The effect of plasma-sprayed calcium phosphate

ceramic coatings on the metal ion release from

porous titanium and cobalt-chromium alloys

Paul Ducheyne and Kevin E. Healy

Dqartrnent of Bioengineering, University of Pennsylvania, Philadelphia, Pennsylvania 29204

Bone tissue ingrowth in porous materials
is enhanced by the deposition of bioactive
calcium phosphate ceramic linings onto
the pore walls. These bioactive coatings
can be deposited using several methods
which yield a variety of coating efficien-
cies and thereby influence the mecha-
nisms and kinetics of ion release from the
metal. We analyzed the effect of plasma-
spraying hydroxyapatite onto titanium
and cobalt-chromium alloys by measuring
the release of Ti, Al, V, Co, and Cr in vitro.
Plasma-sprayed coatings significantly re-
duced the Ti and A1 release from titanium-
based alloy specimens. The tendencies of
release from the cobalt-based specimens
are less pronounced. The data substanti-
ate that neither localized enhanced pas-
sive dissolution of metal ions nor ceramic
shielding of the metal occurs. The Scan-
ning Auger Electron Microprobe Spectro-
scopic data suggest that the dissipation of
thermal and kinetic energy of the ceramic
particle at the time of impact can produce
compositional and structural changes in
the metal surfaces. The resulting effects
are significant for the titanium alloy but
less significant for the Co-Cr alloy system.

INTRODUCTION

A prevailing method of achieving component stability of orthopedic joint

replacement devices is by the use of porous coatings.2It was through the
pioneering efforts of Hulbert and his associates that this concept was cre-
ated. According to estimates it will be used on 500,000 joint prostheses an-
nually by 1990.?Calcified tissue invades the pores provided requirements of
minimum pore size, materials compatibility, initial component stability,
and growth potential of the surrounding tissue are met.4The bone tissue in-
growth produces a mechanical interlock with the surface porosity of the
component and thus provides for component bonding.
Compared to fixation by poly(methylmethacry1ate) bone cement (PMMA),
bonding by tissue ingrowth is not achieved at the time of surgery. During
the intial period after surgery, restricted activity or even immobilization of the
patient is imposed for bone tissue ingrowth to proceed. As a consequence
the concept of a biologically fixed implant creates a dichotomy with the goal
of restoring motion to a diseased joint, since an extended period of activity
restriction may affect joint functionality. Furthermore, irrespective of
proper design, careful patient selection, and correct operative technique,
Journal of Biomedical Materials Research, Vol. 22, 1137-1163 (1988)
0 1988 John Wiley & Sons, Inc. CCC 0021-93041881121137-27$04.00
1138 DUCHEYNE AND HEALY

the initial stability may gradually disappear before bone ingrowth has set in
fully. Thus bone tissue ingrowth must occur expeditiously and certainly be-
fore any initial postoperative stability would be lost.
Calcified tissue ingrowth can be enhanced by the use of a thin lining of
calcium phosphate ceramic within the porous n e t ~ o r kIn . ~this study, a
layer of about 20 pm thick was slip cast on a porous metal fiber structure
with an average pore size of 110 pm thus retaining an open pore. In com-
parison to the unlined porous metal specimens, a pronounced increase of
the strength of the interfacial bond the few weeks following surgery was
measured. This effect disappeared at 12 weeks after implantation since the
ingrowth without enhancement had caught up with the much faster com-
plete ingrowth with a bioactive ceramic lining.
This effect of bone tissue ingrowth enhancement has recently been con-
firmed by various In these studies adhesion of the ceramic
to the metal surface was sought by plasma-spraying the ceramic powder on
to the porous metallic target.
In his studies, Hulbert favored the use of ceramics for the porous mate-
rial, since in his view the fully oxidized state of ceramics would provide
better biocompatibility than easily ionizable metals. These views were
heightened by studies showing measurable amounts of ion release from
metals. Furthermore, it was observed that the larger the specific surface
area, the greater the amount of release. Thus the question arises whether
using calcium phosphate ceramic linings on porous metals would provide
the advantage of reducing or eliminating metal ion release in addition to the
beneficial effect of tissue ingrowth enhancement.
Plasma-spraying the calcium-phosphate ceramic onto the porous metal
can alter the ion release kinetics in a variety of ways. The ceramic can shield
the metal, nonuniform coatings can create a local exposure of the metal, the
coating can degrade with time therefore provoking a temporal variation of
release, and the metal surface structure can be affected. Thus, this paper
addresses the possible plasma-spraying-induced modification of metal ion
release kinetics. Specifically, the effect of plasma-spray coating hydroxy-
apatite on porous Ti and Co-Cr made with spherical powders is reported.
The release rates are measured in an in vitro model with enhanced ion re-
lease potential.

METHODS AND MATERIALS

The specimen substrates were disk-shaped buttons with a diameter of

8 mm and a thickness of 2.75 k 0.25 mm. The porous coating provided an
additional 0.5 mm in thickness. The compositions of the Ti alloy rod and the
Ti powders used are given in Table I, the ones of the Co-Cr rod and Co-Cr
powders in Table 11. Powders 1 and 2 were used for the plasma-sprayed or
unsprayed specimens, respectively. Commercial purity (c.p.) Ti powders
rather than Ti alloy powders were chosen since clinical porous coated
devices use c.p. Ti powders.
CERAMIC COATINGS AND METAL ION RELEASE 1139

TABLE I
Chemical Composition of the Ti-Base Materials (in % by Weight)
Material Description A1 V C N Fe 0 H Cu W Ti
Ti 6 4 ELI rod* 6.0 4.2 ,011 .008 .19 .13 .0005 - - bal.
Spherical powder #It - - ,017 ,021 .13 .16 ,002 ,0030 ,0007 bal.
Spherical powder #2+ - - ,042 ,007 .33 .27 .002 ,0028 ,0007 bal.
Vendors: *Titanium Industries, Tempe, AZ; Nuclear Metals Inc., Concord, MA.

TABLE I1
Chemical Composition of the Co-Cr Base Materials (in % by Weight)
(nd: not determined)
Material Description Cr Mo Ni C Mn Si Fe N2 Co
Rod* 27.13 5.27 .01 .25 .47 .86 .15 .05 bal.
Spherical powder #1 28.7 5.7 .3 .30 .29 .49 .09 nd bal.
#2+ 28.0 5.5 <.1 .29 <.05 .7 c.10 <.02 bal.
Vendors: *Precision Castparts, Portland, OR; Nuclear Metals Inc., Concord, MA.

Upon cutting from bar stock, the flat disks were ground and smooth
blasted in a similar fashion as implant devices. Porous coatings were de-
posited by slurry casting and vacuum sintering at temperatures above
1200C. The mesh size of the atomized powders was -60 + 80, represent-
ing a particle size between 177 and 250 pm. The resulting pore size and
porosity are .275 pm and 40%. All specimens (including the one for spray-
ing) were made at once in order to exclude possible batch-induced effects.
Obviously this procedure reduces the pore size subsequent to spraying.
Subsequently the specimens were passivated following ASTM standard^.'^
Hydroxyapatite powder with an x-ray diffraction pattern without supple-
mentary peaks compared to ASTM powder diffraction file 9-432 and a mesh
size <325 mesh was used for the plasma spraying in air (A.P.S., Dayton,
OH). Figures la and b are scanning electron micrographs of the non-coated
and the plasma sprayed coated porous titanium specimens. Figure l c is a
cross-sectional micrograph after plasma spraying, showing the alloying of
the beads and the retention of open porosity after spraying. The Ca-P lin-
ing was irregular and averaged 100 pm. Through viscous flow it reached
the underlying substrate. Prior to immersion all specimens were ultrasoni-
cally cleaned in mild detergent following ASTM recommended practice . I 3
The solution was a 1: 1000 mixture of a detergent containing less than 0.5%
of phosphates. Subsequently they were rinsed in ultrasonically agitated
water three consecutive times.
The effect of plasma spraying was determined by immersion of plasma-
sprayed and nonsprayed specimens in Hanks balanced salt solution
(HBSS)I4with 1.5 mM DS-EDTA [(ethylene diamino) tetra acetic acid di-
1140 DUCHEYNE AND HEALY

Figure 1. Scanning electron micrographs of the surface of porous titanium

) with (b) a plasma sprayed calcium-phosphate ceramic
without (a) ( 8 0 ~and
coating (80X); (c) micrograph of a cross section after sintering and plasma
spraying (100x).
CERAMIC COATINGS AND METAL ION RELEASE 1141

(c)
Figure 1. (continued)

sodium salt)] for periods of 1, 2, and 4 weeks. The pH of this solution was
6.4. For each time period, groups of 7 Ti, Ti-HA, Co-Cr, and Co-Cr-HA
specimens were inserted in disposable poly(tetrafluoroethylene) (PTFE)
specimen holders and immersed in poly(propy1ene)or linear poly(ethy1ene)
vials with 50 mL of test solution. Thus a total of 112 specimens were im-
mersed. The vials were covered with gas permeable poly(styrene) petri
dishes as lids. The PTFE specimen holders were 20-mm-diameter, 5-mm-
high disks with a central &mm-diameter, 3-mm-deep recess. The specimen
surface was flush with the PTFE surface. Thus only the top surface of the
specimen was exposed to the solution. The vials were placed in an incuba-
tor at 37C with ambient air containing 5% CO,. Prior to the experiment it
was decided not to use mechanical stirring. If the ceramic coating would
debond, this would allow it to remain in situ, similar to the in v i m situation.
If it could rub over the metal surface, it could have induced mechanically
enhanced ion release. Control solutions were identical to the test solutions,
except that the specimen holders did not contain any specimen. Seven vials
of control solution were placed in the incubator at the same time as the test
solutions and 250-pl aliquots were withdrawn from each at weeks 1, 2, and
4. One 10-liter batch of solution was used for the control, test, and standard
calibration solutions needed in the EAAS analyses. The vials, lids, speci-
mens holders, and all other laboratory ware contacting the test solutions
were acid washed prior to their use. Only ultra-pure chemicals were used
for all of the procedures.
1142 DUCHEYNE AND HEALY

Upon expiration of the appropriate immersion periods, the sample and

control solutions were vortexed and shaken prior to the chemical analysis.
Aluminum was measured first to minimize the risk of contamination arising
from any additional handling.
The trace metal content in solution was measured in triplicate using Elec-
trothermal Atomic Absorption Spectrophotometry (EAAS) [ (Perkin-Elmer
model 4000 with a HGA 400 furnace programmer, automatic sampler AS-40,
and double pyrolytically coated graphite tubes and l'vov platforms (Nor-
walk, CT)]. The trace metal content was measured directly using calibration
curves made each time at the time of the actual analyses. Error bars corre-
sponding to 95% confidence intervals on the means were calculated to as-
sess the accuracy of the chemical analytical measurements. '"I7 The EAAS
programs used are summarized in Appendix 1. The specifics of the statisti-
cal methods are summarized in Appendix 2.
In order to characterize the surface of the metal, i.e., that part of the speci-
men that is actually undergoing the release, selected immersed and nonim-
mersed specimens were analyzed in a Scanning Auger Electron Microprobe
Spectroscope (SAEMS)(PHI Multiprobe 600, Perkin Elmer, Eden-Prairie,
MN). Typical operating conditions were: vacuum < mm Hg, 5 kV elec-
tron beam voltage, 0.4 pA electron beam current, and Ar-ion beam voltage
at 2 kV. Characteristics of the ceramic are discussed separately."

RESULTS AND DISCUSSION

Analytical sensitivity

Addressing the issue of parametrically induced variations of ion release, a

twofold limitation arises. First, the amount of ions released from the
metals in current use for permanent orthopedic devices are very small.
Thus, only the most sensitive of analytical detection methods can be used.
Second, even with extreme sensitivities and minimal detection limits, it is
possible that the release rates are still below the level of normal instrument
capabilities.
The sensitivity and detection limit of the alloying elements of the Co-Cr
alloys are considerably below the ones of the Ti-6A1-4V alloy. This is sub-
stantiated by Table I11 which summarizes these values for the elements ana-
TABLE I11
Detection Limits and Sensitivities for EAAS Analysis of Ions in
Hanks'-1.5 mM DS-EDTA
Detection Limit Sensitivity
WmL) (ng/mL)
Ti 1.1 0.50
V 0.54 0.20
A1 0.42 0.05
CO 0.17 0.04
Cr 0.06 0.01
CERAMIC COATINGS AND METAL ION RELEASE 1143

lyzed in the present study. Cobalt and chromium are present for about 60%
and 30% by weight in the alloy and display a detection limit and sensitivity
one to two orders of magnitude lower than the major alloying element, Ti,
of the other alloy system. As far as A1 and V are concerned they are present
for about 6 and 4% by weight only in the bulk metal and to smaller concen-
trations in the protective oxide. In addition their detection limits are consid-
erably higher than that of Co, and especially the one of Cr.

Metal ion release enhancement

We circumvent the experimental limitation of the small amounts of re-

lease based upon the assumption that ions which are released from the
metal surface in viuo do not remain in solution around the metal surface. In-
stead they undergo complex forming reactions, and/or are transported
away through fluid motion. As a result there is no steady state of dissolu-
tion and redeposition established at the metal surface in vim. We model this
divergence from steady state by continuously complexing the metal ions
with DS-EDTA. Thereby we effectively eliminate the ions from being avail-
able in the solution, and thus, we remove them from the site of release.
The functional groups of the DS-EDTA provoking the complex formation
are amino- and carboxyl-groups. They are the same as the protein structures
thought being responsible for metal ion complex forming reactions i~ uivo.
A 1.55 mM concentration complexes the same amount of Ca ions as the
average concentration of albumin in human plasma. The addition of DS-
EDTA is essential since without it, Al and Ti concentrations in solution at
1 week of immersion are below the detection limit.
The addition of DS-EDTA to simulate the removal of the ions from the
site of release, i.e., the metal surface, is justified since, in this article, we do
not draw conclusions based on absolute data, but on comparisons between
plasma-sprayed and nonsprayed materials. In this respect it is important to
know that DS-EDTA binds to all cations of the study in a 1:l ratio, regard-
less of their valency. The equilibrium constants for the formation of the
cation complexes (Kml) is pH and cation-dependent. At the pH of this ex-
periment (6.4) the Kmys for the heavy cations (Ti, Al, V, Co, and Cr) are
much greater than those for the cations initially present in the HBSS
(Ca, Mg, K, Na), thus representing a stronger cation-EDTA complex. In ad-
dition, the Km;s are thermodynamic constants which do not give informa-
tion about the specific cation-EDTA reaction rates. Therefore, the validation
of the use of DS-EDTA for the assessment of absolute release values awaits
a comparative assessment of the order of kinetics of ion release in vitro and
in uivo. If they would be the same there is presumptive evidence that the re-
lease mechanisms in vitro and in viuo are similar. Whether, in addition it
would be possible to establish the time factor of enhancement associated
with the use of DS-EDTA is not known at this time.
1144 DUCHEYNE AND HEALY

Ions in solution

The extreme sensitivities and the enhancement of ion release are essential
considering the results of ions in solution with a n d without a plasma
sprayed hydroxyapatite coating. Table IV and Figures 2a,b,c, and 3a,b rep-
resent the Ti, Al, V, Co, and Cr ion concentrations at 1, 2, and 4 weeks of
immersion for the control specimens, the porous metal specimens and the
porous metal specimens with plasma sprayed ceramic coating. The detec-
tion limit (DL) for each element is indicated on the right side of the dia-
grams. The bars without confidence interval are the result of values below
the detection limit. These bars, as represented, are made equal to the DL.
Statistically significant conclusions with respect to the effect of plasma
spraying can be drawn for those elements of the Ti alloy with an ion con-
centration exceeding the detection limit. As Table V shows, the coating re-
duces the ion release of Ti (at 1, 2, and 4 weeks), A1 (at 1 and 4 weeks), and
V (at 2 and 4 weeks). A statistically significant difference was not obtained
for V at 1 week since the absolute amounts of ions in solution was below
the detection limit. The data of the A1 release after plasma spraying at
2 weeks included two outliers which could not be removed statistically. We
consider this the reason for the absence of a statistically significant result for
the A1 release at 2 weeks.
In assessing these data it is important to bear in mind that A1 and V origi-
nally are not present in the powder of the porous coating but only in the
substrate. Sintering produces A1 and V diffusion into the first layer of
beads. The release of A1 from the substrate increases with time, both with
and without the plasma spray coating, but the reducing effect of the ce-
ramic coating remains. It is also interesting to note that the magnitude of
the reduction of Ti release is considerably greater than the one of Al. This
could mean that the effectiveness of the sprayed ceramic layer is much less
pronounced on the substrate than on the beads,

TABLE IV
Group Grand Means and Standard Deviations (in Parentheses) of Ions in Solution
(ng/mL except Ca: pg/mL) (nd: not detectable)
Ions in Solution
Time of Specimen
Immersion Type Ti A1 V co Cr Ca
1 week w/o CaP 28.2(4.8) O.g(O.3) nd 5.0(1.0) 0.17(0.09)
w CaP 1.2(0.50) 0.4(0.1) nd 4.0(0.5) 0.37(0.09) 23.9(0.3)
Control nd 0.6(0.2) nd nd nd
2 week w/o CaP 59.2(12.6) 1.5(0.7) l.l(O.4) 4.7(1.2) 0.39(0.20)
w CaP 1.9 (0.5) 1.5(1.3) 0.7(0.4) 5.0(0.8) 0.69(0.20) 38.1(0.4)
Control 1.8 (0.7) 0.3(0.3) 0.6(0.4) nd nd
4 week w/o CaP 78.3 (6.6) 3.6(1.1) 1.5(0.5) 7.1(1.5) 0.39(0.28)
w CaP nd 1.9(0.6) nd 6.1(1.5) 0.49(0.24) 54.0(0.6)
Control nd l.l(O.3) nd nd nd 3.8(0.2)
Note: The standard deviations reflect both the variation within each sample popula-
tion and the error on the analytical chemical measurements.
CERAMIC COATINGS AND METAL ION RELEASE 1145

~D.L.
1 2 3 4
TIME (WEEKS)

(4

1 2 3 4
TIME (WEEKS)
(b)

2.0 I I

D.L.

1 2 3 4
TIME (WEEKS)
(c)
Figure 2.(a, b, c ) The release of Ti, A1 and V from the Ti alloy after 1, 2 and
4 weeks of immersion; D.L. indicates the detection limit of each element. Er-
ror bars represent the 95% confidence interval on the means.
1146 DUCHEYNE AND HEALY
n

1 2 3 4
TIME (WEEKS)
(a)

1.0 1
=
&
E 0.8
-
C

z
0 0.6
5U
2 0.4
YZ
0 0.2
0 D.L.

0.0
1 2 3 4
TIME (WEEKS)
(b)
Figure 3.(a, b) The release of Co and Cr from the Co-Cr alloy after 1, 2 and
4 weeks of immersion; D.L. indicates the detection limit of each element. Er-
ror bars represent the 95% confidence interval on the means.

The effects with the Co-Cr alloy are not as predominant. As far as the
temporal variations or plasma-spray-induced variations of Cr in solution are
concerned, there is no clear tendency. There is an increase of Cr release as-
sociated with the plasma spraying but the level of significance decreases

TABLE V
Level of Statistical Significance of the Difference between Release Rates with and
without Plasma-Sprayed Coating Using a Two-Tailed t-Test. The Differences Are a
Plasma-Spray-Induced Reduction for Ti, Al, V, and Co Release, and an Increase for
the Cr Release
Time of immersion Ti A1 V co Cr
1 week 0.001 0.02 n.m. (") 0.05 0.002
2 week 0.001 N.S. 0.2 N.S. 0.02
4 week 0.001 0.01 0.002 0.5 0.5
*n.m.: not measurable.
CERAMIC COATINGS AND METAL ION RELEASE 1147

with time of immersion. The absolute amount of Co in solution is an order

of magnitude larger than the one of Cr. As a result error bars for Co, which
represent chemical analytical accuracy, are much smaller; yet, again there is
no major tendency since the plasma-spraying-induced decrease of Co re-
lease is more significant at 1 than at 4 weeks, and statistically insignificant
at 2 weeks, as shown in Table V. The time dependent variation of release
appears to be a rapid initial release during the initial period of immersion
(the first week) followed by a much slower release subsequently. This time
dependent pattern compares to recent findings. l9
We do not attempt to compare the absolute values of metal ion concentra-
tion between the Ti and Co-Cr based groups, since the relative kinetics of
complex formation of each element is not known for the present experimen-
tal conditions. However, we can compare the concentration of each metallic
element between, prior to, and subsequent to plasma spraying. The Ti sam-
ples show a significant reduction in Ti release due to plasma spraying, the
Co-Cr samples do not.
The difference in the plasma-spray-induced variation of ion release be-
tween the two alloy groups indicates that there is an effect of the plasma
spray coating. Irrespective of this observation, there are additional reasons
to exclude other possible explanations why we measured a significant re-
duction of ion release from the Ti-A1-V alloy system. We excluded the ef-
fect Ca released from the ceramic coating might have on the accuracy of the
electrothermal atomic absorption spectroscopic measurements. Specifically
this would mean that the Ti signal is reduced, but the Co and Cr signal is
largely unaffected. We measured that titanium absorbance values were not
affected by the Ca concentration.
Another hypothetical effect was the reduction of the complexing activity
of the solution as a result of competitive binding of the DS-EDTA by the Ca
ions released from the calcium-phosphate ceramic. Using flame AAS we
measured Ca in the Ti and Co--Cr samples solutions at 1, 2, and 4 weeks.
These data are included in Table IV. If we would assume that all of the mea-
sured released Ca binds in a l : l ratio to EDTA, there would be approxi-
mately 60, 50, and 25% of the activity remaining at 1, 2, and 4 weeks.
However, this assumption is overly conservative since the concentration of
released Ca is still an order of magnitude smaller than the Ca originally
present in HBSS. If the assumption would have been correct, there would
not have been any activity of the EDTA in the original solution. Further-
more, as indicated before, the Ca-EDTA complex a t the pH of the experi-
ment is very weak, and displacement by the heavier cations can occur
easily. Other complexes formed, specifically with Mg and alkali metals, are
even weaker. Thus, any reduction of free EDTA, if it occurs, cannot pro-
duce the reduction of ion release observed after the plasma spraying.

Surface structure and composition

Based on theoretical precepts, plasma spray coating porous metals with a

bioactive ceramic coating can have either a beneficial or an adverse effect on
1148 DUCHEYNE AND HEALY

the metal ion release. Similarly, any other type of ceramic coating can have
an effect which can not be predicted theoretically.
We can exclude any adverse effect for the present alloy systems due to
plasma spraying hydroxyapatite for u p to the time (though enhanced) cov-
ered by our experiments. Nonuniform coatings do not create localized areas
of metal exposure with a concomitant increase of metal ion release. Simi-
larly, crevices at the metal-ceramic interface which could be formed during
the biodegradation of the ceramic do not measurably affect the release from
the exposed metal surface.
The beneficial effect can be explained theoretically by either one or both
of two mechanisms. The plasma spray coating either shields the metal,
thereby effectively hindering any transfer of metallic ion into solution, or
the plasma spray deposition of hydroxyapatite onto the metal changes the
mechanisms or the kinetics of metal ion release.
Scanning Auger Electron Microprobe Spectroscopy (SAEMS) in conjunc-
tion with Scanning Electron Microscopy (SEM) provides strong evidence to
exclude the first hypothesis. After immersion in the reaction-enhancing so-
lution the porous metal lost a portion of its bioactive ceramic coating. Fig-
ure 4 is a scanning electron micrograph illustrating this phenomenon. Our
SAEMS analyses indicate that the metal is exposed, because Ti is detected
prior to any Ar ion sputtering. Furthermore, the amount of coating lost
varied considerably among the Ti specimens immersed for 4 weeks. Thus,
each specimen had its surface exposed to a different extent after immersion.

Figure 4. Scanning electron micrograph of a Ti specimen after 4 weeks of

immersion; note the exposure of the metal powders ( 8 0 ~ ) .
CERAMIC COATINGS AND METAL ION RELEASE 1149

This variability in coverage, however, was not manifested in a large stan-

dard deviation of the metal ion release. Thus the protective shielding effect,
if it ever existed immediately after implantation, is one of short duration
and certainly does not play a predominant role.
The key element to explain the reduction of metal ion release associated
with plasma sprayed calcium phosphate ceramic coatings is probably the
change of the structure and composition of the protective metal oxide. We
performed SAEMS analyses to obtain the compositional information. Such
measurements suffer from considerable experimental difficulty. First only a
limited number of beads has what appears macroscopically to be the metal
surface exposed. Second, analyses can be performed on the outside of the
top layer of beads only; the second and subsequent layer of particles and
the substrate cannot be analyzed. Third, only part of the surface of the
highly curved particles is well oriented and in line of sight with respect to
the Ar ion sputter beam and the Auger electron detector.
The plasma spray deposition technique is a line of sight process which
produces a nonuniform coating when applied to porous surfaces made with
spherical beads. Our SAEMS analyses with parallel SEM observations of the
same beads show that three types of Ti-bead surfaces can be identified after
plasma spraying: type I and I1 areas, as defined below which were hit by
impinging particles either directly or as a result of the viscous flow of the
calcium phosphate ceramic immediately after impact; and type I11 areas
which were not affected at all by the plasma spray process. Figure 5 shows
a bead typical for type 111. A SAEMS depth profile (100 x 100 pm2 area

Figure 5. The central bead is typical for a type 111 surface.

1150 DUCHEYNE AND HEALY

analysis) neither indicated a measurable concentration of Ca or P at the sur-

face, nor into the oxide.
Figure 6a represents a sphere covered by remnants of the ceramic.
Point 1 was chosen in what was believed to be a thin film (about 0.1 to
1 pm) of ceramic. The spot analysis reveals the combined presence of P, Ca,
Ti, and 0, and the P peak is situated at 120 eV (Fig. 6b). Therefore, the
white streak in Figure 6a is in reality an interfacial reaction layer between
the ceramic and metal particles: this follows from the depth of resolution of
the SAEMS technique (approximately 0.002 p m ) and the simultaneous
presence of elements of the ceramic (Ca and P) and the metal (Ti). Addi-
tional evidence follows from the position of the I' peak, since it was shown
that when P is present in the Ti oxide surface its peak position is 120 eV;
when present in the ceramic phase there is a peak shift to lower values.2"
We consider this a type I interface. The spot analyses at points 2 and 3 of
Figure 6a are made at areas that appear to have surface ripples and slightly
flattened areas as a result of the impact of the ceramic particles. Morphologi-
cally there is no ceramic film visible at points 2 and 3, but the SAEMS analy-
sis clearly indicate Ca incorporation: Figure 6c is a survey spectrum of

(4
Figure 6 . SAEMS spectra and SEM micrographs from a Ti, HA coated
specimen immersed for 4 weeks.
a. detail of the sphere of analysis
b. spectrum at location 1 of (a) (spot analysis)
c . spectrum at locatiun 2 of (a) (spot analysis)
Both spectra were recorded after 8% minutes of sputtering at a shallow an-
gle.
CERAMIC COATINGS AND METAL ION RELEASE 1151

(4
Figure 6 . (continued)
1152 DUCHEYNE AND HEALY

point 2 illustrating the still intensive Ca peaks at 291 eV. These areas 2 and
3 illustrate the type I1 surface: in essence, a type I1 surface does not appear
morphologically as a ceramic film; however, Ca has diffused into it. Fig-
ure 7 is another illustration of the juxtaposition of type I and I1 surfaces.
Figure 7a and b show a bead (secondary electron image) and its corre-
sponding Ca-Auger electron intensity image. The area of high Ca intensity
(white) is much larger than the flake of ceramic remnant present in the
lower left corner of (a). These type I1 areas can have both a Ca and C con-
centration which level off slowly upon penetrating into the oxide as is illus-
trated by Figure 8. Figure 8a, b, and d are increasingly larger magnifications
of the area of analysis; Figure 8c is the Ca Auger electron intensity image
corresponding to Figure 8b; Figure 8d is a detail of the area indicated in Fig-
ure 8b; it represents the total area (13.5 X 13.5 pm2)of the surface survey of
Figure 8e, indicating a clear Ca and C presence. These Ca and C intensities
decrease only slowly upon Ar ion etching as shown in Figure 8f.
The incorporation of Ca is not due to the immersion in a Ca containing
solution, since Ti specimens without plasma sprayed coating, analyzed
prior to and after 4 weeks of immersion do not show an indiffusion: Fig-
ures 9a and 9b. Similar observations of Ca incorporation as made with the
Ti alloy specimens were made for the Co-Cr bead surfaces prior to and
after plasma spraying.
Having shown that a reacted layer on the Ti bead surfaces can be ob-
served after a 4-week immersion period, implies that even when a chemical
interaction took place between metal and ceramic upon plasma spraying,
the interface is easily exposed.
In analyzing the present data it is of value to reconsider the experimental
limitations cited above. Due to the variable orientation of the areas analyzed
the depth profiles cannot be expressed in sputtering depths. Also since a
considerable portion of the surface is not accessible or is covered by ce-
ramic, the relative ratios of the type I, 11, and I11 surface cannot be deter-
mined. It appears though that type 111 occurs primarily at the apexes of
exposed beads. By virtue of the absence of these quantitative data, the
present results cannot serve as conclusive evidence that Ca-incorporation is
the ion-release determining factor. However, the persistent presence of this
phenomenon supports the hypothesis that ion diffusion and migration
through the oxide film is an important mechanism governing metal ion re-
lease phenomena. Indeed, diffusion and migration of ions across the pro-
tective oxide which are controlled by compositional gradients and electrical
potential differences respectively are widely accepted mechanisms opera-
tive during passive dissolution.21,Ionic transport through the oxide is then
affected by the characteristics of the oxide crystal lattice, including thick-
ness, defect density, and presence of coherent precipitates.*
Hoar and M e a r ~ showed
~ ~ that under potentiostatic conditions over a pro-
longed period of weeks, the rate of passive dissolution of a wide variety of
titanium alloys showed progressive diminution by orders of magnitude for
the duration of the test. The obvious explanation was that the oxide surface
became progressively thicker, Ti diffusion became increasingly more diffi-
CERAMIC COATINGS AND METAL ION RELEASE 1153

Figure 7. Secondary electron image (a) and corresponding Ca-Auger elec-

tron intensity image (b) of a Ti, HA coated specimen immersed for 4 weeks.
1154 DUCHEYNE AND HEALY

Figure 8. Spectra and Auger electron intensity micrographs of one location

on a Ti plasma sprayed specimen immersed for 4 weeks.
a. secondary electron image: general view
b. secondary electron image: area of surface analysis of (c)
c. Ca-Auger electron intensity image of (b)
d. Secondary electron image of the area indicated on Fig. b
e. Auger electron surface survey of the area of Fig. d: note the distinct Ca
peak
f. Auger electron depth profile of the area of Fig. d: note the slowly decreas-
ing Ca intensity.
CERAMIC COATINGS AND METAL ION RELEASE 1155

(4
Figure 8. (continued)
1156 DUCHEYNE AND HEALY

(4
Figure 8. (continued)
 1157

Figure 9. Depth profiles of the Ti surface, non plasma sprayed, prior to (a)
and after 4 weeks of immersion (b).
1158 DUCHEYNE AND HEALY

cult and oxide formation rates decreased for increasing time of experimental
observation.
It was shown recently that ion release from titanium is associated with
oxide thickening and that a passivation treatment reduces the amounts of re-
lease c o n ~ i d e r a b l y .Another
~ ~ , ~ ~ recent experiment, now with Co-Cr alloys,
shows that another ceramic coating, specifically one type of C-coating en-
hances the release from this alloy, due to a preferential segregation of Mo to
the surface of the alloy and a resultant change in electrochemical properties.26
It was also shown that ions from the calcium-phosphate ceramic can be
incorporated into the titanium surface. After electrophoretic deposition and
low temperature sintering of calcium phosphate coatings on titanium, the
combined use of x-ray diffraction, Auger electron spectroscopy and energy
dispersive x-ray analysis unequivocally established indiffusion of Ca and P
into the Ti
We suggest the following explanation for our present observations: the
plasma spray process required a substantial dissipation of thermal and ki-
netic energy at the time of impact of the calcium-phosphate particle on the
metal surface. This energy is released in part to the oxide, by either promot-
ing diffusion of ions of the ceramic into the metal surface or by modifying
the defect density of the oxide crystal structure.28The former phenomenon
can be measured by SAEMS analyses, the latter cannot. Conceivably, both
can have an effect on the ion diffusion kinetics from the bulk through the
oxide to the surface of the metal. Our data bear out that such effects are sig-
nificant with respect to the ion release from titanium alloys and much less
pronounced for the cobalt-chromium alloys.

CONCLUSION

Using an in vitro model with enhanced ion release capabilities, and ex-
treme chemical analytical sensitivities, it is shown that, under the condi-
tions of this experiment, plasma spraying an original hydroxyapatite
powder on titanium porous coated specimens significantly reduces the ion
release rate. Such observation pertains primarily to Ti and A1 release. Fur-
ther it is shown that no major change is produced in the ion release kinetics
from Co-Cr alloys, specifically the Co and Cr release.

APPENDIX 1

Characteristics of the electrothermal atomic absorption spectroscopical

(EAAS) analyses of Ti, Al, V, Co, and Cr in HBSS with 1.55 mM DS-EDTA.
Specific conditions can vary from instrument to instrument.

Titanium

Spectral wavelength: 365.3 nm.

Spectral width: 0.2 nm low.
CERAMIC COATINGS AND METAL ION RELEASE 1159

Determination off tube wall.

Lamp current: 35 mA.
Peak height measurement.
Drying Charring Atomization Cleaning
Step 1 2 3 4 5 6 7

Temperature (C) 80 100 120 350 1200 2800 2700

Ramp time (s) 1 20 15 15 10 0 1
Hold time (s) 3 30 15 5 30 7 2
Baseline reading (s) 26

Vanadium

Spectral wavelength: 318.4 nm.

Spectral width: 0.7 nm low.
Determination off tube wall.
Lamp current: 35 mA.
Peak height measurement.
Drying Charring Atomization Cleaning
Step 1 2 3 4 5 6 7
~ ~

Temperature (C) 80 100 120 300 1300 2700 2700

Ramp time (s) 1 20 15 15 20 0 1
Hold time (s) 3 30 15 5 30 7 2
Baseline (s) 36

Aluminum

Spectral wavelength: 309.3 nm.

Spectral width: 0.7 nm low.
Determination on Lvov platform.
Lamp Current: 20 mA.
Peak area measurement.
Read time: 10 s.

Drying Charring Atomization Cleaning

Step 1 2 3 4 5 6 7
Temperature (C) 100 130 180 500 1500 2500 2700
Ramp time (s) 1 20 15 20 10 0 1
Hold time (s) 3 20 15 10 40 10 2
Baseline (s) 33

Cobalt

Spectral wavelength: 240.7 nm.

Spectral width: 0.2 nm low.
1160 DUCHEYNE AND HEALY

Determination off tube wall.

Lamp current: 25 mA.
Peak height measurement.
Drying Charring Atomization Cleaning
Step 1 2 3 4 5 6 7

Temperature ("C) 80 100 120 300 1100 2400 2600

Ramp time (s) 1 10 20 20 20 0 1
Hold time (s) 3 10 30 10 20 7 2
Baseline (s) 26

Chromium

Spectral wavelength: 357.9 nm.

Spectral width: 0.7 nm low.
Determination off tube wall.
Lamp current: 25 mA.
Peak height measurement.
Drying Charring Atomization Cleaning
Step 1 2 3 4 5 6 7

Temperature ("C) 80 100 120 300 1200 2600 2600

Ramp time (s) 1 10 20 20 20 0 1
Hold time (s) 3 10 30 10 20 7 2
Baseline ( s ) 26

APPENDIX 2: STATISTICAL ANALYSIS

1. It is assumed that the observations are statistically independent and

that they are normally distributed. Thus, the analysis is based upon the
behavior of the true means and their associated variances.
2. The true value of the elemental concentration in one sample solution,
p, is related to the analytical chemical error:

p=c+E (1)
with c = the estimate of p
E = the random error associated with
the EAAS analysis
The true concentration mean, G, i.e., the true mean for an infinitesimally
large population, is related to both, the analytical chemical error, E , and the
error associated with limited sampling arising by measuring a population of
npsolutions only. Thus the variance on the true concentration mean, V ( p )is
given by
CERAMIC COATINGS AND METAL ION RELEASE 1161

V(F)= V G ) + V(E) (2)

with V representing the variance
V G ) the variance of the population,
with estimated means C

3. The estimated mean of sample i of the population, ci,is obtained as the

estimation of the X-value (concentration) using calibration lines constructed
from standard concentration solutions. The determination of V ( E )is given
by Eq. (4).A full derivation of this equation can be found in chemical ana-
lytical texts.17

with n,: number of samples with standard concentrations

(in our analyses: n, = 10)
m: number of analyses per sample
(in our analyses: m = 3; i.e., triplicate
measurements)
y;: absorbance of the mean concentration of the
unknown samples (C) calculated from the
calibration line
-
y : average measured absorbance of standards
x k : individual concentration standards
-
x : mean of concentration standards
The regression line y i = b . c, + a
SYix: standard deviation of y-residuals
4. The variance V(C)is given by

9 (Cj - C)

5. Inserting (4) and (5) into (2) yields the variance on ji. In our analyses,
the magnitude of the variance primarily was determined by the variance of
the population, not the one of the chemical analysis.
6. The null hypothesis was tested using the two-tailed Students t-test
with = V(F)
n = npl = np2 = 7 (in our analyses)
the calculated t value is equal to
1162 DUCHEYNE AND HEALY

Cl - cz
t =

with the number of degrees of freedom

df = 2n - 2 .

This work was supported by grants from Miter, lnc., NSF DMR-8519059 and
NSF ECE 85-15031. Bruce Rothman's assistance with the Auger analyses is gratefully
acknowledged.

References
1. S. F. Hulbert, F. A. Young, R. S. Mathews, J. J. Klawitter, C. D. Talbert,
and F. H. Stelling, "Potential of ceramic materials as permanently im-
plantable skeletal prostheses," J. Biomed. Mater. Res., 4, 433-456 (1970).
2. J. 0. Galante, W. Rostoker, R. h e c k , and R. Ray, "Sintered fiber metal
composites as a basis for attachment of implants to bone," J. Bone Joint
Surg., 53A, 101-114 (1971).
3. The Orthopaedic lmplants and Allied Products Market in the U.S., Frost &
Sullivan, New York, 1986.
4. P. Ducheyne, "Biological fixation of implants," in Fiinctional Behaviour of
Orthopaedic Biomaterials, Vol 11. Applications, P. Ducheyne and G. W.
Hastings, (Eds.), CRC Press, Boca Raton, Fla., 1984.
5. P. Ducheyne, L . L . Hench, A. Kagan, M. Martens, A. Burssens, and
J. C. Mulier, "The effect of hydroxyapatite impregnation on skeletal
bonding of porous coated implants," I. Biomed. Muter. Res., 14, 225-237
(1980).
6. D.P. Rivero, J. Fox, A. K. Skippor, R. M. Urban, J. 0. Galante, "Calcium
phosphate-coated porous titanium implants for enhanced skeletal fixa-
tion," J. Biomed. Muter. Res., 22, 191-202 (1988).
7. M.D. Mayor, J. B. Collier, and C. K. Hanes, "Enhanced early fixation of
porous coated implants using tricalcium phosphate," Trans. 32nd ORS
Mtg., 1986, 348.
8. S . D. Cook, K. A. Thomas, J. F. Kay, and M. Jarcho "Hydroxyapatite-
coated porous titanium for use as an orthopaedic biologic attachment
system," Clin. Orthop., 230, 303-312 (1988).
9. S.F. Hulbert, J. J. Klawitter, and R. B. Leonard, "Ceramics in severe en-
vironments," Mater. Sci. Xes., Vol. 5, W. Wurth Krieg and H. Palmor,
111, (Eds.), Plenum, New York, 1971.
10. A. B. Ferguson, Jr., P. G. Laing, and E. S. Hodge, "The ionization of
metal implants in living tissues," J. Bone Jnt. Surg., 42-A, 77-90 (1960).
11. J.L. Woodman, J. Black, and S. Jimenez, "Isolation of serum organo-
metallic corrosion products from 316 L SS and HS-21 in vitro and in
vivo," J. Biomed. Mater. Res., 18, 99-114 (1984).
12. T. Smith, "Morphological characterization of porous coatings." J.
Lemons, Ed., In Quantitative characterization and performance of porous im-
plants for hard tissue applications ASTM STP 953, Philadelphia, 1988.
13. "Standard recommended practice for surface preparation and marking
of metallic surgical implants," ASTM Standard F86, ASTM, Philadel-
phia, 1976.
14. J. H. Hanks and R. E. Wallace, "Relation of oxygen and temperature in
the preservation of tissues by refrigeration," Proc. SOC.Exp. Biol. Med.,
71, 195-200 (1949).
CERAMIC COATINGS AND METAL ION RELEASE 1163

15. G. W. Snedecor and W. G. Cochran, Statistical Methods, 7th ed. Iowa

State University Press, Ames, 1980, pp. 169 ff.
16. A. G. Worthing and J. Geffner, Trratment of Experimental Data. John
Wiley & Sons, New York, 1943, pp. 195-196.
17. J. C. Miller and J. N. Miller, Statistics for Analytical Chemistry. Ellis Hor-
wood Limited, Chichester, 1984.
18. P. Ducheyne, S. Radin, K. Healy, and J.M. Cuckler, The effect of
plasma spraying on the structure and properties of calcium phosphate
ceramics, Trans. 34th ORS, 1988, p. 50.
19. J. L. Bence, A. M. Konowal, and J. Black, In vitro release from Co-Cr
microspheres in contact with saline and human plasma, Trans. 11th
A n n . M f g . SOC. Biomat., 158, 1985.
20. W. Van Raemdonck, P. Ducheyne, and P. De Meester, Auger electron
spectroscopic analysis of hydroxyapatite coatings on titanium, 1. A m .
Ceramic Soc., 67, 381-384 (1984).
21. J. C. Scully, The Fundamentals of Corrosion. Pergamon Press, Oxford,
1975.
22. P. Lacombe, Corrosion and oxidation of Ti and Ti alloys, in Titanium
and Titanium Alloys, J. C . Williams and A. F. Belov (Eds.), Plenum
Press, New York, 1982, pp. 847-880.
23. T. P. Hoar and D. C. Mears, Corrosion resistant alloys in chloride solu-
tions: materials for surgical implants, Proc. Roy. SOC., A294, 486-510
(1966).
24. P. Ducheyne, M. Martens, W. Colen, and Delport, P. The effect of sur-
face condition and environment on the Ti ion release, Trans. 10th Ann.
Mfg. SOC. Biomat., 1984, p. 185.
25. P. Ducheyne and K. E. Healy, Surface spectroscopy of calcium phos-
phate ceramic and titanium implant materials, in Surface Characteriza-
tion of Biomaterials, 8. Ratner (Ed.), Elsevier, AmsterdamiNew York,
1988.
26. J. Black, P. Oppenheimer, D. M. Morris, A.M. Peduto, and C. G .
Clark, Release of corrosion products by F-75 cobalt base alloy in the
rat 111: Effects of a carbon surface coating, J. Biomed. Muter. Res., 21,
1213-1230 (1987).
27. P. Ducheyne, W. Van Raemdonck, J. C. Heughebaert, and M. Heughe-
baert, Structural analysis of hydroxyapatite coatings on titanium, Bio-
materials, 7, 97-103 (1986).
28. W. Gopel, J . A . Anderson, D. Frankel, M. Jaehnig, K. Phillips, J.A.
Schafer, and G. Rocker, Surface defects of TiOz (110): a combined XIS,
XAES and ELS study, Surf. Sci., 339, 333-346 (1984).

Accepted at Wiley August 1, 1988