As I’ve written about at some length previously, some of the most promising computational fabrication technologies at macroscopic scale are 2-D cutting processes like laser cutting, waterjet cutting, and CNC plasma table cutting. See Layers plus electroforming for notes on the scaling laws. Cooper Zurad has prototyped an electrolytic 2-D cutting process using a needle-shaped cathode, but his process is very slow and imprecise because he’s cutting at a single point, and because he’s not doing the usual ECM things: closed-loop control of the process gap, using pulsed current, or vibrating the electrodes to reduce the tradeoff between flushing and cutting speed. Traumflug did similar experiments in 02011.
However, even without gap control, pulsing, and vibration, if you’re cutting over a large area rather than just a point, you should be able to get a proportionally higher material removal rate. And if you’re cutting a thin sheet of material, even a low material removal rate might be adequate.
Unlike other ways of cutting a thin sheet of material, this sort of electrolytic cutting leaves no burrs and does not produce heat-affected zones or mechanical stresses in the material being cut — though for very thin sheets the surface tension of the electrolyte may be big enough to plastically deform the workpiece, possibly requiring supercritical drying if the workpiece is ever to be removed from water.
In addition to cutting sheet metal, this process can be used to selectively remove a metal coating (as on a printed circuit board) or to etch or anodize a surface.
One way to do this is, as I wrote in Dercuano in 02016, to have an array of separate cathodes close to the anode workpiece, controlling the voltage or current of each cathode to either dissolve the workpiece near it, or not. A different way is to prepare the pattern of the cuts in a material form, for example as a printed circuit board, a network of wires on an insulating plate, or a pattern of apertures in an dielectric mask placed over a continuous-sheet cathode, and make this pattern the cathode. Then you can “print” this cut pattern on a series of sheets of metal.
One particularly interesting cathode-patterning possibility is to produce the insulating “mask” by laser-printing on paper, ideally paper that will not fall apart when soaked with the electrolyte. If the laser printing is sufficiently solid to be used for other toner-transfer methods, it should also work for this electrolytic sheet cutting approach; unlike the other toner-transfer methods, it might be possible to get more than one metal copy from a single paper pattern.
Thermal wax printers may or may not produce a better dielectric pattern.
In either case, the cathode is placed very close to the anode workpiece; the most practical way of doing this is probably to separate them with either some sort of fabric, such as a paper towel or other nonwoven cloth or a woven cloth, or with a thin porous membrane full of holes, for example a thin porous layer of polyethylene. (In the case of laser printing on paper, the paper itself provides the separator.) Once the current is turned on, the cuts are all made simultaneously, though some may take longer to finish than others. Voltammetry should be adequate to determine when the process completes.
With this approach, the porous separator, plus any space it produces around the cathode, need to be able to absorb the metal salts produced by the cutting, since vibration or indeed flushing at all would be very difficult.
In cases where the cut pattern contains some cuts that are completely surrounded by others, for example holes “drilled” in a part to be cut out, there is a potential problem. If we feed the anode current in from the edge of the anode workpiece, it may not be able to reach the inner cuts if the outer cuts happen to complete first. There are several possible ways to solve this problem:
We can do the cutting in two phases, first cutting out the inner contours and then later the outer contours. This is a common tactic in single-point 2-D cutting processes like those I mentioned above, because it prevents the outer piece from falling out of plane before you’ve cut the holes in it. This would result in slower cutting but prevents the problem. It requires setting up the cathode pattern in two or more electrically separate parts.
We can feed in the current to the anode with an inert conductive backing anode, thus coupling in current to the workpiece over its entire back surface rather than only at the edge. This will reduce the efficiency of the process and perhaps eliminate the usefulness of voltammetry for measuring its completion.
We can try to leave “tabs” around the outside of each cut, so that when the process is finished all the desired parts are still all connected together and must be removed from their support in a separate step. This is commonly done in casting, molding, and, especially in the case of thin materials, 2-D cutting by means such as lasers. It might be possible to calibrate the process with enough precision that the tabs are barely thick enough to maintain electrical continuity until the inner contours are all cut out, but continuing the process a little longer severs the tabs.
As one example, as I wrote in “Dead bugging” in Derctuo, it’s easy enough to find 100-μm-thick copper conductors (a) in stranded copper wire. You could paste these in the desired pattern (b) onto a high-impact polystyrene surface (c) with PVA glue (d), bridging disconnected parts of the cut pattern with fine varnish-insulated magnet wire (e). After allowing the PVA glue to dry, you make it water-insoluble by treating with a borax solution. You lay a paper towel (f) onto the pattern and soak it with sodium nitrate (g). You lay a sheet of 10-μm-thick household aluminum foil (h) onto the paper towel and press the whole assembly together, then apply a low voltage power supply for long enough to dissolve more than 10 μm of aluminum. The results should be, for example, if your current density is enough to cut 100 μm/s, that the cut is completed in 100 ms. Applying electricity for less time will result in etching the surface rather than cutting through.
You can heat up example 1 to make the cutting go faster, using a heated press such as are commonly used in dye sublimation, laundry pressing, and vinyl transfer onto cloth.
For the foil (h) in examples 1 or 2 you can substitute heavy-duty aluminum foil, commonly available in thicknesses such as 50 μm; aluminum flashing; aluminum sheet from aluminum cans (typically 100 μm) after cleaning nonconductive contaminants off of one of its two sides; aluminum sheet as is commonly available from metal vendors like Metals Depot, commonly in thicknesses as low as .032" (81 μm) or any other source of sheet aluminum. Thicker sheets will take proportionally longer to cut, produce less precise cuts, and, above a certain thickness, will also require a thicker separator (f). Cutting multiple stacked layers of metal (h) in a single run is a possibility that may increase the efficiency of the process in several ways, such as amortizing the setup time over multiple produced pieces, but will reduce the precision achieved.
For the sodium nitrate (g) in examples 1, 2, or 3 you can substitute any other soluble salt whose anion forms a soluble aluminum salt or aluminate, such as sodium chloride, azanium acetate, iron sulfate, or potassium hydroxide, among dozens of other possibilities; particularly appealing are the chloride, acetate, sulfate, and hydroxide salts of azanium and the alkali metals, due to their high solubility and low toxicity; the corresponding salts of iron and zinc are also relatively safe and, except for the hydroxides, soluble. More toxic options include sodium perchlorate. In cases such as potassium hydroxide which are capable of corroding the aluminum rapidly without electricity, it will be necessary to stop the reaction, for example by washing the pieces thus produced with water, a buffer solution, or an acid that will not attack the aluminum. Salts which produce a passivating “anodized” layer on the aluminum at lower voltages may be preferable, because although they reduce efficiency, they will restrict the electrolytic etching to areas at sufficiently high voltages, improving the precision of the process. It is probably also useful to include additives such as metal borates, metal EDTA, metal tetrasodiumglutamatediacetates (GLDA) to prevent the formation of aluminum hydroxide, metal cyanides, SPS (CAS 27206-35-5), MPS (CAS 17636-10-1), ZPS (CAS 49625-94-7), polyethylene glycol, polyvinyl alcohol, polyvinyl acetate, glycerine, propylene glycol, dipropylene glycol, DPS (CAS 18880-36-9), surfactants (such as SLS, alkali stearates, EN 16-80 (CAS 26468-86-0), or EA 15-90(CAS 154906-10-2)), UPS (CAS 21668-81-5), PPS (CAS 15471-17-7), NAPE 14-90 (CAS 120478-49-1), sodium benzoate, saccharin, coumarin, metal tartrates, metal citrates, metal sulfonates not otherwise mentioned, metal urates, thiazole, benzaldehyde, thiourea, quaternary azanium salts, phthalimide, metal methanesulfonates, metal ethylene sulfonates, depolarizers (such as manganese dioxide, metal sulfates, silver oxide, or metal chromates and dichromates, among many other possibilities), the acid forms of the anions mentioned here, or these anions’ salts with organic cations or azanium, as well as other additives used in electrodeposition and ECM. Also, the solvent in which the salt is dissolved can be replaced with any other solvent suitable for the salts employed, such as DMSO, ammonia, ethyl acetate, THF, DCM, acetone, acetonitrile, DMF, formamide, acetic or formic acid, alcohols (such as methanol, ethanol, and isopropanol, among many others), organic carbonates (such as propylene carbonate, ethylene carbonate, diethyl carbonate, or dimethyl carbonate, among many others), glycerol, nitromethane, molten methylsulfonylmethane, deep eutectic systems, or other ionic solvents, among hundreds of others, or mixtures of these, with or without water; such substitution can permit the use of higher temperatures or electrolyte salts that either react undesirably in water or will not dissolve in it, and may be able to reduce the surface tension to less mechanically damaging levels. Generally the more important solubility consideration will be the solubility of the salts produced at the anode workpiece, since you cannot choose their cations, rather than the electrolytic etchant (g).
For the paper towel (f) in examples 1, 2, 3, or 4 you can substitute any other porous material that will not be attacked too rapidly by the salts and is not too electrically conductive except ionically; for example, asbestos, fiberglass, carbon fibers, carborundum fibers, rock wool, basalt fiber, ordinary paper, buckypaper, nonwoven polypropylene, nonwoven polyester, nonwoven cotton, nonwoven rayon, onion-skin paper, other thin papers such as crepe paper and those used for tracing drawings and rolling cigarettes, perforated polyethylene film, perforated PET film, perforated polypropylene film, hydrogels (such as gelatin, agar, borated polyvinyl alcohol, or silica gel), woven textiles of the above-mentioned fibers, and porous ceramics such as glass frits or unglazed fired clays, among dozens or hundreds of other possibilities. Woven textiles will tend to add their weave pattern to the etched pattern, which may be considered a form of error in some applications. You can stack more than one such layer; for example, a layer of perforated polyethylene film can be used to separate a layer of borated PVA hydrogel from the cathode, preventing adhesion. Perforated boPET or polyethylene films can easily be made under 10 μm in thickness, a feature which might enable reproducing details not much larger than that.
For the varnish-insulated magnet wire (e) in examples 1, 2, 3, 4, and 5, you can substitute wire insulated by other means such as thin layers of dielectric polymers, or you can pierce holes in the dielectric backing (c) to pass through conductors from a region devoid of electrolyte or at least separated from the workpiece by a dielectric or by distance.
For the borate-crosslinked PVA glue (d) in examples 1, 2, 3, 4, 5, and 6, you can substitute any other material that will hold the pattern conductors in place while permitting electrolytic access to them; for example, agar, gelatin, cross-linked starch, hydrogels used for contact lenses (such as silicone hydrogels, hydroxyethyl methacrylate), sodium polyacrylate as used in maxi pads, polyethylene glycol (perhaps treated to crosslink it into an insoluble gel as is commonly done for cell encapsulation), and porous ceramics such as glass frits or those made by unglazed fired clay. Alternatively, the separator material (f) can simply be bonded permanently to the cathode, which would require the electrolyte to be washed out between runs rather than merely replacing the separator as you would normally do. Alternatively, instead of holding the cathode pattern in place with any kind of continuous material, you can hold it in place with occasional thin fibers of dielectric material either bonded to the dielectric separator (c) or passing through it, as is done in embroidery or furniture decoration to hold certain kinds of thread or piping on the surface of the material.
For the dielectric backing material (c) in examples 1, 2, 3, 4, 5, 6, and 7, you can substitute any other dielectric material that will not be too readily attacked by the electrolyte and in particular the alkaline solution that will tend to form in contact with the cathode, such as glass, polyethylene terephthalate, poly(methyl methacrylate), polymerized linseed oil, shellac, polyethylene, polypropylene, epoxy resins, teflon, fluorinated ethylene propylene, other polyester resins, aluminum oxide, or other metal oxides, among many others. A stack of such layers may be useful. Extremely inert backing materials such as teflon introduce the problem that firmly adhering the cathode to them with the PVA glue (d) or its alternative may be more difficult; stacking a readily adherable material such as HIPS on top of a more inert material such as polyethylene is one possible solution, and welding the backing (c) to the glue (d) will also improve adhesion in difficult cases.
For the fine copper conductors (a) in examples 1, 2, 3, 4, 5, 6, 7, and 8, you can substitute nearly any other conductive material at all as long as it’s sufficiently cathodically protected; for example, copper, aluminum, gold, silver, platinum, palladium, rhodium, tantalum, niobium, vanadium, molybdenum, graphite, glassy carbon, non-glassy amorphous carbon, nickel, stainless steel, chromium, lithium metal, sodium metal, or ordinary steel, or mixtures of these, among hundreds of other possibilities, in the form of fine wire, foil, thin film, or plating. Some of these possibilities rule out the use of certain electrolytes; for example, sodium metal probably cannot be used in contact with water regardless of how well it’s cathodically protected. The use of nobler metals such as tantalum and gold does not affect the anodic dissolution process and permits the use of more aggressive electrolytes. Thinner metals such as gold leaf, especially together with thinner separator layers and thinner workpieces, permit finer patterning of the workpiece. Additionally, you can provide the pattern instead by using a continuous layer or mesh of any of these materials as the cathode, superposed on a selectively nonporous mask of some dielectric material, such as the laser-printer toner mentioned earlier or materials such as those listed in example 9 above.
In examples 1, 2, 3, 4, 5, 6, 7, and 8, as an alternative to a pattern (a) supported on a dielectric backing (c) as described above, you can use a conductive plate (i) made out of any conductive material such as those mentioned in example 9 above with a selectively patterned impermeable dielectric “stop-off” or “mask” (j) on it, so that electrolysis can only proceed where the mask is absent or at least porous. For example, you can use sheet steel or any other sheet metal with nail polish selectively painted onto it; or a dielectric photoresist deposited on it and optically patterned in the way that is common for fabricating integrated circuits or printed circuit boards; or laser printer toner transferred onto it; or a dielectric coating selectively deposited by inkjet printing, perhaps then baked to improve the coating; or “permanent” marker ink; or “dry erase” marker ink; shellac (an idea due to Mina); cellophane tape; paraffin; powder coat paint, as commonly used for painting industrial machinery; glass, as in cloisonné; dried soluble silicates, if heated between uses to drive out excess water; a layer of a passivating compound formed from the surface of the conductive plate (i) itself, for example by heating or anodizing; polymerized linseed oil; photoresists; teflon; rosin; spray paint; shellac; or any other dielectric that is sufficiently resistant to the electrolyte. Many of these dielectrics can be applied in a continuous layer and then selectively removed by laser ablation, for example with a low-wattage laser cutting and engraving machine like those commonly used for cutting MDF, or by some other method such as stamping, grinding, abrasive jet blasting, or scraping. The mask (j) can be a separate removable layer rather than firmly adhered to the conductive plate, as in the earlier example of laser-printed paper; the screens used in silkscreening or the waxed fabric in batik would work well for this. A nonwoven thermoplastic cloth can combine the functions of the mask (j) and the electrolyte bearer (f) by being melted in the regions to be “masked”, rendering it nonporous, as is commonly done to join nonwoven thermoplastic cloths.
In example 10, instead of protecting parts of the pattern electrode surface (i) with a solid dielectric, you can protect parts of the pattern surface by recessing them far enough that when the conductive pattern plate is brought into contact with the electrolyte-soaked porous material (f), the recessed parts are separated from it by an air gap.
In example 10, instead of protecting parts of the pattern electrode surface (i) with a solid dielectric, you can cut spaces in the electrolyte-soaked porous material (f), or otherwise pattern it to fill only a part of the space between the two electrodes. For example, a thin stranded string of fiber can be shaped into the desired pattern, moistened with electrolyte, and squished between the two plates before applying the power.
In examples 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10, if you use an electrolyte that can dissolve the aluminum workpiece at zero voltage, such as alkali-metal hydroxides or hydrochloric acid, you can reverse the voltage to cathodically protect the workpiece rather than the pattern. This will result in dissolution of the workpiece in a positive pattern (leaving workpiece where the pattern is present rather than where it is absent) rather than a negative one. In this case it may be convenient to first subject the whole workpiece to cathodic reduction (using it as a cathode with an unpatterned anode, possibly with a different electrolyte) to eliminate possible passivating oxide films, before reversing the polarity. This approach can also result in the dissolution of the pattern, possibly in an uneven fashion resulting from parts of it remaining connected longer, as with the tabs mentioned earlier. (A similar consideration applies to ensuring the electrical continuity of the protected part of the workpiece until the end of the process.) This can be prevented by using a nobler material for the pattern than for the workpiece and operating at a moderate enough voltage to prevent the pattern from being attacked. Stainless steel wire or graphite is probably the most convenient pattern material in cases where copper is insufficiently noble. As an alternative to preventing this electrolytic pattern erosion, if the pattern is thick enough, you can alternate between patterning a workpiece cathode in this way, and electrodepositing new metal on the pattern to replace the lost metal.
In examples 1, 2, 3, 4, 5, 6, 7, 8, 9, 12, and 13, as explained earlier, rather than using a dielectric backing (c) with a patterned electrode (a) on it, you could use an array (k) of independently controlled electrode pixels insulated from one another.
In examples 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, and 12, rather than dissolving the workpiece, you can deposit oxides or other insoluble metal salts on its surface by a suitable choice of electrolyte (g), known as “anodizing”. This can be used, among other things, to selectively passivate it for future use as a pattern electrode (as mentioned in example 10), to selectively passivate it to selectively resist some other etching process or reaction, to selectively harden the surface, to color it with opaque compounds, or to color it with iridescence by controlling the layer thickness, a process which is less effective for aluminum than it would be with some other metals. It may be possible to modulate the current density over time to spatially modulate the density of the oxide layer to form a rugate filter.
In examples 1, 2, 3, 4, 5, 6, 7, 8, 9, and 14, instead of a dielectric backing (c) with a patterned cathode, you can use one or more movable cathodes (m) that electro-etches the anode workpiece (h) where it touches the porous material (f) and not elsewhere. Some useful forms of patterned cathode for this purpose might include one or more a narrow rollers like pizza cutters, which cut along a line rather than at a single point while exerting minimal friction on the porous material (f); one or more needles which are touched to the surface of the porous material (f) at different points at different times; a metal ball like that used in ballpoint pens and ball bearings, which can roll like the pizza cutters; outlines of various forms, such as circles and semicircles of different diameters, the edges of razor blades, the whole shapes of parts, logos, letters, cartoon characters, and halftone patterns, which can be placed at different points on the material at different times and etched to varying depths. These “stamps” can be made in many different ways, including engraving or etching a solid metal or graphite surface, and bending wire. A soft wire brush is another candidate cathode, as in brush electroplating. The roller approach and the seal approach can be combined in a rolling seal. A wire or metal tape can be used to etch a straight line of variable length all at once, either by hand or under the control of a machine similar to an old automatic wire-wrap machine.
In example 15, the porous material (f) can be attached to the movable cathode or cathodes (m) rather than to the workpiece (h), and the pattern can be in the porous material rather than the cathode, as in example 12.
In examples 1, 2, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, or 16, the aluminum sheet (h) can be replaced by a sheet of almost any other metal or conductive ceramic with a corresponding change of electrolyte (g), though platinum and a few related substances are considered impossible to dissolve anodically. This opens up more interesting anodizing possibilities, such as bluing steel, or depositing iridescent layers on metals with transparent oxides with high refractive indices such as titanium. This cutting process is particularly appealing for very hard metals, ceramics, and cermets.
In the cases where the porous material (f) is held fixed relative to the workpiece (h), the result of the process includes not only the etching of the anode (h) but also metal ions impregnated into parts of the porous material (f). In some cases this directly produces a visible pattern on the paper or similar material, but whether it does or not, onchrome or afterchrome dyeing can be used to produce a permanent, colorfast pattern. Mordants commonly used for dyeing include salts of copper, tin, iron, aluminum, chromium, and tungsten, so electro-etching anode metals containing these metals can be used to selectively mordant a textile in this way — either one that is dyed afterwards (“onchrome”) or one that is pre-impregnated with the dye (“afterchrome”) and then merely washed to reveal the pattern. For this it is necessary to use a dye that cannot be effectively mordanted with whatever cations are present in the electrolyte before electro-etching; I think azanium ions are safe for all dyes.
Ferrous (Fe²⁺) ions from the photodecomposition of organic ferric (Fe³⁺) complexes (ferrioxalate, ferricitrate, ferric oxalate, ferric tartrate) are commonly used in this way to tone siderotypes with various kinds of vegetable pigments, as explained in Mike Ware’s Cyanomicon §3.5 (p. 77). It also explains the possibility of using the ferrous ions to “reduce the compounds of a ‘noble’ metal, such as platinum, palladium, silver or gold, to the metallic state”, which also leaves an indelible mark in the paper. Below about pH 9, the Pourbaix diagram for iron says there’s a wide stability range for ferrous ions; unfortunately this range extends below 0 V, to about -0.6 V, so if we look only at the equilibrium, this range can only be used directly in the examples above in the “positive” process, where the workpiece is a cathode that is dissolved except where it is cathodically protected. However, I think that in practice the spontaneous reaction rates may be low enough to permit the use of the “negative” process where the iron electrode is anodically dissolved, particularly if there’s a little bit of postprocessing to cathodically draw some of the dissolved iron out of solution.
It would be exciting to be able to do the same trick with an ion that could be further oxidized to reduce something a little less noble (and expensive) like copper. But I can’t think of anything.
Imaging is not the only way of using this selective ion impregnation. The polyvalent cations thus obtained can selectively catalyze other reactions if the porous material is supplied with the reagents, and they can (noncatalytically) harden soluble silicate and phosphate solutions, which can thus selectively stiffen the porous material, thus forming a ceramic/fiber composite object with flexible blades of fibrous material joining stiff blades of fiber-reinforced ceramic.
Ware’s book also reports (§4.2 “Pellet’s process”, p. 89):
The new feature that Pellet introduced to solve the problem of fixing the positive-working process was based on earlier observations by Alphonse Poitevin in 1863, that ferric salts cause gum and similar colloids to harden and become insoluble in water, whereas ferrous salts do not.
In §4.10 on p. 113 he explains the gel lithography process:
Gel lithography was also variously known in its hey-day by several proprietory [sic] names: Lithoprint, Ferro-gelatine, Fotol, Ordoverax, Velograph or Fulgur printing. The method took an over-exposed, but unprocessed negative blueprint image as its source, which was lightly squeegeed into contact with a matrix of moist gelatine, known as a “graph”, containing a ferrous salt. Diffusion of excess potassium ferricyanide out of the lightly-exposed and unexposed regions of the cyanotype (the image shadows) formed Prussian blue in the gelatin matrix, with the effect of hardening it locally (see §4.1.1), where it became receptive to a greasy lithographers’ ink, which was still repelled by the moisture in the unhardened regions of the gelatin. After a minute or so, the cyanotype was peeled off and the jelly surface inked with a roller. About 25 positive copies of the original image could be ‘pulled’, using little pressure, from the jelly, which was re-inked between each.
I’m not sure whether the effect of hardening the gelatin was due to the complexed ferric ions in the ferricyanide (but Ware asserts that it is, and we can probably trust him) or some other effect of forming the Prussian blue.
Aluminum has only one common oxidation state (3+), so the Faraday efficiency of the setup should be near perfect with aluminum. It’s 27.0 g/mol. At 96485.34 coulombs per mole, times three, we have 10.7 megacoulombs per kilogram:
You have: avogadro 3 e / (aluminum g/mol)
You want: MC/kg
* 10.727928
/ 0.093214641
At an ordinary electroplating current density of 10 A/dm² and 2.70 g/cc this is almost a year per meter:
You have: avogadro 3 e / (aluminum g/mol) / (10 A/dm^2) * 2.70 g/cc
You want: days/m
* 335.24776
/ 0.0029828685
This works out to only 35 nm/s, which would take 300 seconds to cut through a 10 μm household aluminum foil. We ought to be able to use a much higher current density for electrochemical machining because we don’t have to worry about forming dendrites, but the 1000 A/dm² we’d need to do the cut in three seconds sounds pretty extreme. Is it?
Considering the 100-μm-diameter copper wires I was talking about at the beginning, how much current are we talking about? Suppose one such wire is 100 mm long; it then covers an area roughly 100 μm by 100 mm on the porous medium (f), 10 mm². Then 1000 A/dm² would be just 1 amp. Suppose we’re feeding it from both ends. At a nominal copper conductivity value of 58 siemens m / mm² (from definitions.units) this works out to 2.20 Ω/m and thus 0.11 Ω in the 50 mm; if all the current came out in the middle it would be 500 mA in each side through that 0.11 Ω, with a resulting voltage drop of about 55 mV, which is probably bearable. But actually the current is hopefully coming out evenly along the length of the wire, so the situation is a little better, with that 500 mA though 2.20 mΩ/mm initially dropping 1.1 mV/mm, but linearly dropping to 0 mA and thus constant voltage in the middle of the wire. Without actually doing the algebra, I think this works out to a voltage drop of 27 mV.
This is a very reasonable voltage drop. I think it also works out to about 7 mW, which normally would be a large enough power to worry about in a tiny wire like this, but maybe not when it’s immersed in water.
The standard electrode potential of reducing 2H₂O to H₂ and 2OH⁻ is -0.8277 V (per electron), and that for oxidizing Al to Al³⁺ + 3e⁻ is -1.662 V (per electron). If I understand this stuff right, which I might not be, that means you need at least 834 mV between the electrodes before you start electro-etching the aluminum. This is a very easy voltage to supply and implies that the overall power needed to do these cuts is only about 800 mW, plus whatever gets wasted on Joule heating of the electrolyte and the cathode (about 3.4% in the electrode in the above example). If you have something in the electrolyte that’s more likely to deposit on the cathode than sodium or aluminum — copper, say — then you might not have to pay even that much; but then your cathode becomes less precise.
If we use the conductivity of seawater, 50 mS/cm and an electrolyte path of 100 μm, we get 2 Ω:
You have: 100 um / (50 mS/cm * 100 um 100 mm)
You want: ohms
* 2
/ 0.5
At 1 A this would be a joule-heating voltage drop of 2 V, giving a total of 2.859 V: 2.000 V in the electrolyte, 0.832 V in the electrolytic interfaces, and 0.027 V in the wire. The conductivity is proportional to the ion mobility, the ion concentration, the ion charge, and the temperature (≈2%/°); with more concentrated solutions, and concentrations with highly mobile ions (hydronium beats sodium 7×), we ought to be able to get it down to 0.2 Ω and thus 0.2 V, so that even at 1000 A/dm² (100 mA/mm²) we spend 80% of the energy on electrolysis. And of course at lower currents the ohmic losses become insignificant.
At significantly higher currents the voltage drop along the wire would become sufficient to provoke different electrolytic reactions in different places, which is not the desired effect. This would also produce different current densities in different places, and thus reaction speeds, cutting speeds, and potentially cut widths; a higher-resistivity electrolyte will tend to avoid this problem, at the expense of wasting more energy as heat.
A power supply that can produce 3 V at 1 A is straightforward to cobble together from common components; in the most primitive form, two resistors and a power-transistor emitter follower can produce this from many USB chargers, though it would produce a lot of heat. A more efficient switcher design is also not very demanding and would be a lot safer.
So in fact cutting through hand-sized aluminum foil in a few seconds with submillimeter precision is eminently attainable, and should be reasonably efficient, using minimally 8 mJ per millimeter and realistically 30 mJ/mm. If you could manage a thinner kerf, it could be even more efficient. Scaling the cutting up to higher speeds, larger workpieces, or very complex cuts might start to be a challenge, though.
Different metals require somewhat different amounts of current, but the density of the electron gas you’re sucking out of the metal doesn’t vary nearly as much as other properties of metals such as hardness, toughness, mass density, and electronegativity; here are my calculations for a selection of metals including the common ones (excluding the air-unstable sodium, potassium, calcium, strontium, and barium and the brittle manganese):
| metal | molar mass | density | valence | current required | melts |
|---|---|---|---|---|---|
| Silver | 107.868 g/mol | 10.49 g/cc | 1 | 9.38 A/mm²/(mm/s) | 1234.93 K |
| Gold | 196.967 g/mol | 19.30 g/cc | 1? | 9.45 A/mm²/(mm/s) | 1337.33 K |
| Lead | 207.2 g/mol | 11.34 g/cc | 2 | 10.6 A/mm²/(mm/s) | 600.61 K |
| Tin | 118.710 g/mol | 7.265 g/cc | 2? | 11.8 A/mm²/(mm/s) | 505.08 K |
| Zirconium | 91.224 g/mol | 6.52 g/cc | 4 | 13.8 A/mm²/(mm/s) | 2128 K |
| Magnesium | 24.305 g/mol | 1.738 g/cc | 2! | 13.8 A/mm²/(mm/s) | 923 K |
| Titanium | 47.867 g/mol | 4.506 g/cc | 4? | 18.2 A/mm²/(mm/s) | 1941 K |
| Zinc | 65.38 g/mol | 7.14 g/cc | 2 | 21.1 A/mm²/(mm/s) | 692.88 K |
| Iron | 55.845 g/mol | 7.874 g/cc | 2? | 27.2 A/mm²/(mm/s) | 1811 K |
| Copper | 64.546 g/mol | 8.96 g/cc | 2? | 27.2 A/mm²/(mm/s) | 1357.77 K |
| Aluminum | 26.98 g/mol | 2.70 g/cc | 3 | 29.0 A/mm²/(mm/s) | 933.47 K |
| Cobalt | 58.9332 g/mol | 8.90 g/cc | 2? | 29.1 A/mm²/(mm/s) | 1768 K |
| Nickel | 58.693 g/mol | 8.908 g/cc | 2? | 29.3 A/mm²/(mm/s) | 1728 K |
| Molybdenum | 95.95 g/mol | 10.28 g/cc | 3? | 31.0 A/mm²/(mm/s) | 2896 K |
| Chromium | 51.9961 g/mol | 7.19 g/cc | 3 | 40.0 A/mm²/(mm/s) | 2180 K |
| Tungsten | 183.84 g/mol | 19.3 g/cc | 6? | 60.8 A/mm²/(mm/s) | 3695 K |
(The unit A/mm²/(mm/s) is equivalently A·s/mm³, GA·s/m³, or GC/m³, but I find these units less intuitive.)
This ordering more closely aligns with those of malleability, ductility, and hardness than with any other property I can think of: gold is the most malleable metal, very nearly the fastest cutting, and soft enough to dent with your teeth (as are lead and magnesium), while tungsten is the brittlest and nearly the hardest, and chromium is actually the hardest and also quite brittle.
If you wanted to design a material to be more rapidly cut by ECM, you’d probably want a composite of two or more phases, such that most of the volume of the material was in a discontinuous phase cemented together by a metallic continuous phase, and you could electrolytically cut the continuous phase without having to cut the discontinuous phase. The discontinuous phase might be a liquid or gas, making the material a gel or foam; it might be some other conductive substance, such as a metal with a more positive electrode potential, in which case it would need to be physically removed from the cut for it to proceed; or it might be an insulator. In any case the grain size of the discontinuous phase would need to be smaller than the desired cuts. A metal volume fraction of 15%, corresponding to a 6× ECM speedup, seems reasonable:
There is some overlap between [metal matrix composites] and cermets, with the latter typically consisting of less than 20% metal by volume.
See below for notes on suitable solid nonconductive reinforcing discontinuous phase materials. Foams are appealing for increasing stiffness without increasing mass or cutting time.
Zirconium is particularly appealing as an electrolyzable matrix material here; though it is not as abundant as iron, aluminum, magnesium, or titanium, it is more abundant (in Earth’s crust) than copper, zinc, nickel, chromium, tin, lead, molybdenum, or tungsten, on par with carbon or vanadium; even as a pure element it is about as strong as steel (230 MPa yield stress, 330 MPa ultimate tensile strength, with Young’s modulus of 94.5 GPa; grade 705 is alloyed with 2.5% niobium to get 500 MPa yield stress, 600 MPa ultimate tensile strength, higher than Zircaloy); while being substantially less dense, having a higher melting point, and being biocompatible; and it should electrolyze twice as fast as iron, copper, nickel, or cobalt — assuming you can sufficiently disrupt its protective oxide layer during electrolysis, a problem which also arises with titanium. You could imagine a zirconium-cemented composite consisting principally of submicron grains of yttrium-stabilized zirconia (assuming cubic zirconia adheres as well to zirconium as the protective oxide layer does) that can be cut electrolytically five times as fast as steel. Zirconium also potentially supports the formation of hardening carbide grains like those in steel, though I’m not sure if there’s a way to form a pearlite-like structure in zirconium. (See Exotic steel analogues in other metals for more thoughts on this theme.)
(Zirconia is notable for its electrical properties, but at room temperature it is an insulator, because its conductance is mediated by the mobility of oxygen ions.)
Metallic magnesium is also appealing here because it has not only a high electrolysis rate but also a very low standard electrode potential (-2.372 V) and many conveniently soluble compounds. It has alloys with reasonable strength: yield strength of casting alloys “typically 75–200 MPa, tensile strength 135–285 MPa … Young’s modulus is 42 GPa.” ASTM A36 steel, for reference, has yield strength 250 MPa, UTS 400–500 MPa, Young’s modulus 200 GPa, so these alloys have a substantial fraction of steel’s strength and (to a lesser degree) stiffness. (Pure magnesium is much weaker, only about 20 MPa, though another source says 65–100 MPa, and some wrought alloys are stronger, as high as 300 MPa yield strength.) Stiffness can be improved with discontinuous reinforcing fillers to a much greater extent than strength. Its greatest drawbacks are its inflammability, its intolerance of high temperatures (worse even than aluminum) and creep. (Fillers tend to eliminate creep.)
Ideally these would be in the form of submicron particles, especially submicron-length whiskers or laminae; they might include carborundum, carbon nanotubes, carbon fibers, halloysite nanotubes, other clays, boron nitride nanotubes, basalt fiber, goethite, asbestos, zirconia, zircon, sapphire, talc, cubic boron nitride, boron carbide, silicon nitride, topaz, diamond, silica, rutile, chrysoberyl, beryl, spinel, mica, aluminum magnesium boride, boron, or iron tetraboride. (Titanium nitride and zirconium nitride are too conductive.) Composites drawing most of their strength from such high-aspect-ratio functional fillers may actually benefit being bonded with a soft, malleable metal (like tin, magnesium, or zinc), rather than a harder, stronger metal (like tungsten, chromium, or cobalt), because, as with intentional weakening of the fiber–matrix bond in ceramic-matrix composites, it allows pullout, impeding crack propagation and distributing the load along the length of the fibers or plates. With this sort of nanostructure it should be possible to take advantage of the extra strength of reinforcement whose thickness is below the critical dimension for flaw-insensitivity.
Laminar functional fillers can enjoy flaw-insensitivity by having only one of their particle dimensions below the critical dimension, and can theoretically provide high strength in two dimensions, thus providing on average twice the strength of the same material as a fibrous filler, but high filler loadings for laminar fillers are only possible by aligning the laminae parallel. I saw a paper about 10 years ago which achieved this with bentonite and PVA (rather than a metal) by depositing them in alternate layers (“layer-by-layer (LBL) assembly”), but I haven’t seen examples since then. (I think some of steel’s strength can be attributed to pearlite and bainite having precisely this structure, with ceramic cementite nanolayers alternating with soft metallic ferrite.) I posted the paper to kragen-fw with the headline “new high-strength composite made of “nanoclay” and PVA”:
Charles Griffiths told me about this October 4 article from Physorg, “New plastic is strong as steel, transparent”:
Apparently, by alternating layers of polyvinyl alcohol and “clay nanosheets”, Nicholas Kotov and a bunch of other people at UMich (many from his own lab), plus some folks at Northwestern (in some earlier research; see below) have fabricated an extremely high-strength composite. It gets its strength from parallel layers of clay nanosheets glued together with thin layers (monolayers?) of PVA. ...
http://www.sciencemag.org/cgi/content/abstract/318/5847/80 doi:10.1126/science.1143176
Science 5 October 2007: Vol. 318. no. 5847, pp. 80-83.
The authors are Paul Podsiadlo, Amit K. Kaushik, Ellen M. Arruda, Anthony M. Waas, Bong Sup Shim, Jiadi Xu, Himabindu Nandivada, Benjamin G. Pumplin, Joerg Lahann, Ayyalusamy Ramamoorthy, and Nicholas A. Kotov, all of whom are from UMich and five of whom are from Kotov’s lab.
In this work, for which Google Scholar finds 1563 citations, by crosslinking the polyvinyl alcohol with glutaraldehyde (widely sold as a disinfectant at 2–2.5% strength under names like Surgibac G and Sertex), they achieved 400 MPa strengths, stronger than many steels. They’d previously done the same thing with a mussel glue amino acid, L-3,4-dihydroxyphenylalanine, achieving lower strengths.
Electrodeposition would seem to offer a low-temperature codeposition route to fabricating such layered structures in bulk rather than a few nanometers at a time: first, compact the mass of filler to a high density, then electrodeposit a metal matrix in its interstices, similar to the molten metal infiltration technique for tungsten carbide, also used for Al/SiC metal matrix composites:
AlSiC metal matrix composites are formed by pressure infiltrating molten aluminum into silicon carbide preforms. This method of casting is typically used in applications where solution requirements include high strength, lightweight, custom CTE and high thermal conductivity. PCC offers AlSiC with a composition varying between 30% to 74% silicon carbide by volume, depending on the application. This flexible material system allows PCC Composites to produce a part that can be tailored to exact solution requirements.
Conceivably electroless plating would work better.
For metal matrix composites or cermets, a crucial question is the adhesion of the metal matrix to the filler; as mentioned above, adhesion that is too strong can propagate cracks into the filler particles, eliminating their flaw-insensitivity, but of course in the limit of weak adhesion the composite is no better than a foam with extra dead weight.
The high filler loadings that would be ideal for electrolytic machinability are more similar to the area of practice generally known as “cermets” than to the area of practice generally known as “metal matrix composites”. Nonconductive reinforcing discontinuous phases used in cermets seem to include sapphire, glucina, magnesia (periclase), zirconia, phosphates of calcium, fluoroaluminosilicate glass, rutile, boron carbide, carborundum, aluminum nitride, sodalite, and quartz.
A truly 2D material like graphene or a MXene would also make a great functional filler for this kind of thing if, like nitrides of titanium and zirconium or like the MAX phases, you could find one that isn’t conductive. The problem with conductive fillers is that, once the surface of the metal is etched, they would screen the electric field from the metallic matrix surface in their interstices, so it would stop being etched.