Retrotechtacular: The Nernst Lamp

After dominating the illumination market for more than a century, it’s easy to think of the glowing filament of the standard incandescent lamp as the only way people found to turn electricity into light. But plenty of fertile minds turned out alternative designs, one of which is the fascinating Nernst lamp, which we’d previously never heard of.

If the name sounds familiar, it’s likely through exposure to [Walther Nernst]’s equation for electrochemistry, or for his “New Heat Theorem” which eventually became the Third Law of Thermodynamics. Pal of [Einstein] and eventual Nobel laureate, [Nernst] was also a bit of a tinkerer, and he came up with a design for an incandescent lamp in 1897 that was twice as efficient as carbon-filament lamps. The video below, from the Edison Tech Center, details the design, which used a ceramic “glower rod” that would incandesce when current flowed through it. The glower, though, was not conductive until it was quite hot, so separate heater coils that gave the glower a start on the process were included; these were switched off by a relay built into the base of the lamp once the glower started conducting.

It’s a complicated design, but its efficiency, coupled with a better light spectrum and the fact that it didn’t need a vacuum bulb since the glower wouldn’t oxidize like a carbon or tungsten filament, gave it certain advantages that let it stake out a decent share of the early market for electric illumination. It was even the light source for one of the first facsimile machines. We find it a very clever use of what were at the time exotic materials, and wonder if this could have lead to something like vacuum tubes without the vacuum.

Thanks to [Zane Atkins] for the tip

55 thoughts on “Retrotechtacular: The Nernst Lamp

      1. That is very neat, so they heat up a chunk of ceramic knife (zirconium dioxide), and then they are current limiting the supply to protect the zirconium dioxide from melting because as it gets hotter its resistance decreases. And because zirconium dioxide is already an oxide it can not oxidise in air.

  1. This was very interesting. Never heard of this before. I wonder if this principle could have been developed further..? Maybe as a replacement for street lights or those mercury vapor lamps. Today we have led’s though.

    1. As I understand, it fell out of favor early because it took a while for the light to come on while incandescent lamp were instant-on. Fluorescent lamp were invented not long after, and as more energy efficient lamp, which made it worth developing further.

  2. “Vacuum tubes without the vacuum” … um nope. The vacuum is not there to prevent the filament from burning through but because electrons would ionize gas molecules from the gasses instead of flowing to the anode/plate. Which is sometimes useful (e.g. in switching tubes – thyratrons or neon lamps) but not useful for amplification.

    There are tubes that work without heated cathode but never without vacuum.

        1. Yes. Duh.

          Vacuum tubes work by emitting thermal electrons out of the filament, which are accelerated by an electric field. Why can’t it work by ionizing (heating) the ambient air and then accelerating those ions? It’s the same principle, only working with opposite polarities.

          1. (Raises hand) Ooo I know! I know! What is because positive ions are still actual atoms and how do you propose to replace those atoms (answer: not with electricity) and what do you do with the free electrons even if this worked Alex?

          2. The positive atoms act as charge carries. If an atom of nitrogen gets heated enough to lose an electron at the hot filament, and the filament is biased at a positive potential, the electron gets taken up by the filament and the positive nitrogen ion starts flying towards the oppositely charged grid electrode where it can pick up an electron and become neutral again. But, like vacuum tubes do, most of these ions will miss the grid and fly past it to the plate where the same thing happens.

            When the nitrogen ion is neutralized, it will diffuse around in the open air until it hits the hot filament again, becomes charged, and starts flying towards the grid electrode.

            Vacuum tubes work with just electrons, but I can’t see why a non-vacuum tube couldn’t work just the same. It’s probably less efficient and lasts a shorter time because of the ion wind blasting at the grid and plate, but it should work.

          3. “the positive nitrogen ion starts flying towards the oppositely charged grid electrode”

            Yeah, uh, no. Now you’ve got nitrogen atoms flying out of a heated rod (made of nitrogen?). Let’s break it down:

            1. Filament gets heated enough to shed an electron.
            2. The filament is biased at a positive potential (which can only be done using electrons)
            3. “the electron gets taken up by the filament” So the recently shed electron returns to the filament it came from
            4. Which causes a nitrogen atom missing at least one electron to start flying (for reasons unknown) to the negatively charged filament, where it acquires at least one electron, meanders around until it [again becomes part of the filament? and] loses at least one electron again, then that electron flies back and the nitrogen atom jets off again.

            It bears reminding that a nitrogen atom, even just the nucleus, is quite a bit heavier than an alpha particle. I think your idea has beaucoup problems.

          4. > Now you’ve got nitrogen atoms flying out of a heated rod (made of nitrogen?).

            No, from the ambient air. Damn, try to understand the point before you comment on it. A hot filament heats the surrounding air enough to ionize molecules of nitrogen that happen to bump into the filament, and these ions can carry current across the non-vacuum tube just like the electrons in a vacuum tube.

          5. Luke, think this one through.
            What’s the ionization energy of nitrogen?
            How hot does that filament have to be to ionize a useful amount of nitrogen thermally? (work this one out: you may be surprised)
            Where does the nitrogen to be ionized come from — random molecules wandering into the filament? (answer: yes) It’s not like you’re boiling them off the filament copiously, like electrons.
            If it’s coming from the ambient gas around the filament, what is the current density or ionization rate you can expect? How dense does the atmosphere need to be to get enough nitrogens bumping into the filament to make a useful ionization rate?
            What is the mean free path of a nitrogen ion at that density?
            What is the drift velocity of the nitrogen ion in the tube at that density and voltage gradient?
            What is the recombination lifetime of a free nitrogen ion at that density?
            How many of those ions survive to be collected at the plate?

            (replace nitrogen with a more suitable gas if you wish, but the answers don’t change usefully).

          6. Luke, you need to get back to Luckenbach Texas and get back to the basics, by which I mean read up on thermionic emission. So many problems, as Paula has explained more eloquently and in more detail than I. I and others can just see that it’s wrong, like dipping an O-ring in icewater, but she gives you avenues to explore so you can see why it’s wrong. I suggest you treat her post as a quiz and post your answers here.

            One other thing: ions don’t carry current, they carry charge. If you don’t understand the difference then JHVH-1 help you. Electricity is as the name suggests, electrons (yes, I know that’s the Classic Comics version).


    1. Tungsten lamps are in argon, not vacuum. IIRC the big fat argon atoms bash any vaporizing tungsten back onto the filament. Otherwise the inside of the glass envelope will be plated with tungsten.

      1. You’re confusing inert argon with the halogen cycle.

        Cheap (glass) light bulbs are backfilled with argon to flush the remaining oxygen out during the evacuation: it’s cheaper and quicker than getting an oxygen-free super-hard vacuum. The argon does nothing to impede the evaporation of tungsten.

        Quartz halogen bulbs are backfilled with a halogen (iodine, bromine) that combines with the evaporated tungsten atoms. The quartz envelope runs hot enough to evaporate any tungsten halide that gets deposited on it. When the tungsten halide eventually hits the hot filament again, it disassociates, leaving the tungsten atom back on the filament and the halogen back into the gas.

        1. My recollection of the explanation pre-dates halogen bulbs. Tungsten filament lamps are at a pressure equal to about 10,000 feet of altitude. It is very easy to pull much better vacuum than that, so why not?

          From the Wiki: “In 1913, Irving Langmuir found that filling a lamp with inert gas instead of a vacuum resulted in twice the luminous efficacy and reduced bulb blackening. “

          1. Ah, you’re correct. I was misinformed by a former manager at a GE light bulb manufacturing plant. It being GE, the cost saving explanation rang true, so given the source I assumed it was accurate. I stand corrected. Thanks. I’ll also wryly note that GE, where Langmuir worked, doesn’t make light bulbs at that plant any more (and soon not anywhere: they announced last month they’re exiting the light bulb business).

          2. I happened yesterday to look at relative efficiencies. I have some Chinese “starfish” I use in a big shop. They make a single standard fixture into a 3 or 5 bulb. I know LED are a lot better ad wasn’t worried about the current, but just to check, incandescents are abut 10 lumens per watt. Florescents are 50 and LEDs are 160! I guess I could put 16 “100” watt LEDs on the same circuit as a single 100W incandescent. The 5 bulb star gives huge amounts of light and still much cheaper than one 100W tungsten. I am going to have to measure the current before I truly believe it :-)

          3. > I could put 16 “100” watt LEDs on the same circuit as a single 100W incandescent.

            Well, no. Mind the power factor and peak current draw. Incandescent bulbs draw sinusoidal current while LEDs draw current at the peaks of the waveform where the voltage exceeds the input filter capacitor voltage (if there is one). Putting 100 Watts of LEDs to replace 100 W of purely resistive loads causes the peak current draw to increase tremendously and that makes makes for power quality problems and other electronics suffering early failures from their PF-correction circuits and filters burning out.

          4. > incandescents are abut 10 lumens per watt

            The irony is that the smaller you make an incandescent bulb, the less efficient it gets. So of course they banned the largest bulbs and left the low-powered ones.

          5. > So of course they banned the largest bulbs and left the low-powered ones.

            Well, I’d bet a lot of those small incandescent bulbs are special purpose for uses LEDs wouldn’t survive (dryers, ovens, stove hoods, …).

  3. The bulb was twice as efficient as carbon filament. That was only impressive until tungsten filaments were invented, which were more than twice as efficient as carbon filaments.

    1. Except that this design has been abandoned since. Not sure doping the ceramic element, exciting it with microwaves has been tried.
      There must be other ways to increase the efficiency.

        1. That’s a really good paper by Toyoda et al. Thermal excitation of resonant structures etched in the hot surface produce spectral peaks that can be different from the Planck peak. It modifies the spectral emissivity (“color”) of the filament to enhance visible spectrum output at the expense of longer wavelengths without having to make the filament hotter. Neat.

    2. Hidrogen+heat? Now I know where the movies got the idea of light bulbs that explode when hit by a shot – or by a ghost!
      Nevertheless is a great article, and a good example of secondary inventions to support a primary intent.

          1. Right, Prince Rupert’s drops style… yeah I forgot they’d actually bang if you dropped them too. Sometime in the late 90s they must have phased in a different glass.

  4. Fun modern use for the Nernst ceramic “filament”: A biowaste resistojet. That’s right: a rocket engine that ‘burns’ astronaut exhaust (CO2, waste water) using electricity to heat the ceramic core. Specific impulse is not too shabby: >200s.

    I’ve always wondered why they don’t put one of these on the butt end of the ISS. They throw several kilograms of waste gas overboard every day, but then carry rocket propellant up to reboost the station several times per year.

    1. I believe they currently strip O2 electrolytically, and dump the hydrogen etc from wastes. However, still no reason why they shouldn’t be using that as reaction mass, microwave excited plasma accelerator deallie or something. They do have over 70kW of solar so it’s not like they don’t have the power to make it worth doing.

      1. They actually dump methane, produced by removing the oxygen from carbon dioxide and water.
        It would take about 1 kW to offset nominal orbital decay of 1 m/s/week with a resistojet. They’d still need something with a bit more oomph for avoidance maneuvers, but that little 0.7N or so continuous thrust could avoid the need for a lot of propellant transfer from earth.

      1. IDK, take one to the bottom of the Marianas trench, drag a sardine in front of it all the way to the surface and maybe it will come out like a Polaris missile.

        On the other hand we could fit an African (or European) swallow with an oxygen tank and see if it can do it from 100,000 feet in a zoom climb.

        1. By the way, to the contrary of a minor sci-fi trope of using canaries to tell if the air is bad, birds and space won’t mix. They need gravity to get their vittles down.

          1. Mostly been seen in scifi short stories, not a whole lot, which is why I tagged it minor, or should that be miner, because they of course referenced the old coal mine use.

          2. So, none? Cool. It’s a figure of speech. I say “a stitch in time saves nine” is a sci-fi trope. Unless there’s sewing in the story it’s just a figure of speech. In each case nothiing particularly sci-fi about it.

            It seems to me that you made the trope (figure of speech) specifically about birds in space when you said “to the contrary of a minor…”

          3. A Niche in Time Saves Stein is a mildly well known story by Asimov though :-D

            By a trope I meant it was a theme used in more than one story, but only a handful, so I called it minor. I guess you would have to plow though a hundred 1940s to 1980s anthologies before you noticed it though.

    2. Ha, we have very different definitions of shabby isp. But not bad for a piss rocket, this is true.

      Run the piss through a nuclear reactor, though. That’s what you’d wanna do.

      1. Why a nuclear reactor? You would get even poorer Isp, because you can’t run a reactor as hot as you can a ceramic heater. The relatively poor Isp is due to the molecular weight of the gas (CO2): Run it with plain water and you’ll get closer to 300. Methane would be even higher but might gum up the works — it might dissociate and leave carbon residue. Plain hydrogen (like NERVA proposed) would be best, of course — Isp over 1000, but it might reduce the ceramic to nothing.

        But why worry about Isp? You’re throwing the mass away anyway. The only limitation is available electrical power, so it’s actually more favorable to use a lower Isp, lower exhaust velocity, lower energy per unit impulse.

        Maybe you want to go rescue that Pu-238 core from Apollo 13 that’s still sitting at the bottom of the Pacific: it’s probably still good for a kilowatt thermal.

        1. You state that you can run a ceramic heater hotter than a reactor. But I feel like the examples you provide run counter to that claim. Either way, the nuclear lightbulb (gas core reactor) is a pretty bright idea ;-)

  5. Chemistry students are likely to have heard of Nernst Glowers for the IR. They are found in Beckman spectrometers and spectrophotometers and need replacement occasionally (couple years).

  6. Silicon carbide resistors also produce a useful amount of light when heated to incandescent temperatures. Like the Nernst lamp, they do not need to exclude air. In fact, they are widely used in furnaces and other gas appliances to light the gas. They are simply switched across a power supply, with no voltage regulation or current limiting.

    Efficiency of these igniters would be low as a light source. But maybe that technology could have produced a usable lamp?

    1. “They are simply switched across a power supply, with no voltage regulation or current limiting.”
      It’s a negative temperature coefficient device. It will self destruct if simply placed across the power supply with no other current limiting.

      In my gas stove, at least, that igniter is wired in series with the gas valve that is the current limiter: once the igniter is hot enough, its resistance drops enough to energize the valve and start the flow of gas. Foolproof, until the igniter ages enough to not get a low enough resistance to actually turn on the gas…

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.