~68~3~ METHOD FOR MAKING A BISMUTH DOPED NEODYMIUM BARIUM TITANATE This invention relates to a method for making a low firing bismuth doped neodymium barium titanate dielec- tric, and more particularly to such a method wherein a particular small amount of bismuth is included in the start materials to effect a large reduction in the minimum sintering temperature that produces a dense mature dielec- tric ceramic. Numerous manufacturers of ceramic capacitors use a neodymium barium titanate with various additives to make a body having a low temperature coefficient of dielectric constant, e.g. ~ 30 ppm/C. Among the additives employed in the prior art are alkaline-earth metal zirconates, stannates and titanates. Bismuth titanate is also added; the bismuth typically accounted for as the oxide amounting to from 1.8 to 407 mole percent of the oxides of all the cations in the body. Others have added oxides of bismuth lead and silicon to a prefired or calcined neodymium ; barium titanate powder before sintering to obtain a lower sintering temperature. ~0 However3 these low-temperature-coe~ficient formulations including a rather large number and substan- tial amounts of additives makes difficult the control of composition and dielectric properties in manufacturing. ., ~k . ~ - - : 0 -- 2 -- ~ feature of this invention is the provision of a method for making a bismuth doped neodymium barium tita- nate wherein a particular small amount of bismuth dopant effects a large reduction in the minimum sintering tempera- ture of the ceramic. Another feature is the provision ofsuch a method wherein the actual oxides of the cations in said ceramic are used as start materials to more surely obtain reactivity at calcining of the start ingredients. In accordance with this invention a low firing temperature neodymium barium titanate is produced by pre- paring a mixture of powdered precursors of a bismuth doped neodymium barium titanate, calcining to react the precur- sors, pulverizing the calcined precursors, forming a body of khe pulverized precursors, and sintering the body at no more than 1200C to form a dense mature ceramic body. The powdered precursors, i.e., the start materials, are all oxides, or oxides equivalen-ts such as carbonates, hydrox- ides, hydrates and oxylates. The precursor containing bismuth, accounted for as the oxide, amounts to between 0.25 and 1.5 mole percent of the molar sum of all the precursors accounted for as oxides. In drawings which illustrate embodiments of the invention, Figure 1 is a sectional view of a wafer capacitor having a body of this invention, Figure 2 is a graph of the minimum sintering temperature versus the amount of bismuth doping for a group of neodymium barium titanate formulations, Figure 3 is a graph of shrinkage curing sintering, a function of sintering time, and sintering temperature for neodymium titanate bodies with various amounts of bismuth doping, Figure ~ is a graph of sintering temperature versus bismuth doping taken from Figure 3 at 6.5% shrink- age for each composition, Figure 5 is a graph of shrinkage versus bismuthdoping taken from Figure 3 at 3% hours into sintering for each composition, -- 3 Figure 6 is a sectional view of a multilayer capacitor of this invention, and Figure 7 is a graph of the percent theoretical density of neodymium barium titanate bodies, all fired at the same temperature, as a function of the amount of bismuth doping. In general, this invention recognizes that the inclusion of bismuth of a particular small amount in the paraelectric crystals of neodymium barium titanate mate- rial dramatically reduces the sintering temperature ofthat material. This invention also recognizes that little or no reaction occurs at sintering of a mixture of cal- cined neodymium barium titanate and bismuth-containing compounds such as bismuth oxide and bismuth titanate. Thus it is also essential in the method of this invention that the bismuth be included among the start materials for forming neodymium barium titanate to effect the wanted bismuth doping. A number of experimental wafer type capacitors were made and are illustrated in Figure 1. The wafer ` bodies were first formed by the conventional steps of combining start ceramic precursor powders, wet milling for two hours, drying, granulating, and calcining at 1200C for two hours. This was followeed by crushing and jet pulverizing to an average particle size of about 1 to 1.5 microns. The fine powder was combined in an organic vehicle and cast squares were formed about 25 mils thick. m ese sq~lares or wafers had minimum sintering temperatures between 1100 and 1350C. A silver electroding paste was applied to the two major faces of each mature ceramic wafer. The electroded wafers were fired at 800C to form wafer capacitors each having bodies 10 and electrodes 11 and 12, as seen in Figure 1. The minimum sintering temperatures given in 50C intervals for densification of each ceramic are shown in Table 1 along with the corresponding compositions of the ceramic in a series of experimental wafer capacitors. No , -- 4 -- sintering fluxes were used and all the start ingredients were co-reacted at calcining. The start ingredients were specifically neodym- ium carbonate, barium carbonate, titania, bismuth trioxide, lead oxide, and zinc oxide. However, whether these mate- rials are carbonates, hydroxides, hydrates, oxalates or oxides, thPy all effectively become a group of reacted oxides after sintering. The particular forms selected from those above-named start materials and their like is of no consequence to the composition and properties of the sintered ceramic. In Table 1 the ceramic compositions are given for convenience as the molar amounts of the oxide forms of the start materials. TABLE 1 Composition (m%) Ex. TiO2 Nd203/Bi203BaO/P~O Min Sinter T 1 67.5 17.0/0.0 15.5/0.0 1350C 2 67.5 15.5/1.5 15.5/0.0 1250C 3 67.5 14.0/3.0 15.5/0.0 1300C 4 67.5 17.0/0.0 1~.0/1.5 1250C 67.5 17.0/0.0 12.5/3.0 1250C 6 67.5 17.0/0.0 11.0/4.5 1200C 7 67.5 15.5/1.5 12.5/3.0 1200C 8 71.44 14.73/0.0 14.28/0.0 1250C 9 71.44 12.78/1.5 14.28/0.0 1150C 71.44 12.78/1.5 12.78/1.5 1200C BaO/ZnO 11 71.44 14.28/0.0 12.78/1.5 1250C 12 71.44 14.28/0.0 11.28/3.0 1250C 13 68.75 15.75/1.2514.25/0.0 1150C In Examples 1 ~hrough 7, the ceramic is basi- cally (17Nd203) (15.5BaO) (67.5TiO2). The compositions of Examples 1, 2 and 3 have increasing amounts of Bi203 substltuted for the Nd2O3. The intermediate amount of bismuth doping of Example 2 gives a substantially lower minimum sintering temperature at 1250C. The compositions of Examples 4, 5 and 6 have no bismuth~ but lead is added at the expense of barium. These increasing amounts of lead doping reduce the minimum sintering temperature also, although no optimum amount of lead is evident. However, although the reduction in sin- tering temperature due to the addition of lead is small, the addition of lead may be useful because it has the effect of making the temperature coefficient of capaci- tance more positive. The further addition of bismuth as in Example 7 to the lead doped material of Example 5 has less effect on the minimum sintering te~perature to that exhibited in the material of Example 2 where no lead was present. The neodymium barium titanate (with a composi- tion of Nd203-BaO-5TiO2) of Example 8 has a relatively low minimum sintering temperature even without bismuth or lead dopants. Nevertheless t a 1.5 mole percent addition of Bi203 in Example 9 drops this sintering temperature by 100C. The urther addition of lead in Example 10 surprisingly raises the sintering temperature. The addition o zinc ratller than lead at the expense of barium in Examples 11 and 12, has no apparent effect on sintering temperature. All the neodymium barium titanate compositions described above yield a ceramic exhibiting an X-ray dif~raction pattern of Nd203-BaO-4TiO2. This observation lead to an e~periment in which that 1-1-4 ceramic with bismuth oxide substituted for some of the neodymium was made, i.e., 14.4Nd203-16.67BaO-66.67TiO2. This material 30( had the expected low sintering temperature and under scanning electron microscope examination showed essentially none of a high-titanium second phase, unlike all of the other compositions wherein a second phase was prominent. This essentially single phase composition may have special advantages over the other compositions although further work is necessary to be sure. .~ .Z~68~(~ -- 6 -- The foregoing experiments suggest that there is something unusual going on in neodymium barium titanate materials when doped with 1.25 mole% Bi203 that has yet not been made fully explicit by the foregoing examples. A composite graph of minimum sintering tempera- ture as a function of the amount of the Bi203 includQd in the ceramic composition is shown as curve 13 in Figure 2. This graph takes into consideration the minimum sintering temperatures of the three basic neodymium barium titanate compositions represented by the groups of Examples 1-7, 13, and 8-12, respectively. The groups taken in that order have a decreasing minimum sintering temperature with a given kind and amount of dopants. Thus, the phenomena in the graph of Figure 2 for a minimum sintering tempera- ture to occur at 1.25 mole% addition of Bi 2 3 is attributed to all basic neodymium barium titanate compositions of the popular titanium rich class that include at least 25 weight percent each of Nd203 and TiO2. The particular vertical scale used in the graph of Fig. 2 is for the basic material of Examples 1-7. Those experiments led to another experimental series to more closely determine the effect on sintering temperature of tl~e bismuth doping of neodymium barium titanate. Using the process described for making wafer bodies in Examples 1 through 13, the materials of Examples 1~ through 22 were made, for which material compositions in mole% are shown in Table II. 12~6E~ -- 7 TABLE II Ex. Composition (mole%) ___ TiO2 Nd203/Bi203 BaO 14 68.75 17.00/0.0 14.25 68.75 16.25/0.85 14.25 16 68.75 15.75/1.25 14.25 17 68.75 15.25/1.75 14.25 1~ 68.75 14.75/2.25 14.25 19 68.75 14.25/2.75 14.25 68.75 13.75/3.25 14.25 21 68.75 13.25/3.75 14.25 22 68.75 12.75/4.25 14.25 As before, the bismuth is purposefully reacted at the outset with the precursors of the basic neodymium barium titanate. However, sintering was done for each bodyJ one at a time, in a dilatometer wherein the thick- ness of each body was continuously measured while the temperature was raised at 6.7C/minute, held for 30 minutes at 1200C, and then heated at the same rate to 1300C. The sintering profile thus generated is shown for each example in Figure 3 wherein the numeral designation for each curve corresponds to the example number. Curve 23 is the oven temperature. A sintered body from each formulation was exam- Z5 ined by scanning electron microscope. For those with bis- muth oxide doping amo~mts greater than 3.25 mole%, a second phase of bismuth titanate began to appear indicat- ing a very limited solubility and affinity for bismuth by neodymium barium titanate. From the experimental information presented in Figure 3, data points were located in the graphs of Figures 4 and 5. In the graph of Figure 4 a curve 25 was fitted to the data points plotting the sintering temperature at which 6.5% shrinkage occurred is shown as a function of Biz03 dopant content. In the graph of Figure 5, a curve 27 was fitted to the data points plotting the percent shrink- age at 3~ hours into sintering. These curves dramatically :' '' : , . . ~Z9681~ " . -- 8 -- point out the extraordinary advantage in this ceramic sys- tem of doping with about 1.25 mole percent Bi203. Furthermore, the trough in the curve 25, centered at about 1.25 mole% ~i203, is low and quite narrow. In fact, the best performance is obtained for dopant amounts less than 1.5 mole% Bi203 and preferably less than 1.4 mole%. A 1.25 mole% Bi203 dopant amount corresponds to sintering temperatures lower than that which is obtain- able using any other Bi203 dopant amounts, as seen by the curve in Figure 4. ` Analysis of some of the sintered bodies of Exam- ples 1 through 12 using the electron microscope and micro- probe, it was observed that the bodies exhibited small patches of a high titanate second phase in a dominant ma- trix of neodymium barium titanate e~cept that essentially no second phase was apparent in bodies of Example 13. The number (N) of cations in the basic neodymium barium titanate are very nearly related by 3 (NNd ~ Nsa) 2 NTi whereby substituting for some of the Nd cations a large Ba cation for each removed Nd cation in the crystal would still give a single phase crystal composition. Since the use of single phase compositions (avoidance of second phases) leads to a simpler and a more easily understood system, it also leads to better control in manufacturing and is preferred. To hold the molar amount o~ any one of these three major cations within 5 or 6 percent of that indicated by the above single phase relationship would seem close enough to obtain the above-noted advantage of control. In order to use the above-noted optimally bismuth- doped neodymium barium titanate compositions to make mono- lithic ceramic capacitors with low ~ost silver palladium ;~ electrodes, it is desirable to add a flux to the green ceramic slurry for reducing the sintering temperature a little further. Conventionally the alloy 70Ag/30Pd (by ; weight) is so employed, which has a melting temperature of 1160C, for which a capacitor sintering temperature should be no greater than 1150C to avoid melting the ~ , ,, ~, ~: . ~Z~368~ . g buried electrodes. The higher silver content electrodes melt at an even lower temperature, but are of course less expensive. Each of several experimental monolithic capaci- tors were made by the following conventional steps. Cal- cined and pulverized powder of a high firing ceramic (name- ly, 68.75 TiO2-1S.75 Nd2O 31. 25 Bi2O 3 14.25 BaO) was mixed with a low melting flux and, in some cases, another addi- tive in a vehicle and binder medium of turpentine, 6% pine oil, and 5% lecithin. This slurry containing about 70% by weight solids was milled for about six hours. Successive coatings of the milled slurry were applied over a glass substrate, drying each layer in turn and screen printing an electroding paste of 70% silver and 30% palladium to the cast dried layer. ~efore apply- ing the next successive dielectric layer, the screened pattern of electroding paste was dried. The body 30, as shown in Fig. 6 with buried electrodes 31 and 32~ was cut from the stack and was fired to ~laturity at 1100C for 2~ hours. ~ silver paste was then applied to either end of body 30 at which edges of the buried electrodes were exposed. The body was fired at 750C Eor a few minutes to :Eorm silver terminations 35 and 36. I'he composition of the ceramic bodies of these experimental monolithic capacitors, Examples 23 through 29, are provided in Table III along with TCC and DF Data. ~96~ - 10 - TABLE III Composition (wt~/o~ Nd Ba TCC Df ~x. Ti Bi Flux and Addivives (ppm) (v/o)- 23 97 3 CdO-2ZnO-B203 N65 0.01 24 95 3 CdO 2ZnO-B203 N50 0.012 2 Bi2Ti207 93 3 CdO-2ZnO-B203 P7 0.007 4 PbTiO3 26 92 3 CdO 2ZnO B203 N4 0.015 3 PbTiO3 2 BaTiO3 27 92 3 CdO-2ZnO B203 N27 0.009 BaTiO3 28 98.5 1.5- BaO-2B203 N47 0.003 29 97.75 2.25 BaO B203 N54 0.007 68 29 SrTiO3 3 CdO-2ZnO-B203 N750 0.01 All of these capacitor bodies are essentially nonporous and dense, having been sintered at 1100C. The dielectric constant of all of the bodies was from 75 to 90. The dissipation factor (DF) was measured at 1 KHz. Small quantities of lead and/or barium titanate are seen to be effective means for adjustlng the temperature coefficient of capacitance (TCC). The high-firing additives, lead titanate, barium titanate, bismuth titanate, and strontium titanate seem to experience little, i any, reactivity with the calcined neodymium titanate at sintering. Other additives including alkaline-earth-metal zirconates and stannates have been used with similar effect. As has been noted, they are useful for adjusting the temperature coefficient of the body, but otherwise have relatively little effect on other properties. For~example, the dielectric constants of the ceramic bodies of Examples 23 to 29 range from 60 to 80 and that of E~am- ple 30 is about 90. The effect of these additives on mini- mum sintering temperature is essentially nil. ~Z~6i~0 - 11 - In yet another experiment (Ex 23A) a material was made similar to that described as Example 23, except no bismuth was used in forming the calcined neodymium barium titanate. Instead, 4.25 weight percent Bi203 was combined with 3 weight percent of the cadmium zinc borate flux so that the constituents of this experimental group (Ex 23A) are the same (e.g., the bismuth oxide amounts to 1.25 mole% of the neodymium barium titanate) as those in the Ex 23 ceramic bodies. Green bodies of both Examples 23 and 23A were sintered for 2~ hours at 1050C. The bodies of Ex 23A were porous, not fully densified and had poor electrical properties (e.g., DF=0.14%) whereas those of Example 23 were dense and otherwise also indistinguish- able :Erom those of this same group that were sintered at 1100C. The barium borate fluxes of Examples 28 and 29 were found to be most effective in reducing the sintering temperature and providing a low DF. Other fluxes that were tried are magnesium borate and cadmium silicate. The silicates proved least effective in reducing sintering temperature, and borate fluxes are preferred. In still another experiment, a calcined and pow- dered bismuth doped barium neodymium titanate, namely: 17 Nd2(CO3)-13.75 BaCO3-1.25 Bi203-68TiO2, was mixed with 4 weight percent PbTiO3 and with 3 weight percent of a powdered flux or sintering aid CdO 2ZnO B203. A slurry of this mixture was used to form the dielectric portion of a group of monolithic ceramic capacitors. The buried electrodes are composed of 70Ag/30Pd alloy. After the organic binder was removed at about 700C the ceramic capa- citors were placed in an alumina crucible with a tight cover and sintered at 1100C for 2~ hours. The capacitance of each capacitor is 1000 pico- farads. At 1 K~z the DF is less than 0.01% and at 10 MHz the DF is about 0.05%. The dielectric constant is 80+2. Insulation resistance is greater than 106megohms and the temperature coefficient of capacitance is 0~10 ppm/C from -55C -to +125C. There has been no degradation of these :,. . :~z~ - 12 - properties after 200 hours of stress at 160 volts per mil of dielectric thickness at 150C. It should be noted that this neodymium barium titanate meets the above-noted cri- teria that three times the sum of the neodymium and barium atoms is within 6% of twice the number of titanium atoms, and the titanate component is essentially a single phase material. The foregoing experiments have compared the pro- perties of different ceramic materials with the introduc~ tion of various amounts of bismuth oxide to the start ingredients wherein the bismuth oxide was subs~ituted mole for mole for neodymium oxide. It has been discovered that by introducing the same quantities of bismuth oxide at the expense of the titania that those same bismuth additions are even more effective in reducing the minimum sintering temperature. Thus, when substituting bismuth for titanium, stoichiometry is apparently more nearly preserved and sintering more easily achieved. For the above-noted 1:1:4 phase the composition may be expressed as (l-x)Nd203-xBi203 BaO-4TiO2. Using the same process described herein for mak- ing wafer bodies in Examples 1 through 13, except for intro- ducing the bismuth at the expense of titanium, five groups of ceramic wafers corresponding to Examples 31 through 35 were made having ceramic compositions given in Table IV. TABLE I~ Composition (mole%) Ex. BaO Nd203 TiO2./Bi203 31 1~.25 17.00 68.75/0.00 32 14.25 17.00 68.00/0.75 33 1~.25 17.00 67.50/1.25 34 14.25 17.00 67.00/1.75 14.25 17.00 66.50/2.25 6~38~ - 13 - Wafers from each group were sintered in separatecrucibles at 1250C at which all wafers were sure to den- sify. Also similarly sintered were representative wafers from experimental groups 14 through 18 wherein the bismuth additives replaced neodymium. The degree of shrinkage was determined for each of the sintered wafers and the percent of theoretical density for each was calculated based on a determination that 100% theoretical density is 5.67 grams per cubic centimeter for these materials. In Figure 7, there is plotted a curve 29 repre- senting the percent ~heoretical density of wafers of Examples 14 through 18 (Table II). A similar plot for wafers of Examples 31 through 35 (Table IV) is given by curve 30 in Figure 7. It is clear from these plots that the substitution of bismuth for titanium (curve 30) is more effective, and that even lesser dopant amounts of bismuth than that indicated by the neodymium displaced examples represented by curve 29 will result in the sub- stantial reduction in minimum sintering temperature. Indeed, it is apparent that at a theoretical density of about 85%, 0.5 percent Bi203 is required to displace neodymium but about 0.25 mole percent Bi203 can displace titanium to produce that same density at which sintering i5 consiclered adequate. Thus, bismuth oxide amounts as low as 0.25 mole% are capable of providing the large reductions in sintering temperature that are character- istic of this invention. :,,., ~ , ,