Microstructure-Hardened SilverNanowiresBin Wu, Andreas Heidelberg, and John J. Boland*

School of Chemistry and the Centre for Research on AdaptiVe Nanostructures andNanodeVices (CRANN), Trinity College Dublin, Dublin 2, Ireland

John E. SaderDepartment of Mathematics and Statistics, The UniVersity of Melbourne, Australia

XiaoMing Sun and YaDong LiChemistry Department, Tsinghua UniVersity, Beijing, 100871, P R China

Received December 8, 2005; Revised Manuscript Received January 20, 2006

ABSTRACT

To exploit the novel size-dependent mechanical properties of nanowires, it is necessary for one to develop strategies to control the strengthand toughness of these materials. Here, we report on the mechanical properties of silver nanowires with a unique fivefold twin structure usinga lateral force atomic force microscopy (AFM) method in which wires are held in a double-clamped beam configuration. Force-displacementcurves exhibit super elastic behavior followed by unexpected brittle failure without significant plastic deformation. Thermal annealing resulte din a gradual transition to weaker, more ductile materials associated with the elimination of the twinned boundary structure. These resultspoint to the critical roles of microstructure and confinement in engineering the mechanical properties of nanoscale materials.

Metal nanowires have stimulated great interest as potentialbuilding blocks in nanoelectronic and nanoelectromechanicaldevices because of their high conductivities and highstrengths. Although a variety of metal nanowires such asgold, copper, and silver have been synthesized successfullyvia wet chemistry or template-directed approaches, much lessis known about the chemical or physical properties of free-standing nanomaterials at a single-object level. One of themain issues involved in the assembly of functional structuresusing these nano building blocks, however, is the relationshipbetween material properties and structural elements such assize, geometry, and microstructure. Many earlier studies havefocused on size-dependent properties.1-7 For example,micrometer-sized whiskers and nanowires have ultrahighstrength compared with that of their bulk counterparts.1-3

The exceptional strength is believed to originate from adecreased defect density resulting from reduced wire size.However, it is well established that microstructures such ascrystallinity, defects, and grain boundaries play importantroles in bulk material properties. Metals in bulk form canbe engineered to have high strengths by cold-work hardening,grain-boundary hardening, and precipitation hardening, tech-niques that rely on restricting and hindering dislocationmotion.8 However, on the nanometer scale strengtheningmethods that rely on the incorporation of impurities may be

ineffective because of facile surface segregation and expul-sion. This leaves microstructure as the best candidate routefor controlling material strength. To date, single crystalline,polycrystalline, and twin structured nanowires9-12 have beensynthesized, but the correlation between the microstructuresand mechanical properties is poorly understood, principallybecause of the difficulties in performing standard tensile orbending tests on individual wires.

Here in this work we demonstrate that microstructurecontrol is a particularly effective means for controllingmechanical properties in nanowire systems. Using a fivefoldtwinned Ag nanowire system, we show that the preciselycontrolled grain orientation and grain-boundary organizationwithin these wires is responsible for their anomalous strengthand brittle failure. The principle slip directions in these grainsintersect with the twinning boundaries that extend along theentire wire length to produce a uniformly hardened structure.These results demonstrate that, in the case of free-standingnanoscale materials, grain-boundary hardening is extraordi-narily effective because of the ability to completely controlthe orientation and boundaries of the limited numbers ofgrains in these materials.

Nanowire mechanical properties were measured using aAFM lateral bending technique developed recently,2 whichin contrast to earlier methods,3,13-23 allows the full spectrumof mechanical properties to be measured, ranging fromelasticity to plasticity and failure. Figure 1a shows a low-* To whom correspondence should be addressed. E-mail: jboland@tcd.ie.

NANOLETTERS

2006Vol. 6, No. 3

468-472

10.1021/nl052427f CCC: $33.50 © 2006 American Chemical SocietyPublished on Web 02/21/2006

magnification transmission electron microscopy (TEM) im-age of pentagonal silver nanowires prepared at a lowertemperature (140°C). Details of the preparation of thesepentagonal silver nanowires can be found elsewhere.9 Thesewires have typical lengths of many micrometers and diam-eters ranging from 16 to 35 nm. The cross-sectional TEMimage in Figure 1b shows the remarkable fivefold twinnedgrain-boundary structure that exists along the entire wirelength. Note that there is a∼2 nm thick carbon coating onthe surface as determined from high-resolution TEM obser-vation9 (data is not shown here).

To perform three-point bending tests on 16-35 nmdiameter pentagonal silver nanowires, we fabricated well-defined trench patterns on a substrate that was coated witha 10-20 nm TiN film. Typically, trenches having widths of650 nm and depths of 475 nm were used in these experi-ments. This selection provides a reasonable ratio betweenthe pinned nanowire length and its diameter while eliminatingproblems associated with wire droop. The pentagonal silvernanowires were dispersed in ethanol and then deposited onthe prepatterned substrates after solvent evaporation. Silvernanowires found to bridge well-defined trenches were locatedby scanning electron microscopy (SEM) in a dual-beam(electron/focused ion beam) system and subsequently double-clamped at the trench edges by electron-beam-induceddeposition of Pt lines. Details of the experimental conditionsand the pinning procedure can be found elsewhere.2 A well-calibrated rectangular cantilever (Budgetsensors) with anaverage normal force constant of 1-3 N/m (75 kHz) wasused in the bending experiments. AFM lateral manipulationswere carried out using a Digital Instruments NanomanSystem with closed-loopx-y-zscanner. By positioning the

AFM tip 450 nm into the trench, that is, 20-30 nm abovethe trench floor, frictional forces between the trench floorand tip were eliminated completely. Lateral bending mea-surements were then performed either as a single-shotexperiment (in which the tip engaged the wire, elasticallyand then plastically deformed it, and finally broke the wirein a single manipulation) or in a series of loading-unloadingcycles (in which the wire was increasingly loaded andunloaded in a series of manipulations so that progressiveelastic and then plastic deformation occurred, followed bywire failure). The normal and lateral force signals wererecorded using a Labview-based program. However, here wefocus on the lateral force because in this geometry the normalforce on the cantilever is less than 5% of the total force.2

Tip velocities were 20 nm/s throughout and all details relatingto the manipulation and tip calibration procedures can befound elsewhere.2

Figure 2 shows a typical set of experimental data includingAFM images before bending and after failure (Figure 2a andb) together with theF-d curves recorded during a series ofloading and unloading cycles (Figure 2c). The two curveslabeled 1 and 2 in Figure 2c (which are displaced relative toeach other for viewing purposes) are essentially nonlinearbut symmetric about the vertical dashed lines that identifythe starting point for unloading, that is, the cantilever’sturning point. This symmetry reflects the full elastic recoveryof the wire and is evident also from AFM images (not shown,but identical to Figure 2a) recorded both before andafterward, which reveal no permanent deformation of thewire. However, after reloading again the wire was subjectto a large manipulation that resulted inF-d curve 3 in Figure2c. In this curve, a sharp force-drop was observed and isassociated with wire failure, which is confirmed immediatelyby the subsequent AFM image (Figure 2b). TheF-d curvesare nonlinear and reproducible; both curves 1 and 2 can beshifted to completely overlap curve 3 in Figure 2. TheseF-dcurves were analyzed in terms of a generalized model thatincludes contributions from wire bending and tensile stretch-ing.24 This approach has the advantage that it provides anaccurate description of the mechanical properties over theentire range of elastic deformation. The exact analyticalsolution is expressed as24

R is related toε by (the detailed description of this approachcan be found elsewhere24)

Figure 1. (a) TEM image of pentagonal silver nanowires. (b)Cross-sectional high-resolution TEM image of pentagonal silvernanowires. (c) Schematic of slip systems for faced-centered cubicfivefold twinned Ag nanowires, showing that one of the slip planes(111) together with three possible slip directions⟨110⟩ inevitablyintersect the grain boundaries. Shaded planes indicate fivefoldtwinned boundaries.

Fcenter)192EI

L3f(R)∆zcenter (1)

wheref(R) ) R

48 -192 tanh(xR

4 )xR

ε ) ∆zcenter2 (AI ) (2)

R )6ε(140+ ε)350+ 3ε

(3)

Nano Lett., Vol. 6, No. 3, 2006 469

where Fcenter is the measured lateral force,E is Young’smodulus, I is the moment of inertia, and∆zcenter is thedisplacement of the wire of suspended lengthL and cross-sectional areaA. Because the wires have a circular crosssection as seen from the TEM image in Figure 1b, themoment of inertia isI ) πr4/4. AFM was used to determinethe diameter and the suspended length of pinned wires. Thediameter was measured at several points along the wires’length and average values based on a typical scatter of(1nm were used for fitting the curves. A complete analysis

also requires a detailed calibration of the AFM tip dimensionsand a determination of cantilever lateral spring constant, theprocedures for which are described elsewhere.2

TheF-d curves in Figure 2 were fitted using eqs 1-3 todetermine the mechanical properties of the nanowire: Young’smodulus, elastic deformation, yield point, and failure. Thedata analysis was performed using a MATHEMATICA 5.2-based algorithm that inputted all measured parameters (wirediameter, the suspended length of wires, lateral force constantof the cantilever, etc.) and in which the Young’s modulus,E, was the only adjustable parameter. Analysis involvedinitially fitting data at small displacement and then extendingthe range to include increasingly larger displacements. Thevalue of the modulus obtained by this method remainedconstant until the yield point associated with the onset ofplastic deformation was reached and after which the apparentvalue of the modulus dropped and the fit became increasinglypoorer. This approach is possible only because the general-ized formula provides a complete description of the elasticproperties over the entire range of displacements. Curve 3in Figure 2c is replotted in Figure 2d together with thecorresponding fit to the generalized theory. The excellentfit demonstrates that these pentagonal silver nanowiresexhibit very good elasticity up to displacements of∼80 nm;that is, these wires can be bent elastically up to an angle of∼13.2° from the wire original direction (see the inset inFigure 2c). The axial strain induced by bending is related tothe bending angle byε ) (1/cos θ) - 1. After the yieldpoint, Figure 2d indicates that the wire undergoes only 30nm of plastic deformation prior to failure. Significantly, wemeasure∼30 nm permanent displacement of wire in the post-failure AFM image Figure 2b, which underscores the validityof this analysis in the determination of both the modulusand yield point.

We have measured the mechanical properties of 24 singlewires, and Figure 3 shows values of the Young’s modulusdetermined for a range of pentagonal silver nanowires withdiameters from 22 to 35 nm. The data scatter falls withinthat typically found using this technique, and details of the

Figure 2. (a and b) Tapping-mode AFM images of a 23.6 nmdiameter pentagonal silver nanowire before bending and after brittlefailure. All scale bars are 250 nm. (c)F-d curves recorded duringthe consecutive manipulation by AFM tip-induced lateral bendingof a 23.6 nm pentagonal silver nanowire. Curves 1 and 2 showthat the wire was elastically loaded and unloaded. The unloadingpoints are identified as vertical dashed lines. Curve 3 is a single-shot experiment and shows nonlinear elastic behavior of silvernanowire, followed by limited plastic deformation and then brittlefailure. Note thatF-d curves are shifted for clarity. Inset: schematicof bending test showing that the bending angle defined as the anglebetween the deformed wire and its original direction. (d) Fit ofF-d curve 3 to the generalized formula, which yields a Young’smodulus of 90 GPa.

Figure 3. Plot of Young’s modulus versus the nanowire radius inthe range of 10-15 nm for pentagonal silver nanowires. TheYoung’s modulus remains essentially the same before (circle) andafter (star) thermal annealing experiments. The dashed line showsthe average value of the Young’s modulus for bulk silver.

470 Nano Lett., Vol. 6, No. 3, 2006

error propagation can be found elsewhere.2 The average valueof the modulus is 102( 23 GPa and is higher than that ofbulk silver (83 GPa). Oxygen plasma experiments thatremoved the 2 nm carbon coating indicated that the coatingis not responsible for the high modulus values of thepentagonal silver nanowires (see the Supporting Information).At present, the physical origin of the increased Young’smodulus is unknown, but similar observations have beenreported for other nanoscale systems.5-7

The observation in Figure 2d that the extent of plasticdeformation at failure is less than 40% of the elasticdeformation is exceptional for a pure metal. In the case ofgold nanowires, the total plastic deformation during bendingis up to 450% of the elastic deformation, an order ofmagnitude greater than that observed for the present silverwires.2 For bulk metals, strengthening typically involves theincorporation of impurities or the modification of themicrostructure. Although the former method is effective forengineering bulk materials, it has limited application on thenanoscale because impurities can be expelled easily fromthe material. The effectiveness of the microstructure modi-fication, however, is limited by physical size in bulkspecimens because there will always be grains with favorableorientations for plastic deformation. In contrast, the finitephysical dimensions and limited numbers of highly orientatednanoscale grains within the present Ag nanowires make themparticularly amenable to strengthening by microstructuremodification. We hypothesize here that the novel fivefoldtwin microstructure of the nanowires is responsible for theunique mechanical properties seen in Figure 2. Wire growthoccurs along the<011> direction of the face-centered cubic(FCC) crystal structure, which is characterized by a slipsystem of four (111) slip planes, along the three<1-10>directions (see Figure 1c). Consequently, the fivefold grainboundaries in these wires necessarily intersect with all ofthe possible slip systems and so the motion of dislocationsassociated with the initiation of plastic deformation alongany slip direction is restricted by the twinning boundariesthat extend into the center of the wire. In this manner, thefivefold twinned silver nanowires are effectively grain-boundary hardened materials, which sacrifices ductility forstrength. Because the twin boundaries exist along the entirelength of wire, the whole wire is uniformly hardened andthere are no defects that limit the strength of the wire.

To test this hypothesis, we carried out nanowire annealingexperiments at 240-250°C in nitrogen gas and then let themcool slowly to room temperature before removing andmechanically testing the samples. These annealing conditionsare sufficient to induce recrystallization even for bulk silver.As a result, atoms at the grain boundaries diffuse rapidly toreduce the interfacial energy, resulting in the gradualelimination of the unique pentagonal grain-boundary struc-ture. Figure 4 shows single-shot mechanical measurementsfor pentagonal silver nanowires that were annealed for 17and 48 h, respectively. These data, which are representativeof a study involving 15 nanowires, illustrate several importantpoints, which are summarized in Table 1. First, the Young’smodulus remains essentially unchanged before and after

annealing (see Figure 3). This can also be clearly seen fromthe solid curves in Figure 4a and b, which represents a bestfit to the generalized formula over the elastic region of thedata (see above). Second, theF-d curve fit for the 17 hannealed sample in Figure 4a shows that the wire is

Figure 4. (a) A typical single-shotF-d curve fit for a 23.2 nmdiameter pentagonal silver nanowire following 17 h thermalannealing treatment in N2. The red curve is the best fit of the elasticregion of F-d data showing three regimes: the elastic bending,grain-boundary hardened plastic bending and failure. The pointwhere the red curve departs from the data is identified as the yieldpoint. The curve fit gives a modulus of 121 GPa. (b) A typicalsingle-shotF-d curve fit for a 17 nm diameter silver nanowireafter 48 h thermal annealing. The modulus is 99 GPa in this case.(c) AFM images of a 16.5 nm diameter silver nanowire after 48 hthermal annealing in N2 during consecutive lateral bending tests.All scale bars are 250 nm.

Nano Lett., Vol. 6, No. 3, 2006 471

elastically deformed up to a displacement of 50 nm, whichcorresponds to a maximum elastic bending angle of∼8.2°,followed by 90 nm of plastic deformation. The extended 48h anneal resulted in an elastic displacement of 40 nm, areduced elastic bending angle of 6.9°, and 130 nm of plasticdeformation (Figure 4b). When compared with the mechan-ical properties of original unannealed wire, the annealed wireshave much reduced elasticity with lower yield strengths (seeTable 1). For these annealed wires, yielding at smalldeformation is induced by mechanical pure bending2 and theyield strength can be estimated asσy ) FyL/2πr3, resultingin values (∼7.3 GPa from Figure 4c) that are substantiallylarger than that for bulk silver (55 MPa). This analysis isnot valid for the original wires because yielding involvesboth stretching and bending, but is indicative of even strongermaterials. Finally, the data in Figure 4 also demonstrate thatthe transition from a brittle to ductile metal is a gradual one.Note that the yield point (the deviation point between thefitted curves and the data) in Figures 2d, 4a, and 4bprogressively moves to smaller displacements and the smallerthe elastic recovery following plastic deformation. This isconsistent with reduced hardening as the grain boundariesdisappear during annealing and is also evident from thevariation in the bending angle at failure, which range from17.9° for the original wire, to 21.9° and 27.1°, for the 17and 48 h annealed wires, respectively (see Table 1). Thisdramatic increase in ductility can also be seen from the AFMimages in Figure 4c recorded following consecutive ma-nipulation of the 48 h annealed wire.

These results demonstrate that the mechanical propertiesof nanowires can be uniquely tailored by controlling theirmicrostructure. Although microstructure is a recognizedmeans to engineering the strength of materials, it is excep-tionally effective on the nanoscale level because it is possibleto assemble materials with oriented, interlocking grains thatare both grain-boundary hardened and oriented so as toeliminate all favorable slip orientations.

Acknowledgment. This work was supported by ScienceFoundation Ireland under grant 00/PI.1/C077A.2.

Note Added after ASAP Publication. A coauthor wasnot included in the version published ASAP February 21,2006; the correct version was published March 3, 2006.

Supporting Information Available: The possible effectof carbon coating on the elastic properties of pentagonalsilver wires was performed by oxygen plasma etchingexperiments. AFM images, height analysis, and correspond-ing F-d curve fitting are shown. This material is availablefree of charge via the Internet at http://pubs.acs.org.

References

(1) Levitt, A. P. Whisker Technology; Levitt, A. P., Ed.; Wiley-Interscience: New York, 1970.

(2) Wu, B.; Heidelberg, A.; Boland, J. J. Nat. Mater.2005, 4, 525.(3) Wong, E. W.; Sheehan, P. E.; Lieber, C. M.Science1997, 277, 1971.(4) Uchic, M. D.; Dimiduk, D. M.; Florando, J. N.; Nix, W. D.Science

2004, 305, 986-989.(5) Cuenot, S.; Champagne, S. D.; Nysten, B.Phys. ReV. Lett. 2000,

85, 1690.(6) Cuenot, S.; Fretigny, C.; Demoustier-Champagne, S.; Nysten, B.Phys.

ReV. B. 2004, 69, 165410.(7) Liang, H. Y.; Upmanyu, M.Phys. ReV. B. 2005, 71, 241403(R).(8) Callister, W. D., Jr.Materials Science and Engineering: An

Introduction, third edition; John Wiley & Sons: New York, 1994.(9) Sun, X. M.; Li, Y. D. AdV. Mater. 2005, 17, 2626.

(10) Chen, H. Y.; Cao, Y.; Zhang, H. R.; Liu, L. B.; Yu, H. C.; Tian, H.F.; Xie, S. S.; Li, J. Q.J. Phys. Chem. B.2004, 108, 12038.

(11) Jiang, P.; Li, S. Y.; Xie, S. S.; Gao, Y.; Song, L.Chem.sEur. J.2004, 10, 4817.

(12) Wiley, B.; Sun, Y. G.; Mayers, B.; Xia, Y. N.Chem.sEur. J.2005,11, 454.

(13) Yu, M. F.; Lourie, O.; Dyer, M. J.; Moloni, K.; Kelly, T. F.; Ruoff,R. S.Science2000, 287, 637-640.

(14) Poncharal, P.; Wang, Z. L.; Ugarte, D.; De Heer, W. A.Science1999, 283, 1513-1516.

(15) Gao, R. P.; Wang, Z. L.; Bai, Z. G.; De Heer, W. A.; Dai, L. M.;Gao, M.Phys. ReV. Lett. 2000, 85, 622-625.

(16) Salvetat, J. P.; Briggs, G. A. D.; Bonard, J. M.; Bacsa, R. R.; Eulik,A. J.; Stockli, T.; Burnham, N. A.; Forro, L.Phys. ReV. Lett.1999,82, 944.

(17) Salvetat, J. P.; Kulik, A. J.; Bonard, J. M.; Briggs, G. A. D.; Stockli,T.; Metenier, K.; Bonnamy, S.; Beguin, F.; Burnham, N. A.; Forro,L. AdV. Mater. 1999, 11, 161.

(18) Kis, A.; Mihailovic, D.; Remskar, M.; Mrzel, A.; Jesih, A.; Piwonski,I.; Kulik, A. J.; Benoit, W.; Forro, L.AdV. Mater. 2003, 15, 733.

(19) Walters, D. A.; Ericson, L. M.; Casavant, M. J.; Liu, J.; Colbert, D.T.; Smith, K. A.; Smalley, R. E.Appl. Phys. Lett.1999, 74, 3803.

(20) Tombler, T. W.; Zhou, C. W.; Alexseyev, L.; Kong, J.; Dai, H. J.;Liu, L.; Jayanthi, C. S.; Tang, M. J.; Wu, S. Y.Nature2000, 405,769.

(21) Minot, E. D.; Yaish, Y.; Sazonova, V.; Park, J. Y.; Brink, M.;McEuen, P. L.Phys. ReV. Lett. 2003, 90, 1564011.

(22) Falvo, M. R.; Clary, G. J.; Taylor, R. M.; Chi, V.; Brooks, F. P.;Washburn, S.; Superfine, R.Nature1997, 389, 582.

(23) Li, X. D.; Gao, H. S.; Murphy, C. J.; Caswell, K. K.Nano Lett.2003, 3, 1495.

(24) Heidelberg, A.; Ngo, L. T.; Wu, B.; Phillips, M. A.; Sharma, S.;Kamins, T. I.; Sader, J. E.; Boland, J. J., submitted for publication.

NL052427F

Table 1. Summary of the Bending Angles in DifferentSamples

maximumelastic

bendingangles

bendingangle atfailure

ratio betweenplastic and

elasticdisplacements

original Agnanowire

13.2° 17.9° 35%

17 h annealedAg nanowire

8.2° 21.9° 169%

48 h annealedAg nanowire

6.9° 27.1° 304%

200-nm Aunanowirea

6.8° 34.0° 432%

a Typical bending angles for a 200-nm Au nanowire were shown forcomparison.2

472 Nano Lett., Vol. 6, No. 3, 2006