Analysis of mold insert fabrication for the processing of microfluidic chip.

INTRODUCTION

In the past decades, there has been an explosion of activity in the field of microfluidic chip. Research and development on transport phenomena in micro-and nano-fluidic channel networks have been a frontier in fluid dynamics. The microfluidic channels of such microdevices were mostly based on silicon, glass, or quartz and are typically fabricated by photolithographic, wet chemical etching, and cover bonding procedures.

Yu et al. (1) used different processing steps to produce various mold inserts (computer numerical control (CNC) epoxy photoresist, photolithography, and electroforming). They studied the application of microstructure from 5 to 100 [mu]m by micro forming method. The results reveal that the mold insert for electroforming process has better replication and demolding. Chou and Kraus (2) used the silicon mold insert to fabricate a nanostructure with a width of 10nm and an aspect ratio of 4. The results showed that the silicon mold insert could be used repeatedly 30 times. Madou and coworkers (3) discussed different microfabrication methods for polymer-based compact dick (CD) micro-fluidic platforms. Their prototyping fabrication methods included CNC-machining and photolithography techniques. For mass production, mold inserts were made by LIGA-like process and CNC-machining of tool steel. The replication techniques used the liquid resin molding, thin wall injection molding, and hot embossing. They have used bonding processes such as vacuum-assisted thermal bonding, adhesive tape bonding, and water-soluble polymer-assisted bonding. Yang (4) used UV-LIGA processing to produce the master for the negative photoresist and then fabricated the nickel (Ni) mold insert by electroforming. The results showed that this method could get the injection mold high precision and long life. Lin et al. (5) indicated that the silicon mold insert that was used to emboss the poly (methyl methacrylate) (PMMA) material and electroforming Ni mold insert embossed the polyvinyl chloride (PVC) material by hot embossing molding. The cycle time of hot embossing was 2 h. Becker and Heim (6) used three methods to fabricate the mold inserts (LIGA, reactive ion etching (RIE), and inductively coupled plasma (ICP)), which were 50-[mu]m in width and depth. The results showed that the demolding temperature was an important factor for surface roughness. If the demolding temperature is too high, the surface roughness is not good. If the demolding temperature is too low, the molding cycle time becomes too long. Nilsson et al. (7) used excite laser to produce the mold insert of microfluidic chip and embossing method to replicate the microhannel on the cyclo-olefin copolymer(COC) material. The results showed that the COC material was very suitable for the optical and biomedical products. Becker and Gartner (8) reviewed the polymer microfabrication methods for microfluidic chip. The research indicated that different polymer materials (Polyamide 66 (PA66), Polycarbonate (PC), Polyoxymethylen (POM), COC, and PMMA), various replication technologies (hot embossing, injection molding, and casting), and different direct techniques (laser-based technology, optical lithography in deep resists, stereo lithography, and layering technique). Mills et al. (9) presented the technology for the fabrication of three-dimensional microfluidic channels in optical transparent substrates consisting of polymers, with different properties. Their results showed that Si[O.sub.2] and S[i.sub.3][N.sub.4] based masters could be used for polymer replication techniques to reduce the amount of sticking occurring between the polymer and the master, as compared to the purely silicon-based masters. Lin and Burns (10) investigated handling in surface modified polyolefin microfluidic devices. The surface was modified from hydrophobic to hydrophilic using the UV-mediated grafting technique in their research. The capillary-driven flow, pulsed pumping, and contact angle hysteresis were indicated, and the separation of double-stranded deoxyribonucleic acid (DNA) samples was demonstrated. Chow et al. (11) used a novel technique for bonding polymer substrates (PMMA) using PDMS-interface bonding in their work. The bonding strength of microfluidic chip was about 0.015 Mpa.

Among these fabrication methods, hot embossing and injection molding are regarded as the low cost mass-production processes to replicate microfluidic chip. However, focused ion beam and modified LIGA techniques are expensive, complicated, and not easily accessible to scientists and industrialists.

In this article, an experimental study, based on the Taguchi orthogonal array design, is conducted to characterize the effect of different processing parameters on the micro-hot embossing of molded micofulidic chip, including embossing temperature, embossing pressure, embossing time, and demolding temperature. The sag height and surface roughness of molded micofulidic chip are also measured and analyzed. This article also discusses the characteristics of molded microfluidic chip using various mold inserts. This article also indicates that the energy dispersive X-ray spectrometer (EDS) and nano indentation for various mold inserts of microfluidic chip. Finally, this article shows that the result of bonding process for sealed microfluidic chip.

FABRICATION OF MOLD INSERT OF MICROFLUIDIC CHIP

The metal mold insert is electroformed from a master whose surface has been patterned with the microchannel structure (see Fig. 1). Figure 1 shows the dimensions of microfluidic chip. The channel defined by points P3 (Buffer) and P1 (Waste) provides the separation situation and that defined by P4 (Sample) and P5 (Waste) is the injection channel. At the end of each channel, there are reservoirs for waste, buffer, or sample. These reservoirs also provide access of the electrodes for high-voltage input. The microhannel has 70[micro]m width and 50 [micro]m depth. The length of separation and injection channel is all 8 mm. The effective separation distance from the intersection to detection point (P2) is 13mm. The simple "cross" geometry is designed for sample injection and separation. As a high-voltage is applied between points P4 and P5, the sample will be driven electro-kinetically and moved across the intersection of the channels. With appropriate switching off this voltage and applying another high-voltage between points P3 and P1 on the separation channel, a specific amount of sample will be injected into the separation channel. with the existence of a high electrical field, samples will be moved and separated subsequently. An optical or electrochemical detection method can be also applied at the end of separation channel to detect separated sample fragments. The steps of fabricating the microfluidic chip are described in detail.

[FIGURE 1 OMITTED]

The first step is photolithography. In conventional photolithography, the patterns on the mask are transferred onto the photoresist on top of the (100)-oriented silicon wafer. The photomask is generated using computer-aided design software (Pro/E) and is printed on a transparent film using a high-resolution laser printer (15,000 dpi). Then a 2-[micro]m thick SU-8 2050 negative photoresist is spun over the wafers at 4000 rpm followed by a 95[degrees]C soft bake for 600 s. The wafer was then exposed through a mask for 5 s. For this exposure, a UV Kar-Suss double side mask aligner is used. The aligner is equipped with an ultra-violet source operating in the 365-405-nm wave length region. The UV intensity at 365 nm is 150 mJ/c[m.sup.2]. The resist patterns are than developed using an AZ 400 k developer, diluted 1 to 4 with deionized water (DI), followed by a thorough rinse in DI. Following the definition of resist patterns, the wafers are baked at 95[degrees]C in an oven for an additional 8 min to harden the resist structure. BY hardening the resist, the feature patterns are formed.

The second step indicates that the Si mater puts into the RIE machine. This process uses C[F.sub.4] and [O.sub.2] to each Si. The third step indicates that the deep etching technology by ICP etches the silicon substrate. When the process gets the needed height by etching, the master can be used to electroform the metal mold insert. The fourth step indicates the electroforming metal mold insert from the master. After the development process, the electroforming seed layer is coated by vaporization. The silicon master puts on the hot vaporization, then uses the vaporization method to deposit the seed layer (Cr/Cu: 200 [Angstrom]/2000 [Angstrom]) on the silicon master. Finally, this article uses the electroforming process to produce the metal mold insert from the silicon master. In this article, the authors discuss various mold inserts (Ni, Ni-Co) that influence the properties of molded microfluidic chip. When this study finishes the metal mold insert, this study uses metal mold insert to fabricate microchannel of molded microfluidic chip by hot embossing process. Finally, this article uses the bonding process to fabricate the sealed microfluidic chip. The details of the process flow chart are shown in Fig. 2.

[FIGURE 2 OMITTED]

INSTRUMENTATION AND EXPERIMENTAL APPROACHES

The micro-hot embossing process is carried out with a commercial hot embossing machine (Hex01, Jenoptik Mikrotechnik GmbH, Germany). A 0.8-mm thick optical grade cycle-olefin polymer (COP, Zeonor-ZF14) film and a glass transition temperature of 136[degrees]C are used.

The micro-hot embossing process can be divided into four stages as illustrated in Fig. 3. The first step is the heating stage. The plastic sheet/metal mold insert stack is mounted on the embossing machine, and the hot plate is heated to an embossing temperature ([T.sub.1]) that is above the glass transition temperature ([T.sub.g]) of the plastic sheet. During the heating process, a low pressure is applied on the sheet to prevent the sheet from creasing. The second step is the constant temperature and embossing stage. A high embossing pressure ([P.sub.1]) is applied when the temperature of the mold reaches the embossing temperature. The softened polymer is filled into the microfluidic chip cavity. The third step is the cooling and packing stage. After the embossing time period ([t.sub.1] ~ [t.sub.2]) has ended, the polymer is cooled down to below [T.sub.g] while maintaining the pressure ([P.sub.1]) to prevent uncontrolled shrinkage and distortion. The final step is the demolding stage. Once the demolding temperature ([T.sub.2]) has been reached, the molded microfluidic chip is removed from the metal mold insert.

[FIGURE 3 OMITTED]

It is very important to understand the effects of different processing parameters on the replication quality and surface roughness of micro-hot embossed micrifluidic chip. Four different processing parameters including embossing temperature, embossing pressure, embossing time, and demolding temperature are selected as the factors for evaluation. Table 2 lists the processing parameters and parameter levels selected in the main experiment. The embossing temperature are 170, 180, and 190[degrees]C, and the embossing pressures are 1.875, 2.5, and 3.125 MPa. The embossing times are 60, 90, and 120 s, and the demolding temperature are set at 40, 60, and 80[degrees]C.

TABLE 2. Tabulation of the sag heights for molded microfluidic chip by
Ni mold insert in the main experiment.

                            Samples

         Sample   Sample   Sample   Sample   Sample
            1        2        3        4        5
           (y1)     (y2)     (y3)     (y4)     (y5)
Runs     ([mu]m)  ([mu]m)  ([mu]m)  ([mu]m)  ([mu]m)  Average  S/N (dB)

1         41.77    41.71    41.73    41.47    41.07   41.5500  -32.372
2         43.44    43.39    43.36    43.37    43.33   43.3780  -32.745
3         44.80    44.84    44.79    44.77    44.81   44.8020  -33.026
4         47.80    47.83    47.79    47.76    47.78   47.7920  -33.587
5         48.25    48.24    48.22    48.23    48.24   48.2360  -33.587
6         48.41    48.43    48.46    48.44    48.40   48.4280  -33.667
7         49.50    49.54    49.55    49.51    49.52   49.5240  -33.702
8         49.61    49.60    49.57    49.63    49.59   49.6000  -33.896
9         49.80    49.79    49.77    49.78    49.76   49.7800  -33.941
Optimum   49.84    49.82    49.83    49.79    49.78   47.8120  -33.947

To identify the relative significance of these four parameters, a large array of experiments with as many as [3.sub.4] runs are required. A statistics-based design of experiments, the Taguchi method (12), is used to reduce the number of experimental runs. The levels of processing conditions and the experimental array used are shown in Table 1. This research uses single-parameter method to discuss the properties of surface roughness of molded microfluidic chip for different processing parameters and various mold inserts.

TABLE 1. The L9 orthogonal array used in the main experiment.

                                    Parameters

                                                        D. De-molding
         A. Embossing      B. Embossing   C. Embossing      temp.
Runs  temp. ([degrees]C)  pressure (MPa)   time (Sec.)   ([degrees]C)

1           1 (170)          1 (1.875)        1 (60)        1 (40)
2           1 (170)          2 (2.500)        2 (90)        2 (60)
3           1 (170)          3 (3.125)        3 (120)       3 (80)
4           2 (180)          1 (1.875)        2 (90)        3 (80)
5           2 (180)          2 (2.500)        3 (120)       1 (40)
6           2 (180)          3 (3.125)        1 (60)        2 (60)
7           3 (190)          1 (1.875)        3 (120)       2 (60)
8           3 (190)          2 (2.500)        1 (60)        3 (80)
9           3 (190)          3 (3.125)        2 (90)        1 (40)

EXPERIMENTAL RESULTS AND DISCUSSION

Mold Insert

In this article, the authors use a noncontact optical interferometer Wyko NT1100 (Veeco, Woodbury, NY) to measure the surface morphology and profile for different metal, mold inserts and molded microfluidic chip. The authors also use the atomic force microscope (DIMENSION, DI-3100) to measure the surface roughness for different mold metal inserts and molded microfluidic chip. Figures 4 and 5 show the characteristics of different metal mold inserts (Ni, Ni-Co). The results show that the morphology of sample reservoir and microchannel are very good for different metal mold inserts. The microchannel is found to be of vertical shape. The results also show that the sag heights of microchannel are 50.92 [mu]m and 47.05 [mu]m, respectively, for Ni and Ni-Co mold inserts. The results also show that the surface rough nesses are 6.124 nm and 0.562 nm for Ni and Ni-Co mold inserts. The surface roughness obtained for metal mold inserts is very good using electroforming process. The results show that the Ni-Co mold insert has low roughness.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Replication

Figure 6 shows the surfaced profile of the molded microfluidic chip for Ni-Co mold insert by micro-hot embossing. the sag height of molded microfluidic chip is 46.42 [mu]m. Table 2 shows that the measured sag height of molded microfluidic chip by Ni mold insert for nine test trials. Five samples are measured for each test trial. The S/N ratios for the sag heights of the molded microfluidic chip are calculated from the main experiment; and the results are shown in Fig. 7. Based on this figure, the optimum factor levels that can statistically result in the maximum sag height for the micro-hot embossed microfludic chip are predicted to be A3/B3/C3/D3. These optimized factor levels represent a processing temperature of 190[degrees]C, an embossing pressure of 3.125 MPa, an embossing time of 120 s, and a demolding temperature of 80[degrees]C. A confirmation experiment is conducted according to the optimized factor levels. The optimum sag height of molded microfluidic chip obtained is 47.81 [mu]m. Table 3 shows the measured data of the sag height of molded microfluidic chip by Ni-Co mold insert. Teh S/N ratios for the sag heights of molded microfluidic chip are shown on Fig. 8. The optimum factor levels that can statistically result in the maximum sag height by micro-hot embossed microfluidic chip are predicted to be A3/B3/C3/D3. The values of optimized processing parameters are the same as that of Ni mold insert. The optimum sag height of molded mirofluidic chip for Ni-Co mold insert is 45.99 [mu]m.

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

TABLE 3. Tabulation of the sag heights for molded microfluidic chip
by Ni-Co mold insert in the main experiment.

                          Samples

                  Sample  Sample   Sample   Sample
         Sample     2        3        4        5
         1 (y1)    (y2)     (y3)    (y4)     (y5)
Runs     [mu]m)  ([mu]m)  ([mu]m)  ([mu]m)  ([mu]m)  Average  S/N (dB)

1        38.02    38.00    37.97    37.95    37.99   37.9860  -31.592
2        39.71    39.79    39.71    39.72    39.73   39.3872  -31.909
3        41.04    41.02    41.03    41.00    41.00   40.7564  -32.205
4        44.07    44.10    44.06    44.09    44.08   43.4696  -32.768
5        44.60    44.56    44.54    44.52    44.54   44.5520  -32.977
6        44.67    44.70    44.66    44.63    44.68   44.6444  -32.995
7        45.73    45.71    45.74    45.78    45.77   45.5336  -33.167
8        45.78    45.77    45.80    45.74    45.79   45.7692  -33.211
9        45.95    45.97    45.94    45.98    45.96   45.9252  -33.241
Optimum  46.02    45.97    45.98    46.01    46.00   45.9960  -33.254

By the previous results, this research shows that the replication of molded microfluidic chip by Ni-Co mold insert is the best, then is the Ni mold insert. From the previous results, it is show that the embossing temperature is the most important factor for replication of molded microfluidic chip, then is embossing pressure, then is embossing time, and the last one is demolding temperature. The results show that the replication of molded microfluidic chip is influenced by various mold inserts and processing parameters.

Finally, this article uses SEM (TESCAN, 5136LS) to measure the profile of molded microfluidic chip as shown in Fig. 9. The shape of microchannel and sample reservoir is very good. The shape of microchannel shows a very good vertical property.

[FIGURE 9 OMITTED]

Surface Roughness

Figure 10 shows the AFM image and surface roughness of molded microfluidic chip by different metal mold inserts. Figure 11 shows that the surface roughness of molded microfluidic chip for different processing parameters by different metal mold inserts. The results show that the surface roughness increases upon increasing different processing parameters (embossing temperature, embossing pressure, embossing time, and demolding temperature) for different metal mold inserts. Figure 11 shows that the surface roughness of molded microfluidic chip is larger for Ni mold insert and smaller for the Ni-Co mold insert.

[FIGURE 10 OMITTED]

[FIGURE 11 OMITTED]

The results show that the demolding temperature is the most important factor for surface roughness of molded microfludic chip with different processing parameters. The reason is that the higher demolding temperature causes the COP film to become soft and the microchannel may be destroyed on demolding process. Therefore, the surface roughness of molded microfluidic chip changes quickly. The embossing pressure is the second factor for surface roughness, followed by embossing temperature. The embossing time is not a very important factor for surface roughness of molded microfluidic chip.

EDS Analysis

The elemental composition of various mold inserts has been measured using EDS (OXFORD, INCA Wave). Figure 12 shows different elements for various metal mold inserts (Ni, Ni-Co) of microfluidic chip. The results show that the Ni mold insert has 96.98% Ni and other elements. The other elements include carbon and oxygen. The results also show that the Ni-Co mold insert has 72.26% Ni and 19.23% Co. It also has the other elements that such as C1, C, O, and Si.

[FIGURE 12 OMITTED]

The results show that this article can get the good metal mold inserts by electroforming method to do the micro-hot embossing process.

Hardness and Young's Modulus

This article uses the nano indenter (Hysitron) to measure the hardness and Young's modulus for different metal mold inserts. The hardness of a nano indenter H is given by

P = [alpha] (h-[h.sub.f]).sup.m] (1)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

[h.sub.c] = [h.sub.max]-[epsilon] [P.sub.max]/S (3)

H = [P.sub.max]/A (4)

where A is the contact area, [h.sub.f] is the residual depth for elastic displacement recovery, [h.sub.c] is the contact depth, [h.sub.max] is the maximum contact depth, [alpha], m, and, [epsilon] are constants. [epsilon] = 0.72 for cylindered-cone indentation head.

The Young's modulus of nano indentation [E.sub.r] is given by

1/[E.sub.r] = [1-[v.sup.2.]/E] + [1-[v.sup.2.sub.i]/[E.sub.i]] (5)

where E and v are Young's modulus and Poisson ratio for the measurement material, and [E.sub.i] and [V.sub.i] are Young's modulus and Poisson ratio for the probe, respectively.

Figure 13 shows the hardness and Young's modulus of Ni, Ni-Co mold insert for microfluidic chip. The results show that the Young's modulus of Ni-Co mold insert has the largest value, followed by Ni mold insert. The results also show that the Ni-Co mold insert has the maximum hardness. The reason is that the Ni-Co mold insert has the Co element.

[FIGURE 13 OMITTED]

This article shows that the Ni-Co mold insert has the largest hardness, and it induces the good replication and smallest surface roughness of molded microfluidic chip. The hard mold insert is suitable for molded microfluidic chip by micro-hot embossing process.

Bonding Analysis

After molding the microfluidic chip by micro-hot embossing process, the molded microfluidic chip must be bonded by cover film and then forms the sealed microfluidic chip. There are several methods for bonding process like thermal bonding anodic bonding, and low temperature adhesion. This article uses COP film for the micro channel of molded microfluidic chip with a glass transition temperature of 136 [dergee]C. cover also uses a COP film. This study uses thermal bonding on the microchannel and cover for microfluidic chip by micro-hot embossing machine. This study uses 5000 N embossing force and 30 min embossing time for the thermal bonding process. The results are shown in Fig. 14. The results show that the shape of surrounding on microchannel is very good. The bonding strength of COP film is measured by Q-Test tensile tester (MTS systems). The effects of bonding temperature and bonding pressure are discussed in this article. Figures 15 and 16 show the bonding strength for various bonding temperatures and bonding pressures. The results show that the bonding strength increases as the bonding temperature increases. The bonding strength increases rapidly when the bonding temperature reaches 126 [degrees] C The microchannel collapses at the bonding temperature larger than 135 [degrees] C in this research. At bonding temperature of 129 [degrees] C, sidewalls of the microchannel draped. To get considerable bond strength and high quality of bonded microchannel, temperature of 123-129 [degrees] C was chosen as the bonding temperature for suitable case. In this case, 126 [degrees] C is selected as the bonding temperature for the experiments shown in Fig. 16 The results also show that the bonding strength increases as the bonding pressure increases. The bonding strength increases quickly as the bonding pressure is larger than 0.9 MPa. The sidewalls of microchannel draped at bonding pressure of 0.9 MPa. So the 0.8 MPa is selected as bonding pressure for suitable case on Fig. 15. The results also show that the effect of bonding temperature is a more important factor for the bonding strength of sealed microfluidic chip. The typical bonding strength of 0.03 MPa is obtained at bonding temperature of 126 [degrees] C, bonding pressure of 0.9 MPa, and bonding time of 30 min.

[FIGURE 14 OMITTED]

[FIGURE 15 OMITTED]

[FIGURE 16 OMITTED]

The cover and molded microfluidic chip both use the COP film. The COP material is transparent: it is difficult to observe the bonding quality by human eye. In this article, color dye is pumped into the microchannel and reservoir of sealed microfluidic chip at a flow rate of 0.5 ml/min. There is no leakage phenomenon in the microchannel and reservoir of sealed microfluidic chip.

CONCLUSIONS

This article indicates that the replication and surface roughness of microfluidic chip are influenced by various mold inserts for micro-hot embossing. The results show that the replication of molded microfluidic chip is the best using Ni-Co mold insert and then Ni mild insert. the embossing temperature is a most important factor for the replication of molded microfluidic chip on different processing parameters. The embossing temperature is 190 [degrees] C, embossing pressure is 3.125 MPa, embossing time is 120 s, and demolding temperature is 80 [degrees] C as the optimal processing for molded microfluidic chip.

The results show that the surface roughness of metal mold insert is smaller of the Ni-Co mild insert and larger for the Ni mold insert. The surface roughness of molded microfluidic chip has the smallest value for Ni-Co mold insert and then for the Ni mold insert. The demolding temperature is the most important factor for the surface roughness of molded microfluidic chip on various processing parameters. The results also indicate that the hardness of Ni-Co mold insert has the biggest value, followed by Ni mold insert. The Ni-Co mold insert has the low roughness and high hardness. The bonding temperature is the most important factor for the bonding strength of sealed microfluidic chip.

REFERENCES

(1.) L. Yu, C.G. Koh, L.J. Lee, K.W. Koelling, and M.J. Madou, Polym. Eng. Sci., 42, 871 (2002).

(2.) S.Y. Chou and P.R. Kraus, Microelectron. Eng., 35, 237 (1997).

(3.) L.J. Lee, M.J. Madou, K.W. Koelling, S. Daunert, S. Lai, C.G. Koh, Y.J. Juang, Y. Lu, and L. Yu, Biomed. Microdevices, 3, 339 (2001).

(4.) S.P. Yang, Study on Micro-injection Molding Technology with Application of Microfluidics, Masters Thesis, Department of Aeronautics and Astronautics, National Cheng Kung University, Taiwan (2002).

(5.) L. Lin, Y.T. Cheng, and C.J. Chiu, Microsyst. Technol., 4, 113 (1998).

(6.) H. Becker and U. Heim, Sens. Actuators, 82, 130 (2000).

(7.) D. Nilsson, S. Balslev, and A. Kristensen, J. Micromech. Microeng., 15, 296 (2005).

(8.) H. Becker and C. Gartner, Electrophoresis, 21, 12 (2000).

(9.) C.A. Mills, E. Martinez, F. Bessueille, G. Villanueve, J. Bausells, J. Samitier, and A. Errachid, Microelectron. Eng., 78-79, 695 (2005).

(10.) R.S. Lin and M.A. Burns, J. Miromech. Microeng., 15, 2156 (2005).

(11.) W.Y. Chow, K.F. Lei, G. Shi, W.J. Li, and Q. Huang, Smart. Mater. Struct., 15, S112 (2006).

(12.) G.S. Peace, Taguchi Methods, Addison-Wesley, New York (1993).

Y.K. Shen, (1) J.D. Lin, (2) R.H. Hong (2)

(1) School of Dental Technology, Bio-Medical Nano/Micro Forming Laboratory, College of Oral Medicine, Taipei Medical University, Taipei 110, Taiwan

(2) Department of Mechanical Engineering, Nano/Micro Forming Laboratory, LungHwa University of Science and Technology, Taoyuan County 33306, Taiwan

Correspondence to: Y.K. Shen; e-mail: [e-mail]

Contract grant sponsor: National Science Council, Taiwan; contract grant number: NSC 93-2218-E-262-001.

DOI 10.1002/pen.21227

Published online in Wiley InterScience (www.interscience.wiley.com)