tag:blogger.com,1999:blog-5624373746968694430.post3363886824949944540..comments2009-07-30T08:47:55.320-07:00Comments on THz Microscopy at Ecole Polytechnique: A Few Observations Under N2 ConditionsAlexhttp://www.blogger.com/profile/12978898101212860021noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-5624373746968694430.post-35645205681470214162009-07-27T03:53:15.089-07:002009-07-27T03:53:15.089-07:00Thank you for the comments on the N2 data -- unfor...Thank you for the comments on the N2 data -- unfortunately, I think I will not have any time to look more closely at this effect. I would really like to try what you mentioned with scanning with N2 and then opening the lid right away and taking more scans to use in comparison.<br /><br />I think that for now I will have to maybe take some scans with small time frame on the ellipsometer setup and try to determine waist size as a function of frequency. I would also like to look at some of the data from before but with truncating the data so as to clean up the spectrum a bit.Alexhttps://www.blogger.com/profile/12978898101212860021noreply@blogger.comtag:blogger.com,1999:blog-5624373746968694430.post-90018015954809860362009-07-21T12:35:48.638-07:002009-07-21T12:35:48.638-07:00I think I know part of the problem with the FFT an...I think I know part of the problem with the FFT and the crazy spectra observed. Look at the 3rd figure, with the several time-domain traces. Do you see how abrupt the amplitude changes are between the data points? That means that the data points are too far apart in time compared to the abruptness with which the data is actually varying. In other words, delta t is too large, or the delay stage is moving by too large of an increment between when it pauses to take data. It is likely that this has been done when the long traces (like 35 ps) are acquired, because otherwise you will wait a very long time to get data points that change a negligible amount relative to their neighboring points in the large 35 ps time window. However, when you want to use data over a smaller time span (like 10 ps or less), you can't just take the relatively low-resolution data from the 35-ps window and look at some smaller percentage of it. You need to scan a smaller delay distance at a smaller step size to get time-domain data that is very smooth. The abrupt changes in amplitude that I see from point to point in that 10 ps window (zoomed from the 35 ps window) are most likely leading to the spectral oscillations, analogous to the way in which the Fourier transform of a step function is a sinc function. You don't have complete step functions in that time-domain data, but unless the data is nice and smooth, then the FFT is going to end up being messy. And actually, it doesn't matter if you zoom in the time window as far as the FFT is concerned - if the time time-domain data is choppy on the time scale that is related to the frequency information of interest, then the spectrum will not come out well.<br /><br />In your Fig. 4 time-domain comparison, I wouldn't place too much faith in any comparison that isn't made with data taken by the same person, with the same experimental conditions, and preferably in the course of the same experimental run (i.e., same day). This is because of system (mechanical, laser, etc) instabilities, consistency in experimental parameters, and so forth. e.g., your pulse could be lower amplitude because the laser was down in power by a little bit. With that said, I can observe that your pulse appears to be a bit wider than the one taken in the N2, which would account for the lower peak frequency in your spectrum. You also don't seem to have as much short-period ringing after the initial cycle of radiation, although it is a little difficult to tell with the pulses overlapping as they do. Both of these features could indicate a loss of high frequency due to water absorption. However, these also look to me like symptoms of broad frequency absorption, and not the narrow absorption lines characteristic of water. <br /><br />At any rate, the best way to do this comparison would be to take a measurement with the N2 environment, then open the chamber so that the humid air became the atmosphere, then take another trace right away. If done consecutively without changing anything else, then these traces could be reliably compared.<br /><br />So, for your last experiments, regardless of what they are, I recommend just taking time-domain windows that include the main initial THz transient and no more than a few picoseconds of the following ringing/noise/reflection behavior, probably total windows of about 5-10 ps, with a small step size for the delay rail. If you feel compelled to take longer traces for some reason, you will still need to use small step sizes to get small delta-t values, if you want to FFT the results more cleanly. <br /><br />Bonne chance!JWhithttps://www.blogger.com/profile/09974922396411206566noreply@blogger.com