I was thinking about soldering irons, temperature controlled high-end soldering irons and primitive ones consisting of a chunk of metal heated in a propane torch, and phase-change materials, and it occurred to me that with a phase-change material, you could get very tight temperature control of the primitive kind of soldering iron heated in a propane torch — perhaps very useful for Ghettobotics: making robots out of trash.
Electronics soldering requires heating the junction to the melting point of the solder — 183° for 63–37 tin–lead solder, according to Filling hollow FDM things with other materials and Wikipedia’s soldering article. To achieve this, the soldering iron has to be hotter than 183°, because if the iron is at 183°, it would take infinite time for the joint to reach 183°, even in a perfectly insulated vacuum; in a situation where you’re losing heat to radiation and air, it will never reach it. So the iron needs to be hotter than the temperature needed at the junction.
But the junction can’t get too far above the melting temperature, in order to reduce damage to the electronic components. Traditional through-hole leaded packages were only connected to the solder joint with a thin copper wire, but surface-mount components are often entirely leadless, and so experience much faster heat flow from the joint. So it’s important for the iron not to be too hot.
Wikipedia’s reflow soldering article tells me that normal reflow temperatures are 20°–40° above the solder’s “liquidus” (the high end of the melting temperature range).
Thermostat-controlled soldering irons are now becoming standard — because the necessary electronics are now much cheaper than the rest of the iron, because of the trickier nature of surface-mount soldering, and because the higher temperatures required for lead-free solder leave less room for error. The other advantage they have, aside from not overheating, is that their tips don’t cool down when you touch them to a joint; the iron kicks up the current to compensate for the heat loss, which is an enormous help when you’re trying to desolder a power transistor from a big ground pour.
Eutectic tin–silver–copper solder melts at 217° and thus requires higher soldering temperatures. Other lead-free solders are in the 211°–227° range, including tin–copper, tin–silver, tin–silver–copper–zinc, and tin–silver–copper–manganese.
Phase-change materials serve as reservoirs of heat over the relatively narrow temperature range of their phase change — in the case of melting, the range between the solidus and liquidus.
For a eutectic, the liquidus and solidus are equal. By adding more of one or the other metal to the eutectic, the liquidus is increased while the solidus remains unchanged; in between, the equilibrium state is a slush with crystals enriched in the excess ingredient mixed with a liquid eutectic. So, for example, 70–30 tin–lead solder has the same 183° solidus as the eutectic, but its liquidus is 193°, and 40–60 tin–lead solder (60% lead) has an even higher liquidus of 247°, which is actually higher than the liquidus of pure tin at 231.93°.
Suppose you have 10 mℓ of 40–60 tin–lead sealed in a container, perhaps one made of 400-μm-thick copper flash-plated with nickel and then plated with, say, 10 μm of iron, the way soldering-iron tips are made. If you heat the container to 247°, the contents will be entirely liquid; if you allow it to cool uniformly to 183°, the contents will have entirely solidified. For its density, let’s say 9 g/cc, since molten lead is 10.7 g/cc, room-temperature lead is 11.3 g/cc, and molten tin is 7.0 g/cc, so this is about 90 g of molten metal, which will yield up its heat of fusion as it solidifies; I don’t know what the eutectic’s is, but lead’s is 4.77 kJ/mol (= 23.0 J/g) and tin’s is 7.03 kJ/mol (= 59.2 J/g), so let’s suppose the mix is somewhere around 20 J/g; that gives us 1800 J of heat released through this transition. (Plus the small amount of sensible heat from the 60° temperature change, about 0.2 J/g/K (lead being 0.129 and tin being 0.227), adding another 18 J.)
A typical non-temperature-controlled hand soldering iron is 15–30 W, although soldering guns are commonly 150 W. 1800 J at 150 W is 12 seconds of soldering; at 30 W it’s 60 seconds of soldering.
So you could heat up such a reservoir (with fire or whatever) and then solder with it for a few seconds to minutes while it stays hot enough to melt tin–lead solder — up to a total mass of solder similar to the 45 g it contains — or a shorter time while it remains hot enough to melt lead-free solder.
Since solidified solder is a much poorer conductor of heat than copper (401 W/m/K for copper, vs. 35.3 W/m/K for lead, 66.8 W/m/K for tin, so solder is probably somewhere in that neighborhood), some copper bars inside the can might be useful in conducting heat from its interior to its surface. In the other direction, some fiberglass or porous ceramic around the can might be useful in preventing heat loss through other than the tip.
I have one of those goofy multicolored ballpoint pens, with eight sliders near the back end to select the color. It’s comfortable to use, but it’s toward the chunky end of what’s comfortable; it’s about 17 mm in diameter and 130 mm long, and the tip curves down to a cone of about 60° included angle; so its total volume is about 30 milliliters.
Electric soldering irons are normally held much further from the tip, which I assume is because the old thermostat-less type only had its temperature limited by losing heat to the air; if they had only a little area exposed, then they could only use a very small amount of power, and consequently would heat up slowly and require very low duty cycles. But this would seem to be unnecessary with any kind of thermal regulation. Certainly it is possible to insulate the iron sufficiently that you can grasp it quite near the tip, and this would be advantageous. So this pen is probably a reasonable way to estimate the comfortable volume that can be used for a phase-change soldering iron.
However, if my estimate of solder’s density above is in the ballpark, a reservoir of 30 mℓ would weigh some 270 g — in folk units, nearly a pound. Finely manipulating such a heavy reservoir would probably tire the hand before long (I did a brief test with a half-full 600mℓ plastic coke bottle, and found fatigue but no diminution of dexterity), so most likely the reservoir-based iron should be smaller than this.
If you only wanted to solder traditional tin–lead solder, maybe instead of a 40–60 tin–lead phase-change reservoir which needs to be heated up to 247° to be fully melted, you should use 70–30 tin–lead, which is fully melted at 193°. This gives you a much more precise temperature, dramatically reducing the maximum temperature the delicate electronic components can reach — but the temperature is a bit low, meaning that heat flow will be slow, and this can actually increase the risk to the components by giving them too much time to heat up.
Better choices for tin–lead soldering might include Sn 89%, Zn 8%, Bi 3% (191°–198°), which has the advantage of being entirely non-precious-metal; Pb 60%, In 40% (195°–225°); the tin–zinc eutectic Sn 91%, Zn 9% (199°); and the near-eutectic ternary alloy Sn 86.9%, In 10%, Ag 3.1% (204°–205°), which is 87% non-precious-metal. (Indium and silver are about the same price.) For lead-free soldering, pure tin (232°, as I said above) or the tin–silver eutectic (3.5% silver, 221°) might be the best phase-change material.
As described in Hot oil cutter, you can heat a thin pipe to a consistent high temperature by pumping hot oil through it; as described in Coolants, sunflower oil, glycerin, and mineral oil are all liquid and fairly stable at the temperatures under discussion as well as at room temperature. So you could perhaps maintain a large “solder pot” partly molten, with a thick coolant pipe serpentining through it to carry off the heat of fusion, connected to a thinner coolant pipe to heat the iron tip. However, this is probably more elaborate than a thermostatically-controlled electric tip would be, so probably isn’t justified.
I suggested above using a thin can of copper to hold the phase-change alloy, but that’s probably unnecessary; a thick can of steel would surely work just as well, and would be cheaper, stronger, stiffer, and not at risk of dissolving in the phase-change medium. As mentioned above, copper’s thermal conductivity is 401 W/m/K, so sending 150 W through a 400-μm wall with a 20° temperature difference would require a 3.8-millimeter-diameter circle of wall with an area of 7.5 mm²; this is about the diameter of a soldering tip, so you could reasonably just expose a conical tip of the can as the soldering tip. That’s the only part of the can that would need to be copper, though.
A more sensible approach, though, is to use a separate conical solid-copper tip to conduct the heat, which communicates with the can through a much larger contact area, at which point the can itself can be steel — despite its lower thermal conductivity (80.4 W/m/K for iron), the can is no longer contributing the majority of the thermal resistance. This also makes it practical to use the plated copper-bar or plated copper-sheet conductors mentioned earlier to conduct heat out of the center of the can.
Earlier I talked about using copper bars to conduct the heat from the midst of the solidifying reservoir to the tip where it is used, because of copper’s five-times-higher thermal conductivity. Thinking about conducting heat lengthwise through a long cylinder, if the copper is to conduct more heat than the rest of the reservoir contents — much less far more heat — it needs to have at least a fifth of the cross-sectional area. That suggests that, for example, if a cylindrical reservoir is to have a diameter of 15 mm, the copper bar must have a diameter of 6.7 mm. That’s not enough to make the idea impractical, but it’s a fucking hell of a lot of copper.
Consider, though, the total power we can get. Suppose my reservoir is 100 mm long and 17 mm in diameter, a bit smaller than the multicolored pen mentioned earlier. If nearly all of it is filled with a massive copper bar, the most heat we can get from the middle of it to the end at a ΔT of 20 K is 401 (W/m/K) · 20 K · 2π(8½ mm)² / (50 mm) = 72 W. This is barely adequate, and becomes less so once more realistic amounts of copper are considered.
Materials are known with higher heat conductivity than copper. They are silver, diamond, carbon nanotubes, and graphene. At present none of these is a practicable alternative.
A possible alternative is the “heat pipe”, called by Wollaston, its inventor, the “cryophorus”. This is an elongated, evacuated chamber, with a little liquid coolant in equilibrium with its vapor, in modern realizations with a wick to carry the liquid throughout the system. When one end of the chamber is cooler than the other, the vapor condenses at that end, lowering the pressure until the liquid vaporizes equally fast at the other; with the wick, this is an endless cycle. Because the vapor carrying the enthalpy of vaporization from one end to the other moves bodily through space, it can travel considerably faster than heat diffusing through a body — the heat transfer power is almost independent of the distance the heat must be transferred.
Most often, heat pipes are used near room temperature and employ water as the coolant; in theory, this works from water’s triple point at 0.01° up to its critical point at about 374°, but as I understand it, water is normally only used up to about 200°, perhaps to avoid dangerously high pressures.
Heat pipes don’t suffer from the drawbacks of the hot-oil approach mentioned earlier, in that they don’t need any pumping and don’t contain any moving parts.
Alternatively, perhaps a much shorter, more bulbous heat reservoir could enable mere copper to deliver the heat adequately.