Team:Penn/Results
Characterizing Wax Droplets
Since droplets are the minimum outputs of our wax printer, we first investigated the property of wax drops to characterize the performance of our printer and optimize mechanical and electrical settings for better performance.
Extrusion of wax droplets was conducted by applying pressure for a set amount of time in the scale of milliseconds. To examine dependent variables of droplet properties, we altered three physical parameters: the extruder temperature, the amount of pressure, and the duration of pressure. We printed 96 drops for each setting and quantified them by measuring their sizes and weights, as well as calculating the success rates of extrusion. At the same time, we briefly assessed how well wax droplets permeate through the filter paper.
Fig1 shows how the size and weight of wax droplets change with altered duration of pressure, where the temperature and the amount of pressure were fixed to 80 [℃] and 5.0 [psi] respectively. It is demonstrated that the volume of wax droplets increases as the duration of pressure increases. A similar trend was observed when the amount of pressure was varied while the duration and temperature was fixed.
Fig1.For each duration of pressure, 96 wax drops were printed and successfully extruded droplets were measured for their weight. Temperature was fixed to 80[℃] and the amount of pressure was fixed to 5.0 [psi].
While printing wax droplets, we found that droplets sometimes do not get extruded or get extruded with extra wax from a previous extrusion, so we calculated the rates of wax droplets to be successfully extruded. We defined the extrusion to fail when there is no extrusion occurring or when the extruded droplet is twice as big as other droplets. The success rate is highly correlated to the pressure pulse length but is not highly correlated to the amount of pressure. This is probably because all of the pressure we tested was above the minimum amount needed to extrude wax from our device. Also, little or no correlation was seen between the success rate and the temperature when testing temperatures higher than 80 [℃].
Fig2. Surface plot of successful extrusion rate (0.0-1.0) where duration or amount of pressure was changed between 20[ms] and 50[ms] or 5.0 [psi] and 10.0 [psi] respectively. Temperature was set to 80 [℃].
Next, we tested how wax permeates through the paper (Whatman®︎ qualitative filter paper, grade 595) at different extrusion temperatures. Fig.3 shows pictures taken from the backside of filter papers where wax droplets (light blue) were printed and dyed water (dark blue) flowed through. As shown in those pictures, wax permeates better when the temperature is higher presumably due to the increased fluidity. Although increasing temperature resulted in larger droplets on paper, it also reduced the minimum extrudable pressure from 4.0 [psi] (80℃) to 2.5 [psi] (90℃), which compensates for the resolution loss (Fig3C).
Fig3. Pictures taken from the backside of paper where wax droplets were printed and dyed water flowed afterwards. Areas where droplets did not permeate through the paper are shown with white circles. A) temperature: 80 [℃] pressure: 5.0 [psi], B) temperature: 90 [℃] pressure: 5.0 [psi], C) temperature: 90 [℃] pressure: 3.0 [psi].
Characterizing Wax Lines
Extruding wax lines with high consistency and resolution is crucial to print functional paper microfluidic devices. We compared two different extruding methods to draw lines: continuous extrusion and drop-wise extrusion. Using each method, our printer either continuously extrudes the stream of wax or intermittently emits wax droplets onto the paper while moving, as if printing pixels.
We first optimized parameters for drop-wise extrusion. In addition to the parameters investigated during characterization, we altered another parameter, the interval of drops. Pictures were taken of drop-wise printed wax lines and, consequently, those pictures were converted into binary images. The linewidth was measured by using our custom ImageJ plugin function that counts pixels at 500 different points of each line, and its median was determined. The line completeness was defined as the probability of the counted pixels not being 0 within 500 values.
Increasing the interval of droplets merely helps printing thinner lines because of the huge variation in the line width itself (Fig4 Left). Another problem is the trade-off between line resolution and continuation of drops: when the interval of drops is longer, the resolution is higher, but the line itself tends to be more discontinuous (Fig4 Right). Theoretically, there should be an optimal interval for this trade-off, but the success rate of wax drop (~95%) was seemingly too low to find the sweet spot.
Fig4. The amount of pressure, duration, or temperature was set to 3.0[psi], 30 [ms], or 90 [℃] respectively. Left) Each data point represents the median linewidth within each line. The line and the band represent linear regression and its 95% confidence intervals. Right) Each data point represents the line completion within each line. The line and the band represent logistic regression and its 95% confidence intervals.
For continuous extrusion we took an additional parameter, the speed of the extruder movement, into consideration. With increased printing speed, line width logarithmically decreased with reasonably small variations among printing processes(Fig. 5).
To note, one drawback of using the continuous extrusion method is that hot liquid wax could move as the print bed moves and spread around the print area. This happens because the printing speed for the continuous method is sometimes too fast for the wax to get solidified before the next print bed movement. The problem could be solved by modifying our munging script to optimize the printing process.
Fig5. Each data point represents the median linewidth within each line. The line and the band represent the logarithmic fitting curve and its 95% confidence intervals. Throughout the experiment, the amount of pressure was fixed to 3.0 [psi], and temperature was set to 90 [℃].
In order to comprehensively compare different extrusion methods and parameters, we analyzed printed lines to determine three features: resolution, consistency, and completeness. The resolution, consistency, and completeness were estimated by measuring the median linewidth, the coefficient of variation of the median linewidth, and the line completion, respectively. Table 1 summarizes those features for potential sets of extrusion method and parameters. Between the two extrusion methods, the continuous method is better in many aspects and among different parameters, the printing speed of 10000 [mm/min] was chosen as best for having high consistency, line completeness, and resolution compared to others.
Table 1. Summary of statistical features of lines printed under different conditions. Lines to measure statistical features were independently printed for three times (N=3). The average linewidth was defined as mean ± standard deviation of three median linewidth. The coefficient of variation was determined by dividing the variance of three median linewidth by the mean value.
Characterizing Channels
Using the continuous extrusion method with optimized parameters, we further investigated how closely two lines could align to create a hydrophilic channel, which is a key component of microfluidics. Printed channels were analyzed by following our line width analysis method, except that dyed water was applied to each channel so as to make binary images.
Although channel completion requires at least 5mm of designed channel width to reliably create microfluidic channels, designed width and actual width of channels clearly demonstrate a linear relationship. For these investigations, we concluded that the resolution limit of our current version of wax printer is 5 [mm], and our resolution loss from the ideal line is about 2 [mm].
Fig6. Left) Each data point represents the median channel width within each printed channel. The solid line and band show the linear regression line and its 95% confidence intervals. The dashed line shows ideal correspondence (y=x) between the designed width and printed width. Right) Each data point represents the channel completeness (0.0-1.0) for each printed channel. The fitted line and band show the logistic regression line and its 95% confidence intervals.
Printing Functional Devices
Fig7shows a microfluidic device printed with our assembled wax printer. In order to avoid using low-quality sections of wax lines, we further modified our munging script. It inserts resting commands when the extruder rapidly changes its direction of movement.
First, our munging code virtually sets the circle depicted as a dashed line in the figures, whose points later become resting areas for the extruder when it pauses extruding. The code inserts resting commands whenever angles between lines exceeds the implemented threshold (>60[deg] in Fig7), while avoiding lines in the design area.
Fig7. Top Left) Virtual demonstration of wax printing. Solid lines represent where wax is extruded, solid points represent where the extruder pauses, and the circle drawn with dashed lines shows where the potential resting areas are. Top Right) Figure of device using the code depicted within the animation. Bottom) Picture of printed device.