Direct Micromachining of Microfluidic Channels on Biodegradable Materials Using Laser Ablation

Laser patterning on polymeric materials is considered a green and rapid manufacturing process with low material selection barrier and high adjustability. Unlike microelectromechanical systems (MEMS), it is a highly flexible processing method, especially useful for prototyping. This study focuses on the development of polymer surface modification method using a 193 nm excimer laser system for the design and fabrication of a microfluidic system similar to that of natural vasculatures. Besides from poly(dimethyl siloxane) (PDMS), laser ablation on biodegradable polymeric material, poly(glycerol sebacate) (PGS) and poly(1,3-diamino-2-hydroxypropane-co-polyol sebacate) (APS) are investigated. Parameters of laser ablation and fabrication techniques to create microchannels are discussed. The results show that nano/micro-sized fractures and cracks are generally observed across PDMS surface after laser ablation, but not on PGS and APS surfaces. The widths of channels are more precise on PGS and APS than those on PDMS. Laser beam size and channel depth are high correlation with a linear relationship. Repeated laser ablations on the same position of scaffolds reveal that the ablation efficiencies and edge quality on PGS and APS are higher than on PDMS, suggesting the high applicability of direct laser machining to PGS and APS. To ensure stable ablation efficiency, effects of defocus distance into polymer surfaces toward laser ablation stability are investigated. The depth of channel is related to the ratio of firing frequency and ablation progression speed. The hydrodynamic simulation of channels suggests that natural blood vessel is similar to the laser patterned U-shaped channels, and the resulting micro-patterns are highly applicable in the field of micro-fabrication and biomedical engineering.


APS Synthesis
The synthesis of APS was carried out as described by Bettinger et al. [43] with slight modification of the reaction parameters. A round-bottom flask was charged with 0.06 mol of 1,3-diamino-2-hydroxy-propane (DAHP), 0.03 mol glycerol (G) and 0.09 mol of sebacic acid (SA) to produce a molar ratio of 2:1:3 of DAHP:G:SA, respectively. The reactants were heated under a nitrogen blanket at 130 • C. The pressure was then dropped to approximately 10 −3 Torr and the contents were allowed to react for 24 h. The product was then stored under a desiccant environment until further use. The product was spread onto silicon wafer molds and glass slides and cured at 180 • C at approximately 10 −3 Torr for 24 h. Film thicknesses of 1 mm were achieved.

PDMS Synthesisar
The PDMS oligomer and crosslinking prepolymer of PDMS agent from a SylgardTM 184 kit (Dow and Corning, Auburn, MI, USA) was mixed in a weight ratio of 10:1 and cast in Petri dishes. After a degas process, PDMS layers with height of approximately 0.1 mm were cured at 65 • C for 24 h.

Laser Ablation System
The channels of pattern were ablated by a New Wave UP193 ArF excimer laser system (Electro Scientific Industries, San Jose, CA, USA). In this study, the working parameters for laser ablation are listed in Table 1. Briefly, the laser wavelength was 190 nm and pulse width 4 ns; ablation progression speed of 100 µm s −1 was used; frequency of laser shots was 10 Hz; the laser energy was between 6-9 J cm −2 ; beam sizes of 10-150 µm were chosen, and the repeated ablations were conducted for 1, 10, 20, 40, 60, 80 and 100 times in this study. During the ablation, the defocus distances were set as 0 (focus), 300, 500, 800 and 1000 µm. The sample was positioned in the ablation cell and could be translated by a computer-controlled XYZ stage to the focused laser beam.

Surface Topography Measurement
Surface topography profilometer (Bruker corp., Billerica, MA, USA) is an advanced thin measurement tool detecting the sample beneath a diamond-tipped stylus. The high-precision stage moves a sample beneath the stylus according to a user-programmed scan length, speed and stylus Polymers 2017, 9, 242 4 of 16 force. In this work, 2000 µm of scan length, 50 µm s −1 of speed, stylus tip size of 2 µm and 10 mg of stylus force were used. The depth of channels in this work was measured by surface topography measurement unless otherwise specified.

Field Emission Scanning Electron Microscope (FE-SEM) Image
Field emission scanning electron microscope (Hitach, Tokyo, Japan) was used to measure the channel depth which is created by repeated ablation. The samples were prepared by sputter coating with palladium for 90 s. Images were taken under 5 and 10-kV accelerating voltages.

Fabrication of Microfluidic Device
A polyethylene tube with inner diameter of 580 µm (Becton Dickinson, San Jose, CA, USA) was glued to a petri dish, and both ends of it were sealed. PDMS prepolymer mixed with crosslinking agent was cast in the petri dish until the entire tube was immersed. After degassing and curing at 65 • C for 24 h, the tube-embedded PDMS was used as the base layer of the flow device. Oxygen plasma etcher (Advanced Research Technology, Hsinchu, Taiwan) was set with oxygen flow rate at 20 cm 3 min −1 for 1 min, chamber pressure at 500 mTorr, and power at 100 W to attach PGS/APS to the PDMS base layer. The sample was ablated by laser system, and the debris in the patterns on PDMS was removed. A connecting channel was created between the PGS/APS patterns and the tubes embedded in PDMS via laser ablation, and the pattern was covered with a thin PDMS/PGS/APS film as described in previous study [9]. Methylene blue solution was used to visualize the flow in microfluidic channels.

Hydrodynamics Simulation of Micropatterns
In this work, the COMSOL Multiphysics ® Modeling Software (5.1, COMSOL Inc., Burlington, MA, USA, 2016) was applied in hydrodynamic simulation of velocity in the micro-channels. The viscosity of fluid and hydraulic diameter were calculated: In Equation (1), flow velocity (v) is equal to the ratio of flow rate (Q) and cross-sectional area (A). The cross-sectional area of squared, U-shaped and circular channels is 2.25 × 10 −8 , 1.91 × 10 −8 and 1.77 × 10 −8 µm 2 , respectively. The hydraulic diameter, D h , can also be calculated by the ratio of four times of cross-sectional area (A) and wetted perimeter of the cross-section (p) (Equation (2)). The wetted perimeter of the cross-section of squared, U-shaped and circular channels is 600, 527 and 471 µm, respectively. The hydraulic diameter can be calculated as 150, 150 and 145 µm for squared, U-shaped and circular channels, and the flow velocity is 1.2 × 10 −2 , 8.7 × 10 −3 , and 9.5 × 10 −3 ms −1 .
When using water as the working medium, the density ( ) was 998.2 kg m −3 and dynamic viscosity (η) was 9.38 × 10 −4 Ns m −2 at 25 • C working temperature. The Reynolds number (Re) was calculated as shown in Equation (3): The Reynolds number of squared, U-shaped and circular channels was found to be 1.9, 1.3 and 1.5, respectively. The entrance length (l θ ) is calculated by Equation (4): To reach the fully developed laminar flow, the entrance length for squared, U-shaped and circular channels is 3.5 × 10 −6 , 2.4 × 10 −6 , and 2.8 × 10 −6 m, respectively. In this study, the flow rates are simulated at 0.05, 0.1 and 0.2 µL min −1 .

Influence of Laser Fluence toward Polymer Patterning
In order to create micro-patterns on polymeric surface via laser ablation, the influences of laser fluence to ablation depth on polymeric materials were examined. In Figure 1, SEM images of PGS, APS and PDMS ablated at 6 J cm −2 are shown, along with PDMS ablated at 9 J cm −2 . At 6 J cm −2 , there are clear channels with microgrooves from laser ablation on both PGS and APS films. In contrast, channels with cracked edges together with rough and porous surfaces along with charred pieces are observed on PDMS at 9 J cm −2 . PDMS ablated at higher energy (9 J cm −2 ) are shown in Figure 1d. More fractures are found in Figure 1d than in Figure 1c, and SEM images clearly indicate that these nano/micro-sized fractures are present along the channels and the density of cavities inside the ablated channels increased with increasing ablation energy. Observation of ablated channels indicates that though the depth of the channels remains steady across different laser fluence, the edge quality of the channels varies widely. In order to create comparable microchannels on PDMS, PGS and APS, it is concluded that the laser fluence capacity used in this study will be set at 6 J cm −2 to avoid creating cracked and burnt PDMS channels.
It is worth noting that these fractures are not only present in the surface of channels, but inside the polymeric materials in a radially out fashion. It is believed that the different ablation efficiency across different materials and the presence of nano/micro-sized fractures were results of a combined effect of two factors: the thermal conductivity of polymeric materials and the total energy received from higher laser fluence. As PDMS was less thermally conductive, high energy was accumulated at the surface of the polymer, creating shallow channels with cracks and burns around the channel. As APS and PGS were both materials synthesized via step growth polymerization under high temperature and low pressure, laser energy conducted and dissipated evenly through a thick surface while maintaining smooth channel surfaces.

Influence of Laser Fluence toward Polymer Patterning
In order to create micro-patterns on polymeric surface via laser ablation, the influences of laser fluence to ablation depth on polymeric materials were examined. In Figure 1, SEM images of PGS, APS and PDMS ablated at 6 J cm −2 are shown, along with PDMS ablated at 9 J cm −2 . At 6 J cm −2 , there are clear channels with microgrooves from laser ablation on both PGS and APS films. In contrast, channels with cracked edges together with rough and porous surfaces along with charred pieces are observed on PDMS at 9 J cm −2 . PDMS ablated at higher energy (9 J cm −2 ) are shown in Figure 1d. More fractures are found in Figure 1d than in Figure 1c, and SEM images clearly indicate that these nano/micro-sized fractures are present along the channels and the density of cavities inside the ablated channels increased with increasing ablation energy. Observation of ablated channels indicates that though the depth of the channels remains steady across different laser fluence, the edge quality of the channels varies widely. In order to create comparable microchannels on PDMS, PGS and APS, it is concluded that the laser fluence capacity used in this study will be set at 6 J cm −2 to avoid creating cracked and burnt PDMS channels.
It is worth noting that these fractures are not only present in the surface of channels, but inside the polymeric materials in a radially out fashion. It is believed that the different ablation efficiency across different materials and the presence of nano/micro-sized fractures were results of a combined effect of two factors: the thermal conductivity of polymeric materials and the total energy received from higher laser fluence. As PDMS was less thermally conductive, high energy was accumulated at the surface of the polymer, creating shallow channels with cracks and burns around the channel. As APS and PGS were both materials synthesized via step growth polymerization under high temperature and low pressure, laser energy conducted and dissipated evenly through a thick surface while maintaining smooth channel surfaces.

Control of Channel Widths and Depths
As the width of channels created via laser ablation depended solely on the size of aperture in a laser ablation system, it is considered the key factor in creating bifurcated networks mimicking blood vessels. Table 2 lists the measured widths of channels created under various beam sizes ranging from 10 µm to 150 µm on PGS, APS and PDMS. It is clear that the accuracy of laser ablation decreased slightly with increasing beam size, but was overall very stable. Overall, it is clear that the channel width created via laser ablation is easily manipulated with high accuracy. Table 2. The widths of linear channels ablated on PGS, APS, and PDMS ablated with 10 µm, 50 µm, 100 µm, and 150 µm beam size at ablation progression speed of 100 µm s −1 , frequency of 10 Hz and energy of 6 J cm −2 . The measured widths are obtained through α-step. All units are in micrometers. Through the modification in beam sizes (aperture size), the width of microchannels is highly tunable, and possesses a wide application, especially in mimicking microvasculature. For example, by utilizing beam size of 10 µm, it is possible to create channels resembling capillary, while 50 µm beams are capable of mimicking the arteriole and venule. Through theoretical calculation, at constant fluence, as the beam sizes increases, the total energy per ablation increases linearly with area. This effect was observed in Figure 2, which measures the variations in depth over different beam sizes (obtained from Surface Topography Profilometer) on PGS, APS and PDMS under fixed ablation fluence 6 J cm −2 . The depth is observed to have increased with beam size, and the coefficients of beam size with depth on PGS, APS and PDMS were 0.9762, 0.9950 and 0.9894, respectively. As discussed in the work of Sizyuk et al. [45], increase in beam sizes (spot sizes) often lead to higher plasma temperature, which may directly result in higher etching efficiency.

Control of Channel Widths and Depths
As the width of channels created via laser ablation depended solely on the size of aperture in a laser ablation system, it is considered the key factor in creating bifurcated networks mimicking blood vessels. Table 2 lists the measured widths of channels created under various beam sizes ranging from 10 μm to 150 μm on PGS, APS and PDMS. It is clear that the accuracy of laser ablation decreased slightly with increasing beam size, but was overall very stable. Overall, it is clear that the channel width created via laser ablation is easily manipulated with high accuracy. Table 2. The widths of linear channels ablated on PGS, APS, and PDMS ablated with 10 μm, 50 μm, 100 μm, and 150 μm beam size at ablation progression speed of 100 μm s −1 , frequency of 10 Hz and energy of 6 J cm −2 . The measured widths are obtained through α-step. All units are in micrometers.

Beam Size
PGS APS PDMS 10 8 ± 1 11 ± 2 12 ± 2 50 48 ± 3 52 ± 3 52 ± 3 100 97 ± 5 104 ± 6 102 ± 6 150 156 ± 9 157 ± 7 157 ± 7 Through the modification in beam sizes (aperture size), the width of microchannels is highly tunable, and possesses a wide application, especially in mimicking microvasculature. For example, by utilizing beam size of 10 μm, it is possible to create channels resembling capillary, while 50 μm beams are capable of mimicking the arteriole and venule. Through theoretical calculation, at constant fluence, as the beam sizes increases, the total energy per ablation increases linearly with area. This effect was observed in Figure 2, which measures the variations in depth over different beam sizes (obtained from Surface Topography Profilometer) on PGS, APS and PDMS under fixed ablation fluence 6 J cm −2 . The depth is observed to have increased with beam size, and the coefficients of beam size with depth on PGS, APS and PDMS were 0.9762, 0.9950 and 0.9894, respectively. As discussed in the work of Sizyuk et al. [45], increase in beam sizes (spot sizes) often lead to higher plasma temperature, which may directly result in higher etching efficiency. The width of ablated channels is directly correlated to the beam size of laser used, and in this work, it ranged between 1 to 150 μm. In order to create wider channels, channels created by overlapping linear laser ablation were investigated. In Figure 3, 350 μm wide channels were created  The width of ablated channels is directly correlated to the beam size of laser used, and in this work, it ranged between 1 to 150 µm. In order to create wider channels, channels created by overlapping linear laser ablation were investigated. In Figure 3, 350 µm wide channels were created with three linear ablations operated with beam size of 150 µm, fluences of 6 J cm −2 , ablation frequency at 10 Hz and ablation progression speed at 100 µm s −1 . Each linear pattern overlapped by 50 µm, and the total width of three ablations with two overlaps was about 350 µm. In Figure 3a, a linear channel of 350 µm wide with two visible overlapped sections are observed under optical microscope. Surface profilometer was employed to further examine the depth of channel made by overlapped patterns, and is shown in Figure 3b. The results show that there are two dimples in the bottom of channels, creating a "w" shaped pattern. It is verified that the position of the dimples in the channel is consistent with the overlapped area of each linear pattern ( Figure 3c). As a result, it may suggest that the two dimples were caused by overlapped ablation, as shown in Figure 3d. The overlapped area was doubly ablated, thus received more energy than other parts, resulting in the two dimples. It is noted that there are micropatterns inside the channels shown in Figures 1a,b and 3a perpendicular to the moving direction of laser beam, but they are relatively shallow compared to the channels, such that they were not observed through the surface profilometer at all, as shown in Figure 3b. with three linear ablations operated with beam size of 150 μm, fluences of 6 J cm −2 , ablation frequency at 10 Hz and ablation progression speed at 100 μm s −1 . Each linear pattern overlapped by 50 μm, and the total width of three ablations with two overlaps was about 350 μm. In Figure 3a, a linear channel of 350 μm wide with two visible overlapped sections are observed under optical microscope. Surface profilometer was employed to further examine the depth of channel made by overlapped patterns, and is shown in Figure 3b. The results show that there are two dimples in the bottom of channels, creating a "w" shaped pattern. It is verified that the position of the dimples in the channel is consistent with the overlapped area of each linear pattern (Figure 3c). As a result, it may suggest that the two dimples were caused by overlapped ablation, as shown in Figure 3d. The overlapped area was doubly ablated, thus received more energy than other parts, resulting in the two dimples. It is noted that there are micropatterns inside the channels shown in Figures 1a,b and 3a perpendicular to the moving direction of laser beam, but they are relatively shallow compared to the channels, such that they were not observed through the surface profilometer at all, as shown in Figure 3b.

Repeated Ablation to Vary Channel Depth
The channel depths of the microchannels are highly tunable by repeated ablation on scaffold. As shown through SEM imaging in Figure 4a, repeated laser ablation on the same position of scaffold created indentations of deeper depth. Good edge quality is observed in PGS in Figure 4b. This is likely due to the thermal stability previously discussed, where the energy from laser source was dissipated evenly into surrounding material instead of accumulating around the pattern creating debris as observed in PDMS (Figure 4c).

Repeated Ablation to Vary Channel Depth
The channel depths of the microchannels are highly tunable by repeated ablation on scaffold. As shown through SEM imaging in Figure 4a, repeated laser ablation on the same position of scaffold created indentations of deeper depth. Good edge quality is observed in PGS in Figure 4b. This is likely due to the thermal stability previously discussed, where the energy from laser source was dissipated evenly into surrounding material instead of accumulating around the pattern creating debris as observed in PDMS (Figure 4c).  The changes in depth are linearly proportional to the number of repeated ablation as shown in Figure 5, with ablation efficiency of 7.64, 6.51 and 3.11 μm per ablation on PGS, APS and PDMS, respectively. As observed above, the ablation efficiency is the highest in PGS and lowest in PDMS, indicating that there may be more precise control over the depth of channels ablated on PDMS, though the edge quality might be slightly lower. Meanwhile, it is most efficient to create patterns using laser ablation on PGS. Overall, laser ablation is capable of creating channels of any depth by controlling the numbers of repeated ablation, easily changing the aspect ratio of any pattern. The changes in depth are linearly proportional to the number of repeated ablation as shown in Figure 5, with ablation efficiency of 7.64, 6.51 and 3.11 µm per ablation on PGS, APS and PDMS, respectively. As observed above, the ablation efficiency is the highest in PGS and lowest in PDMS, indicating that there may be more precise control over the depth of channels ablated on PDMS, though the edge quality might be slightly lower. Meanwhile, it is most efficient to create patterns using laser ablation on PGS. Overall, laser ablation is capable of creating channels of any depth by controlling the numbers of repeated ablation, easily changing the aspect ratio of any pattern.
Through the analyses in Figures 1-5, it is found that even though PDMS is one of the most commonly used materials in the fabrication of microfluidic devices, it is also the least thermally stable amongst the three, thus less susceptible toward laser ablation. APS and PGS both are thermally stable under laser ablation, and PGS is especially useful as a laser ablation material due to its high laser ablation efficiency and high edge quality. Therefore, for the remaining studies, PGS was chosen as the main material of investigation.
indicating that there may be more precise control over the depth of channels ablated on PDMS, though the edge quality might be slightly lower. Meanwhile, it is most efficient to create patterns using laser ablation on PGS. Overall, laser ablation is capable of creating channels of any depth by controlling the numbers of repeated ablation, easily changing the aspect ratio of any pattern. Through the analyses in Figures 1-5, it is found that even though PDMS is one of the most commonly used materials in the fabrication of microfluidic devices, it is also the least thermally stable amongst the three, thus less susceptible toward laser ablation. APS and PGS both are thermally stable under laser ablation, and PGS is especially useful as a laser ablation material due to its high laser ablation efficiency and high edge quality. Therefore, for the remaining studies, PGS was chosen as the main material of investigation.

The Relation between Defocus Distance and Ablation Efficiency
With the direct writing capacity of laser ablation, laser beams were adjusted along the xand y-axes to create microchannels, while repeated ablation enabled the deepening of channels along the z-axis. However, further investigation by adjusting laser focal point along the z-axis was considered vital, since laser focal point offsets may result in defocusing of laser beam during the laser patterning process. In Figure 6, the laser with beam size of 150 µm, fluence of 6 J cm −2 , ablation frequency at 20 Hz and ablation progression speed of 100 µm s −1 was employed to create channels on PGS scaffolds at focus (0 µm) and defocused at 300, 500, 800 and 1000 µm each. The cross-sectional depth and width were measured with surface profilometer and the area was integrated over the measured depth and width. Figure 6a clearly illustrates the increase in width and decrease in depth as laser beams become more negatively defocused, while Figure 6b illustrates the result of positively defocused laser beams. It is clear that the two figures are nearly identical, indicating that when laser beam is defocused from the surface of the polymer surface, positive and negative defocus are equivalent in polymer removal rate. Further investigation of the effect of laser defocus on depth of ablation is shown in Figure 6c, the width of channel increased with the defocus distances, and this phenomenon depends on the size of plasma formed by the laser shot, which is determined by defocus distance of laser from the surface of substrate. A proportional linearity is thus expected. Figure 6d depicts the ablation depth created at different defocus distances. It is observed to remain steady around 10 µm per ablation within 500 µm of defocus, and linearly decreased beyond 500 µm. On the other hand, the width of the channels increased with increasing defocus distance proportionally. Thus the depth of channels were affected by the energy dissipation of plasmas to the substrates at constant fluence, and energy density of the plasmas dropped significantly for larger defocus distance, resulting in reductions in ablated depth and area (Figure 6e). estimation, the defocus distance must be readjusted within depth of 500 μm. However, since Figure  6c clearly illustrates that the width of channels increases with defocus distance linearly, the error margin for width may be well over 200%, making it essential to readjust the defocus distance earlier than 500 μm. In this work, it is suggested that the defocus distance be adjusted every 75 μm ablated in depth, so that the variation in channel width be maintained within 20%.  Figure 7 demonstrates the laser ablation shape and corresponding depth of channels ablated at different ratios between firing frequency and ablation progression speed. In Figure 7a, the channels ablated at ablation progression speed of 100 μm s −1 appear almost doubled in depth than that of 200 μm s −1 when the firing frequency is 10 Hz, indicating an inverse correlation between channel depth and ablation progression speed. The corresponding depth of channels ablated at different ratios between firing frequency and ablation progression speed are presented in Figure 7b. The concept of ablation densities along the moving axis is proposed by calculating the total number of ablations occurring in 1 μm distance. The results suggest that the higher the ablation density, the deeper the depth of channels. This indicates a direct result of increased accumulation of energy per ablation. With the high correlation (R 2 = 0.9951), it is concluded that channel depth is inversely correlated to ablation progression speed. As the ablation progression speed and beam firing frequency is adjusted, By combining the observations in Figure 6d,e, it is clear that for reliable ablation depth estimation, the defocus distance must be readjusted within depth of 500 µm. However, since Figure 6c clearly illustrates that the width of channels increases with defocus distance linearly, the error margin for width may be well over 200%, making it essential to readjust the defocus distance earlier than 500 µm. In this work, it is suggested that the defocus distance be adjusted every 75 µm ablated in depth, so that the variation in channel width be maintained within 20%. Figure 7 demonstrates the laser ablation shape and corresponding depth of channels ablated at different ratios between firing frequency and ablation progression speed. In Figure 7a, the channels ablated at ablation progression speed of 100 µm s −1 appear almost doubled in depth than that of 200 µm s −1 when the firing frequency is 10 Hz, indicating an inverse correlation between channel depth and ablation progression speed. The corresponding depth of channels ablated at different ratios between firing frequency and ablation progression speed are presented in Figure 7b. The concept of ablation densities along the moving axis is proposed by calculating the total number of ablations occurring in 1 µm distance. The results suggest that the higher the ablation density, the deeper the depth of channels. This indicates a direct result of increased accumulation of energy per ablation. With the high correlation (R 2 = 0.9951), it is concluded that channel depth is inversely correlated to ablation progression speed. As the ablation progression speed and beam firing frequency is adjusted, it is hypothesized that the similar "ratio" between the two is a more dominant factor than the absolute numbers. To analyze from an ablation per area perspective, a laser beam firing at 10 Hz and moving at 100 µm s −1 fires the same number of shots as one ablating at 20 Hz, 200 µm s −1 . The main difference between the two is that the amount of energy received in the same amount of time is doubled. As shown in Figure 7c, the ablation shape made by the different ablation progression speed and beam firing frequency is the same regardless of the absolute ablation progression speed and beam firing frequency. This phenomenon is also observed by Arutinov et al. with respect to nanosecond and picosecond lasers [16].

Hydrodynamic Simulation of Laser Patterned Channels
As described in Figure 4a-c, microfluidic channels created with laser ablation are generally Ushaped. In order to understand the difference between the laser-ablated micro-channels and natural blood vessels, an analysis using COMSOL hydrodynamic simulation is performed. Three shapes of channels were chosen for the hydrodynamic simulation: squared, U-shaped and circular crosssectional channels as shown in Table 3. These three shapes were chosen as square and U-shaped

Hydrodynamic Simulation of Laser Patterned Channels
As described in Figure 4a-c, microfluidic channels created with laser ablation are generally U-shaped. In order to understand the difference between the laser-ablated micro-channels and natural blood vessels, an analysis using COMSOL hydrodynamic simulation is performed. Three shapes of channels were chosen for the hydrodynamic simulation: squared, U-shaped and circular cross-sectional channels as shown in Table 3. These three shapes were chosen as square and U-shaped channels share the linear top edge, while circular channels are most similar to that of the natural blood vessels. The simulation results suggest that geometric variations of microchannels present nearly no influence on the hydrodynamics except for the maximum velocity. The maximum velocity of the flux was 156, 180 and 191 µm s −1 in squared, U-shaped and circular channels, respectively, when flow rate is fixed at 0.1 µL min −1 . The maximum velocity between U-shaped and circular channels was different by 6.11%, but the maximum velocities between U-shaped and squared channels were different by 13.33%. This suggests that the performance of channels made by laser ablation (U-shaped) is more similar to that of natural blood vessel (circle) than a squared channel, and the assembly process of the U-shaped microfluidic device is very similar to those fabricated via MEMS-Fab.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

The Fabrication of Simple Microfluidic Systems via Laser Ablation
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 µL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.
Three microfluidic patterns-single-channels, honeycombs and mesh-like patterns-were prototyped using laser ablation on PDMS, PGS and APS, as shown in Figure 8. The microfluidic device with single-channel patterns may serve as micro-reactor, and consume less reagent and sample than bulk systems [46]. The honeycombs and mesh-like patterns were adapted from studies of biomaterial scaffolds [47,48], which proved that laser ablation is useful in creating micropatterns. In this study, patterns were tested at 0.1 μL min −1 for a minimum of 24 h, and a minimum of three flow experiments were repeated to prove the robustness and repeatability. The tests of flow in the channels are displayed in the animation (Video S1). However, it should be mentioned that the flows in the microfluidic patterns on PDMS were blocked more often than those on PGS and APS due to the excess debris in PDMS channels, largely due to the poor edge quality from laser ablation.

Conclusions
In this study, laser ablation was applied as a simple and convenient method for the fabrication of microfluidic flow devices. By removing the steps of masks/molds production in photolithographic or soft photolithographic technique, this fabrication method holds great potential as a prototyping method. The relations of ablated width as well as depth-to-laser parameters were investigated, and the relations between laser energy and surface changing on the polymeric materials were examined. The desired depth of pattern was adjusted by repeated ablations, and accompanied by defocus adjustments, as it was found that the defocus distance affects the channel width and depth, significant error when larger than 500 μm. Therefore, adjustment in focus is suggested to be made every 75 μm for stable ablation efficiency. Further investigation toward depth manipulation using laser ablation also indicated that ablation density based on the combination of firing frequency and ablation progression speed is another factor in controlling laser ablation efficiency. Although traditional MEMS-Fab is well known for high-precision manufacturing of microchannels with a wide range of width, it is highly limited by the photolithographic property toward depth variation within the same mold. Meanwhile, one common obstacle in reverse molding of polymeric material is the challenge in delaminating the patterned films without damaging the micro/nanopatterns. Instead, flat polymeric films, which are much easier to produce, are employed after delamination in direct writing using

Conclusions
In this study, laser ablation was applied as a simple and convenient method for the fabrication of microfluidic flow devices. By removing the steps of masks/molds production in photolithographic or soft photolithographic technique, this fabrication method holds great potential as a prototyping method. The relations of ablated width as well as depth-to-laser parameters were investigated, and the relations between laser energy and surface changing on the polymeric materials were examined. The desired depth of pattern was adjusted by repeated ablations, and accompanied by defocus adjustments, as it was found that the defocus distance affects the channel width and depth, significant error when larger than 500 µm. Therefore, adjustment in focus is suggested to be made every 75 µm for stable ablation efficiency. Further investigation toward depth manipulation using laser ablation also indicated that ablation density based on the combination of firing frequency and ablation progression speed is another factor in controlling laser ablation efficiency. Although traditional MEMS-Fab is well known for high-precision manufacturing of microchannels with a wide range of width, it is highly limited by the photolithographic property toward depth variation within the same mold. Meanwhile, one common obstacle in reverse molding of polymeric material is the challenge in delaminating the patterned films without damaging the micro/nanopatterns. Instead, flat polymeric films, which are much easier to produce, are employed after delamination in direct writing using laser ablation, greatly reducing the risk of damaging the patterns. Through the application of laser ablation, wide and deep channels can be produced alongside narrow and shallow channels in one simple process, creating microfluidic systems that are much more analogous to natural vasculatures.
Three prototypes of microfluidic designed in this study were revealed, and channel patterns were created on the polymeric materials for modular bio-analytical chip applications. Using laser ablation, device fabrication and prototyping were fast and precise. The technique may become useful in analyzing liquid sample, simulating blood vessels and sample pretreatment or mixing device. It is proven that laser ablation is an effective method for the creation of microfluidic devices with micro-even nano-sized to mimic the blood vessels. The reproducibility and quality of devices are excellent for biological and chemical researches. Simulation suggested that the performance of channels made by laser ablation is similar to that of natural blood vessel. By carefully controlling the parameters of laser, high-precision micro-channels without cracks/debris were produced. The predictable ablation performances on PGS as well as APS reveal the potential in application of ArF laser in the microfabrication of polymeric materials and bioengineering studies. However, much similar to the result of predecessors, PDMS is again proven a poor material of choice for laser ablation due to its low thermal conductively and poor edge quality.