Laser-Based Fabrication of Carbon Nanotube–Silver Composites With Enhanced Fatigue Performance Onto a Flexible Substrate

Flexible electronics, where electronic circuits are fabricated onto flexible substrates such as polymers, have several advantageous features, such as compactness, low weight, low energy consumption, and easiness in their integration with other devices, and the relevant market is expected to grow very fast [1–6]. Flexible electronics may find good applications in sensors, displays, flexible solar cells, and other areas [1,5,6]. However, during their services or applications, flexible electronic devices often need to go through repeated deformations (such as bending) and related stresses and strains. This may cause the initiation and development of cracks in metallic components (e.g., metallic interconnects) that often exist in many flexible electronic devices [3,6,7], which could obviously affect the long-term reliability and durability of the devices. Therefore, an effective approach to enhance the fatigue performance and durability of metallic components in flexible electronics is desired.

It has been found that carbon nanotube (CNT)–metal composites could exhibit better mechanical properties than metals, such as improved strength and/or ductility [8,9]. Reference [10] reported a CNT–copper composite that has an electrical conductivity similar to copper, but has a 100 times better ampacity (i.e., the capacity of carrying current). However, the previous work in the literature has been limited on producing CNT–metal composites onto flexible substrates. Obviously, several typical processes for CNT–metal composite fabrications, such as hot pressing and plasma spraying [11,12], are difficult to use in this situation because they may damage the fragile flexible substrates and/or often lack the desired spatial resolutions. In Ref. [13], CNT–metal composites were produced onto a Kapton polyimide film through aerosol-jet printing followed by furnace sintering, while Ref. [14] reported research work on the fabrication of CNT–metal composite films onto a polyimide substrate through flash light sintering.

This paper reports a new (to the authors' best knowledge) approach of producing CNT–metal composites onto flexible substrates through laser sintering of CNTs and metal nanoparticles (NPs), which was proposed by the corresponding author of this paper Wu [15]. Laser sintering has several potential advantages: (1) laser spot scanning trajectories on a flexible substrate can be easily controlled, programed, and changed, and different and complicated composite patterns can be flexibly produced without the need of a mask or vacuum; (2) a laser beam can be focused to a very small spot, leading to a potentially very high spatial resolution; and (3) different from a furnace, laser beam can achieve precise and localized energy delivery during sintering, with small or little undesirable thermal effects to the flexible substrate and the other surrounding regions. Although research work about laser sintering of metal nanoparticles (without CNTs) on polymer or other substrates has been performed by researchers (e.g., [16–18]), the reported fabrication of CNT–metal composites onto a flexible substrate through laser sintering has been relatively rare to the authors' best knowledge. Reference [19] reports laser sintering of conductive flexible adhesive (CFA) onto a flexible substrate. The CFA contains silver flakes, CNTs decorated by silver nanoparticles and a nitrile-butadiene-rubber matrix. The laser-sintered material has a conductivity as high as 25,012 S cm⁻¹. As introduced later, the novel approach in this paper is different from that in Ref. [19] and does not use CFA. Instead, a suspension of CNTs and silver nanoparticles (without a nitrile-butadiene-rubber matrix) in liquid is dispensed onto the flexible substrate, dried, and then laser-sintered. The obtained CNT–silver composites under the studied conditions can have an electrical conductivity of ∼1.1 × 10⁵ S cm⁻¹, which is ∼4.4 times that obtained in Ref. [19].

The major goal of the work reported in this paper is to verify and demonstrate the feasibility of the novel laser-based approach to produce CNT–metal composite lines onto a polymer substrate, and the effectiveness of CNTs in enhancing the bending fatigue performance of the lines on the substrate. Future work is certainly still needed on the laser-based composite fabrication process.

As shown in Fig. 1, the fabrication process of CNT–silver composite lines onto a polyimide substrate mainly consists of the following steps:

Step 1: Preparation of CNT and silver nanoparticle (NP) suspension in ethanol.

Step 2: Dispensing of the prepared suspension onto a polyimide substrate to obtain CNT and silver NP mixture lines.

Step 3: Sintering of the CNT and silver NP mixture lines into CNT–silver composite lines through laser irradiation

Step 1. Preparation of CNT and silver NP suspension

In preparing the CNT and silver NP suspension, 95% ethanol has been used as the solvent, and the prepared suspension has a silver nanoparticle concentration of ∼0.45 g/ml (that is, ∼0.45 g for every milliliter ethanol added to the suspension). To disperse CNTs and decrease their possible agglomeration, sodium dodecyl sulfate (SDS) is used to help the dispersion of CNTs in the suspension. When mixing CNTs, SDS, and silver NPs, their weight ratio is chosen to be 1:4:100, where the CNT to SDS ratio is selected based on information from Ref. [20]. The CNTs are multiwalled CNTs (Product code: TNM5, Chengdu Organic Chemicals Co., Ltd, Chinese Academy of Sciences, Chengdu, China), which have diameters typically in the range of ∼20–30 nm and lengths in the range of ∼10–30 μm with a purity >95% based on the information from the vendor. Figure 2 shows a scanning electron microscope (SEM) image of the silver nanoparticles (from Inframat^® Advanced Materials™, LLC, Manchester, CT) purchased for this work (the silver nanoparticles are put into acetone, and a certain amount of the formed suspension is dropped onto a copper tape, and the SEM observation is made after the acetone is evaporated).

First, the SDS and ethanol are mixed and stirred using a magnetic stirrer for a period of 20 min. Next, CNTs are added and the obtained suspension is stirred using the stirrer for 10 min and then ultrasonicated using an ultrasonic cleaner for 30 min. After this, silver nanoparticles are added to the suspension, which then go through a three-round treatment. In each round, the suspension is stirred for 10 min and then treated by ultrasonication for 30 min. After the three-round treatment, the suspension is stirred again using the magnetic stirrer for a period of ∼20 min or longer. After this, the suspension is ready to be dispensed onto the polyimide substrate.

The suspension of silver nanoparticles in ethanol without CNTs has also been prepared following a similar procedure without using SDS and CNTs (and hence without the involved relevant steps). The prepared suspension also has a silver nanoparticle concentration of ∼0.45 g/ml.

Step 2. Dispensing of the CNT and silver NP suspension onto polyimide

The dispensing of the suspension onto a polyimide substrate is achieved through a Nordson EFD Ultimus II dispensing system, which is a pressure-based fluid dispensing system. The system includes a control unit, a syringe, and a dispense tip (see Fig. 1). The tip orifice used in this work has an inner diameter of 0.2 mm, and the tip—polyimide distance is typically controlled to be less than roughly around 0.1 mm during the dispensing process. The dispensing system is operated in the continuous mode with a dispense pressure of ∼0.1 psi and a vacuum pullback of ∼0.06 psi. During the dispensing process, the polyimide film (which is bonded onto another plastic layer through an intermediate adhesive layer for mechanical support when it comes from the vendor) is positioned on a three-dimensional motion stage, and the linear motion of the stage leads to a line of dispensed suspension on the polyimide surface. Then, the polyimide samples with deposited suspension lines are kept in the ambient air for over ∼24 h to dry. After this, the suspension lines become solid mixtures containing CNTs and silver nanoparticles, which are ready for the following laser sintering step. When dispensing silver NP suspensions without CNTs, the major process parameters, such as the dispense pressure and vacuum pullback applied in the dispensing device and the tip-substrate distance, are about the same. For both the suspensions with and without CNTs, the moving velocity of the dispense tip relative to the substrate is typically in the range of ∼20 mm/s to ∼40 mm/s.

Step 3: Laser sintering of CNT and silver NP mixture lines into CNT–silver composite lines.

The sintering of the dispensed CNT and silver NP mixture lines into CNT–silver composite lines is achieved through laser beam irradiation. The laser (SPI G3.0) used in this work is a fiber laser that can be operated in either a pulsed mode or a continuous-wave mode. In this work, the laser is operated in the continuous-wave mode with a power of around ∼0.29 W to ∼0.46 W. The laser beam, which has a ∼1064 nm wavelength, is delivered onto the CNT and silver NP mixture line on the polyimide surface through a laser scan head (ScanLab, HurryScan 14), which has a lens with a focal length of f = 100 mm. During the sintering process, the sample is put in a small container filled with argon and covered by a piece of glass, through which the laser beam can pass (the container is not drawn in Fig. 1 for simplicity). The laser spot diameter on the sample surface is estimated to be approximately ∼30 μm using the knife-edge method (neglecting the effect of the aforementioned glass cover). During the sintering process, the laser spot is moved using the scan head at a speed of ∼20 mm/s on the sample surface along the dispensed CNT and silver NP mixture line, and the laser beam irradiation can sinter the mixture line into a CNT–silver composite line on the polymer substrate. The width of laser-sintered region is typically narrower than the width of the initially deposited CNT and silver NP mixture line, and after the sintering process, the materials in the un-sintered region can be washed away using an organic solvent. During the laser sintering process, the plastic layer (as mentioned earlier) is still bonded below the polyimide film for mechanical support.

The sintered CNT–silver composites have been observed using a SEM (FEI NOVA nanoSEM Field Emission SEM). The electrical resistance of the sintered lines is measured using the four-point probe method with a multimeter (Keysight Technologies Digital Multimeter 34461A). The approximate average width of a sintered line is determined through the observation using an optical microscope. The approximate average thickness of a sintered line is measured using a white-light interferometer (Bruker ContourGT). Based on the electrical resistance, the length, and the average width and thickness of a sintered line, the average electrical resistivity of the sintered line can be determined.

Fatigue bending tests of laser-sintered silver lines and CNT–silver composite lines on the polyimide substrate have been performed. Figure 3 shows the schematic diagram of the experimental setup for the bending test. The polyimide thin film (with laser-sintered line on its top surface) is bonded onto a plastic support layer through an intermediate adhesive layer. The thickness of the polyimide film, the adhesive layer, and the plastic support is ∼50 μm, ∼10 μm, and ∼75 μm, respectively. One end of the plastic support is fixed, while the other end is connected with a moving panel driven by a linear motor. The original undeformed length of the plastic support is typically ∼50 mm. The motor drives the back-and-forth linear motion of the panel, which has a travel time of ∼1.3 s per cycle and a total travel distance of ∼90 mm per cycle (i.e., d_t given in Fig. 3 is equal to ∼45 mm). During each cycle, the nearest distance between the fixed and the moving panels, d_min, is ∼2 mm. Hence, the smallest radius of curvature for the sintered line on the polyimide top surface is roughly estimated to be ∼1 mm in the region near x = 0 during each cycle (see Fig. 3(c) for the definition of x = 0). If it is roughly assumed that the sample's neutral axis is located at its middle-depth plane, then the maximum tensile strain in the laser-sintered line during the bending test can be approximately estimated as around ∼7.2%.

where $ρ0$ is the initial average electrical resistivity for the laser-sintered line measured before the bending test, $Δρ(t)$ is the real-time average resistivity change for the sintered material in this region, W and T are, respectively, the average width and thickness of the laser-sintered line measured before the bending test, and L is the length of this region and is equal to 5 mm. The real-time total resistance change ΔR(t) can be obtained from the electrical resistance measurement in the bending test, and then based on Eq. (1), a plot of normalized electrical resistivity versus the number of bending cycles can be obtained for each bending test. As introduced earlier, each bending cycle is ∼1.3 s long.

For the bending test, copper wires are bonded with the laser-sintered line using highly electrically conductive silver paint (from Ted Pella, Inc.), and then the copper wires are connected to the wires of the measurement device during each bending test. The silver paint, after being dried, can have sufficiently strong adhesions to the copper wires, the sintered line, and the polyimide substrate, such that there is no contact failure during the tests for the studied cases. It should also be noted that during the bending test, the copper wire-sample contact point is far away from the severe bending region of the sample, and hence, is expected to be subject to relatively small bending-induced strains.

In this study on the fabrication of CNT–silver composites through laser sintering, the following questions are very important and will be discussed based on the experimental results: (1) Has the laser irradiation effectively led to coalition of silver nanoparticles into a relatively more continuous medium? (2) Do CNTs still exist in the sample after laser sintering? (3) What are the electrical resistivities of laser-sintered CNT–silver composite lines compared with silver lines? (4) How does the electrical resistivity change with the number of bending cycles during cyclic bending deformations (i.e., what is the bending fatigue performance) for laser-sintered CNT–silver composite lines compared with silver lines?

The first two questions can be answered based on the SEM images shown in Fig. 4 for laser-sintered CNT–silver composite samples. It can be clearly seen from the SEM images that the laser irradiation during the sintering process has led to coalition of silver nanoparticles into a relatively more continuous medium, although some pores do still exist. CNTs can be clearly observed in the SEM images, and they appear to be bonded with the matrix phase in the images. The CNT-matrix bonding will be further investigated in future work.

Figure 5 shows the measured electrical resistivity for laser-sintered CNT–silver composite lines and silver lines. The lowest average electrical resistivity in the figure for laser-sintered silver line and CNT–silver composite line is around ∼11 μΩ cm and ∼9 μΩ cm, respectively. The values are within the electrical resistivity range obtained from the previous studies on laser sintering of silver particles in Refs. [21–23] (higher than those obtained in Refs. [21,23] and lower than that in Ref. [22]). The resistivity is a few times higher than that for bulk silver (1.59 μΩ cm [23]), and this is likely due to one or more factors, such as pores in the sintered material and electrical resistance due to granular boundaries [13].

Figure 5 shows that for both single-scan and double-scan laser sintering, the electrical resistivity typically decreases as the laser power increases. One important reason is expected to be that higher laser power can yield a better coalition of silver nanoparticles and a lower porosity in the sintered material. However, when the laser power is too high and reaches ∼0.46 W, it has been found in the experiments that clearly observable damages can be induced on the polyimide substrate. It can also be seen from Fig. 5 that under the same power, the double-scan sintering typically leads to a little lower electrical resistivity than the single-scan sintering. One important reason could be that when laser beam scans the sample for the second time, the laser irradiation can lead to a further enhanced nanoparticle coalition and porosity reduction.

A very interesting finding from Fig. 5 is that for all the given conditions, under the same laser power and number of laser scans, the laser-sintered CNT–silver composite lines have a lower average electrical resistivity than the laser-sintered silver lines without CNTs. For example, for the double-scan sintering results given in the figure, the electrical resistivity for the CNT–silver composite lines is typically ∼14% to ∼18% lower than that for the silver lines without CNTs under the same laser power. The addition of CNTs may affect the electrical resistivity of the sintered material due to one or more factors that may include (but may not be limited to) (a) the CNT–silver interface in the composites may induce additional electrical resistance and (b) the CNTs that run though the pores or granular boundaries in the sintered material may provide additional paths or channels for electrical currents to flow [13]. Factor (a) may increase the material electrical resistance, while factor (b) may decrease the resistance. The measurement results in Fig. 5 imply that factor (b) has likely played an important role in the studied cases. Certainly, future work may still be needed to better understand this interesting phenomenon.

Figure 6 shows the normalized real-time electrical resistivity versus the number of bending cycles during fatigue bending tests for laser-sintered silver lines and laser-sintered CNT–silver composite lines on polyimide substrates. As introduced earlier, the plotted normalized electrical resistivity is for the sintered line region of x = −2.5 mm to +2.5 mm (within which most of the major cracks are expected to mainly occur) and is obtained based on the experimentally measured total electrical resistance change, ΔR(t), and Eq. (1). Each of the eight plots in Fig. 6 gives the results for each of the eight comparison groups. Each comparison group includes one laser-sintered silver line and one laser-sintered CNT–silver composite line produced using about the same laser power, the same number of laser scans, and with a similar line thickness after sintering as shown in Table 1.

It can be seen that in each plot in Fig. 6, the overall trend is that the normalized electrical resistivity increases with the number of bending cycles, which is expected to be mainly due to the generation and growth of cracks in the sintered material (including possible material loss that might be associated with the cracks). It can be seen that for most of the tested cases given in Fig. 6, the overall electrical resistivity increase is much slower for the CNT–silver composite line than the silver line in the same group (see group 1, 3, 6, 7, and 8). If it is assumed that a fatigue life is defined as the number of bending cycles when the normalized electrical resistivity reaches 10, then the fatigue life for the lines in each comparison group can be determined based on the data for Fig. 6 and is given in Fig. 7.

In Fig. 7, the fatigue life of almost all the tested CNT–silver composite lines is longer than that for the silver line in the same group (except group 5, where they are close to each other). Six out of the eight tested CNT–silver composite lines have a fatigue life longer than 500 cycles, while none of the tested silver lines has a fatigue life reaching 500 cycles. Half of the tested CNT–silver composite lines have a fatigue life that is close to or longer than ∼1000 cycles. Therefore, the fatigue bending-test results in Figs. 6 and 7 show that under the studied conditions and for the tested samples, overall, the laser-sintered CNT–silver composite lines on the polyimide substrate have a much longer bending fatigue life (as defined earlier) than the laser-sintered silver lines (in other words, the addition of CNTs has greatly enhanced the bending fatigue life of the laser-sintered material on the polyimide substrate).

Figures 8 and 9 show the SEM images for the CNT–silver composite line and the silver line, respectively, for group 7 after the bending test in certain areas within the region of x = −2.5 to +2.5 mm. The silver line has been bent for ∼580 cycles, and the peak real-time normalized electrical resistivity is over 100 during the bending test. The CNT–silver composite line has been bent for ∼2800 cycles, and the peak real-time normalized electrical resistivity during the bending test is still not too far from 10. Cracks can be observed in the lower-magnification SEM image in Fig. 8, left, for the CNT–silver composite line. It is expected that the cracks have contributed to the increase of the electrical resistivity. However, from the higher-magnification SEM image in Fig. 8, right, it can be seen that there appear to be CNTs that run across the crack in the image. It is expected that these CNTs will (1) provide channels for electrical currents to flow across the crack and (2) help inhibit or slow down the further growth of the crack [24–26]. These are expected to be among factors responsible for the better bending fatigue performance shown in Figs. 6 and 7 for the CNT–silver composite lines. Certainly, future work may still be needed to confirm this and also check whether or not other mechanism(s) also plays an important role. Figure 9 shows the SEM images for the silver line, where it can be seen that the crack width in the images is larger than that in the images in Fig. 8. Because there are no CNTs, the corresponding electrical current channels do not exist.

It should also be noted that in this study, multiple samples (without or prior to bending tests) have been observed using an optical microscope at a magnification of 500×. The observations show that very few surface cracks (if any at all) are clearly observable for both laser-sintered silver lines and CNT–silver composite lines.

Figure 10 shows an SEM image of one as-printed CNT–silver nanoparticle line and a cross section profile for another as-printed line. The lines are dispensed onto a polyimide surface, and then dried in air, but not sintered. The cross section profile is an averaged profile measured by a white-light interferometer over a 0.5 mm long section of the line. As mentioned earlier, the laser spot diameter on the sample surface is typically much smaller than the width of the as-printed line, and hence, the eventually sintered line width is typically much narrower than that for the as-printed line. Hence, the exact width of the as-printed line does not matter much for this study. The printed line thickness is typically in the range of around ∼1.0 to ∼2.5 μm. For laser-sintered composite lines, Fig. 4 shows that the CNTs and the matrix phase appear to be bonded. It is expected that the major CNT–silver bonding mechanism in the composite might be through the Van der Waals force [27], and the exact bonding mechanism may still require further study.

This paper has reported research work on a novel laser-based approach to fabricate CNT–metal composites onto a polymer substrate. Suspensions containing CNTs and silver nanoparticles (NPs) are deposited onto the polymer substrate through a dispensing device and dried in the ambient air. Then, the obtained mixture lines containing CNTs and silver NPs on the polymer are laser-sintered into CNT–silver composite lines. Under the investigated conditions and for the tested samples, it has been found:

1. (1)

   The SEM observations of the laser-sintered composites show that (a) CNTs still exist after the sintering process; (b) the sintering process has led to coalition of silver NPs, but some pores still exist; and (c) CNTs and the matrix phase appear to be bonded in the SEM images.

2. (2)

   The addition of CNTs does not degrade the sintered material electrical conductivity; instead, the conductivity has even been slightly increased.

3. (3)

   Overall, the laser-sintered CNT–silver composite lines have a much better bending fatigue performance than the laser-sintered silver lines.

4. (4)

   It is expected that the underlying mechanisms for the observed better fatigue properties may include the following: (a) some CNTs could run across cracks and provide channels for electrical currents to flow across the cracks, and (b) CNTs can help inhibit the further growth of the cracks.

This study suggests that the laser-based approach of fabricating CNT–metal composites onto a flexible substrate has a great potential to enhance the durability and reliability of metallic components in flexible electronic devices without degrading their electrical properties. Certainly, lots of future work (including, but not limited to, further sintered material characterizations) is still needed on the laser-based fabrication process.

The authors would like to thank Mr. Yifan Zhu for his early-stage relevant work.

• Division of Civil, Mechanical and Manufacturing Innovation (CMMI), Directorate for Engineering, the National Science Foundation (Grant No. CMMI 1542376).