7-21 Titanium Steinemann

Periodontology 2000, Val. 17, 1998, 7-21 Printed in Denmark. All rights reserved

Comriaht 0 Munksaaard 1998

PERIODONTOLOGY 2000 ISSN 0906-6713

Titanium - the material of choice? SAMUEL G. STEINEMANN

The foreign body

Surgical fracture-treatment implants have the func- tion of a temporary splint. In the form of a screw or a plate or a pin, the implant stabilizes the fracture and supports forces, including those of functional load. Other types of implants, such as endoprosthes- es, are permanent devices. In dental practice, the life expectancy of a reconstruction or an implant should at least be several decades. Yet any implant is a “for- eign body”. The expression has a straightforward and simple meaning. An implant is not like living tissue, where no free electrons exist and where metals, such as essential trace elements, occur in a bound and not in an elemental state. Is this foreign body a chemical, physiological or mechanical “insult” to the living tissue?

Corrosion and soft tissue reaction

In the chemist’s view, corrosion is the visible de- struction of metal. It may cause a structure to rup- ture or lose function, such as by breakage of an im- plant. That was before the 1960s. This aspect is not important for modern metals in surgery because an attack is so small that a material loss is not visible and cannot be weighed. More sensitive electro- chemical methods are needed to measure corrosion. But experiments intended to reproduce the real con- ditions for a surgical implant in tissue are not simple.

The polarization resistance method has been used for in vivo experiments (52) ; the results are shown in Fig. 1. The method requires minimally invasive pro- cedures, and a characteristic feature is that reduction and oxidation reactions on the metal are not forced and run in the open-circuit mode.

The noble metals silver and gold have a resistance to corrosion that is about in the middle of a practical scale. In the logarithmic scale of the polarization re- sistance, the number is 5 to 6, and thus about 2 units, or a factor of 100, lower than high-grade stain-

less steel and titanium. Gold and silver resist oxi- dation in air but resist corrosion much less in sea water and biological fluids. It is common experience that these metals lose polish after some time.

Metals having lower corrosion resistance than sil- ver and gold, such as aluminum, molybdenum and iron, show visible attack or oxidation in living tissue and abnormal cells are observed in their vicinity. This tissue reaction is equivalent to a “chemical in- sult”. In fact, the metals from iron through silver cor- rode so rapidly that the supply and migration of oxy- gen cannot keep up with the consumption of the oxidant (this state is called the diffusion limit) so that the contact tissue to the foreign body is starved of oxygen. This direct effect of corrosion is not speci- fic to the metal of the implant. However, metals are released by the corrosion process and some of them are toxic to cells; a sterile abscess and cell death are

[ohmcm*]

Toxicity

I c o

CoCrNiMo cw316LESRw

Ti alloys Inertness

‘A1 Mo

Fe Sequestration

I I

Tissue reaction

Fig. 1. Data from in vivo corrosion experiments for various metallic elements and for practical alloys. The diagram has the two coordinates tissue reaction as abscissa and polarization and corrosion resistance, as the ordinate. Tissue reaction is grouped according to the three distinct forms of “toxicity”, “sequestration” and “inertness”. Cor- rosion resistance is roughly proportional to the measured polarization resistance (to obtain corrosion current den- sity, use j=0.03 Vlpolarization resistance). The useful scale in chemistry and biology is always the logarithmic one. Thus, differences over the series, such as from cobalt (Co) to silver to titanium (Ti), amount to factors of 1 to 100 to 10,000 in corrosion resistance. CoCrNiMo is wrought cobalt-base alloy, cw316LESR is cold-worked, remelted stainless steel and the titanium alloys are Ti&l,Mo, Tid4V, Ti15Mo.

7

Steinemann

Fig. 2. Micrographs of tissue in contact with implants made of commercially pure titanium (left) and TiAlV alloy (right)

observed. Among the four elements shown in Fig. 1, vanadium is the most toxic and copper the least. This reaction is equivalent to a major “physiological insult”.

Stainless steel, cobalt-base alloy and titanium have similar levels of polarization resistance, but tissue reactions differ. High corrosion resistance is apparently not sufficient to suppress the minor re- jection reaction observed for the two classical alloys that comprise the cell-toxic nickel and cobalt as es- sential components. Another example is the Ti&14V alloy, a standardized metallic biomaterial. Four per- cent of toxic vanadium suffice to elicit a prominent foreign body reaction (Fig. 2). These observations suggest an alloy rule (53): avoid toxic components! Corrosion is essentially an atomic process so that, in principle, the interaction for all components of a metal must be considered. On the other hand, path- ologists question the existence of a threshold for lo- cal and especially systemic (allergic) reactions.

The unwanted reaction of the foreign body

The species of metal compounds ingested with nu- trition and passing into the bloodstream, that is, being metabolized, as well as those finally stored in organs and tissue are incompletely known. But most metals (other than alkali) in body fluids and tissue are bound to organic matter and exist in a stable, electrically uncharged form. Fig. 3 suggests that metal release from implants involves a different path; it can be associated with the entry of the metal through a wound and then corrosion of the metal. In that state, the metal is definitely in its elemental form and in a different chemical state than the metal bound in living matter.

Oxidation-reduction reactions (redox reactions) play an important role in bioenergetics: the energy transformations in living organisms. These reactions

involve electrons. Chemical reactions are the basis of metabolic and growth processes; they involve the second elementary particle of chemistry, the proton or hydrogen ion, and the technical term for this is acid-base reactions. The foreign body interferes with these processes of life.

Redox reaction

In water and tissue fluid, corrosion occurs as an elec- trochemical process in which oxidation of the metal is coupled to reduction: an electron gain of electrolyte components. The dominant form for the latter reac- tion in neutral solutions is the oxygen reduction, which yields a hydroxide that precipitates on the metal surface. Other reduction reactions with inor- ganic and organic components of tissue electrolytes are possible. The crucial characteristic is that electron exchange occurs at the metal surface (and only there, but not in the liquid, where electrons do not exist as free entities). It is a foreign body reaction of a chemi- cal kind that can lead to denaturation of the tissue in contact with metallic implants. The method for study- ing these reactions is voltammetry, or a polarization experiment. Interesting systems are electrolytes con- taining complex formers, such as chloride, cyanide and EDTA; gold shows massive reactions for any of these complexes, whereas titanium is fully inert.

Dissolution of the corrosion product

The first-formed reaction products of corrosion are hydroxides, hydrous oxides and oxides, that is, spar- ingly soluble salts, and occasionally complexes (such as halides). These salts can be soluble or not in the tissue fluids (which are aqueous electrolytes) and

Internal fixation

Dental implant a Food 97 \

Metabolism

Foreign body

Fig. 3. Metals released from implants follow another reac- tion path than metals entering metabolism with nutrition

8

Titanium - the material of choice?

MI I

6 m M - .- - pM:

6 n M - E

P M - , , , , , , , , , , , , , - 0 4 7 10 14

nM S

physiological u pH value

Fig. 4. Distribution of hydrolysis products in solutions saturated with respect to nickel hydroxide [Ni(OH),]. The full line is the solubility limit expressed as the total con- centration of nickel. Data for the concentration of nickel in serum (S), in muscle (M), in contact tissue around stainless steel implants (I) and toxicity levels (T) are added at right margin. Sources: Steinemann & Mausli (55) and Steinemann (56).

they can be toxic or not (55, 56). To determine the effects of this corrosion burden, the identity and sta- bility of the hydrolysis products must be considered. A question arises. What is the fate of the unwanted reaction product of corrosion?

The word hydrolysis is applied to chemical reac- tions in which a substance is split or decomposed by water. New forms of precipitates and new species, as bare ions and as charged and uncharged hydroxo- (OH-containing) and 0x0- (0-containing) com- pounds are found in solution. It is common experi- ence that salts and oxides dissolve easily in strong acids and in strong bases and that the solubility of oxides is small for intermediate acidity (pH) values. The mechanism behind this behavior is proton ex- change. In solution, water molecules are attached to metal ions and these aquaions tend to deprotonize. The removal of a proton from the hydration sheath results in a negatively charged hydroxyl, and thus a reduction of electrical charge of the aquaion. The de- protonized species become hydroxo complexes; Fig. 4 and 5 show these reactions in the case of nickel and titanium. Hydrolysis data are best represented as the solubility or distribution curves of the various hydrolysis products. The diagrams have the coordi- nates acidity (pH of the solution) as abscissa and the molar concentration of dissolved and precipitated species as the ordinate, both in logarithmic scale what gives the straight lines.

Nickel. Nickel is a main component of stainless steel, the metal largely used for fracture treatment im- plants. The 2 + oxidation state of this metal is the important one, and in an aqueous environment the hydroxide is the first-formed corrosion product. Its

dominant hydrolysis product under physiological conditions (pH of about 7) is the unhydrolyzed nickel cation, with a concentration of about 1 mM at the limit of hydroxide precipitation. The unwanted reaction product of corrosion is an ion.

In serum, the nickel concentration is about 10 nM, and in human skeletal muscle it is about 3 pM and less than the solubility limit for physiological pH values. The metal concentration in the contact tissue around implants is strongly enhanced and far above the content in normal muscle tissue; the concen- tration is of the order of the toxicity threshold for nickel. The sequestration reaction for stainless steel implants is the consequence.

Titanium. Titanium is a reactive metal; in air and aqueous electrolytes, it forms spontaneously a dense oxide film at its surface. The unwanted reaction product becomes a potent barrier against dissol- ution of the metal.

The constant solubility of titanium dioxide above pH 3 and up to pH 12 suggests that an electroneutral species dominates in solution (Fig. 5) . The core of this dissolved neutral species Ti(OH),(aq) can be pictured as a titanium ion with its four positive charges sur- rounded by four negatively charged hydroxyl mol- ecules. At physiological pH values, the first charged species is the cation Ti(OHI3+ with a concentration of not more than 0.1 nM, which is three orders of magni- tude lower than the concentration of the hydrogen ion, which is always present in solution.

The unwanted reaction product of corrosion is not an ion. This is an important finding because un- charged hydrolysis products have no affinity for re- action with organic molecules. Corrosion of titanium

M i I

P M t , , , , , , , , \ , , , , , { 0 4 7 10 14

physiological U pH value

Fig. 5. Solubility behavior of hydrous titanium dioxide measured in sodium chloride and chlorate electrolytes. The full line is the solubility limit and dashed lines are partial concentrations for the named species. Data for the concentration of titanium in serum (S), in muscle (M) and in contact tissue around implants (I) are added at right margin. Sources: Steinemann & Mausli (55) and Steine- mann (56).

9

Steinemann

becomes in fact no chemical burden, and its inert reaction in tissue is a sign of the basically different chemistry in solution.

In serum, the titanium concentration is about 0.1 pM, and in human skeletal muscle it is about 5 pM. The muscle concentration of titanium equals the upper limit of solubility for the aqueous hydroxide, which is also about the lower limit for precipitation of the solid oxide (about 3 pM). Solution chemistry thus provides a stringent, even simple homeostatic mechanism for the regulation of titanium in tissue: titanium is at saturation in tissue! The concentration of titanium in the contact tissue around implants is about 100 times higher than that of muscle tissue. These high concentrations seem representative for larger and loaded implants and include fretting and wear debris and residues from surface treatment. These metal and oxide particles beyond the solu- bility limit are just deposited in tissue, and retrieval studies give no indication of any adverse reaction. A statement is in order: no case of local or systemic reaction for titanium has been documented.

Complexes in fluids and tissue

The contact between tissue and an implant raises the local metal concentration by orders of magni- tude. The capacity of homeostatic mechanisms can be exceeded. Hydrolysis of the corrosion products means proton exchanges in the tissue electrolyte and pH shifts, which can cause denaturation of organic matter. But binding of metal ions with other ligands competes with hydrolysis and complexes can be formed by substitution reactions. This substitution is a third form of reaction, insidious in nature, of how a foreign body interacts with living matter. The distinction between ions or uncharged species as re- action product of the corrosion becomes important. Cell toxicity is well known for nickel, cobalt, copper and vanadium, whose hydrolysis products are cat- ions and anions respectively. Metals can act as hap- tens: the ion can unite with a protein to form an antigen, and nickel, cobalt and chromium are known allergens (32). Titanium behaves totally different in that no proton exchange and liganding with biologi- cal molecules is possible.

Cell reaction in presence of metals

A small corrosion rate is not a sufficient condition for tissue compatibility; the fate of the unwanted re- action product (of corrosion) must be considered

(18, 55). The reaction products are available and can be taken up by the cells. This may or may not be toxic to the cells. Studying the effect of solid and dis- solved metals on the development of embryonic bone rudiments indicates that there is at least some effect on skeletal development (13). The variety of chemical interactions in tissue near the foreign body and at the artificial interface is great.

Maurer et al. (33) investigated the chemical and biochemical aspects of cell adhesion and cell pro- liferation in the presence of many metals. Reactions of fibroblast and osteoblast cultures were followed for two modes of exposure in three experiments (Fig. 6): (i) fibroblasts cultured on metal discs, (ii) osteo- blasts cultured on a neutral substrate in metal satu- rated media, and (iii) osteoblasts cultured on metal discs. An influence of the physical surface topogra- phy was circumvented by preparing all metal samples and the reference support made of a neutral polymer to the same fine finish. The pure metal samples included in experiments (i) and (iii) were ground under water irrigation and then cleaned ul- trasonically in deionized water and dried in a jet of nitrogen gas. For the culture experiments, the metal discs were rinsed with modified Eagle’s medium be- fore cell seeding. For experiment (ii), the chlorides of tin, aluminum, titanium, zirconium and tantalum were added to water in concentrations somewhat higher than the saturation limit of the oxide in water (4, 5 5 ) . The solutions were titrated to a pH between 6.4 and 7.0 and served to make the supporting media for the cell cultures. W o cell lines were used for the three experiments, one rat fibroblastic line and one rat osteoblastic line, and both were cultured with modified Eagle’s medium as base. Cells were har- vested at day 5 for cell counting.

The results of the three experiments are shown in Fig. 7-9. In all figures, the abscissa is the polarization resistance for the metals that had been measured in a separate in vitro experiment. The polarization re- sistance of a metal is proportional to its corrosion resistance or the reciprocal of the corrosion current density. Metals vary greatly (by a factor of lO ,OOO) , and the logarithmic scale provides a good presen-

I Metal In aqueous On metal

Fig. 6. Cell reactions are studied for two different modes of exposure: the metal is dissolved in the supporting medium and the cells are in direct contact with the solid metal.

10

Titanium - the material of choice?

- - 2 r 0

0.5

= 0.2 8 I 0.1

b .05

lrj > c .-. I

3 0

!A

1 -

- - - -

- Control level -

- 1 2 c 0 E 0.5

= 0.2 8 15 2 0.1

.05

> c I

- 0

8 -

Iv - Detection limit

- I - I 1- - - -, - Control level - - - Ti Zr

- Sn

Ta - Ni I N b I I

- -Detection limit

- F e l -

Ta I -

I I I I I I 3 4 5 6 ?

Polarization resistance, log ohmcmz

.02 ‘ Fig. 7. Results from experiment (i): fibroblasts cultured on metal discs. The abscissa is the logarithm of the polariza- tion resistance of the metal considered and is a good scale for its corrosion resistance. The ordinate is the normalized cell count, also in logarithmic scale. Copper and va- nadium are toxic to fibroblasts, and corroding molyb- denum strongly reduces the proliferation rate. Titanium, niobium, zirconium and tantalum do not interact.

tation (see also the preceding section on corrosion and soft tissue reaction). This scale also provides a good grading, as any metal located below number 5 will show visible attack in chloride containing elec- trolytes and a strong redox reaction, both processes equivalent to a ”chemical insult” in living tissue. The ordinate in the three figures, cell growth ratios, are also drawn in a logarithmic scale for simplicity of interpretation. The cell proliferation rate observed for the inert support and without metal salt is used as the reference.

Fibroblast cells in contact with titanium, niobium, zirconium and tantalum can proliferate but do not in proximity with molybdenum, copper, vanadium (Fig. 7 ) . These are reactions of “inertness” or “sequestration” and “toxicity” on the other hand, as they are found for soft tissue in contact with im- planted metals (Fig. If. Osteoblasts cultured in modified Eagle’s medium with dissolved metals dis- played no inhibition of cell growth for the five metals tested (Fig. 8). The metal concentrations in the cul- ture medium ranged from 0.4 to 6 pM, equal to satu- ration and a maximum possible level of dissolved hydroxo compounds for pH values in the physiologi- cal range. In the experimental series of osteoblasts cultured on pure metal discs, growth inhibition is absent for titanium and zirconium, relatively weak for tin and aluminum, and strong or total for zinc, iron, copper, molybdenum, vanadium, nickel, silver, niobium and tantalum (Fig. 9). One experiment of six with titanium resulted in unusually low cell counts, but two others showed extreme proliferation where the osteoblast layer reached confluence and lifted off the surface.

.02 I I I I I I

3 4 5 6 7 Polarization resistance, log ohmcm2

Fig. 8. Results from experiment (ii): osteoblasts cultured on a neutral substrate in metal-saturated media. The ab- scissa is the logarithm of the polarization resistance of the metal considered and is a good scale for its corrosion resistance. The ordinate is the normalized cell count, also in logarithmic scale. Osteoblasts cultured in modified Eagle’s medium with dissolved metal hydroxides displayed no inhibition of cell proliferation.

3 4 5 6 7 Polarization resistance, log ohmcmZ

.“L

Fig. 9. Results from experiment (iii): osteoblasts cultured on metal discs. The abscissa is the logarithm of the polar- ization resistance of the metal considered and it is a good scale for its corrosion resistance. The ordinate is the nor- malized cell count, also in logarithmic scale. Growth inhi- bition is absent for titanium and zirconium but strong for the corrosion-resistant metals niobium and tantalum. In- hibition is observed for all lesser corrosion resistant metals, such as silver, molybdenum, aluminum, tin, iron, zinc and the toxic metals nickel, vanadium, copper. The compatible metals are titanium and zirconium.

Different modes of exposure, cells in a metal-satu- rated electrolyte or cells in contact with solid metal, elicit different responses. The conditions are not the same. In the first case, the cells are in an environ- ment of completely hydrolyzed metal species at low concentration, what might be termed “weak interac- tion”. Furthermore, it follows from basic principles of chemistry that an interaction would be possible only, and only when, the dissolved species is electri- cally charged: an ion (55). This is the case for only one, perhaps two of the elements considered in Fig. 8. Conversely, the direct contact of the cells with the

11

Steinemann

Fig. 10. Transmission electron micrograph of a nondemin- eralized specimen showing the intimate contact between the mineralized bone matrix (B) and the titanium coating (T). Source: Listgarten et al. (31).

solid metal is sort of a “strong interaction” (Fig. 9). Indeed, direct communication via cellular processes between osseous cells and dental implants has been documented (5 1).

A selective response is found whenever the cells are in contact with the solid metal. Growth inhi- bition is found for fibroblast cells (Fig. 7) and osteo- blast cells (Fig. 9) if the polarization resistance of the metal considered is below a logarithm of about 5 , that is, under conditions of strong corrosion and strong redox activity. The observation of seques- tration and toxicity is then an expected parallel situ- ation for tissue reaction (Fig. 1). But the remaining metals titanium, niobium, zirconium and tantalum elicit different reactions for the two cell lines, even if these elements have a same high corrosion resist- ance. The biochemical interactions differ in the case of niobium and tantalum. Fibroblasts are hardy cells compared to the more differentiated osteoblasts. The inhibition of osteoblast cell growth for any metal other than titanium and zirconium suggests that these elements only have the capacity for osseoin- tegration, that is, a rather distinct property.

Osseointegration

Titanium has the surprising property that it can bind to living tissue and to bone. It is interesting to read Leventhal in 1951 (29):

Bone reaction was studied by the insertion of up to 80 screws into the femora of rats ... At the end

of six weeks, the screws were slightly tighter than when originally put in; at twelve weeks, the screws were more difficult to remove and at the end of sixteen weeks, the screws were so tight that in one specimen the femur was fractured when an attempt was made to remove the screw ... From these studies it would appear that titanium is a metal which may be useful in surgery, be- cause of its strength and its failure to cause tissue reaction. The fact that bone becomes attached to titanium may be a disadvantage in cases where screws or pins are placed temporarily. In the past, the use of some prostheses has not become popu- lar because it has been felt that these would re- main separate from the bone and eventually loosen. Since titanium adheres to bone, it may prove to be an ideal metal for such prostheses.

A new form of prosthesis, tooth root analogues made of titanium, become a reality in the late 1960s. This was a major advance in clinical dental treatment and engineering initiated by Brfinemark and collabor- ators in Goteborg. The word “osseointegration” was coined and defined as “a direct structural and func- tional connection between ordered, living bone and the surface of a load carrying implant” (6).

Structural aspects and cell reactions accompany- ing osseointegration have been studied at all scales, by light-optical histology through high-resolution electron microscopy (1, 20, 30). Listgarten et al. (31) comment on their transmission electron micro- scopic image of the interface (Fig. 10) by saying that “there is no evidence of any space between the met- allic surface and the bone”. This finding suggests the possibility of a direct chemical bond.

Osseointegration allows the efficient stress trans- fer from implant to bone. It implies that no relative motion occurs at the interface, and in case of a chemical bond between the metallic implant and bone, the displacement is restricted to atomic dis- tances.

Torque Pull - off F A ..$

Fig. 11. Forces (F) on a dental implant have axial (A) and transverse (T) components which generate shear and ten- sile stresses at the interface between the metallic implant (M) and the bone (B). The ultimate stresses for the bond between metal and bone are measured as push-out load, as release torque and as pull-off load.

12

Titanium - the material o f choice?

Natural teeth and implants that support a dental prosthesis will generally be loaded by both forces and moments (torque). The mechanical force is a vector quantity and it has a magnitude and a direc- tion (Fig. 11). Mericske-Stern et al. (36-38) made simultaneous force measurements on endosseous implants along three axes and found that transverse forces cannot be neglected. Using the Pythagorean theorem, it is found that transverse components reach 10-50% of the vertical force. The loading of an implant is skewed, with angles up to 30”. In conse- quence, different kinds of forces, shear, tension and compression, will act in the implant-bone interface. To determine the failure limit of the interface, “push- out” tests of cylinders and “release torque” tests of screws in bone are performed to give an interfacial shear strength (Fig. 11). “Pull-off’’ tests, on the other hand give the interfacial tensile strength. Selected examples of such measurements for various healing times are shown in Fig. 12.

Mechanical aspects

An implant in the form of a screw will resist push- out, but this shape requires a bond if it is to resist torsion, which tends to loosen the implant along the threads. The resistance to loosening is measured as a removal torque (Fig. 12, left). It is noted that the figure reports measurements without any normaliza- tion for screw geometry and bone thickness. The re- lease forces show a latency period of 2 to 3 weeks,

followed by a rapid increase of torque and a maxi- mum or leveling out of this torque after about 5 months. This timetable for incorporation of a dental implant is similar to that of bone fracture healing, whereby a short resorption period is followed by bridging a gap with less organized bone and then stabilization through generalized bone remodeling. These processes can be followed by sequential label- ing. Using this technique in the late healing-in phase of a titanium implant, it was found that hard bone formation occurs from the metal surface outwards, a clear indication of total integration.

The removal torque and the push-out force are proportional to the interfacial shear strength. The length of the contact between implant and bone en- ters the formula and in the case of a screw, this true length can be expressed through an effective pitch height leff=1.5 1, the factor being obtained from geo- metric arguments or empirically (17). Representative results from push-out and removal torque tests and different surface preparations are collected in Table 1. At 12 weeks after insertion, the interfacial shear strength ranges from 9 to 19 MPa for cortical bone and from 2 to 6 MPa for trabecular bone. The results clearly indicate that surface roughness and texture is a primary factor in the anchorage of dental implants. For longer times, as displayed in Fig. 12, the maxi- mum interfacial shear strength reaches 24 MPa for machined implants and nearly 50 MPa for special preparations. These ultimate bond strengths are high.

Time after insertion of implant, weeks

I I I

0

TPS and corundum blasted

Time after insertion of implant, weeks

Fig. 12. Shear and tensile resistance of the titanium-bone bond. Left. Raw data for the removal torque of screws (21, 54, 62); average values are shown. Surface preparation: machine cut, titanium plasma-sprayed (TPS), corundum-

blasted-etched (SLA). Right. Measurements of the pull-off force for titanium plasma-sprayed (TPS) and blasted sur- faces (44, 54).

13

Steinemann

I I Table 1. Interfacial shear strength for bone-implant fixation made of commercially pure titanium

Experiment (12 weeks after insertion) Implant Animal, bone removal torque push-out load shear strength References

Machined screw rabbit, corticalb 38N.cm 9.3 MPaa 60 Screw, blasted A1203 25 pm rabbit, corticalb 42 N.cm 10 MPaa 61 Screw, blasted A1203 75 pm rabbit, corticalb 50N.cm 12 MPaa 61 Cylinder, blasted A1203 fine minipig, trabecular 130 N 1.9 MPaC 64 Cylinder, blasted A1203 rough minipig, trabecular 260 N 3.9 MPaC 64 Cylinder, blasted A1203 rough+etched minipig, trabecular 390 N 5.8 MPaC 64 Screw, titanium plasma-coated sheep and dog 200 N . cm 15 MPad 54, 62 Screw, blasted N203 rough+etched sheep, corticale 250 N . cm 19 MPad 62

The thickness of bone is 1.25 mm. The relationship between push-out load P and interfacial shear strength T is P=s T, with surface area s=d x .1=67 mm2, The relationship between removal torque T and interfacial shear strength T is T=c T, with c=80 mm3 per pitch height (1.75 mm). The thickness of bone is 3 mm.

a The relationship between removal torque T and interfacial shear strength T is T=d/2 d x leff. T = C . 7, with c= 19 mm3 per pitch height (0.6 mm).

I Table 2. Interfacial tensile strength for bone-implant fixation made of titanium

Experiment Implant Animal, bone duration pull-off load tensile strength References

Titanium plasma-coated disc, 20 mm* Macacu sp., ulna 30 weeks 76 (30-120) N 2.7 MPa 54 Polished Ti&V disc, 39 mm' dog, cortical 43 weeks 25 (041) N 0.6 MPa 58 Hydroxyapatite-coated disc, 39 mm2 dog, cortical 43weeks 107N 2.7 MPa 58 Commercially pure titanium, blasted

and plasma coated rabbit, epiphysis 24 weeks 107 N 1.6-3.1 MPa 44 The relationship between pull-off load P and interfacial tensile strength a is P=s a, where s is surface area.

Interestingly, the pull-out interfacial shear strengths for freshly inserted BrHnemark (Nobel Bi- ocare AB, Goteborg, Sweden) screw implants (7, 8) and small bone fixation screws (49) are less than the above-reported results: 31 MPa and 34-39 MPa re- spectively. These latter figures can represent a "notched shear strength' and a practical limit when acute threads engender high local stresses. Likewise, an implant inserted in the mandibula or maxilla will be subject to similar effects that limit the anchorage. For comparison, human bone has a shear strength of 68 MPa when measured in torsion along the bone axis, but the property falls strongly for other orien- tations (45).

Pull-off tests have the characteristic feature that a strong interaction between the metallic implant and living hard tissue, that is, the adhesion reaction, is observed only beyond about 10 weeks (Fig. 12, right). Perfect immobilization of the test implant is always required. Table 2 collects data for conditions when bonding is completed. For titanium, the interfacial tensile strength is 2.5 MPa on average, and hydroxy- apatite-coated titanium does not fare better. The

figure is of the order of the compressive strength of human trabecular bone (2-10 MPa). Fragments of bone are commonly observed on the fractured test pieces, which indicates a less dense bone in the con- tact zone.

To give an idea how strong this bond is, reference can be made to a classical work in physics. The Mag- deburg half-spheres experiment measured the force necessary to break vacuum; it needs the (negative) pressure of 1 atmosphere, equal to 0.1 MPa. The ad- hesion between two bodies can never exceed this stress if there is empty space or free water present in the bond region. The interfacial tensile strength of the metal-bone bond is 25 times larger than this limit. Thus, one may ask what is the glue?

Material aspects

Corrosion can be the cause of a tissue reaction. An example is gold, which has a strong redox activity in the chloride-containing tissue fluids and whose corrosion current density is not far above a practical limit for visible attack (see earlier). Sequestration of

14

Titanium - the material of choice?

the foreign body is observed in soft tissue, and im- plants made of gold provoke bone resorption (59). Stainless steel, CoCrMo (vitallium) and Ti&14V have much better corrosion resistance; nevertheless, sequestration and a local reaction are observed in soft tissue because of leaching, even in minute quan- tities, of toxic components from the alloys. Bone screws made of stainless steel loosen after some weeks, and screw implants made of vitallium and ti- tanium alloy have reduced removal torque (Table 3). Push-out tests in trabecular bone all show the effect of surface texture, but the enhancement of anchor- age by etching pure titanium is absent for the alloys (compare with Table 1). It is concluded that vitallium and all titanium alloys used for bone surgery have no capacity or a reduced capacity for total integration. A chemical effect of the metal is manifest. On the other hand, zirconium and niobium have interfacial shear strength values not less than that of titanium.

Chemistry of osseointegration

Titanium is a reactive metal. This means that, in air, water or any other electrolyte, an oxide is spon- taneously formed on the surface of the metal (Fig. 13). The native oxide film has a thickness of about 4 nm, or 20 times the interatomic distance. Its mode of growth is specific in that the oxygen ions migrate towards the metal and react with the counter-ion ti-

Metal ion J, J Oxygen ion

0 . 0 0 . 0 0 . 0 . .

Fig. 13. Surface oxide on a metal. In the case of titanium, the oxide grows by the transport of oxygen from the exter- nal surface towards the metal-oxide interface where the oxidation reaction proper takes place.

tanium at the base of the oxide (35). This mechanism is unique for titanium and some other elements of valency IV, such as silicon and zirconium. Ordinarily, both the oxygen anion and the metal cation migrate when a metal undergoes oxidation, or corrosion. The specific mode of oxide growth on titanium has the positive effect that no metal ion will reach the sur- face and be released into the electrolyte.

Titanium oxide is a good insulator even in the form of a very thin surface film, which may be in a crystalline or in a glassy state and which can occlude anionic impurities such as chlorine, fluorine or phosphates (12, 34). This behavior differs from that

Table 3. Interfacial shear strength for bone-implant fixation made of titanium alloys and other materials

Experiment (12 weeks after insertion)

Implant removal push-out shear

Animal, bone torque load strength References Vitallium, machined screw Zirconium, machined screw Niobium, machined screw Ti&14V, machined screw Ti&14V, cylinder, blasted A1203, fine Ti&14V, cylinder, blasted A1203, rough Ti&14V, cylinder, blasted rough+etched Ti&17Nb, cylinder, blasted A1203, fine Ti&17Nb, cylinder, blasted A1203, rough Ti&17Nb, cylinder, blasted rough+etched Hydroxyapatite-coated Ti, cylinder

rabbit, corticalg rabbit, corticalg rabbit, corticalg rabbit, corticalg minipig, trabecular minipig, trabecular minipig, trabecular minipig, trabecular minipig, trabecular minipig, trabecular minipig, trabecular

12 N . cma 26 N . cmb 33 N . cmc 16 N. cmd

120 N 230 N 230 N 110 N 280 N 230 N 830 N

3.4 MPae 6.5 MPae 9.1 MPaf 4.3 MPae 1.8 MPah 3.4 MPah 3.4 MPah 1.6 MPah 4.2 MPah 3.4 MPah

12 MPah

24

26

25

22

64

64

64

64

64

64

64

a The measurement for commercially pure titanium gives 25 N . cm. Commercially pure titanium has 26 N cm. Commercially pure titanium has 25 N cm. Commercially pure titanium has 23 N . cm.

The relationship between removal torque T and interfacial shear strength T is T=d/2 . d n I,#' T = C . T, with (9 c=17 mm? per pitch height (0.6 mm), and ( f ) c=19 mm3 per pitch height (D.6 mm). 8 The thickness of bone is 1.25 mm.

The relationship between push-out load P and interfacial shear stress I[ is P=s .I[, with surface area s=d . B. 1=67 mm'.

15

Steinemann

of the passive film on gold, stainless steel and vitalli- um, which is an electron conductor. The availability of electrons at the outer interface towards the elec- trolyte or living contact tissue implies that redox pro- cesses, that is, electron exchange processes, can pro- ceed. Or, this is one mechanism leading to denatura- tion of macromolecules and sequestration at a larger scale. Gold, stainless steel and cobalt-chromium al- loy show this response; titanium does not.

Most metal oxides are hydroxylated at room tem- perature and when water or its vapor has access to the surface (5, 39). In fact, a water molecule adsorb- ed on a bare metal site loses a proton, which jumps to the neighboring surface oxide ion, leaving a hy- droxyl (OH) on the metal ion (Fig. 14). These reac- tions lead to two types of hydroxyl groups. In the case of titanium dioxide, one is bound to one ti- tanium-ion site (terminal OH), and the other to two such sites (bridging OH), and these would be ex- pected to exhibit quite different types of chemical behavior. The bridging groups must be strongly po- larized by the cations and therefore be acidic in character, whereas the terminal OH group can be predominantly basic and exchangeable with other anions. This surface has an amphoteric nature: it has the ability to bind with both acids and bases. The acid groups react, for example, with amines (such as ammonia and methylamine), and basic OH groups

H r-r

H Oxygen ion H H H

Fig. 14. Surface of a metal oxide. Top. Naked surface of oxide. Middle. In the presence of water, the surface metal ions bind water molecules. Bottom. Dissociation and chemisorption of the water molecules leads to an ap- parent monolayer of OH groups, a “hydroxylated surface. Source: Schindler (47).

exchange with phosphates. Further, alcohols react fairly easily with the acidic OH groups to form esters.

The amphoteric or zwitterionic nature of the oxide surface can be weighed by counting the effective positive and negative charges at all surface sites, and this can be measured by potentiometric titration. A characteristic value for which the positive and nega- tive charges at the surface cancel each other is called the zero point of charge. For TiO, in the two struc- ture variants of rutile and anatase zero point of charge =pH 6, that is, near to neutrality (9, 40). For other oxides of interest, A1203 and ZrO,, zero point of charge =pH 9, and for Si02 zero point of charge =pH 2 (39). At physiological pH values, the surface of titanium oxide has no charge, whereas aluminum oxide and zirconium oxide have negative and quartz has positive charges. Zeta potentials, connected to the zero point of charge, have also been measured for natural and synthetic apatites, giving a pH of 6- 7 (50), similar to that of titanium oxide.

Schindler et al. (14, 47, 48, 63) have studied the adsorption of amino acids and some metal ions on crystalline titanium oxide and give a complete pic- ture of how the surface hydroxyls exchange charges, that is, their protons, with the dissolved ligands. These exchanges are controlled by the pH of the sol- vent, that is, the environment. Strong interactions are involved. In fact, the energy calculated from re- action constants derived from adsorption coverage numbers (63) for amino acids amounts to 20-50 kJ/mole, and the energy necessary for thermal de- sorption of glycine on platinum indicates binding energy values of roughly 100 kJ/mole (11). These numbers are characteristic of real chemisorption processes involving the formation of covalent or ionic chemical bonds. The necessary proof that such bonds exist comes from photoelectron diffraction experiments with adsorbed glycine on single-crystal rutile, which show the correct distances and angles among carbon atoms of the amino acid (41).

The spontaneously formed thin surface oxide film on titanium metal is apparently in an amorphous or glassy state (35). But photoelectron spectroscopy in- dicates that the stoichiometry of the crystalline oxide is preserved and that the amorphous oxide film is also hydroxylated at its surface (34). Adsorption and desorption experiments with the oxidized metal by Gold et al. (15, 16) showed a similar amphoteric re- action, as displayed by crystalline oxides (5, 14, 63). In a first series of experiments (Fig. 15), the adsorp- tion of cysteine in a salt solution of variable pH was monitored by X-ray excited photoelectron spec- troscopy. In a second series of experiments (Fig. 16),

16

Titanium - the material of choice?

Ti O2

d 0 I

I

0 4 7 10 pH 14

Fig. 15. Cysteine adsorption on titanium vs pH as deter- mined by photoelectron spectroscopy. Proton exchange reactions on the amino acid occur at pH=2 and =ll. Source: Gold et al. (15).

f f + -

loo- .- P I t

-0.4 0.0 E-E (H*/Oz), V I .o Fig. 16. Desorption of soft tissue residues on titanium after exposure to solutions of variable pH (top) and of variable redox potential (bottom). The thickness of the or- ganic films before the experiments was 6.3-8.7 nm. Posi- tive potential differences correspond to oxidizing electro- lytes. Source: Gold et al. (16).

the adherent tissue on retrieved titanium implants was examined by controlled desorption treatments in solutions of variable pH and of variable oxidation strength (redox potential); again, the organic matter at the surface was monitored by photoelectron spec- troscopy. Adsorption and desorption show a type of mirror response, as expected (compare Fig. 15 with Fig. 16). The proton exchanges on carboxyl and amino groups of the amino acids favor binding in the first experiments, but the opposite, that is, a re- lease of proteinaceous residues in the second series of experiments. The results further indicate a strong influence of the redox reaction on desorption; strong oxidizers such as potassium permanganate, hydro- gen peroxide and chloric acid can remove adherent

organic matter, probably by destroying a primary structure.

The strong adhesion between bone and tissue is quite a puzzle. The structure of a tissue and interac- tion in cohesion occur at rather different scales. In fact, the space between bone trabeculae or vessels in soft tissue is of the order of mm, and the size of cells is of the order of pm, whereas the interaction distance of chemical forces, the Debye length, is only in the nm range. One wonders how the chemistry at the substrate relatively remote from a cell surface “protrudes” through to influence interaction that re- sults in binding.

Organic compounds comprise various groups of atoms that account for the compounds’ character- istic reactions (3). These groups, about 15 in number, are called functional groups. Among them, the rela- tively simple chemical groups such as hydroxyl (OH), carbonyl (CO), carboxyl (COOH), amine (NH,) and sulfonate (S03H) can change the surface properties of organic matter, including rendering these surfaces hydrophobic or hydrophilic, and they can modulate cell adhesion on polymers and on metals (10,27, 57).

The ground substance identified in the near con- tact zone of bone toward titanium, zirconium, ni- obium, tantalum metals (2,23,30) has the properties

+- I 2- I I 20 I O O J Arg . R I ;f+ I ?+ -

OxidexG [ q I Oxidex D 1 1 OxidexR 7 1

1 I I PH 0 7 14

Fig. 17. Affinity for surface complex formation between titanium oxide and the RGD series of amino acids (argi- nine-glycine-aspartic acid). Bar 1. Surface charge of the oxide. Bars 2-4. Charge of glycine, aspartic acid, arginine. Bars 5-7. Binding, as represented by folding the surface charges of the oxide with the charges on the binding sites of the amino acid, for the whole molecule and for the side chain only (cross-hatched). No charged aqueous titanium hydroxide species do exist for pH between 2.5 and 11.2. The signature of charges for the carboxyllamino and for the side group of the amino acid is shown (+ positive, - negative, 0 uncharged). The dissociation constants are from Lehninger (28).

17

Steinemann

Ti02 f - f + 1 Ca (11) Caz+

OxidexCa

OxidexPh 1 1 I I I

PH 0 7 14

Fig. 18. Affinity for surface complex formation between ti- tanium oxide and inorganic calcium and phosphate. Bar 1. Surface charge of the oxide. Bars 2,3. Charge of dissolved Ca(I1) and P (V) species. Bars 4,5. Binding as represented by folding the surface charges on the oxide with the charges of the chemisorbed ligand. No charged aqueous titanium hydroxide species do exist for pH between 2.5 and 11.2. The hydrolysis data are from Pourbaix et al. (43).

of a general “glue” and certainly comprises adhesion proteins having the ability to promote cell adhesion and to interact with matrix molecules and with a for- eign material. A simple electrostatic model of the in- teraction of an adhesion protein with an oxide sur- face having receptor sites will now be given.

Fibronectin has emerged as a prototype adhesion protein, and the identification of the tripeptide se- quence arginine-glycine-aspartic acid (RGD) as the focal domain where cells attach to the surface of the large molecule was a major breakthrough in molecu- lar biology (42, 46). The charge structure of the three amino acids and their apparent dissociation con- stants, that is, the pH values at which protonation and deprotonation of the functional groups occur, are shown in Fig. 17. If the organic molecule adheres to the surface, its charges will interact with the am- photeric oxide in a strictly local manner. A proton is exchanged and electrostatic energy is gained. The mechanism is selective for the functional group of the amino acid and for the surface site of the oxide;

the carboxyl group can only bind with the basic site (terminal OH) and the amino group can only bind with the acidic site (bridging OH), whereas the side chain functional groups will exchange protons with one or the other surface site matching in character. The latter case corresponds to the molecular state with the intact peptide links. This great selectivity of interactions is the consequence of the zwitterionic property of both reactants.

Experiments indicate that titanium oxide has four or five reactive groups of acidic and basic character per nm2 of surface (5). On the other hand, the vol- ume of an amino acid molecule in the Arg-Gly-Asp tripeptide is 0.1 to 0.2 nm3; thus, if these amino acids are spread on a surface, about 4 molecules will cover an area of 1 nm2. This number of molecules matches well with the number of available bonds on the inor- ganic substrate.

It is known from photoelectron spectroscopy that calcium and phosphate ions chemisorb on the ox- ide-covered titanium (19). A binding mechanism can also be modeled by electrostatic interaction (Fig. 18). It is not calcium phosphate that precipitates on the metallic substrate; the 2-valent Ca ion binds to oxy- gen after deprotonation, and the phosphate makes a bidentate link to surface atoms. As in the case of the organic ligand, these selective interactions are the consequence of the zwitterionic behavior of the oxide.

The simple electrostatic force model proved to be a versatile tool in understanding the true mechanism of osseointegration. It turns out that the native oxide on titanium is a receptor of short- and perhaps also long-range interactions for adhesion-promoting fac- tors of living tissue.

Conclusion - the dedicated metal

For modern metals, the surgeon expects full tissue compatibility. Unalloyed titanium is the reference.

Table 4. Mechanical data for a dedicated metal

Yield strength to Notched tensile strength Material Yield strength elastic modulus to (plain) tensile strength

316L stainless steel, cold-worked 730 MPa 0.38% 1.7 Commercially pure titanium, grade 2 330 MPa 0.32% 1.8 Commercially pure titanium grade 4, cold-worked 690 MPa 0.66% 1.6 Titanium alloys ( T a V Ti&Nb) 920 MPa 0.84% 1.4 Cortical bone 0.67%

!8

Titanium - the material of choice?

But that metal has limited mechanical strength. High mechanical properties are needed for structural ef- ficiency in surgical and dental implants. Their vol- ume is in fact restricted by anatomic realities, which require good yield and fatigue strength of the metal. On the other hand, an implant “bridges” the forces in bone: it may reduce the normal physiological level of forces. This should give advantage to a “less-rigid’ bone plate or hip prostheses so as to retain a func- tional load on bone. Frequently this suggestion is as- sociated with a low elastic modulus, comparable to bone. The conjecture is misplaced; the implant must “guide, transport and distribute” forces, and to do this it must be stiff. The useful comparison of prop- erties refers to functional aspects (Table 4). The yield strength has to do with load capacity and possible distortion failure of the implant and associated ele- ments. A second quantity is the elastic modulus, re- lated to the stiffness of the implant. The ratio of these two properties, a dimensionless quantity, is the permissible strain, which has to do with deformation capacity of a mechanical construct. It equals 0.67% for human cortical bone. Differences between ma- terials are substantial. High yield stress and permiss- ible elastic strain do characterize the “forgiving metal” in the clinic; in fact, implants should have spare strength to overcome an unfavorable anatomic situation and lack of cooperation by the patient. The penalties associated with implant failure are always great. Another entry in the table has to do with the safety of the implant. The notch sensitivity is import- ant for design; this parameter must exceed 1.2, to avoid cracks in the shape variations of common im- plants. It is found that the properties of titanium metals are favorable for making surgical implants in- trinsically safe and damage tolerant.

The metal titanium must not be a foreign body in living tissue, at least regarding a chemical and physiological “insult”. For integration, it has unique properties among all metals in the periodic table. The question at the outset of this work can be answered: yes - titanium is the material of choice!

References

1. Albrektsson T, Brlnemark PI, Hansson HA, Lindstrom J. Osseointegrated titanium implants. Acta Orthop Scand

2. Albrektsson T, Hansson HA, Ivarsson B. Interface analysis of titanium and zirconium bone implants. Biomaterials

3. Atkins PW, Beran JA. General chemistry. New York Scien-

1981: 52: 155-170.

1985: 6: 97-101.

tific American Books. 1992: 887-917.

4. Baes CF, Mesmer RE. The hydrolysis of cations. New York John Wiley & Sons, 1976.

5. Boehm Hl? Acidic and basic properties of hydroxylated metal oxide surfaces. Disc Faraday SOC 1971: 52: 264-275.

6. Brinemark PI, Zarb GA, Albrektsson T, ed. Tissue-inte- grated prostheses. Osseo-integration in clinical dentistry. Carol Stream, IL Quintessence Publishers, 1985: 11-76,

7. Brunski JB. Biomechanics of dental implants: future re- search directions. J Dent Educ 1988: 52: 775-787.

8. Brunski JB, Skalak R. Biomechanical considerations. In: Worthington r: Brlnemark PI, ed. Advanced osseointegr- ation surgery: applications in the maxillofacial region. Carol Stream, I L Quintessence Publishers, 1992: 15-39.

9. Cornell RM, Posner AM, Quirk JF? A titrimetric and electro- phoretic investigation of the PZC and the IEP of pigment rutile. J Colloid Interface Sci 1975: 53: 6-13.

10. Curtis ASG, Forrester Tv, Clark F? Substrate hydroxylation and cell adhesion. J Cell Sci 1986: 86: 9-24.

11. Ernst KH, Christmann K. The interaction of glycine with a platinum (111) surface. Surface Sci 1989: 222: 277-310.

12. Frey N, Buchillier T, LC VD, Steinemann SG. Properties of surface oxides on titanium. In: Froes FH, Caplan I, ed. Ti- tanium ’92 Science and Technology. Warrendale, PA Min- erals, Metals & Materials Society, 1993: 2135-2141.

13. Gerber HW, Moosmann A, Steinemann S. Bioactivity of metals - tissue tolerance of soluble or solid metal, tested on organ cultured embryonic bone rudiments. In: Buch- horn GH, Willert HG, ed. Technical principles, design and safety of joint implants. Seattle: Hogrefe & Huber, 1994:

14. Gisler A. Die Adsorption von Aminosauren an Grenzflachen Oxyd-Wasser. Dissertation. Berne: University of Berne, 1981.

15. Gold JM, Schmidt M, Steinemann SG. X P S study of amino acid adsorption to titanium surfaces. Helv Phys Acta 1989: 62: 246-249.

129-143.

248-254.

16.

17.

18.

19.

20.

21.

22.

Gold JM, Schmidt M, Steinemann SG. X P S study of re- trieved titanium and Ti alloy implants. In: Heimke G, Sol- tksz U, Lee AJC, ed. Clinical implant materials. Amsterdam: Elsevier Science Publishers, 1990: 69-74. Gotfredsen K, Nimb L, Hjorting-Hansen E, Jensen JS, Hol- men A. Histomorphometric and removal torque analysis for Ti0,-blasted titanium implants. An experimental study on dogs. Clin Oral Implants Res 1992: 3: 77-84. Gwynn 1010 A€? Cell biology at interfaces. J Mater Sci Mater Med 1994: 5: 357-360. Hanawa T, Asaoka K, Ukai H, Murakami K. Calcium phos- phate precipitation on titanium in electrolyte and its accel- eration by calcium-ion-implantation. In: Kovacs rl Iste- phanous NS, ed. Compatibility of biomedical implants. Corrosion and organic and biological electrochemistry di- visions. Proc Vol. 94-15. Pennington, NJ: The Electrochemi- cal Society, 1994: 126-137. Hillmann G, Donath K. Licht- und elektronenmikroskopis- che Untersuchung zur Biostabilitat dentaler Titanimplant- ate. Z Zahnarztl Implantol 1991: 7: 170-177. Johansson C, Jacobsson M, Albrektsson T. Removal forces for osseointegrated titanium implants. In: de Putter C, de Lange GL, de Groot K, ed. Implant materials in biofunction: advances in biomaterials. Amsterdam: Elsevier Science Publishers, 1988: 8: 87-92. Johansson CB, Albrektsson T, Thomsen P. Removal torques

19

Steinemann

of screw-shaped c.p. titanium and Ti&14V implants in rab- bit bone. In: Heimke G, SoltBsz U, Lee AJC, ed. Clinical im- plant materials: advances in biomaterials. Amsterdam: El- sevier Science Publishers, 1990: 9: 87-92.

23. Johansson CB, Hansson HA, Albrektsson T. Qualitative inter- facial study between bone and tantalum, niobium or com- mercially pure titanium. Biomaterials 1990: 11: 277-280.

24. Johansson CB, Sennerby L, Albrektsson T. A removal torque and histomorphometric study of bone tissue reactions to commercially pure titanium and vitallium implants. Int J Oral Maxillofac Implants 1991: 6: 437-441.

25. Johansson CB, Albrektsson T. A removal torque and histo- morphometric study of commercially pure niobium and ti- tanium implants in rabbit bone. Clin Oral Implants Res 1991: 2: 24-29.

26. Johansson CB, Wennerberg A, Albrektsson T. Quantitative comparison of screw-shaped commercially pure titanium and zirconium implants in rabbit tibia. J Mater Sci Mater Med 1994: 5: 340-344.

27. Kothari S, Hatton PV, Douglas CWI. Protein adsorption to titanium surfaces. J Mater Sci Mater Med 1995: 6: 695-698.

28. Lehninger AL. Biochemistry. New York Worth Publishers, 1975.

29. Leventhal GS. Titanium, a metal for surgery. J Bone Joint Surg 1951: 33A 473.

30. Linder L, Alberktsson T, Brinemark PI, Hansson HA, Ivars- son B, Jonsson U, Lundstrom 1. Electron microscopic analysis of the bone-titanium interface. Acta Orthop Scand

31. Listgarten MA, Buser D, Steinemann SG, Donath K, Lang NF: Weber HP Light and transmission electron microscopy of the intact interfaces between non-submerged titanium- coated epoxy resin implants and bone or gingiva. J Dent Res 1992; 71: 364-371.

32. Luckey TD, Venugopal B. Metal toxicity in mammals. Physiological and chemical basis. New York Plenum Press, 1977: 1: 103-128.

33. Maurer AM, LC VD, Steinemann SG, Guenther H, Bille J. Surface chemistry and cell reactions. 10th Eur Conf Bio- materials. Davos: in press.

34. Mausli PA, Bloch PR, Geret V, Steinemann SG. Surface char- acterisation of titanium and Ti-alloys. In: Christel P, Meun- ier A, Lee AJC, ed. Biological and biomechanical perform- ance of biomaterials. Amsterdam: Elsevier Science Pub- lishers, 1986: 57-62.

35. Mausli PA, Steinemann SG, Simpson J€? Properties of sur- face oxides on titanium and some titanium alloys. In: Lac- ombe P, Tricot R, Beranger G, ed. 6th World Conference on Titanium. Les Ulis: Les editions de physique, 1988: 1759- 1764.

36. Mericske-Stern R, Geering AH, Burgin WB, Graf H. Three- dimensional force measurements on mandibular implants supporting overdentures. Int J Oral Maxillofac Implants 1992: 7: 185-194.

37. Mericske-Stern R, Assal P, Buergin W. Simultaneous force measurements in 3 dimensions on oral endosseous im- plants in vitro and in vivo. Clin Oral Implants Res 1996: 7:

38. Mericske-Stern R, Piotti M, Sirtes G. 3-D in vivo force meas- urements on mandibular implants supporting overden- tures. Clin Oral Implants Res 1996: 7: 387-396.

39. Parfitt GD. The surface of titanium dioxide. Progr Surface Membr Sci 1976: 11: 181-226.

1983: 54: 45-52.

378-386.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

Parks GA. The isoelectric points of solid oxides, solid hy- droxides, and aqueous hydroxo complex systems. Chem Rev 1965: 65: 177-198. Patthey L. Photoelectron diffraction and photoemission studies of some semiconductor and oxide surfaces. PhD thesis. Lausanne: University of Lausanne, 1995. Pierschbacher MD, Ruoslahti E. Cell attachment activity of fibronectin can be duplicated by small synthetic fragments of the molecule. Nature 1984: 309: 30-33. Pourbaix M, de Zoubov N, van Muylder J, Deltombe E, Schmets J, Vanleugenhaghe C, Moussard M, Besson J, Brenet JE Burgers WG, Charlot G, Carrels RM, Hoar TP, Jolas F, Kunz W, Maraghini M, Piontelli R, Schwabe K, Valensi G, van Rysselberche I: Pitman AL. Atlas dkquilibres electro- chimiques. Paris: Gauthier-Villars, 1963. Probster L, Voigt C, Fuhrmann G, Gross VM. Tensile and torsional shear strength of the bone implant interface of titanium implants in the rabbit. J Mater Sci Mater Med

Reilly RT, Burstein AH. The elastic and ultimate properties of compact bone tissue. J Biomech 1975: 8: 393-405. Ruoslahti E, Hayman EG, Pierschbacher MD. Extracellular matrices and cell adhesion. Arteriosclerosis 1985: 5: 581- 594. Schindler PW. Surface complexation. In: Sigel H, ed. Metal ions in biological systems. New York Dekker, 1984: 18: 105- 135. Schindler PW. Co-adsorption of metal ions and organic ligands: formation of ternary surface complexes. In: Hoch- ella ME White AE ed. Mineral-water interface geochemis- try. Reviews in mineralogy. Washington, DC: Mineralogical Society of America, 1990: 23: 281-307. Shelton JC, Loukota RA. Pull-out strength of screws from cortical bone in the maxillofacial region. J Mater Sci Mater Med 1996: 7: 231-235. Somasundaran P, Wang YHC. Surface chemical character- istics and adsorption properties of apatite. In: Misra DN, ed. Adsorption on and surface chemistry of hydroxyapatite. New York Plenum Press, 1984: 129-149. Steflik DE, Sisk AL, Parr GR, Lake RT, Hanes PJ, Berkery DJ, Brewer €? Transmission electron and high-voltage electron microscopy of osteocyte cellular processes extending to the dental implant surface. J Biomed Mater Res 1994: 28: 1095- 1107. Steinemann SG. Corrosion of titanium and titanium alloys for surgical implants. In: Lutjering G, Zwicker U, Bunk W, ed. Proc 5th Int Conf Titanium. Oberursl: Deutsche Ges Metallk, 1985: 1373-1379. Steinemann SG, Perren SM. Titanium alloys as metallic bio- materials. In: Lutjering G, Zwicker U, Bunk W, ed. Proc 5th Int Conf Titanium. Oberursl: Deutsche Ges Metallk, 1985:

Steinemann SG, Eulenberger J, Maeusli PA, Schroeder A. Adhesion of bone to titanium: biological and biomechan- ical performance of biomaterials. In: Christel Meunier A, Lee AJC, ed. Advances in biomaterials. Amsterdam: Elsevier Science Publishers, 1986: 6: 409414. Steinemann SG, Mausli PA. Titanium alloys for surgical im- plants - biocompatibility from physicochemical principles. Proc 6th World Conf Titanium. In: Lacombe F: Tricot R, BC- ranger G, ed. Les Ulis: Les Bditions de physique, 1988: 535- 540. Steinemann SG. Tissue compatibility of metals from

.

1994: 5: 314-319.

1327-1334.

20

Titanium - the material of choice?

physicochemical principles. In: Kovacs P, Istephanous NS, ed. Compatibility of biomedical implants. Corrosion and organic and biological electrochemistry divisions. Proc Vol. 94-15. Pennington, NJ: The Electrochemical Society, 1994:

57. Sukenik CN, Balachander N, Culp LA, Lewandowska K, Merritt K. Modulation of cell adhesion by modification of titanium surfaces with covalently attached self-assembled monolayers. J Biomed Mater Res 1990: 24: 1307-1323.

58. Taylor JL, Brunski JB, Hoshaw SJ, Cochran GVB, Higuchi Kw. Interfacial bond strengths of Ti&14V and hydroxyapa- tite-coated Ti&14V implants in cortical bone. In: Laney WR, Tolman DE, ed. Tissue Integration in Oral, Orthopedic and Maxillofacial Reconstruction. Proc 2nd Intern Congr Tissue Integration. Carol Stream, IL: Quintessence Publishers Co.,

59. Thomsen E: Larsson C, Ericsson LE, Sennerby L, Lausmaa J, Kasemo B. Structure of the interface between rabbit cor- tical bone and implants of gold, zirconium and titanium. J Mater Sci Mater Med 1997: 8: 653-665.

1-13.

1992: 125-132.

60. Wennerberg A, Albrektsson T, Andersson B. An animal study of c.p. titanium screws with different surface topo- graphies. J Mater Sci Mater Med 1995: 6: 302-309.

61. Wennerberg A, Albrektsson T, Lausmaa J. Torque and histo- morphometric evaluation of c.p. titanium screws blasted with 25 and 75 pm sized particles of A1203. J Biomed Mater Res 1996: 30: 251-260.

62. Wilke HJ, Claes L, Steinemann S. The influence of vari- ous titanium surfaces on the interface shear strength be- tween implants and bone. In: Heimke G, Soltesz U, Lee AJC, ed. Clinical implant materials, Advances in Bio- materials, vol. 9. Amsterdam: Elsevier Science Publishers

63. Wirz UE. Die Adsorption proteinogener Aminosauren an TiOz (Anatas). Dissertation University of Berne 1984.

64. Wong M, Eulenberger J, Schenk R, Hunziker E. Effect of surface topology on the osseointegration of implant ma- terials in trabecular bone. J Biomed Mater Res 1995: 29:

1990: 309-314.

1567-1575.

21