I spent Monday working in the lab more with Antoine to try and see a change in photocurrent between having the fs beam (non-THz) incident on the wafer as opposed to blocking it.
[Before this we talked more about chopping the beam and the lifetime of the charge carriers, but the conclusion was to not worry about this too much and instead just read about lock-in amplifiers a little].
As far as what we tried to do in the lab, nothing really worked. We first tried a GaAs wafer in the same THz setup as before (the ellipsometry-esk one). This was done using the same setup as on Friday with the only exception being a variable aperture placed in front of the GaAs wafer which was used to limit the size of the THz beam that was incident on the wafer. This was used today and not on Friday because the Si wafer was much larger in diameter than this GaAs wafer, and so we did not have to worry about part of the beam going around the object and the other part going through. I think we measured a pulse throught this GaAs wafer, but I know that we did not detect any change in photocurrent with the beam blocked vs. unblocked.
We then tried this same thing with another THz microscopy setup, this one being for spectroscopy. We thought that maybe the more tightly-focused THz beam would make a difference -- but we found the same null results as with the other microscopy setup for both the Si and the GaAs.
The final thing that we tried was making the optical path of the two beams (the fs beam used to excite the electrons and the THz beam) almost equal, but still so that the fs beam would be incident on the Si wafer first. This we did on the initial ellipsometry setup. The length of the fs beam was about 330 cm and the length of the THz beam was about 350 cm. The hope was that this would ensure the creation of charge carriers right before the THz beam was on the wafer. However, this again showed us no difference in the signal.
I think that at this point the best thing to do is get some more reading done and try to think more about what the possible problems might be. I also plan on emailing the group who published a paper on this method to ask a few questions: one about depth of modulation and the other about the type of doping used in their design.
Some other things I want to look at are:
* How important is it that the excitation laser be cw (which is what it was in the paper)?
* Is the excitation beam stretched too much (in the x- and y- directions) due to reflections off the mirrors?
* Do we have the right type of semiconductor/laser light combination?
* Is the THz beam focused enough on the wafer?
* Is there simply too much noise to detect such a small effect? And if so, how to remove the noise?
I hope to have some time this week to read through some articles and answer these questions.
Subscribe to:
Post Comments (Atom)
Welcome to research! One often starts off with an experiment they believe should work, only to find that numerous compromises need to be made before an effect is observed. However, once an effect is found, then optimization can take place, and variables can be changed in order to learn more.
ReplyDeleteI have many comments/questions, but not enough time now to formulate most of them.
1) You should be able to see if the spatial profile of the excitation beam is 'stretched' too much. Just look at it on a card with an IR viewer, attenuating the beam as necessary so that it isn't saturating the viewer (blinding it by making the spot appear to be a large, bright ball.
2) What is the wavelength of the excitation beam? It should be shorter than 1.1 um or so for Si, and shorter than 870 nm or so for GaAs.
3) I would actually think that the pulsed beam would be better for excitation than a cw beam, at least for the Si. The high peak power would geerate a lot of carriers, and the carrier lifetime is so long for Si that the high reflective state should remain even when the short laser pulse has turned off.
4) What is the chopping frequency that you are using (also the same as the reference frequency of the lock-in).
5) What kind of GaAs are you using? Semi-insulating, I assume.
In response to some of your questions/comments:
ReplyDelete1. I am able to use the IR card to look at the spatial profile, and I think that we will need to focus this excitation beam to a smaller area (for instance, the Japanese group measured the FWHM of their excitation beam to be 710 µm and 210 µm in the x- and y- directions, respectively).
2. The wavelength of the excitation beam is around 800 nm, though I was not given an exact figure.
3. This makes sense... and the lifetime should still not matter then... or do we have to worry about the path length/timing/etc of the two beams?
4. The chopping frequency is about 300 Hz (maybe a little less). Compared to the repetition rate of the fs laser (70MHz), this allows almost a quarter million pulses through before each chop (233,000).
5. The GaAs that we used was just a sample that we borrowed from someone else in the lab. It was not LT-GaAs, but we are not sure if it was semi-insulating or not... I can try to find this out (and would like to read more into this).
Alex
One thing to keep in mind is that GaAs or Si that is undoped has a background, unintentional doping level. This, however, is not enough to make the material very conductive, so it can therefore be considered insulating and not nearly as reflective as something metal-like. If the semiconductor is intentionally doped, it will probably be somewhat conductive and also somewhat reflective (like, a lot more than you would expect from a Fresnel reflection). So, if the THz transmission you see corresponds to that expected when there is only a Fresnel reflection from the semiconductor, then it is probably not doped, or at least not enough to make it look at all like a metal.
ReplyDeleteOn #3, I am thinking that it is not necessary to time up the arrival of the excitation laser pulses on the semiconductor with the arrival of the THz pulses. This is certainly true for silicon, which has such a long carrier lifetime that a pulsed laser with a 70+ MHz repetition rate will essentially keep the material in a state such that lots of electrons are maintained in the conduction band and the material is highly reflective. When the chopper blocks that excitation beam, then the carriers die out within a millisecond or so of when the light is blocked. I believe that the response with a pulsed excitation beam will be about the same as when a cw beam is used. So, you basically have the Si acting like a shutter, blocking the THz (or a tiny part of it when using a small optical beam) for the whole time that the excitation beam is passed thru the chopper, and passing more of the THz for the whole time when the excitation beam is blocked by the chopper. This would actually suggest to me, however, that they were then just looking at the total THz signal received during the time intervals described by the inverse of 300 Hz.
Can you tell me what the detector is that is used by the Japanese group in their experiment?