plasmonic frequency divider

Hi,

Also wondering if anyone has any ideas on how to implement a "plasmonic mixer", fed with two light signals (L1 and L2), and having the same functionality as an RF mixer in that the output would be L1 + L2 and optionally L1 - L2.

cheers, Jamie

At one point I was working on making metal-insulator-metal tunnelling varactors. That works by making a junction with one positive barrier height and one negative, e.g. Ni-NiO-Y or Ni-NiO-Ce. As in a semiconductor junction, on the negative-barrier side the charge spills out of the metal into the oxide, and only a huge E field restrains it.

Changing the bias voltage changes the position of the depletion zone, just as in a normal diode, so it should work as a varactor. The reason this is interesting for optical purposes is the huge charge density (rho). Metals have enormously more free electrons than semiconductors, so the inner edge of the oxide on the negative-barrier side has a gigantic charge density as well.

The response speed of such a varactor is fundamentally limited by the plasma frequency in the oxide, which goes as the sqrt(rho). In a metal, that's up in the ultraviolet someplace, so the varactor should respond up into at least the mid-infrared.

What I was hoping to do was to rent all the old varactor parametric devices, such as frequency dividers and paramps, and do the same thing in the infrared. So yes, if you can actually make tunnel junctions with negative barrier heights that are chemically stable, it ought to be possible to make circuit-style frequency dividers.

(I may revisit this if I get any customer interest--I was doing a seedling with a big defence company a couple of years ago, but the follow-on DARPA program never got funded. I think you can do a lot of cool stuff that way.)

Cheers

Phil Hobbs

Jamie M wrote in news:kc0eaj$dh5$ snipped-for-privacy@speranza.aioe.org:

There are some crystals able to do that. I once bought a infrared detector, where a crystal layer doubled/triplet the frequency of the infraared falling on it. It was rather insensetive however, just enough for the laser i was experimenting with. You might try mixing with that non-linear crystal sheet.

That would be very neat. I could imagine, say, for the utmost accuracy, start with a Rb or Cs microwave source, amplify if necessary, then start multiplying harmonics out. By the time you get to infrared, you've got zillions of channels, phase locked to a highly stable and pure reference (depending on how much phase noise adds up in the process, hmm), and the same thing on the receiver side can demodulate each band down in parallel for slower processing (i.e., gigabits/ch into an FPGA). Downsides include, once you get to the 20th harmonic from multipliers and you've no power left, how do you amplify the THz without another amp (as such)? But then, that's what the parametric mumbo jumbo is for.

How wide is the average IR "channel"? Say, a few percent out of 300THz is a *lot* of bandwidth. And the coherent wideband (or comb spectrum) could do very interesting optical interferometry, especially with a phase sensitive detector -- correct me if I'm wrong, but that's kind of a holy grail among optics, right? :)

Tim

I did a subharmonic oscillator with a single diode, can't recall why. It could surely be tuned for a nicer output, but it clearly divides down.

Are there any cases of synchronous fluoresence? That should be possible.

Version 4 SHEET 1 880 680 WIRE 16 80 -16 80 WIRE 64 80 16 80 WIRE 208 80 128 80 WIRE 304 80 208 80 WIRE 368 80 304 80 WIRE 416 80 368 80 WIRE 528 80 480 80 WIRE 592 80 528 80 WIRE -16 128 -16 80 WIRE 208 144 208 80 WIRE 304 144 304 80 WIRE 480 144 480 80 WIRE 592 144 592 80 WIRE 416 160 416 80 WIRE 432 160 416 160 WIRE 432 208 416 208 WIRE 416 240 416 208 WIRE 480 240 480 224 WIRE 480 240 416 240 WIRE -16 272 -16 208 WIRE 208 272 208 208 WIRE 304 272 304 224 WIRE 416 272 416 240 WIRE 592 272 592 224 FLAG -16 272 0 FLAG 208 272 0 FLAG 304 272 0 FLAG 416 272 0 FLAG 592 272 0 FLAG 16 80 IN FLAG 368 80 TANK FLAG 528 80 OUT SYMBOL ind 288 128 R0 SYMATTR InstName L1 SYMATTR Value 0.01 SYMBOL cap 192 144 R0 SYMATTR InstName C1 SYMATTR Value 1µ SYMBOL schottky 128 64 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName D1 SYMATTR Value 1N5819 SYMATTR Description Diode SYMATTR Type diode SYMBOL voltage -16 112 R0 WINDOW 3 -358 -31 Left 0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value SINE(2 2 3185.71 0 0 0 1000) SYMBOL e 480 128 R0 SYMATTR InstName E1 SYMATTR Value 1000 SYMBOL res 576 128 R0 SYMATTR InstName R1 SYMATTR Value 1e6 TEXT -288 152 Left 0 !.tran 0 0.1 0

Am 02.01.2013 05:48, schrieb Jamie M:

I have seen that in action > 10 years ago. Two lasers shining into a receiver diode and on the spectrum analyzer one could see the beat at 60 or 70 GHz or so. They tuned one of the lasers somewhat and the beat frequency moved by 10 or 20 GHz. Nice stunt.

regards, Gerhard

of

^^^^ I put in 1uF, but that didn't work. (Well, only a frequency doubler)

George H.

ed text -

It should indeed be 1 uF.

It's actully an injection-locked parametric oscillator, or something. The gain element is the diode's nonlinear capacitance.

cy of

ere

er

ason

ors,

ions

be

of

OK I don't understand how the 'f' that works.

George H.

ed text -

of

I don't either, but I had a hunch that it would work, and it does.

I don't have to understand things; I just have to make them work.

Hi,

That is really neat, what would the best practical efficiency in the light absorption stage be? Could you make a plasmonic mixer with that stuff (ie with a reference light source as well as the input light source)? If you could and also have good quantum efficiency, that would be great, I have a feeling that there might be big losses in the circuitry though, not sure however, maybe with nano-scale stuff the losses disappear too. Really interesting work you've done!

cheers, Jamie

I'm not sure if this meets your definition of average but DWDM used for communications at (nominally) 1550nm often uses a 100GHz, 50GHz, 25GHz or

The frequency is related to the channel number as

where n is the channel number and M is 0, 1, 2, etc. to suit the grid size.

Here's the relevant ITU-T recommendation for DWDM:

Regards, Allan

Tunnelling free-space mixers have been used for that since the 1970s, I think. Back before the Haensch-Hall comb generator, the NBS/NIST clock guys used to use frequency chains, harmonic-locking some ridiculous methane vapour laser to an equally ridiculous colour centre laser, to a passive cavity-stabilized dye laser, dot dot dot, to measure laser line frequencies using a caesium clock. Possible, but you really have to want to do it. It isn't a plasmonic technique, though.

Cheers

Phil Hobbs

You can make a degenerate optical parametric oscillator (OPO) that divides by two. It's not usually done, though, because it's hard to separate the signal and idler beams (for obvious reasons).

OPOs aren't really synchronous devices, though, because their line width is governed by the phase matching condition in the volume of the crystal, just like a volume hologram. A 1-cm OPO crystal will give you a line width of roughly a wave number (1/cm, i.e. 30 GHz).

Cheers

Phil Hobbs

^^^^

It's those damn Greek mu things again. Not traveling well on Usenet.

Mine come out as "1/u00b5", when copied and pasted, although they appear correctly in the newsreader.

Character set conflicts. I always check the "convert Greek mu" option, some others don't.