SLA USES IN MOLDING FOR RAPID PROTOTYPING

Jim Mishek
Vista Technologies LLC
White Bear Lake, MN

7 September 1997

Introduction
As recent as three years ago the experts were saying that injection molds made of stereolithograhy (SLA) produced materials would never hold up. Fortunately, the machine owners did not listen to the experts. We are no longer debating over if it will work. Instead the race is, who can develop molds that produce a wide variety of part designs in a wide variety of materials for the greatest number of shots in the shortest period of time.

Wide variety of part designs, wide variety of materials, greatest number of shots, shortest period of time, and we should add lowest in price. Because this is a tough combination of parameters, these goals seem inherently paradoxical. To make it easier to grasp, we have grouped these parameters into a three dimension matrix, see figure 1. The Y-axis is for the variety of materials, the X-axis is for the number of shots obtained, and the Z-axis is for the variety of design criteria.

Figure 1

Before we go too far into the experimentation, it might be good to back up and to examine why we are even interested in these processes.

Background
Fit, Form, and Function are the three key words defining the need for prototyping. With the advent of the various rapid prototyping machines, particularly SLA, the need for Fit and Form has been fulfilled. The problem is Function. In the proper application, we have succeeded in building entire functioning models using the SLA material. But are the properties the same as the materials to be used in the final product? NO.

Due to this dilemma came the advent of new technologies; the Selected Laser Sintering (SLS) and the Fused Deposition Modeling (FDM). With both of these technologies, you are able to assimilate the approximate properties of the approximate materials that are to be used. These methods come closer, but they still are not the real thing.

The search then reverted to trying to find some way to quickly manufacture molds which could mold the real thing. The use of short run aluminum tooling was examined, as well as the development of KelTool and the further development of the SLS technology, enabling the manufacturing of metal tooling. Even in these cases you are generally looking at a two to three week turn around and then you still have to do additional machining which adds to the time.

The search continued. As mentioned earlier, despite the sage advise from the experts, experimentation was carried out using solid blocks of the SLA material as the injection molds themselves and it did work. This technique is referred to as a solid ACES insert. ACES is an acronym for "Accurate Clear Epoxy Solid", which is a build style for the SLA machines.

The major problems encountered with this technique were the amount of time it took to make a solid mold, easily 30 hours, and the lack of heat transfer.¹ The length of the build is height and volume dependent.

The lack of heat transfer greatly reduces the variety of design parameters and reduces the life of the mold faces due to the softness of the SLA material at elevated temperatures above 60^oC/140^oF, and the material property changes at its glass transition temperature of 70^oC/158^oF. At these elevated temperatures you can have mold degradation through erosion or from spalling, where the mold material sticks to the plastic and spalls off at ejection.² Also, if you can control the mold temperature more efficiently, you may be able to reduce mold pressures which will increase mold life.³

The slow heat transfer, 300 times slower than tool steel, also means longer cooling times for the parts before ejection. Periods of five minutes were not unusual.

The next step was to go to a hollowed out SLA shell instead of a solid block, these are called an ACES shell. In this method a mold shell is built on the SLA, then copper tubing is inserted into the hollowed out area behind the mold face, then the entire shell is filled with aluminum filled epoxy. This is the point from which we began our experimentation to fill in the matrix.

Experimentation
Experiment 1: These trials involved using a very simple shallow mold for a golf ball mark repair tool, seeing which materials we could successfully run, and what tolerancing and longevity we could get from the mold. The mold was made from an ACES shell with a face thickness of 1.5mm/.060" and side wall thickness of 2.5mm/.100", see figures 2 & 3.

Figure 2: The ACES shell mold unfilled

Figure 3: The ACES mold completed with molded parts

Experiment 2: The next tests were to study the design parameters. We developed molds using low melting alloy as the fill material rather than epoxy, then we combined this with molds that featured different design parameters such as a thin, deep web, another tool that had very thin walls was designed into a family mold, and a similar golf tool as in Experiment 1.

Experiment 1 was an interesting test, we did everything wrong and still got good results. When completing the SLA mold shell, we allowed the shell to sit in normal room conditions for two weeks before we tried to fill it. During this time, the mold walls warped, approximately 3mm/.125", Requiring us to do extra clean up before we could mount these shells into the mold base. Then, as we tried to put the holes for the water lines in the shell wall by drilling, we cracked the mold. This required us to glue the shell back together using the same resin and a high intensity UV light.

The SLA material is very difficult to machine in the shelled condition because of the vibrations. However, once the shell is filled the vibrations are dampened, it is easy to machine. We performed milling, drilling, fly cutting, and grinding with no complications.

The experimentation was run in a Cincinnati Milacron, 33 ton, Vista Sentry injection molding machine, see Figures 4 & 5. In all of these tests the mold was short shot to begin with and then adjusted accordingly until good parts were produced. The shots took more pressure at the beginning until the mold warmed up.

Figure 4: The mold mounted into the machine	Figure 5: Measuring the temperature of the mold face

The table below shows the running parameters. The cooling was sufficient enough to stabilize the temperature of the mold faces and to keep the temperature of the mold faces below 57^oC/135^oF. With this style mold the cooling times were approximately twice that of a steel mold. The polycarbonate required the highest pressure, at one time reaching 15,390 psi, yet the mold held up. The proper backing is very important, as you will see in the second experiment.

Experiment #1 Molding Results

Material	Avg. Press.	Press Temp	Water Temp	Max Mold Temp	Cycle Time	# Pcs
Polypropylene	5,280 psi	220oC/425oF	13oC/55oF	48oC/120oF	120 sec	25
ABS	5,470 psi	220oC/425oF	13oC/55oF	50oC/125oF	100 sec	25
Cycoloy C2950, ABS/PC	7,450 psi	265oC/510oF	43oC/110oF	57oC/135oF	80 sec	14
Polycarbonate	11,110 psi	293oC/560oF	43oC/110oF	57oC/135oF	100 sec	10
Nylon 8018, 14% glass	7,530 psi	293oC/560oF	43oC/110oF	54oC/130oF	100 sec	3

As we ran each test, we numbered each part. Later each part was measured and inspected, see Figures 6 & 7. The chart below shows the results.

Experiment #1 Dimensional Stability of Parts

Material	Avg Meas A	Avg Meas B	Avg Meas C	Avg Meas D
Polypropylene	.698 +/- .001	.180 +/- .001	.238 +/- .002	.241 +/- .001
ABS	.709 +/- .001	.189 +/- .001	.254 +/- .001	.251 +/- .001
Cycoloy C2950, ABS/PC	.705 +/- .001	.185 +/- .001	.249 +/- .002	.246 +/- .002
Polycarbonate	.705 +/- .001	.184 +/- .003	.248 +/- .003	.246 +/- .002
Nylon 8018, 14% glass	.703 +/- .001	.178 +/- .001	.236 +/- .003	.239 +/- .003

As you can see, the dimensional stability is quite good. When examining the parts, you will see that the mold had started to erode. This happened when we started to shoot the polycarbonate. Whether this is just from the conditions of molding polycarbonate or if it was a cumulative effect of the previous shots, we can not conclude at this time.

The eventual mode of failure was twofold. There was the erosion of the faces and there was a blow out through the edge of the mold near the tips of the tines of the tool.

The second cause of failure points out that good mold design is still a must. This is proven in the second experiment as well. This mold should have had two small ejectors near the tips of the tines to aid in the ejection as well as to serve as vents. The mold face shows the effect of the hot gases being forced into the tines and then having to find its way out - eventually the thin wall gave way. On the mold faces you are able to see the traces of the gases escaping through the parting lines.

Figure 7: This diagram shows the location of the measurements.	Figure 8: The parts were identified for future inspection. Notice the flashing near the tines on the last parts molded.

The conclusion from these tests is: the ACES shell mold will accept a wide variety of materials with some higher heats and pressures and will produce good parts, the quantity appears to vary dependant on the material, more experimentation for life expectancy needs to be done for each material.

We decided that before we chased down this path, we wanted to try to perfect the mold design aspects first. We believe if we can cool the part quicker, we can obtain longer life (due to the less time that the mold faces spend in their vulnerable temperature range).

Therefore, Experiment 2 concentrated on trying to get the low melting metal to work as a backing material. The low melting metal (lmm) is a mixture of lead and bismuth. Depending on the alloying percentages, you can have an alloy which melts between 50^oC/120^oF and in excess of 100^oC/210^oF.

The lmm was much tougher to work with than the epoxy. If we were not careful we could overheat the lmm and then when we tried to pour the lmm into the mold face, the face warped and we lost dimensional integrity. We did successfully mount the faces into the mold bases, see figures 8 & 9.

Unfortunately, we never had any totally successful shots using this technique. We had mold face spalling and we had mold face deformation. See figure 10. The fascinating problem with the lmm is that it does not bond to the SLA material which makes up the mold face. When the face tried to move due to thermal forces, there was nothing there to hold it. The mold with the deep web failed when the face of the web pocket distorted and froze the part into the mold. The lmm also proved not to be an adequate backing material. Our mold faces were pushed back .5mm/.020", which ruined our shut offs and created excessive flashing.

Figure 8: This is the face of the watch mold.	Figure 9: The cavity of the golf repair tool mounted into the mold base.

There were minor successes in these failures. The detail we managed to achieve in the thin wall sections was excellent and it proved once again how important mold design and mold preparation is for success. The one thin watch part would have been successful except for the lack of venting, gas built up in the center of the part. This same part began to show erosion in one of the cored areas because enough care was not taken in the mold preparation. This area still had stair stepping, so when the parts were ejected they eroded the mold face. See Figures 11-13.

We did achieve our goal of fast cooling. The low melting metal did allow us to run parts at similar speeds to metal molds, while keeping the mold face temperatures at near our cooling water temperature.

Conclusion
The ACES shell design is an adequate design for producing parts for a wide variety of materials. Future testing should be done using the aluminum filled epoxy rather than the low melting metal. However, the goal should still be to get the heat away from the SLA mold faces quicker. Additional work should be pursued on coatings as well. The most recent tests published have shown good results with nickel coatings when running ABS, but minuscule gains when running polycarbonate.⁴

A final note that keeps reappearing in all of our tests, mold design is of a premium. The fact that you can not run 50 parts to fine tune your mold with a SLA mold, requires the mold designer to over design the mold for vents and ejection. It is also important to develop a range of operating parameters so the press operator can produce successful parts quickly.

Figure 10: Notice the deformation in the center of the part due to a break out, also notice the pieces of spall on the thumb indent and by the gate.	Figure 11: These parts had thin webs, over an inch in depth.
Figure 12: Note the thin wall sections and the fine detail. This came from a family mold.	Figure 13: the molded watch case and the original SLA prototypes.

Acknowledgments
Vista Technologies wishes to thank Imperial Custom Mold for their design efforts, molding expertise, and press availability. We, also wish to thank Spectrum Mold for their expert mold design and molding expertise along with Donnelly Custom Manufacturing for their molding expertise and machine availability.

References
1. Jacobs, Paul, 1996, Recent Advances in Rapid Tooling From Stereolithography, 3D Systems, pp. 4-7.
2. Decelles, Paul, 1996, Direct AIM Prototype Tooling Procedure Guide, 3D Systems, p. 9.
3. Dell’Arciprete, John, 1997, Cavity Pressure Studies for Stereolithography Produced Tooling, IPI, University of Massachusetts Lowell, p. 1.
4. Burns, David T., 1997, Analysis of Metal Coating Effects on Stereolithography Tooling for Injection Molding, IPI, University of Massachusetts Lowell, pp. 1, 10-12.