On Twitter Sunday morning, the National Society of Black Physicsts account retweeted this:
— ✨The Solar System✨ (@The_SolarSystem) September 28, 2014
I recognized the title as a likely reference to the use of optical frequency combs as calibration sources for spectrometry, which is awesome stuff. Unfortunately, the story at that link is less awesome than awful. It goes on at some length about the astronomy, then dispenses with the physics in two short paragraphs of joking references to scare-quoted jargon from the AMO side. The end result is less a pointer to fascinating research than an instructive example of what not to do if you're hired to write copy about science outside your area.
The worst part of this is that now I have to take the time to do a better job of this. Which I can ill afford, but reading a description of the comb calibration process as "magical gizmo fun" leaves such a bad taste in my mouth than I can't let it go.
I dunno, at least this gives us a chance to do some physics. I mean, we hardly ever talk any more. I miss you, man... OK, that's a little weird. Anyway, let's get on with this.
Fine. All right, what's the issue here? Well, the astronomy part of that story is, from what I know, reasonably good. It's about the exoplanet-hunting group at Yale, led by Debra Fischer, and their search for ever more Earth-like planets. Their particualr technique is the redshift method, which looks at small changes in the wavelengths of light emitted by distant stars due to the gravitational tug of an orbiting planet on the star. This was the first method used to find extrasolar planets, and it's been used to locate and characterize dozens of planets.
So, when do they zap these planets with lasers? Well, since they're many light-years away, never. The lasers would be used only on Earth, at the telescopes being used to hunt for the planets in the first place.
See, the shift in the spectral lines is tiny, even for a big planet-- at a long-ago talk I saw about this ("long-ago" here meaning "1998"), they were talking about shifts of a quarter of a pixel on the CCD camera they were using to measure the spectrum. The sensitivity has gotten better, of course, but this is still a very demanding process. And what's more, it demands long-term stability if you want to see planets that are Earth-like in both mass and orbit-- you need to watch a star over a period of years, and know that any changes you see are due to its motion, and not drifts due to Earthbound effects.
Can't you just compare the lines from the star to the same element here on Earth? Yes and no. In principle, that's what they do, but there are a lot of complications. Among other things, the atoms emitting light in the stars are, well, in stars, which means they're in a very different environment than we can easily produce here on Earth. There are a lot of other effects that can shift the lines by a little bit, and you need to worry about that stuff.
There's also the fact that you want to know that your spectrometer is behaving nicely over the full range-- that it's not responding in a different way in different parts of the spectrum. So you need some kind of source with lines at lots of different wavelengths, as a check on that. This is traditionally done with lamps filled with a mix of gases-- thorium and argon is a common one.
And what's the problem? Well, there's a bit of black art to the making and maintaining of these-- the people who do it are amazingly good, but when you get to looking for the tiny shifts the exoplanet hunters want, and tracking them over several years, you worry that the calibration lamp will drift due to changes in the pressure, temperature, other gas leaking in, etc..
So, beaming a laser into the telescope works better? Because, like, you can watch the frequency be stable for years? Yes and no. Lasers can do better, not because you watch a single frequency for years, but because you can make a laser that produces a wide range of lines at frequencies you can measure absolutely.
Wait a minute. I thought lasers were a single color of light? Continuous lasers are pretty much monochromatic, it's true, but a pulsed laser is actually a collection of a large number of regularly spaced frequencies, and the shorter the pulse, the wider the range. It uses the same adding-lots-of-waves physics as the Heisenberg Uncertainty Principle video I did for TED-Ed.
But why are there lots of waves? Well, a laser works by sticking something that amplifies light between two mirrors facing each other. The light bounces back and forth, getting amplified a little bit on each pass, and some of the light leaks out on each bounce, because it's impossible to make a perfect mirror.
This leaking-out process is what produces the many different frequencies you get with a real laser. If your cavity is about half a meter long, the time to go from one mirror to the other and back will be around three nanoseconds, so every new light being produced at any given instant is being added to light that was created three nanoseconds earlier, and has been reflected. And also light from six nanoseconds earlier that's made two passes, and nine nanoseconds earlier, etc.
OK, but how does that create multiple frequencies? It doesn't create them, it filters out which frequency are allowed. If the frequency of the light pulse is just right, when new light waves and the reflected light waves come together, they interfere constructively-- the peaks from one align with the peaks from the other, and the two waves reinforce each other. If the frequency is a little bit off, though, the waves interfere destructively-- the peaks of one fill in the valleys of the other, and cancel out.
So, a laser only works at a single special frequency, like I said. No, a laser can only work at any of an infinite number of special frequencies, determined by the length of the cavity and the speed of light. These form a regularly-spaced "comb" (so called because the usual representation of the spectrum (like the figure above) is an array of spikes indicating high intensity at some frequencies and no light in between) of allowed laser modes (a very flexible jargon term that here just means "light of a particular frequency that will interfere constructively in a given laser cavity").
If you use the right amplifier material inside your laser, it will amplify many of these modes, producing light at a wide range of different frequencies. And when you add all those frequencies together, it produces a regular train of short pulses. The time between pulses is equal to the round-trip time for light in the cavity-- effectively, each pulse constructively interferes with the reflected light of previous pulses. This is called a "mode-locked" laser, because the rate at which the pulses occur and the spacing of the modes are both fixed by the length of the cavity.
The length of the pulses depends on the length and the properties of the amplifying medium. The wider the range of frequencies you amplify, the shorter the pulse, and vice versa. It turns out that if you make a laser whose pulses are only a femtosecond or so in length (that is, 0.000000000000001s), the range of frequencies spans the entire visible range of the spectrum. You can think of the pulse as the sum of millions of little lasers with slightly different frequencies that are all locked together.
Oh, and that's what you need for a spectroscopy calibration! Right. The cool thing about these comb sources is that you can lock their frequency in an absolute sense-- you compare one of the modes of the comb to the light absorbed or emitted by a particular atom, and adjust your cavity length as needed to keep that one mode at the same frequency as the atoms. This gives you a comb of frequencies whose frequencies are known as well as the frequency of your reference atoms; if you're really clever, you use something like an atomic clock as your reference and then you know the frequency of any given mode to ridiculous precision.
This is where you cite some old blog posts about clock stuff, right? Right. Such as this cool measurement of relativistic effects with a pair of aluminum ion clocks, or this demonstration of time transfer good to eighteen or nineteen decimal places.
They don't demand quite this level of precision for exoplanet-hunting, but even a less technically demanding comb system can give you a broad range of regularly spaced lines that can be tied to an atomic reference. This is exactly what you want to calibrate Doppler shift measurements over a wide range of frequencies. And the precision ultimately traces to the stability of atomic clocks, which are what we use to define time, and thus guaranteed to be stable unless the fine structure constant starts doing wacky things.
So, that's what they brushed off as "magical gizmo fun"? Not quite. The specific "magical gizmo" reference was to a second step of the process, that uses the same physics. You see, a mode-locked laser made at a reasonable size will produce too many modes for a typical telescope spectrometer to resolve cleanly, so they need to get rid of some of them. They do this by using a Fabry-Perot cavity, which is just a pair of mirrors facing each other with nothing in between them-- a laser minus the amplification medium.
The same bouncing-pulse physics applies to the empty cavity-- only very specific frequencies will interfere constructively, and make it through to the other side of the cavity. So they set up a Fabry-Perot that only transmits light at special frequencies spaced by, say, a hundred times the spacing between laser modes. This gets you every hundredth laser mode, which is a spacing that works better in astronomical instruments.
That's a gizmo, all right, but it doesn't sound all that magical. Or fun, for that matter. "Fun" is a matter of personal taste, but science is, after all, magic without lies. The word "magical" should never be used to gloss over actual science content. Not even ironically. That's what annoyed me enough about this story to write all this up.
To be fair, it did take you about 1500 words to explain all this. You can hardly expect them to devote that much space to laser physics. No, but I don't think it's too much to expect some explanation, rather than just dumb jokes making fun of the jargon terms. Here's a quick attempt at something better, in approximately the same amount of space:
Fischer's group plans to calibrate their system using an optical frequency comb, a special laser that produces ultra-fast pulses of light only a few femtoseconds in duration, containing many different frequencies. The "modes" of this laser are evenly spaced across a wide range of frequencies in the red region of the visible spectrum where the EXPRES spectrometer operates. Setting the frequency of one of these modes to the natural absorption of a particular atomic transition fixes the frequency of all the modes of the laser.
The original laser actually provides too much of a good thing-- it has so many modes that they run together on the spectrometer. They fix this problem by using a Fabry-Perot etalon, a device consisting of a pair of mirrors facing each other. The light waves bouncing around between the mirrors interfere with each other, which filters out all but every Nth mode of the original laser, giving a comb with a wider spacing between modes. The end result is a regularly spaced set of lines across the XX nanometer range of interest, each line with stability comparable to an atomic clock. This is an ideal calibration source for the long-term measurements needed to pick out the tiny wobble of an Earth-like planet in an Earth-like orbit over many years of observations.
That's not perfect, I know, but it took me half an hour to write, and it's not insultingly dumb. With some revision (and some data to fill in the experimental parameters that the original article doesn't see fit to give us), it could be compact but also informative.
Yeah, I see what you mean. And keep in mind, I banged this out on the basis of background knowledge only, having no contact with the actual group doing the research. You would think that a writer with access to the research group in question-- and he definitely had that, because there are quotes from them earlier in the article-- would be able to do better. This ought to be better, and the fact that it isn't reflects very poorly on the writer, and on the Planetary Society for not demanding better.
I find this particularly annoying because it has this "all these big words! Optical physics is Hard!" vibe to it. It would be easy enough to do the same thing with the astronomy side, cracking wise about stellar classifications and the like, but they would never consider doing that, because that's their business. When it comes to physics, though, they have no qualms about dropping into Barbie mode, and I find that really annoying.
Well, I'm sorry you're annoyed, but it was nice talking physics again. Let's not wait so long next time, ok? I'll try, but I don't have as much control over my schedule as I would like. I sincerely hope, though, that our next conversation originates in something more positive than flippant and lazy science writing.
Very nice explanation of the use of frequency combs for exoplanet searches. I think the article you referenced was actually so bad that it didn't even convey, to an expert in the field, what they are doing. The Yale group is trying to do away the complexity of a frequency comb and simply use a stabilized Fabry-Perot and white light to generate the calibration lines. See their more detailed webpage -
or a a paper on it -
About lasers in general:
I didn't realize that a laser could put out multiple frequencies at once until you explained it, but it makes sense. Actually, the clearest explanation to me of what is coherent light is William Beaty's: not that it's "in phase", but that it's just "pointsource" light, *spatially* coherent: