Faster Stereolithography

[Enthalpy]

Hello everybody!

3D printing spreads out as a means to obtain parts without the shape, production volume, delays constraints of machining:
https://en.wikipedia.org/wiki/3D_printing
Among several processes, the older stereolithography still makes the most detailed parts
https://en.wikipedia.org/wiki/Stereolithography
and here I propose methods to make it faster (or less slow).

Stereolithography produces a part as successive layers from a bath of a prepolymer that hardens by light. The part sinks by a small amount like 0.1mm, a wiper defines an accurate thickness of liquid, and UV light hardens it where the solid shall be, by mirrors steering the beam of a laser.

Successive layers take time. Acousto-optic beam deflectors would be faster than mirrors but the He-Cd laser limits the speed anyway, with optical power and efficiency like 100mW and 0.01%. 325nm wavelength is supposedly needed to stop the light within a part's layer, and continuous wave avoids material ablation, leaving little choice among lasers.

----------

A first proposal irradiates the part with a lamp and a mask instead of a steered laser beam. The many masks - one per layer - must be produced first, speaking against rapid and cheap prototypes. Then, the exposure needs no laser, and >kW lamps, 10-40% efficient exist at 308nm (XeCl), 254nm (lp-Hg) and 365nm (Led and hp-Hg), to produce quickly each layer. A shutter helps with slow lamps.

A polymer film can hold the successive masks. Fluorinated polymers offer UV resistance and transparency; some are decently strong. Moving the film between two rolls changes the mask; the accurate position needs active measure and adjustment.

The mask can be lowered close to the future layer after the wiper passes, needing no extra optics. 2mm above the resin, diffraction by a 100µm wide line adds only 2*10µm, while a D=20mm lamp at 1m adds 2*20µm and outputs tens of watts.



Or an optics projects an image of the mask on the future layer and the masks stay permanently at stable and clean distance from the resin. Semiconductor lithography can provide inspiration but the scale is easier here. Diffraction isn't a limit, and the lamp can be wide hence powerful. The masks can be smaller (or bigger) than the part with proper optics. The masks could be reflective instead of transparent, with the light source below.



Up to now, 3D printing doesn't spring to mind for series production, but some usual functions do need fine intricate shapes: heat exchangers, filters, batteries, catalyst holders, injectors, chemical microreactors and more - as the final polymer part or as a support for a metal layer.

Marc Schaefer, aka Enthalpy

[Danijel Gorupec]

I am interested if you investigated if it could be possible to use LCD screen (or some other screening technology) as a mask?

[Enthalpy]

To go faster but keep the rapid prototype ability, the second proposal sweeps an array of switchable sources or faders over the new part layer. It sounds logical - though not mandatory - to sweep the array together with the wiper and have the array cover the machine's width. Then, the movement resembles much a flat bed scanner.

 

One mature candidate as a fader array are Mems, micromachined actuators. Some can route an external light source like the ones developed long ago for telephone router; then a common lamp could feed many faders, for instance through a light guide.

Liquid Crystal Displays too can align many small pixels, reflective or transmissive. I just doubt that they survive the UV.

UV Led or laser diodes are the proper solution, but depending on the necessary wavelength, they may already exist or not quite. Led for 385nm and 365nm are excellent in 2015, for instance Nichia
http://www.nichia.co.jp/en/product/uvled.html
offers almost 50% power efficiency and 3.5W light in 45mm², so that an array of 30 chips of hundred 50µm*50µm Leds side by side would emit 1.5W. Led provide up to now only 1mW at shorter waves (260nm) but this will improve, and lasers follow Led few years behind.

Whatever the array's sources or faders, they can have lenses to maximize the light and keep some distance to the prepolymer bath. One lens per chip seems logical, and if it widens the image a bit, it permits clearance between the chips. Putting the chips on several rows also eases the distance requirement.

All sources or faders arrays can have several rows where several active elements illuminate a point of the part successively. This increases much the optical power (for instance *20 with as many Led rows on a chip) hence speed and gives redundancy. If the active elements are smaller than the part's definition, software can correct movement imperfections.

Marc Schaefer, aka Enthalpy

[Enthalpy]

[...] use LCD screen [...] as a mask?

 

Thanks for your interest!

 

You're right, an Lcd panel whose image is projected on the future layer looks more interesting than sweeping a 1D-Lcd. Screens offer only 96 pixels per inch for instance, but even if the optics reduces the image to 100µm per pixel, a resolution of 1920*1200 pixels permits 192mm*120mm. Well-aligned optics lets use several screens.

 

But would the weak liquid crystal molecules resist UV light? This is worth a trial, with a hp-Hg lamp (emits 365nm, 385nm and 395nm) or even a UVB lamp (its phosphors emit at 310nm).

 

My old patent EP0564012 describes an electrochemical cell supposed to become opaque with current

http://www.freepatentsonline.com/EP0564012.html

http://www.freepatentsonline.com/EP0564012.pdf

being anorganic, it must resist UV better, but I didn't even begin to try it, and whether it's reversible many times is doubtful.

 

----------

 

I've already polymerized Mma to Pmma using normal fluorescent lamps, so 365nm must suffice though it wasn't very fast. This would enable existing Led as well, as a transition. Maybe some materials don't stop the light within one layer, but opaque fillers solve that. For instance, very short carbon fibres, which also improve the strength of the polymer.

 

Marc Schaefer, aka Enthalpy