Corrosion Science 42 (2000) 481±503

Corrosion behaviour of porous titanium±

graphite composites designed for surgical
implants
D.J. Blackwood a,*, A.W.C. Chua b, K.H.W. Seah b,
R. Thampuran c, S.H. Teoh b
a
Department of Materials Science, National University of Singapore, 10 Kent Ridge Crescent, Singapore
119260, Singapore
b
Department of Mechanical and Production Engineering, National University of Singapore, 10 Kent Ridge
Crescent, Singapore 119260 Singapore
c
Institute of Materials Research and Engineering, National University of Singapore, 10 Kent Ridge
Crescent, Singapore 119260, Singapore
Received 20 January 1999; accepted 12 August 1999

Abstract

Porous titanium is a popular surgical implant material. Its low elastic modulus
encourages regular bone growth, whilst its porosity enables bone in-growth. Unfortunately,
porous titanium has poor friction and wear properties. The addition of graphite, to lower
the friction coecient and titanium carbide to increase wear resistance, could produce a
multi-component material that overcomes these disadvantages. However, polarisation
measurements revealed that the graphite causes an increase in the titanium's corrosion rate.
Therefore, the extent that graphite can be used in porous titanium biomaterials will be a
compromise between the bene®ts of improved friction properties against the disadvantage
of increased corrosion. # 2000 Published by Elsevier Science Ltd. All rights reserved.

Keywords: Titanium; Metal matrix composite; Polarization; SEM; Passive ®lms

* Corresponding author. Tel.: +65-776-3604; fax: +65-874-6289.

E-mail address: masdjb@nus.edu.sg (D.J. Blackwood).

0010-938X/00/$ - see front matter # 2000 Published by Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 0 - 9 3 8 X ( 9 9 ) 0 0 1 0 3 - 1
482 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

1. Introduction

1.1. The need for multi-component biomaterials

Wear of joint replacements is presently recognised as the biggest challenge to

the long-term success of these devices [1]. This is fundamentally a materials
engineering problem. Despite the use of advanced metals, polymers and ceramics
retrieval studies [2] have indicated that an insidious and progressive wear process
occurs during implantation. This often leads to bone resorption and loss in
biofunctionality of the implants.
The keys to better orthopedic joint implants are fourfold. First, the stresses
accommodated by the biomaterials should be as physiologic as possible [3].
Second, the biomaterial should maintain a low friction tribosystem during service;
this implies that it should be self-lubricating. Third, the biomaterial must be wear
resistant. Fourth, the biomaterial must be fracture resistant such that it does not
fail catastrophically in service. This point must be a major concern with the use of
bioceramics in load bearing applications, because while these may be defect free
after processing, stresses during service could initiate defects. Since ceramics are
notoriously notch sensitive, this could lead to catastrophic brittle failure.
Porous titanium become a popular surgical implant [4,5] material since it has a
low elastic modulus that modulus encourages regular bone growth, whilst its
porosity enables bone in-growth. Unfortunately, porous titanium has poor friction
and wear properties. Although these properties could be improved by using
particular reinforced composites, the weak interface between the particles and
matrix often results in debonding or loosening of the particles. In the wear
process, the release of these hard particles can cause further damage to the
implant and a cascade of cellular activity that degenerates bone.
To address some of the concerns with existing biomaterials, Teoh et al. [6]
developed a triphasic titanium±graphite composite. This system attempted to
obviate the weak interface between a hard particle phase and a soft matrix by
controlled di€usion of graphite into a titanium matrix resulting in a dual-phase
composite of hard titanium carbide and soft, ductile pure titanium. It was
reported that this composite comprised of: ductile pure titanium; a colony of
hard, wear resistant titanium carbide; and self-lubricating, replenishing free
graphite. It was shown that the wear performance of this composite is signi®cantly
better than pure titanium due to the particulate-like presence of the titanium
carbide. The interface between the titanium carbide and titanium matrix is non-
discrete because the former is derived from the di€usion of graphite in titanium.
This limits the debonding of these hard particles during the wear process. Friction
studies, in unlubricated conditions, revealed that the titanium±graphite composite
was capable of achieving low friction because of the gradual release of the solid
lubricant (free graphite) that adheres to the articulating surfaces.
Just as important as the mechanical properties of biomaterials is their corrosion
resistance. Multi-component systems introduce additional variables to the kinetics
of corrosion during electrochemical activity. In this paper, the in¯uence of
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 483

graphite and titanium carbide on the corrosion resistance of pure titanium is

investigated.

1.2. Corrosion of titanium-based biomaterials

Hoar and Mears [7] concluded that solid titanium and some of its alloys
should withstand exposure to body ¯uids for an inde®nite period. This is
because in near neutral solutions, including those containing the chloride ions
which attack many other metals and alloys, titanium forms a protective passive
oxide ®lm that reduces the corrosion rate to very low levels [8]. However,
in¯ammation around the implant can result in the production of hydrogen
peroxide [9], which can accelerate the titanium's corrosion rate [10±12].
Furthermore, the problem of corrosion within the human body is further
complicated by the need to ensure that toxic levels of metal ions (leaching out of
the metal implant) do not accumulate. Such toxic levels can occur even at
corrosion rates that would be considered insigni®cant with respect to the
physical performance of the implant.
Considerable work has been done to improve the corrosion resistance of
titanium by the use of noble metal alloying additions. For example Ti-0.2Pd has
considerably greater resistance to corrosion in reducing environments than that of
unalloyed titanium [13]. Noble metals such as platinum and palladium, are much
better catalysts for the reductions of both oxygen and water, the two main
supporting cathodic reactions for corrosion in aqueous environments, than
titanium. This increase in cathodic eciency leads to a positive shift in the open-
circuit corrosion potentials adopted by these alloys, thereby encouraging the
formation of a thicker passive oxide ®lm. The ®lm thickness is directly
proportional to the potential drop across the titanium/solution interface [14]. An
increase in ®lm thickness should lead to a further decrease in the already low
metal dissolution rate, whilst the titanium is in the passive state, a desirable
feature for an implant material.
Graphite, like the noble metals, is an excellent cathode. Hence, its presence in a
composite should cause a positive shift in the free corrosion potential. In the case
of solid titanium in body ¯uids, this positive shift should cause a thicker passive
®lm to form, and thus, decrease the corrosion rate. However, it has been
previously reported [15] that porous titanium does not exhibit ideal passive
behaviour in simulated body ¯uids. Therefore, it is possible that the positive shift
in the corrosion potential expected on galvanic coupling to graphite could cause
an increase, rather than a decrease, in the titanium's corrosion rate, as would be
the case for a metal undergoing simple active dissolution, for example iron in
hydrochloric acid [16].
The presence of titanium carbide in the porous titanium matrix is not expected
to signi®cantly a€ect the corrosion behaviour. This is because, it is electrically
insulating and thus cannot support the electrochemical reactions involved in
corrosion.
484 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

2. Experimental method

Disc-shaped porous titanium±graphite specimens 19.2 mm in diameter and 9.5

mm thick were prepared from commercially pure grade 4 titanium powder plus
8% high pure graphite powder, such that the total weight of the mixture was 8 g.
The particle size of the powders was analysed using feature measurements (ASTM
E20) with an advantage image analyser (Omnimet). This revealed that the
irregularly shaped titanium powder had an average particle size of approximately
150 mm (Fig. 1), whilst the graphite had an average particle size of about 100 mm
(Fig. 2).
In order to obtain an even distribution of titanium and graphite, the two
powders were blended using a twin shell blender (Apex Construction) at 30 rpm
for 45 min prior to compaction. Compaction was achieved using a punch and die
combination with a hydraulic press set at one of four di€erent pressures: 5, 10, 14
and 18 ton, to obtain the desired variations in porosity. To allow relocation and
sliding to take place between the particles, the desired compaction was only
applied in three steps. Each of the intermediate pressure steps was applied for
about 10 s.
The relative density of the compacts (RD) is de®ned as
Dgc Wgc
RD ˆ 100  ˆ 100 
…QTi DTi ‡ Qc Dc † Vgc …QTi DTi ‡ Qc Dc †

Fig. 1. Particle size distribution for the titanium powder.

 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 485

Fig. 2. Particle size distribution for the graphite powder.

and the percentage porosity (P ) of the compacts is de®ned as

P ˆ …1 ÿ RD †  100

where Dgc, Wgc and Vgc are the density, total weight and volume of the green
(non-sintered) compact, whilst DTi, Dc, QTi and Qc are the densities of titanium
and graphite and the weight percentages of titanium and graphite, respectively.
For sintering, the green compacts were placed in a small ceramic tube, the ends
of which were then covered with stainless steel foil before the whole assembly was
inserted into a vacuum furnace (Carbolite). The heating cycle used for the
sintering process is shown in Fig. 3. Care was taken to reduce the heating rate at
6008C so as to ensure that the temperature does not overshoot the desired plateau
value of 12508C. The phase transformation temperature of titanium is 8828C [17].
Hence, if transformation were to occur, it would happen during the constant
heating rate phase, although it is expected to be reversed during the slow cooling
cycle. After sintering, the dimensions of the compacts were measured again to
allow new values of their densities and porosities to be computed.
The compositions of the sintered compacts were analysed using X-ray
di€raction (XRD) performed at 2y sweeps from 20 to 508 at a scan rate of 28/
minÿ1, and their pore morphology was investigated with both, a scanning electron
486 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

Fig. 3. Heating cycle for sintering titanium and titanium±graphite composites.

microscope (Jeol JSM-T330A) and an optical microscope. Prior to observations

under the optical microscope, the specimens were mechanically polished using
silicon carbide grit size 1000 (coarser silicon carbide was found to cause pore
closure) followed by ®ne polishing, using 5 and 0.3 mm alumina, with ®nal
cleaning being performed ultrasonically in de-ionised water.
For electrochemical tests, the titanium±graphite composites were mechanically
polished with 5 mm alumina to remove the ®ne scratches and irregularities on the
surface of the porous specimen. After polishing, the specimens were meticulously
washed with acetone and de-ionised water, followed by ultrasonic cleaning in de-
ionised water for 15 min and ®nally rinsed in de-ionised water. To allow
comparison to be made between the composites and solid titanium, specimens of
commercial grade 2 titanium were also prepared in the same way as the composite
samples.
Attempts to electropolish titanium±graphite composites proved unsuccessful.
Trials using a mixture of acetic acid and perchloric acid as speci®ed in ASTM
E1558-93 [18], a composition commonly used for the electropolishing of titanium,
resulted in the surface of the composites being badly stained (oxidised to a
purplish colour with black graphite particles) at high voltages (40±60 V), whilst at
low voltages (20±25 V), a thick yellowish oxide ®lm was formed. Other
electropolishing recipes proved equally unsuccessful.
Electrical connection to the titanium±graphite composites was achieved by
attaching a circlip that was tightened around the specimen at mid-height [19]. An
insulated copper wire was then soldered onto the circlip, and electrical connection
con®rmed by means of a multimeter. The specimens were next encapsulated in a
thermosetting polyester resin (hardener: methylethylketone peroxide), that was
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 487

allowed to cure for 24 h at room temperature before use. Before the resin was
poured over the specimens, polytetra¯uoroethylene (PTFE) tape was used to cover
the test face; this eliminated the need to grind o€ the resin after curing. Finally,
insulating lacquer was applied at the resin specimen interface to prevent seepage
of electrolyte down the sizes of the specimen.
A single-compartment three-electrode cell was used for the electrochemical
experiments. The reference electrode was a saturated calomel electrode (SCE), and
all potentials quoted in this paper are based on this reference system. The
auxiliary current carrying electrode was a platinised titanium wire looped around
the specimen. The test electrolyte was either a saline solution of 0.9% NaCl
(AnalaR quality) in de-ionised water (4.0 MO cmÿ1) or commercially purchased
lactated Ringer's solution (OptoPharm), the composition of which is shown in
Table 1. Lactate Ringer's solution is similar to Ringer's solution except that 28
mmol of chloride is replaced by 28 mmol of lactate. Lactated Ringer's Solution is
a popular solution both, for the evaluation of materials for in vitro use [20,21]
and for treating dehydration associated with acidosis [22]. Fresh solution was used
for each electrochemical test, and stock solutions were discarded if exposed to air
for more than one week (mainly to reduce the likelihood of contamination by
atmospheric carbon dioxide). A review by Solar [23] concluded that inorganic
solutions based on diluted NaCl were satisfactory substitutes for human body
¯uids when studying the anodic behaviour of passive metals.
Potentiodynamic polarisation experiments were carried out at a temperature of
378C using a commercial ®eld machine (ACM instruments) which incorporates a
potentiostat, a frequency analyser and a sweep generator into one instrument.
After being allowed to stabilise under open-circuit conditions for 1 h, the
electrode potential of the specimens was swept from ÿ300 to 2500 mV, at a rate
of 25 mV per minute. The sweep was terminated once a negative current was
detected on the reverse cycle; the potential at which this occurred was termed
Ecorr, back. Each electrochemical experiment was repeated three times, with a fresh
specimen being used for each test.
Particles released from the specimen during the polarisation experiments were
analysed using a 2 MeV nuclear microprobe, for which the full details are
available elsewhere [24].

Table 1
Composition of lactated Ringer's solution

Dissolved ion Concentration (mmol dmÿ3)

Na+ 131
Clÿ 109
K+ 4
Ca2+ 2
Lactate 28
488 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

Fig. 4. XRD intensity peaks for sintered titanium-8%graphite specimens: (a) 5 ton; (b) 10 ton; (c) 14
ton; (d) 18 ton.
D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 489

Fig. 4 (continued)
490 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

3. Experimental results and discussion

3.1. Specimen morphology

XRD analysis con®rmed the hypothesis that the controlled sintering of binary
powders of pure titanium and graphite resulted in a triphasic composite of pure
titanium, graphite and titanium carbide, was produced (Fig. 4a±d). The intensity
ratio between the carbon and titanium carbide peaks was observed to decrease
with increasing compaction pressure. Likewise, the intensity ratio between the
titanium and titanium carbide peaks decreased between 5 and 14 ton, but
increased slightly between 14 and 18 ton. Since before sintering, the composite
contains an excess of Ti over carbon, the vast majority of the carbon might be
expected to be converted to TiC regardless of the compaction pressure. Thus, any
additional dependence on compaction pressure would have a stronger in¯uence on
the C peaks than on TiC or Ti peaks. Thus, it is tentatively concluded that the
percentage of free carbon converted to carbide increases with compaction
pressure.
The compacted specimens appeared visually smaller in volume after sintering,
and this is believed to be due to neck growth occurring at points of particle
contact. Sintering provided thermal energy to bring about surface and bulk
di€usion that caused the densi®cation of the specimen, leading to a drop in its
porosity. The calculated porosities of titanium±graphite before and after sintering
are shown in Table 2. It can be seen that sintering caused a drop in porosity and
the amount of densi®cation decreased with increasing compacting pressure. This
agrees with previous work done [25,26] on porous materials, which concluded that
at higher compacting pressures, the titanium particles would have been plastically
deformed to take on a ¯attened appearance, with fewer voids. The reduction of
total surface energy via surface area reduction would thus be impeded and hence
the amount of densi®cation decreased.
Fig. 5(a) and (b) show photomicrographs (200) of the surfaces of sintered,
porous titanium±graphite system at compacting pressures of 5 and 18 ton,
respectively. With increasing compacting pressure, the pores became smaller and

Table 2
E€ects of sintering on the mean and standard deviation porosity of the titanium±graphite compacts

Specimen Titanium±graphite (8%)

Percentage porosity Densi®cation

Compacting pressure (tons) Unsintered Sintered

5 36.5(0.5) 26.8(0.3) 9.7

10 27.0(0.3) 19.5(0.4) 7.5
14 22.0(0.3) 15.7(0.5) 6.3
18 18.2(0.4) 13.6(0.5) 4.6
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 491

Fig. 5. Photomicrographs at 200 magni®cation of sintered titanium-8%graphite compacts: (a) 5 ton;

(b) 18 ton.

more isolated: the 5 ton compacts have large (typical diameters in the range 100±
200 mm), interconnected pores while the 18 ton compacts have small (typical
diameters in the range 10±80 mm) isolated pores. The light grey matrix in these
photographs represents metallic titanium, whilst the slightly darker more grainy
regions are thought to be the titanium carbide. There are also a number of black
specks within both the titanium and titanium carbide phases and these are most
likely unreacted graphite particles. The majority of the titanium carbide forms on
the periphery of the pores (the graphite being located within the pores between the
Fig. 6. SEM photographs of sintered titanium-8%graphite compacts at 75 magni®cation: (a) 5 ton;
(b) 18 ton. (c) 5 ton at 350 magni®cation.
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 493

titanium particles prior to sintering); hence it covers a higher fraction of the true
surface area of the sintered compact than the original graphite weight percentage
might suggest. However, in all cases, the majority of the pores have at least two
components on their walls; titanium and titanium carbide.
At ®rst sight, SEM photographs (Fig. 6a and b) of the titanium±graphite
compacts appear to contradict the optical pictures, in that the pore sizes appear to
be much smaller (typical diameters 10±20 mm) and independent of compaction
pressure; with only the number of pores being an inverse function of compaction
pressure. However, examining the higher magni®cation, SEM photograph shown
in Fig. 6(c) can solve this apparent contradiction. This is an image of a specimen
compacted to 5 ton. In the centre of the image, there appears to be one large
shallow pore with a diameter in the region of 200 mm. At the base of this pore,
well-resolved particles can be observed, and it can be estimated that the pore
depth is equal to only a few particle widths; that is in the order of 100±200 mm.
The SEM image also shows about 20 much smaller pores, with diameters in the
range of 10±20 mm, that appear simply as black holes indicating that their base is
too deep to be brought into focus. Many of these smaller pores are located within
the single large pore, suggesting that the compact consists of a matrix of large
pores connected by a number of much smaller channels.

3.2. Electrochemical experiments

The open-circuit corrosion potential (Ecorr) adopted by the specimens will

depend on the balance of the kinetics between the anodic corrosion reaction of the
titanium metal and the cathodic supporting reactions. In aqueous solution, these
cathodic reactions are usually the reduction of either dissolved oxygen or water.
The titanium carbide, being an insulator, is not expected to take part in the
electrochemical reactions that control the corrosion behaviour of the composites.
However, the kinetics of cathodic reductions of both oxygen and water are more
than an order of magnitude faster on graphite than on titanium [27]. Therefore,
the value of Ecorr is expected to be controlled by the ratio between the exposed
surface areas of the free graphite and the unreacted titanium.
Table 3 shows the range of Ecorr values adopted by the specimens in various

Table 3
Mean and standard deviations open-circuit corrosion potentials obtained from specimens after stabilis-
ing for 1 h

Specimen Solution Corrosion potentials (mV vs. SCE)

5 ton 10 ton 14 ton 18 ton

Porous titanium 0.9% NaCl ± ÿ19 (35) ± 68 (14)

Porous titanium Ringer's ÿ47 (10) ÿ63 (23) ÿ64 (12) ÿ92 (18)
Titanium-8%graphite 0.9% NaCl 124 (14) 134 (15) 127 (15) 111 (14)
Titanium-8%graphite Ringer's 40 (12) 74 (17) 52 (34) 49 (22)
 494
D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

Fig. 7. Potentiodynamic curves for titanium-8%graphite in 0.9% NaCl saline solution. Sweep rate 25 mV minÿ1. For comparison, the polarisation curve of
a commercial grade 2 solid titanium specimen is also shown.
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

Fig. 8. Potentiodynamic curves for titanium-8%graphite in lactated Ringer's solution. Sweep rate 25 mV minÿ1. For comparison, the polarisation curve of
495

a commercial grade 2 solid titanium specimen is also shown.

496 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

test solutions. Although there appeared to be no correlation between compaction

pressure and corrosion potential, the data in Table 3 shows that the presence of
graphite in the composite did cause the corrosion potential to shift in the expected
positive direction by more than 100 mV. Table 3 also shows that the corrosion
potentials adopted in 0.9% NaCl saline solution were consistently positive of that
adopted in Ringer's solution. This result was rather unexpected, since it has
previously been reported that lactate does not in¯uence the kinetics of anodic
dissolution [20,21,23]. In passivating media, a positive corrosion potential
encourages the formation of a stable protective oxide ®lm. However, if active or
localised corrosion (e.g. pitting) occurs, then a positive corrosion potential causes
the corrosion rate to increase.
Figs. 7 and 8 show potentiodynamic polarisation curves for titanium±graphite
composites in 0.9% NaCl saline solution and Ringer's solution. There were no
major di€erences between the general type of electrochemical behaviour of
titanium±graphite composite in the two solutions. The main data derived from
these curves, along with data for porous titanium extracted from previous
publications [15,19,28] are summarised in Tables 4±6.
One point of additional interest in the polarisation curves is the small amount
of hysteresis observed in the 5 ton specimens. However, this is not likely to be due
to pitting corrosion for the following reasons. Firstly, there was no rapid increase
in the current density during the forward scan (except that due to oxygen
evolution that was present at all compaction pressures) which is characteristic of
pitting corrosion. Secondly, in 0.9% saline solution, the forward and reverse scans
cross three times, rather than the once expected in pitting corrosion. Thirdly, on
reversing the potential sweep, the current density starts to fall almost immediately,
whilst with pitting corrosion it usually continues to increase. Finally, titanium is
usually resistant to crevice corrosion in sea water [16], thus, even if the conditions
within the pores were very acidic, titanium would still not be expected to undergo

Table 4
Summary of the mean and standard deviations of the forward and backward corrosion potentials
obtained from the potentiodynamic scans (data for porous titanium specimens extracted from Refs.
[19,28]

Specimen Titanium±graphite Porous titanium

(8%)

0.9% NaCl saline Ringer's solution Ringer's solution

Compacting Pressure Ecorr, forw Ecorr, back Ecorr, forw Ecorr, back Ecorr, forw Ecorr,back
(tons) (mV) (mV) (mV) (mV) (mV) [28] (mV) [19]

5 ÿ57 (22) 707 (39) ÿ76 (24) 588 (34) ± 542

10 ÿ17 (51) 611 (81) ÿ34 (27) 592 (30) ÿ235 433
14 ÿ5 (53) 584 (82) ÿ36 (7) 560 (80) ± 229
18 ÿ4 (22) 530 (85) ÿ14 (10) 505 (91) ± 81
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 497

Table 5
Summary of the mean and standard deviations for the corrosion current densities and corrosion rates
obtained from the potentiodynamic scans

Specimen Titanium±graphite
(8%)

0.9% NaCl saline Ringer's solution

Compacting pressure Icorr Corrosion rate Icorr Corrosion rate

(tons) (mA cmÿ2) (mm/year) (mA cmÿ2) (mm/year)

5 29.3 (3.1) 255 (27) 18.9 (4.1) 163 (35)

10 16.6 (2.5) 144 (21) 7.1 (2.7) 61 (23)
14 10.0 (1.9) 86 (16) 3.0 (1.3) 26 (11)
18 6.7 (1.7) 58 (15) 2.6 (0.8) 22 (7)

pitting corrosion. Instead, the observed hysteresis is thought to be due to the gas
evolving from the oxygen evolution reaction forcing out some free graphite
particles, thus increasing the surface area (see later). The rate of oxygen evolution
was greatest on the 5 ton specimens (highest current density) hence, more graphite
particles are likely to be moved in these than the other specimens.
The corrosion potentials obtained on the forward scan (Ecorr, forw: the right
hand minimum in Figs. 7 and 8) are more negative than those found in the open-
circuit experiment (compare Tables 3 and 4). This could be due to the natural
oxide ®lm being either partially reduced or damaged at the initial negative
potentials. Unlike the open-circuit Ecorr measurements, the values of Ecorr, forw
generally became more positive with increasing compaction. Examination of the
optical photographs shown in Fig. 5 indicates that the percentage of the periphery
of the pores that comprises titanium carbide increases with increasing compaction
pressure. Since the remaining percentage of the pore wall is predominantly free
titanium, this may mean that the true surface area of titanium decreases with

Table 6
Comparison of the mean (and standard deviations) current densities ¯owing in the pseudo passive
range in the potentiodynamic scans of titanium based materials in 0.9% saline solution (data for porous
titanium extracted from Ref. [31])

Current density at speci®ed potential (mA cmÿ2)

Specimen Titanium-8%graphite Porous titanium Solid titanium

Compacting pressure (ton) 600 (mV) 1100 (mV) 600 (mV) 1100 (mV) 600 (mV) 1100 (mV)

5 694 (132) 1930 (30) 351 593 2.0 1.8

10 442 (39) 1104 (90) 48 77 ± ±
14 222 (42) 611 (154) 43 53 ± ±
18 130 (31) 353 (24) 15 28 ± ±
498 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

increasing compaction pressure. One possible reason why there is no dependence

of Ecorr on compaction pressure found earlier is that, some of the pores may have
initially been blocked during the mechanical grinding process. On evolving
hydrogen at negative potentials, these blockages could have been cleared exposing
more free titanium. It is likely that the lower compaction pressure specimens
would have had the highest number of blockages due to the larger pore diameters
in these composites.
As the titanium in chloride solutions generally shows passive behaviour, it can
usually be assumed that the more positive the corrosion potential adopted, the
lower the corrosion rate. This is because, a positive shift in corrosion potential
will both reduce the driving force of the cathodic reaction and increase the
thickness of the passive oxide ®lm [29]. A thicker ®lm represents an increase in the
barrier for metal dissolution and thus can be expected to reduce the corrosion
rate. However, these remarks only apply to composites of the same composition,
since changing the graphite content can alter the rate of the cathodic supporting
reaction, and thereby shift the corrosion potential in either direction without
necessarily changing the rate of the anodic corrosion reaction.
The corrosion potentials obtained on the return scan (Ecorr, back: the left hand
minimum) are several hundred mV more positive than their respective forward
scan values. This re¯ects the formation of the stable oxide ®lm at positive
potentials. The data in Table 4 shows that when like compaction pressures and
like solutions are compared, the Ecorr, back values of titanium±graphite are positive
of porous titanium compacts. However, the general trend of decreasing Ecorr, back
with increasing compaction is much less pronounced in the titanium±graphite
composites than has been reported for porous titanium [19].
The trend of decreasing Ecorr, back with increasing compaction pressure (the
exact opposite to the general trend observed for Ecorr, forw) suggests that the degree
of corrosion protection o€ered by the anodic oxide ®lms decreases with increasing
compaction. For porous titanium, this trend has previously been explained in
terms of de®cient supplies of dissolved oxygen within the pores leading to a
reduction in oxide thickness as the compaction pressure increases (smaller pores)
[15].
However, an alternative explanation for the dependence of Ecorr, back on
compaction pressure can be deduced from de Levie's porous electrode model [30],
in which it is shown that electrode potential within the pore decreases with depth,
and that the rate of decrease is a function of the lateral radius of the pore. An
increase in the compaction pressure has been found to decrease the pore radius
(Fig. 5). Hence, the electrode potential at the base of pores in a highly compacted
specimen can be expected to be less than in a light compacted one. Since oxide
thickness is directly dependent on the magnitude of the applied potential, the
oxide near the base of pores in highly compacted specimens should be thinner,
and therefore less protective, than that found in the pores of lightly compacted
samples. In the hypothetical extreme, the compaction pressure could be increased
to the extent where the size of the pores is reduced to the level where aqueous
solution is unable to penetrate, the value of Ecorr, back should only be dependent
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 499

on the thick oxide external of the pores and should thus be expected to register a
much more positive value. This situation is analogous to a solid, and indeed it has
been reported that the Ecorr, back of solid titanium is more positive than that of
any of the porous compacts [19]. If it is assumed that the pressure (P ) within the
pores is one atmosphere and that the surface tension …g† of the aqueous electrolyte
is about 0.07 N mÿ1, solution will not be able to penetrate pores with a radius
…2g =r † less than about 1 mm. Both explanations for the trend in Ecorr,back values
suggest a thinner oxide ®lm within the pores, and it is likely that in practice the
two e€ects occur simultaneously.
Corrosion current densities and corrosion rates can often be obtained from
plots of applied potential versus log current density by extrapolating linear regions
in either the anodic or cathodic (or both) current back to the corrosion potential
[16]. These linear regions correspond to the potential regimes where the
electrochemical reactions occurring at the electrode's surface are purely under
electron transfer control, a process referred to as Tafel behaviour. However, for
true Tafel behaviour the linear regions should extend over at least two orders of
magnitude of current and should have a slope not greater than 120 mV per
decade of current (higher values being an indication of some mass transport
control) [27]. These conditions were not satis®ed in any of the plots generated for
the titanium±graphite composites (Figs. 7 and 8). However, very short (less than
half an order of magnitude of current) linear regions were usually seen in the
cathodic currents, and occasionally in the anodic currents, extrapolation of these
back to Ecorr,forw does allow an estimation of the corrosion current densities to be
obtained (Table 5).
Because the slopes of these regions were always much greater than 120 mV
decÿ1 (typically 300 mV decÿ1) the estimated corrosion current densities given in
Table 5 (and the subsequent corrosion rates calculated from these) are likely to be
higher than the true values. Although the corrosion rates in Table 5 are probably
over-estimations, it is likely that the errors in obtaining the corrosion current
densities are consistent for all the samples and hence conclusions based on
comparisons between the data should still to be valid.
The corrosion current densities and corrosion rates tabulated in Table 5 clearly
decrease with increasing compaction pressure. This could be as much due to the
much larger real surface areas obtained at low compaction pressures [31], as to
any true dependence of corrosion rate on compaction pressure. Similar
polarisation scans using grade 2 solid titanium revealed a corrosion rate of about
1.7 mm yearÿ1 in reasonable agreement with the 1 mm yearÿ1 or less that has
previously been reported for the corrosion of solid titanium in a neutral non-
reducing environments [16,32]. Table 5 thus shows that the titanium±graphite
composites corrode at a rate up to a factor of two hundred higher than solid
titanium. It is unlikely that such a large increase is solely due to an increase in
surface area. Attempts are currently being made to accurately assess the true
e€ective surface areas (that is the area directly in contact with the aqueous
electrolyte) of the compacted specimens, although to date these have proved
unsuccessful.
500 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

Corrosion rates in Ringer's solution were about a factor of three less than in
0.9% NaCl, suggesting that the earlier observed cathodic shift in Ecorr on
changing from saline to Ringer's solution was due to the lactate in the latter
inhibiting the cathodic reaction.
Although the corrosion rates shown in Table 5 do not represent true metal
penetration rates (since the true surface areas are unknown), these do indicate the
amount of metal dissolution per unit external surface area, an important
parameter in the design of surgical implants [33]. Therefore the increased metal
dissolution observed with lower compaction pressures and higher graphite levels
may o€set some of the potential advantages of a highly porous composite implant
material (e.g. low friction). Hence the optimum levels of graphite and compaction
pressures are likely to have been compromised.
Referring back to the potentiodynamic scans shown in Figs. 7 and 8, it can be
seen that the porous titanium±graphite specimens did not show ideal passive
behaviour. This is normally characterised by a region of low current, the value of
which is independent of the applied potential, i.e. a region in which di=dV ˆ 0
[16,32]. Although the value of di=dV signi®cantly decreases at around +600 mV,
indicating the formation of an oxide ®lm, it does not fall completely to zero. Solid
titanium does show such ideal passive behaviour. However, previous work on
porous titanium revealed results similar to those of porous titanium±graphite,
with the dependence of the magnitude of the current density on compaction
pressure being explained in terms of the variation in surface area [7].
A comparison of the current densities measured within the pseudo-passive
region (600 to 1100 mV) on titanium-8%graphite in 0.9% NaCl saline solution
with previously published data [31] for porous titanium is shown in Table 6. This
reveals that the presence of the graphite caused the current density to increase by
up to an order of magnitude (up to 3 orders of magnitude higher than the solid
titanium). This is more than can be reasonably explained by any di€erence in the
true surface areas of the specimens. It is thus concluded that the presence of
graphite caused an increase in the titanium's corrosion rate rather than a decrease
as originally expected. The reason for a lack of ideal passive behaviour in porous
specimens has not yet been fully explained. However, it is likely to be related to
the variation in electrode potential with depth as mentioned above [30].
Above 1300 mV the value of di=dV was observed to increase for the titanium±
graphite specimens, due to oxygen evolution (the evolution of gas bubbles being
observed). The presence of graphite, which is a much more ecient electrode for
this reaction than titanium, meant that the currents reached in this region were
much higher than would be expected on pure titanium specimens. The oxygen
evolution also caused black particles to be dislodged from the composites. After
the cyclic polarisation test had been completed, the surfaces of all the specimens
seem to have been roughened with dispersed black particles on them, some of
which appeared to be embedded in the surface. Analysis of the free black particles
by nuclear microprobe revealed a predominance of graphite, with some titanium
dioxide. It is likely, this graphite was dislodged from within the pores of the
composite, where it may have been blocking the small channels earlier postulated
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 501

to interconnect the larger pores. The removal of the graphite could thus allow the
electrolytic solution access to a larger area of titanium metal. It is possible that
oxygen gas evolution within the pores could have led to the build-up of sucient
pressure to locally rupture the composite, and this could account for the observed
roughening.
The release of graphite debris from the composite is a worrying concern, since
its physiological e€ects on the surrounding tissue are unknown. One possible
solution to this problem is to increase the length of time used to sinter the
composite. This would convert more, or all, of the graphite to titanium carbide.
Although, this would produce a material without the advantage of a low friction
coecient, it would still have a higher resistance to wear than porous titanium. At
the same time, reducing the free graphite content would also reduce the galvanic
corrosion problems associated with the current material.

4. Conclusions

Sintering titanium±graphite compacts produced a triphasic porous material,

with titanium carbide being the third constituent. Some evidence was obtained
that allowed it to be postulated that the percentage of titanium converted to
carbide increases with compaction pressure. It was found that the extent of
densi®cation resulting from the sintering process decreased with increasing
compaction pressure or increasing graphite content.
The pore morphology within the titanium±graphite specimens appears to consist
of a number of large chambers in the region of 100 mm diameter connected by
smaller channels with diameters around 10 mm. As might be expected, increased
compaction pressures caused the pores in the titanium±graphite system to become
smaller and more isolated.
The presence of graphite in the titanium composites caused the corrosion
potential to shift in the positive direction. Although such an e€ect should help
stabilise the oxide ®lm and decrease corrosion rate of a passive metal, the porous
structure of the titanium±graphite composites prevented the complete passivation
of the titanium. As a result, the presence of the graphite caused an increase in the
titanium's corrosion rate.
Increasing the compaction pressure was found to reduce the corrosion rate of
the composites, although even the specimens compressed at the highest pressures
used (18 ton) still had corrosion rates signi®cantly above that of solid titanium.
The composite was observed to release graphite debris during oxygen evolution.
Changing the sintering process in order to convert more of the free graphite into
titanium carbide could possibly prevent this. Although, this would produce a
material without the advantage of a low friction coecient, it would still have a
higher resistance to wear than porous titanium.
In summary, it appears the extent to which graphite can be used in porous
titanium biomaterials will have to be a compromise between the bene®ts of
improved friction and wear properties and the disadvantage of increased corrosion
502 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503

rates. However, if a method could be found to induce porous titanium based

materials to simulate ideal passive behaviour in body ¯uids, perhaps by alloying
or anodising prior to implanting, this disadvantage could be eliminated.
Alternatively, higher compaction levels could be used to reduce the corrosion rate
of the composites. However, increased compaction is likely to lead to surgical
implants with a higher elastic modulus and poorer bone-ingrowth characteristics.

Acknowledgements

The authors would like to thank /Prof. Frank Watt and Dr. Thomas Osipowicz
of the Department of Physics, National University of Singapore, for their analysis
of debris particle with the nuclear microprobe.

References

[1] M. Long, H.J. Rack, Biomaterials 19 (1998) 1621.

[2] J.A. Savio, L.M. Onecamp, J. Black, Clin. Mater 15 (1994) 101.
[3] M. Spector, M.J. Michno, W.H. Smarook, G.T. Kwiatkowski, J. Biomed. Mater. Res 12 (1978)
665.
[4] L.C. Lucas, J.E. Lemons, J. Lee, P. Dale, J.E. Lemons (Ed.), Quantitative Characterization and
Performance of Porous Implants for Hard Tissue Applications, ASTM STP 953, American
Society for Testing and Materials, Philadelphia, 1987, p. 137.
[5] K.J. Bundy, R.E. Luedemann, J.E. Lemons (Ed.), Quantitative Characterization and Performance
of Porous Implants for Hard Tissue Applications, ASTM STP 953, American Society for Testing
and Materials, Philadelphia, 1987, p. 137.
[6] S.H. Teoh, R. Thampuran, K.H.W. Seah, Wear 214 (1998) 237.
[7] T.P. Hoars, D.C. Mears, Proc. R. Soc. London, Ser. A 294 (1966) 486.
[8] J. Pan, D. Thierry, C. Leygraf, Electrochim. Acta 41 (1996) 1143.
[9] T.J. Krause, F.M. Robertson, R.S. Greco, J. Biomed. Mater. Res 27 (1993) 65.
[10] B. Waelivaara, I. Lundstroem, P. Tengvall, Clin. Mater 12 (1993) 141.
[11] J. Pan, D. Thierry, C. Leygraf, J. Biomed. Mater. Res 30 (1996) 393.
[12] J. Pan, D. Thierry, C. Leygraf, J. Biomed. Mater. Res 28 (1994) 113.
[13] D.F. Williams, in: M.J. Donachie (Ed.), Titanium and Titanium Alloys, vol. 1, American Society
for Metals, Metals Park, OH, 1982, p. 12.
[14] D.J. Blackwood, R. Greef, L.M. Peter, Electrochim. Acta 34 (1989) 875.
[15] K.H.W. Seah, X. Chen, Corros. Sci 34 (1993) 1841.
[16] M.G. Fontana, in: Corrosion Engineering, 3rd ed, McGraw-Hill, New York, 1986, p. 491.
[17] M.J. Donachie (Ed.), Titanium a Technical Guide, American Society for Metals, Metals Park,
Oh, USA, 1988, p. 21.
[18] ``Standard guide for electrolytic polishing of metallographic specimens: E1558-93''. In Annual
Book of ASTM Standards. American Society for Testing and Materials, Philadelphia, 1993, vol.
03.01.
[19] K.H.W. Seah, R. Thampuran, S.H. Teoh, Corros. Sci 40 (1998) 547.
[20] P. Kovacs, R.D. Jamison, L.A. Gilbertson (Eds.), Composite Materials for Implant Applications
in the Human Body: Characterisation and Testing, ASTM STP 1178, American Society for
Testing and Materials, Philadelphia, 1993, p. 41.
[21] G.R. Maharaj, R.D. Jamison, R.D. Jamison, L.A. Gilbertson (Eds.), Composite Materials for
 D.J. Blackwood et al. / Corrosion Science 42 (2000) 481±503 503

Implant Applications in the Human Body: Characterisation and Testing, ASTM STP 1178,
American Society for Testing and Materials, Philadelphia, 1993, p. 86.
[22] R.F. Burton, in: Ringer Solutions and Physiological Salines, Wright, Bristol, UK, 1975, p. 1.
[23] R.J. Solar, B.C. Syrett, A. Acharya (Eds.), Corrosion and Degradation of Implant Materials,
ASTM STP 684, American Society for Testing and Materials, Philadelphia, 1979, p. 259.
[24] F. Watt, I. Orlic, K.K. Loh, C.H. Sow, P. Thong, S.C. Liew, T. Osipowicz, T.F. Choo, S.M.
Tang, Nucl. Instrum. Methods Phys. Res., Sect. B 85 (1994) 708.
[25] B.H. Amstead, P.F. Ostwald, M.L. Begeman, in: Manufacturing Processes, 8th ed, Wiley, New
York, 1987, p. 196.
[26] Lee, C.T.G., 1997. B. Eng. Thesis, Department of Mechanical and Production Engineering,
National University of Singapore.
[27] Southampton Electrochemistry Group, in: R. Greef, R. Peat, L.M. Peter, D. Pletcher, J.
Robinson (Eds.), Instrumental Methods in Electrochemistry, Ellis Horwood, Chichester, 1985, p.
233.
[28] Chooi, S.K.M., 1998. B.Sc. Thesis, Department of Materials Science, National University of
Singapore.
[29] K.V. Ashok, in: Electrochemistry of Metals and Semiconductors, Marcel-Decker, New York,
1973.
[30] R. de Levie, in: P. Delahay (Ed.), Advances in Electrochemistry and Electrochemical Engineering,
vol. 6, Wiley-Interscience, New York, 1967, p. 329.
[31] K.H.W. Seah, R. Thampuran, X. Chen, S.H. Teoh, Corros. Sci 37 (1995) 1333.
[32] D.J. Blackwood, L.M. Peter, D.E. Williams, Electrochim. Acta 33 (1988) 1143.
[33] J. Kruger, B.C. Syrett, A. Acharya (Eds.), Corrosion and Degradation of Implant Materials,
ASTM STP 684, American Society for Testing and Materials, Philadelphia, 1979, p. 107.