(Issued May 1975).

and

1. Introduction

Two years ago a review article on ceramics in medicine by one of the authors reported that such materials are seldom used for chronic applications in medicine [1].1 However, pioneering efforts of the last few years have changed this situation. Dr. P. Boutin of Pav, France recently reported over 500 successful cases of total hip replacements using high density, high purity Al2O3 components over a 4 year period [2]. Application of the alumina. ceramic joint was based upon studies of very low wear rates of alumina in contact with alumina when both ball and socket are finely polished to a 1 surface finish. Biocompatibility of the high purity (99.3%) alumina was first established in dogs and humans using both acicular and button configura

1 Figures in brackets indicate the literature references at the end of the paper.

tions of the ceramic. Only a thin fibrous membrane several cell layers deep was present at the implanttissue interface. Similar results of tissue response and low rates of wear of dense high purity Al2O3 implants have been demonstrated by a West German research team [3-5]. These efforts have led to a systematic development of a total hip prostheses that is all Al2O3 ceramic. This is in contrast to the Boutin prostheses that utilizes a Ti metal femoral stem pressure fitted onto an alumina ball for the femoral portion of the hip prostheses.

A combined West German and Austrian research team reports total replacement of vertebrae, humerus, and radius in humans with high density, high purity Al2O3 ceramics [6]. Successful use of the implants for times in excess of 1 year are recorded.

With application of ceramics in humans commencing, it becomes imperative to understand fully interfacial reactions between ceramic materials

and bone. The studies reported by Hulbert and Klawitter et al. [7-10], and those cited above have demonstrated that very close apposition of mineralized bone occurs at the implant interface. However, if dense implants are used, taking advantage of the 40,000 psi (0.28 GPa) tensile strength of high density alumina, a thin fibrous membrane is always observed. The presence of the membrane causes one to reflect on the possibility of implant loosening during long term application with attendant stress localization and bone resorbtion currently characteristic of long term metallic implants. Will the membrane remain firmly attached to the ceramic interface or will interfacial fluids develop with stress-corrosion fatigue at the the ceramic grain boundaries? Can compositional, microstructural, or surface modifications be made to the alumina ceramics to avoid fibrous membrane formation without degrading strength? Answers to these and other fundamental questions are only beginning to be obtained. Alumina with 200 μm pore diameters and 50 percent porosity appear to be limited to 10,000 psi (0.07 GPa) tensile strength. Will 10,000 psi tensile be strong enough at the interface to transmit the loads required in weight bearing applications? Successful load bearing segmental bone replacements tested in monkeys in our laboratory using a bioglass-ceramic of less than 10,000 psi tensile strength suggest that mechanical design limits for such materials are presently not understood [11]. In other words, we do not know how strong a ceramic prosthesis must be to withstand chronic applications in humans. The data in animals is only beginning to be achieved in spite of much recent effort [1-15]. Much additional effort directed toward understanding interfacial reactions, strength requirements, and strength deterioration [16-19] must be completed before large scale chronic human applications seem warranted.

Alternative directions towards achieving satisfactory ceramic prostheses also continue to show promise. The research of Graves et al., at the University of Dayton has demonstrated considerable control over osteogenesis in monkeys using variable POs content in calcium-aluminate ceramics [12]. Strength enhancement in this system also seems quite probable.

The efforts of our laboratories have concentrated on the investigation of specially designed bioglass and bioglass-ceramic compositions [20-25]. This research will be reviewed in the remainder of this paper. Special emphasis will be placed upon the results of combining the osteogenic features of the surface active bioglass system with the high mechanical strength of 316L surgical stainless steel. The results indicate that flame spray coatings of bioglass on stainless steel or high density Al¿Os may provide a solution to many of the interfacial tissue and strength problems described above.

The surface ion release leaves behind a polymerizing SiO rich gel [22, 23, 26, 27]. The surface gel appears similar in composition to that preceding hydroxyapatite mineralization in normal bone growth [28]. As a consequence of the surface activity, collagen bonding and bone mineralization are initiated at the implant surface [20-25]. The bone growth from the implant-bone interface results in an interfacial chemical bonding which is sufficiently strong that the bone or the implant will fracture before the interface does [11].

Although the tensile strength of the bioglassceramic material is in the range of 10,000 psi when crystallized in a controlled manner [29], there are numerous prostheses applications where greater strength or impact resistance may be desirable. This is especially true for plates, nails, pins, and femoral head shafts where a small crosssectional area of the implant is required. Stainless steel devices satisfy the strength requirement for these applications. However, a fibrous tissue reaction at the steel interface results in loosening of the implant with time and corrosion of the implant produces metallic particles which become incorporated in the neighboring tissues.

An investigation was conducted to combine the superior strength and impact resistance of surgica! stainless steel with the biocompatibility of Na, Ca. P containing bioglasses. This combination is achieved by applying the bioglass as an intimate coating on the surface of stainless steel prostheses. The coating is applied by a flame spraying method developed in collaboration with the ceramics group at IITRI. This paper presents the development of

Ihmous Institute of Technology Research Institute, Ceramics Division, C m. S. W. Fresman and E. 1. Onesto, Investigators.

the bioglass flame spray coating process, the design of a flame spray coated partial hip prosthesis, surgical results, and a histological comparison of tissue responses adjacent to the partial hip prostheses with those of a standard surgical stainless steel screw.

Flame Spray Process

Flame spraying has been in use for well over half a century in the application of metallic coatings while efforts to apply ceramic coatings, with emphasis on refractory crystalline oxides, began about 20 years ago. The flame spray process consists of passing a material in powder form through a flame where it is heated to a temperature at which it becomes plastic and is readily deformed. The molten material is then impinged onto a substrate at high velocity with consequent deformation and adherence of the particles. The combination of heat and impact which is available in flame spraying permits rapid particle-to-particle sintering at temperatures and times where normal sintering would not be obtained. Consequently, the substrate does not reach a high temperature. For this reason, substrates can be flame spray coated without deformation and loss of tolerance. Diffusion processes between the substrate and the coating are also retarded. Both of these factors are important in producing a high quality coated medical prosthesis.

The spray efficiency and the quality of adherence of a flame sprayed coating depends on many factors, but of particular importance are the thermophysical properties of the sprayed material, and the surface conditions, thermal conductivity and thermal expansion of the substrate. For a given material and substrate, spray conditions are readily optimized empirically as to spray distance, gas flows, powder feed rate, etc.

3. Flame Spray Equipment

The flame spray apparatus consists of an oxyhydrogen torch 4,5 in conjunction with a powder feed hopper, shown schematically in figure 1. The oxygen line to the torch passes through the feed hopper which is equipped with a variable speed screw-feed mechanism to regulate powder flow. The powder is entrained in the oxygen stream and passes through the torch. The powder laden oxygen is combined with the hydrogen at the torch, the powder passing through the resulting oxyhydrogen flame. Since the theoretical flame temperature is approximately 3000 °C, the powder is rapidly heated and becomes molten.

When the material being sprayed is a glass, a special situation exists. Glasses differ markedly from metals and most crystalline ceramics in that they display high optical transparency and relatively high viscosity above the softening temperature. These differences are very significant to the flame spraying process, as was observed in previous work at IITRI [30].

As the particles are projected through the flame, they are heated by convection from the hot gases and by radiation from neighboring particles. The lack of the radiation component to heat transfer in the case of transparent glasses results in the evident low temperature of the glass particles relative to spray particles of opaque metals and ceramics. On the other hand, too high a particle temperature may destroy the glass, particularly compositions containing phosphate such as the bioglasses. This is due to the high vapor pressure of the phosphate groups. Also, the low thermal emittance of the particles imparts a slow cooling rate, hence more deformation and flow after impact (provided that the viscosity is sufficiently low).

The long softening range of glasses is advantageous where the substrate is hot or where the thermal conductivity is low because flow can continue for a longer period, resulting in better adherence and a tighter structure. A glass with high viscosity will not flow even when well over the softening temperature. The more rapid the quench rate of the particle, the lower the viscosity must be.

It has been found that a low viscosity is required for sintering as well as for flow of glass particles. Because of the smoothness of the molten glass surface the free energy is small, and the material must flow in order to sinter. This situation is not observed with metal and ceramic particles at high

temperature where surface defects give rise to a high free energy for sintering [31].

The relationship between softening range, quench rate and viscosity of a glass important to flame spraying is shown in figure 2. The volume decrease associated with the cooling of the molten droplets is shown as dark curves in the figure. If the liquid does not crystallize upon cooling to Tm, it can retain its random, liquid-like structure through a supercooled liquid regime, B, and solidify into a glassy material within the transformation range, Tg. The transformation temperature of a given flame sprayed glass is dependent upon its quench rate. A rapid quench produces a high T, and a less dense glass, curve C. Likewise, a slower quench rate results in high densities and lower transformation temperatures, curves D and E.

substrate be nearly the same as the substrate or slightly lower. A lower expansion will provide thermal shock resistance and a minimum of tensile stress established within the glass when the coating shrinks on cooling. However, as shown in figure 3, if there is large expansion mismatch between the glass and the substrate, tensile stresses will develop at the interface when the glass cools below its transformation temperature. Cracking will result if the stresses become too large [32].

FIGURE 3. Development of interfacial stresses between a flame spray coating and a prosthesis due to thermal expansion mismatch.

If a glass has a high viscosity before the glass transformation occurs, such as the dashed curve #1, the flame sprayed particles cannot flow and create a uniform glassy coating on the substrate. However, when the viscosity of the glass is low (curve #2), spreading and interparticle flow occurs prior to solidification of the glass. This condition must be achieved by adjusting the glass composition. There is a limit on compositional adjustment, however, since crystallization at Tm will occur if sufficient three-dimensional network formers are not present in the molten glass particles. Because of the large volume change associated with crystallization (curve F) it is very difficult to obtain a smooth, coherent crystalline coating.

To supplement the low viscosity, a low surface tension of the glass is also required for good wetting of the substrate. Particle flow and sintering after impact will be a function of both viscosity and interfacial tension.

It is most important that the coefficient of thermal expansion of the glass sprayed on a relatively cold

5. Development of Bioglass Composition for Flame Spraying

Initial flame spraying experiments were conducted using flat plates of 316L surgical stainless steel as the substrate for the 45S5 bioglass (table 1) originally developed for bulk glass-ceramic prostheses. All substrates were grit blasted with 20 mesh SiC at a pressure of 80 psi to completely roughen the surface before spraying. The devices were then thoroughly scrubbed in a detergent solution and rinsed first in water and then in isopropyl alcohol to remove any grit, blasting debris, dirt and oils picked up in handling the devices.

A variety of experimental variables were examined in an effort to achieve a high adherence coating of the 45S5 bioglass [33]. None succeeded. It was concluded that the major reason for the lack of success in flame spraying the 45S5 glass was its very high viscosity at the spraying temperature. This high viscosity prevented the sprayed material from flowing and forming an adherent glassy layer. The high viscosity of the glass is probably due to its relatively high silica content. Previous studies had shown that glasses which form the best flame sprayed layers are those that are low (<5%) in SiO 2 [30].

This interpretation was borne out by an analysis of the thermal expansion coefficient of the 45S5 bioglass and the dilatometric softening point. As shown in figure 4 (determined on 2-inch specimens with an Orton Automatic Recording Dilatometer), the expansion of the 316L stainless steel substrate and the 45S5 bioglass are well matched. However, the softening of the glass under stress, revealed by the hump in the curve, occurs at a relatively high temperature (530 °C) for this composition. Thus, the glass viscosity is too high for a successful flame sprayed coating.

Two approaches were taken to decrease the glass viscosity; substituting B2O3 for SiO2 in the glass and CaF for CaO (table 1). Both changes decreased the viscosity as shown by the lower softening points in figure 4. The decrease in softening point to 470 °C was the same for both the fluorine and boron containing bioglasses. The magnitude of the thermal expansion coefficient was also the same for both compositional series. However, the higher temperature fluidity of the fluorine-containing bioglass 45S5F was less under the flame spraying conditions and produced a smooth, uniform glassy coating on the stainless steel with excellent adherence.

Since it is required that implant devices be virtually encapsulated in a glass coating, it was necessary to flame spray each device individually. Samples were hand held in forceps modified to grip them with minimum contact. In this manner,

front, back, and interior surfaces (in the case of cylinders) could be uniformly preheated and coated with a well fused, glassy appearing coating.

Coatings on the prostheses were obtained by preheating the substrate above the softening point of the glass, then initiating the powder feed through the torch and allowing the glass particles to flow, rather than quench, as they struck the surface.

Altering the bioglass composition in order to obtain suitable flame spray behavior does not yield a unilateral change in properties. Consequently, evaluation of solubility characteristics and in-vivo responses of the new compositions were necessary before it was justified to produce and test coated prostheses. Previous publications show that the in-vivo behavior of both the fluorine and boron containing compositions were suitable for prostheses applications [21-25].

200

300

400

500

600

700