IE47548B1 - Cubic boron nitride compact - Google Patents

Abstract

A process for making a sintered polycrystalline cubic boron nitride (CBN) compact by direct conversion of a pyrolytic, preferably substrate nucleated, boron nitride (PBN) comprises subjecting the PBN to temperatures and pressures within the CBN stable region. The crystallites of the compact produced by this process are preferentially oriented. Also, such compacts have a high thermal conductivity between about 2 watts/cm DEG K and 9 watts/cm DEG K and thus are useful as a heat sink for electronic devices. The size of the CBN crystallites of the compact may be between about 1 x 10<3> ANGSTROM to 1 x 10<5> ANGSTROM .

Description

Tnis invention relates to processes for making polycrystalline abrasive compacts and, more particularly, it relates to a direct conversion process for .iiaking a cubic boron nitride (CBN) compact from hexagonal boron nitride (HBN) and the resulting product.

Three crystal!-ine forms of boron nitride are known: 1) soft graphitic (hexagonal) form (HBN) similar in structure to graphitic carbon, 2) a hard wurtzite (hexagonal) form (WBN) similar to hexagonal diamond, and 3) a hard zincblende (cubic) form (CBN) similar to cubic diamond. lhe three BN crystal structures may be visualized as formed by the stacking of a series of sheets (layers) of atoms. In the low pressure graphitic structure the stacking layers are trade up of planar fused hexagonal (like bathroom tile) in which the vertices of the hexagons are occupied alternately by boron and nitrogen atoms and stacked vertically such that the B and N atoms also alternate in the stacking [001] direction as illustrated in FIG. IA. In the more dense CBN and WBN crystal structures the atoms of the stacking layers are puckered out-of-plane and the two dense structures - 2 47548 result fran variation in the stacking of the layers. As illustrated in FIGS. IB and IC, the layer stacking sequence of the CBN and WBN structures can therefore be symbolized as •. .A 5 C A... and . ·Α B A B.. · respectively.

In HBN and WBN crystals the layers are stacked along the /5()17 direction (i.e. the c crystallographic axis is perpendicular to the layers) whereas in the CBN crystal the layers are stacked along the ^117 direction. These layers are referred to as hexagonal stacking layers or planes. In HBN, bonding between the atoms within the layers is predominantly of the strong covalent type, but with only weak van der Waals bonding between layers. In WBN and CBN, strong, predominantly covalent tetrahedral bonds are formed between each atom and Its four neighbors.

Hard phase BN compacts are of two general types: a cluster ccsipact and a conposite canpact.

A cluster compact is defined as a cluster of abrasive crystals bonded together either (1) in a self-bonded relationship, (2) by means of a bonding medium disposed between the crystals or (3) by means of sane combination of Xl) and (2). Reference can be made to U. S. 3,136,615 and U. S. 3,233,988 for a detailed disclosure of certain types cluster compacts and methods for making same.

A conposite conpact is defined as a cluster compact bonded to a substrate material, such as cemented tungsten carbide. The bond to the substrate can be formed either during or subsequent to the formation of the cluster compact.

Reference can be made to U. S. 3,7^3,489 and U. S. 3,767,371 for a detailed disclosure of certain types of conposite compacts and methods for making same. — 3 — ( / Known process for making CBN conpacts can be generally classified In four categories and are so defined as used herein. (1) catalytic conversion process, a one-step process in which a catalyst metal or alloy aids in the transition of HBN to CBN simultaneously with the formation of the conpact; (2) bonding medium process, a two-step process In which the first step comprises the conversion of HBN to CBN and the second step ccnprises the formation of a conpact from cleaned CBN crystals mixed with a metal or alloy which aids In the bonding of the CBN into a compact; (3) direct sintering process, a twostep process which is the same as process (2) except that conpact is formed without addition of metal or alloy to aid in bonding CEN crystals; and (4) direct conversion process, a one-step process In which substantially pure HBN is directly transformed to a CBN compact without the aid of a catalyst and/or bonding medium.

The catalytic and bonding medium processes are generally disadvantageous because the catalysts and bonding medium are lower In hardness than CBN and are retained ln the resultant mass which reduces the hardness and abrasive resistance of the masses. Particular reference can be made to U. S. 3j233,988, col. 4, line 3, through col. 6, line 41 and to U. S. 3j918,219 for a more detailed discussion of catalytically formed CBN compacts and to ϋ. S, U. S. 3s767j371 for the details of CBN compacts utilizing a bonding medium.

The direct conversion process, while theoretically possible, has been found, in practice, to have high losses because it is difficult to consistently achieve a sufficient number of crystal to crystal bonds distributed uniformly throughout the compact. Without such, the strength and density of the canpact are less than ideal. - 4 47548 Hie direct conversion under static pressure conditions of HBN to the more dense wurtzitic or cubic (zineblende) phases at pressures of 100 tears and ahove is described in detail in J. Chem, Phys., 38, pp. 1144-49, 1963, Bundy, et al., and In U. S. 3,212,852. A disadvantage of this method is that in the pressure range above 100 tear the effective reaction volume is limited which limits the size of the converted polycrystalline compacts products.

More recently, numerous reports and patents have been published concerning the direct conversion of HBN to CBN cluster compacts at pressure below 100 kbar. Representative of these publications are: 1. Wakatsukl et al., Japanese Patent No. Sho 49-27518. 2. Wakatsukl et al., Japanese Patent No. Sho 49-30357. 3. VfetatsukL et al., Japanese Patent No. Sbo 49-22925. 4. Wakatsukl et al., U. S. Patent No. 3,852,078.

. Ifekatsuki et al., Synthesis of Polycrystalline Cubic Boron MLtid.de, Ifet. Res. Bull., 7, 999-1004 (1972). 6. Iehinose et al., Synthesis of Polycrystalline Cubic BN (V), Proceedings of the Fourth International Conference on High Pressure, Kyoto, Japan (1974), pp. 436-440. 7. Wakatsukl et al., Synthesis of Polycrystalline Cubic Boron Nitride (VI), Proceedings of the Fourth International Conference on High Pressure, Kyoto, Jaoan (1974), pp. 441-445.

No. 8. Sirota, N. Patent Specification^36384 Process for the Production of Cubic Boron Nitride, Ifey 23, 1973.

Publications Nos. 1 through 7 report direct conversion occurs' at pressures>50 tear (preferably 60 kbar and above) and temperatures^1100°C, while Publication No. 8 reports conversion at pressures of 60 tear and higher over the tenperature range fran l800°C to 3000°C.

The publications generally used HBN powder as the starting material. Two publications (Pubs. 6 and 7) reported the use of pyrolytic boron nitride (PBN) as the starting material. Reference can be made to U. S. 3,152,006 and U. S. 3,578,403, respectively, (which are hereby incorporated herein by reference) for a mare detailed description of PBN and R-PBN and acceptable processes for making it.

Publication No. 6 reports the use of PBN as a starting material for the synthesis of CBN cluster connects in a direct conversion process practiced at a pressure of 69 kbars and a tenperature of between l800°C and 1900°C. The resultant product (TABLE 1, p. 436) was characterized as a soft mass having varying amounts of unconverted HBN.

Publication No. 7 also reports the use of PBN as a starting material for the synthesis of wurtzitic boron nitride (WBN) and CBN. There were no reported results of the successful formation of either WBN or CBN using PBN as a starting material. See TABLE 1, p. 442.

PBN is a low pressure form of HBN made typically by thermal composition of BClg + NH3 vapors on a graphite substrate. As deposited, it has a high purity of 99.99 + %, a density between 1.8 to 2.28 and normally 2.0 to 2.18 g/cm3 (compared to 2.28 for crystalline HBN), a crystallite size oetween 50 and 100 A and a preferred crystallite orientation between 50° and 100° in the /0017 direction (c-axis). The structure of PBN, as with analogous pyrolytic carbon in the carbon system, is not well understood. Various models have been proposed to explain the structure of PBN and pyrolytic carbons. According to one of the more popular models, termed turbostatic state, the B and N atoms form more on less parallel stacks of fused hexagon graphite BN like layers, but with stacking being random in translation parallel to the layers and random in rotation about the normal to the layers. Other models emphasize’inperfections and distortion within the layers. The increased interlayer spacing in the pyrolytic materials - 6 47548 (3.^2 fi for PBN compared to 3.33 X-for crystalline HBN). is attributed to the disorder in the stacking direction resulting in attenuation of the weak van der Waals attraction between the layers.

Although highly disordered, PBN is not conpletely devoid of crystallo5 graphic order (not amorphous). There is, though inperfect, organization of the B and N atoms into graphite-like layers: it is the ordered stacking arrangement? of the layers which is most conspicuously absent. Extensive structural transformation is required to convert pyrolytic BN to the HBN structure shown in PIG. 1.

The as deposited type of PBN will be referred to hereinafter as unrecrystallized PBN (U-PBN).

Another known type of PBN is recrystallized PBN (R-PBN). It is formed by compression annealing of PBN and has a theoretical density of 2.28 g/cm3, a highly crystalline structure with an interlayer spacing of 3.33 X, a purity of 99.99+%» and a preferred crystallite orientation of about 2° in the /0017 direction (c-axis).

Each type of PBN is made and ccomercially available in the form of a solid continuous sheet having the hexagonal stacking planes of each crystallite aligned with rajor planes of the sheet to the degree of preferential orientation Thus the hexagonal stacking planes (001) of U-PBN are disposed at angles varying between about 50° and 100° with major planes of the sheet, while the (001) planes of the R-PBN are disposed at angles varying between about 2° or less with the major planes of a sheet.

R-PBN is further described in U. S. 3,578,403 which is hereby incorporated by reference herein. - 7 47548 PBN can be also classified as either ’’substrate nucleated or continuously renucleated. Substrate nucleated PBN is characterized as material substantially free of co-deposited gas-phase formed particles which act as new nucleation sites. Continuously renucleated material is characterized by the presence of co-deposited gas-phase formed particles which result in continuous renucleation during the deposition process. The concentration of co-deposited gas-phase formed particles and thus the degree of renucleation is reflected in the size of the growth cones developed during the deposition process. Large growth cones are characteristic of substrate nucleated material Snd is thus associated with a low degree of renucleation, and vice versa. The growth cone structure can be observed under low power magnification. The terms substrate nucleated and continuously renucleated PBN define more or less end point types of microstructure. A range of the microstructures exist between the continuously renucleated microstructure containing a high concentration of co-deposited gas-phase formed particles to the substrate nucleated structure free of co-deposited particles.

Also, the aforementioned U. S. 3,212,852, col. 10, lines 19 - 24, discloses tha use of PBN as starting material in direct conversion processes practiced at ‘ pressures above 100 kbars.

It has been found through experiment that the cluster compacts produced in accordance with the teaching of the foregoing prior art publications still fail to achieve desired performance levels in tests designed to measure the effectiveness of such compacts for cutting tool inserts.

Additionally, the trend to miniaturization in electronics has led to the need for inproved heat dissipating substrates for solid state devices. For exanple, in nearly all microwave devices, heat generated during operation leads to decreased efficiency; and dissipation of the heat generated is the critical - 8 47548 factor limiting operation. A canmonly used heat sink material oxygen free high thermal conductivity copper has a thermal conductivity of approximately 4 W/cm°C at room tenperature. For applications where excellent dielectric properties are required, sintered beryllium oxide is canmonly used even though its thermal conductivity is only about one-half that of copper. A combination of high thermal conductivity and good dielectric properties is highly desirable ln a new substrate material.

Type Ila single crystal diamond has the highest room tenperature thermal conductivity of any known material and is currently being used to a limited extent for some microwave devices. Known applications for Improved heat dissipating diamond substrates range from heat sinks for solid state microwave generators, such as Gunn and ΙΜΡΑΤΓ diodes, through solid state lasers, high power transistors and Integrated circuits. It is not extensively used because of cost and difficulty ln shaping.

A high thermal conductivity material, less costly than Type Ila single crystal diamond would be highly desirable if it also had good dielectric properties and could be molded in larger pieces than Type Ila diamond.

In addition to diamond, CEN has been suggested as a possible dielectric heat sink material. In Slack, J. Phys, Chem. Solids 34, 321 (1972), pure single crystal CBN was predicted to have a room tenperature thermal conductivity of ^13 W/cm°C. Until recently maximum values of only —2W/cm°C have been reported for sintered CBN ccnpacts. In Japanese Patent No. 50-61413, however, cnemial conductivity values as high as 6.3 W/cm°C for sintered compacts •or isotope enriched sintered CBN compacts have been reported compared with 1.7 W/cm°C for sintered compacts of naturally occurring isotope concentration.

Accordingly, an object of this invention Is to produce strong, abrasive resistant CBN cluster eonpacts with improved performance characteristics. - 9 • 4 7 54 8 Another object of this invention is to produce large sized CBN cluster compacts by the direct conversion of HBN under high pressure and high tanperature (HP/HT) conditions.

Another object of this invention is to produce CBN cluster compacts in 5 sufficiently large sizes to be useful in material removal applications and at HP/HT conditions which are more economical.

Another object of this invention is to prepare CBN cluster compacts having a room tenperature (300°K) thermal conductivity (k) greater than 2 watts/cm°K and preferably greater than 6 watts/cm°K.

Another object of this invention is to prepare CBN cluster compacts of high thermal conductivity by direct conversion of HBN to CBN in which the crystallite size, is larger than the room tenperature phonon mean free path length and the thermal resistance between grains (crystallites) is not increased by grain boundary oxide contamination.

Another object of this invention is to economically produce CBN cluster conpacts having thermal conductivity values suitable for electronic heat sink application.

Another object of this invention is to produce high thermally conductive cluster conpacts of high electrical resistance, low relative permittivity and low dielectric loss tangent.

Another object of this invention is to produce high thermal conducting polycrystalline CBN conpacts free of secondary binder or sintering aid phases.

Another object is to produce polycrystalline CBN conpacts free of Impurities (particularly oxygen and nitrogen impurities) which would act as phonon scattering centers and thus limit the thermal conductivity/ - 10 47848 According to the present invention we provide a high temperature and pressure process for making a sintered polycrystalline cubic boron nitride compact, said process comprising the steps of: (a) placing preferentially oriented pyrolytic hexagonal boron nitride in the form of a bevel-edged disc in a reaction cell, said boron nitride being substantially free of catalytically active materials, said cell having means for shielding said pyrolytic boron nitride from contamination during transformation; (b) compressing said cell and the contents thereof at a pressure between 50 kbars and 100 kbars; (c) heating said cell and the contents thereof to a temperature of at least 1780°C. within the cubic boron nitride stable region of the boron nitride phase diagram; (d) maintaining said pressure and temperature conditions of steps (b) and (c) for a period of time sufficient to transform said pyrolytic boron nitride into a sintered polycrystalline cubic boron nitride compact; (e) ceasing the heating of said cell; and (f) removing the pressure applied to said cell.

The pyrolytic boron nitride preferably has a density between 1.8 and 2.28 g/cm3.

The compact may consist essentially of cubic boron nitride crystallites having (111) planes preferentially oriented and a thermal conductivity of at least 2 watts/cm°C (preferably at least 6.3 watts/cm°C. Thermal conductivity is measured at room temperature (about 25°C) except where otherwise stated. The preferential orientation is desirably between 50° and 100° or between 2° and 0°. The company may have a Knoop hardness between 4000 and 8000 kg/mm and may have a density of at least 3.40 g/cm . 4-7548 Figures IA, IB, and IC are schematic illustrations of the atomic crystal structure of HBN, CBN and HBN, respectively.

Figure 2 is a fragmentary axial cross-sectional view of a preferred embodiment of a HP/HT reaction cell used in the practice of this invention.

Figure 3 is the Bundy-Hentorf boron nitride phase diagram.

Figure 4 is a boron nitride phase diagram with data points illustrating the preferred operating region of this invention.

Figure 5 is a perspective view of a cutting tool insert incorporating a cluster compact made in accordance with the invention.

Figure 6 is a graph of crystallite size versus processing temperature of a cluster compact in accordance with this invention.

Figure 7 is a graph of the thermal conductivity versus the material temperature of various materials including compacts in accordance with this invention.

Figure 8 is a graph of the thermal conductivity versus processing temperature of a cluster compact in accordance with this invention.

Figure 9 is a schematic illustration of an electronic device with a CBN compact heat sink in accordance with the features of this invention.

Figure 2 is a cross-sectional view of cylindrical reaction cell 201 suitably for use with a conventional belt type HP/HT apparatus used in the formation of a CBN cluster compact according to the features of this invention. One preferred embodiment of the type of belt apparatus is fully disclosed in U.S. Patent No. 2,941, 248 which is hereby incorporated by reference herein. 47S48 Figure 2 shows a second preferred embodiment of a reaction cell 201 for the practice of the invention. Reaction cell 201 includes a lava cylindrical bushing (not shown) as in Figure 2. Positioned concentrically within and adjacent to the bushing is a contamination shield tube 205 of tantalum. Within tube 205 there is in turn concentrically positioned a graphite resistance heater tube 207.

A bevelled edged disc 209 of PBN starting material is centrally disposed within heater tube 207. A pair of spacer discs 211, 213 of carbon are disposed one on each side of disc 209. A pair of contamination shield discs 215, 217 of tantalum are disposed outwardly of the spacer discs 211, 213.

Outwardly of shield discs 215, 217 are, in order, a pair of electrically insulating hot pressed BN discs 219, 221; a pair of carbon discs 223, 225; and a second pair of hot pressed .BN discs 227, 229. This cell construction has been found to be superior to that of Figure 2 in that lamination of the Ta tube 205 to the compact formed from starting material disc 209 is avoided by the disposition of heater tube 207 intermediate disc 209 and tube 205 and fewer cracks are found in the compacts produced in such cells.

Other metals which do not interfere with the conversion/sintering process and prevent impurity penetration into the PBN could also be used as shielding material. Other metallic shields may include, but not limited to, Group 4 metals such as titanium and vanadium, Group 5 metals such as zirconium, molybdenum and niobium, and Group 5 metals such as hafnium and tungsten.

In accordance with another feature of this invention, it has been found to be advantageous, in order to additionally reduce cracks in the cluster compacts produced in such cells, to bevel the circumferential edge of the starting material disc 209. This is believed to relieve edge stresses incurred during decompression of the . 47548 cell, thereby yielding a further reduction in the incidence of compact cracking.

The bevel edge shape of the disc 209 is retained in the converted cluster compact.

In forming a cutting tool from such a cluster compact the bevel edge may be ground av/ay if desired.

Operational techniques of simultaneously applying both high pressures and high tenperatures in the apparatus described hereinabove are well known to those skilled in the superpressure art. The foregoing description relates to merely one HP/HT apparatus. Various other apparatuses are capable of providing the required pressures and tenperatures that may be enployed within the scope of this Invention.

FIG, 3 shows the boron nitride phase diagram as published by Bundy and , Wentorf (J. Chem. Phys., 38, 1144-1149 (1963)). In this diagram, AB is the phase equilibrium boundary for CBN and HBN. At pressures above EB in the region EEC spontaneous conversion of HBN to either WBN or CBN was found to occur. At the lower tenperatures, to the left of the hashed marked area EB, in Region EBE, the predominant mode of conversion was to WBN. At the higher tenperatures to the right of EB, in Region EEC, the predominant mode of conversion was to CBN.

In practicing the process of this invention a reaction cell containing a PBN sanple Is placed in a HP/HT apparatus, conpressed and then heated under pressure at values of tenperature and pressure below the Bundy and Wentorf direct conversion region (I.e., below line EB) of the phase diagram (FIG. 3).

The HP/HT conditions are maintained for a length of time sufficient for transformation of the PBN to a strongly sintered CBN cluster conpact. The sanple is then allowed to sufficiently cool under pressure to prevent reconversion prior to relieving the pressure.

Rectangle M shown in both PIG. 3 and PIG. 4 (discussed below) shows the general relationship of the preferred operating region shown in PIG. 4 relative to the conplete Bundy-Wentorf phase diagram (FIG. 3 ).

FIG.4 shows the results of a series of direct conversion and reconversion experiments in the lower pressure region. Ihe direct conversion experiments were run on PEN samples In FIG. 2 type cells for 10 minute heating times. The reconversion experiments were run on CBN cluster ccnpacts previously prepared by direct conversion of PBN to CBN also in a FIG. 2 type cell for 10 minute heating times. In FIG. 4 conversion of PBN to CBN was obtained within the region JHI and reconversion of CBN to HBN was obtained at tenperatures above the line GHI. Also shown in FIG. 4 ia a section, KL, of the CBN/HBN equilibrium line AB from the Bundy-Wentorf diagram. Ihe present results, indicate that the CBN stability region extends beyond the Bundy-Wentorf equilibrium line KL. Although partial conversion of PEN to CBN was obtained at tenperatures of 185O°C to 1900°C, in practice tenperatures above about 2000°C were found necessary to obtain CBN compacts suitable for machining applications.

In selecting a PBN starting material for practice of this invention, it has been discovered that substrate nucleated PBN should preferably be used in order for the transformation to proceed and for large, strong well-sintered masses to be more consistently produced. If continuously renucleated PBN Is used as the starting material, conversion is inhibited.

As used herein, the term substrate nucleated PBN means material in which the concentration of co-deposited gas-phase formed particles (characteristic of continuously renucleated material) is sufficiently low so as not to interfere with conversion of the PBN to a strong tightly bonded, CBN cluster compact. 47S48 The following types of behavior have been observed .with PEN disc shaped sarroles: A. The PEN does not convert at all.

B. Essentially complete conversion to a strongly bonded (well sintered) cluster conpact occurs, but only over a more limited tenperature range at tenperatures above about l800°C to 2000°C.

C. Conversion with strong bonding occurs in layers paralleling the PBN disc top and bottom surfaces with the remainder of the sanple remaining unconverted. As in B, this layer type conversion only occurred at tenperatures above l800°C to 2000°C and the converted layers were usually located at either the top or bottom surfaces of the PEN disc.

D. PBN essentially - completely converted over a more broad tenperature range to a poorly sintered-cluster conpact.

Tne difference in appearance and properties between the two completely converted types of conpacts, B and D above, are quite distinct. Type B conpacts are black in appearance and transmit red light while D are opaque grey to milky white in appearance, very similar in color to conpacts formed by the direct conversion of HBN powder. Type B conpacts are more dense and significantly harder than type D conpacts.

It is theorized that'variation in conversion behavior is explained by twa factors: 1. Variation in the microstructure of PBN. 2. Contamination of the initially pure PBN during conversion under HP/HT conditions apparently by diffusion of active species into the PBN from surrounding cell parts at the high tenperature.

In a reaction cell designed to prevent contamination of the PBN by surrounding the PBN with a diffusion barrier only type A, B and C results were obtained. It Is believed that the purity of the PBN starting material (99.99+%) Is maintained during transformation and thus a very high purity cluster conpact 99-99+% is believed to be produced by the practice of this invention.

In cells where contamination was possible, all four types of conversion were obtained with type D conversion predominant. These results indicate that type D conversion may be contamina10 tion induced. It was also determined that the variation in conversion behavior in clean environments (type A, B and C above) may be related to the microstructure of PBN. In particular, the conversion behavior ls believed to be correlated to the size of the growth cones observed in PBN, which converted to type B, having noticeably larger (under microscopic observation) growth cone structure than that which did not convert.

The type C layer conversion results were obtained with PBN disc in which variation of the microstructure through the disc was observed with the converting layer having larger growth cones than the unconverting layers. If the material used to obtain type C layer conversion is subject to higher tenpera20 tures and pressures type B conversion is obtained. However, the use of such material is not preferred because the use of the higher temperature and pressures significantly increases the cost and difficulty of production of good cluster compacts.

It was further determined that the introduction of various oxides (AI2O3, and BgO^) into otherwise uncontaminated experiments resulted in conversion at temperatures down to the 1500°C to 1700°C range, irrespective of the microstructure of the PBN sanple3. However, the compacts formed in this 47S48 manner are less dense and of considerably less strength than eonpacts prepared from substrate nucleated PEN under clean conditions and are similar to the type D eonpacts obtained in contaminated environments. These results appear to indicate that oxides are the source of contamination and are active ln promoting conversion to CBN, but deterimental to sintering.

The eonpacts prepared by oxide addition (or in contaminated environments) are similar in appearance and strength to eonpacts prepared by HP/HT conversion of HBN powder. In experiments with continuously renucleated PBN with oxide additions run at tenperatures less than that required for conversion to CBN, the continuous turbostratio structure of the PBN plate was found to be recrystallized to HBN powder having tha Ideal hexagonal structure. These results suggest that in oxide enhanced conversion of PBN, the PBN is first recrystallized to powder form with subsequent conversion of the recrystallized (and contaminated) powder to CBN. This type of mechanism would account for the similarity of eonpacts obtained by oxide induced conversion of PBN and conversion of HBN powder.

The najor disadvantage in forming large polycrystalline masses with powdered starting material appears to be that surface contamination of the Individual particles inhibit the sintering (bonding) between-.the particles and decrease the strength of the resulting conpact. The presence of oxide contamination in the reaction cell, including B2O3 and moisture, is particularly detrimental to the sintering process.

It is known (e.g., Pub. No. 6) that moisture (H20) has a catalytic effect on the conversion of HBN powder to CBN, but with detrimental effects on sintering. In accordance with this invention the same effects were observed with Various oxides (AlgOg, MgO, BgOg) on the PBN conversion.

In accordance with another feature of this invention, it has been discovered that the structural relationship between the preferentially oriented FBN starting material and the CBN cluster compact is maintained during transition and thus the CBN cluster conpact produced is also preferentially oriented.

As discussed in the Background of the Invention, the R-PBN and U-PBN plate material show a preferred orientation of the crystallite c-axes relative to an axis drawn perpendicular to the major planes of the plate sample. R-PBN has a preferred orientation of about 2° or less and U-PBN has preferred crystallite orientation between about 50° to 100°.

In the direct transition of R-PBN to CBN, the epitaxial relationship between, the original and converted forms Is a parallelism of the hexagonal stacking layers, i.e., the R-PBN (001) plane Is substantially parallel to the CBN (111) plane. For U-PBN the orientation of the hexagonal stacking planes is also thought·, to be the same after conversion to CBN.

For both U-PBN and R-PBN, the activation energies are thought to be about 200 kcal./gm. (which corresponds to the energy of vaporization). These high activation energies imply that the direct conversion process essentially requires disrwtion of the PBN lattices before the atoms can reform into CBN.

The epitaxial relationships observed for both types of PBN indicate that the passing disnption and reforming into CBN proceeds in a regular fashion without passing through an intermediate disordered phase. X-ray diffraction scans of the converted CK? surfaces, normal to the pressing direction for U-PBN plate, show well-developed CBN (111) and (220) reflections indicating a relatively wide variation in the orientation of the hexagonal stacking planes (or c-axes) is maintained and corresponds to the relatively wide variation in the orientation of the original U-PBN plate. This implies that the hexagonal stacking planes of the CBN are not affected by the direction of the applied pressure (which is applied perpendicular to the hexagonal stacking planes) because they do not „ 4-7 548 become aligned normal to the direction of the applied pressure. With the much narrower c-axes angular distribution in the R-PEN only very weak CEN (200) diffraction is seen. These results indicate that while ndcro-recrystallizaticn of the U-PBN nay occur prior to transfoznation to CBN at high pressure, the U-PBN does not first recrystallize before conversion to the CBN into a highly oriented structure analogous to the structure of R-PBN.

The invention will be better understood by reference to the following 10 Exanples: EXAMPLES 1-16 In Examples 1-14 a plurality of cluster compacts were prepared from bevel-edged disc shaped sanples of U-PBN in a reaction cell such as shown in Figure 2. All examples, except Example 16 which was run at 45 to 50 kbars, were run at about 70 kbar at the temperatures and time tabulated in Table 1. The times listed in Table 1 are times at the maximum temperature, i.e., each value given is the total heat treatment time less the time required to reach the maximum temperature.

Examples 15 and 16 represent a prior art directly converted HBN powder compact and a prior art CBN composite compact, respectively, and are included in Table 1 for purposes of comparison.

In Example 15, directly converted HBN powder compact was prepared by direct conversion of a 1.4 g sample of HBN powder (Carborundum Company Grade HPF). The sample was placed in cell shown in Figure 2 and processed at the temperature and time indicated in Table 1.

After conversion, the compacts of Examples 1 - 15 were surface ground flat and parallel and analyzed by X-ray diffraction.

The diameter of the directly converted samples varied from approximately 11.7 to 12.4 mm. with thicknesses ranging from 1.57 to 3.66 mm. The samples densities tabulated in Table 1 were then determined. Except for the examples 10, 13 and 14 prepared at the lowest temperatures, the densities are, within experimental error, equal to the single crystal density. . 47548 In Example 16 a cluster conpact sanple was prepared from a conposite canpact by removal of the carbide substrate. Ey grinding and lapping until all carbide traces were gone. The final thickness was 0.94 nm.

The effective crystallite size of the eonpacts was determined by analysis of X-ray diffraction line (or peak) width. Analysis by this technique is based upon the fact that deviations from perfect crystalline structure, such as decreased crystallite size, lattice distortion resulting from non-uniform strain or lattice imperfections may result in extra peak width broadening. Because of the nature of the X-ray scattering process line width broadening only occurs If the crystallite size or lattice Inperfection separation is less than about 1000 X (0.1 micron), i.e., the diffracting beam cannot detect lattice distortion at distances greater than about 1000 X.

In non-metallic insulating crystals thermal energy is conducted by lattice waves (phonons). In good thermally conducting Insulating crystals such as high purity diamond or CBN single crystals the room tenperature mean free path of the thermal conducting phonons may be of the order of 1000 X or greater, with the mean free path increasing at lower tenperatures and decreasing at higher temperatures. Since the thermal energy in these crystals is transferred by lattice waves,' lattice Imperfections on the same scale as the phonon mean path, for perfect crystals tends to reduce the phonon mean free path and thus decrease the thermal conductivity (thermal conductivity is directly proportional to the phonon mean free path).

Lattice Imperfections on a scale which results in X-ray diffraction line width broadening is of the same order (or less) than the room temperature phonon· mean free path in CBN, i.e., lattice Imperfections on a scale which causes X-ray diffraction line width broadening may be expected to affect the room tenperature thermal conductivity in a negative manner.

For line broadening due to crystallite size reduction only, the effective crystallite size or relative crystalline perfection can be estimated frcm the relation: where t is the crystallite thickness perpendicular to the diffraction planes, β the diffraction angle, λ the wavelength of the X-radiation and B is related to the peak width by where is the peak width measured at half maximum intensity (FWHM) and Bg ia the peak width of a reference standard of large crystallite size (I.e., instrumental peak width).

While the above formula applies only to line broadening due to crystallite size effects and is not very accurate even in that case, tending to underestimate the crystallite size, It is useful as a parameter for comparing relative values of effective crystallite size, or relative crystalline perfection on a microscale.

X-ray diffraction line width broadening scans were taken of the CBN (111) and CBN (220) diffraction peaks of each of Examples 1-16. The calculated effective crystallite thicknesses in the CBN [111] direction are tabulated in Table 1 in which a more or less general increase in the crystallite size with increasing processing teiperature Is noted for the PBN compacts. A similar increase in size with increasing processing tenperature was also observed in the [220] direction. The residual conpressed HBN peak intensities observed for the thermal diffUsivity sanples are tabulated in Ibble i.

X-ray analysis of Exanple 15 and of other HBN powder converted compacts formed in a FIG. 2 type cell at various tenperatures also revealed that crystallite size increases with increasing tenperature. Important differences, however, were observed in the higher tenperature region·. With the PBN starting material a gradual increase in crystallite size occurred with increasing temperature until tenperatures in the region of about 2200°C was reached where the growth increased much more dramatically with increasing tenperature. With the HBN powder starting naterLal, the rate of increase of size with increasing tenperature was constant up to the reconversion temnerature.

FIG. .6 illustrates the crystallite size as a function of HP/HT processing tenperature (Table 1, column 5) of Examples 1-16 and a plurality of other cluster conpacts similarly prepared by direct conversion of ϋ-ΡΒΝ and HBN pow10 der. In the lower tenperature solid line bounded U-PBN (X-ray) region the crystallite size in the ϋ-PBN conpacts was determined by X-ray diffraction line broadening. The spread in this area represents the variation found in the crystallite size versus tenperature data which is believed to be due to lot-tolot variation in the structure of the U-PBN starting material. At higher tem15 peratures the crystallite size in the U-PBN conpacts becomes too large to be determined by X-ray diffraction. In the high tenperature region, designated U-PBN (SEM) in FIG. 6 scanning electron microscope (SEM) analysis indicated crystallite sizes in the 10 to 20+ micron range for a large crystallite size compact. The dashed line U-PBN region connects the X-ray and SEM regions. In this region where the sharp upswing of the curve occurs, the crystallite size is too large to be determined by X-ray diffraction and too snail to be investigated with the SEM available. With the HBN powder converted conpacts the crystallite size remained sufficiently small to be determined by X-ray analysis up to the highest tenperatures.

The different crystal growth behavior of the U-PBN and HBN pwoder converted conpacts may be speculated on as follows: The HBN powder starting material is conposed of individual platelet type particles of suhmicron thickness. On conversion of the individual particles the crystallite size in the particle is diminished. Crystallite growth can then occur within each particle. However, the extent of crystallite growth is limited by the Individual particle boundaries, i.e., crystallite growth does not proceed across particle to particle interfaces and thus the maximum crystallite size is limited by the size of the individual powder particles in the starting HBN powder.

With PBN, however, no discrete particles exist. Ihe PBN structure, , although highly disordered, is continuous in three dimensions. The PBN starting material disc may be viewed as one large, albeit highly imperfect crystal and therefore resultant crystallite growth after conversion to CBN is not limited by individual particle boundaries as with the HBN powder starting materials. It is thought that conversion in PBN proceeds directly from the turbostratic PBN structure to a quasi-amorphous CBN structure from which ciystal growth then proceeds, I.e., the conversion is not thought to proceed by initial recrystal15 lization of the turbostratic structure to the hexagonal structure prior to conversion of CBN. The turbostratic structure of PBN is stable to very high temperatures - recrystallization to the HBN structure does not occur at atmospheric' pressure up to the sublimation tenperature (2300°C to 2400°C). Recrystalliza±lon nay occur under low pressure uniaxial conpression, but only at temperatures of 2300°C or greater, I.e., higher than the high pressure PBN to CBN conversion tenperature (1700°C to l800°C).

Thermal diffusivity measurements on the conpacts were made using a flash heating method. The flash method involves subjecting the front face of the compact to a short energy pulse and monitoring the resultant tenperature rise of the rear face. A solid state laser Is preferably used as the energy source and the thermal diffusivity is calculated from the rear surface temperature history. Measurements- were made over the tenperature range from -100°C to 650°C. 548 Die measured thermal diffusivity values, °<, are converted to thermal conductivity, k, using the defining relation where Cp is the specific heat and sthe mass density. Densities were determined by the iranersion of sink float technique and known values were used for the specific heat.

Die thermal conductivity results are plotted in FIG. 7. Also shown in FIG. 7 are ths thermal conductivities of high purity copper, polycrystalline and single crystal BeO as well as roan tenperature thermal conductivity values for Type la natural single crystal diamonds of various nitrogen concentrations for reference.

From Table l and FIG. 7 an increase in the thermal conductivity with increasing processing tenperature Is noted for the U-PBN coipacts. Diis is shown graphically in BIG. 8' where the thermal conductivity at -50°C of Examples 2-4, 4, 6, 7, and 10-14 are plotted as a function of maximum processing tenperature.

Acconpanying the increase in k with increasing processing tenperature is a more or less general increase in the effective crystallite size and a decrease in the amount of unconverted residual conpressed HBN phase. Both of these effects contribute to the decrease in k at the lower processing tenperatures.

In the room tenperature region a factor of 3 to 4 inprovements in the thermal conductivity is observed between the high and low tenperature U-PBN conpacts. Diis difference is attributed to increased phonon scattering resulting from crystalline inperfections (decreased crystallite size) and increased thermal resistance resulting from the residual unconverted HEN phase in the lower temperature conpacts.

The room tenperature thermal conductivity of the best U-PBN conpacts is higher by a factor of 6 - 8 ccnpared to the directly converted HBN powder compact (Exanple is) and by a factor of about 10 conpared to the composite compact (Exanple 16). In addition, the conductivity of the HBN powder converted eom5 pact is significantly lower (by a factor of about 4) than PBN compacts showing similar X-ray line width broadening. Ihe extra thermal resistance of the HBN powder converted compact is attributed to increased intergrain thermal resistance in this conpact. likewise, the thermal conductivity of the conposite compact is considerably less than U-PBN conpacts of comparable X-ray line width broadening which is again attributed to increased intergrain thermal resistance in the conposite conpact.

Relative to copper, a PBN canpact in accordance with this Invention has higher thermal conductivities, approaching a factor of 2 improvement in the room tenperature to 200°C range. likewise the thermal conductivities are well in excess of that for polycrystalline BeO over the tenperature range investigated (a factor of about 4 ixprovement in the room tenperature region).

PIG. 9 shews in schematic form an embodiment of this invention which illustrates how a CBN cluster conpact (illustrated as thermal conductor 253) having a high thermal conductivity as disclosed herein may be used to provide a heat sink for an electronic device 251. For heat sinking device 251, a high thermal conductivity material 253 (CBN cluster caipact) is bonded by brazing alloy layers 255 intermediate device 251 and a large heat sink 257.

Techniques for forming layers 255 include metallization of this substrate material using a very thin sputtered epitaxial nickel film as disclosed in Hudson, J. Phys. D: Appl. Phys. 225 (1976), and the use of a high thermal conductivity silver based bonding alloy· Λ4 ta •P •H rM o a! Φ P o q-j ta n fel cn Λ cn ι CM in VO cm σ\ c*· cn ac vo in ay Cn On CH m CM CM CM in CM VC in 0 in 0 0 0 0 * on co co co σ t— on jst 0- CM c*- co vo .ar in ar cn cn c— cn cn h cxi ci on κ ci d> w CQ 0 in O 0 o . O O in O ,0 0 0 O O XT t*- VC •ar on 0> VO * CM VC vo vo co 4T O ar CM cn ay r—{ i—i on * 0 cc co co f-~ in CM CM CM CM CM CM CM CM H H H rd CM rH jSb cn on cn co oo cn J3· 4T JT cn cn cn on cn m on =r ar cn cn CM VO ar “ cn a> tg 40 ca 3* cn cn cm cn Φ r— S. cu <1 EX S g fe ffl ffl w Mi ill g 2 3 CU CU fU CM CU (X* ti » fi £> O ti «Τ in vo c-»

Claims (20)

1. A high temperature and pressure process for making a sintered polycrystalline cubic boron nitride compact, said process comprising the steps of: 5 (a) placing preferentially oriented pyrolytic hexagonal boron nitride in the form of a bevel-edged disc in a reaction cell, said boron nitride being substantially free of catalytically active materials, said cell having means for shielding said pyrolytic boron nitride from contamination during transformation; (b) compressing said cell and the contents thereof at a pressure between 50 kbars and 100 kbars; (c) heating said cell and the contents thereof to a temperature of at least 1780°C. within the cubic boron nitride stable region of the boron nitride phase diagram; 15 (d) maintaining said pressure and temperature conditions of steps (b) and (c) for a period of time sufficient to transform said pyrolytic boron nitride into a sintered polycrystalline cubic boron nitride compact; (e) ceasing the heating of said cell; and 20 (f) removing the pressure applied to said cell.

2. The process of Claim 1 wherein said pyrolytic hexagonal boron nitride is formed by substrate nucleation.

3. The process of Claims 1 or 2 wherein said pyrolytic boron nitride is greater than 99.99% boron nitride. 25

4. The process of Claims 1, 2 or 3 wherein said pyrolytic boron nitride has a density between 1.8 to 2.28 g/cm .

5. The process of any one of Claims 1 to 4 wherein said shielding means is a metallic shield for surrounding said boron nitride during transformation.

6. The process as claimed in any of Claims 1 to 5 wherein the cell and ics contents are heated to a temperature within the range 2100 to 2500°C.

7. A compact made in accordance with any of Claims 1 to 6 and consisting essentially of cubic boron nitride crystallites having (111) planes preferentially oriented and wherein said compact has a thermal conductivity of at least 2 watts/cm°C.

8. The compact of Claim 7 wherein said preferential orientation is between 50° and 100°.

9. The compact of Claim 7 wherein said preferential orientation is between 2° and 0°.

10. The compact of Claim 7 having a Knoop hardness between 4000 and 8000 kg/mm 2 .

11. The compact of Claim 7 having a density of at least 3.40 g/cm .

12. The compact of Claim 7 having a purity greater than 99.99%.

13. The compact of Claim 7 wherein the average crystallite size is at least 1000 A.

14. Λ compact as claimed in Claim 7 consisting essentially of CBN crystallites having a thermal conductivity greater than 6.3 watts/cm°C.

15. A process as claimed in any of Claims 1 to 6 substantially as herein described.

16. A compact as claimed in Claim 7 substantially as herein described.

17. A process for preparing S^fejfon nitride compact substantially as described in any of the Examples 1 to 14.

18. A cubic boron nitride compact when produced substantially in accordance with any of the Examples 1 to 14.

19. A cubic boron nitride compact when produced by the process of any of Claims 1 to 6.

20. A cubic boron nitride compact when prepared substantially as herein described with reference to the accompanying drawings.