The bonding of bioglass to a cobalt-chromium medical and dental alloy

MISSING IMAGE

Material Information

Title:
The bonding of bioglass to a cobalt-chromium medical and dental alloy
Physical Description:
xi, 172 leaves : ill. ; 28 cm.
Language:
English
Creator:
Lacefield, William R ( William Randolph ), 1946-
Publication Date:

Subjects

Subjects / Keywords:
Glass-metal sealing   ( lcsh )
Metal bonding   ( lcsh )
Sealing (Technology)   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1981.
Bibliography:
Includes bibliographical references (leaves 164-171).
Statement of Responsibility:
by William R. Lacefield, Jr.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000295402
notis - ABS1747
oclc - 07883902
System ID:
AA00003454:00001

Full Text














THE BONDING OF BIOGLASS TO A COBALT-CHROMIUM
MEDICAL AND DENTAL ALLOY










BY

WILLIAM R. LACEFIELD, JR.


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA


1981
















ACKNOWLEDGMENTS

The author wishes to express his gratitude to his committee for

their criticisms and suggestions, and especially Dr. Larry Hench

for his guidance and encouragement throughout this study. Special

thanks are given to Dr. Hamdi Mohammed for his role as cochairman and

advisor in the area of dental materials. The author also appreciates

the assistance of Dr. Fumio Ohuchi and Dr. Paul Holloway in inter-

preting the AES results, and Alice Holt for her help in preparing and

typing the manuscript.

This research was funded in part by Howmedica, Inc.
















TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS............................................ ii

LIST OF TABLES............................................. vi

LIST OF FIGURES........................................... viii

ABSTRACT................................................... x

CHAPTER

I INTRODUCTION.................................. 1

Metals, Ceramics and Glass as Implant
Materials.................................... 1
The Development of Bioglass.................... 3
Bioglass-Coated Implants....................... 4
The Development of Bioglass-Coated
Metal Implants.............................. 6
Objectives of this Research.................... 10

II GLASS METAL BONDING............................ 14

Glass-Metal Devices and Applications........... 14
Methods of Coating............................. 15
Nature of Welting.............................. 18
Theories of Adherence .......................... 20
Bioglass Coating Methods....................... 25
Summary........... ......................... 27

III EXPERIMENTAL PROCEDURES........................ 31

Laboratory Procedures, Equipment and
Materials.................................... 31
Thermal Expansion Measurements................. 34
Microscopes and Specimen Preparation
Techniques................................... 35
Compositional Analysis Techniques............... 36











BOND STRENGTH TESTING..........................


Nature of Bond Testing......................... 42
Porcelain Enamel Tests......................... 43
Glass-Metal and Ceramic-Metal Seal
Testing ...................................... 44
Testing of the Porcelain-Metal Crown............ 45
Suitability of Tests for Measuring
Bioglass-Vitallium Bond Strength............. 52
The Push Thru Shear Test....................... 56

V THE METAL-METAL OXIDE INTERFACE................ 60

Introduction.................................. 60
Oxidation of Cobalt-Chromium Alloys............. 61
Comparison of Cobalt-Chromium with
Other Alloys................................. 63
Changing of Nature of the Metal-Metal
Oxide Bond.................................. 65
Analysis of the Oxide Layer on
Vitallium.................................... 68

VI EFFECT OF COATING PROCESS VARIABLES ON
GLASS-METAL BOND STRENGTH...................... 77

Immersion Process Variables ................... 77
Cleaning Cycle for Metal Substrate.............. 78
Surface Roughness of Metal...................... 79
Temperature and Time of Oxidation
of Metal....................................... 84
Temperature of Molten Glass and Time
of Immersion................................. 88
Post Immersion Annealing Cycle................. 89
Oxygen Pressure During Oxidation............... 92
Degassing of Metal Substrate.................. 94

VII GLASS COMPOSITIONAL AND COATING
MODIFICATIONS ................................. 102

Enameling and Double Coating of
Vitallium with Bioglass..................... 102
Addition of Adherence Oxides to
Bioglass.................................... 107
Bioglass Containing Fluorine................... 110
Coating of Vitallium with Dental
Porcelain............ .................... 112
Effect of Bioglass Composition on
Interfacial Porosity......................... 114












VIII BIOLOGICAL, ENVIRONMENTAL, AND FATIGUE
TESTING....................................... 116

Fatigue of Glass.............................. 116
Fatigue Behavior of Bioglass.................. 117
Bioglass-Vitallium Orthodontic Implants....... 120
Implant-Tissue Interfacial Study.............. 123
Bioglass-Vitallium Endosseous Implants........ 126

IX ANALYSIS OF THE GLASS-METAL INTERFACE......... 129

Objective of Interfacial Analysis............. 129
Electron Microprobe Analysis.................. 130
Scanning Electron and Light Microscopy........ 130
Auger Electron Spectroscopy................... 137
Discussion of AES Results..................... 142
Factors Important to Bond Strength............ 148

X SUMMARY AND CONCLUSIONS....................... 155

Bioglass-to-Vitallium Bonding................. 155
The Suitability of the Immersion Process
for Coating Vitallium Implants with
Bioglass................................... 156
Conclusions.................................. 160

REFERENCES ..... ........................ .............. 164

BIOGRAPHICAL SKETCH...................................... 172
















LIST OF TABLES


Table Page


1 Chemical Analyses of Cast and Wrought 11
Vitallium Alloys

2 Bioglass Compositions 12

3 Free Energy of Formation of Metal Oxides 28

4 Glass-Metal Bond Strengths from Various 53
Researchers

5 Auger Electron Spectroscopy of Vitallium 72
Specimen Oxidized in Air

6 Auger Electron Spectroscopy of Vitallium 73
Specimen Oxidized in a Partial Vacuum

7 Effect of Cleaning Agent on Bond Strength 80

8 Effect of Surface Roughness on Bond Strength 83

9 Effect of Oxidation Time and Temperature on 86
Bond Strength

10 Effect of Glass Temperature on Bond Strength 90

11 Effect of Immersion Time on Bond Strength 91

12 Effect of Annealing Cycle on Bond Strength 93

13 Effect of Oxidation Pressure on Bond Strength 95

14 Bond Strengths of Various Special Test 105
Specimens

15 Thermal Expansion of Coefficients of Bioglass 111
and Vitallium

16 Fatigue Strength of the Bioglass-Vitallium 119
Bond











Table Page

17 Bioglass-Vitallium Wire Bond Strengths 122

18 AES Analysis of Metal B (Poor Bond) Fracture 139
Surface

19 AES Analysis of Metal A (Good Bond) Fracture 140
Surface

20 Effect of Glass Heat Treatment on Bond 150
Strength


vii
















LIST OF FIGURES


Figure Page


1 Typical AES analysis of Vitallium 38

2 Typical EMP analysis of Bioglass-Vitallium 41

3 Tests of glass-metal seal strength 46

4 Shell-Nielsen shear test 50


5 Dental porcelain-metal bond strength tests 51

6 The push thru shear test 57

7 Scanning electron micrographs of grit blasted 69
Vitallium surfaces

8 Auger electron spectrographic analysis of 72
an air oxidized Vitallium specimen

9 Auger electron spectrographic analysis of 74
oxidized Vitallium specimens

10 Scanning electron micrographs of Vitallium 82
surfaces roughened by various techniques

11 Bioglass-Vitallium bond strength as a 87
function of time and temperature of
oxidation

12 Scanning electron micrographs of a Vitallium 106
specimen enameled with a layer of frit then
immersed in molten Bioglass

13 Scanning electron micrographs of Bioglass- 109
Vitallium interfaces of frit-enameled
specimens

14 Scanning electron micrographs of a Vitallium- 125
Bioglass implant


viii










Figure Page

15 Scanning electron micrographs of porosity 127
at the interface of a Bioglass-Vitallium
canine implant

16 Electron microprobe analysis of a Bioglass- 131
Vitallium interface

17 Scanning electron micrograph showing Vitallium 133
fracture surface with scattered glass pieces

18 Scanning electron micrographs showing the 134
junction of glass pieces to the oxidized
metal surface

19 Scanning electron micrographs of metal 135
fracture surfaces

20 Scanning electron micrographs of pores in 136
the glass section of a fracture surface

21 Auger electron spectrographic analysis 141
showing the ratio of metallic ions to
oxygen in the glass

22 Auger electron spectrographic analysis of 143
a Bioglass fracture surface

23 Auger electron spectrographic analyses of 144
the metal interfacial region of good and
poor bond specimens

24 Auger electron spectrographic analyses of 145
specimens from Groups I, II, and III

25 Auger electron spectrographic analyses of 146
a Bioglass-dental porcelain specimen and
a ceramed Bioglass specimen
















Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



THE BONDING OF BIOGLASS TO A COBALT-CHROMIUM
MEDICAL AND DENTAL ALLOY


By

William R. Lacefield, Jr.

June, 1981


Chairman: L. L. Hench
Cochairman: H. A. Mohammed
Major Department: Materials Science and Engineering


The use of Bioglass as a coating on a cobalt-chromium surgical

implant alloy gives a composite material which has both the good

mechanical properties of the metal and the ability of the glass to

form a chemical bond with bone. Good adherence of the glass coating

is essential to assure the proper functioning and integrity of the

loaded glass-metal implant in vivo.

A shear strength test was developed so that the effect of various

factors (such as surface roughness) on glass-metal bond strength could

be determined quantitatively. Test results showed that adherence

between glass and metal is more dependent on chemical bonding than

on mechanical interlocking.











Analyses of the glass-metal interface by Auger electron spectros-

copy (AES), electron microprobe (EMP), and scanning electron microscopy

(SEM) were used to determine the compositional profiles and surface

characteristics which are associated with high bond strength. The

thickness of the oxide layer formed on the alloy prior to the coating

operation is the key factor which determines the strength of the bond

formed between glass and metal. For test specimens with high bond

strength, fracture typically occurs between the metal and metal oxide

layer. Specimens with low bond strength either fracture a) within the

oxide layer if this layer is too thick or b) between the glass and

metal if no oxide layer is present. Prolonged heating of a coated

specimen with good bond strength decreases glass-metal adherence due

to loss of metal oxide saturation in the glass or the formation of

voids at the interface.

Fatigue strength of the glass-metal bond was found to be lower

for coated specimens tested in vivo, with the observed loss of strength

attributed in part to the presence of porosity at the interface. In

vivo tests using Bioglass-Vitallium implants showed that 1) bond

strength of the coating under various types of loading is more than

adequate, 2) the reactivity of the Bioglass is not affected by the

immersion coating operation, and 3) there is no incidence of increased

corrosion at the glass-metal junction due to metal sensitization.
















CHAPTER I
INTRODUCTION



Metals, Ceramics, and Glass as Implant Materials

The use of metal and ceramic materials in the human body has be-

come increasingly widespread in the past twenty years. Metals are

commonly used in medical devices such as hip and knee prostheses, heart

valves, and electrodes; and in dentistry as crowns, implants, and fill-

ing materials. Ceramics are used both as permanent structural implants,

such as high density alumina for femur replacements, and as temporary

support implants which soon dissolve in the body (e.g., calcium phosphate).

The use of glass in the body has been traditionally restricted to appli-

cations which require proper esthetics--such as the porcelain used to

simulate tooth enamel.

At the present time the primary use of glass-metal composite devices

or restorations in the human body is the porcelain-fused-to-metal crown

used in dentistry. This type of restoration is widely used for anterior

teeth because of the desirable combination of the strength of the metal

and the natural, toothlike appearance provided by the enamel (glass).

A requirement of all implant materials is that they are tolerated

by the body after implantation. Any adverse reaction in response to the

implant, such as inflammation or increased incidence of infection, can

lead to its failure and removal. Even with a relatively well-tolerated,










inert material such as alumina, the body typically responds by isolating

the implant from the rest of the system by surrounding it with a fibrous

capsule.

Many of the problems associated with the use of biomaterials are

caused by the methods of fastening the implants in their proper position.

Structural or load-bearing devices are usually fastened in place by

cements such as polymethylmethacrylate (e.g., for hip prostheses) and

zinc phosphate (e.g., for metal crowns), or by mechanical means such as

screws and fasteners on a metal splint, or by press-fitting the specially

shaped implant into a prepared space in the bone. Often these methods

of attachment cause bone resorption around the implant leading to loosen-

ing and possibly fracture and thus necessitating the removal of the

device. Other undesirable effects which can lead to subsequent removal

of the implant are the exothermic reaction and monomer release to the

system resulting from the use of PMMA cement, and the wear particles

and corrosion produces in the tissues from metallic devices and fasteners.

Porous materials which allow tissue ingrowth to provide mechanical

fixation of the implant in its proper position have been developed in

order to overcome the problems associated with other methods of fixation.

Tissue ingrowth and partial fixation have been observed with both metallic

and ceramic porous materials, but these materials have been generally

unsatisfactory for use as load-bearing implants because their increased

surface area leads to more corrosion and dissolution, and their porous

structure results in an implant with much lower mechanical properties

(especially fatigue strength) as compared to the dense materials.










The Development of Bioglass

The use of both porous material and the conventional methods of

mechanical fixation of the implant to bone has been observed to be un-

satisfactory in many cases. This has lead to the development of a new

class of bioactive material which forms a direct chemical bond with bone.

A series of glasses and glass-ceramics whose compositions fall within a

specific range has been developed by L. L. Hench and others and given
tB1-4
the name "Bioglass."14 These glasses contain primarily CaO, P205, Na20,

and SiO2, with some limited substitution of other oxides such as CaF2 and

B203 possible in order to modify the reactivity or fluidity of the glass,

if necessary. The composition and reactivity of Bioglass are such that

the body makes no attempt to isolate it from the rest of the system by

the formation of a fibrous capsule.

The mechanism by which Bioglass forms a strong bond with bone has
5,6 + 2+
been extensively studied and is well documented.6 Initially, Na Ca,

and P5+ ions leach out of the glass leaving a silica gel structure, the

thickness of which depends on such factors as time of exposure of glass

to the solution. As osteoblasts begin to lay down new bone substance

around the implant, the Ca and P leached from the Bioglass precepitate

out of the solution and form an amorphous layer on the silica gel matrix.

Thus a graded interface is formed consisting of Ca and P from the Bioglass

and the Ca and P (hydroxyapatite) of the new bone, and a chemical union

between the Bioglass and the bone is established. Microscopic examina-

tion of this interface shows highly elongated hydroxyapatite crystals

bridging the space between the gel layer of the implant surface and

mature bone.










The bond between bone and Bioglass typically exhibits a strength

of 75% of that of dense cortical bone, and the bond area itself is often

stronger than the surrounding bone if sufficient time is allowed for

full development of the bond (generally around 6 weeks).7 Factors such

as vibration or movement of the implant8 or the presence of undesirable

9
elements such as alumina at the bone-Bioglass interface during early

stages of bond formation can be detrimental to bone formation and little

or no bonding will occur.



Bioglass-Coated Implants

The ability of Bioglass to form a direct chemical bond to bone

with only a small surface zone (usually about 100 pm thick) being

affected by leaching means that the original strength of the bulk

material is retained. Unfortunately, Bioglass, like most glasses, is

relatively weak in tension and is subject to crack formation, propaga-

tion, and brittle failure. Therefore, its use as a bulk material for

load-bearing applications in orthopaedics or dentistry is doubtful.

The ideal use of Bioglass in load-bearing applications is as a coating

on a stronger material such as stainless steel or high density alumina.

The thin Bioglass coating reacts to bond the implant to bone, while

the substrate metal or ceramic provides the strength necessary to with-

stand the normal forces to which the implant is subjected. This

necessitates the formation of a good bond between the Bioglass and the

substrate so that the forces are transmitted from the weaker outer

layer of glass to the bulk of the implant.










There are a number of factors to be considered when selecting an

implant material to be coated with Bioglass:

1. The material must have high strength to withstand the complex
tensile, compressive, and torsional loads it will be subjected
to as an implant.

2. It must have a high modulus of elasticity, as excessive bend-
ing under load will cause fracture of the glass coating.

3. It must have been previously certified as an acceptable implant
material, both to facilitate the acceptance of the coated
device with the medical profession and to avoid the long times
necessary to certify a new material for use in the human body.

4. It ideally should have a thermal expansion coefficient similar
to Bioglass because of the necessity of heating and cooling
the coated device during the coating operation.

High density alumina is a high strength, high modulus material

which shows promise for use in both orthopaedics and dentistry. There

are several drawbacks to the use of alumina, however, one of which is

its observed loss of strength in a physiological environment. Fatigue

tests on 65% dense alumina in simulated body solutions showed a reduction

in strength of an average of 35% after 12 weeks,0 and other tests have

shown that there is also reason for concern about the loss of strength of

high-density alumina in various solutions.11

A limitation of the use of alumina as a substrate for coating with

Bioglass is the large difference in its thermal expansion coefficient

(about 8 x 10-6/C) as compared to that of Bioglass (13-15 x 10-6 /C).

Not only is there an excessively large stress level produced at the

interface during cooling from the coating temperature, but also the

lower expansion of the substrate puts the glass layer in tension, thus

enhancing its tendency to crack.










To overcome this alumina-Bioglass expansion mismatch and avoid
12
possible failure of the coated device, Greenspan2 developed a method

of double coating the alumina substrate. In this method the inner

layer of Bioglass takes up the expansion mismatch so that the second

layer has little strain imposed on it during cooling and an expansion-

graded interface is established which, while not ideal, appears to be

satisfactory. After a series of successful tests of the Bioglass-coated

alumina system as dental implants in primates,13 clinical trials are

now being designed for humans. However, development of successful

Bioglass-coated alumina orthopaedic devices appears many years away.

There are several advantages to using metals such as stainless

steel or cobalt-chromium alloys rather than alumina for Bioglass-coated

implants. Metals have long been used in load-bearing applications in

the body, have good fatigue strength in a physiological environment,

and are not as subject to brittle failure as ceramics. Also, the

higher thermal expansion of these metals (13-16 x 10-6/C) is very

similar to that of Bioglass.



The Development of Bioglass-Coated Metal Implants

The coating of metals has long been a part of the Bioglass develop-

ment program. Some of the original work was performed at the University

of Florida using 45S5 Bioglass for coating a stainless steel substrate.

An enameling method was used in which the glass was placed on the metal

as a frit, and the substrate taken to a temperature where the frit

particles fused and formed a uniform glass coating. This method was










unsuccessful because metal ions diffused from the substrate through

the glass to the surface.

A second direction was taken which involved the use of a commer-

cial ground coat (Ferro and Pemco Corporations) applied first to the

stainless steel substrate with Bioglass applied as a second coating.

However, metal ion migration was again a problem as the transition

metals in the ground coat and Fe and Cr from the steel were found on

the Bioglass surface after heating to the high temperatures necessary

to fuse the frit particles and to form a smooth, glossy surface. The

concentration of the metal ions on the Bioglass surface was sufficient

to render the coating unsatisfactory for biological applications.14

A third approach was taken a few years later by the Leitz group

in Germany, using a surface active glass similar in composition to

Bioglass, but termed Ceravital.15 They applied and fired a ground

coat to stainless steel using standard enameling techniques, then

fused crystallized glass-ceramic Ceravital granules to the ground

coat in a separate process. This two-step procedure caused the forma-

tion of a rough granular surface but limited the migration of Fe and

Cr. Unfortunately, recent results have shown that there is a contin-

uous biological attack of the glass-ceramic granules and breakdown

of the interfacial structure in contact with bone.

A flame-spray coating method was developed at the University of

Florida in an attempt to get a uniform coating of Bioglass without

having to heat the metal substrate to the glass softening temperature.1

This method was partially unsuccessful because of the porosity which










developed at the glass-metal interface as the hot glass particles

struck and fused to the metal surface. Another problem with flame

spraying Bioglass was the high viscosity of the glass at the coating

temperature. Lower-viscosity Bioglasses containing fluorine and

boron were developed which enabled the production of smooth, uniform

glass coating on the steel substrates.7 However, porosity at the

interface was still a problem and in vitro tests showed the coating

to have low fatigue strength as well as excessive corrosion at the
18
glass-metal interface.8 Also, a number of flame-spray-coated hip

implants failed after in vivo testing in monkeys.

In an attempt to overcome some of the problems inherent in the

flame spray coating method, Buscemi and Hench developed an immersion

process for coating Bioglass to metal.19 In this technique the metal

substrate is immersed in molten Bioglass for short times (e.g., 3 sec)

and then withdrawn at a rate which allows the excess glass to flow off

leaving a fairly uniform coating of the desired thickness. Porosity

is typically observed at the glass-metal interface of immersion coated

specimens, but not to the extent seen in flame-sprayed specimens.

The bond strength of immersion-coated specimens appeared to be

higher than flame coated ones as judged by observing the integrity of

the coating after quenching the coated specimens in water from various

temperatures. In vivo tests showed immersion coated 316L stainless

steel implants withstood tensile loads of 60 and 137 lbs after 8 weeks

implantation time without failure of either the glass-metal or glass-
bone interfaces.20 However, quench test results on immersion coated
bone interfaces. However, quench test results on immersion coated










Vitallium* indicated that Bioglass formed a better bond with this

surgical grade Co-Cr alloy than with 316L stainless steel. In addi-

tion, the corrosion susceptibility and interfacial bond strength of

Bioglass-coated Vitallium chips implanted in rats were superior to

similarly coated stainless steel implants from previous tests.21

After evaluating the results of all the tests of bond integrity

as well as the preliminary in vitro and in vivo testing, it was

determined that Vitallium was the metal of choice for use as a sub-

strate for coating with Bioglass.



Characteristics of Vitallium

Vitallium is an alloy of cobalt, chromium, and molybdenum which

falls into the general classification of a super alloy. Some industrial

alloys of this group, such as Haynes Stellite-21, are very similar to

Vitallium in composition. These alloys have been developed primarily

for applications which require metals with high creep-rupture strength

and corrosion resistance at temperatures over 8000C. Typical applica-

tions are as either forging or castings for turbine discs and blades,

nozzle vanes in jet engines, and in sheet metal assemblies such as

22
combustion chamber lines, tail pipes, and after burners.22 Molybdenum

is added to these alloys as a solid solution strengthener at room temp-

erature, and imparts much better creep resistance at higher temperatures.

Chromium gives these alloys their excellent corrosion resistance.




*Howmedica, Inc., Chicago, Illinois










Vitallium and similar Co-Cr alloys were first developed for re-

movable partial denture castings in about 1930, and by 1950 over 80%

23
of all partial dentures were made from these alloys.23 Because of its

excellent corrosion resistance and relative inertness in the body,

Vitallium has found wide use as a surgical implant alloy for such

applications as total hip replacement and dental implants.

A wrought form of Vitallium (FHS) has recently been developed to

give increased strength to the alloy for applications in which a load-

bearing implant is required. Table 1 gives a comparison of the composi-

tion and properties of both the wrought and standard cast Vitallium

alloys.



Objectives of This Research

The primary objective of this research is to determine the nature

of the bond formed between Bioglass and Vitallium using suitable

analytical techniques and to correlate the findings and observations

of this study to the prevailing theories on glass-metal bonding. Some

of the secondary objectives of the overall research program are:


1. Optimize the immersion coating process to obtain a Bioglass-
Vitallium bond of maximum strength, and thus increase the
reliability of the composite implant device for use in animal
and human tests.

2. Develop a suitable testing method to monitor progress in
increasing bond strength and to give a quantitative shear
stress measurement which can be used to predict the reli-
ability of a glass-metal composite in actual service
conditions in the body.

3. Determine if the coating process alters the reactivity or
bone-bonding ability of the Bioglass.








11












u I ;(
00

cd 4c











-4
0 n
0i










b-l

C
M

o 0 o


*l





44
0




o0 o
w*



a rt
















O N *
tU)
CO










-H






co
c-4 -n 4


U) U






4_) @ g
-4 -4 o
t0 a 2)
u-

















Table 2

Bioglass Compositions



Nomenclature: 52 S 52% Si
4.6 Ca/P Mole Ratio

52S4.6 Bioglass

Oxide Wt %

Si02 52

CaO 21

Na20 21

P205 6



45S5F Bioglass

Oxide Wt %

SiO2 45

CaO 12.25

CaF2 12.25

Na 0 24.5

P205 6


University of Florida






13



4. Establish the susceptibility of the glass-metal interface
to corrosion and fatigue.

5. Determine the microstructural and surface compositional
factors responsible for good interfacial bonding.

6. Attempt to resolve some of the controversy in the litera-
ture about the exact nature of glass-metal bonding and the
relative importance of chemical and mechanical factors on
bond strength.
















CHAPTER II
GLASS-METAL BONDING



Glass-Metal Devices and Applications

Devices which include glass-bonded-to-metal composites are used

in a wide variety of applications. For example, there are a number of

electronic devices in which proper glass-metal sealing is an important

part in the ability of the component to function properly. One common

application is the sealing of metal wires to glass envelopes in forming

connections such as in light bulbs or electron tubes. Also, glass frit

in a silver conductive paint is used to form the seal (when fused)

between metal lead wires and ceramic capacitor bodies.

Porcelain-enameled steel is another commercial product in which

good glass-metal bonding is essential. Enameled steel or iron is used

for such household items as refrigerator and stove side panels, cooking

pans, and bathtubs. The porcelain coating is added to enhance the

abrasion and corrosion resistance of the steel, as well as to improve

its appearance. The porcelain coating (usually about 0.5 mm in thick-

ness) typically consists of a blue ground coat and outer coat of white

or colored opaque material.

The porcelain-fused-to-metal crown used extensively in the restora-

tion of anterior teeth also depends on a good glass-metal bond for proper

functioning. A thin metal casting (either of gold alloy or a non-noble










alloy such as nickel-chromium) fits over the preparation of the re-

maining tooth structure and provides strength to the restoration. The

natural tooth-like appearance of the composite restoration is provided

by the translucent porcelain. In preparing the glass-metal crown, a

layer of opaque porcelain is put on as a frit then fired so that a

thin, opaque coating is formed which masks the color and reflectivity

of the metal. One or more layers of properly colored porcelain frit

are then added and fired so that the finished restoration has the

translucency and color of the surrounding natural teeth.



Methods of Coating

There are basically four methods of coating a metal substrate with

glass:


1. Application of the glass in particle or frit form, then firing
to form a uniform coating.

2. Flame spray coating of glass particles on the metal substrate.

3. Immersion of the metal substrate into molten glass.

4. Sputter coating a thin layer of glass onto the metal substrate
in a vacuum.

Of these methods, fusion of fritted glass on the metal surface is

by far the most commonly used. Commercial enameling is accomplished by

first dipping or spraying the metal piece with porcelain frit suspended

in a liquid solution, then firing the frit-coated metal until a homo-

geneous glass layer is formed. A blue ground coat containing cobalt and

nickel oxides is often put on the metal first to promote better adherence

between the final decorative coat and the metal substrate.










Glass-to-metal sealing of electronic components is also accom-

plished by fusing low-melting glass frit to the metal and glass

surfaces to be joined. Important considerations are that the glass

should wet the metal and the two should have similar thermal expansion

coefficients since the whole device must be heated to allow melting of

the glass frit.

In the fabrication of a dental crown, the porcelain is also put

on the metal substrate as a frit and the entire restoration fired to

a suitable temperature (usually in a partial vacuum). In this firing

operation, as in all cases in which the glass is put on the substrate

as a frit, it is imperative that the metal has a substantially higher

softening temperature than the glass so that creep and deformation of

the metal substrate do not occur.

Flame spray coating is a technique more often used for the coating

of higher-melting glasses and refractory oxide ceramics onto metal sub-

strates. An example is the alumina coating applied by flame spraying

to a titanium alloy for orthopaedic implant applications. The principal

advantage of this technique is that the metal substrate does not have

to be taken to a high temperature to allow fusion of the glass or ceramic

particles. The coating particles are fused while passing through a suit-

able flame (such as that of an oxy-hydrogen torch) before striking the

preheated metal substrates. This combination of heat and impact causes

the glass to flow and form a homogeneous coat at lower temperatures than

are possible using the conventional method of fritting and then firing

the coated metal substrate.










Immersion of the metal substrate into molten glass is a third

technique for coating metal with glass. This method is rarely used

in commercial product coating because it is expensive and does not

lend itself well to a production line operation. Coatings by this

method tend to be somewhat thicker than with other techniques, and

the coating of complex shapes is more difficult. An advantage of

using this technique is that the metal substrate can be preheated to

a low temperature and the time of contact with the molten glass can

be kept short so that the metal undergoes little change in micro-

structure or mechanical properties.

A fourth coating method is the sputtering of a glass onto a metal

substrate in a vacuum. By properly locating a glass piece in the path

of charged ions in a bell jar, the rapidly moving ions knock off small

pieces of the glass sending them in all directions within the chamber.

If a metal substrate is placed in the path of the heated glass molecules,

a coating on the metal surface is slowly built up with time. Precision

coatings can be applied by this method, but it is the most expensive of

the four coating techniques on a cost-per-piece basis. Only thin coat-

ings are usually produced by this method because the rate of deposition

of glass is so low that it is difficult to get a thick coating in a

reasonable time. For example, a new technique of ion beam coating
0
(using argon ions) produces a glass coating at a rate of only 5000 A

per hour.










Nature of Wetting

Whatever the method used for applying a glass coating to a metal,

the molten glass must first wet the surface of the metal in order for

the coating to form a bond with the substrate. The degree of wetting

can be estimated by measuring the contact angle that a drop of molten

glass makes on the surface of the metal substrate. This angle can be

measured by use of a goniometer24 or calculated by the relation:


YLV cos e = YSV YSL


where 6 is the contact angle and yLV' YSV' and ySL are the interfacial

energies between the liquid-vapor, solid-vapor, and solid-liquid phases,

respectively. A 0 of less than 90 is generally indicative of good

wetting, whereas larger contact angles indicate the absence of wetting.24

Simply having an acute contact angle is not enough to guarantee good

wetting however, as 6 must approach 0 to get spreading or proper wetting

on some surfaces.

One problem in the bonding of glass to metal or vice versa is that

the surface energies of the two materials are greatly different in value.

For example, surface energies for most silicate types of glasses range
2
from about 250 to 400 ergs/cm2. The surface energies of metals are

2 25
higher, ranging from about 1000 to 1800 ergs/cm This difference

leads to poor wetting, which may be desirable in certain cases such as

non-wetting of a refractory ceramic furnace liner by molten metal. In

attempting to bond glass to metal, however, it is absolutely essential

that the molten glass wet the metal substrate. If a condition of poor










wetting exists, the glass will not penetrate into the valleys of the

surface roughness (on a microscopic scale). If this condition exists,

only the peaks will be contact points and only a fracture of the

existing metal surface area is available for chemical bonding or

mechanical interlocking.

There are essentially two ways of overcoming the surface energy

differencesof glass and metal so that good bonding can occur. The

first is by saturating the glass with the oxide of the substrate metal.

King, Tripp, and Duckworth26 found that porcelain frit applied to an

iron substrate and fired in an argon atmosphere produced little or no

wetting of the iron by the enamel. However, if the saturation limit

of 43% FeO was added to the porcelain composition, excellent wetting

(and bonding) of the iron was observed under identical conditions. As

the iron oxide content of the porcelain was reduced to 40% FeO, the

wetting became poor and no adherence of the porcelain to metal was

observed. The impracticability of adding the saturation limit of metal

oxide to every glass to be enameled to metal is obvious and would not

be possible in cases where the glass compositions are rigidly established.

Fortunately, the other means of producing better wetting between

molten glass and metal is much easier to accomplish and is ideal for

most situations. It involves simply oxidizing the metal substrate so

that the molten glass comes in contact with an oxide rather than a

metallic surface. As the similar structure of glass and metal oxide

gives rise to similar surface energies, good wetting of the glass on

metal usually occurs. Excellent glass-to-metal bonding is now possible

if other conditions have been properly controlled.










Theories of Adherence

There have been numerous studies in past years on glass metal

bonding, but there is still a lack of agreement among researchers as

to the exact nature of the bond formed when a glass coating is applied

to a metal substrate. For example, some studies have shown that

adherence is caused by reactions which are chemical in nature, while

other investigations have demonstrated that adherence is primarily a

function of mechanical factors such as glass-metal interlocking. Much

of the published work on glass-metal bonding has been concerned with

27,28
understanding and improving the enameling of glass coatings on steel.2728

Other prominent areas of investigation are the bonding of dental porcelain
29-31
to metals (especially gold), and the improvement of glass-metal
32
seals for electronic devices.3

Many of the older studies (e.g., 1930-1950) on glass-metal bonding

tended to emphasize the importance of the mechanical interlocking of the

glass to the metal surface as the principal cause of adherence. In

porcelain enameling of steel, for example, major emphasis was placed on

the proper roughening of the metal surface prior to the application of

the glass frit. A comprehensive investigation by the National Bureau

of Standards in 1953 was in agreement with earlier studies which con-

cluded that mechanical factors best explained the observed glass-metal
33
adherence. In that study the character of the interfacial roughness

and the measured adherence between glass and metal was correlated with

both the density of the anchor points of the glass at the metal surface

and the specific surface area of the metal. Their conclusion that










adherence is primarily mechanical in nature was based upon results

which showed that specific surface area (or the area available for

chemical bonding) is much less important than the increased mechanical

interlocking provided by a higher anchor point density.

The surface roughness of a metal can be increased by such tech-

niques as grit blasting, or it can be produced on a microscopic scale

by the phenomenon of electrogalvanic corrosion. The increase of sur-

face roughness in the porcelain enameling of steel by electrogalvanic
34,35
corrosion has been observed by a number of investigators.3435

Roughening of the metal surface has been observed when the porcelain

ground coat contains NiO or CoO. The metals of these oxides precipitate

from the glass layer next to the steel forming a short circuited local

cell with iron as the anode. The equations for a typical reaction involv-
34
ing CoO can be written as3


Fe + CoO + FeO + Co
2+ 2-
2 Co + 02 Co2 + 2 02
2
Co2 + 2e -> CoO

Fe 2e -* Fe2+


The current flows from the iron to the cobalt at the interface and back

to the iron completing the circuit. Once these cells are established

they continue during firing in air because there is sufficient oxygen

and anodic iron available. This reaction causes iron to go continually
2+
into solution in the glass (as Fe ) with a corresponding roughness of

the surface and anchoring of the glass into the newly formed holes.










The same mechanism which is responsible for electrogalvanic corro-

sion can also cause the precipitation of metal dendrites in the glass.

These dendrites can occur as isolated specks in the glass or they can
36
extend into the glass while connected to the metal surface. In the

latter case, they can act as anchors since they project into the glass

phases. However, the effect that these dendrites have on glass-metal

bonding has not been established, and some researchers feel that they
26
do little to increase adherence. On the negative side, in some cases

the presence of a precipitated metal at the interface could act as a

parting layer between the glass substrate, thus causing a loss of

adherence.

In recent years an increasing number of investigators have con-

cluded that chemical bonding is the primary cause of adherence between

glass and metal. Good adherence is caused by chemical bonds (i.e.,

ionic, covalent, or metallic) which are established at the glass-metal

interface rather than the much weaker Van der Waals bonds. One way to

characterize the nature of chemical bonding is to represent it as the

presence of a continuous electronic structure across the glass-metal

interface. To have metallic bonding and a continuous electronic struc-

ture, metal ions must be present in the glass, oxide layer, and metal

substrate.

There are several ways of establishing a condition of chemical

bonding between glass and metal assuming good wetting of the substrate

has occurred. One necessary requirement is that the interfacial area

be in equilibrium. For equilibrium to be present, one condition is










that the glass interfacial layer must be saturated with the lowest

valent oxide of the metal substrate. For example, in the porcelain-

steel system this would mean that FeO should be present rather than

Fe203

Another way of establishing the condition of equilibrium at the

interface is to have at least a monolayer of oxide between the metal

and glass. This is the ideal situation to have if the temperature

cycle and other processing conditions can be controlled so that the

molten glass just dissolves the oxide down to the last molecular layer.

In porcelain enameling of steel, however, it is very difficult to

terminate the firing cycle so that a monolayer of oxide is left between

glass and metal. Too little time at temperature leaves a thick oxide

layer (and bonding is usually weak between iron and its oxide), while

too much time results in the entire oxide layer being dissolved in the

glass.
37
Borom and Pask37 found that if a glass-metal interface is held at a

high temperature after the oxide layer has completely dissolved, dif-

fusion of the metal oxide away from the interface occurs. Thus the

equilibrium condition at the interface is destroyed because there is

no longer a saturation of the metal oxide in the layer of glass next

to the metal surface. The effects of this loss of saturation are in

agreement with the study of King et al.26 which found that all adher-

ence between glass and metal was lost if the percentage of FeO in

sodium disilicate glass was reduced from 43% to 40%. Hoge et al.38

also found that maximum adherence occurs when the glass at the interface










is saturated with the oxide of the lowest valent cation of the substrate

metal. It appears that saturation of the metal oxide is necessary only

in the glass layer next to the metal substrate, and if this condition is

altered by diffusion of the metal ion (or oxide) away from the interface

then bonding strength decreases.

The measurement of contact angles of glass on metal in some studies

has demonstrated the importance of maintaining saturation of the metal
39
oxide. Cline et al. observed that the contact angle of sodium

silicate glass on iron decreased from 550 to 200 when an oxide layer was

formed (Fe304), but increased back to 550 when this oxide layer was

completely dissolved. Adams and Pask40 showed that the glass must be

saturated by the oxide of the substrate metal (not just any oxide), as

increasing the FeO content in silicate glass to the saturation point

did not have any effect on the contact angle or adherence of the glass

on a platinum substrate.

The other means of establishing chemical equilibrium at the inter-

face is by setting up proper conditions so that redox reactions occur

between the metal substrate and some metal oxide in the glass. This

is often necessary in porcelain enameling of steel, since after the

glass layer forms no more atmospheric oxygen can get to the interface

to oxidize the metal substrate. If heating is continued past the point

where the oxide layer is completely dissolved, diffusion of the metal

oxide away from the interface leads to a loss of equilibrium and thus

a replacement of chemical bonds with Van der Waals bonds.41 If oxygen

can be supplied by the reduction of an oxide in the glass by a suitable










redox reaction, oxidation of the metal substrate can continue without

the need of atmospheric oxygen and interfacial saturation can be main-

tained.

Thus the key to maintaining chemical equilibrium by interfacial

saturation is to either stop the dissolution of the substrate oxide

layer before it is completely dissolved by the glass, or to have some

metal oxides (called adherence oxides) present in the glass which can

be reduced to give free oxygen. An example of these adherence oxides

is the NiO and CoO present in the ground coat of commercial porcelain

enamel.



Bioglass Coating Methods

There are several reasons for the selection of the immersion

method as the preferred technique for coating Vitallium with Bioglass.

The frit enameling method of obtaining a glass coating is unacceptable

because the relatively long time at elevated temperature required to

cause softening and flow of the glass frit results in diffusion of

metal ions completely through the thin glass coating. Also, one of the

objectives of this research is to optimize the coating of the wrought

Vitallium FHS alloy as it has potentially more uses as an orthopaedic

or structural implant material than does the cast Vitallium alloy. The

wrought alloy cannot be heated over 6500C for any significant time if

its high mechanical properties are to be retained. The same maximum

temperature applies to the cast Vitallium alloy as well, as higher

temperatures cause excessive carbide precipitation and a corresponding










loss of ductility in this alloy. This eliminated the frit enameling

method of coating either the cast or wrought Vitallium alloys since

a temperature in the range of 10000C is required for Bioglass to

soften and flow.

The flame spray coating method has been used with some success

in past coating work, but specimens coated by this technique typically

show a greater number of voids and bubbles at the Bioglass-metal inter-

face as compared to specimens coated by the immersion method. The

sputter coating of glass on metal by ion beam bombardment is a tech-

nique which is still in the development stage and requires equipment

which is not commonly available. The coating of Bioglass on metal,

ceramic, and polymer substrates by this technique is currently taking

place at NASA-Lewis Research Center, but a major research effort may

be required to optimize the coating variables so that a satisfactory

coating is obtained. This method will only be practical for obtaining

relatively thin coatings, and the coating of complex shapes containing

internal or shielded surfaces is not possible.

Although the immersion method appears to be the most satisfactory

coating technique for the Bioglass-Vitallium system, there are a number

of reasons for conducting a concurrent study of coating by the frit

enameling method. The most valuable information to be gained by such

a study is a better understanding of the nature of the glass-metal inter-

face formed. The two coating techniques are quite different in such

factors as time of interface at coating temperature, for example, and

the direct comparison of the metal-glass interfacial characteristics










and resulting bond strengths observed may aid in understanding the

relative importance of such factors. Also, a correlation of the

results from the Bioglass-Vitallium system with the much studied

porcelain-steel and dental porcelain-metal systems produced by the

same coating method can be obtained.



Summary

There are a number of reactions that occur in the porcelain

enameling of steel-which cause increased adherence of the glass to

the metal. Increased surface roughness and thus the potential for

more mechanical interlocking is caused by the reduction of certain

oxides in the glass which have an oxidation affinity which is lower

than that of the substrate metal. From the standpoint of chemical

adherence, reducible oxides in the glass are also required so as to

maintain the interfacial saturation of the glass with the oxide of

the metal substrate.

The standard free energy of formation of metal oxides can be

used to determine which oxides in the glass will be reduced when in

contact with the metal substrate at a given temperature. Thermody-
42
namic data42 for selected oxides at three temperatures are given in

Table 3 in order to show the change in oxidation/reduction potential

across the glass-metal temperature gradient. The lowest temperature

(7000C) is slightly higher than the average preheat temperature of

the metal, while the highest (1300C) is close to the temperature of

the molten glass. During immersion the cooler metal substrate is













Table 3

Free Energy of Formation of Metal Oxides


Energy (K cal/mole 02
Reaction 700 1000 1300


2 Ca + 02

4 Li + 02

4/3 Al + 02

Si + 02

4 Na + 02

4/3 Cr + 02

4/5 P + 02

2 Fe + 02

2/3 Mo + 02

2 Co + 02

2 Ni + 02

2 Cu + 02



C + 02

2 C + 02


- 2 CaO

- 2 Li20

- 2/3 Al203

- Si02

+ 2 Na20

S2/3 Cr203

- 2/5 P205

- 2 FeO

+ 2/3 MoO3

+ 2 CoO

- 2 NiO

+ 2 CuO


CO2

2 CO


-255

-226

-218

-166

-134

-138

- 98

- 95

- 94

- 78

- 74

- 39


- 95

- 94


-242

-207

-202

-153

-105

-125

- 87

- 87

- 83

- 67

- 60

- 31



- 95

-107


-226

-188

-185

-140

---

-112

- 70

- 78

- 69

- 56

- 46

- 27



- 95

-119










heated by the molten glass, drawing some heat away from the surround-

ing layer of glass. An intermediate temperature such as 1000C may

best represent the actual condition at which the given reactions

will occur.

Oxides with higher negative free energy of formation are more

stable and will not be reduced when in contact with any metal lower

on the chart. A metal higher on the chart will reduce a lower oxide

to base metal if favorable thermodynamic conditions are present.

For example, in the firing of a commercial porcelain ground coat at

8700C, it is obvious from the chart that CoO and NiO will be reduced

when in contact with the steel substrate.

For the Bioglass-Vitallium system, however, the glass does not

contain any oxide which can be reduced by cobalt (the major element

in Vitallium). Cobalt is below calcium, sodium, phosphorous and

silicon in its affinity for oxygen so reduction of any of the oxides

of these metals is not possible in the range of temperatures present

during the coating process. If the metal substrate was pure cobalt,

such mechanisms as electrogalvonic corrosion (causing increased sur-

face roughness) would not take place. Vitallium does contain a

significant amount of chromium (25-30%), and this metal has the

potential for reducing two components in the glass under proper

conditions. From Table 3 it can be seen that P205 is the most

likely oxide to be reduced by chromium, with Na20 being the other

possibility. Vitallium also contains a small amount of molybdenum,

but its oxidation potential is below that of any component in the

glass.










The addition of an oxide such as CuO to the basic Bioglass compo-

sition would make available an oxide in the glass which can be reduced

by either the cobalt or chromium of the substrate metal. However,

modifications in the chemical composition of Bioglass are generally

not acceptable because there is only a narrow range of compositions

in which bonding of the glass to bone will take place. Even minor

chemical modifications would require extensive animal testing to

recertify the altered material.

Although the bulk of the experimental work is concerned with

optimizing the immersion process for the coating of wrought Vitallium

with Bioglass of the standard 52S4.6 composition, it is desirable to

include the testing of a modified glass composition by a conventional

frit enameling procedure in order to gain more insight into the rela-

tive importance of such factors as increased surface roughness on

adherence. Some specific questions which can be answered by this

approach are:

1. Can interfacial roughening occur by reduction of certain
oxides in the standard Bioglass by the chromium in the
metal substrate, or does it take an oxide which is below
both cobalt and chromium in free energy of oxide forma-
tion?

2. Does the long time at temperature of the frit-enameling
method cause significantly more interfacial roughening
than the immersion method with its short contact time?

3. Does increased interfacial roughness increase the glass-
metal bond strength?

4. Do conditions necessary for chemical equilibrium to be
established (e.g., a glass layer saturated with metal
oxide at the interface of the substrate) vary depending
on the coating method?
















CHAPTER III
EXPERIMENTAL PROCEDURES



Laboratory Procedures, Equipment, and Materials

The Bioglass used for most of the experimental work in this study

was a special high purity glass of the composition 52S4.6. An analysis

of this glass is given in Table 2. Some of the initial work for this

study was accomplished using 52S4.6 glass which contained a higher level

of impurities and could readily be distinguished from the purer, color-

less glass by its slight greenish color. The purity of the Bioglass had

no effect on glass-to-metal bonding, and the only concern with using a

slightly impure glass would be the effect of the trace metallic impuri-

ties on the formation of the Bioglass-bone bond.

In the initial stages of this research study, several types of cru-

cibles were used in order to determine which type was most suitable for

containing molten Bioglass for long periods of time. For test purposes,

Bioglass of the 52S4.6 composition was brought to 13500C simultaneously

in alumina, magnesia, zirconia, and fused silica crucibles and held at

this temperature for periods of time as long as four days. Upon examina-

tion, each of these crucible types showed evidence of having been dis-

solved to some extent by the Bioglass. The fused silica crucibles were

initially chosen as most acceptable because of 1) the relative low dis-

solution of this type by the glass and 2) the presence of slightly higher










silica in the Bioglass composition should have a minimal effect on

bone-Bioglass bonding. Alumina crucibles are the least desirable as

even a small amount of alumina in the Bioglass has been shown to be

9
detrimental to its ability to form a chemical bond with bone. The

effects of ZrO2 and MgO on the bonding of Bioglass to bone are not

known at this time. Although silica crucibles were found to be sat-

isfactory, platinum crucibles were chosen for all of the immersion

process optimization studies because of their ability to resist thermal

shock and their much lower dissolution rate. The only effect that small

amounts of platinum (e.g., 0.005%) was determined to have on Bioglass

is that it gives the glass the ability to crystallize at a lower temper-

ature. In fact, platinum in amounts of less than 0.01% by weight was

added to several experimental Bioglass compositions in order to obtain

crystallization of the glass at 650C rather than the 8000C temperature

normally required.

Melting of the Bioglass stock was accomplished by heating to 1300C

to 1360C in an electric furnace with the glass allowed to remain at

temperature for at least 60 minutes to allow bubbles in the melt to

escape before the coating of the metal substrates. The crucible was

normally kept covered to minimize the vaporization of sodium from the

melt.

The composition of the cast Vitallium which was used for all the

immersion process optimization work is given in Table 1. The standard

substrates used in this study were 45 mm long, 6 mm diameter Vitallium

rods. These specimens were sawed from 150 mm high purity Vitallium

castings made in air using the lost wax process.










Oxidation of Vitallium substrates selected to be air-oxidized

was accomplished by the use of an electric furnace. The test speci-

mens were suspended by a Vitallium wire into the hot zone of the

furnace with a platinum/Pt-Rd thermocouple used to measure the temp-

erature at that location. A Jelcraft HT furnace* was utilized for

those specimens which were oxidized in a controlled oxygen pressure.

This type of a furnace is typically used in dental laboratory prac-

tice to fire porcelain-to-metal crowns, and is capable of providing

a vacuum of 30 inches of mercury or better (lower than 1 torr pres-

sure).

Attempts to degas the Vitallium substrates prior to immersion

were carried out in a MRC bell jar set-up** with a diffusion pump
-5
capable of obtaining pressures of lower than 10- torr. Vitallium

rod specimens to be degassed or heat treated were placed in a 3 inch

long ceramic boat and nichrome wire ribbon wound around the boat

from end to end to provide uniform heating. A powerstat was regu-

lated so that a maximum of 10 amps flowed through the wire during

heating. A Micro optical pyrometer*** was used to monitor the temp-

erature of the ceramic boat. Nine hundred degrees centigrade (9000C)

was the maximum temperature obtained with this set-up, with a vacuum
-5
of better than 10-5 torr being easily reached for degassing purposes.




*J. F. Jelenko and Co., New Rochelle, NY
**MRC Manufacturing Corp., Orangeburg, NY
***Pyrometer Instruments Co., Bergenfield, NJ










Thermal Expansion Measurements

Knowledge of the thermal expansion coefficients of Vitallium and

Bioglass is important in estimating the magnitude of the interfacial

stress which occur during cooling due to the expansion mismatch of the

two materials. An Orton Automatic Recording Dilatometer* was used to

obtain the thermal expansion coefficients of the glass and metal. The

standard 2 inch specimens used for test purposes were obtained by

casting 6 inch long, inch diameter glass rods in a graphite mold. A

2 inch section of the casting free of bubbles was sawed and polished

down to exactly 2.00 inches. Each glass specimen to be measured was

placed in the dilatometer between two platinum contacts. The specimen

was heated at a prescribed rate and the increase in length with temp-

erature was plotted on a chart.

A silica correction factor can be included in the plot of the

temperature-expansion relation in order to compensate for the thermal

expansion of the glass specimen holder. In order to check the cali-

bration of the dilatometer, a Lucalox (alumina) bar was first tested

on an identical Orton dilatometer at General Electric in Cleveland.

The same specimen was then run in Gainesville using the silica correc-

tion factor in one case and without this factor in another case.

Comparison of the results of these tests showed that a correction

factor was necessary, but one smaller in magnitude than the normal

machine correction factor.




*Edward Orton, Ceramic Foundation, Columbus, OH










A 2-inch Vitallium test specimen was prepared so that a direct

comparison of glass and metal thermal expansion coefficients could

be made. A section was cut from a standard cast Vitallium rod and

the ends of the two pieces were polished until the length was exactly

2.00 inches. This specimen was run using a procedure identical to

the one used for measuring the thermal expansion of the glass speci-

mens. Direct comparison was obtained by placing the plot of one

temperature-expansion curve over the curve for the other material,

with the main region of interest being from 500*C down to room temp-

erature.



Microscopes and Specimen Preparation Techniques

Both reflection and transmitting optical microscopes were used

for the examination of sectioned and polished Bioglass-coated Vitallium

specimens. All specimens prepared for examination using an Olympus*

reflection microscope were first mounted in PMMA, then polished using

180, 320, 400, and 600 SiC paper followed by 6 pm and 1 pm diamond

paste. A Nikon** transmitting microscope was used to examine the

thinly sliced Bioglass-Vitallium implants surrounded by bone.

Thebone-implant specimens were prepared by first drying in a series

of alcohol solutions, then embedded in polymethylmethacrylate and

sliced into 300 pm sections using a watering saw with a low concentra-

tion diamond blade. Each section was ground to 150 pm using a precision

swivel-head grinder. The thin sections were then cemented to a glass




*Vanox, Olympus Optical Co., Tokyo, Japan
**Biphot, Nikon, Nippon Kogaku, K.K., Tokyo, Japan










slide and ground to 50 pm. A polychromatic stain was used to prepare

the tissue for microscopic observation.

A JEOL* scanning electron microscope was used to obtain high-

magnification, high-resolution photomicrographs of oxidized metal

surfaces and metal and glass interfacial fracture surfaces. Speci-

mens examined by the SEM were in an unpolished condition and were

coated with approximately 50 A of gold-palladium by vacuum deposi-

tion. Specimens were fixed to the holder by silver conductive paint.

Typical excitation voltage was 25 KV with the SEM image formed by

secondary electrons. A 100 um aperture was used to get better

depth of field at magnifications of 1000 or less.



Compositional Analysis Techniques

Auger electron spectroscopy (AES) has been found to be a valuable

tool for surface analysis of materials because the limited escape depth

of the Auger electron restricts the analysis to only the surface layer

(50 X or less). Auger electron spectroscopy is also useful for detec-

tion of elements of low atomic number (e.g., oxygen). In recent studies,

AES has been used in conjunction with ion milling to determine the

variation in composition with the depth from the surface for Bioglass

and other materials.4346

Auger electron spectroscopy measurements in this study were made

using a Physical Electronics Thin Film Analyzer** in a residual vacuum




*JM-35C, JEOL, Ltd., Tokyo, Japan
**CMA 10-155, Physical Electronics Industries, Inc., Edina, MN.










-8
pressure of 2 x 10 torr. An electron beam of 3 KeV with a diameter

of about 600 pm (defocused condition) was used. Total beam current

was measured using a Faraday cage. Sputter profiling (ion milling)

was accomplished using 2 KeV argon ions with the chamber back filled
-5
to 5 x 10- torr.

Vitallium specimens analyzed by AES to determine oxide layer

composition were 5 mm-thick slices taken from the test rods and sawed

on one end to allow clamps to hold the specimen in place on the

carousel. Most of the Bioglass specimens examined by AES were the

curved glass coatings which had been stripped off the metal rods

during shear testing. These small glass pieces were attached to metal

tabs on the carousel by the use of silver conductive paint. A partial

coating of silver paint was also applied to the extremities of the top

surface of the glass to minimize charging during electron bombardment.

A plot of the derivitive of the number of counts versus the elec-

tron energies was obtained for each time period as the electron beam

scanned the specimen (as in Fig. 1). Each plot was obtained for a dif-

ferent time of ion milling, and this time was correlated to depth from

the surface by using 50 A per minute as an estimate of the sputtering

rate. Characteristic peaks of each element of interest were selected

(Fig. 1) and measured to determine the occurrence of the particular

element in the area analyzed by the beam. A normalization factor

was calculated for each element so that peak height could be converted

into a number which could then be compared directly with similar
























UofF AES
li IIII I

Co





C 0 Cr




















Date: Sample: IB Time: min. Ep= PtP= Tc=

Electron Energy leVI


Typical AES analysis of Vitallium
selected for measurement.


surface showing peaks


Figure 1.










numbers obtained for other elements in order to get relative occurrence

of each. The volume fraction ratio of Co to Cr was determined by

converting the ratio of the weight percentages (65/28) into the ratio

of the atomic percentages (2.63). The ratio of selected peaks of Co

and Cr obtained under equivalent conditions was found to be 34.6/15.5

or 2.23. Therefore each Co measurement, for example, could be normal-

ized by multiplying by 2.63/2.23 or 1.18. Because of differences in

specimen placement and beam angle, peak heights could not be compared

directly from one specimen to another without first normalizing the

data.

An electron microprobe* (EMP) was used to determine the extent of

the diffusion of ions across the metal-glass interface. The microprobe

has generally a better quantitative capability than AES, but is limited

in its ability to analyze points close together (i.e., less than 5 pm).

Also, the escape depth for X-ray is much larger than that for the Auger

electrons, so EMP is not as useful for the analysis of surfaces.

Specimens to be analyzed by the EMP were first polished to 600

grit SiC, then to 1 pm diamond paste. The polished specimens were

coated with approximately 100 A of carbon by vacuum evaporation and

electronically connected to aluminum discs by the use of silver con-

ductive paint. The coated specimens were placed in a stage inside

the EMP vacuum chamber which was moved perpendicular to the electron

beam at a specified rate of 20 pm per minute during analysis. An




*Electron Probe X-ray Microanalyzer, Model MS-64, Acton Labs.,
Acton, MA









-7
electron beam diameter of 1 pm, a beam current of 10- amps, and a

volt differential of 20 KV were typical operating conditions. An

argon-methane flow proportional detector with various crystals (e.g.,

quartz, mica, LiF) was used to measure radiation intensities. These

intensities were represented as peaks and plotted as in Fig. 2 as

counts per second with a set chart speed of 20 Vm/min.

Normally the microprobe can analyze points no closer than 5 pm

apart because of beam size and reflection characteristics. This is

a problem as the interfacial zone of Bioglass-Vitallium specimens

coated by the quick immersion process is typically less than 5 Pm

wide. A technique commonly employed in the study of oxide layers

on semiconductors was used to prepare several specimens so that more

readings could be taken in the interfacial zone. Selected specimens

were polished through the glass at an angle of 10 until the metal

was contacted, thus increasing the surface of the oxide layer and

other areas of interest in the interfacial zone. Using calculations

based on the curvature of the metal away from the polished surface,

the actual distance of each point of analysis from the true interface

was determined.

















100




l ,Ca Interface
S80


Si Co

S60 Cr




GLASS METAL
4 40


LU
Nj




rCr

I I a I
0 20 40 60 80 10

DISTANCE FROM REFERENCE POINT (vm)



Figure 2. Typical EMP analysis of a Vitallium specimen
immersion-coated with Bioglass.
















CHAPTER IV
BOND STRENGTH TESTING



Nature of Bond Testing

A quantitative determination of the interfacial strength should

be a key factor in any study which seeks to investigate and improve

glass-metal bonding. However, in many investigations of metal-glass

adherence only qualitative-type tests are used. In reviewing the

literature it quickly becomes obvious that there is no universally

accepted testing method for measuring interfacial strength between a

metal substrate and a glass coating. Various researchers appear to

have developed tests which best suit their purposes, but often their

results cannot be duplicated by others using similar tests. The

large scatter of values reported in many adherence tests may be due

to the complex nature of the glass-metal bond and indicates that

measurement of true interfacial strength is difficult and is influ-

enced by extraneous factors such as porosity at the interface.

There are several types of products or devices in which strong

glass-to-metal bonding is of major importance. Three of the most

important of these are commercial porcelain enamels, glass-to-metal

seals in the electronics industry, and porcelain-metal crowns in

dentistry. Adherence tests developed in each of these areas will be

discussed in order to analyze relative strengths and weaknesses and










to determine which type of a test is most suitable for measuring the

bond strength of Bioglass-coated-Vitallium specimens prepared by the

immersion method.



Porcelain Enamel Tests

There are a number of ASTM tests which have been developed to

judge porcelain adherence to metal:47


1. ASTM C313-59 describes a method for testing the adherence
of porcelain enamel and ceramic coatings to sheet metal.
A hydraulic jack is used to push a coated sheet metal piece
into a stationary steel ball at a controlled rate causing
deformation of the coated test specimen. Measurements are
made on the deformed area by an electronic probe to deter-
mine size of deformed area and integrity of remaining coat-
ing. This method gives no direct measurement of actual
bond strength and is not applicable for metals over 2 mm
thick.

2. ASTM C409-60 describes a method for testing the torsion
resistance of porcelain-enameled iron or steel. Metal speci-
mens bent to a 90 angle and coated on both sides with porce-
lain are twisted on a special machine until the glass first
chips. The angle of twist necessary to chip off the porce-
lain coating is a measure of adherence when compared to
similarly tested specimens.

3. ASTM C385-58 measures the thermal shock resistance of porce-
lain-enameled utensiles. Coated metal specimens are quenched
from a selected temperature into water repeatedly until frac-
ture of the coating occurs. This procedure is repeated at
various temperatures, and the number of cycles necessary to
produce fracture of the glass coating is compared to similarly
tested specimens.

4. ASTM C633-69 describes a test for flame sprayed coated metal
but it can also be used for porcelain enameling. In this
test the coated metal piece is attached on both sides to
blocks on a tensile testing machine. The composite is loaded
at a controlled rate and the load required to break the por-
celain-metal bond is divided by the cross sectional area to
give a measure of the interfacial strength.










One of the tests most commonly used to judge the adherence of

porcelain enamel to metal is the simple bend test. In this test por-

celain is coated on one side of the metal strip, the coated strip is

bent so that the porcelain side is put in compression, and the bend-

ing force is increased until the porcelain shatters. This test can

also be set up so that the porcelain side is in tension. In either

case the actual force required to fracture the porcelain is not a good

measurement of bond strength, and varies greatly depending on such

factors as the modulus and thickness of the metal strip. The usual

means of evaluating the porcelain-metal bond using this test is to

observe whether the porcelain pieces adhere well to the metal strip after

fracture occurs.

King, Tripp, and Duckworth2 used a falling weight test to make

some of the adherence measurements for their study on porcelain-metal

bonding. The test consisted of dropping a cylindrical hammer with a

inch diameter ball on the end onto the back of a coated metal test

specimen. Adherence was judged by calculation of the fraction of the

enamel which remained on the deformed metal after impact. These same

investigators also used a tensile test similar to that specified in

ASTM C633-69, and were able to get quantitative measurements of bond

strength up to the limit of the epoxy used (about 6000 psi).



Glass-Metal and Ceramic-Metal Seal Testing

Most of the testing of glass-metal seals have been accomplished

by the use of either a tensile test or a peel test. A standard ASTM










test (ASTM F19-64)48 has been developed for testing a glass seal between

ceramic or metal (Fig. 3a). Twentyman4 used a cup and washer test

assembly and provided a tensile force to break the seal (Fig. 3b), while

in a similar test, Floyd50 used hydralic pressure to apply the force to

the seal (Fig. 3c).

All of these tests employ a tensile force normal to the glass-

metal interface, but in many cases measurement of shear strength by a

force parallel to the interface would be more valuable. A case in

point would be the measurement of the force necessary to pull a metal

lead-through wire out of its glass seal.

Peel tests are also used quite frequently to estimate the bond

strength of glass to metal. One such test involves sealing a thin metal

strip to a glass piece, then measuring the force which is necessary to

initiate and/or continue the peeling of the metal from the glass. Cole

and Hynes1 developed an unusual test (Fig. 3d) in which pressure is

applied by plungers which cause the rubber discs to force the sealed

piece outwards. The force necessary to break the seal is taken as a

measure of its bond strength. This test, like some of the others men-

tioned, gives only an indirect measurement of bond strength, and it

useful only for comparison of similar specimens.



Testing of the Porcelain-Metal Crown

Some of the more sophisticated tests for measuring glass-metal

interfacial strength have been developed for testing the integrity of

the dental porcelain-fused-to-metal crown. A high metal-glass bond














T



















GLASS


METAL







METAL CUP


HARDENED STEEL
SLEEVE


SEAL HARDENED
STEEL PIN





Figure 3. Tests of glass-metal seal strength.
(a) ASTM tensile test assembly
(b) Cup and washer tensile test assembly


I -- 29 MM











HYDRAULIC PRESSURE

1I


HARDENED STEEL


GLASS SEAL

-CERAMIC


I-- 20 MM -)


Figure 3. continued


(c) Hydraulic test of seal strength
(d) Metal-glass-ceramic seal test


ALUMINA




GLASS
SEAL










strength is necessary to assure proper functioning of this device when

it is subjected to the complex stresses which occur during mastication.

One of the most commonly used bond strength tests for porcelain-

metal crowns was developed by Shell and Nielsen.52 In this test por-

celain is built up around a 14 guage wire (1.63 mm diameter) to a

depth of 2.5 mm (Fig. 4). The amount of force required to pull the wire

(rod) out of the porcelain is recorded, and the known surface area of

the wire which is covered by the porcelain is used to find the inter-

facial strength (force/unit area). The porcelain-metal bonding area is

kept small so that the interfacial bonds will break before the yield

strength of the wire is reached. One drawback to use of this type of

test is that the porcelain is typically under compression so the gripping

force of the porcelain "collar" may add to the observed bond strength.

Lavine and Custer53 used a bend test which consists of coating

porcelain on a flat metal strip and bending the composite so that the

porcelain side is in tension. The major deficiency in this test is

that the maximum tensile stress in the composite is at the surface of

the porcelain so failure is likely to start there. The glass-metal bond

area is under much less tensile stress as it is near the neutral plane

where the stress is zero, so true interfacial strength is not measured

by this test.
54 55
Anthony et al.54 and Leone and Fairhurst55 modified the Shell-

Nielsen test by removing porcelain from the end of the rod (wire),

thus eliminating a source of error arising from differential shearing










stresses. These investigators also used a coated-strip bend test

similar to that of Lavine and Custer to observe the fracturing of the

coating in tension.

Knap and Ryge56 used a standard, threaded metallurgical tensile

bar coated with porcelain in a test of porcelain-metal bonding. They

coated porcelain on the center of the bar, machined off the excess

porcelain so that a cylindrical diameter remained, and loaded the

coated specimen in tension until the porcelain fractured. The load

at fracture was taken to be the point at which circumferential cracks

first ringed the specimen. This was not an exacting test, as cracks

sometimes started near the threads and around the ends of the coated

area, and many specimens had to be discarded. Also, reported values

(Table 4) were much too high to be true interfacial strength measure-

ments, as the modulus and strength of the metal bar contributed to

the final test values.

Sced and McLean57 developed a bond strength test to overcome

some of the problems inherent in the Shell-Nielsen and other tests.

A conical-ended test piece is coated with porcelain and pulled in

tension until fracture (Fig. 5a). The 90 cone allows the angle of

fracture in the porcelain to approximate the direction of maximum

shear stress, with interfacial strength being the load divided by

the porcelain-metal area.

Kelly, Asgar, and O'Brien58 developed a tensile test for measur-

ing porcelain-metal adherence (Fig. 5b), but found that bond strengths

measured by this test were only half of those of the Shell-Nielsen
















METAL ROD


- -GINGIVAL PORCELAIN


OPAQUE PORCELAIN



.. RUBBER CUSHION


Figure 4. Shell-Nielsen shear test.

















FIXED -







MOVEABLE


RUBBER SLEEVE

PORCELAIN

METAL


DENTINE PORCELAIN


OPAQUE


CAST METAL SHAFT


10' TAPER


I APPLIED TESTING LOAD


Figure 5. Dental porcelain-metal bond strength tests.
(a) Sced and McLean's test.
(b) Kelly, Asgar, and O'Brien's test.










test due to the primarily tensile mode of failure (Table 4). Nally59

measured the tensile strength of porcelain-metal bonds by a rupture

test in which two small metal cylinders were joined end-to-end by

baking a layer of porcelain in between. However, an interfacial

strength could not be measured as all fractures occurred in the por-

celain, not at the porcelain-metal bond. Johnson et al.60 used a

similar tensile test in which two rods of cast gold were held in

axial alignment and joined by porcelain. The problem with this type

of test is keeping the metal rods in perfect alignment during firing,

as misalignment contributes to the large scatter commonly observed in

the data.



Suitability of Tests for Measuring Bioglass-
Vitallium Bond Strength

As can be seen from a review of the literature, there are a

number of different methods used to estimate the interfacial strength

between a glass and metal. Bond strength results of tests made on

similar systems (e.g., porcelain-to-gold) often vary widely between

the different testing methods. This is due partly to the fact that

these tests are actually measuring different parameters. For example,

it makes a great deal of difference whether the interface is put in

tension, compression, shear, or torsion during testing. If the

applied load is parallel to the interface, then interfacial roughness

is an important factor (e.g., the glass "peaks" may be sheared off

easily). If a force normal to the interface is applied, the glass-

metal junction is in tension with a separating force tending to



















IrQ \
If1 aN
+1 -.


0 C -4
C'4 ,-


0 0

-l


O' Um o0 C'M :r --T %0 Cf) Lf C4 1-4 Ln C0
cvn cMi eN c' cn as a% o rP, -T m c14
(C'4 r-4







CC
C H M T
rd Cc rj H H H
wO CO 4O d O C O ca o C C-4
) (3 QOUU 0 (w a) a) 0 ) a) 0) bo -H
SU u W U U U U U U U C
p w 4 00 0 W p w p w w o r,--I





00 Co 60 u Z 00 N bfl 00 Br 4-- 00 -,


U)

(v




0

Ur
o




0
r-l


0


0






C

-4







0(


'-4
co
~0
4-1 4-1 4-1 4-1
0 3 3 0 :j
-4 0 0 0 0
i-4
U) c a) c J <
-4 0 -4 r-4 -4 -
p- 4- 'A
(0 **d 0. 0. 0. 0.

4-1 Q)
C 0 4 9- -
0 .4 -4H 4 -
U i :3 :3 : :o
w ns S & S S


co





r-4 1- -4
%D0 -4 eo


-4 r- 0 mO --
a -o LrI r4 '- _4



3 o) 1 4 co H D >0

4' 0 4- 3-o I C


SC4 C E0 3w -I
U CU d (B 1 1 0




00 o 0 0 3 r
CH 0. L) 0 r-l >


)-4, > n 0 S4r4 0


u






1-4


9-K
4-1
3


(U

0
E-i
K


I


CY


'0
o


1-4 $
0 I
vt U)

S0(U
*1n 'o




1 C












-4
CM C CO

<^\





10
rl o



a C C

-4 co ai
'- 0' U
-4 ^
SO T3



0~ tO
'.0 Z
o L) C










overcome the mechanical interlocking. If the test introduces too

high a level of tensile stress in the glass coating, the glass will

fracture before the glass-metal interface fails.

A number of the most commonly used tests from the literature

are given in Table 4 along with reported values of glass-metal bond

strengths of various composites. The great variability in the nature

of the test methods is revealed by comparing bond strength results

obtained from similar systems. For example, average bond strengths
2 2
for gold-porcelain range from 16 Kg/cm to 2348 Kg/cm2, a factor of

almost 150. Also, the range of scatter in the data is quite large

in many cases, as is commonly observed in the testing of brittle

materials.

Some investigators have attempted to develop a test which gives

the highest possible bond strength for a given system. This is often

accomplished by negating the effect of flaws such as microcracks or

porosity which occur at the interface or throughout the glass. This

approach is not always desirable as one of the primary objectives of

the test should be to evaluate the strength of the glass-metal inter-

face under conditions which are similar to those occurring in actual

service. As the presence of flaws has a pronounced effect on the

failure of composites and brittle materials, their effect on bond

strength should be evaluated by the test and not minimized except in

cases where fracture tends to occur in the glass coating rather than

at the interface.

The use of the immersion technique as a coating method greatly

restricts the type of test which can be used to give a quantitative










measurement of glass-metal bond strength. In reviewing the available

adherence tests to evaluate their suitability for testing immersion-

coated Bioglass-Vitallium specimens, the following observations can

be made:

1. A bend test which puts the glass layer in either tension or
compression is not a quantitative test for the determination
of bond strength and is not satisfactory for any type of
coated specimen.

2. The joining of two metal rods by holding them in alignment
during immersion is impractical as the entire length of one
rod and the alignment fixture will be covered with glass.
Also the extensive machining required to remove excess glass
would introduce additional flaws into the glass coating.

3. The threaded tensile bar (as described by Knap and Ryge)56
could be immersed in Bioglass with one end coated with a
substance (e.g., colloidal graphite) which would not allow
the glass to wet the metal. However, the test method itself
is undesirable because it does not measure interfacial
strength. In this test the strength of the substrate metal
is of primary importance, which accounts for the fact that
bond strengths obtained by this method are a factor of 10
higher than other adherence tests.

4. The use of a tensile test such as described by King et al.2
would require the machining or grinding of the excess glass
off the immersed end of the metal rod, and the use of an
epoxy which would bond strongly to both metal and glass.
Even if a successful means of fastening (e.g., epoxy) is
used, this test measures only the bond strength for a load
applied normal to the interface and not the potentially more
severe case in which the load is applied parallel to the
interface.

5. A pull out shear test such as the one developed by Shell and
Nielsen is one of the few testing methods which can be adapted
for making quantitative bond strength measurements on immer-
sion coated specimens using a load parallel to the interface.
A minimum amount of machining would be required to prepare
immersion coated specimens for testing by this method or a
similar technique.










The Push Thru Shear Test

A "push-thru" shear test similar to the one used by Shell and

Nielsen was developed in order that a quantitative evaluation of the

glass-to-metal bond could be made. To prepare specimens suitable for

testing, Vitallium rods 45 mm long and 6 mm in diameter are coated

to approximately 2/3 of the total length (30 mm) with Bioglass. The

glass is removed from one end with a diamond saw and grinding tool

so that a bare metal section is left (Fig. 6). Enough glass is removed

so that there is exactly a 25 mm coated section with a flat shoulder

remaining. Precautions must be taken to insure that the ends of the

metal rod are cut perpendicular to the axis, and that the glass shoulder

is exactly flat and perpendicular to the axis of the rod so that the

full measured shear force occurs at the glass-metal interface.

The shear force is provided by an Instron testing machine* used

in the compression mode. The testing fixture consists of 2 steel

cylinders, one of which has a hole through the center which is slightly

greater than 6 mm (the diameter of the coated metal rod). The prepared

test specimen is placed so the flat glass shoulder rests on the surface

of the steel cylinder with the bare end of the rod protruding down into

the shaft. The other section of the test fixture holds the specimen

perpendicular so the Instron forces the two sections to come together

at a rate of 1.25 mm/min. The load required to strip the glass coat

off the metal rod is recorded and divided by the total surface area




*Table Model-TM, Instron Corp., Canton, MA


























I Flat
25mm Shoulder
25\mm


Glass Coat


Steel Test Fixture


Figure 6. The push thru shear test.


Coated
Rod


I-IJ


.c 1il


i


- I










(475 mm2) to obtain the shear stress. Other specimens with one-half

and one-fourth this total surface area are also tested to determine

if similar values of shear strength are obtained for a range of sur-

face areas.

A typical Instron test on a Bioglass-Vitallium specimen produces

a stress-strain curve which is linear from zero until the point of

fracture. This linear stress-strain behavior is typical of a brittle

material such as glass, and indicates the lack of occurrence of such

mechanisms as plastic deformation at the glass-metal interface. As

the load is applied to the specimen, the stress-strain (time) plot

follows a straight line which is representative of the elastic range

of the composite. There is sometimes a slight dip or chatter in the

curve just before fracture, but usually the glass coating is stripped

off the metal almost instantaneously after fracture starts.

Another test which has been used in previous work involves

quenching the glass coated rods from a range of temperatures to deter-

mine the maximum temperature change the glass-metal interface can

endure without fracturing. The numerical shear stress needed to

fracture the bond can be estimated by the following relation: Shear

stress = 1.33 E AaAT where E is the average elastic modulus, Aa is

the difference in thermal expansion coefficients between the metal

and glass, and AT is the temperature change necessary to fracture the

glass at the interface.62 The greater the change in temperature that

the specimen can withstand without fracturing, the higher is the

estimated strength of the glass-metal interface. Results from this






59



test can be compared directly to those from identical specimens

tested by the "push-thru" shear test and a correlation between the

two test methods can be determined.
















CHAPTER V
THE METAL OXIDE-METAL INTERFACE



Introduction

During the initial testing of the bond strength of Bioglass-coated-

Vitallium specimens, the glass was observed to fracture off the metal

leaving a clean, shiny metal surface. Prior to immersion in the molten

glass, the metal surface had been covered with an oxide layer which was

a dull green, gray, or blue color. Examination of the glass pieces

revealed that the surface which had been next to the metal was covered

with a thin gray, metallic layer. These results suggested that the

metal oxide-to-metal bond might be the weakest link in the glass-metal

interface.

The oxidation of the metal substrate to achieve satisfactory wet-

ting of the molten glass is one of the most important factors in

establishing the proper conditions for good bonding between glass and

metal. Variables of the oxidation process (such as temperature and

time of oxidation) can be controlled so that an oxide layer of an

acceptable thickness is formed on the metal. But proper thickness

may not be enough, for in some cases conditions must be controlled so

that the strength of the bond between the metal and its oxide is max-

imized.

The glass-oxide bond is generally strong in a well designed system

as the molten glass absorbs part of the oxide layer as it wets the metal










substrate. However, many metals such as iron form only a weak bond

with their oxide under normal conditions. Therefore, in commercial

porcelain enameling of steel it is desirable to continue heating until

the iron oxide layer is completely dissolved in the molten glass. In

this case there are special mechanisms, such as the reduction of

adherence oxides in the ground coat, which continue the oxidation of

the iron substrate and keep the interface saturated with the metal

oxide. In other cases, however, these mechanisms are not available

and care must be taken to leave a thin oxide layer undissolved by

the glass so that interfacial equilibrium is maintained. These are

the situations in which it is important to have a strong metal-metal

oxide bond for maximum adherence of the glass coating to the metal.



Oxidation of Cobalt-Chromium Alloys

An understanding of the oxidation process of cobalt-chromium alloys

is important in determining if the Bioglass-Vitallium bond strength can

be improved by proper control of the oxide layer. Oxidation of cobalt

alloys containing chromium is similar to that of stainless steels and

nickel-chromium alloys. If these alloys contain sufficient chromium

(15-20%) a protective layer of Cr203 is formed which limits further

oxidation. Thus these alloys are used for a variety of high tempera-

ture applications such as turbine blades for jet engines and as heating

elements.

In cobalt-chromium alloys, CoO usually forms on the metal surface

first during heating because of the higher initial mobility of the Co2+










ion. The oxidation process starts by electrons moving to the surface

of the metal where they ionize the chemisorbed oxygen molecules. This

sets up a condition in which the metal ions can now move to the surface
2-
to meet the 02- ions more freely because of the influence of both

chemical and electrical potential gradients. The defect chemistry of

the alloy is an important part of the oxidation process, as electrons

initially hop between normal lattice sites and electron holes followed

by metallic ion diffusion through the cation vacancies (the cobalt

alloy is a metal-deficient p type semiconductor). Even though the

presence of some high valence impurities as chromium may accelerate

the oxidation rate, Wright and Wood63 found the rate of oxidation of

cobalt-chromium alloys to decrease as the chromium content was increased

to over 15%.

During the oxidation process, the migration of the metal cations

to the surface causes the movement of vacancies in the opposite direc-

tion. These vacancies tend to stop and coalesce at the oxide scale-

metal interface so that voids are formed which may be a major cause of

the weak bonding between the oxide and metal. Thus thicker oxide

layers which have been oxidized for longer times are more likely to be

poorly bonded to the metal.

The scale (oxide layer) of a cobalt-chromium alloy is predomi-

nately Cr203 if the alloy contains at least 15% chromium. Elements

such as chromium, aluminum, and silicon tend to be selectively oxidized

because of their high affinity for oxygen. However, because the initial
2+ 3+
diffusion of Co2+ is higher than Cr3+, the outer oxide layer of cobalt-

chromium alloys typically consists of CoO with some spinel particles










(CoCr20 ). If there is sufficient chromium in the alloy (e.g., 20%),

islands of Cr203 formed initially will eventually develop into a con-

tinuous layer. This "healing" or protective layer of Cr203 slows

down the diffusion of metal and oxygen ions so the oxidation rate is

greatly reduced after its formation.64



Comparison of Cobalt-Chromium with Other Alloys

As cobalt-chromium, iron-chromium (stainless steel), and nickel-

chromium are often used in similar applications, a comparison of the

oxidation behavior of each may be useful in predicting which should

form the strongest bond with its oxide, and thus have more potential

for forming a strong glass-metal composite.

Unalloyed cobalt oxidizes at a rate of 25 times that of nickel.65

For cobalt alloys containing sufficient chromium, the growth rate of

CoO is reduced but is still higher than that of NiO or Fe203 in similar

alloys. This means that an outer layer of unprotective CoO is formed

faster and vacancy diffusion to the scale-metal interface is greater,

potentially causing more extensive void formation and reduced oxide-to-

metal bonding in a cobalt-based alloy.

The alloy interdiffusion coefficient determines how rapidly the

chromium (which is the preferentially oxidizing element) can be supplied

to any surface or developing healing layer. The cobalt-chromium system

has a lower alloy interdiffusion coefficient than the nickel-chromium

system, thus the healing Cr203 layer takes longer to form and the cobalt-

chromium alloy may oxidize more extensively before its formation. Also










it is more difficult to form the protective Cr203 layer on cobalt-

chromium alloys because the growth rate (and hence encroachment rate)

of CoO is higher than that of NiO or Fe20 66

The oxygen solubility and diffusivity in an alloy determine its

internal oxidation characteristics, with a high value causing more

internal oxidation. Iron-chromium alloys, because of their lower oxygen

solubility, tend to form a protective Cr203 layer closer to the surface

than cobalt-chromium or nickel-chromium alloys. Thus, the oxide layer

on cobalt-chromium and nickel-chromium alloys is generally thicker than

that formed on stainless steel under the same conditions. Thicker

oxide layers generally cause more spelling of the oxide during cooling,

but Wood66 found that a nickel-chromium alloy with a slightly thicker

oxide layer had better scale adherence than cobalt-chromium or iron-

chromium alloys because a more irregular and interlocking alloy-oxide

interface was obtained under identical conditions.

It appears that a strong bond between cobalt-chromium and its

oxide is more difficult to achieve than between nickel-chromium and

iron-chromium and their oxides because of the values of such variables

as bare metal oxidation rate, alloy interdiffusion coefficient, and

oxygen solubility. This means that oxidation conditions must be more

tightly controlled for cobalt-chromium alloys in order to get a satis-

factory degree of bonding between the metal substrate and its oxide

prior to coating with glass. There is little direct comparison avail-

able as to which of these three alloy systems forms the best bond with

glass or ceramic coatings. In one instance, however, McLean67 found










that nickel-chromium alloys formed a strong bond with dental porcelain,

while cobalt-chromium alloys coated with porcelain failed at the metal-

metal oxide interface.



Changing the Nature of the Metal-Metal Oxide Bond

There are a number of variables which can play an important role in

determining the nature of the bond formed between the metal and its oxide

layer. The optimization of each of the factors which can be varied for

the Vitallium oxidation process may be important in achieving the ulti-

mate goal of a strong glass-metal bond.

The addition of other elements to cobalt-chromium alloys has been

shown to increase oxidation resistance, enhance the formation of a pro-

tective layer, and increase oxide-to-metal bonding. Kumar and Douglas68

found that the addition of silicon (e.g., 4%) to alloys containing

chromium caused the formation of a more protective scale but imparted

brittleness to the alloy. Wood and Shott69 found that the addition of

aluminum to cobalt-chromium alloys formed an Al203 layer which was more

protective than Cr203 because of less voltilization at higher tempera-

tures. Other elements with lower oxygen affinity have a lesser effect on

the nature of the oxide layer formed. In cobalt-chromium alloys, the

presence of molybdenum in amounts up to 10% (Vitallium has 5.5% Mo) has

been shown to have little effect on the oxidation behavior.70

The most promising means of strengthening the oxide-metal bond in

cobalt-Chromium alloys is by addition of small amounts of rare earth ele-

ments. Stringer and Wright71 added 3% Y203 (by volume) to a cobalt-

chromium alloy and observed a marked decrease in oxide layer growth and an









72
increase in scale adherence. Allam et al. added hafnium and HfO2 to

a similar alloy and found that the adherence of the oxide to the metal

was greatly increased and was superior to alloys containing yitrium
73
and cerium additions. Wright et al. added 1% ThO2 to a nickel-

chromium alloy and found that increased scale adherence was due partly

to the formation of "keys" of rare earth oxides which anchored the

scale to the metal. It has been observed by various researchers that

the improved adhesion of oxide layers on cobalt-chromium and similar

alloys by the addition of as little as 0.1% of a rare earth element or

oxide is due to 1) a slower oxidation rate, 2) keying on internal

oxides, 3) a convoluted alloy-oxide interface, 4) the increased mechan-

ical properties of the alloy-scale composite, 5) smaller oxide grain

size and 6) the ability of the internal oxide particles to absorb in-

ward flowing vacancies (so no voids are formed).

Cold working of a cobalt-chromium alloy before oxidation can

assist in the formation of a protective oxide layer by increasing the

alloy interdiffusion coefficient so that the Cr203 layer forms more

quickly. Also, the cold-worked alloy is a better sink for vacancies

so void formation at the oxide-metal interface is reduced and increased

adherence of the scale results. On the negative side, however, spal-

ling of the oxide layer may occur more readily because of the higher

residual stresses in the alloy after the cold working operation.66

Reduction of the oxygen pressure during the oxidation process

tends to form a more adherent scale by favoring the formation of the

Cr203 layer. Because of the lower external oxygen pressure, the










oxygen flux into the alloy is reduced whereas the flux of the oxidizing

element to the surface remains unchanged. This tends to cause the for-

mation of a compact protective layer on the surface rather than an

internal oxide which only later coalesces to form the healing layer.

Thus, it may be advantageous to form the oxide layer in a low oxygen

environment to establish a strong protective layer before heating the

alloy to an air atmosphere. Wood66 found that an alloy which was

brought to temperature in a partial vacuum, cooled, and then reheated

in air, oxidized very little because a thin adherent Cr203 layer was
74
initially established. Eoer and Meier observed that specimens

oxidized at a low oxygen pressure developed smoother, more uniform

scales than identical specimens oxidized at normal atmosphere pressure.

Of the three principal ways of increasing oxide adherence, the

use of a reduced oxygen pressure to form a tight, protective Cr20

layer appears to be the most desirable for Vitallium. A typical

sequence would be to heat the Vitallium substrate in a low oxygen

atmosphere, cool to room temperature, then preheat in air or a con-

trolled atmosphere prior to immersion in Bioglass.

The amount of mechanical work in the Vitallium specimen cannot

be altered because the alloy is either in the wrought form for high

strength applications or has been cast to a precision shape for such

uses as dental crowns. However, operations such as grit blasting in-

duce a certain degree of cold work to the surface layer of the metal,

and thus direct comparison can be made between the oxide layer forma-

tion on a grit blasted specimen as compared to an electropolished or

untreated specimen.










Improvement of oxide adherence to Vitallium by major alloy modifi-

cations is not feasible at this time. Vitallium has been certified in

extensive clinical tests and has been used for many years for medical

and dental applications. Any significant change in its composition

would require a recertification procedure, which usually involves years

of animal and human tests. A minor modification, such as the addition

of less than 1% of a rare earth element, might require only minimal

tests for recertification. As this approach appears to be most promis-

ing for major strengthening of the oxide-metal bond, future studies for

the development of new Vitallium alloys containing these elements are

recommended. However, the major objective of this research is to

optimize the bond between Bioglass and Vitallium without changing the

composition of either, so any alloy modifications will be left for

future work.



Analysis of the Oxide Layer on Vitallium

Both scanning electron microscopy (SEM) and Auger electron spec-

troscopy (AES) were utilized to characterize the surface of the Vitallium

after oxidation. Scanning electron microscopy was used to visually

determine the effect of the pre-immersion oxidation treatment on the

topography and nature of the substrate surface. Figure 7 shows a grit

blasted specimen a) before oxidation, b) after oxidation at 6000C for

10 min., and c) after oxidation at 800C for 3 hrs. There is little

discernable difference at 1000X between the unoxidized specimen and

the one oxidized for only 10 min. at 600C. The magnitude of the

oxide layer in Fig. 7b is much less than the peak-to-valley height of

















(a)












(b)










"(c)













Figure 7. Scanning electron micrographs (1000X) of grit
blasted Vitallium surfaces.
(a) Unoxidized
(b) Oxidized at 6000C for 10 min.
(c) Oxidized at 8000C for 3 hrs.










the surface profile resulting from the grit blasting operation. How-

ever, after 3 hrs. at 8000C, the surface features are partly obscured

by the thick oxide layer which formed.

Auger electron spectroscopy is used to characterize the oxide

layer profile of Vitallium specimens which had been oxidized in air

or a partial vacuum. Information regarding the depth profile of metal

ions in the oxide layer is necessary for correlation with AES analysis

of the glass-metal interface after the oxide layer is dissolved in the

glass. A description of the AES technique including operating condi-

tions and sample preparation is given in Chapter III.

The variation of chromium, cobalt, and oxygen with time of ion

milling is shown in Fig. 8 for a Vitallium specimen which had been

oxidized in air at 600*C for 10 min. A typical peak for each element

of interest (see Fig. 1) was selected, measured, and plotted versus

time of ion milling. Tables 5 and 6 list the measured and normalized

peak heights for an air-oxidized and a partial vacuum-oxidized specimen,

respectively. The ion milling of each specimen was continued until the

unoxidized alloy was reached. The depth of each data point from the

surface can be estimated by using 50 A/min. as the rate of ion milling.

Based on this milling rate, the thickness of the oxide layer on the

air-oxidized specimen is about 0.35 pm.

A comparison of the AES analysis of a Vitallium specimen oxidized

in air at 6000C for 10 min. to that of a similar specimen oxidized in

a partial vacuum (about 1 torr pressure) for the same time and tempera-

ture is given in Fig. 9. The following observations can be made from

the normalized data:






























20r


Oxygen


Cobalt
^_ --


/
/


" *.. /
.. /
1..
I.'*...
/


Chromium
........ ................................


/
/
/


4 8 12 16 20 24 28 32 36 40 4448 52 56 60 64 68 72
Time of Ion Milling in Minutes


Auger electron spectrographic analysis of the surface
of a Vitallium specimen oxidized at 650C for 10 min.
in air.


Figure 8.


i I I I I I I I I


-










Table 5

Auger Electron Spectroscopy of Vitallium
Specimen Oxidized in Air


Peak Heights in mm, Time of Argon Milling in Min.
Time 0 Cr Co Cr (Normalized)

0-2 11.4 1.35 6.75 4.59

3.4-6.1 15.6 2.3 0.6 7.83

8.4-10.4 16.85 2.9 0.5 9.86

10.7-13.3 17.2 3.2 0.25 10.89

16.2-18.4 19.2 1.6 0.7 10.72

18.7-20.4 17.9 3.1 1.4 10.54

20.8-22.2 16.5 2.9 2.4 9.86

23.0-24.8 15.3 2.55 3.6 8.67

25.5-27.0 12.9 2.2 5.3 7.48

27.5-29.0 11.0 1.85 6.5 6.29

29.7-31.3 9.2 1.75 7.85 5.95

32.5-34.0 7.9 1.6 9.2 5.44

34.5-36.0 7.1 1.5 9.9 5.1

39.0-41.0 5.8 1.3 11.2 4.42

45.5-47.0 5.2 1.4 11.95 4.76

50.0-51.5 4.8 1.5 12.3 5.1

56.5-58.0 4.35 1.5 12.6 5.1

68.0-69.5 4.2 1.6 12.7 5.44










Table 6

Auger Electron Spectroscopy of Vitallium Specimen
Oxidized in a Partial Vacuum


Peak Height in mm, Time of Ar Milling in Min.
Normalized
Time 0 Cr Co Cr Co 0

0-2.8 14.1 3.1 0.3 10.23 0.35 16.67

3.6-5.9 15.6 3.3 0.3 10.89 0.35 18.43

6.5-9.0 16.7 3.2 0.6 10.56 0.71 19.72

9.5-11.5 15.6 2.8 1.7 9.24 2.01 18.43

14.6-16.0 11.7 2.1 4.3 6.93 5.08 13.82

16.5-18.0 6.6 1.4 7.95 4.62 9.39 7.80

18.2-19.6 4.05 1.2 9.85 4.00 11.64 4.78

20.5-22.0 3.15 1.1 10.65 3.63 12.58 3.72

23.0-24.5 2.9 1.15 10.8 3.79 12.76 3.43

27.5-29.0 2.6 1.2 11.0 3.96 12.99 3.07

38.0-39.5 2.25 1.45 10.9 4.78 12.88 2.66

44.0-45.5 2.35 1.65 10.75 5.44 12.70 2.78
































----- Prtial Vocuum
................... Oxidized Specimen


Air Oxidized Specimen


COBALT


/ COBALT -
/ COBALT /


CHROMIUM



CHROMIUM


12 24 36 48 60 72
TIME OF ION MILLING IN MINUTES


Auger electron spectrographic analyses of the surfaces
of Vitallium specimens oxidized at 650C for 10 min.
1) oxidized in air, and 2) oxidized in a partial
vacuum (1 torr).


20 r


16




E12
E



zr
-x 8
a
4



4


Figure 9.










1. The maximum chromium peaks for both specimens are approxi-
mately equal and significantly greater than the chromium
level of the unoxidized alloy. This indicates that a
healing Cr203 layer formed for both specimens, the only
difference being the distance from the surface at which
this layer is located. The Cr203 layer is much closer to
the surface for the specimen oxidized in a partial vacuum
(5 min. milling or "250 A vs. 15 min. milling or n750 A),
indicating that this layer formed earlier in the oxidation
process than the case of the air-oxidized sample.

2. The relative amount of cobalt at the surface is much higher
for the specimen oxidized in air (16% vs. 3% relative to the
chromium level). This indicates that more CoO had a chance
to form at the surface of the air-oxidized specimen before
the healing Cr203 layer was formed.

3. There is a chromium-depleted region for both specimens
between the chromium-rich oxide layer and the bulk metal.

4. The thickness of the oxidized layer is greater for the speci-
men oxidized in air as more oxygen was able to diffuse into
the specimen during the 10 min. firing time. It took about
70 min. of ion milling to reach bulk metal in the specimen
oxidized in air as compared to only 45 min. for the other.

These observations concur with results from the literature which

show that a chromium-rich oxide is formed during the oxidation of

cobalt-chromium alloys. However, most of the literature studies are

concerned with the oxidation of cobalt-chromium alloys at temperatures

in excess of 10000C for periods of several hours (e.g., conditions

which are likely to be encountered in jet engines). Under these condi-

tions thick oxide layers are formed (on the order of 100 pm or more)

and problems such as spelling of the oxide layer during cooling are

much more severe. A thin oxide layer (less than 2 um) such as is re-

quired for the immersion coating of Vitallium with Bioglass is more

likely to be tightly bonded to the metal than a thick layer because

1) the thermal mismatch between oxide and metal is more easily relieved






76



during cooling, and 2) there is minimal void formation at the metal-

metal oxide surface. It remains to be determined whether special

techniques to improve the bonding of the thin layer of oxide to the

bulk Vitallium alloy are effective in increasing the strength of the

Bioglass-Vitallium bond.
















CHAPTER VI
EFFECT OF COATING PROCESS VARIABLES ON
GLASS-METAL BOND STRENGTH



Immersion Process Variables

One of the objectives of this research program was to determine

what effect such factors as surface roughness have on the strength of

the bond formed between a glass and metal. To accomplish this objective,

the processing variables which are important in the immersion technique

(and similar to those of other coating methods) were optimized so that

maximum bond strength of the coated test specimens was obtained. A

secondary result of this optimization study was the corresponding

increase in the coating adherence and integrity of Bioglass-Vitallium

implants.

The most important coating process variables are:

1. Pre-immersion cleaning cycle for metal substrate.

2. Surface condition of metal.

3. Temperature and time of oxidation of metal.

4. Oxygen pressure during metal oxide layer formation.

5. Temperature of molten glass and time of immersion.

6. Post-immersion annealing cycle.

Each of the above will be discussed in regard to its effect on the

glass metal bond strength.










Data obtained from the push thru shear test were used to evaluate

the effect of each of the immersion process variables on the strength

of the Bioglass-Vitallium bond. Means and standard deviations were

calculated using the bond strength results from each test, but a more

sophisticated statistical analysis was needed in several cases to

determine if the observed differences between the means were significant.

For each test group (e.g., cleaning agents), an analysis of variance

gave an F value and a probability factor (PR > F) which were used to

indicate whether there was a significant difference between any of the

means within the group. A high F value and a low PR > F factor is an

indication that at least one of the means is significantly different

from the others. On the other hand, a low F value and a high proba-

bility factor indicate that there is no difference between any of the

means. Duncan's Multiple Range Test was used to define which means

were significantly different from the others in test groups where the

F value was sufficiently large. In the tables of results given in

this chapter, the means are ranked in order of bond strength and those

within a test group which are not significantly different will be

bracketed together using the same letter (A, B, etc.).



Cleaning Cycle for Metal Substrate

The pre-immersion cleaning cycle used for the Vitallium test

specimens was found to be one of the least important of the variables

listed above. Surface cleanliness is usually of primary importance

when coating one material with another, as impurities on the substrate










surface often cause a reduction of wetting and subsequent decrease of

adherence. It was expected that at least some of the organic matter

on the metal surface would be volatilized during the oxidation/preheat

operation. A test was set up to determine the effectiveness of chem-

ically cleaning the metal surface prior to oxidation. The following

,cleaning agents were used for ultrasonically cleaning the Vitallium

test rods:

1. acetone

2. trichloroethylene

3. No-San* caustic solution used to clean dental crowns

4. Alconox** detergent followed by acetone

All these agents were judged to be satisfactory for cleaning Vitallium

(Table 7), and there was not a significant difference between any of the

mean shear strength values for this test group. Test pieces which were

deliberately handled and oxidized without cleaning showed faint prints

in the thin oxide coating, emphasizing the need for a cleaning cycle of

some kind. Acetone was selected as the cleaning agent for use in the

remainder of the program, due to its observed effectiveness in removing

organic contamination and its less toxic nature as compared to tri-

chloroethylene.



Surface Roughness of Metal

The nature of the surface of the metal substrate before oxidation

has been shown to have an important effect on glass-metal bond strength




*Trio-Dent, Inc., Union, NJ
**Alconox, Inc., New York, NY
















Table 7

Effect of Cleaning Agent on Bond Strength


F = 0.33; PR > F = 0.8053
N Std. Dev.


Mean (Kg/cm2 )


Trichloroethylene 8 6.6 65.04 A

Acetone 8 4.3 64.53 A

Alconox/Acetone 6 3.1 64.00 A

No-San Caustic 5 4.2 62.04 A


Agent











in many cases. It has been shown by a number of researchers3335 that

increased surface roughness is beneficial to bond strength because of

added mechanical interlocking of the glass coating. Others have found

that increased surface roughness decreases bond strength if 1) the

glass is too viscous to penetrate into the small surface cavities,75

or 2) the surface is so rough that the sharp reentrant angles between

glass and metal increase the stress concentration at the bond.58

To determine the effect of surface roughness on shear strength

and coating behavior, as-received cast Vitallium rods were polished

with SiC paper, roughened by grit blasting, or electropolished in

acid to obtain test surfaces of uniform roughness.


a) Silicon carbide paper with a grit size of 180 was used to
obtain a metal surface characterized by coarse, parallel
ridges.

b) A mixture of 50% A1203 particles and 50% glass beads under
a pressure of 90 psi was used to grit blast selected
Vitallium test specimens to obtain a uniformly roughened
surface.

c) A solution of perchloric acid was used to electropolish the
test specimens at 65 amps for 60 sec.


Typical microstructures of SiC-polished grit blasted, and electro-

polished Vitallium specimens showing various degrees of surface rough-

ness are shown in Fig. 10.

A shear strength test of coated rods in which the surfaces were

conditioned in the three ways mentioned above was completed prior to

the main study of the immersion process. Results from Table 8 show

that the grit blasted and SiC polished specimens had a higher shear

strength than electropolished specimens. However, all of the specimens






















































Figure 10.


(a)











(b)











(c)










Scanning electron micrographs (1000X) of Vitallium
surfaces roughened by various techniques:
(a) SiC
(b) Electropolished
(c) Grit blasted


















Table 8

Effect of Surface Roughness on Bond Strength


F = 5.06; PR > F = 0.0173
Roughening Method N Std. Dev. Mean (Kg/cm2)


Grit Blast 8 2.7 64.14 A

Silicon Carbide 7 6.0 62.04 A

Electropolish 7 6.7 55.43 B










in this test had higher shear strengths than the cast rods which had

been coated in the as-received condition, indicating that removal of

the outer layer of the cast metal has an important effect on glass-

metal adherence.

From these results it appears that some degree of surface rough-

ness is beneficial to increasing glass-metal adherence in the Bioglass-

Vitallium system. The grit blasting method of surface roughening was

selected over polishing by hand using SiC paper because of the ease and

quickness of this technique and the more uniform nature of the roughened

surfaces. Except for a few special test specimen, the remaining optimi-

zation of the immersion process was accomplished with the use of grit

blasting as the surface roughening technique.



Temperature and Time of Oxidation of Metal

Two of the most important variables in the establishment of good

glass-metal adherence are the temperature and time of the oxidation of

the metal substrate prior to immersion in molten glass. These factors

are responsible for the thickness of the oxide layer formed during

oxidation of the metal at a given oxygen pressure. Oxide layer thick-

ness is of critical importance, as an excessively thick oxide is usually

weak and failure may occur within the oxide layer itself or between the

metal and oxide. On the other hand, an oxide layer which is too thin

may be completely dissolved by the glass, causing loss of saturation of

metal oxide at the glass-metal interface.










To determine the effect of time and temperature of oxidation on

shear strength, approximately 150 Vitallium rods oxidized at nineteen

different time-temperature combinations (5-15 min., 350-8000C) were

coated and then prepared and tested according to the procedures de-

scribed in Chapter IV. All other process variables remained constant:

a) cleaning cycle--acetone (ultrasonic)

b) surface condition--grit blasted

c) oxygen pressuring during oxidation-atmospheric

d) molten glass temperature--1330C to 1340C

e) immersion time--3 sec.

f) annealing cycle--5000C for 4 hrs., furnace cool

The shear strength results of this controlled test are given in

Table 9. The optimum time-temperature oxidation cycle could not be

established from this test because of the scatter in the test data.

The eight cycles with the best shear strength means were grouped to-

gether as being statistically the same. Any of the time-temperature

combinations from this group should give acceptable results.

A plot of the shear strength vs. time of oxidation is shown in

Fig. 11 for some of the oxidation temperatures used in this test. In

general, for a given temperature in the range of 5000C to 6500C, the

shear strength increases with increasing time of oxidation to a certain

point then decreases. For a temperature of 800C, shear strength values

decrease as oxidation time is increased over 5 min. A 350C oxidation

temperature gives low bond strengths regardless of oxidation time.

These results show that an optimum time of oxidation occurs for

each temperature, thus indicating that a certain thickness of an oxide












Table 9

Effect of Oxidation Time and Temperature on Bond Strength


Oxidation Cycle


500C -

5750C -

620C -

620C -

650C -

540C -

575C -

5000C -

650C -

620C -

540C -

650C -

575C -

5400C -

800C -

500C -

8000C -

800C -

3500C -


15

10

10

15

10

15

15

10

5

5

10

15

5

5

5

5

10

15

10


Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.

Min.


F = 15.97; PR > F = 0.0001
N Std. Dev.


3.6

3.6

2.6

4.1

4.3

5.9

2.7

3.7

2.2

9.1

4.7

5.3

0.9

5.0

8.3

4.7

7.9

5.1

5.7


Mean (Kg/cm )


64.76

64.33

63.13

63.00

62.90

62.35

61.38

59.10

55.34

55.27

54.98

49.09

49.00

48.84

47.26

44.50

38.06

33.58

26.37





























Shear Strength


- -


Shear Strength as a function of
temperature and time of metal
oxidation.




500 C
650*C -




-0 0C
*800"C


350* C


Time of Initial Oxidation, minutes


Figure 11.


Bioglass-Vitallium bond strength as a function of
time and temperature of oxidation.


Kg2

80



60



40



20










layer gives maximum bond strength. An oxidation temperature in the

range of 500C to 650C is the most suitable for Vitallium, with oxida-

tion times being 8 to 15 min. depending on the temperature. The use of

a time-temperature cycle in this range gives about 40% better shear

strength test results than the previously-used standard oxidation cycle

of 800C for 10 min. The use of 800C as an oxidation temperature re-

quires very short times which tends to produce irregular oxidation of

the metal substrate. In addition, temperatures over 6500C cannot be

used for either cast or wrought Vitallium specimens if the desired

mechanical properties are to be maintained. Temperatures below 500C

give low shear strength values due to insufficient oxide layer forma-

tion and excessive temperature mismatch between molten glass and pre-

heated substrate causing the glass to chill before it can completely

wet the substrate.



Temperature of Molten Glass and Time of Immersion

The temperature of the molten glass affects both its fluidity and

the rate at which the metal oxide layer is dissolved. Higher glass

temperatures cause an oxide layer of a given thickness to be dissolved

in a shorter time, thus requiring a change in other process variables

to assure that some oxide remains on the metal. The most important

effect of glass temperature on the coating process, however, is on the

thickness of the glass coating which remains on the metal substrate

after immersion. The glass temperature must be maintained within a

certain range to 1) avoid an excessively thick Bioglass coating










(i.e., greater than 2 mm) on the Vitallium substrate and 2) be below a

temperature at which significant vaporization of Na20 from the glass

melt can occur. The two constraints put the acceptable glass tempera-

ture in the range of 1300C to 13600C.

The results of a study designed to show the effect of glass temp-

erature on shear strength are given in Table 10. There was a range of

coating thickness (0.8 mm to 1.9 mm) observed for rods immersed in

glass at temperatures from 13000C to 13500C, but there was no signifi-

cant difference in the average shear strength of the coated specimens.

The time of immersion has been found to have a greater effect on

bond strength than the temperature of immersion. A test was designed

to determine whether longer immersion times could cause increased bond

strength by the removal of part of the oxide layer on metal substrates

which had been excessively oxidized. The coating of Vitallium rods by

immersion in Bioglass for 15 sec. and 30 sec. as compared to the normal

3 sec. immersion was accomplished on a number of specimens. Shear

strength results (Table 11) show a slight but significant decrease in

strength with increasing time of immersion, perhaps as a result of the

increase in porosity observed at the glass-metal interface. Also, a

longer immersion time is undesirable because of the alteration of the

metal structure caused by subjecting the substrate to the extremely

high temperature of the molten glass (about 1350C).



Post-Immersion Annealing Cycle

An annealing cycle is necessary to relieve the stresses in the

glass coating prior to cooling to room temperature. The glass layer