micromechanical properties of human trabecular bone: a hierarchical investigation using...

Micromechanical properties of human trabecular bone:A hierarchical investigation using nanoindentation

Jonathan Norman,1,2 Joe G. Shapter,2 Ken Short,3 Lachlan J. Smith,1,4 Nicola L. Fazzalari1,41Bone and Joint Research Laboratory, Division of Tissue Pathology, Institute of Medical andVeterinary Science and Hanson Institute, Adelaide, Australia2School of Chemistry, Physics, and Earth Sciences, Flinders University of South Australia, Adelaide, Australia3Materials and Engineering Science Division, Australian Nuclear Science andTechnology Organisation, Menai, Australia4Discipline of Pathology, School of Medical Sciences, University of Adelaide, Australia

Received 4 February 2007; revised 27 June 2007; accepted 18 September 2007Published online 17 December 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.31766

Abstract: The ability to assess the risk of fracture, evaluatenew therapies, predict implant success and assess the influ-ence of bone remodeling disorders requires specific mea-surement of local bone micromechanical properties. Nanoin-dentation is an established tool for assessing the microme-chanical properties of hard biological tissues. In this study,elastic modulus and hardness were quantified using nanoin-dentation for human trabecular bone from the intertrochan-teric region of the proximal femur. These properties were

demonstrated to be heterogeneous and highly correlated atthe intraspicule, interspicule, and interspecimen levels. Theresults of this study have important implications for currentunderstanding of structure–function relationships through-out the trabecular bone structural hierarchy. � 2007 WileyPeriodicals, Inc. J Biomed Mater Res 87A: 196–202, 2008

Key words: trabecular bone; structural hierarchy; micro-mechanical properties; nanoindentation

INTRODUCTION

Bone material has a complex hierarchical struc-ture. The ability to assess the risk of fracture, evalu-ate new therapies, predict implant success, andassess the influence of bone remodeling disordersrequires specific measurement of local bone micro-mechanical properties. Anatomical variations in cort-ical and trabecular bone material properties havebeen extensively quantified at the continuum level.1

Observed mechanical heterogeneity suggests that tis-sue level variations may exist in the intrinsic mate-rial properties of trabecular bone; however, thenature of these variations at the microscopic levelremains largely unknown.

Nanoindentation, a technique widely applied inmaterials science, is capable of describing the micro-

mechanical properties, including hardness and elas-tic modulus, of material surfaces.2,3 There have beenseveral studies in which nanoindentation has beenapplied to examine the mechanical properties ofcortical bone; relatively few, though, have studiedtrabecular bone.4–7 None of these have examined thedistribution of the micromechanical properties asthey relate to multiple levels of the trabecular struc-tural hierarchy, that is within an individualtrabecular spicule (intraspicule), between spicules inthe same specimen (interspicule), and betweenspicules from different specimens (interspecimen).

The objectives of this study were to determine thehardness and elastic modulus of trabecular spiculesfrom nondiseased human bone obtained from theintertrochanteric region of the proximal femur, andto examine the heterogeneity in these properties atthe levels of the hierarchy described.

METHODS AND MATERIALS

Specimen collection and preparation

Human bone specimens from the intertrochantericregion of the proximal femur (Fig. 1) were obtained

Correspondence to: N. L. Fazzalari; e-mail: [email protected] grant sponsors: National Health and Medical

Research Council of Australia, University of Adelaide,Flinders University, Australian Institute of Nuclear Scienceand Engineering

� 2007 Wiley Periodicals, Inc.

at autopsy with the approval of the Royal AdelaideHospital Research Ethics Committee and theinformed consent of the next-of-kin. Donors werefive males aged 50, 64, 65, 71, and 85 years. Using X-rays, case notes, and qualitative histological assess-ment, all cases were determined to be withoutevidence of skeletal disease. The intertrochantericregion was selected as the preferred sampling sitebecause it is remote from articular surfaces and com-mon fracture sites, further limiting any possibleinfluences of pathology on results.

Specimens were initially fixed and dehydratedusing serial ethanol concentrations in a manner con-sistent with previous studies.5,8,9 They were thenpolymerized in a mixture of polymethylmethacrylateand polyethylene glycol. The face of each specimenblock, vertically oriented parallel to the anterior faceof the femur (Fig. 1), was polished using abrasivepapers with progressively decreasing silicon carbidegrit size from 30 to 2 lm until the specimen surfaceshowed a smooth, mirror like finish, which wasessential prior to scanning electron microscopy(SEM), atomic force microscopy (AFM) imaging andnanoindentation.

SEM

SEM (JSM 5000, JEOL, Japan) was used to imageeach entire test surface. Specimens were initiallycoated with a thin layer of carbon. Images, taken at153 magnification at a working distance of 15 mmand with an accelerating voltage of 15 kV, were col-

lated into a montage to form a spatial map of thesurface of each specimen. These montages were usedto select the spicule locations for nanoindentation.

Preparation for AFM and nanoindentation

Following SEM imaging, specimen surfaces wereirrigated with water and 200 lm sections wereremoved using an Isomet low-speed saw (Buehler,Dusseldorf, Germany). The carbon coating wasremoved using abrasive paper in the same manneras previously described, followed by a dry aluminamicropolish (1, 0.3, and 0.05 lm) (Buehler, Dussel-dorf, Germany). Between each successive grade ofmicropolish, the specimens were sonicated in waterfor 10 min to remove any remaining polish on thesurface, in a similar fashion to previously establishedmethods.3

AFM

Specimens were reduced in size (laterally, to fit onthe AFM head) using the low-speed saw, and doublesided tape was used to fix them to steel pucks.Nanoscope IV and Dimension 3000 scanning probemicroscopes (Digital Instruments, Santa Barbara,CA) were used for all AFM imaging. Specimenswere imaged under tapping mode using silicon tipswith resonant frequencies of 275–350 kHz (TM300,Digital Instruments, Santa Barbara, CA) so that nodamage occurred to the bone surface. A J-scanner(120 lm scan range) and E-scanner (10 lm scanrange) (Digital Instruments, Santa Barbara, CA) wereused for large (10–120 lm) images and small images(1–10 lm), respectively.

Nanoindentation

Nanoindentations were performed using a Nano-indenter II (MTS Systems, Oak Ridge, TN) nano-indentation system, with a Berkovich tip in load-controlled mode. Prior to conducting the experiment,the indentor was calibrated using fused silica.Hengsberger et al.10 reported that drying time sys-tematically affects the modulus and hardness of thebone specimens. A drying time of 5 days at roomtemperature in a standardized environment was con-sidered sufficient for time-associated wetting effectsto stabilize and enable sample comparisons. Sectionswere affixed to aluminum stubs prior to testing.

The load-time procedure was developed from lit-erature values of loading/unloading rates and maxi-mum loads.9 A maximum load of 6 mN correspond-ing to an approximate indentation depth of 700 nmwas used (Fig. 2), and indents were spaced 50 lm

Figure 1. Representative X-ray image of the proximalfemur showing the sampling site and analysis orientationin the intertrochanteric region.

MICROMECHANICAL PROPERTIES OF HUMAN TRABECULAR BONE 197

Journal of Biomedical Materials Research Part A

apart to prevent interference. A loading rate of 350lm s21 was selected to minimize the effects of vis-coelasticity on the measured values of elastic modu-lus, and optimal holding times were selected tominimize the effects of creep and time-dependantplasticity on the measured values of hardness andelastic modulus, respectively.11 The testing protocolconsisted of three initial load/unload segments, fol-lowed by a 40-s holding time at peak load and thenunloading to 10% of the peak load and further hold-ing for 20 s. This repeated loading pattern wasdesigned to further reduce the time-dependent visco-elastic effects.11

Within each of the five specimens, between 2 and8 randomly selected individual trabecular spiculeswere investigated. Two-dimensional arrays ofbetween 10 and 20 indents (depending on spiculesize) were then made across each region of interest,facilitating an analysis of variations in mechanicalproperties in the transverse and longitudinal direc-tions (Fig. 3). This methodology facilitated the hier-archical approach, which has been described (intra-spicule, interspicule and interspecimen). Hardnessand effective elastic modulus were calculated fromthe initial 45% of the unloading curve, with anassumed Poisson’s ratio for bone of 0.3, using themethods of Oliver and Pharr.12

Statistical analysis

Statistical analyses were performed using SASsoftware (SAS Institute, Cary, NC). Comparisons ofthe variations of both elastic modulus and hardnessmeasurements at the intraspicule, interspicule, andinterspecimen levels were undertaken using one-wayANOVA, with Student–Neumann–Keuls post hoc anal-ysis applied where significance was detected. Signifi-cance was reported at a confidence level of 95%.

RESULTS

Figure 2 illustrates a typical load-displacementprofile for an indentation made in a trabecular spi-

Figure 2. A typical cyclic load–displacement profile foran indentation in trabecular bone.

Figure 3. (A) Two-dimensional hardness map and (B) modulus map illustrating heterogeneous mechanical properties forthe spicule cross-section shown in (C). Indent positions were defined according to a transverse (1 to 3)–longitudinal (1 to10) coordinate system, indicated by the grid in (C). The indents are spaced at 50-lm intervals. Results are omitted for gridpositions (1,1) and (3,5), which correspond to the locations of reference indentations.

198 NORMAN ET AL.


cule. When indented directly, the embedding mate-rial adjacent to spicules exhibited mechanical proper-ties which were significantly lower, enabling cleardifferentiation between the two materials.

Multidimensional mapping of elastic modulus andhardness in the transverse and longitudinal direc-tions (short and long axes respectively, as theyappear on the section) of individual trabecular spi-cules revealed that the micromechanical propertieswere heterogeneous in both these spatialities(example shown in Fig. 3). In the transverse direc-tion, the mean elastic modulus of the inner medialbone of trabeculae was 17% higher than that of theouter margin (inner 5 14.22 6 1.07 GPa, outer 5

12.25 6 1.01 GPa, p < 0.01). Similarly, the meanhardness of the inner medial bone was 32% higherthan that of the outer margin (inner 5 0.90 6 0.18GPa, outer 5 0.68 6 0.11 GPa, p < 0.01).

Although the expected peak in both the hardnessand elastic modulus was observed in the transversedirection, variations along the length of the spiculewere also observed. The coefficient of variation forhardness was 20% along the inner medial region incontrast to 15% along the outer margin. The coeffi-cient of variation of the elastic modulus was invari-ant at 8% along both the inner medial region andouter margin; variations in hardness and elastic

modulus were highly correlated (r 5 0.94, p <0.001).

Individual spicules within a single specimen forall cases studied also show a significant positive cor-relation between hardness and elastic modulus (p <0.04). In the 65-year-old specimen (representative ofall five cases), the coefficient of variation betweenthe mean hardness of individual spicules was 8%whereas that of the elastic modulus was 17%. AnANOVA assessment of the mean hardness revealedno significant difference between individual spicules[Fig. 4(C)] at the eight locations indicated [Fig. 4(A)].In contrast, consistent with the magnitude of thecoefficient of variation above, the elastic moduluswas significantly different between spicules, withpost hoc analysis revealing that this variationoccurred between four distinct groups [Fig. 4(B)]. Re-ferring to the SEM montage [Fig. 4(A)], the mini-mum elastic modulus was observed to be associatedwith an axial cross-sectional cut through a spicule(region 1).

AFM images (Fig. 5) were qualitatively assessedfor visible heterogeneity in surface structure betweensampling sites. It was observed that within a 10 lm2

sampling zone, surface structure varied from beingheterogeneous, in which distinct elements were visi-ble [Fig. 5(A)], to broadly homogeneous [Fig. 5(B)].

Figure 4. A: Scanning electron micrograph montage of a specimen from the intertrochanteric region of the proximalfemur for a 65-year-old male showing discrete sampling sites, labeled 1–8, with accompanying plots of (B) elastic moduli,and (C) hardness for each trabecular spicule (mean 6 SD). Post hoc analysis revealed elastic moduli were significantlydifferent between groups I, II, III, and IV.



The variations in hardness and elastic modulusbetween each specimen were also found to be highlycorrelated (r 5 0.84, p < 0.03). The pooled coeffi-cients of variation were 14% and 22% for mean spec-imen hardness and elastic modulus, respectively.

The results of an ANOVA assessment of interspe-cimen hardness showed no significant variation[Fig. 6(B)]. Conversely, statistically significant differ-ences were demonstrated for elastic moduli [Fig.6(A)]. Post hoc analysis revealed that the 64- and 71-year-old males had an elastic modulus significantlylower than those aged 50, 65, and 85 years [indicatedby I and II in Fig. 6(A), respectively].

DISCUSSION

In this study, elastic modulus and hardness valuesmeasured using nanoindentation are reported for

trabecular bone from the intertrochanteric region ofthe human proximal femur. Variations in these prop-erties at multiple levels of the trabecular bone struc-tural hierarchy are described.

The results reported here are comparable withthose from previous studies. Zysset et al.7 reportedan average trabecular spicule elastic modulus of 11.46 5.6 GPa in the femoral neck and hardness valuesranging from 0.23 to 0.76 GPa. Their study, however,was carried out under wet conditions, under whichsamples were constantly irrigated with a phosphate-buffered saline solution. When indenting in the lon-gitudinal direction of trabeculae, Rho et al.5 reportedan average of 15.0 6 2.5 GPa for the elastic modulusand 0.52 6 0.08 GPa for the hardness. Rho et al.also examined the elastic modulus of transverselyindented vertebral trabeculae from males and re-ported the average elastic modulus to be 13.4 6 2.0GPa and 0.47 6 0.10 GPa for the hardness, where

Figure 5. Representative AFM images of trabecular bone demonstrating that over similar areas, the matrix structure mayappear (A) heterogeneous or (B) homogeneous.

Figure 6. Interspecimen variability in (A) elastic modulus and (B) hardness for trabecular spicules from the intertrochan-teric region of the proximal femur (mean 6 SD). Post hoc analysis revealed elastic moduli were significantly differentbetween groups I and II.

200 NORMAN ET AL.


indentations were carried out under dry conditions.6

The results of the current study, also obtained underdry conditions, are broadly consistent with these re-ported data, with variations most likely attributableto intrinsic material differences in the tissues underanalysis, and to minor differences in the methods ofpreparation and testing.

It was found that the ranges of elastic moduli andhardnesses both across spicules in the same speci-men and across spicules in different specimens werecomparable with the skeletal site differences thathave been previously reported. Hoffler et al.4

reported significant differences between various ana-tomical locations with respect to both the elasticmodulus and hardness in nine males aged 45–85years. They reported trabecular elastic modulus andhardness values of 8.02 6 1.31 GPa and 0.32 6 0.06GPa, respectively in the vertebral body, 10.5 6 1.6and 0.38 6 0.06 GPa in the femoral neck, and 13.756 1.67 and 0.48 6 0.07 GPa in the distal radius. Thiscurrent study reports intraspecimen and interspeci-men variability of the micromechanical properties oftrabecular spicules for five males aged 50–85 years.The variability in the mean elastic modulus, 15.3 6

3.6 GPa, and hardness, 0.72 6 0.10 GPa, was compa-rable with that previously reported, and may be in-dicative of the range of elastic moduli and hardnessin aging male trabecular bone.

Observed variations in mechanical propertieswithin a trabecular spicule are likely to be attribut-able to differences in biochemical composition, pri-marily being mineral and collagen, the strength ofthe structural bonds between these constituents andtheir relative microarchitectural arrangement.13,14

Bone remodeling results in nonuniform distributionof mineral within spicules, as the rate of bone matrixturnover is greater at the spicule surface than at itscore.15 Bone mineral content, which is correlatedwith the rate of matrix turnover, is exponentiallyrelated to the tissue stiffness at the continuum level.13

The results of this study suggest that this relationshipholds true for individual trabecular spicules.

Compressive modulus and hardness may also beaffected by the anisotropy of the bone structure,attributable in part to differences in collagen fiberorientation between lamellae.16 The effects of anisot-ropy on the mechanical properties were recentlydemonstrated for the osteonic and interstitial lamel-lae of cortical bone.8 The likelihood of such effectsconfounding the data presented here is consideredsmall due to the statistical significance of the varia-tions, which were observed over a large number ofdistinct zones, however, the premise could be testedin future studies by combining nanoindentationexperiments with polarized light microscopy. Addi-tionally, regions of diffuse microdamage sufferedin vivo at locations within the trabecular structure

under high physiological stress17 may exhibit alteredcompressive mechanical properties, although thiswould also be unlikely to significantly confound ourresults for the same reasons presented for aniso-tropy.

The physiological forces experienced by eachlamella are unique according to their relative posi-tions in the spicule, and it has been shown that arelationship exists between bone lamella compres-sive moduli and the magnitude of the functionalstrains they experience.18 It is possible that the varia-tions observed within individual spicules reflect theadaptation of the matrix to an ideal homeostaticstate in response to the nonuniformly distributedexternal mechanical stimuli. For example, lamellaeproximal to the surface of the strut experiencegreater strains in lateral bending than those near thecenter.

There are a number of possible functional conse-quences, both advantageous and detrimental, for tra-becular structures with micromechanical properties,which vary from maxima along their spicules’ cen-tral axes to minima at their outer margins. In bend-ing, such a design would reduce the propensity ofthe outer margins of spicules, where strains arehighest, to suffer brittle failure, representing a possi-ble functional advantage. At the same time, a weakermodulus at the periphery would likely reduce thebending resistance of the spicule. It is possible,therefore, that the described variations represent afunctional compromise of the bone matrix designedto balance the propensity of the spicule to fail byway of a number of distinct mechanisms. Alterna-tively, the variations and their functional consequen-ces may be an unavoidable effect of the remodelingprocess.

An uncontrolled variable in this investigation wasthe inability to define exactly the position of eachindividual indent. Indents were instead definedaccording to a two-dimensional array having a ran-domly assigned origin within the spicule under anal-ysis. On a ‘‘rippled’’ substrate, such as the bone sam-ples studied, this could result in a particular indentbeing centered directly on the crest of a ripple, andanother in a trough. Such variation could in theorychange the area function of the indent and hence thevalues obtained. Although this was a potential prob-lem, it did not result in any great variations in ourdata, as load versus displacement across all speci-mens were typically smooth. In reality, the indentsin many cases cover both peaks and valleys of thesubstrate to yield average material properties of thetested region of bone regardless of the exact point ofcontact.

Bone hardness, predominantly determined by theinorganic mineral component,15 was assumed to beinvariant over the area of the indent. Elastic modulus,



however, is likely to be a function of the inter-connectivity of organic and inorganic matrices, andmay vary considerably due to protein compositionand, perhaps more likely, the strength of the bondsat the component interfaces. AFM images (Fig. 5)indicate that regions of protein and inorganic min-eral vary in size. For example, regions consisting ofmany thin strands of interconnected matrix [Fig.5(A)] may have material properties that are distinctfrom regions having thicker, more homogeneousstructure [Fig. 5(B)]. It is possible that these observeddifferences may be related to variations in remodel-ing rates, which in turn result in heterogeneousmaterial properties.

In this study, the Poisson’s ratio used to calculatethe effective elastic modulus was assumed to be con-stant for all indents. In reality, it is likely that thestructural heterogeneity of the specimen surfacewould result in some variations in the Poisson’s ra-tio, possibly adding an additional element of uncon-trolled variability to the results.

There is a question as to whether mounting speci-mens in a soft polymer matrix influences theobserved material properties, even though this is theapproach most commonly used. The results of arecent study demonstrated that choice of embeddingmaterial does not significantly effect measurementsof elastic modulus, and that hardness measurementsare only affected when using embedding materialsof relatively low viscosity.9

In this study, we have shown that the microme-chanical properties of trabecular bone are heteroge-neous within the bounds of individual trabecularspicules, with hardness and elastic modulus greatestin the inner medial regions. In addition, these prop-erties were demonstrated to be heterogeneous acrossindividual spicules both within a single specimenand across multiple specimens. Elastic modulus andhardness were found to be highly correlated at eachlevel of the hierarchy defined. Correlations withbetween these properties and bone mineral (calcium)content could in future be evaluated using quantita-tive backscattered SEM, as has been done in othermusculoskeletal tissues.19 Our results provide anovel perspective as to the mechanical consequencesof bone remodeling on bone strength and stiffness,and further studies are required to determine howthese variations affect the whole bone mechanics.

The authors thank the mortuary staff at the Institute ofMedical and Veterinary Science for assistance with collec-tion of autopsy specimens, and Dr. Peter Sutton-Smith forhis assistance with electron microscopy. Insightful com-ments provided by Prof. John Currey are also gratefullyacknowledged.

References

1. Reilly DT, Burstein AH. The mechanical properties of corticalbone. J Bone Joint Surg Am 1974;56:1001–1022.

2. Oliver WC, Pharr GM. Measurement of hardness and elasticmodulus by instrumented indentation: Advances in under-standing and refinements to methodology. J Mater Res 2004;19:3–20.

3. VanLandingham MR. Review of instrumented indentation.J Res Natl Inst Stand Technol 2003;10:249–265.

4. Hoffler CE, Moore KE, Kozloff K, Zysset PK, Brown MB,Goldstein SA. Heterogeneity of bone lamellar-level elasticmoduli. Bone 2000;26:603–609.

5. Rho JY, Roy ME, Tsui TY, Pharr GM. Elastic properties ofmicrostructural components of human bone tissue as mea-sured by nanoindentation. J Biomed Mater Res 1999;45:48–54.

6. Rho JY, Tsui TY, Pharr GM. Elastic properties of human cort-ical and trabecular lamellar bone measured by nanoindenta-tion. Biomaterials 1997;18:1325–1330.

7. Zysset PK, Guo XE, Hoffler CE, Moore KE, Goldstein SA.Elastic modulus and hardness of cortical and trabecular bonelamellae measured by nanoindentation in the human femur.J Biomech 1999;32:1005–1012.

8. Fan Z, Swadener JG, Rho JY, Roy ME, Pharr GM. Anisotropicproperties of human tibial cortical bone as measured bynanoindentation. J Orthop Res 2002;20:806–810.

9. Mittra E, Akella S, Qin YX. The effects of embedding mate-rial, loading rate and magnitude, and penetration depth innanoindentation of trabecular bone. J Biomed Mater Res A2006;79:86–93.

10. Hengsberger S, Kulik A, Zysset P. Nanoindentation discrimi-nates the elastic properties of individual human bone lamel-lae under dry and physiological conditions. Bone 2002;30:178–184.

11. Fan Z, Rho JY. Effects of viscoelasticity and time-dependentplasticity on nanoindentation measurements of human corti-cal bone. J Biomed Mater Res A 2003;67:208–214.

12. Oliver WC, Pharr GM. An improved technique for determin-ing hardness and elastic-modulus using load and displace-ment sensing indentation experiments. J Mater Res 1992;7:1564–1583.

13. Currey JD. The effect of porosity and mineral content on theYoung’s modulus of elasticity of compact bone. J Biomech1988;21:131–139.

14. Currey JD. Role of collagen and other organics in themechanical properties of bone. Osteoporos Int 2003;5:29–36.

15. van der Linden JC, Birkenhager-Frenkel DH, Verhaar JA,Weinans H. Trabecular bone’s mechanical properties areaffected by its non-uniform mineral distribution. J Biomech2001;34:1573–1580.

16. Ziv V, Wagner HD, Weiner S. Microstructure–microhardnessrelations in parallel-fibered and lamellar bone. Bone 1996;18:417–428.

17. Fazzalari NL, Kuliwaba JS, Forwood MR. Cancellous bonemicrodamage in the proximal femur: Influence of age andosteoarthritis on damage morphology and regional distribu-tion. Bone 2002;31:697–702.

18. Goodwin KJ, Sharkey NA. Material properties of interstitiallamellae reflect local strain environments. J Orthop Res 2002;20:600–606.

19. Gupta HS, Schratter S, Tesch W, Roschger P, BerzlanovichA, Schoeberl T, Klaushofer K, Fratzl P. Two different corre-lations between nanoindentation modulus and mineral con-tent in the bone–cartilage interface. J Struct Biol 2005;149:138–148.

202 NORMAN ET AL.


micromechanical properties of human trabecular bone: a hierarchical investigation using...

Documents