Abstract:Inkjet printing is a promising technique for printed micro-electronics due to low cost, customizability and compatibility with large-area, flexible substrates. However, printed line shapes can suffer from bulges at the start of lines and at corner points in 2D line patterns. The printed pattern can be multiple times wider than the designed linewidth. This can severely impact manufacturing accuracy and achievable circuit density. Bulging can be difficult to prevent without changing the ink-substrate-system, the… Show more
“…All of these ink properties affect the jetting and the pattern formation on the substrate. Pattern formation on the substrate can be further manipulated by changing the properties of the substrate and the spacing and order in which drops are deposited [10][11][12][13][14][15][16][17]. Here, we will focus on the ejection of drops from the nozzle, which is the first step in the inkjet process.…”
Machine learning (ML) as a predictive methodology can potentially reduce the configuration cost and workload of inkjet printing. Inkjet printing has many advantages for additive manufacturing and printed electronics including low cost, scalability, non-contact printing and on-demand customization. Inkjet generates droplets with a piezoelectric dispenser controlled through frequency, voltage pulse and timing parameters. A major challenge is the design of jettable inks and the rapid optimization of stable jetting conditions whilst preventing common problems (no ejection, perturbation, satellite drop, multiple drops, drop breaking, nozzle clogging). Material consuming trial and error experiments are replaced here with a machine learning based jetting window. A data set of machine and material properties is created from literature and experimental data. After exploratory data analysis and feature identification, various (linear and non-linear) regression models are compared in detail. The models are trained on 80% of the data and root mean square error (RMSE) is calculated on 20% test data. Simple polynomial relationships between the input and output features yield coarse prediction. Instead, small ensembles of decision trees (boosted decision trees and random forests) have improved predictive power for drop velocity and radius with RMSE of 0.39 m/s and 2.21 µm respectively. The mean absolute percentage error (MAPE) is 3.87%. The models are validated with experimentally collected data for a novel ink where no data points with this ink were included in the training set. Additionally, several classification algorithms are utilized to categorize ink and printer parameters by jetting regime (‘single drop’, ‘multiple drops’, ‘no ejection’). Categorization and regression models are combined to improve overall model prediction. Machine learning enables efficient material and printing parameter selection speeding up the development of novel ink materials for printed electronics by eliminating jetting experiments that are money, time and material intensive.
“…All of these ink properties affect the jetting and the pattern formation on the substrate. Pattern formation on the substrate can be further manipulated by changing the properties of the substrate and the spacing and order in which drops are deposited [10][11][12][13][14][15][16][17]. Here, we will focus on the ejection of drops from the nozzle, which is the first step in the inkjet process.…”
Machine learning (ML) as a predictive methodology can potentially reduce the configuration cost and workload of inkjet printing. Inkjet printing has many advantages for additive manufacturing and printed electronics including low cost, scalability, non-contact printing and on-demand customization. Inkjet generates droplets with a piezoelectric dispenser controlled through frequency, voltage pulse and timing parameters. A major challenge is the design of jettable inks and the rapid optimization of stable jetting conditions whilst preventing common problems (no ejection, perturbation, satellite drop, multiple drops, drop breaking, nozzle clogging). Material consuming trial and error experiments are replaced here with a machine learning based jetting window. A data set of machine and material properties is created from literature and experimental data. After exploratory data analysis and feature identification, various (linear and non-linear) regression models are compared in detail. The models are trained on 80% of the data and root mean square error (RMSE) is calculated on 20% test data. Simple polynomial relationships between the input and output features yield coarse prediction. Instead, small ensembles of decision trees (boosted decision trees and random forests) have improved predictive power for drop velocity and radius with RMSE of 0.39 m/s and 2.21 µm respectively. The mean absolute percentage error (MAPE) is 3.87%. The models are validated with experimentally collected data for a novel ink where no data points with this ink were included in the training set. Additionally, several classification algorithms are utilized to categorize ink and printer parameters by jetting regime (‘single drop’, ‘multiple drops’, ‘no ejection’). Categorization and regression models are combined to improve overall model prediction. Machine learning enables efficient material and printing parameter selection speeding up the development of novel ink materials for printed electronics by eliminating jetting experiments that are money, time and material intensive.
“…Then, the next segment starts with the fourth drop placed at a distance of (DSp + 2CDSp) relative to the center of the first segment (the third drop), leaving 2CDSp of vacant space between two segments for a connecting drop to be filled in after all the segments have been printed. Previous work has experimentally shown that CDSp should be smaller than DSP, and the optimum CDSp varies between 0.6 and 0.95 times DSp for different substrate and ink combinations [20]. Here, the symmetric printing process is implemented for arbitrary patterns using a vision pipeline shown in figure 6(b).…”
“…Following the printing of the outer two drops of each segment, the central drop is printed and does not experience a pressure gradient due to the symmetry of the segment. Subsequently, the three-drop segments are joined with a connecting drop while maintaining pressure equilibrium on either side of the pattern [20]. Figure 6(a) compares the raster and symmetric drop ordering methodologies.…”
“…There has been significant work to understand the fluid mechanics of these nonidealities and avoid them for simple designs such as lines or squares by changing the location and order in which drops are printed. Inkjet pattern control has been thoroughly investigated for lines [12,[15][16][17][18][19][20][21] and 2D shapes [22][23][24] by characterizing ink and substrate properties and manipulating drop and lineto-line spacing [25,26], which can enhance printed morphologies. However, automated implementation of optimized drop and line-to-line spacing for arbitrary complex circuit patterns and industrial inkjetmanufacturing is still a challenge.…”
Section: Introductionmentioning
confidence: 99%
“…The morphology of micro-scale liquid tracks, especially bulging, has been investigated before using fluid dynamics methods [15,19]. Pattern segmentation and symmetric printing of segments can inhibit bulging in line patterns compared to traditional raster printing [20]. Preprinting the contour of any feature improves corner morphology through enhanced pinning and additional anchoring in front of each segment can significantly subdue the bulging effect [21,26].…”
Inkjet printing technology for printed microelectronics suffers from a number of non-idealities due to unwanted ink flow on the substrate. This can be mitigated and pattern fidelity can be improved by using an optimized drop placement sequence in contrast to the standard raster-scanning approach. However, it is challenging to auto-generate such printing sequences for complex printed patterns. Here, the generation and evaluation of the printing sequence are turned into a computer-vision problem. The desired printed pattern is taken as an input image and converted into a printing sequence using contour, symmetric, and matrix sequencing and corner compensation. After printing, pattern defects are detected by automated image processing to evaluate the printed pattern against the designed ground truth image and to determine the best possible algorithm for printing sequence generation for different pattern types. The machine vision-based experimental approach identifies the best solutions for solving the printing and defect optimization problem in terms of precision, recall, and accuracy. This methodology will enable the automated design of electronic circuits for applications such as wearable sensors, low-cost radio-frequency identification tags, or flexible displays.
While inkjet printing on many hydrophilic surfaces can be achieved through control of drop spacing and droplet deposition delay alone, the same for hydrophobic substrates prove to be challenging. The low surface energies of hydrophobic surfaces lead to dewetting, unwanted coalescence of wet drops, and bulging, preventing intact and uniform lines of low‐viscosity ink to form. In this paper, we have adapted the stacked coin strategy, a well‐established and successful technique for hydrophilic surfaces, for use on hydrophobic surfaces. Stacked coin morphology occurs when the time taken to deposit a subsequent overlapping droplet in a pattern is longer than the evaporation time of the prior droplet. On hydrophobic surfaces, it is considerably more challenging and the parameter window for successful printing is smaller than on hydrophilic surfaces, thus requiring in‐depth study to implement this methodology on hydrophobic surfaces. We conducted experiments using a custom‐built inkjet printer with variable stage speed and stage temperature, where silver nanoparticle ink was used to print on Teflon‐AF substrates. Our study identified the following regimes: isolated droplets, isolated multi‐droplets, broken line, true stacked coin, and delamination. The relationship between substrate temperature, drop spacing and droplet deposition delay controls the printability window of true stacked coin morphology and its surrounding regimes. The key to the success of this strategy on hydrophobic surfaces is rapid drying of individual droplets to minimize movement due to hydrophilic coalescence of overlapping dried and wet droplets. Our study identified 180⁰C as the critical temperature for instantaneous drying of the studied ink, and a maximum drop spacing of 20µm.This article is protected by copyright. All rights reserved.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.