journal of the mechanical behavior of biomedical materials 41 (2015) 261–270

Available online at www.sciencedirect.com

www.elsevier.com/locate/jmbbm

Research paper

Strain-rate sensitivity of the lateral collateral

ligament of the knee

Timothy J. Bonnera,b, Nicolas Newellb, Angelo Karunaratneb, Andy D. Pullenc,

Andrew A. Amisd,e, Anthony M.J. Bullb, Spyros D. Masourosb,n
a
The Academic Department of Military Surgery and Trauma, The Royal Centre for Defence Medicine,
Birmingham B15 2SQ, UK
b
Department of Bioengineering, Imperial College London, London SW7 2AZ, UK
c
Department of Civil and Environmental Engineering, Imperial College London, London SW7 2AZ, UK
d
Department of Mechanical Engineering, Imperial College London, London SW7 2AZ, UK
e
Department of Musculoskeletal Surgery, Imperial College London, London W6 8RF, UK

art i cle i nfo ab st rac t

Article history: The material properties of ligaments are not well characterized at rates of deformation
Received 31 March 2014 that occur during high-speed injuries. The aim of this study was to measure the material
Received in revised form properties of lateral collateral ligament of the porcine stifle joint in a uniaxial tension
28 June 2014 model through strain rates in the range from 0.01 to 100/s. Failure strain, tensile modulus
Accepted 1 July 2014 and failure stress were calculated. Across the range of strain rates, tensile modulus
Available online 9 July 2014 increased from 288 to 905 MPa and failure stress increased from 39.9 to 77.3 MPa. The

Keywords: strain-rate sensitivity of the material properties decreased as deformation rates increased,

Ligament and reached a limit at approximately 1/s, beyond which there was no further significant

Stress change. In addition, time resolved microfocus small angle X-ray scattering was used to

Modulus measure the effective fibril modulus (stress/fibril strain) and fibril to tissue strain ratio. The

Strain rate nanoscale data suggest that the contribution of the collagen fibrils towards the observed

Injury tissue-level deformation of ligaments diminishes as the loading rate increases. These
findings help to predict the patterns of limb injuries that occur at different speeds and
improve computational models used to assess and develop mitigation technology.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction with associated treatment costs alone at an average of €1727

Human ligament injuries are common and can cause sig-
nificant morbidity and long-term disablement (Gelber et al.,
2000). The EUROCOST reference group estimate the inci-
dence of knee ligament injuries at 0.04% population/year
per injury (Polinder et al., 2005). The types of joint injuries that
occur may vary according to health, age, sex, anatomy,
mechanism of trauma and rate of loading (Ytterstad, 1996).
Prevention of significant joint injuries requires us to under-
stand why different patterns of injury may occur with

n
Corresponding author. Tel.: þ44 20 7594 2645; fax: þ44 20 7594 9817.
E-mail address: s.masouros04@imperial.ac.uk (S.D. Masouros).

http://dx.doi.org/10.1016/j.jmbbm.2014.07.004
1751-6161/& 2014 Elsevier Ltd. All rights reserved.
262 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270
different traumatic mechanisms and rates of loading. Com-
putational modelling of joint injuries is one method that may
be useful to help predict different patterns of injury and
assess mitigation technologies. However, reliable biomecha-
nical measurements of the connective tissues of joints are
required if the models are to be accurate and useful.
Ligaments are visco-elastic materials made of collagen
fibres, which change in strength and stiffness relative to their
rate of loading (Yannas and Olson, 1972; Tipton et al., 1967).
Tensile modulus and failure stress are useful measurements
to compare the material properties of different ligaments.
These material properties are important input parameters to
computational models of human joint injuries. Previous
laboratory studies have found that the tensile modulus and
failure stress of ligaments both increase as the rate of loading
increases (Danto and Woo, 1993; Yamamoto et al., 2003;
Yamamoto and Hayashi, 1998; Ng et al., 2004). However,
many of the previous studies have focused on the failure
characteristics at quasi-static loading rates or assessed only a
few different loading rates (Danto and Woo, 1993; Noyes
et al., 1974; Crowninshield and Pope, 1976; Kennedy et al.,
1976).
This strain-rate dependent material behaviour of ligament
tissue cannot be understood without considering the hier-
archical nature of the structure. Small angle X-ray diffraction
(SAXD) has been performed previously on collagenous tissues
such as tendons, bones and cartilage in an attempt to
quantify the viscoelastic properties of the tissues at a fibrillar
level and utilize them to explain their typical macroscopic
behaviour (Fratzl et al., 1997; Gupta et al., 2010; Puxkandl
et al., 2002). Fratzl et al. (1997) proposed a simple model
explaining why the ratio of fibril-to-tissue strain increases
with strain rate in the quasi-static range. They suggest that
the proteoglycan-rich matrix becomes stiffer due to an
increase of the viscous component as strain rate increases.
Unfortunately, the maximum strain rate at which they tested
was 0.001/s and so their observations are not adequate to
demonstrate or explain potential changes in strength and
modulus at strain rates experienced at injury.
Whilst work at slow rates is useful to understand beha-
viour in normal joint function and to choose replacement
grafts, the application of these results to high-speed injury
modelling may not be valid; significant error may occur if low
strain-rate material properties are applied to simulations of
traumatic injury. The limitations of previous work are likely
to be caused by the technical difficulties of measuring stress
and strain at rates that simulate high-speed injuries, such as
motor vehicle collisions or battlefield injuries due to blast
(Nagasaka et al., 2003; Ramasamy et al., 2011).
The aim of this study was to investigate the material
properties of ligaments in a uniaxial tension model at strain
rates in the range from 0.01 to 100/s. A porcine stifle joint
ligament experimental model was designed to simulate the
strain rates that may occur during a full range of different
human knee ligament loading. The hypothesis was formu-
lated that the strain-rate sensitive material properties of a
ligament would diminish as strain rate increased. Studying
ligament properties over a large order of magnitude of strain
rates also provides an insight into the different structural
explanations for their visco elasticity. Furthermore, time
resolved synchrotron small angle X-ray scattering on human
ligaments was used to investigate the deformation mechan-
isms at the nanoscale in an attempt to explain the strain-rate
dependent behaviour of ligaments.
2. Methods
2.1. Specimen preparation
Ligaments of the porcine stifle joint were selected because of
their similarity in morphology, size, structure, material prop-
erties and physiological loading to the human knee joint
(Xerogeanes et al., 1998). Sixty porcine hind limbs were
delivered to the laboratory on the day of slaughter from a
local abattoir. Excess muscle bulk was removed from the
limbs, which were then stored at  20 1C. All limbs were
utilized within one month of slaughter to minimize any
potential deterioration in their mechanical properties
(Masouros et al., 2009). All limbs were from healthy female
large-white pigs, aged between 9 and 12 months. The demo-
graphics of the pigs were controlled to limit the physiological
variation in material properties, which is known to occur
between sexes, age groups, pig breeds and in unhealthy
subjects (Germscheid et al., 2011; Noyes and Grood, 1976).
Each hind limb was thawed at room temperature on the
day of testing. The lateral collateral ligament (LCL) of the
porcine stifle joint was isolated by removing skin, muscle,
other joint ligaments and tibia, thus leaving the femur, LCL
and fibula intact. A hand saw was used to cut a 15  15  25
mm3 bone block around the femoral attachment of the LCL. A
similar bone block was created with the fibula by removing its
rounded proximal margin and dividing it transversely at
40 mm in length. A thin longitudinal round segment of
ligament was isolated along the posterior margin of each
LCL, such that each ligament's fascicles were easily aligned,
similar in length and would reliably fail in its mid-substance.
The unwanted anterior segment was removed by separating
the ligament via blunt dissection in line with the fascicles,
and divided transversely both proximally and distally when
the fascicles could no longer be easily separated; thus ensur-
ing no structural damage. This created a test specimen with a
long thin middle section of a relatively constant cross-
sectional area, with a broad anchor at either end made of
the bone blocks and fibrocartilage transition zone (Fig. 1).
Cross-sectional area was measured using a previously
validated technique for use in soft tissues (Goodship and
Birch, 2005). Each specimen was held under 1 N of tension
and the mid-substance of the ligament was cast in a quick-
setting, stiff alginate paste (Blueprints cremix, Dentsply
DeTrey, Germany). The solid alginate paste was cut perpen-
dicular to the long axis of the construct after removal of the
ligament. Digital photographs were taken of the cut sections
of alginate paste at three different sites. Each photograph was
converted into binary code, based on whether or not each
pixel contained an image of the ligament cast. The number of
pixels in each photograph was counted using a custom
computer code (MatLAB, MathWorks Inc., Natick, MA, USA).
The cross-sectional area was calculated by comparing the
 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270
Fig. 1 – Photograph of one ligament sample.
mean of the three pixel counts against a calibration photo-
graph of a shape with a known cross section.
The bone blocks were placed into cylindrical aluminium
pots and secured by alignment screws. Positioning of the
ligament sample was carefully adjusted such that it was
coaxial with the uniaxial tension test. The bone blocks were
then set in polymethyl-methacrylate (PMMA) bone cement.
Petroleum jelly and saline soaked gauze were used to
keep the ligament hydrated and particularly to protect its
263
attachments from the heat generated during cement poly-
merization. Multiple small black dots were made across both
ends of the ligament with permanent black ink (Staedtler
Ltd., UK) during set up on the tensile testing machines. The
samples were then sprayed with water to ensure they
did not dry out before testing. The length of each ligament
sample was measured once with digital callipers before
testing to provide an estimate of the deformation rates
needed to achieve the required strain rates. All tests were
performed at room temperature.
2.2. Tensile testing
Tests were carried out to achieve target strain rates of 0.01,
0.1, 1, 10, and 100/s.
2.2.1. Quasi-static tensile tests
A screw-driven electro-mechanical (5866; Instron, Canton,
MA, USA) and a servo-hydraulic (8872; Instron, Canton, MA,
USA) materials testing machines were used for tensile tests at
0.01–0.1/s and 1/s, respectively. Pre-conditioning of the liga-
ments was performed for the quasi-static tests with cyclic
loading between 1 and 10 N at 10 mm/min, and repeated five
times, then held at 0 N for 10 s (Masouros et al., 2009). The
specimens were loaded to failure at extension rates of 0.47,
4.7 and 47 mm/s to achieve strain rates in the region of 0.01,
0.1, and 1/s, respectively. Two specimens were extended at
0.39 mm/s and 0.55 mm/s because the ligaments were shorter
and longer, respectively, than the rest. Force-extension data
was measured simultaneously using the built in load cell and
extensometer at a sampling rate of 50 Hz, then recorded with
the machine's accompanying materials testing software
(Bluehills v2.11, Instron, High Wycombe, UK).
2.2.2. Impact tensile tests
Tensile tests at strain rates in the region of 10 and 100/s were
performed using an Instron Dynatup 9250-HV spring-assisted
drop-weight rig (Instron, High Wycombe, UK). A custom-
made impact tensile adaptor (ITA) was manufactured for
the experiment (Fig. 2). The upper aluminium pot was
attached by a spherical rod-end connector to a fixed steel
crossbeam. The lower pot was attached to an aluminium
rectangular fixture, free to move vertically in the axis of the
tensile test, which was rapidly accelerated by a 7.45 kg
impactor with a 50 mm diameter head for each test. The
velocity of the impactor could be altered by changing its drop
height, or with the addition of accelerator springs. A pair of
diametrically-opposed bi-axial strain gauge rosettes (FCA-6;
Techni Measure, Warwickshire, UK) were bonded to the
upper aluminium pot in a Wheatstone full-bridge configura-
tion in order to measure load. The strain gauge output was
calibrated in a series of preliminary tests against known
loads. A PXIe data acquisition system in conjunction with a
custom-written LabVIEWs software program (NI instru-
ments, Austin, TX, USA) was used to record strain gauge
output with a sampling rate of 16 kHz–2 MHz. Strain rates in
the region of 10 and 100/s were achieved with an impactor
velocity at impact of 1 m/s and 8–12 m/s, respectively. The
impactor velocity for the 100/s tests was increased from 8 to
264 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270
Fig. 2 – Diagram of the drop rig apparatus and impact tensile
adaptor used for measuring the material properties at the
highest strain rates.
12 m/s after six tests because the average strain rate was less
than the target rate.
2.3. Data analysis
Strain was measured directly from the mid-substance of each
ligament using high-speed photography (Phantom V12.1,
frame rate 27–47000 fps). Two points were selected at either
end of the ligament and were tracked across frames using
digital tracking software (Phantom Camera Control Applica-
tion 1.3, Vision Research, NJ, USA) by noting their pixel
position at every frame. For each frame, strain was calculated
as the change in pixel separation of the pair of points, divided
by the pixel separation of the two points in a reference frame.
This procedure was repeated for an additional 2 pairs of
points, and ligament strain was taken as the mean value of
the 3 pairs.
Strain rate was calculated as the gradient of the linear part
of the strain-time curve for each test. Stress was calculated as
the recorded force over the calculated original area. Stress–
strain curves were created and used to calculate the tensile
modulus, and stress and strain at failure for each sample.
Pearson's correlation coefficient was calculated to establish
the linear relationship between time and strain before failure.
Changes across the range of strain rates of strain at failure,
stress at failure and tensile modulus were examined using
one-way analysis of variance (ANOVA) with post-hoc Bonfer-
onni testing; po0.05 was taken to indicate a significant
change.
2.4. Nano-scale experiments
In an attempt to understand the strain-rate sensitivity at a
fibrillar level we conducted mechanical testing utilizing
synchrotron SAXD on human LCLs. Ethical approval had been
secured prior to testing from the local research ethics com-
mittee. We used the same experimental protocol to previous
studies (Fratzl et al., 1997; Gupta et al., 2010; Karunaratne
et al., 2012, 2013) to measure fibrillar deformation at 0.001,
0.005, 0.01, and 0.05/s. Twenty human LCL samples were
prepared (5 per strain rate) and gripped in a micro-tensile
tester (Karunaratne et al., 2012) that was mounted on a 2 axis
motorized stage beam-line I22 at Diamond Light Source, UK.
A synchrotron X-ray beam (wavelength 0.886 Å, beam cross
section 10  12 mm2) was used to measure the SAXD patterns,
which were collected by a Pilatus detector system. The
sample-to-detector distance was 1 m. For each strain rate a
SAXD pattern with 1 s exposure time was collected at every
1% applied external (grip-to-grip) strain up to failure. The
fibril strain was measured as described elsewhere (Fratzl
et al., 1997; Gupta et al., 2010; Puxkandl et al., 2002) by
tracking the change in the D-periodicity (  67 nm) of the
meridional banding patterns in the collagen fibrils arising
from the intrafibrillar tropocollagen packing (Fig. 3).
3. Results
The data-sets resulting from 41 ligament tests were available
for analysis; the remaining data-sets were not used due to
either trigger failure during testing, inadequate image cap-
ture, early sample failure around the bone block, or use in
preliminary testing. One further sample was excluded during
the analysis because the sample was a gross outlier and is
likely to have failed in an atypical way.
The time to failure of the ligaments ranged from less than
1 ms at the higher strain rates, to greater than 20 s at the
lower strain rates. The cross-sectional areas of the samples
was 3.6 (SD¼ 1.1) mm2. The relationship between strain and
time was linear for all tests (mean R2 ¼0.99 (SD¼ 0.02),
po0.001); hence constant strain rate was achieved across all
experimental setups. Fig. 4a presents typical stress–strain
curves for each strain rate. A non-linear – often called ‘toe’ –
region was pronounced at lower strain rates, requiring up to
3–4% strain before a linear gradient was observed. A ‘yield’
point could be observed with some samples at the two lowest
 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270 265

Fig. 3 – (a) Schematic view of the synchrotron experimental setup. (b) Detail of the tensile testing apparatus. (c) The ligament
sample ends are potted in acrylonitrile butadiene styrene molds using dental cement (Filtek Supreme XT, 3M ESPE, USA).
(d) By irradiating the collagen fibril with X-rays a series of Bragg reflections appears on the 2D scattering pattern due to the
staggered axial arrangement of tropocollagen molecules.

Fig. 4 – (a) Typical stress–strain curves for 5 different specimens, each being an order of magnitude apart in strain rate.
(b) Mean stress–strain curves per strain rate. The vertical error bars without end caps represent 1 SD. The horizontal and
vertical crossed error bars with the circular end-caps represent the mean71SD stress and strain at failure per strain rate.

strain rates, but complete structural failure occurred shortly
afterwards with minimal further deformation. Toe region and
‘yield’ point were not observed at the two fastest strain rates.
There was a general trend towards a negative relationship
between the failure strain and strain rate (Table 1). Fig. 4b
presents the mean curves at the five target strain rates of the
study. Failure stress was found to increase at the three
slowest strain rates, but further change was insignificant
after approximately 1/s. This was supported by ANOVA tests,
which showed that there were statistically significant
increases in tensile modulus and stress at failure between
each consecutive strain-rate group up to 0.94/s (po0.05), but
not beyond. This suggests that a strain-rate sensitivity limit
occurs at approximately 1/s.
266 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270

Table 1 – Material properties of the groups of ligament samples; mean (standard deviation).

Strain rate (/s) # of samples Tensile modulus (MPa) Failure stress (MPa) Failure strain (%)

0.01 (0) 7 288 (83.9)n 39.9 (11.0)§ 17 (3)

0.1 (0) 7 364 (86.6) 56.5 (8.2) 18 (3)
0.94 (0.11) 7 656 (82.4)n 72.8 (11.1)§ 14 (2)
10.6 (2.2) 7 763 (141.3) 75.9 (9.6) 11 (3)
129.9 (52.8) 12 906 (195.2) 77.4 (15.4) 9 (2)

There was a statistically significant difference in tensile modulus (n) and failure stress (§) as strain rate increased from 0.01 to 0.94/s (po0.01).

Fig. 5 – (a) Failure stress and (b) tensile modulus of the porcine LCL across strain rates. The solid line represents a bilinear fit
and the dashed straight line a logarithmic (with base 10) fit to the experimental data. Note that a plateau would be observed at
high strain rates when the curves are plotted with a linear horizontal axis (inset).

Tensile modulus and failure stress as a function of strain
rate are shown in Fig. 5. Tensile modulus and failure stress of
the ligament increased by an average of approximately 3- and
2-fold respectively over the strain rates tested, but the change
occurred almost entirely over the three slowest strain rates.
The relationship between tensile modulus and failure stress
with strain rate was fitted with logarithmic and bilinear
curves. The bilinear relationship was found to fit the data
with a smaller error than the logarithmic. Specifically:
Tensile modulus (in MPa, R2 ¼ 0.76) equation
( rates up to a limit of approximately 1/s, beyond which this
384_ε þ 292; ε_ o_ε0 ¼ 1:3=s
E¼
0:68ð_ε  ε_ 0 Þ þ 790; ε_ 4_ε0
Failure stress (in MPa R2 ¼ 0.68) equation.
( low strain rates (o1/s) are in broad agreement with previous
32_ε þ 44; ε_ o_ε0 ¼ 0:97=s
σf ¼
0:05ð_ε  ε_ 0 Þ þ 75; ε_ 4_ε0
Tensile modulus (in MPa, R2¼0.73) equation
E ¼ 158:55 log10 ε_ þ 584:76
Failure stress (in MPa, R2 ¼ 0.62) equation
σ f ¼ 9:97 log10 ε_ þ 63:88
Fig. 6 presents stress–tissue strain and fibril strain–tissue
strain relationships for human ligament tissue from the
SAXD experiments. Fibril strain for measured tissue strain
significantly reduced at higher strain rates compared to quasi
static strain rates. Finally, Fig. 7 shows the increase in
collagen-fibril modulus (slope of the stress–fibril-strain curve)
with strain rate.
4. Discussion
The material properties of the lateral collateral ligament of
the porcine stifle joint were found to be sensitive to strain
effect became insignificant. This is the first study to report
the strain-rate dependency of the material properties of
ligament tissue across 5 orders of magnitude. The results at
laboratory studies on the strain-rate sensitivity of animal and
human ligaments (Danto and Woo, 1993; Yamamoto et al.,
2003; Yamamoto and Hayashi, 1998; Ng et al., 2004; Noyes
et al., 1974; Crowninshield and Pope, 1976; Kennedy et al.,
1976; Woo et al., 1990; Crisco et al., 2002; Schenck et al., 1999).
For example, Woo et al. (1990) found that the tensile failure
stress of the lapine medial collateral ligament (MCL)
increased by up to 40% from 0.0001 to 2.2/s. Ng et al. (2004)
reported that the failure stress of murine flexor digitorum
superficialis tendon increased from 50 to 70 MPa from 0.0005
to 1/s, respectively. However, the strain-rate sensitivity limit
 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270 267

Fig. 6 – (a) Two different deformation mechanisms of intra- and inter-fibrillar structures proposed for low (top) and high
(bottom) strain rates. At low strain rates fibrils start from an unloaded state (i), go through the toe region (ii), and exhibit intra-
fibrillar gliding at high strains (iii). At higher strain rates intra-fibrillar sliding occurs due to the early debonding of the matrix
from the collagen fibrils. (b) Stress–tissue strain curve (filled symbols) and fibril strain–tissue strain curve (open symbols) for
low and (c) high strain rate.

at high strain rates (41/s) observed in this study has not been
reported previously.
Earlier studies have recognized that the strain-rate sensi-
tivity of ligaments diminishes as they approach realistic
traumatic injury strain rates (Yamamoto et al., 2003; Crisco
et al., 2002). Crisco et al. (2002) recognized that the stiffness, k
and failure load Ff of lapine stifle joint ligaments were
insensitive to strain rate at two different high loading rates
(at 36.6/s: k¼145 N/mm, Ff ¼434 N; at 140/s: k¼136 N/mm,
Ff ¼443 N). Elsewhere, a logarithmic relationship between
strain rate and failure stress, σf up to strain rates of an order
of magnitude less than in the current study has been reported
for the MCL of the lapine stifle joint ðσ f ¼ 14:2log10 ε_ þ 85:9Þ
(Yamamoto et al., 2003). Interestingly, the logarithmic fit of our
data results in coefficients that are not very dissimilar to the
study on the lapine MCL (Yamamoto et al., 2003).
Many of the previous studies investigated the failure
characteristics of a whole bone–ligament–bone structural
complex, rather than the material properties of the ligament
itself. This approach renders some comparisons with this
study difficult because the material properties of ligament
tissue could have been underestimated if the structures failed
at the bony attachment of the ligament or within the bone.
Yamamoto et al. showed that the tibial attachment of the
lapine MCL demonstrated strain-rate sensitivity, but the fail-
ure load was less than that of the mid-substance of the
ligament at all strain rates (Yamamoto and Hayashi, 1998).
The strain-rate sensitivity limit described in our study helps
us to understand the changes in the site of failure in a
ligament structural complex, which has been described at
different strain rates (Noyes et al., 1974). For example, a cross-
over from bone failure to ligament failure may occur as strain
rate increases if the tensile failure stress of a ligament
reaches its strain-rate sensitivity limit, but the strength of
the surrounding bone continues to change relative to strain
rate. This finding is supported by the clinical literature, which
shows that the failure characteristics of ligaments vary
amongst different ligaments and with different mechanisms.
For example, mid-substance tears of the anterior cruciate
ligament of the knee are very common in low speed sporting
injuries, but both mid-substance rupture and avulsion frac-
tures are common in high speed injuries associated with
knee dislocations after motor vehicle collisions (Kennedy
et al., 1974; Twaddle et al., 2003). In contrast, Noyes et al.
found in a laboratory study that avulsion fractures occurred
at very slow strain rates and mid-substance tears of the
268 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270
Fig. 7 – Fibril modulus of the human LCL across strain rates
(5 samples at each strain rate). Error bars represent 71SD.
anterior cruciate ligament (ACL) were more common at
higher speeds (Noyes et al., 1974; Noyes, 1977). This apparent
anomaly may be explained by either differences in experi-
mental design or, perhaps, due to variable strain-rate sensi-
tivity between different tissues, ligaments, species, joints and
health of the specimens. This might be supported by studies
that have observed that the strain-rate sensitivity varies between
different ligaments from around the same joint (Yamamoto
et al., 2003; Kennedy et al., 1976; Schenck et al., 1998).
The stress–strain curves from the low strain-rate macroscale
experiments presented here are typical of ligament behaviour;
they exhibit an initial non-linear, toe region up to 3–4% strain
followed by a linear region (Fig. 4a). It has been suggested that
the toe region is due to the macroscopic uncrimping of collagen
fibrils followed by sequential straightening of molecular kinks
(Fig. 6a) within the gap regions of collagen fibrils (Diamant et al.,
1972; Misof et al., 1997; Liao and Belkoff, 1999). The linear part of
the stress–strain curves corresponds to the elongation of the
collagen fibrils (Fig. 6b) and sliding between collagen fibrils and
fibres, probably controlled by the surrounding non-collagenous
matrix (Screen et al., 2004). The increase in axial periodicity is
due to the stretching of the collagen triple helices and their
cross-links, with side-by-side gliding of adjacent molecules
(Mosler et al., 1985; Sasaki and Odajima, 1996). Furthermore,
the proteoglycan matrix between collagen fibrils also deforms
during this linear phase and becomes stiffer when the applied
strain rate increases (Puxkandl et al., 2002). Therefore, the
matrix is likely to be playing an important role on the strain-
rate sensitivity of ligaments (Scott and Thomlinson, 1998). We
observed a reduction in the size of the toe region with increasing
strain rate (Fig. 6b and c); this is likely due to the respective
decrease in macroscale crimping; in contrast to the slower
strain-rate synchrotron experiments we conducted, at the
higher strain rates elongation of the collagen fibrils was
observed immediately as the fibrils started to stretch (Fig. 6c).
Crucially, the synchrotron experiments we conducted show that
the collagen-fibril modulus increases with increasing strain rate
in a logarithmic-type fashion (Fig. 7); this enables us to speculate
the relationship between strain rate and tissue-level tensile
modulus up to 1/s. At the higher loading rates the collagen fibrils
debond from the highly viscous matrix and fibrillar sliding
occurs leading to macroscopic failure (Puxkandl et al., 2002).
Collagen–fibril load sharing is dependent on matrix crosslinking
(Fessel and Snedeker, 2011), and so the rapid disruption of
interconnections would lead to a poor transmission of tensile
stresses via crosslinking. This in turn would promote fibrillar
sliding, hence reduced fibrillar strain (Fig. 6c).
The structure-function relationship for the strain-rate
sensitivity of ligaments is also supported by evidence that
increasing hydration increases the strain-rate sensitivity of
the human patellar tendon (Haut and Haut, 1997). The
authors explain these findings by the interaction of the fluid
and solid phases of the tissue, leading to ‘hydraulic stiffening’
as strain rate increases.
An explanation for the observed strain-rate sensitivity
limit requires further investigation. However, the importance
of the structural arrangement of fibrils, cross-links, water and
proteoglycan matrix at lower strain rates indicates that this
may also determine why a strain-rate sensitivity limit occurs.
The results from this study are important to understand
the mechanics of human ligament failure, to aid traumatic
injury modelling and to develop mitigation strategies. If a
strain-rate sensitivity limit is common among all ligaments,
then high strain-rate material properties of any ligament can
be measured at a one or two deformation rates, so long as the
strain rate is known to be above this limit. This would
decrease significantly the number of future experiments that
would need to be performed, decrease the number of cada-
veric samples required, and increase the delivery of reliable
biomechanical measurements for traumatic injury modelling.
However, the application of the observed relationships and
strain-rate sensitivity limit to human injury modelling
requires some caution. The geometry of the ligament due to
joint position affects substantially the number of collagen
fibres recruited and should be considered in injury modell-
ing (Sidles et al., 1991). Furthermore, the values for tensile
modulus and failure stress of the porcine ligament we
obtained in this study are broadly similar to previous human
studies. There is no concrete evidence to support a strain-rate
sensitivity limit of human ligaments at this time; nonethe-
less, the nanoscale experiments presented here suggest
that the human ligament might exhibit a similar strain-rate
sensitivity limit. Unfortunately, synchrotron experiments at
strain rates high enough to investigate the material beha-
viour at the nanoscale beyond the strain-rate dependency
limit we found in our macroscale experiments are not yet
possible as current imaging technology is limited in terms of
sampling rates. Synchronous testing at high strain rates and
appropriate imaging of the microstructure would elucidate
the fibre–matrix relations that are likely responsible for the
macroscale strain-rate sensitivity limit we observed.
5. Conclusion
This study found that the material properties of the lateral
collateral ligament of the porcine stifle joint were strain-rate
dependent up to a limit of approximately 1/s, beyond which
 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270
there was no further significant change. This limit might be
due to hydraulic stiffening and rapid disruption of matrix–
fibril bonding at strain rates beyond 1/s that would impede
the transmission of tensile stresses through the collagen
fibres. As human ligament injury occurs at high strain rates,
our findings and experimental techniques can form the basis
for capturing human ligament material properties at injur-
ious strain rates.
Acknowledgements
Part of this work was conducted under the auspices of the
The Royal British Legion Centre for Blast Injury Studies at
Imperial College London. Therefore the financial support of
the Royal British Legion for NN, AK, SM and AMJB is grate-
fully acknowledged. The financial support of the Defence
Medical Services (DMS) for TJB, of the Royal Centre for
Defence Medicine (RCDM) for the acquisition of equipment
and consumables, of BBSRC for NN, and of ABF – The Soldiers’
Charity for SM are kindly acknowledged.
Annex A
The data (tensile modulus and stress at failure) were fitted to
bilinear and logarithmic mathematical models using the
Nelder–Mead (or polytope) algorithm for function minimiza-
tion (Nelder and Mead, 1965). The objective function (the one
to be minimized) was defined as the squared difference
between the mathematical model prediction and the experi-
mental point for every sample tested, whereby the indepen-
dent variable is strain rate and the dependent variable is
tensile modulus and stress at failure for each fit, respectively.
N
OF ¼ ∑ ðyi  f ðxi ÞÞ2
i¼1
for N data points ( ¼ number of samples tested), where
yi is the experimental dependent variable (i.e. tensile
modulus or stress at failure)
xi is the experimental independent variable (i.e. strain
rate)
f (x) is the mathematical model to be fitted. This is
f ðxÞ ¼ a1 log x þ a2
for the logarithmic fit, and
(
a1 x þ a2 ; xox0
f ðxÞ ¼
a3 ðx x0 Þ þ a4 ; xZ x0
for the bilinear fit, with the constraint x0 ¼ a4 a2 =a1 in order
for the two lines to meet at x¼ x0.
The parameters to be fitted are the ai's for both mathe-
matical models.
r e f e r e n c e s
Crowninshield, R.D., Pope, M.H., 1976. The strength and failure
characteristics of rat medial collateral ligaments. J. Trauma 16,
99–105.
269
Crisco, J.J., Moore, D.C., McGovern, R.D., 2002. Strain rate
sensitivity of the rabbit MCL diminishes at traumatic loading
rates. J. Biomech. 35, 1379–1385, http://dx.doi.org/10.1016/
S0021-9290(02)00167-7.
Danto, M.I., Woo, S.L., 1993. The mechanical properties of
skeletally mature rabbit anterior cruciate ligament and
patellar tendon over a range of strain rates. J. Orthop. Res. 11,
58–67, http://dx.doi.org/10.1002/jor.1100110108.
Diamant, J., Keller, A., Baer, E., Litt, M., Arridge, R.G., 1972.
Collagen; ultrastructure and its relation to mechanical
properties as a function of ageing. Philos. Trans. R. Soc. Lond.
B 180, 293–315, http://dx.doi.org/10.1098/rspb.1972.0019.
Fratzl, P., Misof, K., Zizak, I., Rapp, G., Amenitsch, H., Bernstorff,
S., 1997. Fibrillar structure and mechanical properties of
collagen. J. Struct. Biol. 122, 119–122, http://dx.doi.org/10.1006/
jsbi.1998.3966.
Fessel, G., Snedeker, J.G., 2011. Equivalent stiffness after
glycosaminoglycan depletion in tendon – an ultra-structural
finite element model and corresponding experiments. J.
Theor. Biol. 268, 77–83, http://dx.doi.org/10.1016/j.
jtbi.2010.10.007.
Gelber, A.C., Hochberg, M.C., Mead, L.A., Wang, N.Y., Wigley, F.M.,
Klag, M.J., 2000. Joint injury in young adults and risk for
subsequent knee and hip osteoarthritis. Ann. Intern. Med.
133, 321–328, http://dx.doi.org/10.7326/0003-4819-133-5-
200009050-00007.
Germscheid, N.M., Thornton, G.M., Hart, D.A., Hildebrand, K.A.,
2011. A biomechanical assessment to evaluate breed
differences in normal porcine medial collateral ligaments.
J. Biomech. 44, 725–731, http://dx.doi.org/10.1016/j.
jbiomech.2010.10.036.
Goodship, A.E., Birch, H.L., 2005. Cross sectional area
measurement of tendon and ligament in vitro: a simple, rapid,
non-destructive technique. J. Biomech. 38, 605–608, http://dx.
doi.org/10.1016/j.jbiomech.2004.05.003.
Gupta, H.S., Seto, J., Krauss, S., Boesecke, P., Screen, H.R., 2010. In
situ multi-level analysis of viscoelastic deformation
mechanisms in tendon collagen. J. Struct. Biol. 169, 183–191,
http://dx.doi.org/10.1016/j.jsb.2009.10.002.
Haut, T.L., Haut, R.C., 1997. The state of tissue hydration
determines the strain rate-sensitive stiffness of human
patellar tendon. J. Biomech. 30, 79–81, http://dx.doi.org/
10.1016/S0021-9290(96)00108-X.
Kennedy, J.C., Hawkins, R.J., Willis, R.B., Danylchuck, K.D., 1976.
Tension studies of human knee ligaments. Yield point,
ultimate failure, and disruption of the cruciate and tibial
collateral ligaments. J. Bone Joint Surg. Am. 58, 350–355.
Kennedy, J.C., Weinberg, H.W., Wilson, A.S., 1974. The anatomy
and function of the anterior cruciate ligament. As determined
by clinical and morphological studies. J. Bone Joint Surg. Am.
56, 223–235.
Karunaratne, A., Esapa, C.R., Hiller, J., Boyde, A., Head, R., Bassett,
J.H., et al., 2012. Significant deterioration in nanomechanical
quality occurs through incomplete extrafibrillar
mineralization in rachitic bone: evidence from in-situ
synchrotron X-ray scattering and backscattered electron
imaging. J. Bone Miner. Res. 27, 876–890, http://dx.doi.org/
10.1002/jbmr.1495.
Karunaratne, A., Terrill, N.J., Gupta, H.S., 2013. Synchrotron X-ray
nanomechanical imaging of mineralized fiber composites.
Methods Enzymol. 532, 415–473, http://dx.doi.org/10.1016/
B978-0-12-416617-2.00019-9.
Liao, H., Belkoff, S.M., 1999. A failure model for ligaments. J.
Biomech. 32, 183–188, http://dx.doi.org/10.1016/S0021-9290(98)
00169-9.
Masouros, S.D., Parker, K.H., Hill, A.M., Amis, A.A., Bull, A.M.J.,
2009. Testing and modelling of soft connective tissues of
270 journal of the mechanical behavior of biomedical materials 41 (2015) 261 –270
joints: a review. J. Strain Anal. Eng. Des. 44, 305–318, http://dx.
doi.org/10.1243/03093247JSA507.
Misof, K., Rapp, G., Fratzl, P., 1997. A new molecular model for
collagen elasticity based on synchrotron X-ray scattering
evidence. Biophys. J. 72, 1376–1381, http://dx.doi.org/10.1016/
S0006-3495(97)78783-6.
Mosler, E., Folkhard, W., Knörzer, E., Nemetschek-Gansler, H.,
Nemetschek, T., Koch, M.H.J., 1985. Stress-induced molecular
rearrangement in tendon collagen. J. Mol. Biol. 182, 589–596,
http://dx.doi.org/10.1016/0022-2836(85)90244-X.
Ng, B.H., Chou, S.M., Lim, B.H., Chong, A., 2004. Strain rate effect
on the failure properties of tendons. Proc. Inst. Mech. Eng. H
218, 203–206, http://dx.doi.org/10.1243/095441104323118923.
Noyes, F.R., DeLucas, J.L., Torvik, P.J., 1974. Biomechanics of
anterior cruciate ligament failure: an analysis of strain rate
sensitivity and mechanisms of failure in primates. J. Bone
Joint Surg. Am. 56, 236–253.
Nagasaka, K., Mizuno, K., Tanaka, E., Yamamoto, S., Iwamoto, M.,
Miki, K., Kajzer, J., 2003. Finite element analysis of knee injury
risks in car-to-pedestrian impacts. Traffic Inj. Prev. 4, 345–354,
http://dx.doi.org/10.1080/714040492.
Noyes, F.R., Grood, E.S., 1976. The strength of the anterior cruciate
ligament in humans and Rhesus monkeys. J. Bone Joint Surg.
Am. 58, 1074–1082.
Noyes, F.R., 1977. Functional properties of knee ligaments and
alterations induced by immobilization: a correlative
biomechanical and histological study in primates. Clin.
Orthop. Relat. Res., 210–242.
Nelder, J.A., Mead, R., 1965. A simplex method for function
minimisation. Comput. J. 7, 308–313, http://dx.doi.org/10.1093/
comjnl/7.4.308.
Polinder, S., Meerding, W.J., van Baar, M.E., Toet, H., Mulder, S.,
van Beeck, E.F., EUROCOST Reference Group, 2005. Cost
estimation of injury-related hospital admissions in 10
European countries. J. Trauma 59, 1283–1291, http://dx.doi.org/
10.1097/01.ta.0000195998.11304.5b.
Puxkandl, R., Zizak, I., Paris, O., Keckes, J., Tesch, W., Bernstorff,
S., Purslow, P., Fratzl, P., 2002. Viscoelastic properties of
collagen: synchrotron radiation investigations and structural
model. Philos. Trans. R. Soc. Lond. B 357, 191–197, http://dx.
doi.org/10.1098/rstb.2001.1033.
Ramasamy, A., Masouros, S.D., Newell, N., Hill, A.M., Proud, W.G.,
Brown, K.A., Bull, A.M.J., Clasper, J.C., 2011. In-vehicle
extremity injuries from improvised explosive devices: current
and future foci. Philos. Trans. R. Soc. Lond. B 366, 160–170,
http://dx.doi.org/10.1098/rstb.2010.0219.
Schenck, R.C., Kovach, I.S., Agarwal, A., Brummett, R., Ward, R.A.,
Lanctot, D., Athanasiou, K.A., 1999. Cruciate injury patterns in
knee hyperextension: a cadaveric model. Arthroscopy 15,
489–495, http://dx.doi.org/10.1053/ar.1999.v15.0150481.
Schenck, R., Kovach, I.S., Agarwal, A., Brummett, R., 1998. Strain
rate sensitivity of combined knee ligament injuries. In:
Proceedings of the 17th Southern Biomedical Engineering
Conference, vol. 55. (10.1109/SBEC.1998.666661).
Screen, H.R.C., Lee, D.A., Bader, D.L., Shelton, J.C., 2004. An
investigation into the effects of the hierarchical structure of
tendon fascicles on micromechanical properties. Proc. Inst.
Mech. Eng. H 218, 109–119, http://dx.doi.org/10.1243/
095441104322984004.
Sasaki, N., Odajima, S., 1996. Elongation mechanism of collagen
fibrils and force-strain relations of tendon at each level of
structural hierarchy. J. Biomech. 29, 1131–1136, http://dx.doi.
org/10.1016/0021-9290(96)00024-3.
Scott, J.E., Thomlinson, A.M., 1998. The structure of interfibrillar
proteoglycan bridges (shape modules’) in extracellular matrix
of fibrous connective tissues and their stability in various
chemical environments. J. Anat. 192 (Pt 3), 391–405, http://dx.
doi.org/10.1046/j.1469-7580.1998.19230391.x.
Sidles, J.A., Clark, J.M., Garbini, J.L., 1991. A geometric theory of
the equilibrium mechanics of fibers in ligaments and tendons.
J. Biomech. 24, 943–949, http://dx.doi.org/10.1016/0021-9290
(91)90172-J.
Tipton, C.M., Schild, R.J., Flatt, A.E., 1967. Measurement of
ligamentous strength in rat knees. J. Bone Joint Surg. Am. 49,
63–72.
Twaddle, B.C., Bidwell, T.A., Chapman, J.R., 2003. Knee
dislocations: where are the lesions? A prospective evaluation
of surgical findings in 63 cases. J. Orthop. Trauma 17, 198–202.
Woo, S.L.-Y., Peterson, R.H., Ohland, K.J., Sites, T.J., Danto, M.I.,
1990. The effects of strain rate on the properties of the medial
collateral ligament in skeletally immature and mature rabbits:
a biomechanical and histological study. J. Orthop. Res. 8,
712–721, http://dx.doi.org/10.1002/jor.1100080513.
Xerogeanes, J.W., Fox, R.J., Takeda, Y., Kim, H.S., Ishibashi, Y.,
Carlin, G.J., Woo, S.L., 1998. A functional comparison of animal
anterior cruciate ligament models to the human anterior
cruciate ligament. Ann. Biomed. Eng. 26, 345–352, http://dx.
doi.org/10.1114/1.91.
Ytterstad, B., 1996. The Harstad injury prevention study: the
epidemiology of sports injuries. An 8 year study. Br. J. Sports
Med. 30, 64–68, http://dx.doi.org/10.1136/bjsm.30.1.64.
Yannas, I.V., Olson, D.A., 1972. Linear relaxation analysis of the
mechanochemical transformation of collagen fibers.
Biopolymers 11, 899–912, http://dx.doi.org/10.1002/
bip.1972.360110413.
Yamamoto, S., Saito, A., Nagasaka, K., Sugimoto, S., Mizuno, K.,
Tanaka, E., Kabayama, M., 2003. The strain rate dependence of
mechanical properties of rabbit knee ligaments. In:
Proceedings of the 18th Enhanced Safety Vehicles, pp. 1–9.
Yamamoto, N., Hayashi, K., 1998. Mechanical properties of rabbit
patellar tendon at high strain rate. Biomed. Mater. Eng. 8, 83–90.