There have been numerous attempts to print conductors. Fab@home, Ed Sells and myself have all tried it previously with very limited success. Whilst I've been able to print a basic circuit from solder, we were unable to achieve the resolution to produce anythin but the most simplistic circuit board. Forrest !is and others have tried to identify a useful non" metallic conductive material but conductivity has always been fairly poor. #onths ao I bloed about usin $ic%el &arbonyl powder for e'actly this purpose. What I didn't blo about was an e'periment I did mi'in the nic%el with a low meltin point alloy. When molten resultin semi"solid material had sinificant viscosity and effects of surface tension seemed minimal. (ust by stirrin the mi'ture with spoon it was obvious to )drian and myself that there was so much control that this li%e a suitable material to print conductors with. I did actually ma%e filament usin this composite material usin the old techni*ue of castin into silicone tubin. I was bettin on the mi'ture not separatin due to the ridiculously small particle si+e of the nic%el carbonyl ,about - microns.. )las I was wron, the mi'ture separated and the powder bloc%ed the no++le but the fundamental concept of usin a semisolid to ain more control over the e'trudate seemed li%e a ood one. /oically this is perfectly sound, it could potentially alloy us to deposit material on top of this semi"solid material without it oin completely fluid and losin its shape. Fortunately non"eutectic alloys also offer this ability to have solid elements suspended within a molten li*uid. Further they have the benefit that as the temperature increases the entire alloy completely melts thus we can further control the viscosity with temperature as well as the alloy composition. )fter some thouht I proposed that our ideal material would have a similar meltin point to 0/)1)2S such that when deposited on top a minimal amount of damae. I thouht at the time ,and this later proved to be wron. that the alloy should be as viscous as I could achieve at the RepRap: Blog Blog for the RepRap project at www.reprap.org - a project to create an open-source self-copying 3D printer. Contributors prusa3r 2u++ Eri% de 2rui3n &hristopher 4lah Ian )d%ins Simon #c)uliffe Wi+ard-5 Sebastien 2ailard e6 67plo7d 3mil 2odan 8ecman Forrest !is nop head 9hys (ones )drian 2owyer Steve 6e:roof Wade 2ort+ (onathan #arsden $eil ;nderwood #arius 8intel <ach Smith =i% 4lliver Site Feed archives #arch ->>? )pril ->>? #ay ->>? (une ->>? (uly ->>? )uust ->>? September ->>? 4ctober ->>? $ovember ->>? 6ecember ->>? (anuary ->>@ February ->>@ initial point the alloy meltsA Then the material would have a radual transition as the temperature is increased. I did have a hunt around but I was unable to find a material that I thouht sufficiently these attributes. 2ut we are now a community of at least B>>> machines, so I don't thin% its too impractical for us to do what any of the bi commercial additive manufacturin companies would do and have our own alloy. I decided to o for an alloy of tin, indium and bismuth few a few reasons. Firstly tin is comparatively cheap, so I was hopin to et the properties we need with very little amounts of indium and bismuth. Secondly indium tends to lower the surface tension of molten alloys,thirdly and most importantly there is a lot of data on the meltin point of alloys containin these elements so it'd be fairly easy to tune it to be comparable that of 0/)1)2S. This would let us to print onto the plastic whilst causin minimal damae. )bove is the phase diaram for the system. #y first alloy was @C.CD tin -C.-D bismuth and >.CD indium ,wt. D.. 0er %iloram the cost of the material was about EC> and I converted it to filament usin the same castin tric%. !owever this was a one off order so I anticipate it bein substantially cheaper in *uantity. The phase diaram predicted that this alloy should be about F>D solid when it beins to melt at about 75> de &, and be fully molten by 3ust under ->> de. The above was printed directly onto glass. I decided at the start of all this that if we are to produce useful circuits,we need to have sufficient control to build the track free form directly on top of PLA. Previously we needed channels to get any sort of the consistent #arch ->>@ )pril ->>@ #ay ->>@ (une ->>@ (uly ->>@ )uust ->>@ September ->>@ 4ctober ->>@ $ovember ->>@ 6ecember ->>@ (anuary ->>G February ->>G #arch ->>G )pril ->>G #ay ->>G (une ->>G (uly ->>G )uust ->>G September ->>G 4ctober ->>G $ovember ->>G 6ecember ->>G (anuary ->>F February ->>F #arch ->>F )pril ->>F #ay ->>F (une ->>F (uly ->>F )uust ->>F September ->>F 4ctober ->>F $ovember ->>F 6ecember ->>F (anuary ->>C February ->>C #arch ->>C )pril ->>C #ay ->>C (une ->>C (uly ->>C )uust ->>C September ->>C 4ctober ->>C $ovember ->>C 6ecember ->>C (anuary ->7> February ->7> #arch ->7> )pril ->7> #ay ->7> (une ->7> (uly ->7> )uust ->7> September ->7> 4ctober ->7> $ovember ->7> 6ecember ->7> (anuary ->77 track so this was already a big step forward. The rounding at the corner is due to the layer height being slightly off, which seems particularly critical when compared to plastics. )fter the initial success, this is where the onslauht of problems bean. Somethin I did not anticipate was that the alloy bean to dissolve the brass no++le due to solubility effects after very little use. In practice this meant that my e'trusion parameters were constantly varyin until the no++le completely disappeared ,this happened within about 7? meters of 5mm filament. . I also attempted an aluminium no++le with the same result. )t this point we were a bit stumped. )drian suested tryin to ma%e an all 0TFE e'truder ,at least as far as the alloy was concerned.. )fter several iterations the best solution was machinin a 0TFE liner with a built in no++le. This was then surrounded by the usual brass heater etc li%e our hybrid e'truders. Whilst I accepted this was not a lon term solution, I thouht that if we could et it to wor% in steady state conditions we could learn more about the process. )s it turns out with e'tended use the alloy was so viscous the alloy physically ripped the 0TFE liner that serves perfectly well in our plastic e'truders. Further it was apparent a blac% o'ide was formin at the entrance to the no++le. )fter a few hours use this ended up in a complete e'truder failure. February ->77 #arch ->77 )pril ->77 #ay ->77 (une ->77 (uly ->77 )uust ->77 September ->77 4ctober ->77 $ovember ->77 6ecember ->77 February ->7- #arch ->7- )pril ->7- (une ->7- (uly ->7- September ->7- 4ctober ->7- (anuary ->75 February ->75 #arch ->75 )pril ->75 #ay ->75 (une ->75 (uly ->75 )uust ->75 September ->75 6ecember ->75 February ->7B )pril ->7B Subscribe to 0osts H)tomI )fter this failure it was fairly apparent that my initial alloy was far too aressive. I therefore scaled it bac% to somethin that would operate at lower temperature ,reducin o'idation times. whilst also havin a lower viscosity to put less stress on the 0TFE liner. I uessed that a ?>D li*uid1solid ratio when the alloy beins to melt would be sufficient, and this turned out to be a composition of ?G.?DSn, B7.5D2i, 7.-D In, aain beinnin to melt at 75>de but finishin by about 7G>. This proved much more reliable and with much less o'idation. It allowed me to produce the results below printed directly onto lass ,it was flat before I peeled it off.. The second imae is particularly interestin as its three layers of trac% on top of one another, which would be re*uired for trac%s connected in parallel. I've also otten similar results printin directly onto a 0/) substrate, in this case it stuc% fairly well but could be removed with a bit of force. Whilst I've achieved fairly decent results with this setup, the transient performance was unsurprisinly poor iven the insulatin nature of the 0TFE and the lare melt +one I needed to et it to wor%. I therefore went bac% to solvin the underlyin problem of the alloy dissolvin the no++le. I've had ao at anodisin an aluminium no++le which is fairly easy, it 3ust re*uires about ->> =dc and the appropriate electrolyte. )t the moment the no++le desin is still varyin a bit, but I've ta%en pictures under a microscope both after anodisin ,first pic. and after e'trudin about 7>m of -mm filament. )s far as I can tell there hasn't been any substantial wear. )fter about 7> different e'truder iterations ,most I haven't mentioned here. It appears that the dissolvin issue has therefore been solved. I haven't *uite finished calibratin this e'truder, but the latest results are below. $aturally its our loo, the trac% si+e is about >.Gmm in diameter. :iven this was fairly successful I also tried doin it with infill, unfortunately it overheated as the filament didn't have chance to cool but I still thin% its *uite successful. I need to try repeatin this with coolin from a fan. I would have continued this build further but I can only ma%e lenths of filament about 5>cm lon and chanin them mid print is *uite tric%y, /abelsJ 0rintin &onductors K posted by 9hys (ones @ 77J?7 )# Comments: I remember that some time ago someone designed a hot end for use with a glass nozzle. The nozzle was made by heating a glass tube with a blowtorch then pulling it to taper the end to a point. The end was then ground down until the opening had the desired diameter. You might try using this type of nozzle to extrude your mixture, as it shouldn't dissolve the way brass does. # posted by whosawhatsis ! "une #$, #%&& &&!&' () I'm curious. *ow is the alloy is dissolving the brass nozzle in the first place +hat reaction is going on here # posted by stan ! "une #,, #%&& &#!#- .) /i0uid metals can act as solvents for other metals below their melting point. # posted by nophead ! "une #,, #%&& &!#, .) .h1 2reat1 You've published1 2ood going, 3hys1 !45 # posted by 6orrest *iggs ! "une #,, #%&& #!-# .) 7ice1 # posted by "eff 8eegan ! "une #,, #%&& 9!:9 () ;eautiful1 .t the <ancouver )a=er6aire last wee=end, probably every -th visitor to my 3ep3ap table ended up as=ing about printing metals. I can't wait to start experimenting with this1 # posted by +ade ;ortz ! "une #', #%&& >!&: .) <?3Y exciting progress1 It's a smart move to dirty up the eutectic alloy to ma=e it more viscous. @eems li=e well tuned formula could be the print medium of our digital futures of physical devices. Aombine that with pic=4n4 place for components Balso pic=4n4place disassembly of obsolete devices5 and we can start ma=ing complex electronic devices with an essentially single4clic= to print process1 # posted by ?ri= de ;ruiCn ! "une #', #%&& ,!:' () This loo=s li=e a bit more complicated to create the in=, but It doesn't ta=e any heat and seems 0uite durable... http!DDnews.illinois.eduDnewsD&&D%$#'silverEpenE"ennifer/ewis.html # posted by T/ane ! "uly %&, #%&& :!:- .) This was a great lin=. Instead of melting metals why not Cust plot it all out with a pen +hat about powdered metals with a binding agent /ay down the binding agent then put the powdered metal over the top of the binding agent +ould the binding agent remove any change of electrical conductivity # posted by +anna;e2ee=ster ! "uly &-, #%&& $!># () I'm worried about all the embodied energy of reduced aluminum, should this alloy ever brea= through the oxide coating. It might be worthwhile to loo= into a coating that stands up to this alloy for extended periods. I =now there are lots of metals that simply don't dissolve in indiumDgalliumDtin alloys Bmercury, for example! it and galinstan can be sha=en up and will phase4separate5. )aybe chrome4plated brass Ahrome plating services are reasonably cheap and accessible, and I don't thin= it's out of the 0uestion as a home enterprise, either. # posted by "oel ! "uly #%, #%&& '!%' () Is it possible that instead of melting metal, other aproach could be electroplating the metal # posted by )emo ! "uly :%, #%&& &#!%' .) *ave you considered using other materials for the nozzle such as ceramics or heat resistant plastics *ave you considered coming at the problem from the opposite direction (rinting a very accurate pattern of groves then filling it with a fluid metal and sealing it with a second layer of plastic to ma=e the circuit board, mercury is the first example that comes to mind but there are toxicity and expansion issues Bobviously5 so maybe ma=e the groves double the size they need to be and pipe solder in then cut it down post printing to ma=e a proper size tidy circuit board, there would be allot of waste but it would wor= as a proof of concept. # posted by .lex ! .ugust :%, #%&& #!%# () *ave you considered using other materials for the nozzle such as ceramics or heat resistant plastics *ave you considered coming at the problem from the opposite direction (rinting a very accurate pattern of groves then filling it with a fluid metal and sealing it with a second layer of plastic to ma=e the circuit board, mercury is the first example that comes to mind but there are toxicity and expansion issues Bobviously5 so maybe ma=e the groves double the size they need to be and pipe solder in then cut it down post printing to ma=e a proper size tidy circuit board, there would be allot of waste but it would wor= as a proof of concept. # posted by .lex ! .ugust :%, #%&& #!%: () Fne option if a faster oxidizing alloy prints better is to use a helium reducing atmosphere which is pretty easy and cheap to set up Bplastic box with helium canister5. (robably too much hassle for experimenting but it might be usable at a later stage. Im currently using this techni0ue to prevent copper oxidation during casting. # posted by TomA ! "une #9, #%&: '!#, () Post a Comment !ome