Pergamon J Biumechanics. Vol. 2X. No. 3. pp. 309 320.

IYY5
Copyright rl‘ 1994 Elsevter Smnce LI*J
Printed in Great Britain. All rights res‘ervsd
0021 9?90.‘95 SY 50 4 0.00

0021-9290(94)0006%7

FRACTURE TOUGHNESS OF HUMAN BONE UNDER

TENSION
T. L. Norman,* D. Vashishtht and D. B. BurrS
*Departments of Mechanical and Aerospace Engineering and Orthopedics, West Virginia University,
Morgantown, WV 26505, U.S.A.; tInterdisciplinary Research Center in Biomedical Materials, Queen Mary
and Westfield College, University of London, London, U.K.; IDepartments of Anatomy and Orthopedic
Surgery, Indiana University and Biomechanics and Biomaterials Research Center, IUPUI, Indianapolis.
IN 46202, U.S.A.

Abstract--The longitudinal fracture toughnesses of human cortical bone were compared to those of bovine
cortical bone to test the hypothesis that although human osteonal bone is significantly weaker and more
compliant than primary (plexiform) bone, it is not less tough than primary bone. The fracture toughness
indices, critical strain energy release rate (G,) and critical stress intensity factor (K,), were determined for
human Haversian bone and bovine bone under tension (Mode I) loading using the compact tension
method. The effects of thickness, crack growth range and anisotropy on fracture indices for slow stable
crack growth in cortical bone were determined. Plane strain assumptions required for application of linear
elastic fracture mechanics (LEFM) to bone were investigated. Longitudinal oriented fracture toughness
tests were used to assess the crack inhibiting effect of human bone microstructure on frature resistance.
Human bone I<, calculated from the stress concentration formula for 2 and 3 mm thick specimens equaled
4.32 and 4.05 MN m-3iz, respectively. Human bone G, calculated from the compliance method equaled
827 N mm ’ for 2 mm thick specimens and 595 N m- ’ for 3 mm thick specimens. It was found that crack
growth range, thickness and material assumptions affect fracture toughness. K, calculated from G, using an
anisotropic relation provided the lowest estimate of K, and equaled 3.31 MNme3’* for 2 mm thick
specimens and 2.81 MN m- 3’2 for 3 mm thick specimens. Both K, and G, were significantly reduced after
being adjusted to ASTM standard thickness using ratios determined from bovine bone. The fracture
toughness of bovine bone relative to human bone ranged from 1.08 to 1.66. This was compared to the
longitudinal strength of bovine bone relative to the longitudinal strength of human bone which is
approximately equal to 1.5. We found that even though human bone is significantly weaker than bovine
bone, relative to its strength, the toughness of human and bovine bone are roughly similar, but the data
were not sufficiently definitive to answer the question of which is tougher.

Keywords: Fracture toughness, crack propagation, human bone, bovine bone, fracture mechanics.

NOMENCLATURE Burr (1989) proposed that, although human osteonal

bone is weaker and more compliant than primary
crack length, measured from the vertical axis bone, it is not less tough than primary bone. It is not
through the loading pins unusual for reduced material stiffness in composite
thickness of the specimen
compliance materials to improve the material’s fatigue properties
crosshead displacement (Fine and Ritchi, 1978).
Young’s modulus in the longitudinal, circum- The matrix of tough fiber-reinforced composite ma-
ferential and radial directions terials transfers internal loads around and between
in-plane shear modulus
critical strain energy release rate fibers (Beardmore et al., 1980; Kelly and Davies,
critical stress intensity factor 1976), providing the greatest toughness for a given
critical load strength (Cooke et al., 1973). Fiber-reinforced com-
specimen width dimension, measured from the posites, like bone, are often highly anisotropic and
vertical axis through the loading pins have low transverse strength because of the direc-
shape factor
poisson’s ratio tionality of the fibers (Beardmore et ul., 1980). This
yield stress design is a compromise between competing needs:
high strength in the predominant load direction. but
not at the expense of properties in the transverse and
INTRODUCTION thickness directions.
Haversian bone is similar in its osteonal arrange-
Cortical bone fracture toughness, or bone’s resistance ment to these very tough fibrous composite laminate
to crack growth, is an important factor in resistance of materials reinforced with discontinuous fibers. Haver-
bone to monotonic and fatigue failure. Martin and sian bone is both laminated, and also contains large
numbers of osteons, each having poor shear strength
and a weak osteon-matrix interface (cement line). The
Received in jinal form 2 May 1994. cement line provides many sites for crack arrest (Burr

309
310 T. L. Norman er al
~‘f al., 1988: Currey, 1962); delamination increases
fracture toughness (Ascenzi and Bonucci, 1967; Mar-
gel-Robertson, 1973; Piekarski, 1970; Pope and
Murphy, 1974: Simkin and Robin, 1974). Corondan
and Haworth (1986) showed that crack propagation
in bone is inhibited by increased numbers of osteons
and by larger osteons. Comparing the morphology of
fractured surfaces of bone with that of sections adjac-
ent to the fracture, they found that bone tends to
break where there are fewer osteons and where os-
teons are smaller. Bone did not break in adjacent
regions where osteon density and size were greatest.
Fracture toughness of human bone in which Haver-
sian systems predominate is not well defined. Based
on the similarities between bone and other fiber-rein-
forced composites that are tough, we hypothesized
that the resistance of human bone to crack growth
was equivalent to that of bovine plexiform bone. Be-
cause specimens from human bone cannot meet
American Society for Testing and Materials (ASTM)
size requirements, a second goal of this work was to
investigate the effects of specimen thickness and crack
length on fracture toughness measurements. A third
goal was to investigate the effect of material property
assumptions on calculated fracture toughness.
EXPERIMENTAL PROCEDURE
The two commonly used fracture toughness indices
are derived from linear elastic fracture mechanics
Lateral
Posterior
a
Anterior
Medial
,- - 7 r - -,
(a)
Fig. 1. Compact tension specimen: orientation with respect to tibia (a) and configuration indicating
dimensions (b). Distance markers were placed on the compact tension specimen with a lead pencil at
distances of 7,8,9,10,11,12 and 13 mm (a,-a,). The specimen width dimension, W, is equal to 14 mm.
(LEFM); the critical stress intensity factor (K,j de-
scribes the stress field near a crack tip. and the critical
strain energy release rate ((;,) measures the energy
required to extend a preexisting crack. Both depend
on load level, crack geometry, and mode of fracture:
Mode I (opening mode). Mode IT (shearing mode),
and Mode III (tearing mode). Linear elastic fracture
mechanics requires the assumption that human bone
is linearly elastic and a continuum. Specimen size
must be restricted to satisfy plane strain assumptions.
One reason that fracture toughness of human bone
in which Heversian systems predominate is not well
defined is that specimens obtained from human bone
cannot meet ASTM size requirements. Cortical thick-
ness and bone curvature reduce the size of test speci-
mens that can be cut from human bone. Relatively
thick specimens are required to meet size standards
when plane strain assumptions are used. According to
the ASTM standard for plain strain fracture tough-
ness testing (Brown and Strawley, 1966),
B> 2S(K,/a,,)’ where B is the specimen thickness, K,
is the Mode I fracture toughness and gYais the yield
stress for the material. Thickness calculated from this
formula can vary greatly depending on the values of
K, and gys chosen (Moyle and Gavens, 1986). The
standard ASTM specimen has thickness. 8, equal to
width, W, divided by two. K, is a thickness-dependent
property, which reaches a constant minimum value
when plane strain conditions are approached.
Ten fresh bovine tibias were obtained within 24 h
after slaughter to study the effect of thickness on
(W
Fig. 2. Illuminated human bone compact tension specimen. The vertical lines on the specimen are the
distance markers The crack is shown fully grown.

311
 Fracture toughness 313

fracture toughness. The age and sex of the bovine were controlled at a crossheaddisplacementrate averaging
not known although typically animals are slaughtered 0.26cmmin- 1 (+ 0.06cm min- ‘) during loading, un-
at about the age of 2 years.. Microscopic inspection loading and reloadingfor compliancemeasurements,
indicated that all tibiae were predominantly primary and automatically loaded at a displacementrate of
bone. A typical specimen, randomly selected, had an 0.05cmmin- ’ for critical load measurements during
osteon population density of 2.80 osteon per mm2 and crack growth. Crack growth in each specimenwas
2.67% porosity. Compact tension specimens (Fig. 1) monitored by a traveling microscopeattached to the
of 2, 3,5,7 and 9 mm thicknesses were machined from testingmachinewith illumination provided by a fiber-
the mid-diaphysis of the tibias to dimensions of optic light to ensureoptimumcrack visualization.The
17.5 x 16.8 mm (W= 14 mm) using procedures out- critical load (Fig. 3) was measuredas the applied
lined by Normen et al. (1991).Test data werecollected compact tension load corrsponding to specific in-
on six specimens for each thickness.The orientation crementsin crack extension. A manually operated
of the specimens correspondedto longitudinal crack switch was triggered as the crack passedspecified
propagation. Specimenswere taken from lateral and intervals to designatecritcial loads. Compliance(C)
mediallocationsof eachtibia. The 7 mm thick speci- calibration cruves(Fig. 4) for bovine and humanbone
mencorrespondedto the minimumstandardoutlined
in ASTM E-399(1985),which specifiesthat thickness,
B, is equal to width. W, divided by two. Each speci-
menwasinspectedunder a microscopebeforetesting
to ensure premature failure did not occur due to
machining. Specimenswere machined within 24 h
after obtaining the bone and testedwithin 36h after
machining.Specimenswerekept moistat all stagesof
the machiningprocesswith physiologicalsalineand
stored in salineat 4°C before testing.
Four tibiae from two humanmales,66 and 68 years
old, were obtained fresh through the Human Gift x r : l l
Registry at West Virginia University. The tibiae were x
:
.

machinedinto 15 compact tension specimenstaken ; g d 6 x 0:

0 x l
from the lateral, medial and posterior corticesof the 0

mid-diaphysisof the tibiae (Fig. 1) with the same .,--.L-.-.l-~L--~-~-.

8 9 10 11 12 13-14
&

overall dimensionsas bovine bone specimenspre- Crack Length, a (mm)

viously described.However, 2 and 3 mm thick speci-
menswere the largest that could be machinedfrom Fig. 3. Critical load plotted against crack length for human
the human bone due to bone curvature and cortical and bovine bone compact tension specimens tested at
bone thickness.Test data were collected on ten hu- a crosshead displacement rateof 0.05cmmin- I. Thesedata
man bonespecimens, five for eachthickness,and five show that the critical load depends on specimen thickness.
specimens wereusedfor systemcalibration. The tibiae
were kept frozen in physiological saline at -20°C
until machining. All specimenswere kept moist in
physiologicalsalineuntil testing.Approximately two
days lapsedbetweenthawing and testing. 2mm Bovine
3mm Bovine
Fracture toughness tests were conducted to 5mm Bovine
measure:(1)compliance,and (2)critical load for differ- 7mm Bovine
ent lengths of crack extension. Three specimensof
bovine boneweretestedat eachthicknessfor compli- 3mm Human 0
ance,and threespecimens weretestedfor criticai load.
Three 2 mm and three 3 mm thick humanbonespeci-
menswere tested to determinecompliance,and two
2 mm and two 3 mm thick human bone specimens
were testedto determinecritical load.
The experimental setup was similar to previous
investigationsthat usedcompact tension specimens 6 7 8 9 10 11 12 13 14
(Behiri and Banfield, 1980,1984;Norman et al., 1991, Crack Length, a (mm)
1922a;Wright and Hayes, 1977).A test specimen,
saturatedin physiologicalsaline,was placedin a fix- Fig. 4. The slope of the loading/unloading curve, compli-
ture (Fig. 2) designedaccordingto ASTM E399(1985) ance, at each increment in crack extension of human and
bovine bone compact tension specimens plotted against
for fracture toughnesstesting, and mounted in an crack length. These data show that compliance depends on
Instron (model TT-C) electromechanicaltesting ma- specimen thickness and that human bone is more compiiant
chine(Canton,MA 02021).Screwspeedwasmanually (less stiff) than bovine bone.
314 T. L. Norman et ul

werecalculatedby C=d/P, the slopeof the displace- where

ment (d) and load (P) curve (Norman et LJ/.,1991).
~11’?33-43 h , 1= %f!?.: - “i .’f!! ’
h,I =---.
(133 u1.3
METHOD OF ANALYSIS a22433 -43 %+33 --UiC>
th2 = , b,,=-m--m-.

a33 fl33
K, is calculated using experimentaldata and the
stressconcentration formula given by for plane strain, and

PCy2
Kc=----BW0.5’ (1)
where P, is the critical load, B is the thickness,W is
the specimenwidth (Fig. 1) and
Y,=29.6(+~‘5-185.5(;J1.5+655.7(;~.5
- 1017($y’s +638.9 ($‘Y
wheresubscripts1,2 and 3 correspondto the longitu-
dinal, circumferential and radial directions, respec-
In the above expression,a is the crack length defined tively. Orthotropic materialpropertiesobtainedfrom
in Fig. 1. G, is calculatedfrom the compliancemethod mechanicaltestsof humantibia wereassumed
and is given by 1978):
(2)
where C is the experimentally measuredspecimen
compliance.
Although compact tension testing correspondsto
Mode I (tension) loading, Mode II (shearing)and
Mode III (tearing)loadingcan alsobepresent.This is
why we report total K, and G, instead of values
associatedwith Mode I only (K,, and G,,). However,
Mode II and Mode III contributions are small for
Mode I configurations(Sun and Manoharan, 1989).
This assumptionwas verified for the current invest-
igation by calculating Mode II G, using the finite
elementmethod (Norman et al., 1992b).
The critical stressintensity factor and the critical
strain energy releaserate are related by well-estab-
lishedrelationshipsfor isotropic materialsin LEFM
under plane strain conditions as
wherev is Poisson’sratio and E is the elasticmodulus.
The planestrain relation agreesreasonablywell with
experimentalresultsfor bovine bone(Behiri and Bon-
field, 1980, 1984).Equation (3) can be extended to
account for material anisotropy for bone. The rela-
tionship betweenMode I K, and G, for anisotropic
materials is written as (Liebowitz, 1968;Sih er a[.,
1965)
for plane stress.In the above expressions,
all = l/E,, a13= -v13/EI = -v3JE3,
a22 = W2, a23 = - ~23/E2 = - vA2iE3,
as3 = l/E,, a12= -vlz,!E1 = -vzlJEz,
a36 = 0 (due to elasticsymmetry), ue6= l/G, 2,
(Knets,
E, = 18.4 GPa, v,,=O.32.
E2=8.51 GPa, v23=0.62.
E3=6.91 GPa, vr2=o.31,
G12=4.91 GPa.
K, and G, are calculatedfrom experimentaldata
usingequation(1)and (2),respectively,for bovine and
humanbone.Human boneK, is alsocalculatedfrom
G, usingthe isotropic [equation (3)] and anisotropic
[equation (4)] relations.
RESULTS
Average K, for longitudinal crack propagation in
bovine bone varied from 4.68 to 6.73MN m- ‘I2 over
the 2-9 mm thickness range tested (Fig. 5) and
equaled4.76MN m- 312(S.D.= + 1.09MN mM3j2)for
the 7 mm thick specimens,the recommendedASTM
sized specimen.Average G, for longitudinal crack
propagation varied lrom 240 to 988Nm-’ over the
samethicknessrange(Fig. 6) and equaled596N m ’
(S.D.= + 134N m-‘) for the 7 mm thick specimens.
The standarddeviations are basedon a least-squares
fit of the data.
AverageK, for 2 and 3 mm thick humanbone speci-
mensequaled4.32MN m 312(S.D.= k 0.58MN m I 1
and 4.05MNmm312 (SD.= f0.76 MNmm312), re-
spectively(Table 1).G, for 2 mm thick specimenswas
equal to 827Nm-’ (S.D.= +I58 Nm-‘) and was
equal to 595Nm-’ (S.D.= $-267Nm-‘) for 3 mm
thick specimens.
A comparisonof relative strengthand toughnessof
human and bovine bone showedthat relative to its
 Fracture toughness

1500 / / I
t

1 / , I , / I ()LL-L-L--L
* k 2 4 6 8 10 0 2 4 6 8 10
Thickness, B (mm) Thickness, B (mm)

Fig. 5. The effect of thickness on the average K, for bovine Fig. 6. The effect of thickness on the average G, for bovine
bone compact tension specimens. The data indicate that K, bone compact tension specimens. The data indicate that G,
exhibits a plane stress stage (Bi3 mm), transition stage exhibits plane stress and transition stages but does not
(3 <B < 6 mm) and plane strain stage (B > 6) commonly ob- appear to reach the plane strain stage, where G, reaches
served in metal and reaches a constant value at about a constant value. The data also show the effect of crack
6-7 mm. The data also show the effect of crack length on length on fracture toughness measurements: increasing crack
fracture toughness measurements: increasing crack propaga- propagation length increases G,.
tion length increases K,.

Table 1. Critical stress intensity factor (K,) and critical strain energy release rate (G,) for
2 and 3 mm thick human bone compact tension specimens

2mm 3mm
Crack length, a
(mm) G,*(N m-t) K,P(MN m-312) G,*(N m-l) K, t(MN rnm3j2)

I 648 3.13 404 3.30

8 884 4.34 480 4.04
9 949 4.88 900 4.82
10 758 4.08 957 5.63
11 514 5.60 820 5.60
12 491 5.12 295 3.95
13 326 4.99 200 4.03
Average 661 4.68 519 4.48
(kS.D.) (220) (0.65) (308) (0.89)
(0.5 <a/W< 1.0)
Average 827 4.32 595 4.05
( + S.D.) (158) (0.58) (267) (0.76)
(OS<a/W<0.7)

*G, calculated using the compliance method [equation (2)].

tK, calculated using the stress concentration formula [equation (l)].

Table 2. Relative fracture toughness calculated of bovine and human bone. The relative
fracture toughness is approximately the same as the relative strength of bovine bone to
human bone (- 1.5) within the variability of the data. Standard deviations are indicated in
parentheses

Relative toughness
Bovine Human (bovine/human)

Thickness G K G
(mm) (Nm-t) (MNmm3/‘) (Nm-‘) (MN?e3i2) G, ratio K, ratio

2 896 6.29 827 4.32 1.08 1.46

(131) (1.40) (158) (0.58)
3 981 6.13 595 4.05 I .66 1.66
(14‘Y (1.59) (267) (0.76)
316 T. L. Norman et al.

strength, the toughnesses of human and bovine bone

are roughly similar. but the data were not sufficiently
definitive to answer the question of which is tougher.
Relative toughness of 2 mm thick bovine and human
bone specimens equaled 1.08 and 1.46 for G, and K,,
respectively (Table 2). The same ratios for 3 mm thick
specimens equaled 1.66 for G, and K,. These ratios
indicate that the absolute toughness of bovine bone is
greater than human bone for both specimen thick-
nesses. When the ratios are normalized by the ratio of
longitudinal strength of bovine bone to the longitudi-
nal strength of human bone which is approximately
1.5 (Cowan, 1989), the data indicate that relative to
its strength, the toughnesses of human and bovine
bone are roughly similar. The 2 mm thick specimens
yielded ratios of 0.72 for G, and 0.97 for K,, indicating
that relative to its strength, the toughness of human
bone is greater than bovine bone. For the 3 mm thick
specimens, the same ratio was 1.11, indicating that
relative to its strength, bovine bone is tougher than
human bone, Thus, although the toughnesses of hu-
man and bovine bone are roughly similar the data are
not sufficiently definitive to answer the question of
which is tougher.
K, and G, varied with crack length, i.e. as a/W
decreased so did K, and G, (Figs 5 and 6). Our tests
were conducted over a crack range of 0.5 <a/ W-c 1.0,
while previous studies (Table 3) were conducted for
a/W<0.7. To minimize differences with previous in-
vestigations and to comply more closely with ASTM
recommendations, we report our results for
OS<a/W<0.7.
K, and G, also varied with specimen thickness. K,
reached a limiting value at approximately B=7 mm
(Fig. 5) using bovine bone specimens, which was the
minimum recommended ASTM geometric criterion
for W= 14 mm used in this study. K, exhibited the
three stages commonly observed for metals (Ewalds
and Wanhill, 1986; Krafft et al., 1961): (1) plane strain
for thicker specimens where K, becomes constant
(2) a transition stage where K, increases with decreas-
ing thickness and (3) a plane stress stage where K,
decreases rapidly with decreasing thickness. By con-
trast, average G, (Fig. 6) exhibited stages 2 and 3, but
the curve did not reach a limiting value (stage 1) even
at B=7 mm.
Human bone K, and G, were significantly reduced
after being adjusted to ASTM standard thickness
using thickness ratios determined from bovine bone.
Using O.S<a/W<0.7 averages, the ratio of K, for
7 mm bovine specimens to K, for 3 mm bovine speci-
mens was equal to 0.71 (Fig. 5). The ratio at 7 to 3 mm
for G, was 0.60. The ratios for 2 mm thick specimens
were equal to 0.76 and 0.66 for K, and G,, respective-
ly. After applying these ratios, human bone K, was
reduced to 3.28 and 2.88 MN rn-‘!’ for 2 and 3 mm
specimens, respectively (Table 4). G, was reduced to
546 Nm-’ for 2 mm specimens and to 357 Nm-r for
3 mm specimens. Although human bone and bovine
bone are different, the effect of thickness on human
 Fracture toughness 317

Table 4. Fracture toughness of human bone compact ten- plane stressand transition stagesbut did not appear
sion specimens adjusted for material assumptions and thick- to reach the plane strain stage.K, measuredfor bo-
ness. The effect of material property assumptions and thick- vine boneexhibited planestress,transition and plane
ness are significant
strain stagescommonly observed for metals. It is
2mm 3mm noteworthy that plane stressconditions begin at ap-
proximately 2-3 mm for bovine bone and approxim-
Gc Kc Gc Kc ately 2 mm for metal.It is reasonableto assumethat
Experimental (Table 2) 827 4.32 595 4.05 for thicknessesavailable for human bone (x3 mm),
(adjusted for thickness) (546) (3.28) (357) (2.88) the bone is in the transition phasebetween plane
Calculated from stressand planestrain.Thus,the fracture toughnessof
equation (4) - 3.31 - 2.81 humanbone requiresmodification if resultsare to be
(adjusted for thickness) (2.52) G-3 comparedto other materialsthat meetASTM thick-
nessrequirements.
Resultsfrom this experiment provided an indica-
tion of what the adjustmentsmight be for bovine
bone.Although this adjustmentis important for com-
paring valuesof K, and G, for humanbone to other
- Fit of Experimental Data (k0.85) materials, it is not necessarywhen comparing the
---Q-- Isotropic, Plane Strain Relation
- * - Orthotropic. Plane Strain Relation
sametype of material, specimenconfiguration and
loading to assess bonefracture resistance.We recom-
mend maintaining the ASTM recommendedthick-
nessfor bovine bone. For human bone, it is not
possibleto meetthis criterion. We suggestthat 2 or
3 mm specimens with W= 14mm be usedfor human
bone.This is about the largestthat can be machined
0 200 400 600 600 1000 1200 and would makecomparisonswith the current study
G (N/m) possible.Similar geometry, crack growth range and
loading rate mustbeusedto makea valid comparison
Fig. 7. K: plotted against experimentally measured G,. Kf is amongspecimens.
calculated from experimental data using the stress concen-
tration formula and from G, using the isotropic and anisot- We also found that crack length, or a/W, greatly
ropic plane strain relations. These data show that K, cal- affectsmeasuredK, and G,. Decreasingcrack a/W
culated from the stress concentration formula gives the high- decreasesaverage fracture toughnessand standard
est prediction of fracture toughness and that K, calculated deviations. ASTM specifiesthat a/W should be be-
from the anisotropic relation [equation (4)] assuming tween0.45 and 0.55 when conductingfracture initia-
orthotropic material properties provides the lowest estimate
of G,. tion experiments.Bonfield et al., (1978, 1980) have
demonstratednear constant valuesof K, within the
range alWc0.7. Previous investigations (Table 3)
only consideredalWc0.7. In the current study, we
bone fracture toughness can be estimated using bo- measuredfracture toughnessover a crack range of
vine bone data assumingthe thicknesseffect is inde- 0.5<a/W < 1.O,which is larger than the rangeusedby
pendent of microstructure. previous investigatorsand recommendedby ASTM.
The stressconcentration formula predicts higher Therefore,we decidedto report K, and G, over the
valuesof Kz than either isotropic or anisotropicrela- entire test range (Table I), but only considered
tions (Fig. 7). The anisotropic relation predicts the 0.5<a/W-co.7 in analysis, comparisonsand dis-
lowest K, of the three methodsinvestigated.K, was cussion.We recommendthat alWc0.7 be usedin
calculated using the latter method for 2 and 3 mm future studies.
humanbone specimens(Table 4). K, and G, are derived from theory based on
LEFM. The critical stressintensityfactor is a measure
of crack tip stressesand, therefore,characterizeslocal
(near tip) behavior. Severalinvestigators(Behiri and
This study investigatedlongitudinal fracture tough- Bonfield, 1980,1984,1989;Bonfield et al., 1978)have
nessof human Haversianbone to assess the effect of concludedthat K, is an adequatemeasureof fracture
human bone architecture on fracture resistance.We toughnessfor bovine bone but its useas a fracture
alsoexplored plane strain assumptions.The effect of parameter for human bone is lesswell defined. In
thicknesscould not be examinedusing human bone a compositematerial like Haversianbone,the plastic
specimens;therefore, tests using bovine bone were zone aheadof the crack tip may bealteredand conse-
used.Bovine bonedoesnot have the samemicrostruc- quently may be quite different from that predictedby
tural featuresas humanbone, but it doesprovide us stressconcentration formulas used to calculate K,.
with an estimateof the effect of thicknesson human This raisesconcern over the suitability of using
bonefracture toughness.G, of bovine boneexhibited a parameter basedon LEFM to calculate fracture
318 T. L. Norman et al.
toughness of human bone. LEFM has been applied to
reinforced polymers in several investigations (Guess
and Hoover, 1973; Hamilton and Berg, 1973; Owen
and Bishop, 1973; Wu, 1967, 1968) and shown to be
an adequate criterion for the onset of fracture.
It is clear, however, that K, does not describe the
same stress field singularity that is determined for an
isotropic elastic solid (Tirosh, 1973). Formulas used to
calculate K, from LEFM theory lead to values of K,
that probably cannot be compared to other loading
conditions or specimen geometries (Grootenboer and
Weersink, 1982). K, calculated from LEFM is an
adequate relative measure of fracture toughness as
long as the same type of bone is compared. e.g..
Haversian bone, where the microstructure among
specimens does not vary greatly. In this study, an
alternative methodology was proposed, in which K,
was calculated from G,, which had been experi-
mentally measured by the compliance method, using
the anisotropic relation [equation (4)]. Adoption of
LEFM to determine G, for anisotropic materials is
permissible (Grootenboer and Weersink, 1982; Guess
and Hoover, 1973; Sun and Manoharan, 1989; Whit-
ney et al., 1982). The strain energy release rate quan-
tifies net change in potential energy that accompanies
an increment of crack extension. It measures global
behavior, and is not influenced directly by local
microstructural features found in Haversian bone.
Beacause of this, G, should provide a good estimate of
fracture toughness for human and bovine bone.
Although G, has been shown to be a valid measure
of fracture toughness for anisotropic materials, G,
calculated for bovine bone compact tension speci-
mens in this study failed to achieve a constant value
(plane strain) even though K, did. This suggests that
assumptions of LEFM differ for cortical bone. How-
ever, in a previous study, Behiri and Bonfield (1984)
showed that K, and G, were linearly related for bo-
vine bone with specimen thicknesses less than 3 mm,
verifying that LEFM is valid over the specified thick-
ness range. At this time, we are unable to adequately
address this discrepancy. However, if we assume that
G, should reach a minimum value at 7 mm, then for
thicknesses greater than 7 mm, G, may be under-
estimated. At the 2 and 3 mm thicknesses used in this
study, failure to achieve plane strain most likely has
negligible effects on fracture toughness.
Experimentally measured G, was compared to
Kf calculated from experimental data using the stress
concentration formula [equation (l)] and from G,
using the isotropic [equation (3)] and anisotropic
[equation (4)] relation. These data showed that K,
calculated from the stress concentration formula gives
the highest prediction of fracture toughness, whereas
K, calculated from the anisotropic relation gives the
lowest. K, calculated from the isotropic relation pre-
dicts K, closes to experimental measurements as ex-
pected; the stress concentration formula and the iso-
tropic relation both assume isotropic material proper-
ties. The anisotropic relation introduces Young’s
modulus in the circumferential (Ez) and radial (15,)
directions that are much less than the longitudinal
modulus (E,) used in the isotropic relation. This cha-
nges the relationship between K,’ and C;<.
Our results for K, and G, for hot:ine hone are within
the range reported by others using the compact teu-
sion method (Table 3). The fracture toughnesses re-
ported in this study for 2 and 3 mm thick specimens
are slightly larger than those previously reported.
Differences among all reported values of fracture
toughness can be explained by considering the effects
of specimen thickness, size (IV/B), crack growth range
(a/IV) and displacement rate. We have tabulated these
with our summary of previous investigations in order
to point out the differences among tests. In the current
study, we have examined the effect of thickness and
crack growth range on fracture toughness and found
these effects to be significant. Displacement rate has
also been shown to significantly affect fracture tough-
ness (Behiri and Bonfield, 1978, 1980. 1984).
The only other investigation of humun hone fracture
toughness was conducted by Bontield et ul. (1985).
Their reported K, of human bone corrected for size
using bovine bone (Charalambides, 1984) ranged from
approximately 4.7 MN m 3’2 at age 25 to approxim-
ately 2.1 MN m - ‘I2 at age 90. Our values of K, for
human bone are within the range reported by Bon-
field et al. (1985). K, calculated using the stress con-
centration formula [equation (l)] for 2 mm specimens
ranged from 4.32 to 2.52 MN m -3!2, depending on the
method of calculation and whether an adjustment was
made for thickness. This range for 3 mm thick speci-
mens was 4.05-2.00 MN m - 3’2.
It has been proposed that although human bone is
mechanically weaker in bending and tension than
bone of most other animals, its osteonal structure has
adapted to prevent crack growth, thereby extending
bone’s fatigue life (Carter and Hayes, 1976; Carter
et al., 1976; Martin and Burr. 1989). Increased fiber
(osteon) discontinuity increases toughness in three-
point bending (Alto and Pope, 1979) suggesting that
a history of active Haversian remodeling may not
reduce bone toughness although remodeling reduces
bone strength. The relatively low material stiffness of
human bone may enhance its fatigue life further (Fine
and Ritchie, 1978). Materials with low shear strength
often have good fatigue and impact properties be-
cause greater displacement is possible along shear
planes before failure. Weakness of the osteonal inter-
face allows it to divert or stop cracks by blunting the
leading edge of the crack and trapping it within the
lamellar structure. Crack deflection serves the dual
purposes of energy absorption and diversion of the
crack from the plane of maximum stress, features that
increases fracture toughness. The value of osteons was
shown by Moyle and Bowden (1984). who found that
the work of fracture was positively although weakly,
correlated (r’=0.31) to the fractional area occupied
by osteons. Osteons only provide great toughness for
a given strength (Cooke et al., 1973).
 Fracture toughness 319

Results from this study show that the absolute Charalambides, B. (1984) M.Phil. Thesis, University of Lon-
values of K, and G, for human bone are lower than don.
Cooke, F. W., Zeidman, H. and Scheifele, S. J. (1973) The
the absolute values of K, and G, for bovine bone.
fracture mechanics of bone-another look at composite
Bovine tibia1 plexiform bone is about 50% stronger modeling. J. biomed. Mater. Res. Symp. 4, 383-399:
than human Haversian bone (Cowan, 1989) in the Corondan. G. and Haworth. W. L. (1986) A fractoaranhic
longitudinal direction and should perhaps be ex- study or human long bone. J. Biomecharzics 19, 207-!?18.
pected to show greater absolute toughness. The frac- Cowan, S. C. (1989) Bone Mechanics, pp. 111-112. CRC
Press, Boca Raton, FL.
ture toughness of bovine bone relative to the fracture Currey, J. D. (1962) Stress concentrations in bone. Quart. J.
toughness of human bone ranged from 1.08 to 1.66. Microscop. Sci. 103, 11 l-l 33.
The longitudinal strength of bovine bone relative to Ewalds, H. L. and Wanhill, R. J. H. (1986) Fracture Mechan-
the longitudinal strength of human bone is approxim- ics. Edward Arnold and Delftse Uitgevers Maatschappij
71.
ately 1.5. Therefore, even though human bone is
Fine, M. E. and Ritchie, R. 0. (1978) Fatigue-Crack Initia-
siginificantly weaker than bovine bone, relative to its tion and Near Threshold Crack Growth. Fatigue and
strength, the toughnesses of human and bovine bone Microstructure. American Society for Metals, Metals
are roughly similar. but the data were not sufficiently Park, OH.
definitive to answer the question of which is tougher. Grootenboer, H. J. and Weersink, A. F. J. (1982) A composite
model of cortical bone for the prediction of crack propaga-
tion. In Biomechanics: Principles and Applications (Edited
AclinowledSement---The authors are grateful to Dr Stanley by Huiskes, R., VanCampen, D. and DeWijn, J.), pp.
Wearden for statistical support, Vince Kish for laboratory 253-260. Martinus Nijhoff, Dordrecht.
assistance and Jeanne Grimm for her help in manuscript Guess, T. R. and Hoover, W. R. (1973) Fracture toughness of
preparation. This work was supported by the National carbonxarbon composite. J. Composite Mater. 7, 2-20.
Science Foundation under grant MSS-9009738. Hamilton, R. G. and Berg, C. A. (1973) Fracture mechanics
of fiberglass laminates. Fibre Sci. Technol. 6. 55-78.
Kelly, A.-and Davies. G. J. (1965) The principles of the
fiber-reinforcement of metals. Metall. Rev. 10, l-77.
Knets, I. V. (1978) Mechanics of biological tissue. A review.
REFERENCES Polymer Mechanics (Translation of Mekhanika Polimerov),
13,434~440.
Alto, A. and Pope, M. H. (1979) On the fracture toughness of Krafft, J. M., Sullivan, A. M. and Boyle, R. W. (1961) Proc.
equine metacarpi. f. Biomechanics 12, 415-421. Symp. Crack Propagtion, Cranfeld, p. 8.
Ascenzi, A. A. and Bonucci, E. (1967) The tensile properties Liebowitz, H. (1968) Fracture. Academic Press, New York.
of single osteons. Anat. Rec. 158, 375-386. Margel-Robertson, D. R. (1973) Studies of fracture in bone.
ASTM (1985) Standard test method for plane strain fracture PbD. thesis, Standard University.
toughness testing of metallic materials. ASTM Annual Martin. R. B. and Burr. D. B. (1989) Structure. Function and
Book of Standards Section 2, Designation E 399-83, pp. Adaptation qf Compact Bone. Raien Press, ‘New York.
898-933. Martin, R. B. and Ishida, J. (1989) The relative effects of
Beardmore, P., Harwood, J. J. Kinsman. K. R. and Rober- collagen fiber orientation, porosity, density and mineraliz-
tson, R. E. (1980) Fiber-reinforced composites: engineering ation on bone strength. J. Biomechanics 22, 419-426.
structural materials. Science 208, 833-840. Moyle, D. D. and Bowden, R. W. (1984) Fracture of human
Behiri, J. C. and Bonfield, W. (1980) Crack velocity depend- femoral bone. J. Biomechanics 17, 203-213.
ence of longitudinal fracture in bone. J. Mater. Sci. 15, Moyle, D. D. and Gavens, A. J. (1986) Fracture properties of
1841-1849. bovine tibia1 bone. J. Biomechanics 17, 203-213.
Behiri, J. C. and Bonfield, W. (1984) Fracture mechanics of Norman, T. L., Vashishth, D. and Burr, D. B. (1991) Mode
bone--the effects of density. specimen thickness and crack I fracture toughness of human bone. Adv. Bioengng 20,
velocity on longitudinal fracture. J. Biomechanics 17, 361-364.
25-34. Norman, T. L., Vashishth. D. and Burr, D. B. (1992a) Effect
Behiri. J. C. and Bonfield, W. (1989) Orientation dependence of groove on bone fracture toughness. J. Biomechanics 25,
of the fracture mechanics of bone. J. Biomechanics 22, 1489-1492.
863-872. Norman, T. L., Vashishth, D. and Burr, D. B. (1992b) 2D
Bonfield, W.. Behiri. J. C. and Charalambides: B. (1985) finite element analysis of Mode I fracture in human bone.
Orientation and age-related dependence of fracture tough- Adv. Bioengng 22, 51-54.
ness of cortical bone. In Biomechanics: Current Interdisci- Owen, M. I. and Bishop, P. T. (1973) Critical stress intensity
plinary Research (Edited by Perren, S. M. and Schnsider, factors applied to glass reinforced polyester resin. J. Com-
E.), pp. 185-189. Martinus NijholT, Dordrecht. posite Mater. 7, 146-159.
Bonfield. W.. Grynpas, M. D. and Young, R. J. (1978) Crack Piekarski, K. (1970) Fracture of bone. J. Appl. Phys. 41,
velocity and the fracture of bone. 1. Biomechanics 1, 215-223.
473-479. Pope, M. H. and Murphy, M. C. (1974) Fracture energy of
Brown, W. and Strawley, J. (1966) Plane strain crack testing bone in a shear mode. Biol. Engng 12, 763-767.
of high strength metallic material. In ASTM: 410. Sih, G. C., Paris, P. C. and Irwin, G. R. (1965) On cracks in
Burr. D. B., Schaffler, M. B. and Frederickson, R. G. (1988) rectilinearly anisotropic bodies. Inc. J. Fract. Mech. 1,
Composition of the cement line and its possible mechan- 189-203.
ical role as a local interface in human compact bone. Simkin, A. and Robin, G. (1974) Fracture formation in differ-
J. Biomechanics 21, 939-945. ing collagen fiber pattern of compact bone. J. Bio-
Carter, D. R. and Hayes, W. C. (1976) Fatigue life of compact mechanics 7, 183-188.
bone I-effect of stress amplitude, temperature and den- Sun, C. T. and Manoharan, M. G. (1989) Growth of de-
sity, J. Biomechanics 9, 27-34. lamination cracks due to bending in (90,/0,/90,) lami-
Carter, D. R., Hayes, W. C. and Schurman, D. J. (1976) nates. Composites Sci. Technol. 34, 365-377.
Fatigue life of compact bone II--effect of microstructure Swanson Analysis System, Inc. (1981) Fracture Mechanics.
and density. J. Biomechanics 8, 21 l-218. pp. 3.1 I-3.15. Technical Publication Rev. 4.4 Tutorial.
320 T. L. Norman et ul

Tirosh, J. (1973) The effect of plasticity and crack blunting on parameters for compact boneeffects of density and spec-
the stress distribution in orthotropic composite materials, imen thickness. J. Biomechanics 10, 4199430.
J. appl. Mech. 40, 785-790. Wu, E. M. (1967) Application of fracture mechanics to An-
Whitney, J. M., Browning, C. E. and Hoogsteden, W. (1982) isotropic Plates. J. Appl. Mrch. 34. 967 974.
A double cantilever beam test for characterizing Mode Wu, E. M. (1968) Fracture mechanics of anisotropic plates.
I delamination of composite materials. J. Reinforced Plas- Composite Materiuls Workshop. Technomic Publishing
tics Composites 2, 297-313. co.
Wright, T. M. and Hayes, W. C. (1977) Fracture mechanics