Concept multimaterial laser printer

This page is a development stub. Please enhance this page by adding information, cad files, nice big images, and well structured data!

After doing a lot of reading for the metal deposition head, there are a number of approaches that are interesting and could be used in conjuction with the multimaterial printer. But it is more clear than ever that laser deposition from a sheer performance (in terms of net capability for parts, excluding cost/performance ratio) standpoint, with a very small melt pool under argon is a highly interesting and capable approach.

One problem is that with the multi head approach, you still cannot really print a bearing or other parts in situ. You still need to assemble them later. In the interest of achieving "type 3" replication as opposed to type 2 (complete automation vs. large labor input to assemble parts) I turn to thinking about other approaches.

So in considering what I might at this point consider the most interesting approach for a multimaterial printer, I'm of course not sure and this is all just thinking at this point. There are a lot of different ways and it's not really possible to tell at this stage which would turn out to be better, but this is my current favorite:

• under argon, atmospheric

• An array of lasers which can be modulated at suitably high speed. The power of each laser is whatever gives the degree of precision (with good melt pool control rather than trying to cram high build rates per melt pool in just have larger number of lower power pools) and resolution required (yet to be determined, see todo list). The resolution is determined largely by melt pool size. The smaller the melt pool the less thermal distortion too. But the smaller the melt pool, the smaller the laser which usually means higher cost per watt (at these power levels, which is a couple watts). More research and calculation is needed to determine the build rate for various laser powers, melt pools sizes and materials to make a good decision. Basically smaller melt pool means higher ratio of heat power lost to the bulk of the object vs. gone in to heating the new material so slower build rate per watt. So some compromises there.
• array mounted on xyz bot

• Deposit powder, heat to weld in place then vacuum it up, then really scrub or vibrate or gas-blast the workpiece to get the last bits of powder off. Then add a different powder for the other material.
  • shoudl include a plastic (probably have to be low viscosity so the particles will consolidate well without any pressure), a durable ceramic like silicon carbide, a support material so leaders can be printed (water soluble ceramic sounds good), and metal, maybe more than one metal - copper and steel.

• proof of concept. Maybe only one laser but of the size that was determined appropriate for the laser array. Probably very small. The size of a hand or fist? Makes the high precision desired easier and cheaper - we know we can do it at a larger size for the right price so no need to make it large.
• Hydrostatic bearings, maybe with ferrofluid. Air (argon) or pumped fluid if needed - woudl be expensive for the prototype but can tehmselves be printed so no problem for next gen.
• extremely high precision/accuracy. Probably use encoders on linear bearings, maybe optical. could be a modified digital readout micrometer? Is no force so that helps.
  • hydrodynamic bearings avoid stiction and normal friction, only highly predicable fluid forces.
  • there must be hydrodynamic leadscrew mechanisms? The same principle can be applied easily and it would provide the desired precision. Otherwise ball screw maybe biassed.
  • Somebody in the forum mentioned that they were part of a project to do a positioning stage that got to 10 nm for microfabrication at low cost. Talk to them about precision positioning on a budget. As long as the encoders are fast enough and the laser can be modulated, the main thing is to get some force on the laser head that is stable and predictable and smooth.
  • maybe use the SEM but with light thing to image the object to some degree. Modulate the laser intensity down, move the laser up so that it is focussed right to a point on the print plane, and scan.
• Would be extremely slow with only one laser most likely, but proof of concept. If laser moves at 1 meter per second and has a 10 micron melt pool and 2 micron layer thickness then 20 mil cubic microns per second, 1 cc is 10,000^3 or 10^12 cubic microns, so 2/10^6 CC per second, (3.6*2)/10^3 cc per hour for the outline of the boundaries between material/the outer boundary. However the bulk areas could be done much faster as no precision is needed. 100 micron melt pool, 20 micron layer thickness and 5 m/s speed, 10x10 array of lasers, (5*3.6*2)*10 =50 cc per hour. Pretty slow but enough for a proof of concept. More lasers still might be added or maybe parts could be made in parallel on different printers then consolidated ultrasonically after being precisely positioned over each other (they could have alignment pegs or cones built in to the top/bottom). Obvious problem is that 10x 10 array of lasers would be quite expensive, but the laser in a dvd writer can't cost more than a couple bucks and is a few watts (2 I think).

• a central assumption is that a gas laser could be produced that was suitable to replace the diode laser that is used during prototyping, which would have a much lower cost per watt. The cavity per se without the end cap, including any and all electrodes in the gas chamber and conformal cooling channels, could be printed in a single glass block. The end caps and lenses could be printed roughly and then would be finished with standard mandrel polishing using a printable companion unit and maybe even printed mandrel. The lens form accuracy ultimately does not have to be perfect because we do not need to focus to a diffraction limited spot anyway, so 1 micron dimensional accuracy may suffice. If the experience with the stages allows higher accuracy still that would be good though.
  • ideally a TEA laser or something else with a wavelength that can use materials that are easy to obtain, not the zinc selenide used by CO2 lasers although ultimately maybe it is not that expensive, have to check. Zinc is cheap and selenium isn't that expensive.
    • problem with tea lasers is that they are pulsed. Either has to be qutie high frequency (like a MHz might be good if the melt pool is 10 microns and the speed of movement is 1 m/s) or CW.
  • maybe something like a CO2 laser but higher wavelength.
  • power supply for a gas laser will not be cheap as they are low efficiency, but the device should have little difficulty printing most of a vacuum tube (require removal of support material, sealing and evacuation as well as addition of a volatile material as charge carrier maybe).
• use extremely fine powders, they are not that expensive. The powder needs to be vacuumed off the surface anyway as you print so it can be stored for re use and also does not need much or any cleaning after, should not escape the print chamber. Might need a mechanism to vacuum any stray powder from the print chamber and build tray to prevent escape and prevent mixing of powders (so powders that are vacuumed up do not need to be sorted before re use)

• heating areas with a large depth to width ratio of valleys that laser can reach for heating. Because of the need for vertical separation between printing materials that could pyrolize or otherwise break down at temperatures required for deposition of other materials, if you wanted to print say a thin wall of wax between 2 metal walls the height of the metal walls could block laser light to an undesirable degree. Might put serious limits on the width of such wax (or other non temperature resistant material) walls. This is one argument in favor of longer stand off distances and smaller aperture numbers for the laser optics i.e. large rforcal length to lens diameter ratio, so the angle of the cone of light is small if you see what I mean.

• • What does that do to the precision that the optics need? With galvo mirrors or lenses the laser can come in from the side sometimes, bbut conversely it is forced to come from the side to some degree in most cases. A lot of laser energy might end up getting blocked one way or the other. Fortunately undesired heating of the materials blocking the light unlikely since they will be higher melting point materials due to the processing temp being higher the reason they protrude in the first place? Easy to identify during computer simulation of the build process.

• a similar problem as the metal deposition head printer has when it comes to printing a horizontal or nearly horizontal ledge: the material underneath the ledge has to be resistant to the metal deposition temperatures. An argument in favor of temperature resistant support material. WRT the dimensioning accuracy, there is no machining here as there would be with the metal deposition head so form and surface finish accuracy would likely suffer little or none I think.

• materials that need higher deposition temperatures should probably be deposited first to avoid melting or burning the base of a neighboring wall (burning (or rather pyrolizing or otherwise causing chemical decomposition since there is no oxygen) is an obvious no no, even if the more heat sensitive material is not chemically damaged distortion could occur as 2 liquids of different density in contact try to flow around each other, or diffuse within each other. Diffusion might not be so bad, depends if it causes an undesirable material in the diffusion zone i.e. excessively brittle or something. In most cases it would just glue the materials together I guess, which might be desirable for improved strength.)

• might be good to print a lubricant, print chamber could be at sub zero temp so lubricant is solid, could be one of the support materials too maybe or different

• have to be able to print thin walls of the void/support material and lubricant to achieve close separation between components when desired
• alternative heating methods, rotating array that is mounted on x and z axis for motion, the galvo lenses or mirrors (scan areas shoudl overlap so not all lasers need to be operable)
• ideally the laser array should be designed so that each area on the surface of the object can be heated by more than one laser so not all the lasers need to work

-exolaind the lack of brighnesss from other light sources, maybe there is a plasma one but likely would be used already by commercial dmls units? Check to see if there are patents on. Also less coherent so focussing not as easy but ocudl use mirrors to avoid chromatic abberation. -explain how the gas laser and machine might go, might need liquid crystal modulation cell if the power output cannot otherwise be modulated at the desired speed. - the staircase problem: the layerwise deposition leaves staircase shaped profile but would prefer smooth. Mayb the staircase side can be tilted by careful control of the melt pool/ irradiation and positioning of focal area? move slightly towards the edge of the object to smear the top layer of the desired material. Could also print up a shallow wall of support material (with a support material that melts at a higher temperature than the desired material) more than one print slice tall above the plane where the desired material is, and the sufrace tension of the melt pool at the edge of the wall gives smoothness and roundness, or even if not could selectively melt the sides of the wall in the area between print slice thicknesses to smooth them out. When the desired material is deposited it takes on the form of the support material have to be sure no gas gets trapped in the area but probably would not

• powder deposition done with precise deposition method. Can't really use imprecise method followed by knife edge or roller going over the surface (have to be very precisely made), ultrasonic lubrication to level it etc. because of the vertical height separation between materials that needs to be maintained during the build process. Powder would fill in the valleys.
  • maybe could use knife, roller etc. if could get the top level of thepowder bed to go down precisely after application, by spraying it off electrostaticaly or something. Spray the surface with high voltage electrons? Process would be different for different materials. Residual static charge has to be removed after each cycle though
  • could maybe do printing upside down? Spray the particles at the surface and maybe they will stick lightly in a very thin and reasonably uniform layer.

• focus on precisely getting suitable material where it needs to be for good voxel control. Size of machine, etc is all optional, can be very small but preferrably done with a larger practical machine in mind. Proof of concept only. Could use the parts from a DVD or CD burner in various ways to keep costs down perhaps.

• Need feedback of surface shape and preferrably melt pool temperature and video of it, at least for during the prototyping process so am not working in the dark as much.

• • small microscope on gantry should do? Mount on another xyz bot but does not have to be more precise than the viewing area. Should be able to focus though. Might need piezo actuator or similar. Maybe the lens focusing thing from a CD writer would do? Also the linear bearings and stepper motors. Small project of it's own. Also preferably it does not block the laser, so can look at the melt pool while it is going.

• • Scanning laser microscope thing much like electron microscope could maybe be used with 2 photosensors to scan in the surface. Focus laser do smallest spot can, scan, focusing done with movement in the z axis of the laser probably since have the precision xyz bot, collect light intensity signal and form image like a SEM. 2 sensors for stereo vision. The lasers can be modulated each at a different high frequency and the signal processed to use multiple lasers at once. Would be quite slow to scan in the whole surface, suppose 1 micron focal area, 1 m/s traversal, that would take 100 seconds per square centimeter (10,000 rows, 100 rows per second) per laser. An argument for a larger number of lasers (if the signal processing can handle it)?
    • could use larger number of photosensors if desired. They could be built into the laser array or just arranged around the periphery
    • make all surfaces black where can, done in the dark of course to reduce signal noise.
    • no video possible
    • can be used to align the overlap of the laser focal areas since the lasers scan over the surface of the same shape the offsets can be calculated easily.
    • another reason to use a laser with short wavelength
    • diffraction limited spot would require precise optics though. May not need to be that small though.

• powder removal with vacuuming and gas jets, brushes, ultrasound or sonic vibes, whatever works - complicating factors include adhesion of particles of materials to like or unlike materials, thin films of lubricants or oxidesor adsorbed films of water , abrasion that may be caused by ceramic particles left behind (very fine powders might help as they would be a lot smaller than the lubricant film thickness),

• list of complicating factors for the deposition material physics
  • crystal structure of metals deposited affects surface roughness, the roughness caused by it may be on the order of the print resolution in this case? Small melt pools might help compensate

• • gas bubbles especially argon trapped under the powder for viscous or low density materials. For metal bubbles would mostly float to the top. Probably have to use low viscosity materials.

• • surface tension (cohesion), wetting or lack of wetting (adhesion) of the melt pool material to the nearby material surfaces etc. all confounding factors but can be useful too (e.g. to reduce step profile as mentioned above), maybe do computer simulation to see how things happen at this size scale or can find info in ref materials. Microscope vision system will help too. Also viscosity of the melted material vs. time that heating is applied (especially with regards to plastics which may be much more viscous than molten metals).

• • adhesion of particles of materials to like or unlike materials, thin films of lubricants or oxides or adsorbed films of water or other substances, may be useful too though e.g. when spreading powder layer as mentioned above.
  • thermal expansion and distortion, small melt pools help.
  • phase transition distortion again small pools help
  • what else? probably quite a few things

• melt pool temperature control done how? Open loop with suitable computed settings for the situations expected applied? With diode lasers could monitor electrical characteristics which will change with backscatter. Backscatter may change when surface melts and also with temperature as the reflectivity of the material changes? See methods on comprehensive search page. If large number of melt pools then might preclude some types of sensors.
  • during prototyping, the microscope might have a set of filters that could be changed to allow temperature sensing crudely.

• support strategy
  • Requirements:
    • Support thing walls and shafts against any forces, probably not much but some during powder removal.
    • Act as a "void" voxel in the final part, i.e. provide spacing etc. between moving parts.
    • Might be nice if it could act as a lubricant after printing but there are no lubricants that can stand the temperatures of metal deposition. Maybe one of the materials in the dual material with different temperature compatibilities approach could be a lubricant.
    • be easy to remove somehow
      • melting, dissolution, sublimation, powderized with ultrasonics, pulverized by submersing in a bath of hard beads and vibrating (might not work or cause problems elsewhere)

---

Todo:

• Check out the the size, traversal rates and power inputs for various materials for good melt/fusion pools and decide on the right laser powers etc. for the array.

• need to nail down the precision and accuracy and feature sizes/resolution needed to print hydrodynamic bearings, other journal bearings, leadscrews, and also maybe the parts for a laser and the optics production machine (so read up about spinning mandrel and other common optics production methods)
• hydrodynamic or dynamic leadscrew and nut exist?
  • better to focus research time on the physics involved because precision xyz units are well known.
  • yes they exist and are practical: http://www.dgte.kr/business/hyprostatic/article_lead_screws.pdf.

• make a list of things want to be able to print, their material and resolution, tolerance (including surface finish) requirements to try and make things a little easier. So it is mostly a proof of concept even if it can't print everything (or even any complete objects).
  • a precision stepper motor with bearings of some sort included
  • Hydrostatic bearing surfaces including the hydrostatic journal bearings and the leadscrew and nut
  • Hydrodynamic bearing surfaces including air bearings and bearings that use water or oil as the fluid.
    • Okay, hydrodynamic bearings should be highly polished if bearing large loads, not sure if this can be accomplished with this printing method, might need to remove the parts and process them separately then. What if not bearing large loads? Also the purpose is to remove asperites
  • a precision leadscrew
  • the mandrel tool for a lens making process - bearing, drive coupling and the form thing to attach
  • the vacuum attach cup thing for the mandrel polishing machine
  • the blank pre lens thing before polishing
  • a PCB. Basically very low tolerance.
    • materials: copper and insulating material that can stand 200 C degrees at least over periods of some minutes, and 275 C during soldering, basically. Copper has to not delaminate from the substrate when soldered. Don't bother with other conductors, the metallurgy of soldering won't be as reliable so not commercial quality boards.
    • Could attach components and weld them onto the board right in the machine with a pick and place tool as long as the vertical height between the bottom of the laser array and the top of the print surface is okay. It might be nice to be able to print resistors, inductors which should be possible and practical. Printing capacitors of any size might be less practical but maybe... and small ones should be doable. Most of a vacuum tube would be printable too, though the sealing evacuation and addition of charge carriers would require post processing.
  • turbine blades for air turbine tools as expanders.
  • other expanders with suitable long lasting bearings, roots and gear, and piston expanders in particular.
  • connectors, no strict requirements there on mechanical tolerances, materials should be good like nickel or gold plating on the surfaces, is stainless steel okay? I don't think so.
  • coils for inductors and motors
  • hydraulic pistons.
  • gearboxes and gears.
  • what gets really interesting is when even a few of these things can be combined in to a single object automatically inside the printer as the print process proceeds, especially bearings and gears and coils and populated pcbs
  • Industrial tools of all sorts, particularly important
    • dies and stamping tools - very high strength materials needed like tungsten carbide etc.
    • casts for molding
    • the machines to make rolling element bearings
    • Particularly interesting: equipment for chemical engineering processes. Often needs to be made of specific materials though for chemical compatibility like teflon or whatever. In cases where it is not possible to make with the material, make whatever equipment it takes to make with the material.
  • obviously structural components of all sorts, materials metals mostly precision pretty low.
  • fasteners and fittings
    • might seem like no point because they tend to be cheap but they are often not that cheap, or are hard to obtain. Plus to print a machine that does not require assembly but can be maintained you sort of need them.
    • tolerance for a bolt is like 50 microns not to strict at all.
    • can probably do without rivets.
    • motor couplers and machine screws etc.
  • every part in a car basically. Need some sort of meltable elastomer for the tires, seals of various sorts.
    • Hard wearing and durable materials for the inner part of the cylinder and piston
    • as mentioned high quality bearings important, the bearings in the piston head and crankshaft are subject to special polishing procedures that basically remove asperites. The thing is they are subject to huge loads (like 12,000 pounds peak) and the higher the load the more polished it has to be - the asperites have to not protrude through the lubricant layer basically, the lubricant layer is smaller at high loads so the asperities have to be smaller. Exactly what is the surface finish needed in Ra?
    • many things fall into the materials zone and fall outside of what such a printer could ever do.
    • many of the things like the differential, transmission, engine and valves, brake pads and hydraulics, door locks, are already covered by capabilities mentioned elsewhere the other things mentioned.
    • painting can be done by bonding a thermoplastic mix with pigment to metal like polypropylene with UV additives (basically is an anti rust method, previously mentioned).
    • durable electrical contacts. The contacts in a motor commutator are usually graphite copper composite that might be hard to deposit with melting. The other end of the commutator is just copper alloy though

for much of the above, a much wider range of materials would of course be needed to make a good one, than only 3 like tool steel, silicon carbide and polypropylene... but for the most part they seem to be melt processable so could fit into this scheme. Exceptions for mechanical parts alluded to or which are usually part of the above that can't be melt processed:

• • rubber obviously, but there are probably elastomers that can replace it and are melt processable.
  • maybe some things that require specific crystalline structures but none that I can think of. Maybe nitinol or something. Parts that require hardening will probably be a problem, but those may be replaceable with softer materials that are coated with ceramics or alloys that are very hard and tough without heat treatment, or maybe it would in fact be possible to control the processing to do heat treatment of a surface or bulk material during building, in situ though there would be limitations on nearby material layer's temperature compatibility etc., maybe not.

• • lubricants of course have to be added after probably, after support structures are removed. Also other liquids and fluids like battery electrolyte or hydraulic fluid.
  • natural materials like wood or cotton not usually involved in any of those things.

• desirable but probably not printable directly
  • already sharp drill and milling bits
  • ball bearings and other rolling element bearings, shafts and raceways for them
  • objects which require certain chemically resistant materials
  • e.g. halogen light bulbs, how would you get the halogen in there without post processing? Still you could do most of the work.
  • optics of any sort. Could it not be conceivable that selective melting of a near net shape glass surface not lead to an optical quality surface? Would take special techniques to achieve extra precision but, basically it is conceivable that most of the hardware in this printer could be reused for a device for (slow) optics production. would require some sort of optical encoder that depends on Time of flight or something perhaps, or the capacity to scan in the optic and determine it's shape to a high degree, then keep machining in the places that need correction, and repeat.
  • I think you could make most of the parts for a gas laser, though you might have to have a special apparatus to help you assemble and align them by hand. There are a lot of people who make homemade lasers but they are maybe importing precision in the optics usually so that's not really proof of being able to make one from scratch....

• interesting things you might be able to make or do, but not that important
  • microturbines
  • composite mesoscale antennas etc. for radio waves or sort of like optics for them.
  • clothes (out of synthetic fibers pretty much, cotton etc is out) The fibers are above the resolution of the printer mostly, though some may be only microns wide you could certainly make clothes of a sort that were functional. Might be really slow though as high res printing tends to be slower due to smaller melt pool. Probably better to just be extra sure you can print a machine that which can make the stuff
  • mesostructured materials. The high resolution that is needed for bearings also could be used to print interesting structures like microhoneycombs. Slow though
  • novel things like very small mechanical systems. Again slow.
  • repair worn or damaged objects by building up the appropriate areas. Laser focus area has to be able to reach though. Still the technology and software developed might even be adapted to robot arm based head that could basically print directly on another object, so the object is welded on and fully consolidated with the large part.
  • put ceramic coatings etc. on existing metal objects, plastic anti corrosion layers etc.
  • functionally graded or composite materials
  • weird or high performance things like a model helicopter blade made from ceramic.

• given the issues and demands of the printing process above, identify suitable materials to print with given the things want to print, conductive, tough structural material, an insulator and a good support/void material.

• put together a support structure strategy that lays out the required roles as did for the metal depot head can copy most of it
• Find out about the surface tension etc. other complicating factors as applies to small scales
• when advanced enough ask for comment etc. on forums.

---

• random notes:
  • support materials
    • The problem is processing/deposit/melting temperature incompatibility . Sometimes you would want to print the support material on top of a low temperature material, and sometimes you need to deposit high temperature material on the support material. So either it melts at a low temperature to avoid melting the low temperature stuff, or it melts at a high temperature to not be melted. However it may be acceptable to have a limited amount of melting of the support material and/or the material on which is it deposited, as long as the melting is localized. It is less than ideal though because it messes up the interface between the two, either from diffusion or undesired flow of them.
      • undesired flow may be quite limited due to the small melt pool size and the surface tension and viscosity of the liquids subsequently overwhelming gravity.
        • Fundamental problem is getting the material from the main storage to the desired volume and only that volume/voxel as close as can, so a printing process again but with special requirements.
      • so options (given the other requirements of a support structure strategy):
        • 1) Could have 2 support materials. Low temperature melting but highly temperature resistant material (preferably insulating, reasonably strong (doesn't take much though since this is mostly force free, though there may be some forces during powder removal) and also otherwise compatible with any materials it touches, e.g. not a metal because it would dissolve other metals) and high temperature: print the low temp where needed and then print high temp on top, that undesirable melting would occur but only at an interface that does not matter (since both materials will be removed after printing anyway).
          • Problem: The combined thickness might be too much for some places in printed parts because want to print bearings. As mentioned in todo file need to find out bearing tolerances and resolution (which would tell the gap layer needs).
        • A material that can be deposited at low temperatures but stands high temperatures. How?

• • • • • • A solvent which was carrying the material, leaves deposit behind. Low resolution probably. Could use piezo inkjet head. Dilute solution could produce thin layer

What about a solvent that has very low adhesion, which carries a solute that makes a suitable support material? If suitably low adhesion it could be deposited as droplets without spreading out on the surface, and could be vacuumed up more easily.

• • • • • • something that cured thermally or due to the light. Anything that could possibly liquefy at low temperature once and then refuse to liquefy again until high temperatures.
          • some sort of treatment with a particular gas, support material reacts with the gas to produce a temperature resistant upper layer (all of which can still be easily removed and/or act as a lubricant). Probably something that can do this, CVD may be a good place to start looking for suitable substances.
          • some other treatment involving something other than temperature. UV, Pressure, vibration, solvents or solutions of chemicals. Solvents in particular sounds like there might be something that can be found that is both volatile and reacts with the support material, and does not contaminate the metal etc. in the chamber when it is time to melt that.
          • a Preceramic polymer or some other substance that when heated enough, converts into another substance that has the desired temperature resistant properties. The obvious problem is that it would melt during the transition, and if it was not the material that was lowermost during the build process it might slump or flow in undesired ways. It could be heated so fast there is not enough time to slump.
    • temperature incompatibilities will necessarily impose some limitations on exactly what can be built. Imagine the whole build process. Metal cannot be laid down on top of polymer because it would normally burn or otherwise decompose the polymer, leading to a fragile layer in between. You could roughly approximate it with a thermally insulating and tolerant material to lay on top of the polymer, or some other material that forms a durable bond layer like a preceramic polymer maybe. If such approximations were strong enough maybe you'd be okay and could print what you want essentially. In the case where you do not want the layers bonded that's not as much of a problem usually because that implies a bit of space between the two layers at least a few microns wide, which can be filled with support material which is similarly temperature resistant but does not have to provide a strong bond.
      • only way I can think of right now to get past the limitation on temperature is to do the welding with firing the particles at the target at extreme velocities or as an ion beam. Extremely small particles might help as they would need less mechanical distortion to come into adequate contact with the surface? question: what sort of speeds would be needed to achieve 100% bonding and near 100% density? Plasma spraying achieves 97 to 99 percent density, but I don't think the bonds are full strength. Plus the plasma impinges on the workpiece so it would produce temperature problems itself. In plasma spraying the particles are melted though - what about firing solid particles and just letting them deform plastically on impact so they form a bond like in explosive or electromagnetic pulse welding? Would need extreme speeds. Wikipedia says EM pulse welding involves 1 km per second speeds. We only need to accelerate small particles. How would you do it?
        • travelling wave inside a chamber, helium or h2, extremely high pressures, the particles ride the front of the wave.
        • Speed of sound in solids can be even higher, a tube with surface waves that are moving towards the front of the tube (then around the outside and back in maybe), accelerate particles by getting them to ride the crests.
        • even plastic and ceramic particles are slightly conductive, maybe possible to accelerate them with lorenze forces like in em pulse welding, and maybe repeatedly along a sort of EM gun? Might at least work for metal particles? From the list of things that want to make metal and plastic are the most important things, if metal can be codeposited with with plastic and to high precision with a support material that is the bulk of it even if ceramics are out.

• • • Dissolving or otherwise removing support material from very narrow areas like between the surfaces of a bearing might be hard or take a long time. Ultimately brownian motion may help with dissolution so eventually you could get it out though. Reacting with a gas similarly. With melting it might be pretty much stuck in there though liquid or not.