size is considerably reduced. It is perhaps significant that Adams and lannucci,16 working with a grade of tantalum containing a higher interstitial content than that used by Tedmon and Ferriss, prepared the specimens of different grain size by annealing in the temperature range 1500" to 2000° C to minimize any differences in dislocation structure, and found that ky had a value of 1.04 x 107 cgs units, independent of testing temperature. Such behavior is consistent with the dislocations being locked by carbide precipitates so that the generation of free dislocations is an athermal process. The recent work of Gilbert et al.17 also shows that in tantalum there is no significant variation of ky with grain size provided it contains 150 ppm of oxygen. In this case, however, the dislocations are not locked by precipitate and ky is temperature dependent. C. S. Tedmon and D. P. Ferriss (authors' reply)— We would like to thank Dr. Smallman for his interesting comments and discussion to our paper, "The Dependence of Yield Stress on Grain Size for Tantalum and a 10 pct W-90 pct Ta Alloy".18 It was suggested that perhaps the relatively small values obtained by us for ky of tantalum could be attributed to the same cause that accounts for the apparently small values of ky that result when it is determined by the Luders Strain technique. Since our values were obtained by plotting the lower yield stress vs the reciprocal of the square root of the grain size, it is not clear how this could be the case. The values of ky in this experiment have been calculated, using the Luders strain technique. With this method, values for ky on the order of 2 x 105 to 5 x lo6 cgs units were obtained. In spite of this rather large variation, the magnitudes are still small, and there appeared to be no good correlation between ky and the grain size or the yield stress, probably because of the difficulty in accurately extrapolating the work-hardening portion of the curve back to zero plastic strain. As was shown in the original data,18 there was little work hardening in any of the curves, at any temperature. In his discussion, Dr. Smallman also points out how ky has been observed to increase with increasing grain size, when determined by the Luders strain technique. There are at least two possible explanations for this. In the first case, if it is assumed that the bulk of the interstitial impurities are concentrated at the grain boundaries, then, of course, the available grain boundary area would decrease with increasing grain size, thus presenting less area for the interstitials, which would then presumably increase the concentration within the grains, thereby increasing the locking of the dislocations. In the second case, the increase in ky with increasing grain size would be attributed to the nature of the grain boundary itself. One of the several ways of deriving the Hall-Petch equation19 is based on the stress concentration arising from a pile-up of dislocations at the boundary. The ability of the stress concentration to unlock a source in a neighboring grain would depend on the strength of the grain boundary. As is well-known, the nature and struc-

Jan 1, 1963

By W. A. Backofen, D. L. Holt

Push-pull fatigue tests have been carried out at 4.2°K, 77oK, and room temperature on two poly crystalline materials of widely different stacking-fault energy (?): pure copper (? - 70 ergs per sq cm) and the Cu-8 wt pct A1 alloy (? - 2.8 ergs per sq cm). Constant stress-amplilude was imposed and measurement was made of the plastic-strain amplitude (ep) at saturation. Lives extended from 104 to 106 cycles. Designating lives at the various temperatures by NRT, N77, and N4.2. the ratios N77/NNT and N4.2/N77 ranged from 3.5 to 18 under the condition of common Ep . Metallo-graphic examination revealed different crack morphology in Cu-8 Al fatigued at room temperature, and at 77" and 4.2oK. At room temperature, cracks lay in or near grain and lain boundavies; at 77o and 4.2oK. cvacks were transcrystalline. Tests on single crystals of Cu-8 A1 showed that such a change in the cracking mode in polycrystallitle material accounted for a factor of- about 3.25 in N77/NRT . The longer life at lower tewperatztre (conslant cp) has heels attributed to two deuelopinents: a reduced production of the dislocation tangles and subgrain boundaries which serve as paths of rapid cracking, and suppression of oxygen chetni-sorption at the crack tip It was concluded that in both materials the luller accounted for an extension of the life at 4.2oK beyond that at room temperature by a factor of 15. XV ECENT experiments on the fatigue of Cu-A1 alloys in the so-called high-cycle range (greater than lo4 cycles) have emphasized the importance of stacking-fault energy (y) as a quantity affecting crack propagation rate and fatigue life.1,2 It was found in comparisons at essentially fixed plastic-strain amplitude that crack growth rate decreased by a factor of about 5 over the composition range from copper (? - 70 ergs per sq cm) to Cu-8 wt pct Al (? - 2.8 ergs per sq cm). The argument was made that, when stacking-fault energy is high, cross slip and climb are favored, so that dislocation tangles and/or subgrain boundaries form more readily under cyclic loading. Since the boundaries and tangles act as paths of rapid crack propagation ,3, 4 life is shortened as a result. However, when stacking-fault energy is reduced (as by alloying), cross slip and climb become more difficult, with the result that substructure formation is retarded and growth rate is also reduced. A purpose of the present work was to investigate the substructure effect in relation to temperature. As temperature is lowered, ? is varied only slightly (if at all), but decreased thermal activation can interfere with cross slip and climb. Thus substructure formation could be curtailed and life increased. Fatigue life in the high-cycle range is also known to be strongly influenced by environment. Working with copper, Wadsworth and Hutchings observed that life in a vacuum of 10-8 mm Hg exceeded life in air by a factor of 20.5 They isolated oxygen as the agent that furthered cracking. While the details are still unclear, a requirement in any mechanism of oxygen-accelerated cracking is that there be chemisorption at the crack tip. That could prevent welding on the compression half cycle,= interfere with reversal of slip,1, 6 or aid in breaking metal-metal bonds at the crack tip.5'7 In the work being reported here, temperature was lowered by immersion in liquid nitrogen and helium, which also served to reduce both the oxygen concentration and chemisorption rate. A possible effect upon life, i.e., a lengthening, had to be recognized. Several researchers have determined fatigue lives at low temperatures presenting their results in the form of stress amplitude (S) vs cycles in life (N) curves.8-11 Such curves reflect, primarily, the fact that metal is strengthened by lowering temperature; effects of substructure and changing environment tend to be masked. The difficulty can be overcome by comparisons based on identical plastic-strain amplitudes, and in the present work the dependence of life on both plastic strain and stress amplitude was established. EXPERIMENTAL Materials. The principal materials were polycrystal-line copper (? - 70 ergs per sq cm)" and the Cu-8 wt pct Al alloy (? - 2.8 ergs per sq cm),I3 the latter being near the limit of solubility of aluminum in copper and having, therefore, the lowest stacking-fault energy in the CU-Al system. Specimens were machined from 0.118-in.-diam cold-swaged rods of high-purity (99.999 pct) copper and the Cu-8 Al alloy, the latter produced initially in a graphite boat by induction vacuum melting a mixture of 99.999 pct Cu and 99.99 pct Al. The machined specimens were annealed to produce mean grain diameters of about 0.070 mm in copper and 0.190 mm in the alloy. Specimen dimensions are given in Fig. 1. Values of the tensile yield stress, ultimate strength, uniform strain (determined by the Considgre construction), and reduction of area, for both materials at 4.2oK, 77oK, and room temperature, are listed in Table I. The tensile apparatus in which these results were obtained has already been described.14 Apparatus. Specimens were fatigued in push-pull with a machine that is illustrated schematically in Fig. 2. The specimen is first soldered into the top grip (1) with Woods metal, and the grip is then screwed into the inner tube (2) which is connected to the drive rod of the Goodmans vibration genera-

Jan 1, 1968

By A. J. Mortlock

Cobalt and nickel have been diffused at tracer concentrations in gold at several temperatures in the range from approximately 700° to 950°C. The diffusion penetration profiles were determined by a serial sectioning technique in which the gold is first anodized and then the anodic layer is dissolved in acid. In this ulay sections as thin as 250A could be removed reproduci-bly. In all cases, the region close to the specimen surface was characterized by irregular behavior in the sense that the logarithm of concentration was not linear in the square of the penetration distance. In sotne cases, there zuas an indication of the operation of very slow dijfusion in this region, while in others the apparent diffusion coejj'icient was negative. Possible reasons for this anomalous behavior are briefly discussed. In recent years it has been found that the region close to the surface of a metal can sometimes exhibit anomalously slow diffusion characteristics relative to the interior of the metal. One of the best examples of this fact is the work of Styris and omizuka,' who showed that the apparent diffusion coefficient for zinc in the region withi: about 1 p of the free surface of copper was about ,,,, that at deeper penetrations. This result is particularly interesting, because it is free from the possibly complicating effects of low solubility of the diffusing tracer in the solvent metal. In the case of diffusion under conditions of low solubilitjr, interpretaticn of the results in terms of lattice diffusion is difficult because of the enhanced short-circuiting produced by segregation to dislocations.2'3 Measurements by Duhl et 1. suggest that cobalt diffusing in gold may also show a near-surface effect of this type. Once again the solubility is high, so that this result could be of great interest. However, the technique used for analyzing the diffusion penetration zones by Duhl, viz. the counting of residual gamma activity in the specimen following sectioning, appears to have indicated a near-surface effect in a parallel experiment on the self-diffusion of gold reported at the same time. The latter result is known to be spurious, since Kidson5 has demonstrated that self-diffusion in gold does not show this effect. Duhl et 01. also reported some measurements on the diffusion of nickel in gold, but failed to give any data for the near-surface region. As the solubility of nickel in gold is high, such data would also be of special interest. We, therefore, decided to conduct another set of experiments on the diffusion of nickel and cobalt in gold, using a sectioning technique that allows the individual sections to be assayed for solute content and thus gives direct determinations of penetration profiles. Also, by sectioning with an anodizing/stripping tech- nique, very thin layers can be removed and the region close to the surface studied in detail. MATERIALS The gold specimens were supplied as single crystal disks $ in. in diam by a in. high by Monocrystals Co. of Cleveland, Ohio. The gold itself was of spectro-scopic purity, i.e., better than 99.99 pct pure. METHOD Specimen Preparation. One flat end face of each gold crystal was spark planed with a Servomet spark erosion machine set for minimum spark energy. Following this treatment the crystals were preannealed for 2 to 4 days at temperatures of either 400" or 700°C. The three crystals preannealed at 700°C showed signs of recrystallization. The spark-planed end face of each crystal was then coated with the appropriate amount of 63i or 60 radioactive tracer. This deposit was laid down in a simple plating bath containing the as-supplied solution of the radioactive isotope as well as sufficient ammonium oxalate to saturate the solution. Some ammonium oxalate remained undissolved on the floor of the bath for this purpose. During plating further additions of ammonium oxalate were sometimes required to allow the plating to continue satisfactorily, perhaps due to passivation of the undissolved oxalate already present. The thickness of the deposited layer was determined by comparison of the apparent surface activity of the plated specimen with that of a similar specimen having a weighable deposit of the isotope on its end face. Correction for self-absorption of the radiation was made in this calculation. Annealing. The deposited crystals were annealed in a hydrogen atmosphere in sealed silica tubes. During this heat treatment they were supported, active face down, on optically flat silica plates. The temperature was measured with calibrated Pt vs Pt-10 pct Rh thermocouples, and the tabulated values can be taken to be correct to Z°C. All the crystals showed evidence of recrystallization following these heat treatments, suggesting that initially they may not have been good single crystals or had suffered strain during delivery. Concentration Profile Analysis. After annealing, the crystals were sectioned by the anodizing-stripping technique.6 The anodizing involved suspension of the specimen with its cylindrical axis horiz6ntal by a gold wire in a 200-ml beaker containing 1 M Hg304. A cathode in the form of a strip of gold sheet, 2 in. wide and positioned to be in contact with the curved side of the beaker, completely encircled the specimen. An anodizing current of 30 ma, corresponding to a current density of 5 ma per sq cm on the surface of the specimen, was passed for times ranging from 5 to 150 min depending on the thickness of gold to be removed; the solution was stirred continuously during this process. Following this treatment, the specimen

Jan 1, 1969

By M. Gell, G. R. Leveran

The high- and low-cycle fatigue properties of the nickel-base superalloy, Mar-MBOO, in columnar-grained and single-crystal forms were determined at room temperature. It was found that the fatigue lives of these materials were greatly affected by the size of preexisting cracks in MC-type carbides contained in the micro structure. Most of the data falls on two curves given by: (zN)'/A€= K, where Nf is the number of cycles to failure, Af is the total strain range, and K is a function of carbide size. No difference was observed in the fatigue behavior of the columnar-grained and single-crystal materials for the same MC carbide size. Matrix slip and crack initiation occurred at precracked MC carbides and, to a lesser extent, at micropores. Fatigue crack propagation was mainly in the Stage I mode, i.e., on cry stallo graPhic slip planes. The Stage I fracture in these materials was unusual in that distinct features were observed on the fracture surfaces. In high-cycle fatigue, these features resembled those commonly observed on the cleavage fracture surfaces of bcc and hcp materials. Yet, in this study, the cracks propagated slowly in a cyclic manner. In low-cycle fatigue, the Stage I facets contained equiaxed dimples, similar to those observed on the tensile fracture surfaces of ductile materials. These observations indicate that both local normal and shear stresses are involved in these Stage I fractures. A model is proposed to explain these results based on the weakening of the cohesive energy of the active slip planes by reversed shear deformation and the fracture of the bonds across the weakened planes by the local normal stress. RECENT developments in casting technology have produced cast nickel-base superalloys in columnar -grained and single-crystal forms.1'2 The tensile and creep properties of the nickel-base superalloy, Mar-M200, cast in these forms have been shown to be superior to the corresponding properties of the conventionally cast polycrystalline material.lp2 This improvement in properties results, in part, from the elimination of grain boundaries in the single crystals and the alignment of the grain boundaries parallel to the stress axis in the columnar-grained castings. As part of a program to evaluate the fatigue properties of nickel-base superalloys cast in single-crystal and columnar-grained forms, a study has been made of the cyclic deformation and fracture of Mar-M2OO at room temperature. M. fiFl I .hininr Mpmher AIMF ic ^pninr Rocoarrh Accn^iata anH I) EXPERIMENTAL PROCEDURE The composition range of Mar-Ma00 in weight percent is: 8 to 10 Cr, 9 to 11 Co, 11.5 to 13.5 W, 0.75 to 1.25 Cb, 1.75 to 2.25 Ti, 4.75 to 5.25 Al, 0.01 to 0.02 B, 0.03 to 0.08 Zr, 0.07 to 0.12 C, bal. Ni. All of the castings met the above specifications. The castings were solutionized for 1 to 4 hr at 2250°F followed by aging at 1600°F for 32 hr which resulted in a 0.2 pct offset yield stress of 150,000 psi at room temperature. The microstructure of the material consisted of cuboidal, coherent particles of ordered, fcc Ni3(A1,Ti) (commonly designated y'), approximately 0.3 p on edge, distributed in an fcc y solid-solution matrix. MC carbides together with shrinkage and gas micropores were also distributed throughout the materials. The MC carbides and micropores were located preferentially in the interdendritic interstices, as well as in the grain boundaries in the columnar-grained castings. The (100) direction of all the single crystals and the common (100) axis of the grains in the columnar materials were aligned within about 5 deg of the specimen axis. Fatigue testing was carried out in the high-cycle (HCF) and low-cycle (LCF) fatigue regions, with the major difference being gross yielding of the specimen occurred during the first cycle in the LCF region. This division also corresponded with the more usual one in which the life of a specimen in LCF is less than lo4 cycles and that in HCF is greater than lo4 cycles. The designs of the high-cycle fatigue and low-cycle fatigue specimens are shown in Figs. l(a) and (c), respectively. The gage sections of both HCF and LCF specimens were electropolished prior to testing. The HCF specimens were tested in an MTS, closed-loop, hydraulic fatigue machine at 10 cps in air. The specimens were cycled between a tensile stress of 5000 psi and a maximum tensile stress which ranged from 35,000 to 125,000 psi, Fig. l(b). The LCF specimens were cycled under strain control from zero to a maximum tensile strain, Fig. l(d), in a Wiedemann-Baldwin testing machine. The experimental procedure has been described elsewhere.3'4 Both HCF and LCF tests were interrupted periodically in order to replicate the development of slip and cracking at the specimen surface. This was accomplished by placing plastic replicating tape around the gage section of the specimen while the specimen was in the mahine. The size of the MC carbides for all specimens was measured on a polished longitudinal section through the gage section after fatigue testing. The method of measurement consisted of carefully scanning the entire polished section in order to locate the largest MC carbides. Photographs were then taken of the six longest carbides oriented approximately normal to the

Jan 1, 1969

By Frank E. Woolley, John F. Elliott

Mathematical models of the transient thermal behavior of a high-temperature solution calorimeter1-3 have been developed. The thermal behavior of the calorimeter is appoxirrzated by linear lumped-parameter models, and hence is described by sets of linear ordinary differential equations with constant coefficients The response of the models to various inputs is shown to agree with the response of the real system. Application of the modeling to experimental design and analysis of data illustrates the usefulness of simple models of complex systems. The early eperiments1,2 with the high-temperature solution calorimeter indicated that the change in the temperature of the bath resulting from the addition of a solute sample to the bath involved not only the direct effect due to the solution process but also possibly a secondary effect arising from the change in coupling between the bath and the induction heating coil. Consequently, an extensive analysis of the calorimeter was carried out, and models of the transient thermal processes of the instrument were developed to aid in improving the design and interpreting the behavior of the system. This paper describes the dynamic modeling; the use of it in treating experimental results has been reported earlier.3 The high-temperature solution calorimeter was constructed to measure directly the partial molar heats of solution of solute elements in a variety of liquid metal solvents.1-3 The calorimeter consists of an induction-heated liquid metal bath into which small samples of a solute element can be dropped. The bath temperature is recorded continuously, and the change in the measured bath temperature with time, dTm = f(t), resulting from the solute addition are the raw data from which the enthalpy change caused by the addition is determined. To extract the rmodynamic results from the data, the temperature change must be compared with that resulting from calibration additions of known enthalpy change. Accordingly, it is necessary to understand the transient thermal processes arising as a result of the addition to the bath. Neither modeling nor experimentation alone could provide the required insight into the working of the calorimeter. The alternate use of both methods in conjunction greatly assisted the design of the equipment and experiments, and the interpretation of the data. THE PHYSICAL CHARACTER OF THE SYSTEM The essential parts of the calorimeter, Fig. 1, for model studies are the thermocouple, the liquid metal bath and the surrounding refractories. The system is the solvent metal bath and those refractories around it which undergo a temperature change as a result of an addition to the bath, and which determine the way the temperature of the bath responds to an input. The inputs are the combined transient thermal effects arising when an addition is made to the bath. They include the thermal effects of the addition itself and the results of changed coupling between the bath and the induction coil. The response is the variation in the measured bath temperature, dTm(t) = Tm(t) - Tm(O), from an initial steady state resulting from the inputs. It was assumed in this study that the physical properties of the various elements of the system are independent of the inputs and time, although these properties may vary as the result of changes in the composition and size of the bath during a series of additions. This separation of inputs and the system is equivalent to assuming that the system is linear, i.e., that its behavior can be described by linear differential equations with constant coefficients. Linear behavior can be expected whenever the departure of each portion of the system from its steady-state condition is small enough to cause negligible changes in the thermal properties of the materials and in the various heat-transfer coefficients. Radiative heat transfer is important in this system, so the assumption of linearity should be valid only for small temperature deviations. Several conclusions were drawn from operation of the calorimeter in earlier experimental studies: 1) Radiative heat transport from the top of the bath is a significant portion of the total heat lost from the bath. However, for small changes in the bath temperature the change in transport by this path could be assumed to be proportional to the change in the bath temperature. 2) A very small portion of the heat input is lost through the thermocouple to its water-cooled holder. The thermal resistance and thermal capacity of the thermocouple protection tube are small, so the temperature of the thermocouple should follow closely that of the bath. 3) The remainder of the total heat lost from the bath will pass by conduction through the crucible to, and through, the other refractories, eventually being absorbed by the water-cooled induction coil or by the water-cooled sides and bottom of the enclosure. 4) The thermal resistance between the bath and crucible is very small. Thus the thermal capacity of the crucible will affect the temperature of the bath very soon after an addition of heat to the bath. 5) The thermal resistance between the crucible and the silica sleeve is large, especially if a radiation shield is placed in the gap. The effect of the thermal capacity of the sleeve thus will be significant only at longer times. The thermal resistance through the packing below the crucible also is large, so the packing and the silica sleeve will have similar effects on the behavior of the system. 6) A large temperature drop exists across the gap containing the water-cooled induction coil. Thus for relatively small changes in the thermal input to the bath, the refractories beyond the sleeve

Jan 1, 1970

By R. G. Davies, P. Beardmore, T. L. Johnston

The flow stress of a series of Ni-Cr-A1 alloys consisting of a dispersion of y' (based on Ni3Al) in a rnatrix of nickel-base solid solution y has been measured at temperatures up to 950°C as a fwzction of the volume fraction of y'. At high temperatures the flow stress is controlled by the amount of Y' in the alloy, i.e., the higher the volume fraction of y', the greater is the flow stress. This simple relationship is not obeyed at low temperatures in so far as a peak in the flow stress-volume fraction relation occurs at about 25 pct y'. The variation in the mechanical properlies of these alloys as a function of both temperature and volume fraction of y' has been correlated with changes in distribution of both the dislocations and y'. The results are interpreted on the basis that at low temperatures the y matrix is strengthened significantly bv the presence of a hyperfine y' precipitate due to decomposition on cooling; at high temperatures the y matrix is a single phase of low strength. It is clearly recognized that the high temperature strength of most nickel-base superalloys depends upon a dispersion of the ordered fcc phase y', based on Ni3A1, in a fcc solid solution matrix y based on nickel. Although the volume fraction of y' varies widely from about 0.2 in Nimonic 80A to about 0.6 in Mar-M200, all such nickel-base alloys manifest an unusual insensi-tivity of the flow stress with respect to temperature. In Mar-M200 for example, the 0.2 pct flow stress remains essentially constant from room temperature to 750°C. The conclusion has been drawn1 that the characteristically low temperature dependence of the flow stress of y-y' nickel-base alloys is obtained when the state of dispersion of y' is such that dislocations are forced to cut through the y' particles at the onset of yielding. When the spacing between the y' particles is so large that the flow stress is controlled by dislocation bowing between particles, then the initial flow stress decreases progressively with an increase in temperature at a rate determined by changes in elastic properties. The same conclusion is inherent in the detailed, mechanistic model of the deformation process in commercial superalloys which has been developed by Copley and ear' in which the temperature independent flow stress is attributed primarily to the contribution of the antiphase boundary energy created in the y' particles during deformation. In this theory the temperature insensitivity of the flow stress is a reflection of the constant antiphase boundary energy as a function of temperature. An important microstructural parameter that is relevant to the explanations that have been suggested' to account for the temperature insensitivity of the flow stress is the volume fraction of y'. To vary the latter to any significant extent in a given commercial alloy is clearly difficult. However, it is possible in a relatively simple Ni-Al-Cr ternary system which manifests analogous microstructures in terms of the distribution of y' in y and contains specific alloys which have flow properties that depend on temperature in a manner quite similar to their more complex commercial counterparts. Hornbogen et . have studied precipitation phenomena and deformation mechanisms in such alloys but only where the y' volume fraction was small (less than 0.2) and the y' particle size varied from less than 100A up to a maximum of -1000A. In the present study, a series of alloys was prepared in which the volume percent of y' at 900°C was varied from 0 to 100 pct with the y' particle size (of the order 0.5 p) comparable to the sizes obtained in commercial superalloys. Particular attention has been given to the relationship between variations in the volume fraction and distribution of y' and the temperature dependence of the flow stress EXPERIMENTAL TECHNIQUES The Ni-Cr-Al system was selected because it is well characterized, bears a close relationship to commercial alloys, and offers the advantage of an extra degree of freedom over a binary system. In the present investigation, a series of alloys across the tie line between NisA1 and Ni3Cr (Ni3Cr is not an in-termetallic compound, the nomenclature is only used to designate the composition) were vacuum cast. The pseudobinary6 and the composition of the alloys used are shown in Fig. 1. It is important to note that the compositions of the y phase and the y' phase in the two-phase alloys was always the same. Alloy compositions were selected from the binary diagram, Fig. 1, in order that aging at 900°C would produce from 0 (100 pct y) to 100 pct y' by volume percent. (The size of the y' particles produced during the equilibrium aging treatment increased as the volume fraction of y' increased, ranging from about 0.2 p at low volume fractions up to about 0.8 p at the highest volume fraction.) The y' phase is based on the inter-metallic compound Ni,A1 which has the fcc LIZ type superlattice structure, and chromium substitutes for aluminum in the structure. The y phase is a disordered fcc solid solution. The alloys were heat treated at 1150°C for 2 hr, air cooled to room temperature, and finally annealed for 16 hr at 900°C. The rods were then centerless ground to 0.25 in. diam and cut into compression samples 0.5 in. long. The compression tests were made on an In-stron machine at a strain rate of 7 x 10"4 sec-'. A rapid heating radiant heat furnace was used which minimized the heating and temperature stabilization time to 10 min for the highest testing temperature. All the tests were stopped after 5 pct plastic strain.

Jan 1, 1970

By N. J. Grant, B. C. Giessen, M. Morris

The slvuclures of alloys in the system In-Bi have been investigated after (levy vapid queuching from the mell (splat cooling) to -190°C. Tuo-phase fields could be suppressed over most of the tota1 concentvalion range; five melastable phases a1, a2, ?, ?1. and ß exisl, of which four have simple elementtike structures. For example, ß has the A5 struclure of white till. Crystal-tographic data and coordination numbers for these phases aye given, as well as addilional information on the equilibrium diagram, where a new phase In5Bi3 teas found. THE phase diagrams of binary combinations of B elements from the groups B2, B3, B4, and B5 are usually of a simple type and show considerably fewer intermediate phases than combinations of these elements with A metals, transition metals, or the B1 noble metals. Aside from phases such as the typical B3-B5 compounds with the ZnS-B3 structure type, few stoichiometric phases are known; among them are In2Bi and 1nBi.l Nonstoichiometric phases include ß (In-Sn), a1, (In-Pb), and the tin-rich ? phases in the In-Sn, Cd-Sn, and Hg-Sn systems.' The first two phases have a close relationship to indium, while the latter group are known to be structurally related to a Sn. Thus, there are generally no immediate experimental data for the correlation of the crystal structures of B metals and B metal combinations with certain monotonously varying parameters, such as the valence electron concentration (VEC), or the average atomic size. Such parameters have been recognized as structure determining, e.g., for the transition 4 Sn — y phase,2,3 where nonintegral valence electron concentrations are encountered. It should be possible to extend such correlations to other binary systems, if nonequilibrium single-phase alloys could be produced over broad valence electron concentration ranges, and if their crystal structures could be regarded as being imposed by their electronic states. It has been shown in the case of the tin-rich 7 phases3 that alloy phases with typical crystal structures can be produced in B metal alloys rapidly quenched from the liquid to -190°C by the splat cooling technique due to Duwez,3-8 and reviewed in Refs. 5 and 6. The In-Bi system was selected because it extends over a valence electron concentration range in which several metastable, nonstoichiometrir phases could be expected to occur. Further, the low melting points of the known intermediate phases, In2Bi and InBi, Fig. 1(a), indicated low binding energies and thus possibly low driving forces for the formation Of the equilibrium structures. EXPERIMENTAL TECHNIQUES AND RESULTS The preparation of the quenched foils followed the practice described in Ref. 3. Master alloys were produced by melting of indium (99.97+) and bismuth (99.99+) in evacuated Vycor capsules or in an inert gas arc furnace; quantities of 20 mg were splat-cooled onto copper and silver substrates held at - 190°C; and crystal structures and lattice parameters were determined on a GE XRD-5 diffractometer using Cu Ka, radiation at -190°C and at room temperature. The duplication of each run with both substrates permitted the elimination of overlap of the substrate diffraction pattern and that of the investigated substance. The XRD patterns were usually taken from sin2 " = 0.03 to 0.35; the substrate was used as a means of internal calibration. The fractional accuracy of the lattice parameters of metastable phases is approximately 10-3 In the following, all percentages are in atomic percent. The stable and metastable phases found after rapid quenching to -190°C are listed in Table I, together with estimated concentration ranges and crystallographic data. The investigated alloys and phases present in them are given in Table 11. Except for the region between indium with 33 and 50 pct Bi, where a revision of the equilibrium phase diagram became necessary to include a new stoichiometric phase In5Bi3, the diffraction patterns taken after heating to room temperature agreed with those expected from the phase diagram, Fig. l(a). This suggests that the new nonstoichiometric phases are not stable at room temperature, thus following the observations made for the y phases based on tin.3 Results concerning the equilibrium phases In2Bi and In5Bi3 and the revision of the equilibrium phase diagram which is made necessary by the inclusion of In5Bi3 will be treated first. The Crystal Structure of In2Bi. Makarov7 had identified In2Bi as belonging to the AlB2-C32 type ; however, in a later paper the structure was revised.8 In2Bio was found to be of the Ni21n-B8ß type,21 with a = 5.496A, c = 6.57.9A, and N = 6 atoms per unit cell.9 This crystal structure was confirmed in the present work; definite evidence for the doubling of the c axis, as compared to the A1B2-C32 type, and for the proposed order structure, was found in the powder pattern. The observed lattice parameters agree well with those of Ref. 8; those measured at -190°C are given- in Table I. The Crystal Structure of In5Bi3. A new, equilibrium, intermediate phase was identified in slow-cooled, as-cast alloys; it was identified as In5Bi3. The structure has been worked out by Giessen and Grant;" lattice parameters at -190°C are given in Table I. This phase is probably identical with "In3Bi2", produced by vapor deposition techniques, but not recognized as an equilibrium phase by Palatnik et a1.11 After completion of the present work, the existence of In5Bi3 has also been demonstrated by superconductivity measurements on In-Bi alloys.20

Jan 1, 1968

By M. S. Azun

M.S. Azun I have many objections to the content of the author's paper. Before discussing it, however, I would like to repeat the property of semivariogram function. Second order stationary properties of regionalized variables (ReV's) such as semivariogram function ?(h) are perfectly known in geostatistics. Also, the kriging equations in the language of mathematical statistics using second order stationary properties are well understood. However, the way to use the sample (estimated) semivariogram function in any one of the kriging procedures is vague. The sample semivariogram function is given as follows: [1 N-hy*(h) = 2(N h) i21 {Xi-Xi+h}Z, h=0, 1, N-1] where N is the total number of samples, Xi is the sample value at the i - th location, X i+h is the sample value at the i +h - th location, and h is the distance among the samples. An estimation variance of sample semivariogram function of first lag is smaller than that of higher order lag. The theoretical semivariogram function reaches the variance of samples asymptotically. But this is not easily observable because of the larger variation involved in the estimate of semivariogram function. In general, an estimation procedure is done for h = 0, 1, 2,…., up to the greatest integer less than N/2, even though sample semivariogram function can be computable through N-1. After estimating semivariogram function, the critical question of how to model sample semivariogram function arises. As seen in the above equation, sample semivariogram function is discrete and can be smoothed by the model being selected. Therefore, modeling of sample semivariogram function is the most important step in geostatistics. It not only smoothes a discrete function but also affects the results of the kriging procedure. When the only aim is to model the semivariogram function, which is the basic point of the author's paper, one can employ any fitting techniques, such as curve fitting, or any ar¬bitrary functions, which are called submodels in the paper. The term "arbitrary function" is used rather than "submodel" because there is no basic understanding of developing them. The author suggests that the sum of those submodels can also be used for the modeling of sample semivariogram function. The combination of any arbitrary functions brings many problems instead of giving an insight of the domain structure considered. The author used two arbitrary functions and the nugget effect in response to sample semivariogram function (Fig. 10). For the same example, he stated that the parameters involved in the mixed arbitrary function model can be accepted when the discrepancy between sample semivariogram function and the model is small visually. For verifying the fitting behavior of any selected model, one should not be contented with the visual satisfactory. Some statistical measure such as goodness of fit has to be used. The author's practice is no more than an exercise in curve fitting without any fundamental understanding or conceptualization of the underlying physical mechanism. Furthermore, the selection of any model is not an easy task if the purpose is the search for the "best" response to the observed second order properties of ReV's. I suggest that the Markovian model (Azun, 1983), on the basis of a theoretical understanding of underlying mechanism, which gives more information about the occurrence of regionalized variables, is used to respond all properties of ReV's. There are a lot of problems for modeling of onedimensional sample semivariogram function. Thus, it is not appropriate to go to higher order dimensional sample semivariogram function modeling. In the meantime, I would recommend that one can connect the values of standardized sample semivariogram function rather than simple values of semivariogram function in the two-dimensional estimation. The standardized values can be computed in dividing the semivariogram function value by the number of sample pairs involved in each lag regardless of the directions. In conclusion, geostatistics is an interdisciplinary area in mining that uses the principles of mathematical statistics. Thus, it should not violate any probabilistic and statistical rules. When Matheron was developing the theory of geostatistical study in the early years of geostatistics, many mining people had a reservation accepting the geostatistical tools. However, this does not mean that we, the geostatisticians, might try to convince those people using some "strange" tools or rules as some authors implied (Baafi and Kim, 1984). Instead, we have to develop and explain the geostatistical tools staying only in the framework of statistical concepts and properties. ? References Azun, M.S., 1983, "Stochastic Process Modeling of Spatially Distributed Geostatistical Data," Columbia University, Ph.D. Thesis. Baafi, E.Y., and Kim, Y.C., 1984, "Discussion - Comparison of Different Ore Reserve Estimation Methods Using Conditional Simulation," Mining Engineering, Vol. 36, No. 3, p. 280. Reply by J.M. Rendu The interactive method proposed by Rendu allows practitioners to develop semivariogram models that take into account not only the numerical information obtained by sampling, but also highly significant additional information that often cannot be quantified. The geology of the deposit - including hypotheses concerning its genesis, sampling methods, assaying methods, and mathematical methods used to calculate the semivariograms - all have an influence on the numerical results obtained and on how these results should be interpreted. If all the information concerning the spatial distribution of values in a mineral deposit was contained in the sample values, it could be argued that statistical techniques alone would produce optimum models. However, this is rarely, if ever, the case. Methods that allow the user to take into account his experience and his geologic understanding of the deposit should not be rejected for the sake of theoretical statistical purity. ?

Jan 1, 1986

By F. W. Luerssen, J. F. Elliott

F LUSHING the early slag from a stationary open Fhearth having a high percentage of hot metal in its charge is necessary in order to remove silica from the system. The flush slag is strongly oxidizing and is somewhat acidic. It has, however, considerable capacity to extract phosphorus from the bath and it also removes considerable manganese. It seems probable that factors which control the distribution of phosphorus and manganese between slag and metal in the refining period also should be dominant in the flush and postflush periods. Several studies, as summarized elsewhere,1,2 support the viewpoint that conditions closely approaching equilibrium for these elements are rather readily established during the refining period. Over the years these studies have repeatedly demonstrated that 1—high slag v01ume, 2—low bath and slag temperature, 3—basic slag, and 4—strongly oxidizing slag favor rapid elimination of phosphorus from the bath to the slag. They also show that the following conditions favor retention of manganese in the bath: 1—low slag volume, 2—high bath and slag temperature, 3— basic slag, and 4—minimum oxidizing power of slag. When it is considered that the flush slag often carries as high as 75 pct of the manganese charged and only 25 to 60 pct of the phosphorus charged, it is evident that in removing silica much manganese is sacrificed but phosphorus removal is far from conplete. Because of overriding circumstances, this is accepted in most operations and actually it is considered to be inevitable. This may account for the fact that little attention has been paid to conditions affecting the elimination of phosphorus and manganese in the flush slag. A recent study of the behavior of various charge oxides has developed considerable information on the flush and postflush periods. Because the data are felt to be of general interest, they have been brought together and Presented in this paper. The object is to show the various factors in the flush and postflush periods which influence elimination of phosphorus and manganese. Physical Conditions During and After Flushing Physical conditions existing during the flush vary from plant to plant, from shop to shop, from furnace to furnace, and even from heat to heat. They are strongly influenced by the physical and chemical character of the charge oxide which is ordinarily necessary to provide sufficient oxidizing power early in the heat. Invariably the period is characterized by a vigorous reaction between the principal re-actants: the hot metal being added and the charge oxide. During the flush, it is probable that the slag acts to some extent as an oxidizer; but, because of the critical influence of the behavior of the charge oxid'e on flushing action, it seems apparent that the oxide itself is the dominant oxidizer. Fig. 1 shows the course of two heats which were selected as being typical of the group studied. Heat A was charged with 55 pet hot metal, based on the total metallics charged, and heat B had 57 pct hot metal. As indicated in Table I and Fig. 1, the melt-down slag, which is not usually voluminous and which is principally FeO, expands greatly in volume and will show rather high levels of SiO2, MnO, and P2O5 very soon after the beginning of the hot metal addition. Simultaneously, large volumes of CO are liberated which cause violent mixing of slag and metal. It is of interest to note that the time required to bring carbon down to a low level is very much longer than that required for the removal of silicon, manganese, or phosphorus. At the end of flush, carbon in the bath is still approximately 2 pct. When strongly reducing hot metal is brought into contact with strongly oxidizing conditions within the furnace! it is probable that the rate of mass transfer to the slag (and atmosphere) of silicon, manganese, phosphorus, and carbon initially depends principally on the rates at which the two participating phases are brought into contact That is, it depends on the nature of the various reactions. Later in the flush period, when the scrap is virtually all dissolved and the action of the bath has settled down to a steady and somewhat gentle boil, it is likely that other factors, such as the transfer of oxygen across the slag-metal interface, become dominant. The temperature of the slag-metal system is far from uniform. Heat is being driven by the flame down through the slag. Bubbling and surging of the metal also frequently brings portions of the bath in contact with the flame. At areas of contact between the ore and liquid metal, or slag and liquid metal, the oxidizing reactions generate much heat. On the other hand, scrap is being melted which tends to absorb large quantities of heat. Because the liquid bath is high in carbon, the steel scrap is brought into solution rapidly. This can proceed at a rather low temperature; and until much of the scrap has been taken into solution, the bath temperature would not be expected to increase appreciably. Consideration of these factors leads to the conclusion that during the flush period the slag should be rather hot and the bath relatively cold. Both observation and temperature measurements bear this out. Experimental Data The extended program of charge oxide evaluation permitted study of the widely varying conditions existing during the flushing period. Slag and metal analyses and bath temperatures reported herein (Tables I and 11) were obtained toward the latter portion of the work. Four different types of charge oxide, sinter, two types of hydraulic cement-bonded soft ores, and a pyrobonded agglomerate were used in the study. Although the heats reported were from only one 205 ton furnace, they show variations in flush slag analyses all the way from 25 pct FeO, which is typical with the use of a hard natural charge ore, to 45 pct FeO which resulted when a very poorly agglomerated fine ore was used. The physical behavior of the flushes showed a correspondingly wide variation from well controlled reactions to violent surges following periods of inac-

Jan 1, 1956

By R. F. Mehl, C. E. Birchenall

R. E. Hoffman and D. Turnbull—The authors have presented evidence which they have interpreted as indicating that the rate of self diffusion is not intrinsically more rapid at grain boundaries than within the grain. Grain-size effects which apparently exist are attributed rather to impurities concentrated at the grain boundaries. In view of our own experiments and the existing evidence, we believe that the support for this hypothesis is not convincing. We have in progress an investigation in which the rate of self diffusion of silver is being measured over an extended temperature range in both single-crystal and polycrystalline specimens. The results of the single-crystal experiments and some preliminary data on fine-grained polycrystalline specimens have already been reported:' and it is anticipated that a complete report will be published in the near future. The self-diffusion coefficient of large-grained polycrystalline silver (1 grain per sq mm) has previously been measured by Johnson" between 730" and 940°C. The diffusion coefficients which we have measured in single crystals (210 plane normal to diffusion direction) agree within experimental error with values calculated from an extrapolation of Johnson's curve down to temperatures as low as 500 °C. However, it has been demonstrated that the overall self-diffusion rate in fine-grained polycrystalline specimens (initial grain size of 0.003 cm) becomes measurably larger than the overall rate in a single crystal at a temperature of 600°C, and the discrepancy between the two rates becomes greater as the temperature is further decreased. In fact, it has been possible to obtain satisfactory penetration curves for polycrystalline specimens using the sectioning technique at temperatures as low as 400°C. At this temperature, the penetration is 50 to 100 times greater in the polycrystalline specimens than in a single crystal. Fisher" has developed an analysis whereby the ratio of the rate of the unit diffusion process at the grain boundary to the corresponding rate within the grain can be calculated from the penetration curves and an assumption as to the width of a grain boundary. This analysis applied to our data indicates that the unit process at the grain boundary is faster by a factor of 10' at 475°C when the grain boundary width is taken to be 5. The silver used in most of these experiments was obtained from the Handy and Harmon Co. and listed as 99.97 pct pure. Preliminary experiments on 99.999 pct silver from the Jarrel-Asch Co. indicate a grain-size effect of the same order of magnitude as in the less pure silver. Nominally, these purities are as good, at least, as that of the carbonyl iron used by the authors, but of course if an impurity effect does exist its magnitude might be very dependent upon the nature of both major and minor constituents. The authors have cited the work of other investigators who have found no grain-size effects. Neither Steigman, Shockley and Nix" nor Maier and Nelson8 were able to correlate self-diffusion coefficients of copper with grain size. However, all their measurements were performed at or above 750°C; and on the basis of our work with silver, no grain-size effect would be expected at temperatures above about 0.7 of the absolute melting temperature unless the grain size were exceedingly small. Likewise, in the investigation of the self diffusion of lead by Seith and Keil,14 the lowest temperature at which the diffusion coefficient was measured in polycrystalline specimens was 207°C, which is still sufficiently high so that the lack of a grain-size effect is not surprising. Finally, in those experiments on iron from which they concluded that there was no grain size effect, Drs. Birchenall and Mehl seem to have no information as to the actual grain sizes immediately prior to and following the diffusion anneal. Without this information, we believe that their own experiments offer little support for their hypothesis. F. S. Buffington, I. D. Bakalar, and M. Cohen—The results given in this paper agree in order of magnitude with those tentatively reported by us.27 However, significant differences exist in the two sets of data, and it may be well to make an explicit comparison. The diffusion studies at M.I.T. were conducted on somewhat higher purity iron (99.98 pct Fe) than the grades used by the authors, but this is undoubtedly not the answer. Fig. 4 shows the diffusion results of both laboratories for the gamma phase, omitting the authors' data on the commercial steels, while fig. 5 presents a similar comparison for the alpha phase. The divergence is much more marked in the latter case than in the former. In connection with the M.I.T. determinations, all of the runs in the gamma range and those above 800 °C in the alpha range were conducted with specimens having a relatively thick (0.002 cm) electrodeposit of radioactive iron. This practice minimizes any possible error due to extraneous diffusion that may occur during the heating to and cooling from the operating temperature. An exact solution of Fick's law for these boundary conditions was used in calculating the diffusion coefficients. At a later time, three runs were made below 800°C, using very thin electrodeposits similar to those of the authors, and the points fell considerably below the values expected from the extrapolation of the results based on the specimens with the thick deposits (compare dash-dot line in fig. 5). However, in the runs with the thin deposits, deviations of 100 pct were found between the individual specimens, whereas the maximum deviation with the thick deposits was less than 25 pct. Accordingly, it is not known at the moment whether the M.I.T. points below 800 °C should be given as much weight as those above 800°C. If this were done, the frequency factor would be of the order of 400 cm2 per sec, which is quite high. In other metals, the frequency factor for self diffusion lies between about 0.1 and 10 cm2 per sec. As the points below

Jan 1, 1951

By C. H. Curtis

HE valuable mineral content of the current feed -*- to the Miami concentrator is as follows: copper, 0.7 pct total; molybdenum, 0.01. Flotation of this ore yields a sulphide concentrate containing: chalco- cite, 44 pct; molybdenite, 0.5; pyrite, 50.0; insol, 5.5. A combination of potassium ethyl xanthate and pentasol amyl xanthate as collectors, and pine oil as frother, are used in this flotation. Rejection of pyrite is encouraged by holding the amount of collectors used to the minimum consistent with copper recovery and by operating at high alkalinity (equivalent to 0.35-0.40 lb CaO per ton solution of pH 11.0). The molybdenum recovery in the sulphide concentrates under the above flotation conditions is approximately 50 pct of that originally present in the ore. Taking into account the acid soluble molybdenum, indicated molybdenite recovery is 75 to 80 pct. The attempt to separate the molybdenite into an acceptable molybdenum product begins with the bulk sulphide flotation concentrate just described. This concentrate is composed of chalcocite, whose floatability has been promoted to the fullest extent possible for the sake of its recovery from the ore, together with the pyrite which has been activated along with the copper mineral. The problem is to deaden the copper and iron minerals, and to float the molybdenite. Obviously in the accomplishment of this end, conditioning and preparation of the pulp, prior to flotation, plays an all important role. The first step is thickening to 50 to 60 pct solids, with milk of lime added to the thickener feed to maintain an alkalinity of the pulp equivalent to a pH of 8.5 to 8.8 during its residence in the thickener. The purpose of the thickening is primarily to reduce the volume of pulp for subsequent treatment. However, the relatively prolonged retention of the pulp in the thickener at the desired alkalinity is known to have a favorable depressing effect upon pyrite. There is a limit for this alkalinity above which a depressing effect upon molybdenite occurs. The thickened pulp (alkalinity: 0.015 lb CaO per ton, pH 8.8), discharges into an agitator, retention time approximately 2 hr, to which additional lime is added to raise the alkalinity to 0.35 to 0.40 lb CaO per ton solution, pH 11.6. This additional lime is required for pyrite depression and can be tolerated without loss of molybdenite because of the limited time of contact in the conditioner tank. The pulp leaving the lime conditioner passes through two successive steaming tanks, which are mechanically agitated, and into which live steam is admitted directly into the pulp near the bottom of the tanks. The temperature of the pulp is maintained as near boiling as possible. The steaming time is approximately 4 hr. The pulp leaving the last steamer has an alkalinity of about 0.04 lb Cao per ton solution, pH 8.7. It is believed that oxidation of the copper and iron sulphides occurs during steaming, the resulting sulphates reacting the calcium hydroxide to calcium sulphate and thus reducing the alkalinity. Since the steamer-feed solution is already saturated with calcium sulphate, the calcium sulphate produced during steaming is precipitated. It is believed that this calcium sulphate is precipitated preferentially on copper and iron mineral surfaces thus decreasing their floatability. Aside from the "lime chemistry" during steaming, pine oil is displaced from the pulp and xanthate decomposed, which has a major effect upon the deadening of the copper and iron sulphides. Following steaming, the hot pulp is admitted to another conditioning tank wherein it is aerated, primarily for cooling, but incidentally for additional oxidation of the copper and iron sulphides. The resulting "deadened" pulp is then diluted to 20 pct solids, a specific collector for molybdenite, ordinary stove oil, is added and the separation of the molybdenite by flotation is undertaken at a pH of 8.5 to 8.8 in standard Miami air-flotation ma-chines. B-22 frother is used when necessary. A re-grind of the thickened rougher concentrates is made prior to the first cleaning operation chiefly for rejection of insoluble in subsequent flotation. The cleaner concentrate is then stepped up to 90 pct MoS, in an 8-cell Denver flotation machine No. 18. Sodium silicate is added to the cleaner circuit. Its effect is to flocculate molybdenite and stabilize the froth. In summary, it may be stated: 1. Separation of molybdenite into an acceptable product from sulphide copper concentrates by flotation involves preliminary preparation and conditioning of the pulp, which is of major importance. 2. This preparation and conditioning consists of several successive steps: (A) Thickening to 50 to 60 pct solids at controlled alkalinity to reduce volume of pulp and to contribute to depression of pyrite. (B) Agitation at high-pulp density for limited time with additional lime to provide for depression of pyrite. (C) Steaming at high-pulp density for decomposition of xanthate and xanthate surface films, evolution of pine oil, and oxidation of sulphide minerals other than molybdenite. The latter involves sulphating of lime with probable precipitation of calcium sulphate preferentially on copper and iron minerals. (D) Aeration at high-pulp density for cooling, and for further oxidation of copper and iron sulphide minerals. (E) Dilution of pulp to 20 pct solids and addition of specific collector for molybdenite, common stove oil. It is hardly necessary to point out that this rather drastic procedure for depression of previously activated copper and iron sulphide minerals, without at the same time depressing molybdenite, is possible due to the inherently high floatability and refractory nature of molybdenite. However, molybdenite is susceptible to depression by excessive lime which must therefore be limited to the amount consistent with satisfactory molybdenite recovery. The steaming procedure is being carried on at Miami Copper Co. under license agreement with Janney, Nokes, and Johnson, holders of letters patent on the process.

Jan 1, 1951

By C. H. Curtis

HE valuable mineral content of the current feed -*- to the Miami concentrator is as follows: copper, 0.7 pct total; molybdenum, 0.01. Flotation of this ore yields a sulphide concentrate containing: chalco- cite, 44 pct; molybdenite, 0.5; pyrite, 50.0; insol, 5.5. A combination of potassium ethyl xanthate and pentasol amyl xanthate as collectors, and pine oil as frother, are used in this flotation. Rejection of pyrite is encouraged by holding the amount of collectors used to the minimum consistent with copper recovery and by operating at high alkalinity (equivalent to 0.35-0.40 lb CaO per ton solution of pH 11.0). The molybdenum recovery in the sulphide concentrates under the above flotation conditions is approximately 50 pct of that originally present in the ore. Taking into account the acid soluble molybdenum, indicated molybdenite recovery is 75 to 80 pct. The attempt to separate the molybdenite into an acceptable molybdenum product begins with the bulk sulphide flotation concentrate just described. This concentrate is composed of chalcocite, whose floatability has been promoted to the fullest extent possible for the sake of its recovery from the ore, together with the pyrite which has been activated along with the copper mineral. The problem is to deaden the copper and iron minerals, and to float the molybdenite. Obviously in the accomplishment of this end, conditioning and preparation of the pulp, prior to flotation, plays an all important role. The first step is thickening to 50 to 60 pct solids, with milk of lime added to the thickener feed to maintain an alkalinity of the pulp equivalent to a pH of 8.5 to 8.8 during its residence in the thickener. The purpose of the thickening is primarily to reduce the volume of pulp for subsequent treatment. However, the relatively prolonged retention of the pulp in the thickener at the desired alkalinity is known to have a favorable depressing effect upon pyrite. There is a limit for this alkalinity above which a depressing effect upon molybdenite occurs. The thickened pulp (alkalinity: 0.015 lb CaO per ton, pH 8.8), discharges into an agitator, retention time approximately 2 hr, to which additional lime is added to raise the alkalinity to 0.35 to 0.40 lb CaO per ton solution, pH 11.6. This additional lime is required for pyrite depression and can be tolerated without loss of molybdenite because of the limited time of contact in the conditioner tank. The pulp leaving the lime conditioner passes through two successive steaming tanks, which are mechanically agitated, and into which live steam is admitted directly into the pulp near the bottom of the tanks. The temperature of the pulp is maintained as near boiling as possible. The steaming time is approximately 4 hr. The pulp leaving the last steamer has an alkalinity of about 0.04 lb Cao per ton solution, pH 8.7. It is believed that oxidation of the copper and iron sulphides occurs during steaming, the resulting sulphates reacting the calcium hydroxide to calcium sulphate and thus reducing the alkalinity. Since the steamer-feed solution is already saturated with calcium sulphate, the calcium sulphate produced during steaming is precipitated. It is believed that this calcium sulphate is precipitated preferentially on copper and iron mineral surfaces thus decreasing their floatability. Aside from the "lime chemistry" during steaming, pine oil is displaced from the pulp and xanthate decomposed, which has a major effect upon the deadening of the copper and iron sulphides. Following steaming, the hot pulp is admitted to another conditioning tank wherein it is aerated, primarily for cooling, but incidentally for additional oxidation of the copper and iron sulphides. The resulting "deadened" pulp is then diluted to 20 pct solids, a specific collector for molybdenite, ordinary stove oil, is added and the separation of the molybdenite by flotation is undertaken at a pH of 8.5 to 8.8 in standard Miami air-flotation ma-chines. B-22 frother is used when necessary. A re-grind of the thickened rougher concentrates is made prior to the first cleaning operation chiefly for rejection of insoluble in subsequent flotation. The cleaner concentrate is then stepped up to 90 pct MoS, in an 8-cell Denver flotation machine No. 18. Sodium silicate is added to the cleaner circuit. Its effect is to flocculate molybdenite and stabilize the froth. In summary, it may be stated: 1. Separation of molybdenite into an acceptable product from sulphide copper concentrates by flotation involves preliminary preparation and conditioning of the pulp, which is of major importance. 2. This preparation and conditioning consists of several successive steps: (A) Thickening to 50 to 60 pct solids at controlled alkalinity to reduce volume of pulp and to contribute to depression of pyrite. (B) Agitation at high-pulp density for limited time with additional lime to provide for depression of pyrite. (C) Steaming at high-pulp density for decomposition of xanthate and xanthate surface films, evolution of pine oil, and oxidation of sulphide minerals other than molybdenite. The latter involves sulphating of lime with probable precipitation of calcium sulphate preferentially on copper and iron minerals. (D) Aeration at high-pulp density for cooling, and for further oxidation of copper and iron sulphide minerals. (E) Dilution of pulp to 20 pct solids and addition of specific collector for molybdenite, common stove oil. It is hardly necessary to point out that this rather drastic procedure for depression of previously activated copper and iron sulphide minerals, without at the same time depressing molybdenite, is possible due to the inherently high floatability and refractory nature of molybdenite. However, molybdenite is susceptible to depression by excessive lime which must therefore be limited to the amount consistent with satisfactory molybdenite recovery. The steaming procedure is being carried on at Miami Copper Co. under license agreement with Janney, Nokes, and Johnson, holders of letters patent on the process.

Jan 1, 1951

By Cloyd M. Smith

B. F. TiLLson*— The manifold difficulties of accurate anemometry in irregular sections of mine passageways, the irregular distributions of velocities in cross sections of the same, and the disturbing influence of the observer upon the air flow, all indicate an undesirable inaccuracy of results obtainable by any standardized method of traversing the section by an anemometer. It seems obvious that another, and simpler, method should be used to determine the volume of air flow in mine passages, namely: 1. At appropriate locations cement or calked framework rings should be installed permanently to equalize the irregularities of sectional contour and provide a place and means of attachment for a temporary cloth brattice which bears a rigid orifice. 2. The measurement of the velocity of air flow through the orifice may then be by anemometer or, preferably, by Pitot tube measurements of the differential pressures on both sides of the orifice in accordance with the standard practices available in engineering 1iterature. † The constants may be determined for various measuring positions in relation to the resulting "vena contracta." 3. The position of the person who makes the measurements is behind the brattice out of the air stream. The Pitot tube does not offer as disturbing an obstruction as the anemometer. A recording gauge may be employed to integrate fluctuations in air flow through that portion of the mine. No traverses are required because the reading may be at a single central point. An anemometer can be used with an orifice flow. The orifice will increase the air velocity at the measuring point, with correspondingly more accurate measurements where the normal air velocity through the passageway is low. Portable brattices might be devised with the cushioning rims which would seal against irregular rock surfaces where permanent rings were not available or feasible. The development by the Ventilation Committee of standard procedures and devices for the orifice measurement of the flow of mine ventilating air might be a desirable project for this coming year. C. M. Smith (author's reply)—Thank you for your discussion of my paper on underground anemometry. Your suggested method of measuring underground air flow is a novel one which might be applicable in some situations. It should be tested along with other suggested methods in any investigation of this subject. G. E. McElroy*—In spite of the adverse publicity that vane-anemometer methods of air measurement have had in the past and that contributed by the present paper, I endorse Mr. Erickovic's statement that anemometer traversing "has proved to be widely applicable, expeditious and simple" and add that available methods are accurate enough for the purposes for which they may be used. The fact that the great majority of minor mine officials assess relative changes in rates of air flow by comparison of crude vane-anemometer measurements, known to average 20 to 30 pct high, has no important bearing on this subject, because state inspection standards were based originally on such methods of air measurement. Federal inspection standards are based on actual rates of flow as determined by traversing, and interest in traversing methods is rapidly increasing. In considering traversing methods, three aspects are of major importance: (1) the absolute accuracy of calibrations; (2) the degree of interference with normal flow conditions introduced by traversing methods designed for accurate measurement by shaft-mounted instruments; and (3) the proper "method" factor to use for approximate measurements by hand-held instruments. With respect to absolute accuracy of calibrations, we have always placed reliance on calibrations made by the National Bureau of Standards, with which manufacturers' calibrations have usually agreed very closely. It is therefore particularly disturbing to find7 that calibrations made previous to June 1947 are presumably about 5 pct in error because of excessive registry caused by the thin flat plates on which anemometers were mounted for calibration. Velocities corrected for calibration have therefore averaged about 5 pct low in all probability. In this connection, it is interesting to note that an anemometer calibrated against Pitot-tube measurement by a single-point method in the Bureau of Mines experimental coal nine in 1923 indicated this same difference of about 5 pct and that the same instrument calibrated by a traversing method in a metal mine some months later indicated a difference in the same direction of about 4 pct. These results are reported by the Bureau of Mines.8 Regarding the degree of interference, or changes in velocity distribution, caused by the position of the observer's body in traversing operations, misconceptions seem to be especially prevalent, resulting in increasing advocacy of methods, such as the "clear section" method outlined in this paper, that cause just the type of interferences that they are designed to avoid. The degree of interference for any method may be gauged easily by a few experiments with a velocity-pressure gauge connected to a Pitot tube or with an indicating velocity meter such as the Velometer. In an experiment cited by McElroy and Richardson,# a decrease of 5 pct was noted at ten widths upstream from a 6-in. plank, whereas an observer's body at about the maximum practical distance of 6 ft downstream from the instrument is only about four widths away. In the Bureau of Standards paper previously mentioned, it is recommended that supports used in calibrations be at least 16 widths downstream. In practice, therefore, a downstream position of the observer is ruled out as far as accurate measurement is concerned. Operation of the anemometer by rigid shaft support from a point outside the section is seldom practicable; however, accurate results can be obtained, with the anemometer rigidly attached to a short shaft and held at arm's length, by an observer advancing across the traversed section while he faces the opposite wall and stands sideways to the current, provided that he keeps the instrument at least 3 ft away from his body at all times and traverses the entire section with it. If the traverse can be started with the observer in a side recess, the entire section can be covered in one operation. Normally, it would be covered in two half-sections. The presence of the observer's body does not, as is commonly supposed, increase the average velocity throughout the remaining part of the section. Rather, the velocities 1 to 2 ft on either side of his body are increased, but the distribution of velocities throughout the rest of the cross section remains normal, and a traverse made as stated gives a true average velocity for normal-flow conditions. Regarding the proper "method" factor for accounting for interference in the approximate methods of traversing with hand-held instruments, here again confusion prevails, for which the writer must assume some of the blame. Comparison of consecutive traverses made by shaft-held and hand-held 4-in. anemometers in field work after the tests reported by McElroy and Richardson' gave method factors

Jan 1, 1950

By W. O. Philbrook, K. M. Goldman, G. Derge

T. Rosenqvist—The most interesting point in this paper is the observed transfer of iron into the slag in the initial stage of the desulphurization process, after which the iron again is reduced to the metallic state. The authors interpret this observation as showing that the sulphur enters the slag as an iron-sulphur compound which subsequently is decomposed by the slag. The present writer has previously suggested the following equation for the desulphurization process: S + O2- ? S2- + O For equilibrium in the blast furnace the oxygen potential is defined by equilibrium with graphite and CO of 1 atm pressure: C + O ? CO [2] During the desulphurization process the reactions proceed in the direction of the arrows. If one assumes eq 2 to be significantly slower than eq 1, the transfer of sulphur into the slag, in accordance with eq 1, will build up a local oxygen potential at the metal-slag interface very much higher than that corresponding to the value defined by eq 2. This is possible because the equilibrium oxygen potential in eq 1 is high as long as the sulphur content in the slag is low. This oxygen potential will again be able to oxidize some iron: Fe + O ? Fe2+ + O2- and an increase in the iron content of the slag will be observed. Adding up eqs 1 and 3 one obtains: S + Fe ? S2- + Fe2+ The net effect is thus in harmony with the experimental observation but is obtained without assuming any close ties between the sulphur and iron atoms during the process. Furthermore, it follows from eqs 1 and 2 that when the sulphur content in the slag increases, and equilibrium with C and CO is finally approached, the local oxygen potential at the metal-slag interface will decrease, and the iron in the slag will be reduced back into its metallic state. C. E. Sims-—The data and conclusions presented in this paper are thoroughly convincing in establishing the mechanism of sulphur transfer from iron to slag as in a blast furnace. The evolution of gaseous CO in step 3 of the reactions given on p. 1112 makes the process virtually irreversible. Assuming that the process is similar in slag-metal systems other than in the blast furnace, it is readily seen why free CaO and re-ducing conditions so greatly favor desulphurization. On the other hand, the very effective desulphurization obtained in oxidizing slags when strongly basic, must be attributed to the relatively high stability of CaS as compared to FeS. The ease and simplicity with which the reactions of classic chemistry agree with the experimental data and explain the mechanism is noteworthy. The concept of molecules of FeS, soluble in both phases (metallic iron is not soluble in the slag), migrating from the iron to the slag and there reacting with CaO, which is soluble only in the slag phase, is clear and uncomplicated. This is likewise true for step 3. Those who would deny the existence of molecules or molecular-type combinations in liquid iron, must strain to provide a mechanism so lucid. In the absence of molecules, the Fe and S exhibit a remarkable collusion. L. S. Darken—The investigation and interpretation of rate phenomena in the range of steelmaking temperatures is a difficult task. Most of the laboratory investigations of steelmaking reactions have been concerned with equilibrium. Having determined the equilibrium, our attention naturally focuses next on the mechanism and rate of approach to equilibrium. The authors seem to have contributed substantially to our understanding of these factors for the case of sulphur transfer. I should like to ask the authors whether they consider that the sulphur transfer reaction is diffusion controlled as many high-temperature reactions seem to be. If so, it would seem reasonable to suppose that the slow diffusion step of the process is the transfer across a pseudo-static layer or film similar to that considered in heat flow problems. As the diffusivity and fluidity are smaller for the slag than for the metal, it may tentatively be assumed that the sulphur gradient exists in a thin layer in the slag adjacent to the slag-metal interface and that the metal and the main mass of slag are each maintained uniform by convection. On this basis the amount of sulphur transferred across unit area per unit time is D p (?S%)/100 ?1, where D is the diffusivity, p the density, (?S%) the difference in percent sulphur on the two sides of the layer, and ?l is the layer thickness. At the beginning of the experiment the main body of the slag and hence one side of the layer contains no sulphur; therefore (?S%) may be replaced by (S%), the sulphur content of the slag at the slag-metal interface, which in turn is equal to L[S%] where [S%] is the sulphur content of the metal and L is the distribution coefficient. The rate of transfer thus becomes DpL[S%]/100 ?l, which the authors designate K[S%]. Equating these two quantities and setting D = 10-6 cm2 per sec, p = 3 g per cm3, L = 40, and K = lo-+ g cm-2 sec-1, it is found that ?l, the film thickness, is about 0.01 cm—a value of the order of magnitude of that found in heat transfer problems in liquids. The uncertainty of the numerical values used leaves much to be desired, but at least it can be said that this calculation tends to support the proposed model involving diffusion through a film. Although this does not seem to affect the general argument, I should like to call attention to the fact that the diffusivity3 of sulphur in hot metal is found (on conversion of units) to be about 10-4 cm2 per sec rather than 104 cm2 per sec as stated by the authors. The three equations written by the authors to express the steps in the overall process of sulphur transfer may alternatively be written ionically as only two Fe + S = Fe++ + S-- Fe++ + O-- + C (graphite or metal) = CO (gas) + Fe where the underscore is used to designate the metallic phase; ionic species are slag constituents. After the authors have so neatly demonstrated that iron and sulphur transfer together (at least initially), this fact seems almost self evident; from eq 4 it is seen that if sulphur acquires a negative charge during transfer

Jan 1, 1951

By Leonard B. Gulbransen, John R. Lewis, W. Martin Fassell, J. Hugh Hamilton

T. E. Leontis (The Dow Chemical Co., Midland, Mich.)—This paper is of particular interest to me because of my own work with F. N. Rhines on the oxidation of magnesium and magnesium alloys a few years ago. The authors are to be complimented on their development of an accurate and reproducible technique for measuring ignition temperatures and on their comprehensive study of the many variables that affect the ignition temperature of magnesium. It is indeed gratifying to see that they have obtained a good correlation between ignition temperatures and the oxidation rates reported by us. The correlation is valid not only with composition within one alloy system but also between alloy systems; that is, alloying elements which effect the greatest increase in oxidation rate also produce the greatest decrease in ignition temperature. There are a few points upon which I would like to comment. In attempting to correlate ignition-temperature data, one must be sure that the same definition of this quantity is used by all investigators. It does not appear to me that such is the case in the authors' comparison of their data with the theoretically calculated values of Eyring and Zwolinski. The equation derived by these investigators defines the ignition temperature, To, as the temperature at the gadoxide interface, whereas the present authors use the metal temperature as the criterion for ignition. The contradiction in the effect of oxide-scale thickness on ignition temperature between the predictions of the Eyring-Zwolinski equation and the observations reported in this paper indicate that some variable has not been taken into consideration. Could that be the geometry and size of the specimen? There is a marked difference in the type of specimen used in this investigation and that used in our work which formed the basis of Eyring and Zwol-inski's theoretical treatment. Another factor which plays an important role in ignition is the vapor pressure or the rate of vaporization. Combustion can safely be assumed to take place in the vapor phase by the reaction between vaporized magnesium and oxygen. Thus, a more accurate theoretical analysis may be made on the basis of the rate of vaporization which may be the controlling rate of the process. The effect of a large number of alloying elements on the ignition temperature has been reported in this paper, but beryllium was not included. Practical experience dictates that beryllium markedly decreases the burning tendency of magnesium. I was wondering if the authors plan to study the effect of beryllium in their future work. The authors predict that concentrations of sulphur dioxide in the furnace atmosphere greater than 5.8 pct would be expected to increase the ignition temperature to values still higher than those they measured. I would like to mention that large concentrations of sulphur dioxide markedly increase the rate of combustion of magnesium once ignition has started. Although it has been shown in the paper that the ignition temperature of magnesium in oxygen increases with increasing sulphur dioxide content up to about 1 to 2 pct, in practice relatively low-melting commercial cast alloys (AZ63A and AZ92A) are being continuously heat treated at temperatures just below the melting point in air containing 0.5 to 0.75 pct SO*. In regard to the change in color of the oxide scale observed on magnesium and magnesium alloys just prior to ignition, I would like to mention that in our work alloying elements were found to color the usually white magnesium oxide even though ignition did not occur. For example, the oxide formed on Mg-A1 alloys was gray, increasing in intensity with aluminum content in the alloy. Finally, I might suggest that the authors indicate their source of the value of 0.8 g per cc for the density of MgO as it is formed on magnesium upon oxidation at elevated temperatures. W. M. Fassell, Jr. (authors' reply)—The comments by Dr. Leontis are very excellent ones and I will attempt to answer them in order. First, the problem of ignition of magnesium is a rather difficult one since many factors are involved. Concerning the comparison of the To in the Eyring-Zwolinski equation, eq 4, with the experimentally determined values, it will be noted that the calculated and experimental values of the ignition temperature in Table I are not self-consistent. In the case of the 1.78 pct A1-Mg alloy the calculated value is 49°C below the experimental value; for the 3.81 pct A1-Mg alloy, 122°C below the experimental value; for Mg with 5x10-' cm film, 19°C above the experimental value; for Mg with 2x10-I cm film, 28 °C below the experimental value. Thus, if it were merely a matter of difference of location of temperature measurement the calculated ignition temperature would always be below the experimental value, the difference being due to the thermal gradient through the oxide film. The possibility of a thermal gradient in the magnesium metal must be considered. From Carslaw and Jaeger,'Y t can be shown that the maximum temperature gradient that could exist between the oxide-metal interface and the center of the sample is of the order of O.Ol°C. The geometry and size of the specimen could certainly have some effect on the ignition temperature. The equation for ignition that has been proposed in reference 14 is of the following type containing terms to account for this and other factors: M dT AHv(T) =Cp--------------\-J(.T—TB) + ZAHl-M A dx where AH is the heat of reaction, v(T) is the velocity of the reaction at temperature, Cp is the heat capacity of sample, M is the mass of sample, A is the area of sample, t is time, J is the total coefficient of heat transfer outward from the reaction zone, TR is the temperature of the bath or furnace, and AH,, is the heat associated with any phase change involved. Prior to the instant of ignition, the vapor pressure of magnesium is of no special significance. After ignition, neither eq 4 nor the above equation is applicable. The actual combination of magnesium cannot safely be assumed to take place in the vapor phase. While experimental data is lacking to support a hypothesis that ignition does or does not occur in the vapor phase, some observation on the pressure ignition experiments may be of interest. At high oxygen pressures, once ignition has occurred, the reaction of magnesium with oxygen approaches near explosive violence, the entire sample being consumed in probably less than 1 sec. At atmospheric pressure it usually requires 15 to 20 sec. Thus it appears that the oxygen concentration becomes the rate determining factor. Further, if burning magnesium is observed through darkened glass (Lincoln Super-visibility Shade No. 12) the magnesium sample is very much hotter than the "smoke" and the outline of the sample is retained perfectly. No "flame" is visible above the metal. No work was done on Mg-Be alloys. We do, however, intend to study this problem in the near future.

Jan 1, 1952

By Douglas J. Robinson

A pilot scale process has been developed to separate mercury, tin, silver, and copper from dental amalgam scrap. Laboratory research led to a process which was operated in 55 gallon drum sized reactors to treat 2000 Tr oz charges of scrap per day. This paper discusses the development work, presents a flowsheet and gives an indication of the equipment necessary to carry out the process on a commercial scale.

Jan 1, 1984

By Thomas M. Barczak

This paper describes a computerized simulation system developed by KETRON, Inc., in cooperation with the US Bureau of Mines. The system is designed for use by mine managers and engineers rather than computer specialists and allows the novice user to arbitrarily create room-and-pillar mine plans, with the computer illustrating the simulation of face activities on a video screen. Extensive visibility of the simulation is provided by visual display of the machine elements and their movement in the section to accommodate the extraction, roof control, place changing, and haulage operations. Performance statistics are also generated to analyze the operations. Both conventional and continuous mining methods can be simulated. In addition to simulating arbitrary scenarios, the system can also be used to display actual time-studied operations and maintain a file of these activities. Practical applications and case studies are provided in the paper to demonstrate the usefulness of the system. The system is designed to operate on a desk-top mini computer and is currently installed on a Hewlett-Packard 9854B system with 450 K bytes of memory.

Jan 1, 1983

By P. J. Clifford, A. B. Brown

The development of computerised mine planning has been hindered by the cost of processing the large data bases associated with mineral deposits and the inherent inconvenience of performing an interactive process on a time-shared computer. A dedicated computer is ideally suited for the purpose but reasonably priced micro/mini-computers have not had the capacity to store the volume of data or the processing capability required. Hardware development has cleared the way for the development of truly interactive mine planning software, covering the complete range of activities associated with mine planning, from geological assessment to short-term scheduling of operations, incorporating them into a single system using a common database. The integrated system, comprising software and dedicated hardware, adds previously unattainable speed, flexibility and clarity to the planning process because it has been tailored exclusively for that purpose.

Jan 1, 1983

By H. E. Wheeler

THE work described was carried out during 1917 and 1918 at the testing laboratory of Watertown Arsenal at the instigation of the Nitrate Division and later with the concurrence of the Cannon Section of, the Ordnance Department, U. S. A. The experiments follow three principal lines of work: First, the effect of nitrogen under pressure on steel containers of various compositions at a red heat; second, the effect of decomposing ammonia on various alloy steels, iron, and non-ferrous alloys; third, a new theory of the erosion of guns in respect to the effect of nitrogen in steel. PART I In the Haber process for the manufacture of ammonia from its elements, it is necessary to have nitrogen and hydrogen of 1500 lb. per sq. in. (105.5 kg. per sq. cm.) at a temperature of 500 to 600° C. The steel containers for these gases gave trouble by failing without apparent cause. When the General Chemical Co. began to develop its method for the production of ammonia, it experienced the same difficulty and, knowing that the Haber process had solved the difficulty by the use of alloy steels, it made several small steel bottles of different compositions and kept them filled with these gases at this pressure and temperature until they failed. The time of service varied from a few days, for the plain steel casting, to two years for a chrome-vanadium forging. Four of these steel bottles were sent to this laboratory for investigation; they were a plain carbon-steel forging, a nickel-steel forging, a chrome-vanadium steel forging, and a chrome-steel forging. The time of service was as follows: plain carbon steel, 4 mo.; nickel steel, 6 mo.; chrome-vanadium steel, over 2 yr.; chrome steel, 4 mo. When these containers were cut open and the cross-section surface polished, they showed an inside zone with a different luster from the rest of the metal. Upon etching, this zone was almost unaffected while the rest of the steel etched normally.

Jan 4, 1920