US3422019A - Method for processing ferrite cores - Google Patents

Description

Jan. 14, 1969 JAMES 5, 535 3,422,019

AND SPACE ADMINISTRATION ADMINISTRATOR OF THE NATIONAL AERONAUTICS METHOD FOR PROCESSING FERRITE CORES Sheet Filed April 30, 1965 n m M m W m o w c w um Y Y W WI V V l T Q m C m m w m m m 4 a P .i N E m m M T 5 E "I R R U T A R E P M E T n n 0 w n 5 w FIG.I.

}TEMPERATURE I TIME PRODUCT IOOO 1 mumIh IOO QUENCH RATE DEG/SEC.

FIG 4 INVENTOR Albert W. Vinol M FTZRNEYS JAMES E. WEBB 3,422,019

ONAL 7 Jan. 14, 1969 ADMINISTRATOR OF THE NATI AERONAUTICS AND SPACE ADMINISTRATION METHOD FOR PROCESSING FERRITE CORES Sheet g of 4 Filed April 50, 1965 .PODQOma m2; I mmnk mmmiub momhmmmo O N mouhwmmO m;

was/93o 31w HoNsIno N QE Fuaooma m2; I wmDh mua2uP 6 5 j '6 50318830 H All/HD8300 OIOHSHHHJ.

INVENTOR Albert W. VincIl Jan. 14, 1969 JAMES E. WEBB 3,422,019

ADMINISTRATOR OF THE NATIONAL AND SPACE ADMINISTRATION AERONAUTICS METHOD FOR PROCESSING FERRITE CORES Sheet 3 of 4 Filed April 30, 1965 PIC-3.5.

FIG.8.

INVENTOR Albert W. VincII on m 0 0 00 0O 0 0 0 on I I M A 'FIZDRNEYS Jan. 14, 1969 AERONAUTICS AND SPACE ADMINISTRATION Filed April 30, 1965 THRESHOLD COERCIVITY (H )-INCREASE "I" PEAK AMPLITUDE (NORMALIZED) JAMES E. WEBB 3,422,019

ADMINISTRATOR OF THE NATIONAL METHOD FOR PROCESSING FERRITE CORES Sheet 4 014 SLOW QUENCH (THICK BOAT) FAST QUENCH (THIN BOAT) o l :60 I000 f -TIME BEFORE QUENCH (SECONDS) FAST QUENCH (THIN BOAT) MED.QUENCH .25"\ CONSTANT TEMP-TIME SLOW QUENCH muck BOAT) PRODUCT lb 160 f -TIME BEFORE QUENCH (sscouos) INVENTOR Fl 6 Albert W. Vinal Q YM United States Patent Oflice 3,422,019 Patented Jan. 14, 1969 METHOD FOR PROCESSING FERRITE CORES James E. Webb, Administrator of the National Aeronautics and Space Administration with respect to an invention of Albert W. Vina], Owego, N.Y.

Filed Apr. 30, 1965, Ser. No. 452,422

US. Cl. 25262.64 Int. Cl. C04 35/26 2 Claims ABSTRACT OF THE DISCLOSURE The present invention relates generally to the manufacture of ferrite cores, and more particularly to an improved firing process that optimizes the response characteristics of ferrite cores of the type used in magnetic memory storage systems.

The use of ferrite cores as memory storage devices (toroids and multi-path core arrangements) is well established in the computer art. Cores of this type are manufactured by compressing finely divided particles of ferrite oxide and one or more other metal oxides such as magnesium, manganese and copper into a solid body of desired size and shape. This body is then subjected to a firing process (sintered) at elevated temperatures and under controlled atmospheric conditions to produce a vitrified, ceramic-like material.

Significant electrical characteristics of ferrite cores are the and 1 responses generated when appropriate windings of the memory system are energized by a specific sequential current pulse program. The peak amplitude of the 1 response during energization is proportional to the magnitude of the energizing field above the critical threshold field H As an example of a particular application, for a random access memory system the address currents which selectively energize one or more cores are set to a specific value in accordance with core properties and system usage. The magnitude of these address currents are determined primarily by the coercivity of the cores and the inner dimensions of the cores through which the currents are passed. The relationship between aperture diameter, coercivity and current magnitude can be expressed by I='YHO1TD, where I is current magnitude, 7 is a drive current factor determined by system configuration, H is the threshold coercivity of the core and D is the aperture diameter. Usually, 'y is .85 for a three-dimensional system and 1.5 for a two dimensional system.

Thus the threshold coercivity is a significant factor for successful memory operation, particularly for a threedimensional coincident current system. The numerical value for H. for a particular core material depends to a large degree upon the sintering of the cores, and the performance and reliability for an array of cores in the system can be jeopardized with I-L, either above or below a spe cific range about a nominal value.

Heretofore, standard practice has been to individually test each sintered core before being assembled into memory planes, and cores that do not meet specific electrical limits are excluded from further use in the system. However, it has been found that a principal cause for rejecting cores has been inadequate control of the sintering in that the threshold coercivity H fluctuates from lot to lot and from core to core within a lot. Thus, with good ferrite material used when the cores are pressed, and care exercised to maintain mechanical uniformity, the acceptable product yield for a specific lot of cores depends on the sintering. For sintered cores possessing a threshold coercivity less than the desired value it was heretofore assumed that nothing could be done to raise its threshold coercivity to meet specifications, and as a consequence of this so called over-firing, the core was discarded for a particular application.

Successful memory system operation also requires a large ratio between the 1 and 0 electrical responses of the cores. This, in turn, may be achieved, for a given threshold coercivity, by minimizing the 0 response of the cores. It has also been found that the 0 response is, to a great extent, dependent upon the depth of an oxide layer of the sintered core, and that a long temperaturetime product during sintering results in an excessively thick oxide layer and a relatively high 0 response.

Accordingly, it is an object of the present invention to provide a new and improved firing process which increases the quality and product yield for ferrite cores of the type used in computer memory systems.

Another object of the invention is to provide an improved firing process for ferrite cores to obtain a desired threshold coercivity and to optimize the 1 to 0 electrical response ratio of the cores.

A further object of the present invention is to provide an improved firing process for ferrite cores which permits the reversible control of their threshold coercivity so that individual cores can be reprocessed to meet specified limits for the threshold coercivity.

Still another object of the present invention is to provide an improved two-step firing process for ferrite cores that allows their 0 and 1 electrical response to be controlled independently.

Other objects, as well as the features and attending advantages of the invention, will become apparent from the following description when taken in conjunction with the accompanying drawings, in which:

FIGURE 1 is a curve depicting the I and 0 electrical responses as a function of the temperature-time product of the sintering step of the two-step firing process of the invention;

FIGURES 2-4 are curves depicting the manner in which the threshold coercivity varies as a function of the temperature-time product of the sintering step and the quench rate of the annealing step of the two-step firing process of the invention;

FIGURE 5 is a diagram illustrating one technique of quench rate control in the two-step firing process of the invention;

FIGURES 6 and 7 are curves depicting the effects of the quench rate control obtained by the technique of FIG- URE 5; and

FIGURE 8 is a diagram illustrating a further technique of quench rate control in the two-step firing process of the invention.

In the manufacture of ferrite cores of the type used in memory systems of computer and data processing equipment in accordance to the present invention, a two-step firing process is utilized which permits reversible control of the coercivity threshold of the cores by controlling the quench rate of cores that have been sintered in accordance with a selected temperature-time product. The first step (or temperature-time controlled sintering step) of the process optimizes the 0 electrical response characteristic of the cores, and the cores thus sintered may be stock-piled or immediately processed by the second step. The second step is a controlled quench rate annealing step which optimizes the threshold coercivity or 1 electrical response characteristic of the cores. The second step is reversible and may be repeated as desired to allow the threshold coercivity to be maintained within a given range by selectively modifying the quench rate. As a result of this two-step firing process the and 1 electrical responses of the cores can be selectively and independently controlled to reduce the number of rejects, and all cores can be reprocessed to provide a different threshold coercivity, if necessary.

GENERAL DESCRIPTION In practicing the present invention, pressed cores of finely divided ferric oxide and manganese, magnesium, copper or other similar metallic oxide powders are subjected to a two-step firing process. For the first step, the cores are sintered at a selected temperature and for a prescribed time in a suitable furnace. Typically this temperature may be in the range of 1300-1500 C. During this step, the furnace temperature is set so that a desired temperature-time product occurs within the prescribed time interval, typically 6075 minutes for Mn-Mg ferrite, and generally in the range of to 120 minutes, depending on the specific material used. This temperature-time product determines the depth of the oxide layer in the cores, which in turn determines their minimum 0 electrical response characteristic.

FIGURE 1 shows the relative peak amplitude of the 0 and 1 electrical responses for typical cores (of a specific constant threshold coercivity H as a function of the temperaturetime product as a result of the first or 0 sintering step of the two-step firing process. Illustratively, the curves of FIGURE 1 (and the curves for other figures to be subsequently described) may be obtained from Mn-Mg ferrite cores, for example, cores of a magnesium-manganese-ferrite consisting of 37.5 mol percent MgO, 22.5 mol percent MnO, and 40 mol percent Fe O It is to be understood that similar responses may be obtained from other ferrites (such as a copper-manganeseferrite), but that the absolute value of the temperaturetime product will differ depending on the critical sintering temperature of the ferrite material used. In any instance the shape and trend of the curves will be substantially the same.

It can be seen from FIGURE 1 that the 0 response has a relatively broad minimum range (depicted by number that corresponds to a point at which the 1 response can be maximized. This provides a Wide latitude in the control of the threshold coercivity H by the second step (quench rate annealing) of the two-step firing process. This is illustrated by FIGURES 24, to be subsequently discussed.

The cores are extracted from the furnace after the desired temperature-time product has been obtained, and may be stock-piled or immediately processed by the second step (the quench rate annealing) of the two-step firing process to produce core types which differ only in threshold coercivity. The variable which allows a desirable threshold coercivity to be selected once the 0" sintering has been completed is the quench rate of the cores and requires additional firing (for approximately three minutes) at temperature less than the first firing or 0 sintering temperatures. Once this second (or annealing) temperture has been reached, the cores are cooled at a controlled quench rate to provide the desired threshold coercivity. Typically this quench rate may be in the range of 5 to 500 C. per second. Preferred techniques for providing the quench rate control will be subsequently discussed.

It should be noted at this point that an important property of the quench rate annealing is that the threshold coercivity may be altered reversibly by repeating the second step simply and selectively by modifying the rate of quench. Thus, the annealing temperature of this step may be raised to, but not exceeding, the temperature of the first step or 0" sintering step, and the time interval involved is only that necessary to elevate the temperature of the cores to the annealing temperature. It should be understood that it is not the purpose of the second firing to trim the temperature-time product of the first firing by adding increments to the overall temperature-time product. This should be avoided, and hence the requirement that the annealing temperature of the second or quench rate annealing step be less than the temperature of the first or 0 sintering step.

The manner in which the threshold coercivity is influenced by both the temperature-time product of the 0 sintering step and the quench rate annealing step of the two step firing process is illustrated by FIGURES 24. FIGURE 2 is a plot of the threshold coercivity versus temperature-time product for a number of quench rates, FIGURE 3 is a plot of quench rate versus temperaturetime product for selected values of threshold coercivity, and FIGURE 4 is a plot of threshold coercivity versus quench rate for selected temperature-time products. The temperature-time product range for a minimum 0 electrical response is depicted by numeral 15 of FIGURES 2-4. Thus from FIGURES 2-4 it is possible to select a temperature-time product for the 0" sintering step and a quench rate for the quench rate annealing step which will produce a desired threshold coercivity (maximum 1 response) and a minimum 0 response. In addition, for a given group of cores the threshold coercivity may be raised or lowered simply by re-cycling the cores through the quench rate annealing step and controlling the quench rate in accordance with the desired threshold coercivity.

FIGURES 2-4 also indicate that there is a pronounced tendency for the degree of control over the threshold coercivity by quench rate annealing to converge as the temperature-time product increases. The optimum temperature-time product for a minimum 0 response has been found to occur at a temperature-time product which allows a high degree of control of the threshold coercivity by control of the quench rate. This permits reversible control over the threshold coercivity and at the same time allows the 0 response to be minimized. This is to be contrasted with present sintering techniques which utilize a relatively long-temperature-time product and a low, uncontrolled quench rate, resulting a lower yield of cores having an acceptable threshold coercivity and a 0 response characteristic.

QUENCH RATE CONTROL Several practical techniques are available for achieving quench rate control for the quench rate annealing step of the above described two-step firing process, as will now be described.

One technique, dual exponential quench, is illustrated by FIGURE 5. Basically this quench method includes cooling of a platinum boat containing a number of cores by contact with a large metal slab acting as a heat sink. With reference to FIGURE 5, the annealing furnace utilized in the second step of the two-step process is representatively shown by numeral 20. Furnace 20 includes hearth plate 21 and a hot zone representatively shown at 23. It is to be understood that furance 20 is adapted to carry out the annealing of the second step of the process, and accordingly hot zone 23 is adapted to anneal the cores at a temperature not exceeding the temperature providing the desired temperature-time product of the first or 0 sintering step of the process. This temperature-time product may be obtained by passing the cores through a furnace similar to that of FIGURE 5, subsequent to which time the cores may be removed and stock-piled or immediately passed through furnace 20 for the second or annealing step of the process.

A platinum boat 24, containing the cores, is rested on thermal mass brick 26 (of suitable refractory material) and both are passed over hearth plate 21 and through hot zone 23 of furnace 20. Thermal mass brick 26 and platinum boat 24 are then taken out offurnace 20 and positioned on hearth plate 27, which is similar to hearth plate 21, but located outside of hot zone 23 in an ambient environment. Platinum boat 24 is then removed from hearth plate 27 and rested on metal plate 29, a large mass of metal functioning as a heat sink.

The temperature T of platinum boat 24 containing the cores to be quenched can be defined as where T is the temperature of hot zone 23, T is ambient room temperature, '1' is the thermal time constant of thermal mass brick 26, and t is time. The purpose of thermal mass brick 26, in addition to providing means for transporting platinum boat 24 through furnace 20, is to provide means for controlling the quench rate of cores in platinum boat 24, as will :be described. It should be understood that a practical thermal time constant 1 for the thermal mass brick 26 should be in the range of 3-15 minutes.

After platinum boat 24 has been removed from thermal mass brick 26 and placed on metal slab 29 its temperature I as a function of time following thermal contact with metal slab 29, can be approximated by where T is the temperature of platinum boat 24 at the instant of contact with metal slab 29, 1- is the thermal time constant of platinum boat 24 in contact with metal slab 29, and K is the thermal conductivity between platinum boat 24 and metal sla'b 29.

The rate of change of the temperature of the cores in platinum boat 24 when in contact with metal slab 29 can be determined by differentiating Equation 2 above with respect to time to provide and the notation t in Equation 4 can be changed to r to signify time before quench, the time between removal from furnace 20 and contact with metal slab 29 for boat 24. Thus Equation 4 becomes and substituting Equation 5 into Equation 3 produces which may be approximated by Equation 7 indicates that the quench rate is exponential and can be controlled exponentially by selectively waiting for a suitable time tBQ before platinum boat 24 is brought into contact with metal slab 29.

The quench rate control obtained in accordance with the above described dual exponential method is illustrated by FIGURES 6 and 7. FIGURE 6 is a plot of the threshold coercivity versus temperature before quench (t of Equation 7) for several quench rates while FIGURE 7 is a plot of the 1 response versus t for several quench rates. This time before quench may be provided, for example, by allowing the cores to cool for approximately 2-120 seconds in an ambient environment. The quench rate, depicted as slow-medium-fast, may be adjusted by modifying such constants as K, 1 and T A further technique for quench rate control, the free fall method, is illustrated by FIGURE 8. The annealing step starts in the'same manner previously discussed by passing platinum boat 24 and thermal mass brick 26 through hot zone 23 of furnace 20 and then onto hearth plate 27. The temperature of the cores in platinum boat 24 at this point is the same as defined in Equation 1 above.

The cores are then removed from platinum boat 24 and caused to fall freely through cylindrical chamber 30 and into funnel end 31. Cylindrical chamber 30 has a height sufficient to allow the cores to cool to less than the Curie temperature before reaching funnel end 31, where they are collected in a suitable container 33.

It has been found that an air or free fall quench provides rapid exponential cooling with extreme uniformity for individual cores. During the free fall the temperature T of each core as a function time 1 may be approximated c= E c+ o( c) where T is the temperature of the core leaving platinum boat 24, 6 is the coefiicient of thermal convection between the cores and the atmosphere, and 1- is the thermal time constant associated with heat dissipation of the cores. This, in turn, is a function of core surface area and mass. The quench rate as a result of the free fall may be determined by diiferentiating Equation 8 above with respect to time to provide It can be seen from Equation 9 that the quench rate is exponential and depends on ambient temperature, exit temperature, core thermal time constant and coefficient of thermal conductivity between the cores and the atmosphere.

The core thermal time constant T is essentially constant, and the remaining factors T and 6 may be used to control the quench rate. Core temperature upon leaving boat 24, in turn, may be controlled in terms of time before quench, t as previously discussed. The coefficient of thermal convention 6 may be changed by controlling the atmosphere and/or gas temperature in cylindrical 30, and the rate of movement thereof. This free fall quenching technique is particularly suitable for use with multi-aperture cores in that all surfaces are cooled at essentially the same rate, thus eliminating the possibility of warping.

A modification of the free fall technique is also considered in conjunction with FIGURE 8. Cylinder 30 is surrounded by a source of radiant energy such as heater coils 34. Coils 34, for example, maybe embedded on the inner walls of cylinder 30, and separated from insulation 35 by reflecting surface 37. Heating coils 34 are arranged to provide a desired temperature gradient along the height of cylinder 30; that is, providing a temperature contour that is linear, exponential, trapezoidal, etc. For example, the temperature at the upper portion of cylinder 30 may be approximately equal to the temperature of furnace 20, varying linearly so that at the bottom of cylinder 30 it is slightly less than the Curie temperature of the cores.

The temperature T of the thermal column within cylinder 30 (assuming a linear temperature gradient) may be approximated by entrance velocity into cylinder 30. For all practical purposes V may be considered 0. Differentiating Equation 10 with respect to time, provides the approximate quench rate It can be seen from Equation 11 above that the quench rate is linearly proportional to the free fall time dt, and the thermal gradient coefiicient K. The rate of change of the free fall quench rate is -32K degrees per sec. With an initial entrance velocity of zero, Equation 11 can be rewritten as (12) quench rate: SKV? where S is the free fall distance below the entrance plane of cylinder 30. The quench rate at the Curie temperature equipment. The improved process optimizes the 1 and 0 electrical response ratio of the cores, and allows the threshold coercivity of the cores to be closely controlled. Cores are sintered by the first temperature-time controlled step to provide a minimum 0 response for a given coercivity and may be stock-pilled or immediately subjected to a controlled quench rate annealing step to provide a desired threshold coercivity. The controlled quench rate anneaing step is reversible so that the threshold coercivity may be altered by repeating the second step with control of the quench rate. This allows the cores to be reproduced to closer electrical tolerances and minimizes rejects.

What is claimed as new and desired to be secured by Letters Patent of the United States is:

1. In the manufacture of ferrite bodies utilized as memory storage devices, said bodies consisting essentially of finely divided particles of ferric oxide and oxides of magnesium and manganese, a process which optimizes the electrical characteristics of said ferrite bodies including the steps of, initially sintering said ferrite bodies by heating in a furnace at a temperature within the range of 1300 to 1500 C. for a time interval of 10 to 120 minutes to produce a time-temperature product which minimizes their 0 electrical response characteristics, removing said ferrite bodies from the furnace, and subsequently providing controlled quench rate annealing by re-heating said ferrite bodies to a temperature no greater than said initial sintering temperature for less than 10 minutes, and cansing said ferrite bodies to fall freely through a space having a temperature gradient with a predetermined contour such that said ferrite bodies are cooled to less than Curie temperature in a time interval providing a desired threshold coercivity for said ferrite bodies to thereby optimize their 1 electrical response characteristic.

2. The process set forth in claim 1 wherein said temperature gradient provides a linear temperature contour.

References Cited UNITED STATES PATENTS 3,220,950 11/1965 DiMarco 25262.64

TOBIAS E. LEVOW, Primary Examiner. ROBERT D. EDMONDS, Assistant Examiner.

Images