Sie sind auf Seite 1von 10

Bone Vol. 22, No.

1
January 1998:57 66

Changes in the Stiffness, Strength, and Toughness of Human


Cortical Bone With Age
P. ZIOUPOS and J. D. CURREY
Department of Biology, University of York, York, UK

Data on the salient mechanical properties of aging bone


material (stiffness, strength) are scanty and in particular little is
known about toughness.19 The concepts of stiffness (Youngs
modulus) and strength (load at failure per unit cross-sectional
area) are intuitively understandable. However, toughness is
less obvious and it refers to the amount of energy required to
fracture a specimen. This energy will be absorbed in three
principal ways. Some energy is absorbed prior to the generation
of a major crack in the form of diffuse damage. Some is required
to start the final fracture crack, the rest is required to drive the
crack through in order to break the material in two (or more)
pieces. The more the overall amount of energy consumed, the
tougher the material.
Toughness is of primary importance, especially in relation to
fractures, and it has been shown that the fatigue strength of older
bone34 and two different toughness measurements (work of fracture
Wf and energy absorbed in impact) reduce considerably with age.8 It
has also been shown that age (obviously through a combination of
concomitant effects) explains the trend for both work of fracture and
impact well.8 An intriguing aspect of this work was that Wf, which
measures the localized amount of energy to create a unit of new
fracture surface and is measured at low strain rates, correlated well
with the overall amount of energy consumed in impact tests at high
strain rates. In tests where fracture and microfractures (denoted
damage) are nonlocalized (as in these impact tests and any other
test in un-notched bone31) toughness, as explained earlier, depends
crucially on the amount of prefailure damage that the material can
sustain before a fatal macrocrack develops. The greater the prefracture damage, the greater the postyield deformation and the tougher
the material.32
It is possible, therefore, that aging bone is less tough because it
is capable of a lesser degree of postyield (or prefailure) damage.7
However, the overall externally perceived toughness usually has a
number of other additional causes. For instance, in tougher bones:
(i) damage accumulates at a very moderate pace prior to the
generation of the macrocrack32,33; (ii) once the crack is started it
experiences an increasing resistance to fracture in terms of its stress
intensity factor KC28 or the energy associated with its growth,22 (iii)
the energy to create a unit of fractured area increases disproportionately with the strain rate,9 something that is of particular importance
for cracks that show a tendency to accelerate.
All this suggests that we need to understand better the
interrelationship of strength and toughness. Processes that increase the stiffness and strength of a material, such as hardening
mechanisms, may either increase or decrease its toughness.
Metallic materials have an inverse relationship between strength
and toughness, because mechanisms that inhibit the motion of
dislocations also reduce the materials ability to relax crack tip
stresses. However, many modern composite materials have a

Aging adversely affects the elastic and ultimate properties of


human cortical bone as seen in uniaxial tests in quasi static
loading, high strain rate impact or fatigue. Little is known about
the full effects of aging on toughness and its relationship with
strength. In the present article the elastic modulus (E), strength
(sf), fracture toughness (KC and J-integral), and work of fracture (Wf) were determined in specimens of male human femoral
bone aged between 3592 years. In this way we investigated
whether fracture of bone in three situations, allowing various
amounts of damage prior to fracture, can provide a better
insight into the fracture process and also the relative importance of these experimental methods for assessing the soundness
of bone material. We found a steady and significant decrease
with age for all these mechanical measures. E fell by 2.3%, from
its value of 15.2 GPa at 35 years of age, per decade of later life;
sf fell similarly from 170 MPa by 3.7%; KC from 6.4 MPa m1/2
by 4.1%; J-integral from 1.2 kJ m22 by 3%, and the Wf from
3.4 kJ m22 by 8.7%. In aging bone there was a deterioration in
the elastic properties of the material. This reduced the (elastically calculated) critical stress intensity level (KC) required to
initiate a macrocrack, or the nonlinear energy associated with
the onset of fracture (J). The macrocrack was preceded by less
damage, and once created needed less energy to drive through
the tissue (Wf). (Bone 22:57 66; 1998) 1998 by Elsevier
Science Inc. All rights reserved.
Key Words: Cortical bone; Aging; Stiffness; Strength; Damage;
Toughness.
Introduction
Humans experience increased incidence of fractures with age.18
This increase is caused either by extraosseous factors such as
reduced proprioceptive efficiency, impaired reflexes, reduced cushioning by fat, by changes in the shape and size of the bones
(diameter and thickness of the cortex), and by a deterioration of the
condition of the bone material per se.12 This last factor, bone
quality,23 has not been examined thoroughly and many believe that
it is a variable that will produce, in combination with bone mass,
a much better prediction of fracture probability than bone mass
alone.21 Although most fractures in elderly people include trabecular bone, information about the quality of aging bone material will
necessarily come primarily from tests on cortical bone.
Address for correspondence and reprints: Dr. Peter Zioupos, Department
of Materials and Medical Sciences, Cranfield University, Shrivenham,
SN6 8LA, UK. E-mail: zioupos@rmcs.cranfield.ac.uk
1998 by Elsevier Science Inc.
All rights reserved.

57

8756-3282/98/$19.00
PII S8756-3282(97)00228-7

58

P. Zioupos and J. D. Currey


Material properties of aging human bone

Bone Vol. 22, No. 1


January 1998:57 66

Table 1. Information regarding the male donors (age, height, weight, cause of death) and their femoral bones

Age (yr)
35
42
46
48
50
52
56
60
70
92

Cause of death

Height
(m)

Weight
(kg)

Anatomical
height

Outer
diameter
(mm)

Inner
diameter
(mm)

Cortical
thickness
(mm)

Drowning
Road traffic accident
Coronary
Cerebral hemorrhage
Acute pericarditis
Coronary
Coronary
Coronary
Respiratory failure
Toxemia

1.68
1.67
1.62
1.60
1.78
1.71
1.64
1.64
1.66
1.64

63.5
65.5
69.5
57.0
86.0
76.5
78.0
79.5
43.3
44.0

md
m
pm
pm
md
m
md
md
m
md

28.8
29.9
26.8
32.0
30.5
27.8
29.4
27.8
27.1
30.3

14.9
14.9
14.8
16.7
16.7
14.7
12.7
12.5
15.5
17.2

6.9
7.0
5.9
7.6
6.8
6.5
8.3
7.6
5.8
6.5

Anatomical height: m 5 medial, pm and md 5 cut towards the proximal or distal end, respectively. The inner and outer diameters of the cylinders and
the maximum apparent cortical thickness were measured along the anterio-posterior and medio-lateral planes and then averaged.

positive relationship between strength and toughness; the tougher


materials are also stronger. Such materials benefit by having a
heterogeneous structure and an ability to accept widespread
microstructural damage without breaking.11
Here we examine changes in five mechanical properties: the
elastic modulus (E), strength (sf), fracture toughness (KC and
J-integral), and work of fracture (Wf) of human femoral bone
aged between 3592 years. Each of the mechanical tests for
strength, fracture toughness, and work of fracture allows different amounts of damage prior to fracture, and this fact may
provide a better insight into the fracture process and also the
relative importance of these experimental methods for assessing
the soundness of aging bone material.
Materials and Methods
Specimens were prepared from 5 6-cm-long mid-diaphyseal
cylindrical cuts from femurs obtained during routine autopsies.
All were from men who had died from causes unrelated to the
condition of the bones (Table 1) and had not been hospitalized.
The bones were transported in plastic bags packed in ice and
washed thoroughly with running water. In between tests and
prior to the preparation the bones were kept at 220C in airtight
plastic containers in the freezer wrapped in cloth soaked in
physiological Ringers solution.
Four sectors, anterior, lateral, posterior, and medial, were cut
under continuous irrigation with an Exakt (Exakt, Otto Herrmann, Denmark) diamond edge band saw. Samples of three
kinds of geometry and size (Figure 1) were prepared from the
anterior, lateral, and posterior sectors of all ten bones. The shape
and curvature of the bones in the medial sector did not allow the
cutting of satisfactory specimens.
From the middle of each sector, plate shaped specimens approximately 50 mm long, 6 7 mm wide, and 2 mm thick were cut on a
longitudinal/radial plane for 3-point bending tests. On either side
two other specimens were prepared for fracture toughness (KCJ)
and work of fracture measurements (Wf). Wf specimens were longitudinal with a 4 3 4 mm cross section. The fracture toughness
specimens were nearly the full thickness of the cortex with the other
dimensions (width and length) prepared according to ASTM
E399-83 standards1 (width equals twice the thickness, length between supports equals eight times the thickness). Analysis of variance of thickness vs. age and sector (by ranking the thickness at a
certain age from 1 to 3) showed that the thickness of the specimens
was unrelated to both age and sector (age: F2,9 5 1.12, p 5 0.40;
sector: F2,9 5 0.24, p 5 0.79). All faces of the samples were
polished by use of progressively finer grades of carborundum paper

in a direction along the long axis of the specimens (longitudinal on


the bone) and finally polished to a mirror finish with Buehler
Micropolish (0.05 mm g alumina) powder. The final dimensions
were prepared to a 0.01 mm accuracy. The whole preparation was
in physiological Ringers solution at room temperature.
All tests were carried out in a purpose-built bath with stainless
steel supports that complied with E-399-831 and D-790-86 ASTM2
standards for testing in 3-point bending. The testing bath was
mounted in a screw-driven Instron 1122 (Instron, High Wycombe,
UK) materials testing machine and was filled with a full Ringers
strength (BDH Lab. Suppl., Poole, UK) fluorescein dye (Fluorescein sodium salt, Sigma Chem. Co., St. Louis, MO) solution to
allow visualization of microcracks.31 The capacity of the inner bath
was about 10 mL and was maintained at constant 37C through
exchange of heat with the surrounding metal jackets in which fluid
from a much larger temperature controlled reservoir was circulating.
The plate-like specimens were used to measure the Youngs
modulus in flexion E and the maximum flexural stress sf (in the
outer layers of the bone, ASTM D790-86 equation-5) prior to
failure. Because of the effects of yield, this calculation of sf
overestimates the actual stress to some extent,5 but it serves as a
useful measure that is comparable to other studies that follow

Figure 1. Specimens were cut from the anterior, lateral, and posterior
sectors of the femora. The plate-like specimen for 3-point bending (E, sf)
was always in the middle of each sector and the Wf and KCJ specimens
were cut right and left on either side of it. Fracture in all three cases
proceeded in a direction across the structural elements driven either by
the direction of bending (E, sf), the chevron notch (Wf), or the sharp
notch (KCJ).

Bone Vol. 22, No. 1


January 1998:57 66

exactly the same methodology. The specimens had a thickness of


less than 2 mm, a width-to-thickness ratio of at least 4:1, and a
ratio of span to thickness of 20:1. The ratio of span to thickness
recommended by current standards is 16:1, but recent work25 has
shown that for bone a ratio of 20:1 is needed to ensure that the
estimated modulus is within 95% of the true Youngs modulus
value for bone. The loading rate was 1 mm min21 (strain rate
'1.3 3 1024 s21) and allowance was made for the (very small)
compliance of the test rig components.
The work of fracture tests were carried out on the square cross
section specimens notched so that a triangular ligament remained 26
by using the diamond band saw. The ligament tip was facing
either radially r or circumferentially c and in either case the crack
would travel transversely across the structural elements. The
orientation (r or c) was randomized and there was no correlation
with either sector or age. Bending was applied at a cross head
speed of 0.1 mm min21. The span of supports was 40 mm for all
Wf tests. In these tests, fracture starts at the tip of the triangular
chevron ligament and advances towards the base of the triangle
in a stabilized manner assisted by the ever-increasing width of
the fracture front. The work done to complete the fracture
process is divided by twice the ligament area to yield the work
needed to create a unit of new fracture surface.
Fracture toughness was assessed using the critical stress intensity
factor (KC) and the nonelastic amount of energy per unit growth of
the crack (J-integral) at the initiation of a macrocrack. A notch was
cut in the KCJ specimens by the use of the finest Exakt band saw
blade. The notch cut across the full thickness of the specimens and
its length was approximately half the width of the specimens so as
to provide a notch length/width ratio a/W ' 0.5 according to current
standards. The length of the notch was measured by both a traveling
microscope and vernier callipers to an accuracy of 0.01 mm. The
width of the notch was roughly 100 mm and its tip was semicircular,
with a minimum radius just under 50 mm. On each specimen, two
methods were used to sharpen the notch tip and precondition the

Figure 2. Force/displacement trace from the posterior sector specimen


of the 46-year-old donor with its preconditioning cycles (pre). The
deviation from nonlinearity was in excess of 5% at the critical load point
PC. PQ was at the intersection of the previous 5% secant line and the
trace. At a height of 0.8 3 PQ, the deviation of the trace (n9) was larger
than a quarter of the deviation (n) at PQ. For the J-integral determination,
the energies (areas under the curve) were determined for a set number of
displacement values (d1-d7) for each specimen and were used to produce
the calibration curve of Fig. 3.

P. Zioupos and J. D. Currey


Material properties of aging human bone

59

specimens. First, the notch tip was sharpened by use of a razorblade


attached to an Instron fixture and the application of 1 N force per
mm of length (across the thickness of the notch tip) for 30 sec. This
method did not produce a visible impression at the notch root. Then
the specimens were cycled (approximately 15 pretest cycles). The
load limit for the preconditioning routine was chosen after a number
of preliminary tests and was about 1/5 of the critical load required
to initiate the major crack. (The critical load is also used for the
determination of KC.) This cycling load was at most 20 N. The
routine produced traces that had no toe region at the very start of
the test (near zero load). The notch width was too small to fit a clip
gauge at the mouth to monitor the opening displacement and,
therefore, we used the displacement motion of the indentor relative
to the supports instead. Supplementary use of microscopic methods
allowed us to established the nature and course of events at the notch
tip.
Microscopic observations of the specimens were made extensively during preliminary tests. Two microscopic methods were
used. Direct observation, by use of a stereoscopic microscope at low
magnification (320), of effects related to the developing microcracks was achieved by intense illumination of the notch tip which
showed the appearance of a whitening front. Also, the test can be
interrupted and the specimen examined in a laser scanning confocal
microscope (LSCM) that reveals the fluorescein-stained microcracks.31 By these means we were able to establish that the deviation
from linearity in the load/displacement curves was coincident with
the appearance of microcracking. We also found that the point
where the notch started growing was coincident with the first
maximum load point on the trace, after which the rapid decrease in
load was caused by the rapid growth of the crack. Separate observations of the route and manner of growth of the major crack
growing from the tip of the manmade notch in the fracture toughness experiments were performed by the use of a Leica-Fluovert FS
microscope (Leica GmbH, Heidelberg, Germany). These observations helped in understanding the interaction of the fracture process
with the various structural elements.
Figure 2 shows a typical (real) force/displacement trace of a

Figure 3. J-integral calibration curve. From each load/displacement


trace, the scaled energy to grow a (notch-similar) crack was examined as
function of notch length and age. The derived energy/notch length slopes
(]U/]a) are plotted here vs. the preselected displacement values to
produce the calibration curve. At the observed critical displacement
values (the one shown is for the specimen from the posterior sector of the
46-year-old donor), the J-integral was derived by dividing by the average
thickness of all 30 specimens of 3.66 mm.

60

P. Zioupos and J. D. Currey


Material properties of aging human bone

Bone Vol. 22, No. 1


January 1998:57 66

Figure 4. Plots for (E, sf, KC, Wf, J) vs. age. A, L, P indicate the anterior,
lateral, and posterior sectors, respectively. The filled symbols are the
average of the previous three values at a certain age. The solid lines are
least squares linear regressions and their 95% confidence intervals. (b)
Dotted line indicates the results compiled by Yamada30 on relatively wet
human femoral bone at room temperature. In (c), the dotted data and lines
are mixed male/female results obtained by Bonfield and Behiri3 at room
temperature on relatively wet human tibial CT specimens along the
longitudinal direction. In (e), the dotted line shows results for a mode-I
critical strain energy release rate obtained for tibia, mixed male/female, by
Norman et al.20 on CT specimens in the longitudinal direction at room
temperature.

fracture toughness test. There is an initial linear region followed


by a deviation from linearity caused by creation of microcracks
ahead of the main notch tip. The load declines as soon as the
notch starts growing (past the critical load point).
The KC values we report here come from the following
formulae:

K C 5 0.031623 3 f(a/W) P CSB 21W 23/2

(1)

f(a/W) 5 1.5(a/W) {1.99 2 (a/W)(1 2 a/W)


1/2

3 [2.15 2 3.93(a/W) 1 2.7(a/W) 2]}


3 (1 1 2a/W) 21(1 2 a/W) 23/2

(2)

Bone Vol. 22, No. 1


January 1998:57 66

P. Zioupos and J. D. Currey


Material properties of aging human bone

61

Table 2. Relationships between material properties and age ($ 35 years old) observed in the present tests; linear and loglog relationships
E (GPa)
sf (MPa)
KC (MPa m1/2)
Wf (kJ m22)
J-integral (kJ m22)
Log E
Log sf
Log KC
Log Wf
Log J-integral

516.36 (0.75)a
21.81/0.000b
5191.1 (10.72)
17.82/0.000
57.33 (0.345)
21.23/0.000
54.42 (0.526)
8.40/0.000
51.34 (0.090)
14.90/0.000
51.41 (0.094)
15.0/0.000
52.62 (0.116)
22.52/0.000
51.25 (0.087)
14.42/0.000
51.61 (0.332)
4.85/0.000
50.38 (0.144)
2.63/0.028

20.0354 (0.013)
22.70/0.027
20.625 (0.188)
23.33/0.010
20.0264 (0.006)
24.37/0.002
20.0295 (0.009)
23.21/0.012
20.0036 (0.0016)
22.31/0.049
20.147 (0.054)
22.71/0.027
20.248 (0.067)
23.69/0.006
20.280 (0.050)
25.59/0.000
20.682 (0.192)
23.55/0.008
20.187 (0.083)
22.25/0.054

age (yr)

R2 5 48%

age (yr)

R2 5 58%

age (yr)

R2 5 71%

age (yr)

R2 5 57%

age (yr)

R2 5 40%

Log age (yr)

R2 5 48%

Log age (yr)

R2 5 63%

Log age (yr)

R2 5 80%

Log age (yr)

R2 5 61%

Log age (yr)

R2 5 39%

SD in parentheses.
t and p values are given underneath each coefficient, probabilities of 0.000 indicate p , 0.001.

where B is the specimen thickness, S the span between the


supports, W the specimen width, and a the notch length, all
measured in mm. PC was the critical load for crack growth in
newtons. In the very few cases in which there was a pop in
effect (the trace shows a small sharp decrease in load prior to
ascending to the overall maximum load), the pop in load value
was taken as PC. The result in (1) is derived in MPa m1/2. The
displacement rate in our tests was 2 mm min21 and was chosen
so as to give a rate of increase of stress intensity in the range of
(0.552.75) MPa m1/2 s21.
The traces did not comply strictly with the ASTM set of
standards for measurement of plane strain fracture toughness KIC.
(i) The point where the load/displacement trace achieved its
critical load was in excess of 5% deviation from linearity and the
shape of the trace showed the presence of plastic effects
(Figure 2). A secant line with a slope 5% more than the tangent
to the initial linear region intersects the trace at a load PQ. The
deviation from linearity at PQ is n. At a height equal to 0.8 3 PQ
the deviation of the trace is n9 and that was larger than a quarter
of the deviation (n) at PQ.14 (ii) The other restriction that makes
use of an estimate of fracture toughness KQ by use of PQ, the
specimen thickness B, and the yield strength of human bone
sys10:
B $ 2.5(K Q/s ys) 2

(3)

also did not hold true. Our specimens had thicknesses in the
2.9 4.9 mm range. Formula (1) produces a minimum thickness
requirement of 7.7 8.5 mm, a restriction that can not be satisfied
given the cortical thickness in human bone (Table 1). Our
mechanical observations were indicative of significant plastic
deformations being present and these were a result of the preinitiation microcracking seen in the vicinity of the notch. The
measurements for KC fracture toughness can, therefore, only be
estimates and not KIC ASTM standard values.
The J-integral was calculated by a graphical method.10 It
estimates fracture toughness through measuring the energy
needed to propagate a crack and its critical value is at the point
where the crack first started growing. The method applies
equally well from the elastic to fully plastic behavior. Each
load/displacement curve is integrated to produce the energy
(work) input in the system up to a certain displacement value.

Each work value is scaled (up or down by using each specimens thickness) to a common value of thickness for all
specimens. For each displacement a plot is made of the scaled
work values vs. each specimens crack length (manmade
notch). The negative slope of this relationship (2]U/]a) when
normalized for the average thickness value is the value of the
J-integral function. The critical value of J is subsequently
determined for the critical displacement value at which the
(manmade-) notch started growing during each test.
Our method followed 4 steps: (i) the work under the load/
displacement curve was determined at a set of seven preselected
displacements (di, i 5 17, Figure 2) for all 30 specimens. (The
displacements were chosen so as to encompass both the linear
behavior and the region of the critical load point for all specimens.) Each of the energy values was scaled to the average
thickness value of 3.66 mm. (ii) For each displacement (i.e., d1),
the scaled energy values (U1,j, j 5 130) were plotted (7
graphs in total) vs. the initial notch length of each specimen.
Because of the separate age effect, the required slope (]U1/]a,
for i 5 1) was derived from a regression of the energy values
(U1,j) vs. both age and notch length. In this way the age effect
Table 3. Correlation coefficients between the mechanical variables
and age

N 5 10
E
sf
KC
Wf
J
N 5 30
E
sf
KC
Wf
J

Age (yr)

sf

KC

Wf

20.690a
20.763a
20.839a
20.750a
20.633a

0.842a
0.676a
0.572
0.378

0.775a
0.761a
0.218

0.709a
0.605

0.572

20.471a
20.602a
20.640a
20.573a
20.406a

0.828a
0.449a
0.458a
0.151

0.533a
0.544a
0.153

0.505a
0.554a

0.252

Data are averaged at each age (N 5 10), or individual A, L, P values


(N 5 30).
a
Significant correlations.

62

P. Zioupos and J. D. Currey


Material properties of aging human bone

Bone Vol. 22, No. 1


January 1998:57 66

Figure 5. (Figure continued on next page)

was factored out at this stage of the method. (iii) The slopes
produced (]Ui/]a, i 5 17) were plotted vs. the preselected
displacement values (di) to produce a J-calibration curve
(Figure 3). (iv) From this calibration curve the critical value
for the J-integral for each specimen is derived at the point of
critical displacement (dC in Figures 2 and 3), where the notch
started propagating.

To reiterate, the various tests have the following characteristics: E measures material stiffness; sf measures flexural strength
and overestimates it to a greater extent the more postyield
deformation the specimen shows, whereas the actual damage is
confined to a mm-wide area; KC is an expression of the intensity
of the stress field necessary to start a crack growing, and assumes
that nonlinear effects are minimal and confined to an area very

Bone Vol. 22, No. 1


January 1998:57 66

P. Zioupos and J. D. Currey


Material properties of aging human bone

63

Figure 5. Photos of LSCM in front of the crack tip: (a) brittle; (b) deflected; and (c) zigzag behavior. The notch site is noted as N; thin arrows
show the path of the major crack. Note that the crack has circumvented the large osteons (r); the diffuse interlamellar damage in some cases (i); and
the fanning out of microdamage (f) in a direction opposite to the notch (which did not, however, cause a secondary macrocrack).

close to the crack tip; J measures the energy necessary to start a


crack traveling, and makes no assumptions about the linearity or
other of events at the notch; Wf measures the work/unit area
necessary to drive a crack right through a specimen. Throughout
this paper the level of significance is taken to be p , 0.05.
Results
Effects of Age on Mechanical Variables
Figure 4a e shows the behavior of E, sf, KC, J, and Wf data as
a function of age. All measures showed a deterioration with age
between 3592 years. A number of experimental factors were
examined to analyze their possible effects on the data. These
were (i) the effect of sector on all measures; (ii) the effect of
direction of the chevron ligament on the Wf data; and (iii) the
effect of specimen thickness on the KC data. (i) Analysis of
variance of each of the mechanical measures (E, sf, KC, J, Wf) vs.
sector and age showed that sector had no significant effect on the
results (F2,9 5 0.51, 0.98, 1.21, 0.75, 1.45, respectively, p . 0.25
in all cases). The linear regression lines and their 95% confidence
intervals that are shown in Figure 4a e are, therefore, least
square fits on the average data values (solid symbols). (ii) As
described in Materials and Methods, the chevron notches in the
Wf experiments were cut at random in the radial or circumferential directions. Analysis of covariance of the two separate Wf
vs. age sets of data showed that the direction had no effect; there
was no statistical difference between the two sets either on the
slopes (F1,25 5 0.30, p . 0.50) or the intercepts (F1,26 5 1.43,
p . 0.10). (iii) The KCJ specimens were prepared with various
thicknesses in order to use as much of the cortex of the bone as
possible (restrictions were imposed by the curvature of the cortex

and the required relative dimensions of length/width/thickness,


ASTM E399-831). Regression of the KC values vs. thickness and
age showed that thickness had no effect (t 5 1.11, p . 0.30).
Alternatively, an analysis of variance of KC vs. age and the ranks
of thickness at a certain age also showed that thickness had no
effect (F2,9 5 1.52, p . 0.25).
Table 2 summarizes the observed behavior of E, sf, KC, J, and
Wf as a function of age by linear and log-log relationships, which
have been produced on the average (filled symbols, Figure 4a e) of
the A, L, P values at a certain age. The linear least square predictions
show that E falls from a value of 15.2 GPa, at 35 years old, by 0.35
GPa, or by 2.3% (of its value at 35 years of age) per decade of later
life; similar values for sf were 1706.25 MPa, ;3.7%; for KC:
6.40.264 MPa m1/2, ;4.1%; for J: 1.20.036 kJ m22, ;3%; and
for Wf: 3.40.30 kJ m22, ;8.7%.
Comparisons of the present data with previous studies, which
offered data on strength or toughness in isolation, are very
favorable. Yamada30 examined relatively wet human femoral
bone at room temperature in bending and observed the behavior
shown in Figure 4b. The regression line of his results is not
statistically different from the present 3-point bending data.
Bonfield and Behiri3 used mixed male/female tibial compact
tension computed tomography (CT) specimens (Figure 4c) with
the major crack oriented along the longitudinal direction (our
cracks travel transversely). These specimens were examined
relatively wet at room temperature. Again, the slope of their
results is not statistically different from the present data. As for
the intercepts, they obey roughly a 2:1 relationship (Figure 4c).
Bonfield and Behiri3 have established that a such relationship
exists in bone in two normal directions, and this is evident here.
In Figure 4e, the dotted line shows results for mode-I critical
strain energy release rate obtained by Norman et al.20 on CT

64

P. Zioupos and J. D. Currey


Material properties of aging human bone

Bone Vol. 22, No. 1


January 1998:57 66

Table 4. Classification of the propagation of the crack in three


major categories
Age (yr)

Anterior

Lateral

Posterior

35
42
46
48
50
52
56
60
70
92

(db-db)
(db-b)
d-zd
(b-b)
(dz-bdz)
d-dz
d-z
(bz-z)
d-d
(dz-bdz)

(db-b)
(b-z)
(bd-z)
(b-d)
(bz-z)
zd-zd
d-z
(bz-zb)
(bd-d)
(d-bd)

(b-b)
d-d a
d-zd
(d-bd)
d-d
d-d
(d-bd)
(dz-db)
d-dz
dz-dz

Brittle (b): straight across in a self similar fashion; deflected (d):


upwards or downwards at a steep angle to the initial notch; zigzag (z):
deflected off its main course to a lesser degree as it met the architectural
barriers, but remaining overall in a straight line. The first entry refers to
a view from the endosteal side of the bone and the second to the view
from the periosteal side. Parentheses are used to highlight those entries
where some brittle behavior has been observed. The bold italic characters
show the final overall behavior of the crack further away from the
starting notch. For instance, dz denotes a crack that showed initially
and locally a deflected fracture to eventually grow overall in a zigzag
fashion. aSample had all its major elements (osteons, lamellae) oriented
at a 10 angle to the notch.

shorter overall length of the crack and of the microscopic deviations.


In terms of time this is indeed a fast route to complete fracture.
However, in terms of energy absorption, it is very expensive to cut
through the various reinforcing elements of bone, which make it
stronger, stiffer, and tougher when loaded in the longitudinal direction. Deflected: the crack deviated to one side and grew at an angle
to the plane of the notch. This kind of crack, although it is longer
than a brittle one, is also macroscopically relatively flat. However,
microscopically it consists of great many small deviations up and
down weak material interfaces, which are orientated mostly longitudinally. In terms of time the fracture process was slightly longer,
but it consumed less energy than the brittle crack. Zigzag: the
crack periodically changed direction up and down at sharp angles,
visible to the naked eye, as it met various architectural barriers
(osteons, cement lines, etc.). However, at the microscale, the various
microelements were actually split cleanly along weak fronts and
interfaces. This mode increases the overall duration of the fracture
process. However, in the deflected cracks the site-specific energy for
fracture along weak fronts is low and therefore, overall, these cracks
also consumed less energy than the brittle ones. Whether the
deflected or the zigzag cracks was the more energy-consuming
depends, of course, on the relative contribution and combination of
deflection and splitting.
A summary of the fracture process as seen in the present tests
is given in Table 4.
(i)

specimens. The specimens were also tibial, mixed male/female,


in the longitudinal direction, and were examined under similar
ambient conditions to Bonfield and Behiri.3 When comparing to
the present results, the slopes and intercepts of Norman et al20
behave in a similar manner as did the KC fracture toughness
slopes and intercepts.
Table 3 lists the correlations between the values of the
mechanical measures E, sf , KC, J, Wf, and also age. The threshold value for attaining significance is R 5 0.632 for 8 of freedom
(averaged values) and R 5 0.361 for 28 of freedom (individual
A, L, P values). There were significant positive correlations
between the stiffness of the bone E, the flexural strength sf, the
elastically calculated critical stress intensity factor to initiate a
macrocrack KC, and also the nonlinear toughness measurement
used to create a new unit of fracture surface Wf. However, the
J-integral, in general, did not correlate much with the rest of the
mechanical measures. There are two possible explanations for
this last result. The J-integral may constitute an intrinsic bone
material property, barely changing with age and as such not an
age-dependent bone quality property. More probably, the fact
that the J values are produced graphically through the use of
regressions and the manipulation of other original data causes a
multiplication of experimental errors that have obscured a real
effect similar to that of the other variables. Certainly the fact that
the Wf, another nonlinear energy-related estimate of toughness,
showed good correlation indicates that the second explanation is
more likely. Overall, it seems that any of the 5 presently used
mechanical variables can be used as an indicator of aging bone
quality but, as judged by the R2 values, they cannot be used
reliably to estimate each other.
The Growth of the Transverse Macrocrack
In the KCJ fracture toughness tests, the crack grew away from the
sharp notch in ways that can be classified broadly into 3 categories.
Examples are given in Figure 5a c. Brittle: the crack moved almost
straight ahead and traveled with little deviation from the intended/
main fracture plane. In a truly brittle and isotropic material this
would have been the least energy-consuming mode because of the

In younger bones, the growth of a major crack in a brittle


manner was more frequent than in older bones. (To clarify:
what we call brittle growth of a macrocrack describes only
the geometry of its fracture path. By contrast, we have
already shown that as far as mechanical properties, toughness, and the overall fragility of the skeleton are concerned,
older bones are more brittle than younger ones.) With age,
the mode of fracture became increasingly either deflected or
zigzag. We attribute this finding to the fact that the younger
bone is a more homogenous material, more compact and
with fewer weak interfaces than the older tissue, which has
many remodeling cavities and secondary osteons.
(ii) The same qualitative result is obtained when viewing the
profile of the crack from either the endosteal or periosteal
sides. There was a majority of cases (23/30) where the more
remodeled inner parts (endosteal, Table 4) exhibited some
kind of deflected fracture when the outer parts (periosteal)
showed either zigzag or brittle behavior. This produced a
twisted fracture front that turned from the original radial/
circumferential plane towards a more circumferential/tangential one.
(iii) The cracks that were easy to characterize were mostly in the
posterior side, where they were either of a simple deflected
or zigzag mode. There again, the specimens for the 3 older
subjects showed a clear predominance of zigzag fractures
caused, presumably, by the increasingly remodeled
architecture.
Discussion
Many workers have shown that human bone reaches maturity by
the age of 30 35 years and then deteriorates.6,1517,24,29 Its
mechanical soundness is reduced and its response to mechanical
stimuli for new bone formation is impaired.27 A number of
studies have produced evidence of senile effects on mechanical
properties, but they did not report different fracture-related properties from the same bones. The present tests have some additional advantages over the previous ones: they were performed at
37C in physiological Ringers solution, the properties were
measured in a direction transverse to the major elements (previ-

Bone Vol. 22, No. 1


January 1998:57 66

ous fracture toughness tests focused almost exclusively on longitudinal cracks), the bones were screened for diseases and
deteriorative effects and, most importantly, all the mechanical
measures came from one set of bones only. Therefore, we know
now not only what the effect of age is on the mechanical
measures, but also, for instance, that a 2.3% reduction in stiffness
is accompanied by a 3.7% reduction in flexural strength, a 4.1%
reduction in fracture toughness KC, and so on. Such relationships
cannot be inferred from the tests reported in previous papers.
Ordinary strength tests may, particularly if the material is
brittle, give values that are misleading, because in materials with
a hierarchical architecture (such as bone), there is a size effect:
smaller specimens may have calculated strengths that are greater
than those of large specimens. Fracture mechanics tests, on the
other hand, attempt to measure material properties on carefully
prepared specimens by either confining damage to a small,
predictable part of the specimen (unlike the situation in real life,
where damage may occur almost randomly throughout the specimen), or by having a shape and size of a notch such that the
stress field around it is mathematically tractable. In this respect,
the present study also brings together measurements of strength
and toughness of the material. In homogeneous metallic materials, the two measures have an inverse relationship,4 whereas
fiber reinforced composites show a positive relationship.11 In the
first case, KC offers information, additional to that provided by
strength, about the conditions that lead to the production of a
macrocrack. However, KC is not useful for predictions of
strength and, in fact, events that initiate a macrocrack are irrelevant to the final fracture or coherence of the structure. In the
second case, the load that creates the first macrocrack is also the
maximum, because afterwards the material simply disintegrates
more or less easily. KC (a variable derived by elastic field theory
equations) is useful, but it is well correlated with the more easily
acquired material strength. KC can therefore be used for strength
predictions, but one should question whether it is conceptually
correct to use it.
Our data and observations of the course of a major crack
partially clarify the fracture process in human bone. All the
3-point bending tests showed that human bone has some postfailure load integrity (failure is at the point where a major
macrocrack is initiated and the material strength is defined) and
therefore cannot be characterized as a brittle material. Fracture
mechanics was originally introduced into bone studies to explain
the differences in stiffness, strength, and toughness observed in
wet and dry bone, and similar though less marked differences in
the properties of fresh and embalmed bone.19 However, in these
cases there is a quantum transformation of the condition of at
least one component of bone (collagen), which can directly affect
its role in bone toughness. Dehydration of bone causes a transition from quasibrittle (wet bone) to a brittle material (dry
bone). Embalming also has a hardening effect that affects the
strength and toughness of the material differently. By contrast,
aging constitutes a simple deteriorative, gradual process, which
only affects the quality of bone material.
All types of bone experience the generation of microcracks
(damage) prior to failure. Microcracks start at yield32 and,
depending on the size, amount, and interaction of such microcracking, bone can be tough to a greater or lesser degree by being
able to prolong its postyield deformation. It is reasonable, therefore, to expect that the decrease in toughness of human bone with
age is related to the kind and amount of microcracking (prefailure damage). However, the clear relationship between toughness
and extent of postyield deformation in normal bone does not
necessarily mean that this is the only possible explanation. The
brittleness of biological materials is a factor of many quantities.
It depends on the amount of prefailure damage,32 on the magni-

P. Zioupos and J. D. Currey


Material properties of aging human bone

65

tude of the critical conditions for initiating a macrocrack,3 the


behavior of such stress- (KC) or energy- (J, Wf) related variables
with the size of the crack,28 and also with the strain rate.9,13 It
seems that aged human bone still has the ability to dissipate
energy around dangerous cracks by microcracking, but it does so
in a less effective manner.
Combined consideration of the present stiffness, strength, and
toughness data allows us to express the overall effects of aging as
follows. With age there was a reduction in the stiffness. This
reduced the (elastically calculated) critical stress intensity level
(KC) required to initiate a macrocrack. The macrocrack was
preceded by less damage (J), and once created needed less
energy to drive through the tissue (Wf). The present findings
reinforce the idea that bone behaves more like a composite
material and less like, for instance, some metals and polymers,
where often fracture toughness and strength have an inverse
relationship. This means that at least in respect to aging and its
effects we have to employ more holistic methods (looking at it
from more angles) in order to understand the process fully. In this
way we may be able to clarify which of the two following
statements, or more likely a combination of them, is true: (i) the
increased intrinsic fragility of aging bone is caused by its being
able to absorb less postyield damage; (ii) the fragility is caused
by a reduced ability to hinder the onset of a macrocrack. There
is a very fine distinction between these two hypotheses.

Acknowledgments: This study was supported by a grant from Wellcome


Trust (No. 040350/Z/94/Z/MP/HA) to investigate age changes in the
toughness of human bone.

References
1. ASTM Standards E399-83. Standard Test Method for Plane-Strain Fracture
Toughness of Metallic Materials. Philadelphia, PA: American Society for
Testing and Materials; 1985; 03.01.
2. ASTM Standards D790-86. Standard Test Methods for Flexural Properties of
Unreinforced and Reinforced Plastics and Electrical Insulating Materials.
Philadelphia, PA: American Society for Testing and Materials; 1990; 08.01.
3. Bonfield, W. and Behiri, J. C. Fracture toughness of natural composites with
reference to cortical bone. In: Friedrich, K., Ed. Application of Fracture
Mechanics to Composite Materials. Amsterdam: Elsevier Science; 1989;
615 635.
4. Broek, D. Elementary Fracture Mechanics. Dordrecht: Martinus Nijhoff; 1986.
5. Burstein, A. H., Currey, J. D., Frankel, V. H., and Reilly, D. T. The ultimate
properties of bone tissue: the effects of yielding. J Biomech 5:34 44; 1972.
6. Burstein, A. H., Reilly, D. T., and Martens, M. Aging of bone tissue:
mechanical properties. J Bone Jt Surg 58A:82 86; 1976.
7. Courtney, A. C., Hayes, W. C., and Gibson, L. J. Age-related differences in
post-yield damage in human cortical bone. Experiment and model. J Biomech
29:14631471; 1996.
8. Currey, J. D., Brear, K., and Zioupos, P. The effects of aging and changes in
mineral content in degrading the toughness of human femora. J Biomech
29:257260; 1996.
9. Currey, J. D., Brear, K., and Zioupos, P. Strain rate dependence of work of
fracture tests on bone, 10th Conference of the European Society of Biomechanics, Aug. 28 31, Leuven, Belgium, 1996.
10. Ewalds, H. L. and Warnhill, R. J. H. Fracture Mechanics. Delft, The Netherlands: Edward Arnold; 1986; 102.
11. Harris, B., Dorey, S. E., and Cooke, R. G. Strength and toughness of fibre
composites. Comp Sci Technol 31:121141; 1988.
12. Heaney, R. P. Is there a role for bone quality in fragility fractures? Calcif Tiss
Int 53(Suppl.):S3S6; 1993.
13. Kasapi, M. A. and Gosline, J. M. Strain rate dependent mechanical properties
of the equine hoof wall. J Exp Biol 199:11331146; 1996.
14. Knott, J. F. Fundamentals of Fracture Mechanics. London: Butterworths; 1973;
140 144.
15. Lindahl, O. and Lindgren, . G. H. Cortical bone in man-II. Variation in tensile
strength with age and sex. Acta Orthop Scand 38:141147; 1967.

66

P. Zioupos and J. D. Currey


Material properties of aging human bone

16. Martin, R. B. and Burr, D. B. Structure Function and Adaptation of Compact


Bone. New York: Raven; 1989.
17. McCalden, R. W., McGeough, J. A., Barker, M. B., and Court-Brown, C. M.
Age-related changes in the tensile properties of cortical bone. J Bone Jt Surg
75A:11931205; 1993.
18. Melton, L. J. Epidemiology of fractures. In: Riggs B. L. and Melton, L. J., Eds.
Osteoporosis: Etiology, Diagnosis and Management. New York: Raven; 1988;
133154.
19. Melvin, J. W. Fracture mechanics of bone. J Biomech Eng 115:549 554; 1993.
20. Norman T. L., and Nivargikar, S. V., and Burr, D. B. Resistance to crack
growth in human cortical bone is greater in shear than in tension. J Biomech
29:10231031; 1996.
21. Ott, S. M. When bone mass fails to predict bone failure. Calcif Tiss Int
53(Suppl.):S7S13; 1993.
22. Peterlik, H., Fratzl, P., Roschger, P., and Klaushofer, K. Unpublished.
23. Sherman, S. and Hadley, E. C. Aging and bone quality: An underexplored
frontier. Calcif Tiss Int 53:S1; 1993.
24. Smith, C. B. and Smith, D. A. Relations between age, mineral density and
mechanical properties of human femoral compacta. Acta Orthop Scand 47:
496 502; 1976.
25. Spatz, C.-CH., OLeary, E. J., and Vincent, J. F. V. Youngs moduli and shear
moduli in cortical bone. Proc R Soc Lond B 263:287294; 1996.
26. Tattersall, H. G. and Tappin, G. The work of fracture and its measurement in
metals, ceramics and other materials. J Mater Sci 1:296 301; 1966.
27. Turner, C. H., Takano, Y., and Owan, I. Aging changes mechanical loading
thresholds for bone formation in rats. Bone 10:1544 1549; 1995.

Bone Vol. 22, No. 1


January 1998:57 66
28. Vashishth, D., Behiri, J. C., Tanner, K. E., and Bonfield, W. Toughening
mechanisms in cortical bone. 42nd Annual Meeting of the ORS, Feb. 19 22,
Atlanta, Georgia, 1996.
29. Wall, J. C., Chatterji, S. K., and Jeffery, J. W. Age-related changes in the
density and tensile strength of human femoral cortical bone. Calcif Tiss Int
27:105108; 1979.
30. Yamada, H. Strength of Biological Materials. In: Evans, F. G., Ed. Baltimore,
MD: William & Wilkins; 1970; 20.
31. Zioupos, P. and Currey, J. D. The extent of microcracking and the morphology
of microcracks in damaged bone. J Mater Sci 29:978 986; 1994.
32. Zioupos, P., Currey, J. D., and Sedman, A. J. An examination of the micromechanics of failure of bone and antler by acoustic emission tests and laser
scanning confocal microscopy. Med Eng Phys 16:203212; 1994.
33. Zioupos, P., Wang, X. T., and Currey, J. D. Experimental and theoretical
quantification of the development of damage in bone and antler. J Biomech
29:989 1002; 1996.
34. Zioupos, P., Wang, X. T., and Currey, J. D. The accumulation of fatigue
microdamage in human cortical bone of two different ages in-vitro. Clin
Biomech 11:365375; 1996.

Date Received: April 8, 1997


Date Revised: July 1, 1997
Date Accepted: August 19, 1997

Das könnte Ihnen auch gefallen