Nature: Research on direct laser lithography of copper using low-cost 3D printers on flexible substrates

The research on direct laser molding of copper (Cu) on a flexible substrate by researchers at Pusan National University was published in Scientific Reports, a sub-journal of Nature, with the title of "Copper laser patterning on a flexible substrate using a cost-effective 3D printer".

Using a 405 nm laser module connected to a 3D printer, the researchers studied the performance of a thin polyimide substrate (PI thickness: 12.5-50 μ m) Effective direct laser modeling of copper (Cu) is carried out (Fig. 1). Researchers studied a 3D printer with a laser module (less than $1000). Polyimide (PI) is used as a lightweight flexible polymer substrate to replace the current glass substrate. Polyimide has many advantages, such as mechanical strength, chemical resistance, heat resistance and thermal stability based on rigid aromatic backbone. Three methods are introduced to find the focal length of the laser: using the beam spot analysis of USB camera, locating the ignition point according to the z-axis height, and using the G-code program to form the Cu model at different z-axis heights.

Figure 1: Laser integrated 3D printer with USB camera.

Figure 2: Focusing laser beam: (a) find the laser focal length using USB camera, (b) determine the focal length using the ignition point on the bare PI film under the 1.6% pulse width modulation (PWM) input signal, and (c) determine the focal length using the G-code program (scanning speed is 1mm/s and 2% PWM) through the Cu model formed on different focal lengths.

Laser power stability is an important factor to improve the quality of the model. The researchers tested the input signal that the laser output depends on the PWM method. The output power of the laser module increases linearly to about 70% of the PWM signal, and then decreases slightly against the input expectation, possibly due to the cooling capacity of the laser module (Fig. 3a). Figure 3b shows that the deviation of the laser output increases with the increase of the input power signal (measured 30 seconds after the laser is turned on). Since the input power used in this study is 38% of the maximum power (about 260 mW), this experiment is calculated at a scanning speed of 4mm/s. Since the "power on" time of the laser is less than a few seconds, it is expected that the power deviation of the laser will be far less than 0.6%.

Figure 3: Characteristics of laser output power changing with input signal: (a) Laser output changing with PWM input ratio; (b) In the "ON" state, the laser output fluctuates under different PWM input ratios.

The shape and defects of Cu model were studied by increasing laser power at a certain scanning speed and fixed focal length (LFL or SFL). For the case of LFL in Fig. 4a (3mm longer than AFL), with the increase of laser power, the line width and grain size of Cu increase. When the input power is greater than 50%, line defects appear again in the Cu pattern. For the SFL case in Figure 4b (2.4 mm shorter than AFL), similar results were observed to the LFL case, except for the laser marking on the Cu model. Interestingly, due to the different angle and diameter of the incident laser beam, the line width of SFL in Fig. 4c seems to increase with the laser power at a lower rate than that of LFL. Based on these results, including other preliminary tests, 38% of PWM power input is selected for copper direct laser molding.

Figure 4: Copper microscope image of plate with input PWM signal ratio changing from 12.5 to 100%: (a) copper wire formed at LFL and 4 mm/s scanning speed, and (b) copper wire formed at SFL and 4 mm/s scanning speed. Microscopic images and linewidth of Cu models formed on various SFL and LFL to find appropriate focal length to minimize defects: (c) focal length (LFL) from 2.4 to 4 mm, and (d) RFL-based focal length (SFL) from − 1.8 to − 3.4 mm. Please note that the Cu model is obtained at 38% PWM input and 4 mm/s scan speed.

The laser irradiation time is directly related to the scanning speed (or speed) of the laser module attached to the 3D printer. The researchers increased the scanning speed at 38% power input and predetermined focal length, and checked the Cu model at the same time. In Figure 5a, the LFL clearly shows that at low scanning speed, the particle size of Cu increases, and the line width decreases with the increase of scanning speed. In Figure 5b, the linewidth of SFL also decreases similarly, but the linewidth change rate is lower than that of LFL (Figure 5c). The SEM images of SFL and LFL in Fig. 5d-e are consistent with the microscope images. It can be seen that the particle size of Cu gradually increases with the decrease of scanning speed. The analysis shows that the content of C in Cu spectra increases with the increase of scanning rate. Based on these results, the scanning speed is determined to be 4mm/s.

Figure 5: Under 38% PWM input condition, microscope and scanning electron microscope (SEM) images of 1~8mm/s laser scanning speed on model copper: (a) copper wire formed by LFL, (b) copper wire formed by SFL, (c) relationship between line width of Cu model and scanning speed, (d) SEM image of LFL, (e) relationship between SEM image of SFL and scanning speed. (a) The illustrations in and (b) are enlarged images of each copper model.

Under proper conditions of determining laser power, scanning speed and focal length, 8 × 8 mm2 square Cu diagram in different laser scanning gaps (50, 70 and 90 μ m) Resistivity under. Figure 6a shows the camera image of the Cu square model that formed the residual carbon strip during the sintering process. Perform a laser scan on the LFL at 50 μ About 830 can be seen at m μ Ω · cm, at 70 μ About 5.4 Ω· cm can be seen at m, at 100 μ About 4.9 Ω· cm can be seen at m. In order to fix this high resistivity and further test the effect of laser treatment, the copper model was laser treated again. Further laser scanning gradually reduced the resistivity of Cu diagram to 70 μ Ω · cm (Fig. 6b). For 70 and 90 μ The resistivity drops sharply to about 70% in the m-line scanning gap μ Ω · cm (Fig. 6c-d). 70 in Figure 6e μ The scanning electron microscope images of m model show that the particle size of Cu increases gradually with the extension of laser scanning time. According to the EDX analysis of the model, the C/Cu ratio also decreases with the increase of laser scanning times.

Figure 6: In different scanning gaps (50, 70 and 90 μ m) And resistivity of Cu square model formed under LFL: (a) Cu square on PI (8 × 8 mm2) camera image and microscope image of the model, (b) 50 μ M resistivity curve of scanning gap, (c) 70 μ M scan gap, (d) 90 μ Resistivity curve of m-scan gap and multiple laser scans, (e) 70 μ M model and SEM image of C/Cu ratio of multiple laser scans.

Figure 7a shows the camera and microscope images of the Cu square model formed in SFL, where there are carbon residual strip lines similar to the Cu model formed in LFL. The resistivity of Cu diagram is completely different from that of LFL diagram. Despite the scanning gap, after a laser scanning, the resistivity of all models is much lower than that of LFL (lower than 60 in Figure 7b-d μ Ω·cm)。 This resistivity is equivalent to that of the Cu square model formed on the glass substrate at 100% input power (Fig. 7c).

Figure 7: In different scanning gaps (50, 70, 90 μ m) And resistivity of Cu square model formed under SFL: (a) Cu square on PI (8 × 8 mm2) camera image and microscope image of the model, (b) 50 μ M resistivity curve of scanning gap, (c) 70 μ M scan gap, (d) 90 μ M Scan gap. Please note that the resistivity map includes the resistivity comparison in the vertical/horizontal direction relative to the laser scanning direction and the resistivity comparison before and after cleaning in each case.

Figure 8: At 25 μ M The size of 5 is prepared on the PI × 30 mm2 square copper, and the bending test was carried out on a self-made bending machine. After 1000 bends (bending radius of 2.5 mm), the resistance of the sample changes from 0.87 to~0.88 Ω/mm, as shown in Figure 8. Change of resistance of laser sintered copper under multiple bending. The bending speed is 100 mm/s and the bending radius is 2.5 mm.

The Cu precursor is coated with a thickness of 12.5, 25 and 50 respectively μ M of PI film, laser sintering. Figure 9a shows the explicit Cu model in all PI films, although the thickness is different. The illustration in Figure 9a shows that the LED works normally. A potential application of this electrode may be used in small bioelectronic devices containing biosensors. Therefore, after connecting the PI and LED with Cu model to the arm skin, the same test was carried out. The results show that the LED can operate well even if the substrate PI is bent during power connection (Fig. 9b).

Figure 9: Various Cu models designed to test the conductive connection with LED: (a) Different thickness (12.5, 25 and 50) on the glass plate μ m) Cu model formed on PI film; (b) LED working on the Cu model on the PI film attached to the arm skin.

The researchers demonstrated the operation of the LED connected to the Cu model on the PI attached to the arm skin. The LED can work normally even if the substrate PI is bent during the power connection. It is expected that this method will be applied in manufacturing bioelectronics, including biosensors.

Article source:
https://www.nature.com/articles/s41598-022-25778-y