Proceedings - Rochester Institute of Technologyedge.rit.edu/content/P10551/public/Final Documentation... · Web viewThis has vast and powerful implications for the future of prototyping

Multi-Disciplinary Senior Design ConferenceKate Gleason College of Engineering

Rochester Institute of TechnologyRochester, New York 14623

Project Number: P10551

NANO INK DEPOSITION SYSTEM

Eric Hettler/Mechanical Engineer Project Manager

William Gallagher/Industrial and Lead Engineer

Joseph Cole/Electrical Engineer Christopher Mieney/Mechanical Engineer

Gregory Ryan/Mechanical Engineer

ABSTRACT

The primary project goal was to modify the preexisting open-source platform known as Fab@Home to deliver a system capable of printing photo-curable epoxy resins with on-the-fly color control. The system has demonstrated the capability to print multiple colors. It has not, however, achieved fluid mixing to deliver a smooth, continuous color gradient.

NOMENCLATURE

Fab@Home – the community based open-source rapid prototyping effort originated at Cornell University

Model 1 – the first generation rapid prototyping platform produced by the Fab@Home effort

FabAmerica – the platform developed at RIT described in this report

Work Part – the physical part being printed by the machine

STL – common rapid prototyping file format using triangular faces to render part geometry

INTRODUCTION AND BACKGROUND

Fab@Home is a community based open-source rapid prototyping effort originated at Cornell University, with the goal of providing consumers with an affordable means of producing functional objects. “The ability to directly fabricate functional custom objects could transform the way we design, make, deliver and consume products. By eliminating many of the barriers of resource and abilities that currently prevent ordinary inventors from realizing their own ideas, fabbers can “democratize innovation.”[4] With this expressed mission, it is important to realize that FabAmerica was intended to remain aligned with the affordable, open-source, community based philosophy of the greater Fab@Home effort.

The Model 1 platform is an additive process rapid prototyping system. As opposed to traditional subtractive machining processes which remove material in order to produce a part, the additive process “builds” a part in a process analogous to printing, layer by layer. The Fab@Home is capable of printing two distinct materials in a single setup. These materials are dispensed through disposable plastic nozzles via positive displacement. A motor turns a lead screw to advance a plunger, which pushes material out of the nozzle. The ability to print two materials -- a “build” material and a “support” material -- allows the system to produce parts with overhanging features. As the part is being built up layer by layer, the support material is deposited in

Copyright © 2008 Rochester Institute of Technology

Proceedings of the Multi-Disciplinary Senior Design ConferencePage 2

order to hold up any overhanging part feature that is not directly supported by the layers of the part already deposited. When the build is complete, the support material is removed by dissolving or other mechanical means, leaving only the finished part. This support system makes the geometry of finished parts virtually limitless.

The scope of the FabAmerica project aimed at expanding upon the multi-material capability of the Fab@Home platform, though with a different end goal. Instead of a build and support material, the finished product would deposit controlled ratios of pre-dyed epoxy resins in order to facilitate on-the-fly color control. Color printing is intended to serve as a vehicle to demonstrate the ability to deposit specific ratios of different nano inks on demand. This has vast and powerful implications for the future of prototyping and additive manufacturing. Spun out to its ultimate conclusion, the ability to print multiple nano inks in specific ratios would enable additive manufacturing machines to produce fully functioning electromechanical devices (i.e. integrated circuit boards, electronics, or even functioning electrochemical cells.) This process could also support a host of other technological advances requiring the development of advanced composite materials, including advancements in energy technologies i.e. fuel cells.

FabAmerica represents the first step in the development of this new technology at RIT, and will surely be continued in subsequent senior design projects. Emphasis here was placed on delivering a low-cost system in keeping with the open-source Fab@Home philosophy, with potential for further development precision and controls.

PROCESS AND METHODOLOGY

Material Delivery Subsystem Design

The material delivery system is responsible for all aspects of handling the resin. This includes storing it in reservoirs, dispensing and mixing controlled quantities of the pre-dyed resin, and depositing resin onto the work part. This subsystem replaces the entire print carriage assembly from the Model 1.

Lee Microvalves

Through a search of commercial products available for micro-fluid dispensing, a suitable candidate was found. The Lee Company makes

micro-fluid dispending solenoid valves (see Figure 1). The valves are simple solenoid valves that can turn a flow on or off, but they can do this very rapidly and on a very small scale. They can operate at frequencies of 1000 Hz or more and have the ability to dispense liquids on the scale of nanoliters. Due to their capabilities and the fact that they are relatively inexpensive, these were chosen to be the best solution to our problem.

Figure 1: Lee microvalve

The use of valves for material control necessitated the use of pressurized air for fluid actuation. The existing positive displacement method employed by Model 1 was not appropriate for use with the valves. The material properties and varying force from the plunger caused erratic dispensing during testing. Microvalves provided an alternative that did not display these types of problem. Constant pressure is essential in a valve system; it is vital for achieving reliable, repeatable and controlled dispensing of material through the valves. Shop air is fed into a pressure regulator to achieve the desired pressure (15 psig), which is then fed into the top of the material reservoirs via specialized fittings designed for applications such as this.

Dispensing

The Lee Company also makes nozzles for micro-fluid dispensing, but it was found that disposable tips made for manual adhesive applications are readily available in sizes comparable to the Lee nozzles. These tips are far cheaper, while still producing similar results to the Lee nozzles. The system was designed to dispense resin in a small bead similar to that of dispensing silicone caulk, only on a much smaller scale. This was achieved by keeping the distance from the tip to the work surface very small (.005”-.010”) and by continuously moving the dispensing tip while resin is flowing out of it.

Initial Testing

Initial testing of the Lee valve was conducted by passing resin through one of the valves plumbed together with .04” ID silicone tubing and Luer Lok fittings to carry the resin and outputting through a 27

Project P10551

Proceedings of the Multi-Disciplinary Senior Design ConferencePage 3

gauge disposable dispensing tip (see Figure 2). This configuration was a good representation of the final assembled manifold since it had similar head loss and internal volume. The fluid was driven pneumatically with a pressure of 15 psig. The Lee valve was driven at a number of frequencies at 50% duty cycle. The valve was run until 10 “drops” of resin were dispensed, and the amount of time it took to do this was measured. The mass of the dispensed fluid was measured and the volumetric flow rate was calculated. Results of this test are shown in Table 1. In the case of this test, a “drop” was defined to be the amount of resin produced when the force of gravity was able to overcome the cohesive forces between the resin and the tip and fall freely from the dispensing tip. Also, in the case of this test, a “droplet” was defined to be the amount of resin produced by one cycle of the valve opening and closing. Due to the very small droplet sizes produced by this test, it was determined that if different colored droplets of such small volumes were placed one next to another, this would result in the illusion of homogenous color to the naked eye - similar to the way that inkjet printing achieves color reproduction (that is, by depositing very small points of distinct colors very close to one another.) This would fulfill the objective of color printing. Because of this, a manifold was designed to attempt to create a queue of different colored droplets of resin that would then be laid down on the surface in a single bead.

Figure 2: Initial test setup

Table 1: Initial Testing Results

Manifold Version I

The first manifold design was configured in a “plus” shape, consisting of four channels: one for each of three colored resins to come in and one for the queue of droplets to exit the manifold (see Figure 3). The exit channel ends in a nipple, which the disposable tips mount onto via a press fit. The design intent was that one valve would open at a time, pushing a droplet of resin down into the exit channel. After the first valve closed, the next valve would open pushing a new droplet down and pushing the first droplet further down the channel. This design was fabricated out of clear acrylic using a CNC milling machine and was tested using the method described above with two different colored resins. Results of this test showed that the resin was far too viscous for this method to work. The manifold simply created a stream of resin consisting of two very distinct streams of different colored resin with very little mixing.

Figure 3: Manifold Version I design (1.5” x 1.5” x .5” acrylic)

Manifold Version II

Due to the failure of the first manifold, it was determined that another feasible alternative for producing color was to mix the resin on the fly as it was being dispensed. Since active mixing seemed to be too complicated to implement in the time remaining, it was determined that a passive, static mixer would be the best solution. A new manifold was designed so that a number of different mixers could be tested. This new design consists of a block with 3 holes drilled perpendicular to the surface for the valves to mount and another hole perpendicular to the surface for the fluid to exit through. Again, the exit channel ends in a nipple, which the disposable tips mount onto via a press fit. This design has a removable mixing plate containing channels to join the fluid streams in a static mixing process. The two mating surfaces are very flat so that a good seal could

Copyright © 2010 Rochester Institute of Technology

Density (g/

cm^3)

# of Drops

Mass (g)

Drop Volume

(nL)

Duty Cycle (%)

Frequency (Hz)

Valve Time Open (sec)

Sec/ Drop

Cycles/ Drop

Droplet Volume

(nL)

1.13 10 .0585 5177 50 20 .025 5 100 51.771.13 10 .0585 5177 50 40 .0125 5 200 25.881.13 10 .0585 5177 50 80 .0062 5 400 12.941.13 10 .0585 5177 50 100 .005 5 500 10.351.13 10 .0585 5177 50 150 .0033 5 750 6.90

Proceedings of the Multi-Disciplinary Senior Design ConferencePage 4

be achieved and no leaking would occur. This back plate can be relatively easily removed and replaced with different mixer designs. The Lee valves are still used in this design, although now their only role is to control the volumetric flow, rather than produce individual droplets. See Figure 4 and 4a for a full drawing of the Manifold Version II.

Figure 4: Manifold Version II design (1.5” x 1.5” x .5” acrylic)

Figure 4a: Assembled print carriage

2D Static Mixers

Obviously the best approach to static mixing would be the crossing helix designs found in static mixers such as those used for two part epoxies (see Figure 5.) However, due to the fact that no commercial mixers of this type could be found that were small enough and due to fabrication limitations, this design could not be used. The best machine to attempt to make small features required for this fabrication was determined to be a laser engraver. With this machine, a piece of acrylic would be cut to a certain depth or all the way through following a 2D pattern. Since the laser engraver was limited to producing planar designs, several were created based

on the work of Bhagat 2007, Hsieh 2008 and Gan 2007. These mixers can be seen in Figure 6. Since the laser engraver melts the material during cutting, the back plates needed to be sanded down with high grit sandpaper to produce a surface flat enough to ensure an air tight seal. These designs were tested using the newly designed and fabricated manifold and two different colored resins. All produced the same result; none of them mixed the colored resin more than just joining the streams together, as in the previous manifold.

Figure 5: Two part epoxy mixing nozzle

Figure 6: 2D Static Mixer Designs (each square is 1.5” x 1.5” x .25” acrylic) (blue is etched 1/16”

deep and black are cut through)

3D Static Mixers

Since the device operates at such low Reynolds numbers and since there was such a small diffusion rate of the dye through the resin, a planar static mixer did not achieve the desired level of mixing. As a result, a more three dimensional approach was attempted. In these three-dimensional mixers, rather than attempting to cause turbulence in the flow, the objective would be to divide the stream into smaller streams, cross the streamlines and join the flow

Project P10551

Proceedings of the Multi-Disciplinary Senior Design ConferencePage 5

again. This would be repeated as many times as possible, so that once the streams were small enough, diffusion would complete the mixing process. Several designs were created to try to implement this, although limitations in the feature size that can be produced with the laser engraver proved to be a limiting factor. The first 3D static mixer (seen in Figures 7A and 7B) was fabricated from two pieces of .25” thick acrylic. It was tested and proved only slightly better than previous designs. This is hypothesized to be due to the fact that the head loss of being sent all the way through the first piece of acrylic, and through the channels in the second, was so much greater than the head loss of the fluid flowing through only the channels in the first piece, that very little resin actually entered the second piece of acrylic. A second design was created to attempt to alleviate this problem (see Figures 8A and 8B). In the second design, all of the fluid paths travel through the first layer and into the second, creating comparable head loss in every stream. This design was cut, but due to limitations with the laser engraver, did not function.

Figure 7A: First 3D manifold design (blue is etched 1/16” deep and black are cut through)

Figure 7B: First 3D manifold concept (blue is etched 1/16” deep and black are cut through)

Figure 8A: Second 3D manifold design (blue is etched 1/16” deep and black are cut through)

Figure 8B: Second 3D manifold concept (blue is etched 1/16” deep and black are cut through)

Controls

The Lee Microvalves are driven by a valve controller/driver manufactured by Lee. Two driver circuits were obtained, but as a result of a long lead time, longer than what the academic quarter would allow, a third was not able to be obtained. As a result a replica would need to be fabricated to drive the third valve.

One of the existing controllers was taken apart and all the part numbers looked up and matching/acceptable substitute components were purchased. The 555 timer, potentiometer and junction block were then unsoldered from the board in order to see all of the electrical traces. After the mapping out of the existing controller a schematic was drawn up so that the circuit could be modeled on a bread board. After verifying that the circuit worked and that all substitute components function as expected, it is now ready for the final stage.

The circuit was drawn up in Express PCB in order to begin building a custom etched circuit board. PCB parts (two double sided copper boards and acid solution, transfer papers were ordered separately) were purchased from RadioShack to build the replica driver. The drawing from Express PCB was then printed out via a laser printer onto the transfer papers

Copyright © 2010 Rochester Institute of Technology

Proceedings of the Multi-Disciplinary Senior Design ConferencePage 6

and ironed onto the copper board. The board is then placed into the acid solution, which must not be put in a metal container, and left until only the traces are left. The board is then pulled out and rinsed of with water. After the board is rinsed and dried the holes for the components are drilled. Once all the holes were drilled the components can be put in. After soldering of the components is complete and all necessary jumper wires are installed the board is ready for testing and if it passes, use.

The signal that is sent to the Lee controller is a 5V signal. When the 5V is on, the valve is open and when it is grounded, the valve is closed. To control this, a BASIC Stamp was programmed to send a signal to the three valves. In this method, only one valve is open at a time. Jumper wires attached to inputs on the Stamp control which valve fires and the length of time it is open for. Unfortunately, due to constraints with the valves, this Stamp was never used during printing and a signal generator was used in place of Stamp.

Curing Subsystem DesignThe expressed customer requirement that the

system print photo-curable epoxy resins necessitated the inclusion of a curing mechanism in the overall design. Because the microvalves used in the material delivery subsystem consumed a large amount of available funds, the overriding consideration in the curing subsystem design was cost. The open-source Fab@Home philosophy also required that ease of construction and implementation be addressed by the subsystem design. In light of these considerations, the components of the curing subsystem are inexpensive and widely available, construction requires no specialized tools, and implementation of the subsystem requires only minimal modification of the existing Model 1 platform.

The light source for the curing subsystem consists of two (2) six-watt fluorescent lighting fixtures, with UV-A black light bulbs installed in them. The fixtures are mounted within a simple aluminum frame using double-sided tape. The frame mounts to the existing Fab@Home print carriage using screws already present on the machine, making implementation very easy for other community members interested in using the design.

Figure 9: A SolidWorks representation of the Curing Subsystem

The peak output of the AV-A bulbs occurs at a wavelength of 352nm – extremely close to the 355nm that the WaterShed XC 11122 resin is design for, resulting in relatively short cure times. The design also attempts to reduce cure times by minimizing the effects of the inverse square law or light, which states that the intensity of energy received from a light source falls off by the square of the distance to that source. As a result, the bulbs were designed to be as close to the work part as possible. Further, reflectors were used in order to re-direct some of the wasted light energy back toward the work part.

Calculations were carried out in order to assess the feasibility of the curing subsystem, and to gain an understanding of the cure times that could be expected. A brief outline of the equations used in these calculations is given below:

The Inverse Square Law:

I 2=I 1∗r 1

2

r22

Where: I 2=irradiance at the work part [ mWcm2 ]

I 1=irradiance at the bulbsurface [ mWcm2 ]

r2=radius at the bulb surface [ cm ]r2=radius at the work part [ cm ]

Estimated surface cure time:

SurfaceCure Time=Ec

I 2∗(number of bulbs )

Where : E c=critical exposure of the resin [ mJcm2 ]

Estimated cure time for a depth of 0.01 inches:0.01 cure depth time = {{E} rsub {10}} over {left ({I} rsub {2} *number of bulbs right )

Project P10551

Proceedings of the Multi-Disciplinary Senior Design ConferencePage 7

Where : E10=exposure required for 0.01 c ure depth left [{mJ} over {{cm} ^ {2}} right

Substituting the necessary values for the proposed curing subsystem design, the time required to cure the surface of a bead of resin was estimated to be 8.3 seconds. The time required for a cure depth of 0.01 inches was estimated to be 39.1 seconds. These cure times seemed very promising for the overall system.

SYSTEM FUNCTIONALITY

Once fully constructed the system was capable of printing in a layer by layer fashion. The open source Fab@Home software was designed to import STL files and execute a slicing process in order to establish print coordinates for each layer. This software serves as a means for motion control for printing. The valves are controlled through an outside power supply. Printing parameters can be changed to alter the properties of the part. For example, the thickness results from the amount of material being deposited on each layer. Controlling this is dependent on the slew speed of the deposition nozzle, the step over command size, and deposition rate of the valves; varying these parameters as shown in Table 2 results in an average layer thickness of .24mm. After deposition, the material is ready to be cured.

Curing takes place by pausing the print cycle and positioning the lights overhead as close as possible to the surface. The lights are turned on for 240 seconds and the layer is allowed to cure enough to facilitate a platform to print the next layer on. Once one layer is cured to a sufficient extent to support the next layer, the table will index down a step to begin printing the next layer. This process is done until all the desired layers are printed. Following the printing of all desired layers, the curing system is left on until the whole entire part is fully cured.

Figure 10: a SolidWorks representation of the Material Delivery and Curing Subsystems mounted

within the Fab@Home system.

RESULTS AND DISCUSSION

Two separate tests were performed to evaluate the system. The first was a single layer color printing test and the second was a multiple layer single color test.

In the first test, seven 1” x 1” x 1 layer squares made up of different ratios of the three different colors, pink, blue and yellow, were printed. The squares will be as follows: pink, blue, yellow, pink/blue, blue/yellow, yellow/pink, and pink/yellow/blue. As it turned out, it was not possible to control the Lee valves with the BASIC Stamp due to the startup cycle required when using the valves. Instead a signal generator was used to control the valves. The printed squares can be seen if Figure 11.

Figure 11: Color printed squares

During the test, several problems occurred including air in the manifold leading to residual pressure, leakage at the valve barbed fittings, inconsistent operation between different valves and excessive head loss leading to changing of the dispensing tip.

Copyright © 2010 Rochester Institute of Technology

Proceedings of the Multi-Disciplinary Senior Design ConferencePage 8

The second test involved printing multiple layers of a single color material on top of one another. To accomplish this, a print cycle was started, one layer was deposited, the print cycle was paused, the layer was cured enough to print another layer on top and the printing was resumed following these steps for each layer. As a result of this test, a z increment between layers was able to be determined.

Both of these tests lead to the determination of printing parameters for the system. These can be seen in Table 2.

Print speed: 10 mm/s (maximum speed)

X/Y step over: .603 mmDriving air

pressure:15 psi

Needle size: 22 gaugeDuty cycle: 15%Z step: .24 mm

Table 2: Determined printing parameters

CONCLUSIONSThe subsystems were constructed and tested

in multiple printing scenarios. The final design is capable of depositing material and then successfully curing it. Although the system can print multiple colors at one time, improvements can still be made to ensure complete mixing. The final project serves as a solid platform that can be used for further research and development.

FUTURE WORKIn future Senior Design projects the goal

could be to develop a mixing system that is capable of mixing a homogeneous solution. Another static mixer with a more complicated design could be use. A dynamic mixing system would incorporate a movable mixing element that could mix without contaminating the print medium. Another necessary revision is the modification of the control software and firmware. The machine should be able to control all aspects of its operation including valve functionality.

ACKNOWLEDGMENTS

The team would like to express its sincerest gratitude to those who have made invaluable contributions to

this project. Thanks to the advisor, Dr. Denis Cormier, for his guidance and support. Additionally, the team would like to express thanks to all individuals who provided prompt and relevant assistance, as well as consultative advice when necessary, including Gerry Garavuso, John Bonzo and the RIT Brinkman Lab, the RIT Aero Club, and of course the Fab@Home community.

REFERENCES

[1] Bhagat, A.A.S. (Dept. of Electr. & Comput. Eng., Univ. of Cincinnati, OH, USA); Peterson, E.T.K.; Papautsky, I.; A passive planar micromixer with obstructions for mixing at low Reynolds numbers; Source: Journal of Micromechanics and Microengineering, v 17, n 5, p 1017-24, May 2007

[2] Hsieh, Shou-Shing (Department of Mechanical and Electro-Mechanical Engineering, National Sun Yat-Sen University, Kaohsiung, 80424, Taiwan); Huang, Yi-Cheng; Passive mixing in micro-channels with geometric variations through νpIV and νlIF measurements; Source: Journal of Micromechanics and Microengineering, v 18, n 6, June 1, 2008

[3] Gan, Hiong Yap (School of Mechanical and Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore); Lam, Yee Cheong; Nguyen, Nam Trung; Tam, Kam Chiu; Yang, Chun; Efficient mixing of viscoelastic fluids in a microchannel at low Reynolds number; Source: Microfluidics and Nanofluidics, v 3, n 1, p 101-108, February 2007

[4] http://fabathome.org/wiki/index.php/Fab%40Home:Overview

Project P10551