Biomaterials 24 (2003) 29412953

Fabrication of a new composite orthodontic archwire and validation by a bridging micromechanics model
Zheng-Ming Huanga,*, R. Gopalb, K. Fujiharab, S. Ramakrishnab, P.L. Lohc, W.C. Foongc, V.K. Ganeshc, C.L. Chewc
b a Department of Engineering Mechanics, Tongji University, 1239 Siping Road, Shanghai 200092, Peoples Republic of China Biomaterials Laboratory, Division of Bioengineering, National University of Singapore, 9 Engineering Drive 1, Singapore 117576, Singapore c Faculty of Dentistry, National University of Singapore, 9 Engineering Drive 1, Singapore 117576, Singapore

Received 30 September 2002; accepted 27 January 2003

Abstract A new technique based on tube shrinkage is proposed for the fabrication of composite archwires. Compared with a traditional pultrusion method, this new technique can avoid any ber damage during the fabrication and can provide the archwire with a required curvature in its nal clinical usage. The present paper focuses on the technique development and mechanical design and validation in terms of constituent materials by using a micromechanics bridging model. Prototype archwire has been fabricated using berglass and an epoxy matrix, with a wire diameter of 0.5 mm and a 45% ber volume fraction. Tensile and three-point bending tests have shown that the mechanical performance of the prototype composite archwire is comparable to that of a clinical NiTi archwire. Another purpose of the present paper is to provide an efcient procedure for a critical design of composite archwires. For this to be possible, the ultimate load especially exural load carrying ability of the composite archwire must be assessed from the knowledge of its constituent properties. However, difculty exists in doing this, which comes from the fact that the failure of the utmost lament of the composite archwire subjected to initially the maximum bending stress does not imply its ultimate failure. Additional higher loads can still be applied and a progressive failure process is generated. In this paper, the circular archwire was discretized into a number of parallel laminae along its axis direction, and the bridging micromechanics model combined with the classical lamination theory has been applied to understand the progressive failure process with reasonable accuracy. Only the constituent ber and matrix properties are required for this understanding. Nevertheless, the ultimate bending strength cannot be obtained only based on a stress failure criterion. This is because neither the rst-ply nor the last-ply failure corresponds to the ultimate failure. An additional critical deection (curvature) condition must be employed also. By using both the stress failure and the critical deection conditions, the predicted load-deection up to the ultimate failure agrees well with the measured data. Thereafter, different mechanical performances of composite archwires can be tailored before fabrication by choosing suitable constituent materials, their contents, and the archwire diameters. Several design examples have been shown in the paper. r 2003 Published by Elsevier Science Ltd.
Keywords: Biocomposite; Archwire; Strength design; Ultimate failure; Critical deection; Sample fabrication

1. Introduction Orthodontics is a branch of dentistry, dealing with the growth, guidance, correction, and maintenance of the dentofacial complex. The orthodontic practice involves the application of corrective appliances, commonly called braces, to move teeth. Main components of the brace (a xed appliance) are brackets and archwires, as shown in Fig. 1. Patients receiving orthodontic treat*Corresponding author. E-mail address: huangzm@mail.tongji.edu.cn (Z.-M. Huang). 0142-9612/03/$ - see front matter r 2003 Published by Elsevier Science Ltd. doi:10.1016/S0142-9612(03)00093-0

ment are most concerned with their esthetic appearance. Current materials in orthodontic applications are mainly metals, such as Stainless Steel, CoCr alloy and titanium alloy, which have color incomparable with teeth. In order to improve the appearance of xed orthodontic appliances, and thus to increase patient acceptance, several aesthetic materials have been used as an alternative to metals in bracket manufacture (Fig. 2). A further improvement would require the use of an aesthetic archwire [1]. Archwires, ideally, are designed to move teeth with light continuous forces. They are placed through the

2942

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

Fig. 1. Fixed orthodontic appliances (braces).

Fig. 3. A schematic pultrusion process for fabrication of composite archwires.

Fig. 2. Comparison between metal and aesthetic brackets.

brackets and retained in position using ligatures and elastic modules (Fig. 1). The archwire should behave elastically over the period of usage, ranging from weeks to months. The duration of use varies with the stage of treatment, namely, initial, intermediate and nal stages of the treatment, and no single archwire is best for all the stages. Springback, stiffness, formability, resilience modulus, biocompatibility and low friction are desirable characteristics of an archwire for optimum mechanical performance during the treatment [2]. It has been recognized that an optimal archwire can be developed using composite technology made from continuous bers and polymer matrix [1,37]. This is because the polymer matrix material, which is available from numerous candidates, can offer the aesthetic feature of the composite archwire. On the other hand, the various mechanical properties of the wires at different stages can be achieved by choosing suitable ber material, ber content, and ber arrangement pattern. For example, by changing the ber diameter and the number of laments, stiffness properties of the wires can be specically tailored to allow good engagement between the wire and the bracket slot from the initial to the nal stage [8]. Most composite archwires reported in the literature have been fabricated by virtue of pultrusion methods [38]. A sketch of such a pultrusion process is shown in Fig. 3. The continuous ber yarn, such as transparent

bio-berglass or ceramic ber bundle, is impregnated in a resin path, which is later drawn through a curing die. Pultrusions with small, clinically relevant round or rectangular cross-sections potentially can be prepared as continuous lengths of wires and subsequently formed into individualized appliances [9]. However, it has been recognized that two drawbacks can be involved with a pultrusion process. The rst is that the ber laments may be readily damaged during the pultrusion. As the composite archwire is a primary load-carrying element, its ber volume fraction has to be made relatively high. The extra polymer resin contained in the impregnated ber bundle must be squeezed out before the bundle can be introduced into the die channel, which has a diameter of about 0.5 mm for a round wire. Thus, the ber damages can occur at the entrance of the die. The second drawback is that the composite archwire thus fabricated is a straight cylinder, which is generally not preferred in a clinical application. In order to eliminate both of the aforementioned shortcomings involved with a pultrusion, we have developed a new method for the fabrication of a composite archwire. It is essentially based on tubeshrinkage. Instead of using a curing die, the resinimpregnated yarn is introduced into a plastic tube, which is heat-shrinkable. As the inner diameter of the tube is larger than that of the yarn, the ber damages can be controlled to minimum. Once heated, the tube shrinks and pushes the extra resin out of the entrance. Before the resin is completely cured at an elevated temperature such as in an oven, the tube containing the resin-impregnated ber bundle is put into a mould with a desired curvature as well as cross-sectional (round or rectangular) shape. A resulting archwire prototype is shown in Fig. 4, which is a curved form more appropriate for the clinical application. Photo micrograph of the newly developed archwire prototype samples indicated that the ber lament arrangements were uniform on the cross-section and there were negligibly small voids as well as other defects. Fiber content of the archwire was determined through a combustion (resin burn out) method and a 45% ber volume fraction was realized. These results show that

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

2943

are then made and are reported in the paper. These diagrams can be helpful for the development of composite archwires used at different stages.

2. Experimental detail 2.1. Materials The prototype composite archwire was made of a unidirectional (UD) ber reinforced composite cylinder. The reinforcement used was a bundle of several E-glass ber yarns, each containing 200 ber laments (the lament diameter=9 mm, Unitica Glass Fiber, Japan). With this limited number of laments in a ber yarn, the diameter and the ber content of the resulting composite archwire can be tailored by choosing suitable number of yarns. A mixture consisting of 68 wt% of an epoxy resin, R50, and 32 wt% of a hardener, H64, provided by Chemicrete Enterprise Pte Ltd. (Singapore) was used as the matrix material in the present study. It should be noted that although epoxy resins have not been acknowledged as fully biocompatible they are much cheaper than more clinical such as dental resins. On the other hand, mechanical functions of the two types of resins (epoxy and dental resins) in the resulting composite wires should not have signicant difference, as the mechanical properties of the composite wires are essentially controlled by the ber properties, ber content, and the ber reinforcement pattern. Thus, in our rst stage of technical development, the use of the epoxy resin is reasonable. The glass bers are biocompatible, and can be used in later clinical development. 2.2. Fabrication The new technique, which has been used to directly fabricate curved archwires, is based on tube-shrinkage. For the fabrication of a prototype composite archwire, a hand-processing method (Fig. 5) was used, and is briey described as follows. A number of resin-impregnated berglass yarns (7 yarns in the present study) were gathered together, and were introduced into a plastic (polyolen) tube, which is heat shrinkable (Fig. 5(a)). As the inner diameter (1 mm inner and 1.5 mm outer diameters) of the tube was larger than that of the gathered yarn bundle, ber damages could be controlled to minimum at this stage. In the next step, a heating element (an electronic soldering iron) was applied to the tube from one end (at the top grip) to the other (at the bottom grip), as shown in Fig. 5(b). Meanwhile, the heated tube shrank, pushing the extra resin out of the bottom end of the tube. Then, the shrunk tube with the resin-impregnated ber yarn bundle inside was loaded into the female part of a mould (Fig. 5(c)), which has the required curvature of the archwire, before

Fig. 4. Prototype of a curved composite archwire.

the tube-shrinkage technique is feasible for the fabrication of composite archwires. Experimental characterization for the prototype samples of a specic size (i.e. diameter) and ber content has been carried out through tensile and three-point bending tests. The test data have been compared with those of a clinically used NiTi (Reexs wire; TP Orthodontics, Inc.) metal archwire. It has been found that the mechanical performance of the prototype composite archwire is comparable to that of the metal wire. Another purpose of the present paper is to provide a useful design procedure for composite archwires. As the archwires will be used as primary load-carrying elements, their strength (tensile and especially bending strength) properties must be targeted by the design. A rational estimation on the stiffness and strength of the resulting composite archwire in terms of its constituent materials and contents and the wire diameter is essential for the efcient design. This is accomplished in the paper by making use of a bridging micromechanics model [10] and the classical lamination theory. The circular wire cylinder is imaginatively separated along the axis direction into a number of lamina layers, all of which have the same mechanical properties but different widths. When the wire is subjected to a lateral (bending) load, each of the discretized layers carries a different load share, whose determination is based on the classical lamination theory. The stress resultants of all the lamina layers must be balanced with the externally applied load (moment) on the wire cross-section. The analysis for each individual lamina is performed using the bridging model, from which the internal stresses in the constituent ber and matrix materials are expressed in terms of the load shared by the lamina. In this way, the mechanical response of the lamina and further of the whole archwire is understood from the knowledge of the constituent responses. The efciency of the theoretical analysis is veried through experiments. The estimated stiffness and strength properties of the composite archwire with some specic ber volume fraction and wire diameter agree favorably with the measured data. Design diagrams for the archwire stiffness and strength versus the ber volume fractions and the wire diameters

2944

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

Fig. 6. Optical cross-section micrograph of a sample composite archwire (100 magnication) with a ber volume fraction of 45%.

Fig. 5. Schematic show of fabrication for composite archwire by tube shrinkage.

being put into an oven (Fig. 5(d)) of 100 C for faster curing. It took about 1 h for the resin-impregnated yarn bundle to be completely cured. Finally, the tube was carefully removed and the composite archwire resulted. With this kind of fabrication process, the aforementioned two drawbacks related with the pultrusion method (Fig. 3) can be overcome. The fabricated prototype composite archwire had a diameter of 0.5 mm. As in subsequent theoretical simulation and design the in situ constituent properties are important, at pure matrix panels were also fabricated through resin casting and cured at the 100 C oven. The panels had a nominal thickness of 4.5 mm. They were then cut into uniaxially tensile, uniaxially compressive, and four-point bending test specimens according to relevant ASTM standards, respectively, using a water-cooled diamond saw. 2.3. Characterization In order to investigate the quality of the developed composite archwire, its cross-section was examined. The composite wire segment was embedded in a Co-cast resin and polished. The polished cross-section was observed under an optical microscope (Olympus BX51) with 100 times magnication. A typical photo micrograph of the archwire cross-section is shown in Fig. 6. From the gure, we can clearly see that the glass ber laments were uniform on the whole cross-section, and there were only negligibly small voids both from amount and from size viewpoints. No other signicant defect existed. The ber volume fraction of the sample composite archwire was measured through a combustion method, and an averaged value of about 0.45 has been found.

Fig. 7. Three-point-bending test of a composite archwire.

Fig. 8. Three-point bending results of a composite wire (0.5 mm diameter) and a NiTi wire (0.406 mm diameter).

Tensile and three-point bending (Fig. 7) tests were carried out for the sample archwire (remember: the wire diameter was 0.5 mm and the ber volume fraction was 45%) developed above. Currently, there is no specic standard for the characterization of an archwire. However, the span size of the three-point bending test was chosen in accordance with the distance of two adjacent brackets xed on the teeth (Fig. 1). This size is usually 14 mm [6,7], and is in the range specied by ASTM D 790 standard, with respect to the wire diameter. Typical load-deection curve up to failure is shown in Fig. 8.

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

2945

As aforementioned, a different stage orthodontic treatment generally uses a different archwire. Our present target is at the initial stage (taking about 34 months), and thus a clinically used NiTi archwire (Reexs wire, TP Orthodontic, Inc.) was also tested. The Reex wires are highly exible for the alignment of teeth, which is employed at the initial stage treatment (lasting for 34 months). They are also less expensive than other wires, and hence are widely used in practice. The metal wire test results are also plotted in Fig. 8 for comparison. Although the composite archwire has a slightly larger diameter (0.5 mm) than that (0.016 in, i.e. 0.406 mm) of the NiTi wire, the bending behavior of the former is highly superior to that of the latter. Testing data for the archwires of these two materials are summarized in Table 1. It is seen that the present composite archwire is comparable in stiffness and strength to those of the NiTi archwire. Further increase in the stiffness and strength of the composite archwire can be expected with the increase of its ber volume fraction, or the number of yarns contained in the tube. 2.4. Constituent properties For subsequent simulation and design purpose, the constituent ber and matrix properties must be specied. In the present case, both the E-glass ber and the epoxy matrix can be considered as isotropic materials. Moreover, the ber is regarded as linearly elastic until rupture. In general, it is difcult to measure directly the mechanical properties of a ber material. From the material data sheet provided by the supplier, the ber used has some comparable mechanical parameters to those of the Silenka E-glass 1200tex given in Ref. [11]. Thus, the elastic properties of the ber were taken from Ref. [11] and are listed in Table 2. These parameters are considered to be the same until rupture at both tension and compression. The ber tensile and compressive

strengths, however, were retrieved from the uniaxial tensile and compressive strengths of a UD composite provided in Ref. [11], using the formulae given in Ref. [12]. The back calculated ber strengths are summarized in Table 2. It is noted that compared with the data (2150 MPa) provided in Ref. [11], only a slight amendment (being 2100 MPa, see Table 2) has been made for the ber tensile strength. In contrast with the ber properties, the monolithic matrix properties are easily measurable. As the composite archwire is mainly subjected to exural loads, the bending behavior of the pure matrix must be understood. Four-point bending tests were carried out. In order to differentiate the mechanical properties of the pure matrix under tension from those under compression, strain gauges have also been used to record the strains at the top and bottom surfaces of the beam, and only a marginal difference between the tensile and the compressive responses has been recognized. The compressive stressstrain curve of the epoxy matrix is stiffer than the tensile stressstrain curve. Four linear segments were used to approximately represent the stressstrain (or load-deection) responses of the matrix material. Based on these representations, the matrix hardening modulus at any load level was found to be
m m m m Em ET i ; when sm Y i1 pse psY i with sY 0 0; m m Em ET 4 ; when sm e XsY 4 ;

where the tangential moduli and the critical stresses corresponding to tension and compression are summarized in Table 3. The remaining question is how to dene the matrix tensile and compressive strengths under bending. The bending test only determined one of these two strengths. To resolve the problem, uniaxial tensile (Fig. 9) and compressive tests have also been performed for the pure matrix specimens. It was found that the matrix has a uniaxial tensile strength of 42.6 MPa and a uniaxial compressive strength of 63 MPa. In other words, the uniaxial compressive strength of the matrix

Table 1 Measured properties of archwires Materials GF/epoxy NiTi

a

Diameter, d (mm) 0.5(0.005)a 0.36

Fiber volume fraction, Vf 0.45(0.001)

Tensile modulus, E (GPa) 31.2(1.8)

Tensile strength, su (MPa) 890(45.3)

Bending modulus, Eb (GPa) 31.8(2.1) 38.9

Bending strength, sb (MPa) 881.5(69.8) 804

Standard deviation.

Table 2 Mechanical properties of E-glass bers [9]

f E11 (GPa)

nf 12 0.2

f E22 (GPa)

f G12 (GPa)

nf 23 0.2

sf u (MPa) 2100

sf u;c (MPa) 1320

74

74

30.8

2946

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

Table 3 Mechanical properties of R50 epoxy matrix under bending (nm 0:414) i Tensile side properties (strength=67.8 MPa) 1 sm Y i (MPa) m i (GPa) ET 28.8 2.98 2 48.9 2.48 3 63.4 1.45 4 67.8 0.56 Compressive side properties (strength=87.8 MPa) 1 35.8 3.31 2 52.9 2.73 3 68.4 1.72 4 87.8 0.66

m m m m m m m i ; when sm Note: Em ET Y i1 pse psY ; with sY 0 0; Em ET 4 ; when se XsY 4 :

bi
i =1

ti ai
i=n

Fig. 9. Uniaxial tensile stressstrain response of R50/H64 pure matrix material.

Z
Fig. 10. A composite cylinder consisting of multilayers of laminae.

material is about 20 MPa higher than its uniaxial tensile strength. Thus, the bending strength of the matrix, 67.8 MPa, measured from the four-point bending test must be the matrix tensile strength under bending. The matrix compressive strength under bending was simply set to: 67.8+20=87.8 MPa, as indicated in Table 3. The Poissons ratio of the matrix, 0.414, was obtained from measured longitudinal and transverse strains under the uniaxial tension.

3. Theoretical background The second purpose of this paper is to provide a useful tool for a preliminary mechanical design of the composite archwire in terms of their constituent material properties and contents. Both the stiffness (modulus) and the ultimate strength especially ultimate bending strength should be targeted in the design. For this purpose, an efcient estimation of those mechanical properties is necessary. In the present paper, we apply the bridging micromechanics model combined with the classical lamination theory to do the estimation. 3.1. Discretization Essentially, the archwire under consideration is a circular composite cylinder, consisting of continuous bers arranged in the same direction and surrounding

polymer matrix. When the wire is subjected to an axial load, it can be regarded as a UD composite the analysis of which is relatively easy [10,12]. However, when the wire is subjected to a lateral (bending) load, the problem becomes much more complicated. In such a case, the stress distribution on the cylinder cross-section is not uniform. The outmost lament subjected to the maximum bending stress fails the rst. The complexity lies in the fact that the failure of this outmost lament generally does not correspond to the ultimate failure. Additional higher load can still be applied to the composite archwire. Thus, a progressive failure process, somewhat similar to that generated in a laminate, will result. This has to be understood before the ultimate bending strength of the composite archwire can be obtained. In order to track the progressive failure process in the UD composite cylinder, we can imaginatively separate the wire into a number of parallel lamina layers along the axial direction. A cross-sectional discretization is shown in Fig. 10. Take a global coordinate system, (X, Y, Z), where X is along the wire axis and XZ is the plane on which the bending load is applied. Suppose that the cross-section of the circular wire is discretized into N 2n layers of an equal thickness, t, which is given by (see Fig. 10) t d =N d =2n;

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

2947

where d is the wire diameter. Referring to Fig. 10, the width of the ith layer, bi ; is determined from q q d 2 4n it 0:5t2 ; i 1; :::; n: bi d 2 4a2 i 2 Now, the composite archwire can be regarded as a laminated composite subjected to an external (in-plane or out of plane) load. Note that in the present case, the longitudinal directions of all the laminae coincide with the global X direction, whereas the other two transverse directions of the laminae can be chosen along the Y and Z directions, respectively. Thus, the global coordinate system is coincident with the lamina local systems. 3.2. Global analysis After the discretization, the classical lamination theory is applicable to the overall analysis. Corresponding to the present load condition, only the in-plane stress and strain increments, fdsg fdsXX ; dsYY ; dsXY gT and fdeg fdeXX ; deYY ; 2deXY gT ; are retained (refer to Fig. 11). The averaged stress increments on the kth lamina can be determined from [10] fdsgk Cij k fdegk ; where Cij k S k 1 ; and fdegk fde0 XX Zk Zk 1 0 dkXX ; de0 YY 2 Zk Zk1 0 dkYY ; 2de0 XY 2 T 0 Zk Zk1 dkXY g 3 4a

equation [10] 9 2 8 > QI QI QI dNXX > > > 11 12 13 > > > 6 I > I I > > > > 6 d N Q Q Q YY > > 22 23 > > 6 12 > > < dN = 6 QI QI QI XY 6 13 23 33 6 II II > > 6 Q11 QII d M Q XX > > 12 13 > > 6 > > > > 6 II II II > > d M Q Q Q > 4 12 > YY > 22 23 > > > : dM ; II II QII Q Q XY 13 23 33 9 8 0 > > d e > XX > > > > > 0 > > > d e > > YY > > > > = < 2de0 > XY ; > > dk0 > XX > > > > > > > 0 > > > dkYY > > > > ; : 2dk0 >
XY

QII 11 QII 12 QII 13 QIII 11 QIII 12 QIII 13

QII 12 QII 22 QII 23 QIII 12 QIII 22 QIII 23

3 QII 13 7 QII 23 7 7 7 QII 33 7 7 7 QIII 13 7 III 7 Q23 5 QIII 33

QI ij

N X k1

bk Cij k Zk Zk1 ;

QII ij

N 2 1X 2 bk Cij k Zk Zk 1 ; 2 k1 N 3 1X 3 bk Cij k Zk Zk 1 ; 3 k1

QIII ij

4b

where Zk and Zk1 are the Z coordinates of the top and the bottom surfaces of the kth lamina. Sk is the lamina instantaneous compliance matrix in its local 0 system, which will be given subsequently. de0 XX ; deYY ; 0 0 0 0 and deXY and dkXX ; dkYY ; and dkXY are the laminate middle plane strain and curvature increments. These middle plane quantities are solved from the following

In Eq. (5), dNXX ; dNYY ; and dNXY and dMXX ; dMYY ; and dMXY are, respectively, the applied total in-plane force and moment increments on the laminate crosssection. When the archwire is subjected to a uniform uniaxial tension, dNXX a0 and all the other load components are zero. Under a simple bending condition, we have dNXX dNYY dNXY dMYY dMXY 0: Only the bending moment increment, dMXX ; exists. For a three-point bending and if the cross-section under consideration is near to the central point, for instance, we have dMXX dPL=4; where P is the load applied at the central point and L is the beam span. The middle plane deection increment, dw0 ; can be integrated from the equation q2 dw0 dk0 XX ; qX 2 7

11 12
x2 d12

d11 x2

22

d22

x1

x1

Fig. 11. Analysis of a lamina layer taken from the cylinder.

together with proper boundary conditions. In the case of three-point bending, dw0 0 dw0 L =0. The total deection is updated from w0 w0 dw0 : Apparently, different layers in the laminate carry different load shares. With the increase of the external load, some layers fail rst before the others. Once some k0 th lamina layer fails, the corresponding overall applied load on the laminate is dened as a progressive failure strength (i.e. the k0th ply failure strength). The failed lamina can hardly sustain any additional load, in general. The additional external load must be shared by the remaining un-failed laminae. Thus, the post-failure

2948

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

analysis is still based on Eq. (5), but with reduced overall stiffness elements given by [10] QI ij QII ij
N X k1; kefk0 g N 2 1 X G 2 bk Cij k Z k Zk 1 ; 2 k1; kefk g
0

G bk Cij k Zk Zk1 ;

QIII ij

1 3 k1;

N X kefk0 g

3 G 3 bk Cij k Z k Zk 1 :

8
Fig. 12. Schematic diagram to show the bridging model for a UD lamina.

In the above, fk0 g represents all those laminae which have already failed and the symbol e stands for not belonging to. Continued in this way, the ultimate failure strength of the laminate can be determined incrementally. For example, the ultimate tensile strength corresponds to the load level at which all of the lamina layers fail. Under a lateral load condition, however, the ultimate bending failure of a laminate generally occurs before its last ply failure [13]. Namely, the ultimate bending strength of the laminate generally corresponds to an intermediate progressive failure strength. In order to gure out this intermediate layer, the failure of which corresponds to the laminate ultimate failure, we need to provide an additional controlling parameter (in addition to the stress failure criterion), i.e. the critical deection of the laminate [14]. This critical deection has to be determined from the measured load-deection curve on which the maximum load is attained. 3.3. Local analysis The local analysis deals with an individual UD lamina, with three purposes. First, we need to provide the lamina instantaneous compliance matrix as required in Eq. (4a). Second, the internal stresses in the constituent ber, and third, in the matrix materials of the lamina must be identied, because without the knowledge of these internal stress states the lamina instantaneous compliance matrix cannot be dened. Furthermore, having known those internal stresses, the lamina ultimate load carrying ability (failure status) can be assessed by checking the ultimate strengths of the constituent materials. Thus, a pre-fabrication design can be made for the composite only based on the information of the constituent properties. All these three purposes can be achieved by using the bridging micromechanics model, which is briey summarized below (see Fig. 12). More details can refer to, e.g., Ref. [10]. 3.3.1. Lamina instantaneous compliance matrix Using the constituent compliance matrices and volume fractions, the instantaneous compliance of the

UD lamina is given by [10] SL Vf Sf Vm Sm AB; 9

where Vf and Vm are the volume fractions of the ber and matrix materials in the composite, respectively. S f and S m are, respectively, the instantaneous compliances of the ber and matrix materials, whose components are given in Appendix A. A is a bridging matrix and B Vf I Vm A1 ; which have the following expressions 2 3 2 3 b11 b12 b13 a11 a12 a13 6 7 6 7 :10 A 4 0 a22 a23 5 and B 4 0 b22 b23 5 0 0 a33 0 0 b33 The elements aij and bij depend on the constituent properties and hence are implicitly dependent on the load level, whose explicit expressions are given in Appendix B. 3.3.2. Constituent internal stresses Suppose that arbitrary external stress increments, fdsg fds11 ; ds22 ; ds12 gT ; are applied to the UD lamina (Fig. 11). The internal stress increments generated in the ber and matrix materials are calculated from [10] 9 8 f 9 8 > > = = < ds11 > < ds11 > B and 11a d s dsf 22 22 > > > ; ; : f > : ds12 ds12 9 8 m9 8 > > = = < ds11 > < ds11 > A B ; dsm d s 22 22 > > > ; ; : m> : ds12 ds12

11b

where A and B are given by Eq. (10). It is noted that in the present case, the local and the global coordinate systems coincide, i.e., 1 X and 2 Y : The stress increments calculated from Eq. (3) can be directly substituted into the right-hand sides of Eqs. (11a) and (11b).

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

2949

The total stresses in the ber and matrix are updated through fsf g fsf g fdsf g and fsm g fsm g fdsm g: 12 Although the present composite archwire was fabricated at an elevated temperature (100 C), not much large variation in temperatures from 100 C to 25 C (room temperature) has been seen. For such a moderately small temperature variation, the inuence of initial thermal residual stresses is insignicant. For simplicity, let us assume that the initial internal stresses in the ber and the matrix, fsf g and fsm g; are both zero. 3.3.3. Failure criterion As the composite consists of only two constituent materials, i.e. the ber and the matrix (Fig. 12), its failure can be considered to occur as long as any of the constituents fails. To detect the constituent failure, one of the most successful and simplest criteria is the maximum normal stress criterion of homogeneous materials. According to this criterion, the ber or the matrix failure is attained if either of the following conditions, q f sf 1 f 11 s22 f 2 f sf 13a 11 s22 4s12 Xsu ; 2 2 q f sf 1 f 11 s22 f 2 f 2 sf 13b 11 s22 4s12 p su;c ; 2 2 q m sm 1 m 11 s22 m 2 f sm 13c 11 s22 4s12 Xsu ; 2 2 q m sm 1 m 11 s22 m 2 m 2 sm 13d 11 s22 4s12 p su;c ; 2 2 f m m is met, where sf u ; su;c and su ; su;c are the ultimate tensile and compressive strengths of the ber and the matrix materials, respectively, which can be obtained, e.g. through uniaxial tension and compression tests. It is f noted that sf u and su;c are the quantities along the ber axis direction.

can be done through convergence analysis for the discretization based predictions. The predicted load (maximum moment)deection curves by using different numbers of discretized layers are plotted in Fig. 13. In the predictions, the wire diameter was assumed to be 0.5 mm and a 45% ber volume fraction was used. From this gure, it is seen that the predictions for the initial parts of the load-deection curve by using differently discretized layers were essentially the same. Discrepancies among them existed only for the later parts of the curve. It seems that the prediction with 16-layers of discretization was close to those with even more layers of discretizations. Thus, the following results are all based on the 16-layers discretization for the wire crosssections. The third quantity that must be specied before designing a composite archwire is the critical deection parameter. As aforementioned, this critical parameter is necessary in order that the maximum bending loadcarrying ability of the composite archwire can be determined. However, the critical deection of the composite archwire can be measured only after a wire specimen has been fabricated. To resolve this problem, let us consider the composite archwire which has been

Fig. 13. Comparison between the predicted results of different lamina layers used in discretizing the cross-section of an archwire.

4. Designing exercise In order to achieve a mechanical design on a composite material, its constituent properties must be rst provided. In the present case, the prototype of the new composite archwire was fabricated using E-glass bers and an epoxy matrix, the properties of which were determined in Section 2 and were given in Tables 2 and 3. Based on those properties and by using the theory described in the previous section, we can easily obtain the mechanical properties of composite archwires with different wire diameters or ber contents. Before applying the theory to the composite archwires, however, a suitable number of lamina layers should be chosen to discretize their cross-sections. This

Fig. 14. Measured and predicted load-deections of a composite archwire.

2950

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

Fig. 15. Load-deection curves of candidate composite archwires with different ber volume fractions.

Fig. 16. Load-deection curves of candidate composite archwires with different diameters.

fabricated and tested (see Section 2). The wire, with the diameter of 0.5 mm and the 45% ber volume fraction, had a typical load-deection curve shown in Fig. 14. The predicted load-deection up to the sixth-ply failure is also shown in the gure. From the gure, it is apparent that the failure load corresponding to the fth-ply failure should be regarded as the maximum load sustainable by the composite archwire under consideration. This is because the predicted deection at the fthply failure is in the nearest close to but greater than the critical deection of the measured curve. This gure also shows that the theoretical modeling as described in the previous section is reasonably accurate. As the same constituent materials and the same number of discretization layers are used, it is reasonable to assume that the ultimate bending failure of a composite archwire occurs at its fth-ply failure. Two designing cases are considered in the present paper. One is the case in which the ber volume

fractions keep unchanged but the diameters of the candidate composite archwires are different. Another is that the candidate archwires have the same diameters but with different ber volume fractions. Predicted loadmiddle span deection curves of these candidate composite archwires up to their fth-ply failure are graphed in Figs. 15 and 16, respectively. Comparing these two gures with each other, it is seen that the inuence of the wire diameter on the resulting bending load-deection is much more signicant than the inuence of the ber volume fraction of the archwire. Table 4 summarizes some designed important mechanical properties of the composite archwires with ber volume fractions varied from 35% to 55%. It must be noted that when designing the tensile strength of the composite archwire, the matrix behavior under a uniaxial tension should be employed. Namely, the matrix tensile strength is 42.6 MPa rather than 67.8 MPa, and the matrix hardening moduli should be

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953 Table 4 Some theoretically designed mechanical parameters of candidate composite archwires (with a diameter of 0.5 mm) Vf (%) Tensile modulus, E (GPa) Bending modulus, Eb (GPa) Tensile strength, su (MPa) Bending rst-ply failure strength, sby (MPa) Bending fth-ply failure strength, sb (MPa) 35 27.8 28.5 782 509 740 40 31.4 32.1 884 575 834 45 34.9 35.8 986 639 929 50 38.5 39.4 1090 704 1023

2951

55 42.0 43.0 1192 773 1120

Fig. 17. Variation of bending modulus and ultimate bending strength of GF/R50 composite archwire versus ber volume fraction.

those listed in Fig. 9. If, otherwise, the matrix hardening moduli given in Table 3 together with the tensile strength of 42.6 MPa were used, the predicted uniaxial tensile strengths of the archwires would be remarkably lower than those reported in Table 4. The reason is that the matrix bending modulus is higher than its tensile counterpart, as can be seen from the comparison between Fig. 9 and Table 3. The higher modulus used would result in more load shared by the matrix when the same amount of overall load is applied to the composite archwire. Thus, the matrix would fail the rst under the uniaxial tension, giving a lower overall load level sustainable by the archwire. Finally, Fig. 17 plots the variation of the bending stiffnesses and the ultimate bending strengths versus ber volume fractions. The gure shows that these two bending properties of the composite archwires are linearly proportional to the ber volume fractions, which they contain.

ber content) were 31.8 GPa and 881.5 MPa.It should be noted that in our present study the E-glass bers (whose modulus is 74 GPa) were used which is lower in mechanical performance than the S-glass bers (whose modulus is 87 GPa [3]) commonly used in the literature. The present study is targeted at the initial stage of an orthodontic treatment. Like many other composite archwires, the present wire is less exible compared with clinical metal archwires, as can be seen from Fig. 8. In practice, some patients teeth may be highly irregular so that some bracket may have to be xed in 45 mm lower (or higher) than an adjacent bracket (Fig. 2). In such case, the brittle archwire may not be applicable. Future improvement will be in the use of more ductile constituent materials. For instance, nanobers have been acknowledged to possess higher stiffness and strength and at the same time more ductile deformation ability than micron bers [15]. In addition, a ductile matrix can also modify the deformation/strain behavior of the resulting composite archwire in a big range of extent. As numerous constituent materials may be available, the designing procedure described in the paper can help choosing suitable ber and matrix materials without much experiment cost and time consuming.

6. Conclusion Composite archwires can be easily fabricated into a curved form similar to that of a metal archwire for clinical usage. Such a fabrication method has been described in the present paper. A prototype archwire thus developed displayed comparable mechanical performance to that of a clinical NiTi archwire. A theoretical modeling procedure to obtain the stiffness and strength, especially the ultimate bending strength, of composite archwires only using their constituent ber and matrix properties has been presented in the paper. Comparison with experiments has indicated its accuracy and efciency. This will make it possible to design a composite archwire before fabrication by choosing suitable constituent materials, ber content, and the wire diameter. Several designing examples have been shown in the paper, which can benet the development of specic composite archwires.

5. Discussion The sample composite archwire was developed using an E-glass ber and an epoxy matrix. Morphology analysis (Fig. 6) indicated that the tube shrinkage technique could result in a composite archwire with satisfactory quality. The bending stiffness and strength of the sample archwire were comparable to those developed in the literature [35]. For instance, the bending modulus and ultimate bending strength of a composite archwire with a 48% ber volume fraction wire obtained by Fallis and Kusy were 34.7 GPa and 1180 MPa whereas those of the present wire (with 45%

2952

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953

Appendix A The ber material can be considered as linearly elastic until rupture, whose instantaneous compliance matrix can be dened using Hookes law, whereas the instantaneous compliance of the matrix is specied using PrandtlReuss theory. Thus, the constituent instantaneous compliance matrices are given as [10] 2 f f f 3 S12 S13 S11 6 f f f 7 Sij 4 S22 S23 5 2 6 4 2 6 m Sij 4 ( symmetric
f 1=E11 f S33 f nf 12 =E11 f 1=E22

Appendix B The elements of the matrices A and B are summarized below [10]
f ; a11 Em =E11 f ; a22 0:51 Em =E22 f ; a33 0:5 1 Gm =G12 ( m E m when sm e psY ; Em m m ET when sm e > sY ; ( m 0:5E m =1 nm when sm e psY Gm m m ET =3 when sm e > sY f m f m S12 a11 a22 =S11 S11 ; a12 S12

0 0
f 1=G12

3 7 5;

symmetric
m S11 m S12 m S22 m S13

symmetric

m 7 S23 5 m S33

a13 a23

d2 b11 d1 b21 ; b11 b22 b12 b21 d1 b22 d2 b12 ; b11 b22 b12 b21

m S m e ; when sm e psY ; m Sm e S m p ; when sm e > sY ;

sm e

q 2 m 2 m m m 2 sm 11 s22 s11 s22 3s12 ; 2 nm m E 1 Em symmetric 3 0 7 7 7 0 7 7; 7 1 5 Gm

b11 Vf Vm a22 Vf Vm a33 =c; b12 Vm a12 Vf Vm a33 =c; b13 Vm a12 Vm a23 Vf Vm a22 Vm a13 =c; b22 Vf Vm a11 Vf Vm a33 =c; b23 Vm a23 Vf Vm a11 =c; b33 Vf Vm a22 Vf Vm a11 =c; c Vf Vm a11 Vf Vm a22 Vf Vm a33 ;
m a11 a33 ; d1 S13

1 6 Em 6 6 S m e 6 6 6 4

Sm p 9
m sm 2 4M T e

2 6 4

s011 s011

s022 s011 s022 s022 symmetry

2s012 s011 2s012 s022 4s012 s012

3 7 5
sij sm ij

m d2 S23 Vf Vm a11 a22 a33 m S13 Vf Vm a33 a12

m MT

m E m ET m; E m ET

m f b12 S11 S11 ; m f b22 Vf Vm a22 S12 S12 ; f f m m b21 Vm S12 S12 S22 a12 Vf Vm a11 S22 :

m f b11 S12 S12 ;

1 s0ij sij skk dij ; 3

i ; j ; k 1; 2 :

References
[1] Kusy RP. A review of contemporary archwires: their properties and characteristics. Angle Orthodontist 1997;67(3):197207. [2] Cohn ER, Dongelli P, Sassouni V, Sassouni A. A simple procedure to assess esthetic preferences for dentofacial treatment. Am J Ortho 1986;89:2237. [3] Karmaker AC, DiBenedetto AT, Goldberg AJ. Continuous ber reinforced composite materials as alternatives for metal alloys used for dental applications. J Biomater Appl 1997: 31828.

f f f ; E22 ; and G12 are the longitudinal, In the above, E11 transverse, and in-plane shear moduli of the ber; E m ; G m ; and nm are Youngs modulus, shear modulus, and Poissons ratio of the matrix; sm e is the von-Mises m effective stress of the matrix; sm and ET are the matrixs Y yield strength and hardening modulus (tangent to its stress-strain curve at the plastic region).

Z.-M. Huang et al. / Biomaterials 24 (2003) 29412953 [4] Jancar J, Dibenedetto AT, Hadziinikolau Y, Goldberg AJ, Dianselmo A. Measurement of the elastic modulus of berreinforced composites used as orthodontic wires. J Mater Sci Mater Med 1994;5:2148. [5] Fallis DW, Kusy RP. Variation in exural properties of photopultruded composite archwires: analyses of round and rectangular proles. J Mater Sci Mater Med 2000;11:68393. [6] Watari F, Yamagata S, Imai T, Nakamura S. The fabrication and properties of aesthetic FRP wires for use in orthodontics. J Mater Sci 1998;33:56614. [7] Toyoizumi H, Watari F, Imai T, Yamagata S, Kobayashi M. Fabrication of aesthetic wires with exural and torsional stiffness by photo curing method. J Jpn Soc Dental Mater Dev 1999;18(6):42940. [8] Zufall SW, Kennedy KC, Kusy RP. Frictional characteristics of composite orthodontic archwires against stainless steel and ceramic brackets in the passive and active congurations. J Mater Sci: Mater Med 1998;9:61120. [9] Goldberg AJ, Burstone CJ. The use of continuous ber reinforcement in dentistry. Dent Mater 1992;8(3):197202.

2953

[10] Huang ZM. Simulation of the mechanical properties of brous composites by the bridging micromechanics model. Composites Part A 2001;32(2):14372. [11] Soden PD, Hinton MJ, Kaddour AS. Lamina properties, lay-up congurations and loading conditions for a range of ber-reinforced composite laminates. Compos Sci Technol 1998;58:101122. [12] Huang ZM. Micromechanical strength formulae of unidirectional composites. Mater Lett 1999;40(4):1649. [13] Huang ZM, Teng XC, Ramakrishna S. Bending behavior of laminated knitted fabric reinforced beams. Adv Compos Lett 2001;10(5):2118. [14] Huang ZM, Fujihara K, Ramakrishna S. Flexural failure behavior of laminated composites reinforced with braided fabrics. AIAA J 2002;40(7):141520. [15] National Science and Technology Council Committee on Technology-Interagency Working Group on Nanoscience, Engineering and Technology (IWGN). Nanotechnology Research Directions: IWGN Workshop Report-Vision for Nanotechnology R&D in the Next Decade, September 1999. http: //itri.loyola.edu/ nano/IWGN.Research.Directions/