Search Documents

By Thomas S. West

A gravel pack completion is described which, in addition to sand screening, permits the exclusion of water and gas from the oil production. A gravel pack composed of alternate layers of different particle sizes is employed which has horizontal but not overall vertical permeability to drilling mud. A packer is run on the tubing and set in the screen member, around which the gravel layers are placed. The hydrostatic pressure of a mud column prevents the entry of fluid above (or below) the level of the packer. The producing level may be changed by raising or lowering the tubing. Vertical coning is controlled by replacing the mud column with oil and continually injecting oil into the sand section above (or below) the packer. Three well histories with this completion are described in detail. Statistics on the first 100 completions of this general type are given. INTRODUCTION Gravel pack type completions have heretofore been almost exclusively for the purpose of preventing sand production. This paper describes and presents well data on a gravel pack procedure which is primarily a selective completion method. Although complete sand screening is achieved, the exclusion of gas and water is the primary objective. The application of this procedure is therefore not limited to unconsolidated sands. Application to hard sand or limestone reservoirs is equally feasible. GRAVEL PACK PROCEDURE Figs. 1 and 2 illustrate the gravel pack procedure employed. The gravel is placed by means of the frangible disc type dump bailer shown by Fig. 1.' This bailer has its lower end closed by a plate glass or steel rimmed cement disc. This disc breaks when the bailer passes over the liner allowing the gravel to be deposited around the liner. The liner employed is usually 2 3/8 in. to 2 7/8 in. OD and 1.995 in. or 2.07 in. ID (ID of 2-in. tubing or 2-in. line pipe). Two tvDes,. of slotted sections have been employed. One type employs horizontal hacksaw cut or milled slots of about 1/32 in. width opposite the producing section. The other type employs the same type,. of horizontal slot. but slots are placed in groups of six at intervals of about 8 in. A cage composed of half round vertical bars, spaced apart, is then welded to the pipe over the slotted sections. This liner has the bottom end closed, usually by bull plugging. The upper end is closed, while placing the gravel, by a cap which is held in place by means of a shear pin. This cap has a mandrel attached to it which extends downward inside of the liner to within a few inches of its lower end. The top of the liner cap is formed to fit a releasing type wire line core barrel overshot. The liner is lowered to bottom on the wire line by means of this overshot and released. The length of the liner is usually suffi- cient to extend to about 20 ft above the highest level to which gravel is to be placed, the upper end of the liner always being in the casing. The dimensions of the liner and bailer are such with respect to the casing size that the bailer will always pass over the liner. After releasing the liner the bailer is run to bottom. Passage of the empty

Jan 1, 1951

By Thomas S. West

A gravel pack completion is described which, in addition to sand screening, permits the exclusion of water and gas from the oil production. A gravel pack composed of alternate layers of different particle sizes is employed which has horizontal but not overall vertical permeability to drilling mud. A packer is run on the tubing and set in the screen member, around which the gravel layers are placed. The hydrostatic pressure of a mud column prevents the entry of fluid above (or below) the level of the packer. The producing level may be changed by raising or lowering the tubing. Vertical coning is controlled by replacing the mud column with oil and continually injecting oil into the sand section above (or below) the packer. Three well histories with this completion are described in detail. Statistics on the first 100 completions of this general type are given. INTRODUCTION Gravel pack type completions have heretofore been almost exclusively for the purpose of preventing sand production. This paper describes and presents well data on a gravel pack procedure which is primarily a selective completion method. Although complete sand screening is achieved, the exclusion of gas and water is the primary objective. The application of this procedure is therefore not limited to unconsolidated sands. Application to hard sand or limestone reservoirs is equally feasible. GRAVEL PACK PROCEDURE Figs. 1 and 2 illustrate the gravel pack procedure employed. The gravel is placed by means of the frangible disc type dump bailer shown by Fig. 1.' This bailer has its lower end closed by a plate glass or steel rimmed cement disc. This disc breaks when the bailer passes over the liner allowing the gravel to be deposited around the liner. The liner employed is usually 2 3/8 in. to 2 7/8 in. OD and 1.995 in. or 2.07 in. ID (ID of 2-in. tubing or 2-in. line pipe). Two tvDes,. of slotted sections have been employed. One type employs horizontal hacksaw cut or milled slots of about 1/32 in. width opposite the producing section. The other type employs the same type,. of horizontal slot. but slots are placed in groups of six at intervals of about 8 in. A cage composed of half round vertical bars, spaced apart, is then welded to the pipe over the slotted sections. This liner has the bottom end closed, usually by bull plugging. The upper end is closed, while placing the gravel, by a cap which is held in place by means of a shear pin. This cap has a mandrel attached to it which extends downward inside of the liner to within a few inches of its lower end. The top of the liner cap is formed to fit a releasing type wire line core barrel overshot. The liner is lowered to bottom on the wire line by means of this overshot and released. The length of the liner is usually suffi- cient to extend to about 20 ft above the highest level to which gravel is to be placed, the upper end of the liner always being in the casing. The dimensions of the liner and bailer are such with respect to the casing size that the bailer will always pass over the liner. After releasing the liner the bailer is run to bottom. Passage of the empty

Jan 1, 1951

By Harvey T. Kennedy, Charles R. Olson

In many investigations of the performance of petroleum reservoirs the assumption is made that the liquid, if below its bubble-point pressure, is at all times in equilibrium with gas. On the other hand, observations by numerous investigators have indicated that gas-liquid systems including hydrocarbon systems, may exhibit supersaturation to the extent of many hundred psi in the laboratory. Up to the present, there has been no reliable data on which ti judge the actual extent of supersaturation under conditions approaching those existing in petroleum reservoirs. The work reported here deals with observations and measurements on mixtures of methane and kerosene in the presence of silica and calcite crystals. Bubbles were observed to form on crystal-hydrocarbon surfaces in preference to the glass-hydrocarbon interface or to the body of the liquid. Statistically, it was found that the number of bubbles formed per second per square centimeter of crystal surface was a function of the supersaturation only. and the function was evaluated graphically. Supersaturations were observed up to 770 psi, under which condition bubbles formed quickly and with considerable violence. With decreasing degrees of supersaturation, the frequency of bubble formation became less, until at 30 psi supersaturation and lower, no bubbles were observed to form, even though the observation at 30 psi was continued for 138 hours. It was found that silica and calcite crystals had identical effects, within experimental error, in accelerating the formation of bubbles, and that small amounts of water and crude oil had no effect on the results. It is shown that the maximum supersaturation that can exist in a reservoir may be calculated from the data presented and from the area of the rock surface. It is also shown that the number of bubbles formed in the reservoir, in order of magni- tude, may be calculated for any rate of pressure decline imposed on the reservoir by production. The bearing of the number and distribution of bubbles on reservoir performance is discussed. INTRODUCTION A liquid system is supersaturated with gas when the amount of gas dissolved exceeds that corresponding to equilibrium at the existing pressure and temperature. The degree of super-saturation may be conveniently expressed as the difference between the bubble-point of the mixture and the prevailing pressure. Thus, if a mixture having a bubble-point of 1,000 psi at a given temperature exists in single liquid phase at 700 psi at the same temperature. it is supersaturated to the extent of 300 psi. There are many examples of high supersaturations, mostly in aqueous solutions, reported in the literature. Thus, Kenrick, Wismer and Wyatt1 showed that water may be saturated with oxygen, nitrogen or carbon dioxide at 100 atmospheres, and the pressure reduced to one atmosphere without producing bubbles immediately. When liquids are in a state of tension, they may be considered as supersaturated at least to the extent of the tension. The tensile strength of water has been reported as 30 atmospheres by Meyer,1 60 atmospheres bv Budgett. 30 to 50 atmospheres by Temperley and Chambers,4,5 200 atmospheres by Dixon6. and 223 atmospheres by Briggs.7 Vincent8,9 determined the tensile strength of a mineral oil as 45 psi. Gardescu1 maintained pressures for short times in a model reservoir at 115 psi below the bubble-point. It should be noted that the high supersaturations observed were obtained on systems carefully purified to remove particles or surfaces which might promote the formation of bubble; These "nuclei" were considered as contaminants which interfered with the determination of a property of the liquid. In

Jan 1, 1952

By D. R. Parrish, W. E. Lamoreaux, W. W. Owens

Several methods are now being used by the industry for determining the gas-oil flow characteristics of reservoir rock samples. Most of the laboratory experirnerltal rlzethods can be classified either as steady-state, unsteady-state, or Hassler-type tests. Although the validity of the results obtained by any specific method is not questiorzed, in general, the complexity of these methods leaves much to be desired. This investigation dealt primarily with the development of a simplified unsteady-state (gas drive) laboratory test method for measuring kg/ko re1ationships of small core samples. A procedure for selecting uniform and representative sandstone rock samples was developetl since it was observed that the reliability of the test resu1ts was dependent upon having a sample which was uniform in porosity and permeability. The importance of factors known to influence the results of this type test, such as oil viscosity, gas expansion, and pressure differential were eva/uated. The method of calculating the test results was found to give valid data over a wide range of gas expansion. The limits of pressure differential and oil viscosity which would insure reliable results were defined. In cornparisorz with other presently used testing methods, the gar drive method requires considerably less apparatus, is less time consuming, and is more readily adaptable to a routine type core test. INTRODUCTION The detcrmination of the gas-oil flow characteristics of reservoir rocks has been given concentrated study by the oil industry for the past decade. Both analytical and experimental methods have been considered. The analytical methods have generally relied on a simplified characterization of porous materials in order that the fundamental laws of fluid flow and capillarity could be utilized. More direct experimental methods have been proposed by numerous production research laboratories with the primary goal being a simple and inexpensive, yet reliable method for measuring the gas-oil flow characteristics of reservoir rock samples. The experimental methods which have been reported in the literature generally can be grouped into one of the following classifications: 1. Steady state1,2,3,4 in which a fixed ratio of flowing fluids is forced through the test sample until equilibrium conditions of saturation and pressure gradient are established. 2. Hassler or stationary liquid method5,6,7 in which a uniform liquid saturation is attained, usually through capillary desaturation, then gas or oil is flowed through the sample at a pressure drop insufficient to disturb the pre-established saturation. :. Non-steady state method1,8,9,10 In which only the gas phase is injected into the sample, causing displacement of the oil phase and resulting in a system of continually changing average saturation and saturation gradient. The steady-state and Hassler methods have been most popular because, no doubt, of the simplicity of the actual relative permeability calculations. The problems involved in these methods were limited to apparatus design and measurement techniques. In the steady-state method, for instance, a constant flowing fluid ratio is established which, at equilibrium, results in an essentially uniform saturation over the entire length of the test sample. Therefore, the basic flow characteristics could be obtained by correlating the average saturation of the test sample with the kg/ko determined at various flowing gas-oil ratios.

Jan 1, 1957

By H. Warlimont

Precipitation of cementite plates from carbon-rich austenite gives rise to localized formation of martensite in the matrix next to the precipitate upon quenching. The growth of this martensite is urmsually restricted. A distinct lattice relationship is found between this martensite and the cementite phase. It is concluded that the martensite is nacleated in the region of lowest carbon concentration at the austenite-cementite integace, with low-energy matching to the cementite lattice. The restriction of the growth of the martensite phase to the vicinity of the cementite plates can be explained by misfit in the austenite -martensite glissile interface and the uphill carbon gradient in the austenite. The investigation was carried out by light- and electron-microscopy and by electron-diffraction of thin foils. THE precipitation of cementite from carbon-rich austenite gives rise to a localized formation of martensite in the matrix next to the precipitate upon quenching. This behavior has been described previously by A. B. Greninger and A. R. Troiano1 for a 1.78 pct C steel ("pseudomorphic martensite"), and by A. Hultgren2 for a Fe-0.99 pct C—5.20 pct Mn steel. By measuring the axial ratio of the martensite phase,' it was found that the carbon content of this martensite was lower than that of the martensite which formed without prior partial transformation to cementite. Therefore, it was concluded that lowering of the carbon concentration in the matrix next to the precipitate raised the M, temperature locally and thus led to nucleation of martensite at the ce-mentite-austenite interface. However, the restriction of the growth of the martensite phase to the vicinity of the cementite plates cannot be explained by localized carbon impoverishment alone. Under ordinary conditions a martensite plate which has been nucleated in a favorable site can continue to grow until it encounters a physical barrier, or until it encounters a compositional barrier in the matrix, where the total free-energy change for the reaction is not negative, since it has been shown by numerous experiments that the activation energy for growth of ordinary martensite is virtually zero.' An investigation was conducted by transmission electron microscopy and electron diffraction to study this extraordinary martensite. It was expected that crystallographic in addition to compo- sitional conditions caused this abnormal mode of martensite formation. MATERIAL AND PROCEDURE An Fe-C-Mn alloy containing 1.15 pct C and 5.1 pct Mn was used for most of the experiments. For metallographic work, specimens 15 by 15 by 3 mm were used. For preparation of thin films, the initial specimen size was 15 by 15 by 0.7 mm. The heat treatment of each specimen consisted of austenitizing for 1/2hr at 1100'~ in an argon atmosphere, subsequent isothermal holding in a lead bath at one of a series of different temperatures, and, finally, quenching into brine at room temperature. For metallographic examination, the specimens were mounted in bakelite (slight tempering effects resulted from the heat in the mounting press) ground, polished, and etched in nital. Thin films for transmission electron microscopy and electron diffraction were obtained by grinding the specimens to approximately 0.2 mm thickness and subsequent electrolytical thinning. The orientation relationships were evaluated from sets of electron micrographs and selected-area electron diffraction patterns of single units of the transformation products and the adjacent matrix. RESULTS In the alloy which was used for the experiments, proeutectoid carbide precipitates over a wide tem-

Jan 1, 1962

By N. J. Grant, C. F. Flo, F. B. Cuff

An investigation of the Ti-Cr system has shown the presence of a complete series of solid solutions in the ß phase, with a minimum in the solid us near 50 pct Cr. An intermetallic compound, TiCr2, forms during cooling from the ß solid solution. There is a eutectoid reaction at the low chromium side of the system. IN view of the recognition of the potentialities of titanium and its alloys as important structural materials there has arisen a need for a systematic investigation of various titanium binary diagrams. Of these, the Ti-Cr system was of particular interest because of the improved properties imparted to titanium by small chromium additions. A literature survey disclosed a partial constitution diagram that had been suggested by Vogel and Wenderott,¹ This diagram, shown in Fig. 1, is the result of work done on the Fe-Cr-Ti ternary system. Alloys containing less than 43 pct Cr were not investigated. In addition to this, a study of alloys in the Ti-Cr system was reported by McPherson and Fontana.² Adenstedt3 furnished a number of dilato-meter curves obtained from specimens containing 5, 10, and 15 pct Cr. Very recently, McQuillan4 submitted a proposed diagram covering the complete range of compositions between titanium and chromium. This diagram was in general agreement with the one described here. Experimental Procedure Sponge titanium (99.7 pct) and electrolytic chromium (Bureau of Mines analysis: 99.03 pct Cr, 0.40 pct Fe, 0.53 pct O2) were used for the initial determination of the system. Subsequent refinement of points, particularly for the low chromium side of the diagram, was accomplished by using iodide titanium and high purity chromium (National Research Corp. analysis: 0.050 pct C, 0.045 pct O2). The Ti-Cr alloys were prepared by melting 25 g charges in an enclosed, water-cooled copper crucible using a movable tungsten electrode. The atmosphere was helium, prepurified by first passing it through a drying agent and then through sponge titanium heated to 850°C. Particular care was taken to avoid any large scale segregation resulting from insufficient mixing of the components. Each charge was melted and kept molten for approximately 1 min, allowed to cool, inverted, remelted, and kept molten for an additional minute. In order to ascertain the amount of impurity pickup (oxygen and nitrogen) during the melting procedure, three iodide titanium specimens were melted in sequence and then checked for increased hardness. The specimens in the as-cast condition showed a relatively large hardness increase over the hardness of the as-received iodide titanium. The scattering on these readings was large. However, after the specimens had been homogenized at 840 °C for 2 hr and slowly cooled, the measured increase in hardness over the as-received titanium was small with a corresponding decrease in scatter. The reason for this hardness change is that in heat treating these specimens below the transformation temperature (885°C) there was a conversion from the as-cast structure to the more nearly equilibrium structure of the annealed specimen. The alloys were analyzed for chromium content and although the actual percentage was invariably low the deviation from the nominal composition never exceeded 0.6 pct Cr.

Jan 1, 1953

By I. Fatt, D. N. Saraf

A method is described for measuring two- and three-phase relative permeabilities in sandstones or sand packs using a nuclear magnetic resonance (NMR) technique to determine fluid saturations. Two- and three-phase relative permeabilities have been determined on Boise sandstone using the NMR technique of saturation measurement. Three-phase relative permeability to water was found to depend only on the water saturation, whereas three-phase permeability to oil depended on both the water and oil saturations. Relative permeability to gas in three-phase flow was found to depend only on the total liquid saturation. INTRODUCTION Three-phase relative permeabilities are extremely useful in calculating field performance for reservoirs being produced by simultaneous water and gas drives. Three-phase relative permeability data are also needed for analyzing solution gas-drive reservoirs which are partially depleted and are being produced by water drive. Some thermal recovery processes involve three-phase flow which require three-phase relative permeability data for predicting reservoir behavior. Unfortunately, three-phase relative permeability measurements have rarely been made. Also, because of the scarcity of three-phase data, it has not been possible to date to relate other measured rock characteristics to the relative permeabilities with a great certainty. Leverett and Lewis,l Reid2 and Snell3 have reported three-phase relative permeability data on unconsolidated sands. Leverett and Lewis used ring electrodes spaced along the length of the sand sample to measure the resistivity of the sample which was assumed to be monotonically related to brine saturation. Gas saturation was determined from pressure-volume measurements. Oil saturation was obtained by material balance on the cell containing the sand sample. This method is involved and time consuming. Another difficulty arises from the fact that the resistivity of the sand is a function not only of saturation of brine but also of the distribution and saturation history of the brine in the pore spaces. Reid2 used a gamma ray absorption technique for measuring liquid saturation. This method has the disadvantage that total liquid saturation rather than oil or brine saturation is all that can be measured and still another method is required to determine the saturations of individual components. snel13 used a neutron bombardment method which also required a separate determination of the individual component saturations. Caudle et al.4 measured three-phase relative permeability on consolidated sandstones using vacuum distillation for determining fluid saturations. Distillation after each reading makes this technique very lengthy and time consuming. Corey et a1. 5 and Naar and Wygal 6 measured three-phase relative permeability on sandstones by the capillary-pressure method. Semipermeable diaphragm assemblies were used at each end of the core specimen to keep the water phase in the core. Gravimetric methods were used to determine fluid saturations. Sarem7 recently reported an unsteady-state method for measuring three-phase relative permeability on sandstones. This method is an extension of Welge's method8 for measuring two-phase relative permeability. Although Sarem's method is simple and comparatively fast, the assumptions involved may oversimplify the problem. Sarem's assumption, that in all rocks relative permeability to each fluid will depend only on the saturation of that fluid, seems to be rather unrealistic. Neglecting capillary effects at the end of the core is also a weak assumption. Donaldson and dean9 measured three-phase relative permeability using a method similar to Sarem's.

By P. Dayal, L. S. Darken

THE migration of carbon in austenite under the influence of an electric current has been observed qualitatively several times.14 Lebedev and Guterman5 recently have reported quantitative measurements; their method consisted of welding a low-carbon to a high-carbon wire, passing current, and later estimating the carbon distribution metal-lographically. We believe the method reported in this paper is superior in principle since electrolytic migration and ordinary diffusion are not necessarily coupled as in the weld method; also for the first time the extent of migration is determined by combustion analysis for carbon. A uniform tubular specimen (S.A.E. 1045 or 1095) was mounted as illustrated in the figures. the whole assembly being in an atmosphere of purified nitrogen. Direct current (129 to 226 amp at 0.65 to 1.0 v) served as source both of heat and of potential gradient for carbon migration. The duration of experiment was 7.5 to 12 hr, during which time the temperature was controlled (±10°C) manually. The experimental conditions are indicated in the first five columns of table I. With this experimental arrangement, the temperature initially has a maximum value at the central plane. Also, in this vicinity (by virtue of the symmetry), the carbon entering any volume element is equal in quantity to that leaving; the loss or accumulation of carbon occurs near the electrodes since the carbon moves uniformly near the center and is "frozen" at the cold ends. Hence, providing the time of experiment is not sufficiently long to seriously influence the conditions at the central plane, carbon crosses this plane only under the influence of the electrical potential gradient and not under the influence of temperature or concentration gradient. The amount of carbon which has crossed this plane was determined by combustion analysis of the two halves of the specimen. The results of these analyses are shown in the sixth and seventh columns of table I. It will be noted that substantial amounts of carbon migrated from the anode portion to the cathode portion (except at 920°C, which temperature was too low to exhibit the effect in the allotted time). In spite of precau-tions taken, considerable decarburization occurred as may be seen by comparing cols. 8 and 5. In three cases (indicated by asterisks in table I) the specimen was ground on one side, polished, and examined metallographically before analysis. One specimen was normalized and re-examined. Micrographs of selected areas in specimens 4 and 6 are shown in figs. 1 and 2, respectively. Evidence of carbon migration from the anode toward the cathode is quite apparent. Fig. 2 shows abundant graphite as well as cementite in a region near the cathode. The temperature of this zone is estimated at about 800°C. In fig. 1, particularly in the micrograph on the extreme right (the right side of which is of unaltered carbon content), there is evidence of slight depletion of carbon in the immediate vicinity of the cathode as contrasted to the general increase in carbon of the cathode half of the specimen. Although this may be attributed in part to decarbur-ization, it is believed that it is occasioned also by thermal diffusion—migration (of carbon) from a cool to a hot zone4—a phenomenon superimposed upon the electrical migration. Evidence of carbon migration was visible particu-larly during the course of the high-temperature,

Jan 1, 1951

By Ralph E. Esarey, Robert G. Reno

DriLLing activity and prospecting for oil and gas increascd in Indiana in 1941 over the previous year. The greater part of the drilling and development continued to centralize in the Indiana portion of the Interior Coal Basin, in the southwestern part of the state, but other parts of the state were not neglected. Many wildcat wells were drilled, including a series of core tests through the mantle of drift on the southern flank of the Michigan Basin. These core tests have been followed by several unsuccessful production tests. Production tests were made also along the axis of the Logansport-Kankakec branch of the Cincinnati Arch. In all, 411 field locations and 141 wildcat wells were completed in Indiana during 1941; that is, a total of 552 wells, an increase of 3 2 wells over the previous year. The field tests resulted in 260 oil wells, 57 gas wells, and 94 dry holes; wildcats included 16 oil wells, 3 gas wells and 122 dry holes. At the end of the year, 53 field wells and 34 wildcats were drilling. Gibson and Posey Counties led the state, with 154 and 155 wells drilled, respectively. However, 125 wells produced oil in Gibson County, while 98 produced oil in Posey County. Twenty-five per cent of all wells drilled were wildcat; II per cent of these produced oil and 5 per cent produced gas. Grifin field continued to lead in drilling activity, with 157 successful wells and 20 dry holes (including important wells defining the field). Production is found in five pay zones (Tar Springs, Waltersburg, Elwren, Aux Vases and McClosky). It produced a major part of Indiana's estimated 7,200,000 bbl. of oil during the year. Heusler field, Posey County, expanded with 11 oil ells and 4 dry holes in the Waltersburg and Tar Springs. Vienna field, Posey County, completed four oil producers in the Mansfield sandstone, between 1200 and 1250 ft. The College field, discovered late in 1940, received three oil producers and two dry holes in the Aux Vases. Two new fields were discovered in Posey County during the year: Lamott pool, in the Tar Springs, at 1950 ft., had initial production of 55 bbl. on discovery. One producer and one dry hole have been drilled since. Mt. Vernon pool found saturation in both the Waltersburg and Cypress formations. The discovery well tested about 1000 bbl. per day; it tested the McClosky at 2764 ft. Seven producers have been drilled since the October discovery. Welborne-Switch pool found oil in the Cypress sand at approximately 2500 ft., with an initial production of 33 bbl. One more well in the area proved a dry hole. Gibson County had the following discoveries in 1941: The Patoka field was opened by a McClosky producer of 300 bbl. Another producing well has been drilled. An attempt was made to extend the Illinois Mt. Carmel field into Gibson County, Ind., and one producer of 24 bbl, was completed on the Indiana side. A dry hole has since been drilled, one location south. One small well was completed in the Mans. field at 1105 ft. in sec. 31, T.2 S., R.12 W., Gibson County promising a new field

Jan 1, 1942

By Robert G. Reno, Ralph E. Esarey

DriLLing activity and prospecting for oil and gas increascd in Indiana in 1941 over the previous year. The greater part of the drilling and development continued to centralize in the Indiana portion of the Interior Coal Basin, in the southwestern part of the state, but other parts of the state were not neglected. Many wildcat wells were drilled, including a series of core tests through the mantle of drift on the southern flank of the Michigan Basin. These core tests have been followed by several unsuccessful production tests. Production tests were made also along the axis of the Logansport-Kankakec branch of the Cincinnati Arch. In all, 411 field locations and 141 wildcat wells were completed in Indiana during 1941; that is, a total of 552 wells, an increase of 3 2 wells over the previous year. The field tests resulted in 260 oil wells, 57 gas wells, and 94 dry holes; wildcats included 16 oil wells, 3 gas wells and 122 dry holes. At the end of the year, 53 field wells and 34 wildcats were drilling. Gibson and Posey Counties led the state, with 154 and 155 wells drilled, respectively. However, 125 wells produced oil in Gibson County, while 98 produced oil in Posey County. Twenty-five per cent of all wells drilled were wildcat; II per cent of these produced oil and 5 per cent produced gas. Grifin field continued to lead in drilling activity, with 157 successful wells and 20 dry holes (including important wells defining the field). Production is found in five pay zones (Tar Springs, Waltersburg, Elwren, Aux Vases and McClosky). It produced a major part of Indiana's estimated 7,200,000 bbl. of oil during the year. Heusler field, Posey County, expanded with 11 oil ells and 4 dry holes in the Waltersburg and Tar Springs. Vienna field, Posey County, completed four oil producers in the Mansfield sandstone, between 1200 and 1250 ft. The College field, discovered late in 1940, received three oil producers and two dry holes in the Aux Vases. Two new fields were discovered in Posey County during the year: Lamott pool, in the Tar Springs, at 1950 ft., had initial production of 55 bbl. on discovery. One producer and one dry hole have been drilled since. Mt. Vernon pool found saturation in both the Waltersburg and Cypress formations. The discovery well tested about 1000 bbl. per day; it tested the McClosky at 2764 ft. Seven producers have been drilled since the October discovery. Welborne-Switch pool found oil in the Cypress sand at approximately 2500 ft., with an initial production of 33 bbl. One more well in the area proved a dry hole. Gibson County had the following discoveries in 1941: The Patoka field was opened by a McClosky producer of 300 bbl. Another producing well has been drilled. An attempt was made to extend the Illinois Mt. Carmel field into Gibson County, Ind., and one producer of 24 bbl, was completed on the Indiana side. A dry hole has since been drilled, one location south. One small well was completed in the Mans. field at 1105 ft. in sec. 31, T.2 S., R.12 W., Gibson County promising a new field

Jan 1, 1942

By Fred D. Gustafson

With the postwar emergency for new housing and for new industrial buildings, much research has been done on lightweight aggregates for use in concrete and plaster. The trend toward lighter weight aggregates to relieve the dead load on the higher, steel framework buildings started toward the end of the last century. Progressively the requirements became more stringent, crushed bank slag, weighing 80 lb per cu ft as compared to 100 lb for crushed rock, was supplemented in some sections by foamed slags and expanded shales weighing from 40 to 60 lb per cu ft. In other sections cinders were used weighing from 40 to 50 lb, including sand. Sections favored with more recent vulcanism turned to pumice weighing from 30 to 60 lb, whereas other localities made use of diatomite, from 28 to 40 lb. Exfoliated vermiculite weighing from only 6 to 10 lb per cu ft came into use. As the search for lighter weights in aggregates continued so did the search for other desirable properties such as nailability, ease of cutting or channeling, and good insulation to both sound and heat. This search eventually led to a widespread interest in the expansion of perlite, or similar volcanic glasses, with the result that in 1946, the Office of the Housing Expediter' listed eleven firms reporting the operation of processing plants either on pilot or commercial scale producing aggregates weighing from 2 to 16 lb per cu ft. Geology Perlite is one of the large family of rocks originating from a granitic or acidic magma and generally has a "rhyolitic composition with a marked perlitic structure."² The name "perlite" is a derivation from "perlstein" originally given "to certain glassy rocks (hyaloliparites, hyalo-rhyolites) with numerous concentric cracks, from the fancied resemblance of broken out fragments to pearls."3 The cause of the perlitic texture is somewhat controversial. Johannsen attributes the texture to strain set up in the glass by cooling and describes the finding of peculiar, glassy balls, called marekanite, more or less rounded and showing concave indentations, which may be left when perlite is broken. Frank Rutley, on the other hand, in 1881, while working on vitreous rocks from Montana, observed that there was incipient or partially started crystallization with segregation of water toward the un-crystallized portion in all specimens of perlite examined. Further he observed evidence of strain around focii of crystallization to which he ascribed the development of the spherical cracking which gives the rock its perlitic texture.4 Widespread over the western and southwestern United States where Tertiary vulcanism is exposed, perlite commonly is associated with andesitic, basaltic, and rhyolitic lavas with inter-bedded tuffs and ash of Eocene age although some deposits are associated with rocks assigned to the Pliocene age.5, 6 n nearly all cases the perlite overlies water laid tuff beds or agglomerates or breccias of explosive nature, an outstanding exception being the occurrence of perlite as the chilled selvage of andesite intrusions.7'8 Available chemical analyses of per-lites from the western and southwestern United States indicate that they are varieties of leucorhyolites rather than of normal rhyolites, leucorhyolites having less than 5 pct of dark minerals, normal rhyolites having more than 5 pct. This conforms with Johannsen's observation that leucorhyolites are the prevailing kind in North America.9 Tables 1 and 2 give

Jan 1, 1950

By W. L. Kerlie, J. H. Richards

The variation in hydrogen content of basic open-hearth steels during refining and the effect of changing from steam to air atomization were studied. The water content of the furnace atmosphere, the slag basicity, and the degree of oxidation of the slag-metal system largely control the hydrogen in the steel. SMALL amounts of hydrogen have an adverse effect on the ductility of steel and a wide range of hydrogen contents has an adverse effect on the internal soundness of both ferrous and nonferrous castings. This has been of major concern to the metal industry, both manufacturers and fabricators, for many years. As one of the results of continuing investigations of the hydrogen problem, much has been learned about methods of reducing the hydrogen content of liquid metals below the relatively high levels that are normally associated with such phenomena as rising ingots and excessive porosity in sand casting. However, it was not until a reliable and simple method for sampling and analyzing liquid metal was developed that much progress was possible in quantitatively evaluating the effect of steelmaking practice on hydrogen contents at levels below 6 to 8 ppm. Throughout all the work described in the following presentation the Geffner pintube technique was used to obtain samples of liquid metal1 * and the tin-fusion method, developed at Massachusetts Institute of Technology,2 was used to analyze the samples. Before attempting a systematic experimental program for minimizing the hydrogen content of open-hearth steels, it was necessary to determine the variation in hydrogen in the open-hearth bath during the normal processing of a heat. The logs of ten heats are illustrated in Figs. 1 through 7. The number at each point on the graphs refers to the number of pintube samples analyzed for hydrogen. An increase in the hydrogen content of the liquid metal during the early period of refining is apparent in all heats. There is evidence of a decrease in hydrogen later in the refining period, see Fig. 2, for example. Therefore, the existence of a maximum hydrogen content at some intermediate time during the refining period is anticipated. To test for the existence of a maximum, the data illustrated in Figs. 1 to 5 were grouped together so that they could be examined on a collectively comparable basis. The grouping was effected, as illustrated in Fig. 8, by superimposing Figs. 1 to 5 so that the apparent hydrogen maxima coincided at a fixed position, indicated by the zero on the time axis. Also, by vertical adjustment until the graphs appeared to the eye to be superimposed, the hydrogen scale was made to show the change in hydrogen content. That is, a standard amount was subtracted from the hydrogen values for each heat. Therefore, Fig. 8 represents a generalized presentation of the data and, for this reason, cannot be used to predict the actual hydrogen content of the bath during refining, but rather to show the probable rate of change in hydrogen during refining. Statistical examination of the grouped data in Fi.g. 8 shows that the apparent. increase and suhsequent decrease in hydrogen are real and confirms the existence of a maximum in general. The average increase In hydrogen during the early period of refining is between 0.8 and 1.3 ppm per hr, and the

Jan 1, 1958

By K. Torssell, U. Lindborg

The kinetics of precipitation deoxidation is considered from a theoretical point of view. The size distribution and the total content of deoxidation products are estimated statistically as a function of time, taking into account Stokes and gradient collisions between the particles and separation at the top surface of the bath. The assumption is made that two particles coalesce into one larger particle as a result of a collision. General expressions are derived and calculations are carried out numerically for the case of silica particles in liquid iron. The analysis predicts a three-stage behavior for the removal of oxygen, an incubation stage, a stage of rapid separation, and a slow final stage where the content approaches the equilibrium value. The agreement with experimentally determined size distributions and deoxidation times is considered to be good. A quantitative estimation of such growth mechanisms as diffusion from the melt and redistribution of oxygen from small particles to large ones shows that these latter mechanisms cannot explain the experimentally observed gvowth of inclusions. This also indicates that collisions provide the important growth mechanism. OXIDE inclusions precipitate when a deoxidizer like silicon or aluminum is added to a steel bath. The inclusions grow and the large particles tend to separate from the bath. For each type of oxide the growth determines the sizes of the final inclusions as well as the total inclusion content of the steel. The mechanism of growth is controversial, however. It has been suggested that an inclusion grows principally by diffusion of atoms through the melt to the inclusion particle or principally by collisions and coalescence with other inclusions. Diffusion growth is always present to some extent but recent experiments and calculations by Torssell,' Kniippel et 'a1.,2 Sano et al. and Miya-shita4 indicate that collisions may be more important than previously realized. A detailed analysis has so far been lacking. The present paper is a quantitative, statistical analysis of the growth and separation of inclusions and is based on a collision theory. Nuclea-tion is not considered. The numerical estimates are based on the experimental conditions of Torssell's' investigation and comparisons are made with his observations. Torssell' studied the growth and separation of silica particles in liquid iron. Samples were taken from a 0.6 kg iron bath at various times after the addition of silicon. By a particle-counting technique developed by Bergh and Lindber~the size distribution of the inclu- sions was determined. The number of particles at the first sampling, 10 sec after silicon addition, was of the order of 3 x lo7 to 10' per cu cm, the largest particles having a radius of 2 to 4 pm. The volume fraction was 0.001 to 0.004; i.e., the dispersion was very dilute. During subsequent holding the largest particles increased their radius by a factor of 10 after about a minute, and the whole size distribution was displaced toward larger values. These results were obtained with allowance for precipitation of the equilibrium concentration of oxygen during solidification. The bath contained less than 0.01 pct of C. Figs. 1 and 2 show the typical appearance of the inclusions. Most particles are very closely spherical and appear to have been liquid or semiliquid in the melt. The observed number of inclusions, about 5x 10' per cu cm, is of the same order of magnitude as found by ~er~h.' This is a large value. It follows that the diffusion distances are very short and that the precipitation

Jan 1, 1969

By Raymond D. Sloan

Dropping from third position among the oil-producing states of the nation in 1942, Oklahoma ranked fourth in 1943 with a total output of 121,431,ooo bbl., a decline of 11.9 per cent from the previous year's total of 137,792,000 bbl. Production ranged from a maximum of 346,900 bbl. daily in February to a low of 322,800 bbl. daily in October, and at the close of the year was averaging 330,000 bbl. daily from 52,096 wells, or an average of 6.3 bbl. per well. The state's well, average at the close of 1942 amounted to 6.7 bbl. At the turn of the year, it appeared reasonable to assume that Oklahoma's producing rate would continue its downward trend during 1944 unless reversed by greater than normal discoveries. To Jan. I, 1944, Oklahoma has produced 5,188,466,-ooo bbl. of crude oil. Development and Exploration A total of 1287 wells was drilled in 1943, representing an increase of 77 wells, or 6.4 per cent over the preceding year's total of 1210. Of the wells completed, 587 (45.6 per cent) were oil wells with an average initial production of 319 bbl.; 127 (9.9 per cent) were gas wells with an average initial of 8,639,000 cu. ft.; and 573 (44.5 per cent) were dry. A total of 333 wildcats was drilled (25.9 per cent of all wells), an increase of 13 (4.1 per cent) over the 1942 total of 320. Of the*wildcats drilled, 46 (13.8 per cent) were oil wells with an average initial production of 235 bbl.; 27 (8.1 per cent) were gas wells with an average initial of 8,249,-ooo cu. ft.; and 260 (78.1 per cent) were recorded as failures. Reconditioning of wells throughout the state led to a total recompletion of 230 wells. Of these, 131 were oil wells with a combined initial production of 8360 bbl., 31 were gas wells with a total initial of 211,950,000 cu. ft., and 68 were abandoned. The 46 successful wildcats opened 33 new producing areas, one of which produced from three separate horizons, and led to the development of production in new horizons in 11 old pools. While none of the 1943 discoveries have as yet developed into what might be termed major pools, the West Moore pool, in Cleveland County, and the West Edmond pool, in Oklahoma County, are the two outstanding new areas opened to production. The West Moore pool, sec. 29, 10 N., 3 W., discovered in December 1943, was completed for an initial production of 248 bbl. of oil in 3 hr. through small tubing chokes in the second Wilcox sand from 8787 to 8800 ft. One well was drilling in this area at the close of the year. The West Edmond pool, an April discovery in sec. 32, 14 N 4 W., is producing from the Hunton lime at a depth of approximately 6950 ft., the world's deepest Hunton lime pool. To the end of the year, 15 oil wells had been completed, with no gas wells and no failures. Accumulated production to Jan. I, 1944, amounted to 474,300 bbl. of 41" gravity oil. There were 29 active operations in the field at the end of the year. The West Edmond and West Moore pool discoveries established the first commercial production west of the Granite Ridge,

Jan 1, 1944

By Raymond D. Sloan

Dropping from third position among the oil-producing states of the nation in 1942, Oklahoma ranked fourth in 1943 with a total output of 121,431,ooo bbl., a decline of 11.9 per cent from the previous year's total of 137,792,000 bbl. Production ranged from a maximum of 346,900 bbl. daily in February to a low of 322,800 bbl. daily in October, and at the close of the year was averaging 330,000 bbl. daily from 52,096 wells, or an average of 6.3 bbl. per well. The state's well, average at the close of 1942 amounted to 6.7 bbl. At the turn of the year, it appeared reasonable to assume that Oklahoma's producing rate would continue its downward trend during 1944 unless reversed by greater than normal discoveries. To Jan. I, 1944, Oklahoma has produced 5,188,466,-ooo bbl. of crude oil. Development and Exploration A total of 1287 wells was drilled in 1943, representing an increase of 77 wells, or 6.4 per cent over the preceding year's total of 1210. Of the wells completed, 587 (45.6 per cent) were oil wells with an average initial production of 319 bbl.; 127 (9.9 per cent) were gas wells with an average initial of 8,639,000 cu. ft.; and 573 (44.5 per cent) were dry. A total of 333 wildcats was drilled (25.9 per cent of all wells), an increase of 13 (4.1 per cent) over the 1942 total of 320. Of the*wildcats drilled, 46 (13.8 per cent) were oil wells with an average initial production of 235 bbl.; 27 (8.1 per cent) were gas wells with an average initial of 8,249,-ooo cu. ft.; and 260 (78.1 per cent) were recorded as failures. Reconditioning of wells throughout the state led to a total recompletion of 230 wells. Of these, 131 were oil wells with a combined initial production of 8360 bbl., 31 were gas wells with a total initial of 211,950,000 cu. ft., and 68 were abandoned. The 46 successful wildcats opened 33 new producing areas, one of which produced from three separate horizons, and led to the development of production in new horizons in 11 old pools. While none of the 1943 discoveries have as yet developed into what might be termed major pools, the West Moore pool, in Cleveland County, and the West Edmond pool, in Oklahoma County, are the two outstanding new areas opened to production. The West Moore pool, sec. 29, 10 N., 3 W., discovered in December 1943, was completed for an initial production of 248 bbl. of oil in 3 hr. through small tubing chokes in the second Wilcox sand from 8787 to 8800 ft. One well was drilling in this area at the close of the year. The West Edmond pool, an April discovery in sec. 32, 14 N 4 W., is producing from the Hunton lime at a depth of approximately 6950 ft., the world's deepest Hunton lime pool. To the end of the year, 15 oil wells had been completed, with no gas wells and no failures. Accumulated production to Jan. I, 1944, amounted to 474,300 bbl. of 41" gravity oil. There were 29 active operations in the field at the end of the year. The West Edmond and West Moore pool discoveries established the first commercial production west of the Granite Ridge,

Jan 1, 1944

By John P. Huger

me phase diagram for the system Ag-Si has been established by thermal analysis. The diagram is a simple eutectic type with a eutectic at 3.0 pct Si and 840°C. From the liquidus curve and with the aid of regular solution theory, the thermodynamic properties of the liquid Ag-Si system at 1420 °C have been determined. Silicon exhibits positive deviations from Raoultain behavior at all compositions, while silver exhibits both positive and negative deviations. MEASUREMENTS of silicon activities in liquid iron alloys by means of distribution studies between the immiscible liquids silver and iron have been reported by Chipman, Fulton, Gokcen, and caskey.' The applicability of this technique is dependent, however, on the accuracy of the silicon activity values in the Ag-Si alloys, particularly in the dilute silicon region. Darken and Gurry2 have shown that log ?si/(1-xSi)2 is approximately linear with composition, but uncertainty exists in the extrapolation to dilute solutions. Their calculations are based on the old Ag-Si phase diagram of Arrivaut3 and any variation in the position of the liquidus curves will strongly influence the extrapolation. The purpose of this study is to redetermine the Ag-Si phase diagram and calculate the associated thermodynamic properties. EXPERIMENTAL The Ag-Si phase diagram was determined by thermal analysis of liquid alloys. The alloys were contained in a 35 mm alumina crucible under a protective stream of purified argon. The crucible was placed inside a 64 mm alumina tube fitted with a gas-tight, water-cooled, copper head. The tube and assembly were heated in a manually controlled globar furnace. The temperature was measured with a Pt, Pt + 10 pct Rh thermocouple in a high-temperature porcelain protection tube. The thermocouple calibration4 was checked against the melting point of pure silver at the start and during the course of the experiments, and against both gold and silver at the completion of the experiments. The alloys were prepared from granular silver analyzing 99.95 pct Ag and powdered silicon analyzing 99.88 pct Si and 0.026 pct Fe. Approximately 150 g of alloy were used for each run. The composition of the alloy was determined by withdrawing samples of the liquid alloy with a vycor tube-aspirator bulb device. Analysis of solidified samples agreed, for alloys containing less than 23 pct Si, within ±0.3 pct Si of the calculated composition. Duplicate samples showed a variance within 50.05 pct Si of the mean analysis. The alloy was heated to 100' to 20pC above the estimated liquidus temperature and then the furnace was adjusted to give an average cooling rate of 2" to 7°C per min. A cooling curve was obtained by taking potentiometer readings at 15-sec intervals. Several analyses were made by recording the thermocouple output on a high-speed, Stip-chart recorder, with the exception of run NO. 11, a minimum of two thermal analyses were made for each alloy and the cooling curves then obtained duplicated the liquidus and eutectic temperatures to about 1°C. RESULTS The results of the thermal analysis are given in Table 1, and are represented as the Ag-Si phase dlagram, Fig. 1. The eutectic composition was found

Jan 1, 1963

By G. G. Priest, B. E. Morgan

The production of oil and gas from wells is often seriously hindered by the compaction of solids in the perforations during the perforating process. To realize the full productive capacity of a well, it is necessary either to remove the plugging material from the perforations or to perforate in a manner which does not cause plugging of the perforations. The latter method can be accomplished by having a non-plugging fluid in the casing at the time the well is perforated. This paper describes the results of laboratory tests, perforating tests, and experimental field tests which demonstrate that certain emulsions serve effectively as non-plugging perforating fluids. The principal components of these emulsions are two liquids: an aqueous phase consisting of a solution of either sodium chloride or calcium chloride, and an oil phase consisting of diesel oil, tetrachloroethylene, or a combination of both. Small amounts of other agents which serve to stubilize the emulsions, to reduce the fluid loss, and to inhibit corrosion are also present. The density of the emulsions is controllable within the range of 7.5 to 12.5 lb/gal; no insoluble solids are present, the filtration rate is low, and the thermal stability is satisfactory at temperatures as high as 200°F. The corrosion of steel by these emulsions is negligible. Tests conducted under conditions which simulate the temperature and the pressure in an actual well demonstrate that these emulsions fulfill the requirements of a non-plugging perforating fluid. In addition, lirnited field experience furnishes evidence that these emulsions remain in position after they are spotted, that they are retained in the casing after the perforations are made, and that they are easily removed from the perforations when the well is opened for production. When the pressure in the borehole is greater than the formation pressure at the time a well is perforated, plugging of the perforations can be avoided by spotting an emulsion adjacent the interval to be perforated. This procedure should eliminate low well productivity caused by plugging the perforations. INTRODUCTION Nature of Problem The most widely used method for the completion of oil and gas wells is to set and cement casing through productive horizons, and then to obtain production by gun-perforating the section of casing opposite the desired interval. This method is gen- erally successful and it was thought at one time that those failures which did occur were due to insufficient penetration by the perforator. Despite recent improvements in perforators which assure ample penetration, unsatisfactory fluid production continues to be a problem. Evidently, the trouble comes from some other source. Recent results obtained by Allen and Worzel1 demonstrated that low fluid productivity may be caused by the perforating process and/or by the fluid in the wellbore at the time of shooting. They found that perforations are often completely plugged by debris from the perforator, pulverized sand from the formation, and compacted solids from the drilling mud when the pressure in the wellbore is greater than the pressure in the formation at the time of perforation. In some cases a differential pressure in excess of 400 psi was required to initiate flow into the well through these plugged perforations. The plugging of perforations by solids not only causes low productivity from wells, but it creates simultaneously other problems in the production of oil and gas. For example, when most of the perforations are plugged, high rates of flow must occur through the few perforations that are open. The resulting high pressure drop may cause production

Jan 1, 1958

By H. H. Rachford, D. S. Howard

In the withdrawal of fluids from tilted aquifers it is of value to be able to predict pressure patterns during the course of the pressure decline. As an example of this, in the displacement of fluids from oil reservoirs being produced by water-drive, prediction of the reservoir pressure is an important factor in fixing production rate. The prediction of pressure patterns in the depletion of an inclined aquifer has conventionally been made by solving the well known flow equations for the pressure Patterns in a horizontal aquifer, and correcting these patterm for the pressure difference due to the hydrostatic head associated with the vertical distance between a point in the horizontal aquifer and the corresponding point in the tilted aquifer. Inasmuch as the equation describing flow in tilted strata is different from that for horizontal flow. there was question as to the validity of the simple corrective procedure current1y in use. To examine the magnitude of error made using this procedure, an inclined aquifer was chosen for which the error was expected to he large, and the equation for flow in this tilted stratum was solved numerically. The results of the conventional procedure of correcting the solution of the equation for horizontal flow was compared with the solution of the correct equation by studying the difference between predicted pressures by the two methods. Three forms of the hydrostatic head correction were studied. The results for all three forms of the head correction are similar. The error in the approximate solution is quite dependent on the position of the horizontal aquifer with respect to the inclined one. The error is greatest at the producing end ad increases with amount of fluid withdrawn. The magnitude of the errors is not large if a good choice is made for the position of the horizontal datum. The most satisfactory position for the datum in all cases is near or slightly about the mid-point of the inclined stratum. Use of the best approximate method studied with the best datum tested yielded predicted pressures in error no greater than 7 psi. INTRODUCTION The movement of water by expansion from an aquifer to displace fluids from a contiguous oil reservoir was first recognized over 20 years ago.9 Since that time much work has been done to relate this mechanism to reservoir behavior mathematically. The result of this effort on the part of many workers has led to an understanding of underground fluid movement which has enabled us to follow and predict the behavior of the water-drive reservoir. One reason that good progress has been made in the application of mathematics to the study of pressure patterns and water movements in aquifers has been that the differential equation which describes this process is of a relatively simple form. The equation for the isothermal, horizontal, viscous flow of water through homogeneous sandstone is identical with that for the diffusion of heat through homogeneous solids, and this equation and means of obtaining solutions for it with many types of boundary conditions have been studied extensively. That this work has been highly successful is clearly emphasized by the large number of solutions developed for problems of interest both by methods of formal mathemat-

Jan 1, 1957

By J. W. Evans

ONE of the most important and frequent calculations that the extractive metallurgist is called upon to make is that of the standard free energy change of a reaction (?F°). For many reactions of metallurgical interest, the lengthy arithmetical calculations have been eliminated by the use of free energy-temperature diagrams which have been constructed by Ellingham,1 Richardson and Jeffes,2 Kellogg,3 and Osborn4 (e.g., for oxides, sulphides, halides) from critical surveys of existing data. The speed and simplicity of making calculations with these graphs make them of great practical value and, this should be stressed, in the majority of cases no greater accuracy is obtained by the individual calculations from specific heat data, entropies, and heats of reaction. With the increasing industrial application of vacuum methods to the winning of metals, notably at present magnesium and the alkaline earth metals, it is necessary to calculate low metal vapor pressures above reaction mixtures. The existing free energy diagrams usually give values where the solid and liquid metals are at unit activity. To obtain the free energy change associated with the reaction Metal (solid or liquid) = Metal (gas) it is customary to calculate this from either vapor pressure equations or from the appropriate free energy functions. This paper presents graphically the free energies of vaporization of a number of metals of economic importance which can be used in conjunction with other free energy diagrams. Certain metals have been excluded on the following grounds: titanium, tungsten, molybdenum, and platinum have very high heats of vaporization and boiling points and the uncertainty of these values arising from difficulties of high temperature measurements makes free energy values of doubtful accuracy; the data for cobalt, vanadium, and zirconium are uncertain; finally, arsenic, antimony, and bismuth could not be conveniently included. because of the presence of mixtures of polyatomic and mona-tomic vapors in the temperature range considered. Free Energy Diagram In Fig. 1, the standard free energies of vaporization per mol (?F°T) have been plotted for the temperature range 0" to 2000°C for the reaction M (solid or liquid) = M (gas) The data have been obtained from the free energy functions — (F-H298)/T for the solid, liquid, and gaseous elements and the heats of vaporization at 298°K, i.e., ?H298 (vapor). The values of these functions have been taken without change from the recent critical survey by Brewer;' the bulk of this data is from Kelley7 with adjustments in the light of more recent work. Values of the function — (F-H298) /T at T = 298°, 500°, 1000°, 1500°, and 2000°K were employed. Where possible, a further independent value was obtained by using the boiling point, i.e., where ?F°T = 0. All calculations are based on the assumption that the metal vapors are monatomic and that the boiling points refer to those temperatures at which the partial pressure of the monatomic vapor above the metal is 760 mm. For the alkali metals, e.g., sodium, potassium, and lithium, diatomic gas molecules are present as minor species; details of these are discussed by Brewer." On the basis of the above values and allowing for the change in slope at the melting point, straight line graphs were constructed and extrapolated to 2000°C. The deviation from a straight line did not exceed ±500 cal, which in all cases appears less than the probable error in the ?F°T values.

Jan 1, 1954

By H. B. Heller, T. J. LaChapelle

Phosphorus nitride has been used as a diffusant for introducing phosphorus into silicon under various conditions. It has a temperature -dependent rate of decomposition beginning in the 500°C range, increasing very rapidly above 1000 C, producing gas -eous phosphorus and nitrogen. Best control was obtained with a two-temperature arrangement, with silicon at the proper higher diffusion temperature. It is an easier compound to use than the usual P2O5 because it is nonhygroscopic and yields temperature-controlled surface concentrations together with excellent final surfaces. Surface doping nearly as high as that produced by phosphorus pentoxide has been observed for sources at 850" to 900°C. Reproducible high sheet resistances were obtained by using slow flow rates (1 linear in. per min) and staggered-baffle gas mixing with the nitride at 550°C. Both inert gas (argon, nitrogen) and oxidizing ambients have been examined. Oxygen reacts to produce anhydrous P2O5, consuming the source at a rapid rate. Nitrogen, except for short diffusion times, suppressed the dissociation reaction and sometimes produced inferior surfaces. Argon appears to be a more useful ambient. PHOSPHORUS is a common impurity atom which may be introduced into a semiconductor lattice to create a p-n junction or an nn contact region. Various investigators1-4 have described techniques which can be used with either elemental phosphorus or certain of its compounds. This paper discusses a newly available compound of phosphorus, phos- some processes silicon wafers are precoated with a phosphate glass which acts as the doping source. Although a nonoxidizing form like elemental phosphorus or a phosphine can be used, most diffusion methods usually involve the use of phosphorus pentoxide. Because of the rapid volatility of this compound, a two-step process is often used, combining a gaseous deposition to introduce a predetermined quantity of phosphorus into the silicon with a later redistribution to obtain the desired distribution and surface concentration. Such a process usually involves predepositing phosphorus at surface concentrations greater than 1020 cm-'. At such doping levels the phosphorus-diffusion coefficient appears to be strongly concentration-dependent.596 Distribution profiles5'9 for these high concentrations of phosphorus do not follow Fick's laws, having extended subsurface regions which appear electrically to be nearly uniformly doped with ionized impurity atoms. Radioactive-tracer experiments also show near-surface anomalies in the distribution profile. Additionally, it has been shown1' that the introduction of very high surface concentrations of impurity atoms into silicon (a step necessitated by the subsequent lowering of concentration during the redistribution) creates dislocations, even in material having no dislocations originally. For device fabrication the use of oxide masking introduces certain difficulties and limitations.7'8 The standard practice of using phosphorus pentoxide as an impurity source adds other problems to these because of the speed and ease with which this compound reacts with moisture present in any gas in contact with it. In fact, P2O5 is one of the fastest and most effective drying agents for gases. Because it is such an excellent desiccant, completely anhydrous phosphorus pentoxide is difficult to obtain and is even more difficult to keep dry. Adequately dry gloved boxes must be used for storage and transfer of the chemical. The presence of even traces of moisture will cause formation of phosphoric acids which, during the high-temperature diffusion, results in polyphosphoric acids of variable composition. Such reaction by-products end up as gradually accumulating viscous fluids in the cooler parts of the diffusion furnace. The calculation of the final solute distribution for

Jan 1, 1964