J O U R N A L O F M AT E R I A L S S C I E N C E : M AT E R I A L S I N M E D I C I N E 1 5 ( 2 0 0 4 ) 1 9 1 ± 1 9 7

Processing of grain-size functionally gradient

bioceramics for implant applications
K. MORSI*
Department of Mechanical Engineering, San Diego State University, 5500 Campanile Drive,
San Diego, California 92182, USA
E-mail: kmorsi@mail.sdsu.edu
H. KESHAVAN
Department of Mechanical and Aerospace Engineering, University of Missouri, Columbia,
MO 65211, USA
S. BAL
Department of Orthopedic Surgery, University of Missouri, Columbia, MO 65211, USA

This paper reports work on the processing of functionally gradient alumina bioceramics with
a continuously decreasing grain size across the thickness, with the view of ultimately utilizing
high-quality nano/ultra®ne powders only at the surface of an implant to provide superior
wear and mechanical properties. A model of disc geometry is used to examine the feasibility
of producing this brand of materials. Wet processing/ball milling and sequential slip casting
procedures were used to de-agglomerate alumina powders and deposit green layers of
varying particle sizes from 50 to 250 nm. Both pressure-less sintering and hot pressing were
evaluated as high temperature sintering/consolidation processes. The results indicate that
pressure-less sintering may not be suitable. Hot pressing, however, achieved very promising
results producing near fully dense product with a grain size that gradually changes across its
thickness.
# 2004 Kluwer Academic Publishers

Introduction currently underway at the University of Missouri

Arti®cial joint replacement has advanced considerably Hospital at Columbia. Ceramic bearings presently used
over the past three decades. Over 300 000 total hip and in hip replacement are composed of micrometer-sized
knee replacements are performed annually in the United grains. Particulate wear generated from such bearings
States to restore function of diseased and damaged could be signi®cantly reduced if nano/ultra®ne grains
articular tissue [1]. The combination of the cobalt± were used at the bearing surface. The drive toward the
chrome femoral head and ultra high molecular weight use of ultra®ne-grained materials and nanomaterials
polyethylene (UHMWPE) acetabular liner re¯ects the results mainly from the improvements in mechanical and
current standard of care in arti®cial hip replacement [2]. wear properties they can offer [4±9]. The mechanical
Despite predictable success for the 10±15 years properties of nanoceramics such as hardness, strength
following the implantation of an arti®cial hip, late and toughness are generally superior to those of
aseptic loosening of the components is virtually conventional ceramics [10]. It is well known that the
inevitable. Clinical results very clearly show that wear characteristics of conventional ceramics can be
UHMWPE wear debris is currently creating problems improved by reducing grain size [11, 12], and even more
in the young and active patients. Eventually, these so by the use of nanoceramics [13]. In fact, the wear
sequences of events lead to loosening and failure of the resistance of nano-composite alumina is up to two orders
prosthesis, requiring major repeat surgery. Efforts to of magnitude superior to that of conventional alumina
increase the longevity of arti®cial hip replacements have [13]. For these reasons, the study of ultra®ne/nano-
focused mainly on reducing the quantity of wear grained ceramic materials for bearing application is of
generated at the bearing surfaces of these implants. high relevance and signi®cant importance to a number of
Improvements in the quality of the metal-on-UHMPE industries in addition to the biomedical (e.g. the cutting
articulation have contributed to reduced wear, and tool industry). One of the main impediments to the
alternative bearing surfaces such as ceramic-on-ceramic sintering of nanopowders to high density has been grain
have demonstrated some of the lowest wear rates growth, resulting in the loss of the nanostructure.
reported in the literature [3]. Clinical trials with total Another major drawback in the application of nano-
hips that have an alumina-on-alumina bearing are materials is the general higher cost of the precursor nano-
*Author to whom all correspondence should be addressed

0957±4530 # 2004 Kluwer Academic Publishers 191

powders used compared to conventional ceramic 1350  C for 2 h with heating rate 20  C/min to 850  C and
powders. Functionally gradient materials (FGMs) are 1  C/min to the sintering temperature. Hot pressing was
materials where the composition gradually or step-wise conducted in a graphite die at 1350  C for 30 min and
changes from the bulk to the surface [14], with the aim of 1400  C for 30 min. In both hot pressing cycles, an initial
gradually changing the material properties from bulk to pressure of 8 MPa was maintained during heating. The
the surface. FGMs have been produced in the past by a heating rate was 5  C/min from room temperature to
variety of processes including chemical vapor deposition 750  C, then the compact was heated at 30  C/min to the
(CVD), plasma spraying, hot pressing [15] and powder hot pressing temperature where the pressure was
stacking methods. Although the CVD and plasma spray- increased to 30 MPa and maintained for 30 min. The
ing processes can generate layers of controlled thickness, pressure was then released and specimens left to cool in
they are, however, expensive production routes com- the die to room temperature. Green densities were
pared to conventional process such as slip casting. In measured by weighing the green compacts and dividing
fact, sequential slip casting has been used successfully to by their volume (measured using a micrometer and
produce functionally gradient materials where composite calipers). Final densities of the sintered/hot pressed
layers of different reinforcement volume fractions are compacts were measured using the Archimedes prin-
deposited on one another [16±18]. Ceramic hip joint ciple. Throughout the paper, all the layers will be
materials can normally be formed through the production referred to by their initial particle sizes, i.e. 50 nm
of a green compact which is then subsequently sintered/ original particle size layer (layer A), 100 nm original
hot isostatically pressed [19]. Alumina (Al2 O3 ) is one of particle size layer (layer B), 150 nm original particle size
the most studied bioceramics due to its combined layer (layer C) and 250 nm original particle size layer
inertness and good mechanical properties [20±22], and (layer D). All sintered/hot pressed specimens were
was therefore used as our model material. Our aim is to sectioned centrally along the hot pressing direction in
apply sequential slip casting to the production of highly such a way as to expose the four layers of the cross-
dense functionally gradient materials with respect to section for polishing operations and microstructural
``grain-size'' i.e. with a transition in grain size from one observations. Specimens were then ground and subse-
side of the material to the other. In the current work a quently polished to a 15 mm, 6 mm, 1 mm and 50 nm
simple disc geometry model is used in order to study the diamond ®nish, each for a duration of 1 h. Micrographs of
sintering behavior, with the view of extending our work the polished and etched microstructures were taken using
to actual orthopedic parts. This paper reports the effect of a ®eld emission electron microscope. The area fraction of
hot pressing and pressure-less sintering on the micro- porosity in the central region within each layer was
structure of the ®nal grain-size functionally gradient measured using a Buehler image analysis software, ®ve
materials (GS-FGMs). Future strategies for achieving areas were analyzed for each layer and a standard
nanostructure at the surface are also identi®ed. deviation calculated. Grain boundaries were revealed by
thermal etching at a temperature 100  C lower than the
sintering/hot pressing temperature for 30 min. Grain size
Experimental procedure was measured using the linear intercept method and
Predominantly a-alumina powders (99.99% purity) multiplying by 1.56. All Knoop microhardness measure-
(Baikowski International Inc., Charlotte, NC) of sizes ments were conducted using a Buehler microhardness
50, 100, 150 and 250 nm were used in our experiments to tester using an indentation load of 500 g.
prepare separate aqueous slips. The solid volume
fractions for each slip were 20 vol % for 50, 100 and
150 nm and 30 vol % for 250 nm particles with the Results and discussions
balance being distilled water. An ammonium polyacry- Pressure-less sintering
late dispersant (Duramax2 D-3005, Rohm and Hass, PA) Fig. 1 shows a sequentially slip cast four-layer (50/100/
was added to each slip to enable good dispersion and de- 150/250 nm) green compact. The optimized four-layer
agglomeration of the powders (indicated by a minimum green compacts were ®rst pressure-less sintered at
in the apparent viscosity measured using a Brook®eld
viscometer) then ball milled for a period of 24 h. All slips
were then degassed in a vacuum desiccator for 10±
15 min to remove entrapped air bubbles left in the
ceramic slips after ball milling. These slips were then
sequentially slip cast in one inch (25.4 mm) internal
diameter short cylinders placed on a plaster of paris base
to form a functionally gradient ceramic green compact
with layers of varying particle sizes along the height. The
quantity of slip poured in each layer was pre-calculated
to obtain ®nal layers each approximately 1 mm thick
assuming full densi®cation after high temperature
consolidation. The resulting green layered compact was
then heated in an oven in air at a temperature of 60  C for
a period of 24 h to remove the residual moisture left in
the green compact. Both pressure-less sintering and hot Figure 1 Four-layer (50/100/150/250 nm) functionally gradient
pressing were examined. Sintering was conducted at ceramic slip cast specimen in the green state.

192
Figure 2 Percent theoretical density for layers A±D, green, pressure less sintered at 1350  C for 120 min, hot pressed at 1350  C for 30 min and hot
pressed at 1400  C for 30 min.

1350  C for 2 h. On removal from the furnace, the

sintered compact appeared intact, however, it appeared
slightly tapered from layer D to layer A, with layer A
having the smaller diameter. The green density of the
layered compact was 54.1 + 0.8% of theoretical
compared to a sintered density of 79.5% of theoretical
(3:98 g/cm3 as quoted by the manufacturer). It was clear
that sintering promoted densi®cation; however, the
individual layers were densi®ed to different degrees.
Fig. 2 includes data on the green and ®nal densities for
each layer. The percent theoretical density for each layer
was found by measuring area fraction of porosity within
each polished and un-etched layer. It can be seen from
Fig. 2 that layer A exhibited the highest density of the
four layers. This increased densi®cation can be
attributed to both a well-dispersed slip and a much Figure 3 Electron micrograph showing poor bonding at interface
higher particle surface area leading to a substantially between layers C and D for pressure-less sintered compact.
higher driving force for sintering than the remaining
layers. The particle surface area is decreased from layer Hot pressing
A to D leading to a decrease in the driving force for The previous section demonstrated the inherent problems
sintering resulting in a corresponding decrease in the associates with pressure-less sintering which includes
density. The overall low sintered density ( * 80% of low density of the ®nal product and poor bonding of
theoretical) of the layered compact raises doubt as to the layers. Hot pressing has the advantage of achieving better
suitability of pressure-less sintering as a viable consolidation by increasing the particle contact stresses
processing route for this class of materials. Higher during sintering. When hot pressing was conducted at
temperatures which should increase the density, how- 1350  C for 30 min, the resulting compact was found to
ever, will be at the expense of considerable grain have a bulk density of 95.6% with all layers perfectly
growth. Sintering under an applied pressure (hot bonded. Fig. 4 compares the thermally etched micro-
pressing) is seen as a more viable processing route. structures taken from the central region within each layer
On examination of the interfaces between different for of the 1350  C hot pressed and pressure-less sintered
layers it was also observed that poor bonding existed in a specimens. It can be seen that in general, the hot pressed
number of regions along the interface line for all the material has a higher density and smaller average grain
layers, an example of which is shown in Fig. 3 for the size. The grain growth observed for the pressure-less
interface between layers C and D. This is believed to be sintered specimen is a direct result of the prolonged
due to the inherent difference in shrinkage rates of the sintering time having been sintered for 90 min more than
individual layers due to their different particle sizes and the hot pressed counterpart.
therefore driving force for sintering, which in turn caused A bulk density of 95.6% will, however, unlikely yield
the poor bonding. These poorly bonded areas resulted in the optimum mechanical properties needed in hip joint
a higher average density when measured using image applications where products closer to full density would
analysis than that measured using the Archimedes be more desirable. An increase in the hot pressing
method, since the former does not take into account temperature by 50  C from 1350 to 1400  C resulted in an
poor interface bonding. almost fully dense (99.8%) ®nal compact. The percent
193
Figure 4 Electron micrographs of polished and thermally etched microstructure for layers A±D after hot pressing (a±d respectively) and pressure-less
sintering (e±h respectively) at 1350  C.

theoretical density of each layer for all the processed measured to be 382 + 98 nm. It is believed that this grain
materials is presented in Fig. 2. It can be seen that hot size can be reduced to below 100 nm by increasing the
pressing was essential in achieving high densities which pressure during hot pressing which would result in a
is highest at 1400  C (30 min). The micrographs in Fig. 5 reduction in the required hot pressing temperature, [23]
clearly demonstrates that highly dense materials with a and also the addition of grain growth inhibitors. The
®nal grain size that increases from the surface inward grain sizes of all layers for all the processed materials are
have been successfully produced by hot pressing at summarized in Table I. It can be seen that hot pressing at
1400  C for 30 min. 1400  C for 30 min achieved a similar grain size in each
The average grain size in the surface layer was layer to that pressure-less sintered at 1350  C for 120 min
194
Figure 5 Grain-size functionally gradient alumina ceramic produced by hot pressing at 1400  C for 30 min.

while at the same time achieving near full density. Hot the original particle size is greatest for the 50 nm layers
pressing at 1350  C gave the best results in terms of grain (layer A) and least for the 250 nm layers (layer D).
size; however, the bulk density of this specimen was only The differences in grain size from one side to the other
95.6% of theoretical. It is well known that the grain can result in differences in strength and hardness (the
growth rate is inversely proportional to grain size [24]. smaller the grain size the higher the strength and
Accordingly, it is expected that after sintering or hot hardness according to the Hall±Petch equation).
pressing, the ®ner the original particle size the larger the Microhardness measurements were therefore conducted
relative grain growth experienced. This was observed for to probe into the mechanical response of our near full
all our processed materials where the ratio of the ®nal density (i.e. 1400  C hot pressed) processed material. It is
grain size after hot pressing or pressure-less sintering to well known that in general, wear resistance may increase
195
T A B L E I Effect of pressureless sintering and hot pressing on the grain size of individual layers

Process Temperature Time Final grain size Final grain size Final grain size Final grain size
(  C) (mins) in layer A (nm) in layer B (nm) in layer C (nm) in layer D (nm)

Pressureless
sintering 1350 120 376 + 51 451 + 92 495 + 59 599 + 59
Hot pressing 1350 30 301 + 56 338 + 48 372 + 74 429 + 58
Hot pressing 1400 30 382 + 98 463 + 44 497 + 97 577 + 78

Figure 6 Micro-hardness as a function of distance beneath the surface of the 1400  C hot pressed disc, showing a smooth transition in the hardness
from surface to bulk layer (i.e. layers A±D, each layer * 1 mm).

with increase in hardness [25, 26] (although other Conclusions

properties including fracture toughness and Young's Experiments were conducted on the processing of
modulus need also to be considered). In the application bioceramics functionally graded with respect to grain
of the ®nal implant materials it is not only important that size. The following conclusions can be drawn:
the surface hardness be maximized through smaller grain
sizes (on the nano-scale) but that there should preferably 1. Pressure-less sintering was found unsuitable under
be no abrupt change in properties from the surface the investigated processing conditions for producing
inwards. Fig. 6 is a hardness pro®le taken from the cross- functionally gradient ceramics with respect to grain size
section of the polished hot pressed material. It is clear owing to poor densi®cation and poor bonding at the
that the hardness changes gradually from one side of the interfaces between layers.
specimen to the other. It must also be mentioned that 2. Hot pressing provided increased densi®cation and
some carbon may have diffused into the specimen good bonding at interfaces; density was highest for the
through the contact with graphite which may have 1400  C hot pressed material.
possibly in¯uenced the hardness. Nevertheless, our 3. The ®nal grain size was successfully made to
material does exhibit a decrease in ®nal average grain decrease gradually from one side of the sintered and hot
size from one side to the other which according to the pressed compacts to the other side, with a hardness
Hall±Petch equation should exhibit a change in hardness. pro®le exhibiting a gradual increase from one side of the
The maximum hardness of approximately 26.9 GPa was specimen to the other.
observed for the 50 nm original particle size layer 4. The extent of grain growth was found to decrease
(having a ®nal grain size of 382 + 98 nm, Table I). from the ®ner to the coarser particle size layers for all our
This is signi®cantly higher than a reported value of processed materials.
* 19 GPa for * 2 mm grain size alumina [27] (a typical 5. The results may provide important implications for
grain size of implanted alumina [28]). This may be the production of grain-size functionally gradient
re¯ected in an increase in wear resistance however; this orthopedic implants with superior wear resistance. This
needs to be veri®ed through a detailed examination of will be investigated and con®rmed through characteriza-
wear properties of these materials (the subject of on- tion of mechanical and wear properties of optimized
going work). materials (the subject of on-going work).
196
Acknowledgments 15. A . K AWA S A K I and R . WATA N A B E , in ``Ceramics: Charting the

The authors would like to thank the Missouri Foundation future: Advances In Science and Technology'', Proceedings of the
8th CIMTEC-World Ceramics Congress and Forum on New
for Medical Research for their much appreciated support Materials, Florence, Italy, 28 June 1994, p. 1715.
in the initiation of this work through grant no. 041801. 16. J . S . M O Y A , A . J . S A N C H E Z - H E R E N C I A , J . R E Q U E N A and
R . M O R E N O , Mater. Lett. 14 (1992) 333.
17. I . M . G . D O S S A N TO S , R . C . M . M O R E I R A , E . R . L E I T E , E .
LO N G O and K . A . VA R E L A , Ceram. Int. 27 (2001) 283.
References 18. H . TA K E B E and K . M O R I N A G A , Mater. Manuf. Proc. 9 (1994)
1. D. A. PULEO and A . NANCI, Biomaterials 20 (1999) 2311.
721.
2. S . B . B A L , D . VA N D E L U N E , D . M . G U R B A and W. H .
19. I . S TA M E N KO V I C and A . S A LO M O N I , Ceram. Acta 10 (1998)
H A R R I S , J. Arthroplasty 13 (1998) 492.
11.
3. E . J . H E N S S G E , I . B O S , G . W I L L M A N N , J. Mater. Sci.-Mater.
20. K . M O R S I , ``Mechanical Properties of Particle Reinforced
Med. 5 (1994) 657.
Alumina'' (Ph.D. Thesis, University of Oxford, UK, 1996).
4. C . S U R Y A N A R AY A N A , J.O.M. 54 (2002) 24.
21. R . I . TO D D , K . M O R S I and B . D E R B Y, Br. Ceram. Proc. 57
5. W. L . G AO et al., J. Inorg. Mater. 15 (2000) 1005.
(1997) 87.
6. M . M I T O M O , Y O U N G - W O O K K I M and H . H I R OT S U R U ,
22. K . M O R S I and B . D E R B Y, Br. Ceram. Trans. 98 (1999) 72.
J. Mater. Res. 11 (1996) 1601.
23. R . S . M I S H R A , E . L . C H A R L E S and K . M . A M I Y A , J. Am.
7. X . H . J I N , L . G AO , L . H . G U I and J . K . G U O , ibid. 17 (2002)
Ceram. Soc. 79 (1996) 2989.
1024.
24. W. D . K I N G E R Y, H . K . B OW E N and D . R . U H L M A N N , in
8. H A E - W O N K I M , YO U N G - H A G KO H and H Y O U N - E E K I M ,
``Introduction to Ceramics'', 2nd edn. (Wiley Interscience New
ibid. 15 (2000) 1478.
York, 1976) p. 454.
9. A . S . E D E L S T E I N and R . C . CA M M A R ATA , in ``Nanomaterials:
25. K . H . W. S E A H , R . S . K U L K U R N I , S . C . S H A R M A and A .
Synthesis, Properties and applications'' (Institute of Physics
R A M AC H A N D R A , J. Mater. Sci. Tech. 12 (1996) 357.
Publishing, London, 1998) p. 332.
26. C . J . O H , S . L E E and M . G . G O L KO U S K I , Met. Mater. Trans. A
10. A . S . E D E L S T E I N and R . C . CA M M A R ATA , in ``Nanomaterials:
32 (2001) 2995.
Synthesis, Properties and applications'' (Institute of Physics
27. S . D . S K R OVA N E K and R . C . B R A D T, J. Am. Ceram. Soc. 62
Publishing, London, 1998) p. 188.
(1979) 215.
11. C . P. D O G A N and J . A . H AW K , Wear 225 (1999) 1050.
28. J . L . K AT Z , Orthopedic applications, in ``Biomaterials ScienceÐ
12. A . K R E L L and D . K L A F F K E , J. Am. Ceram. Soc. 79 (1996)
An Introduction to Materials in Medicine'', edited by B. D.
1139.
Ratner, A. S. Hoffman, F. J. Schoen and J. E. Lemons, (Academic
13. J . R O D R I G U E Z , ibid. 82 (1999), 2252±2254.
Press, California, 1996) p. 343.
14. Y. M I YA M O T O , W. A . K AY S S E R , B . H . R A B I N , A .
K AWA S A K I and R . G . F O R D , in ``Functionally Graded
Materials: Design, Processing and Applications'' (Kluwer Received 31 December 2002
Academic Publishers, Boston,1999) p. 1. and accepted 3 April 2003

197