Re: Standard deviation of beamsplitter ratio

JohnTS wrote:

Hi to all,
I am doing experiments where I measure the transmittance of thin metal films as a function of light intensity at different wavelengths (the name of the technique is z-scan). I split a reference to normalize my measurements, and I measure signal (the portion of the beam that is transmitted by the metal film) and reference with two identical silicon detectors. The beam is constituted by ~ 20 ps pulses at 10 Hz repetition frequency.

I have a question on how to make the std of the signal/reference (S/R) ratio as small as possible; I have to be more precise: I am using an OPA pumped by the third harmonic of an Nd:YAG. I spatially filter the OPA beam using a 15 micron pinhole. Now here comes the problem: if I tune the OPA to 532 nm the std of the S/R is around 1% while if I tune at another wavelength (for example 540) it is 4% to 6%. In my opinion it is clear what is the critical difference between the two wavelengths: at 532 nm the OPA is seeded by a little bit of light that leaks out from the second harmonic of the laser, while at 540 the OPA is seeded by noise, so it is more unstable. The thing that I don't understand is by what mechanism precisely the higher instability in the pulses makes the std of S/R worse. In fact I can, using the acquisition software, select a subset of the pulses (for example the pulses in which the reference falls within +/- 10% of a given value) but this does not have much effect on the std of the S/R.

The best explanation that I have for the moment is the following: I did not build a good enough spatial filter and between different shots there are spatial variations in addition to energy variations. The splitting ratio is not a constant but is a function of the spatial shape of the pulse, because of small imperfections in the microscope cover slip that I use to split the pulses and the detectors do not do anything else but register this variation; in addition there could be some effects of the spatial shape of the pulse on the response of the detector (which has a large area), but my guess is that this is a smaller contribution (note: the problems are exactly the same if I use a wedge beamsplitter in which there are no interference effects between the two sides, so I would say that the fact that the reflection from the beamsplitter contains fringes is not critical; the fringes do not appear in the portion of the beam that is transmitted by the cover slip because the reflection is small).

I am going to observe the spatial shape of the pulses with a camera to check my hypothesis (in fact by eye nothing is evident), but in the meantime could someone offer some comment on this?

Thanks - Giovanni

Is this a lamp-pumped system? I have a somewhat similar setup, with a seeded OPG tunable from 420 microns to 2.4 um (with a difference frequency generator, currently broken, that gets it out to 10 um).

You're probably right about the seeding, but there's another effect too--the correlations in the noise of the second and third harmonics.

If you triple directly, the TH energy is proportional to the cube of the pump energy. Thus if a pump pulse is (1+epsilon) times the average energy, the THG energy goes as (1+3*epsilon). Most triplers are actually doublers followed by sum frequency mixers, which makes the whole thing quieter:

pump: (1+epsilon)
SH: (1+2*epsilon)
After SHG, the pump is at ~1-epsilon due to the quadratically increased conversion efficiency, so the sum frequency (TH) energy goes as the product of the SH energy times the remaining pump:

TH: ~(1-epsilon)(1+2*epsilon) = ~(1+epsilon).

Thus the TH energy variation is about the same as the pump's. Something similar may be happening in your setup: with a pump pulse of 1+epsilon, your TH is ~1+epsilon, and if your residual SH is ~1-epsilon, the two will tend to cancel.

From the above oversimplified argument, one might expect that the residual SH in the TH beam would go as 1+3*epsilon, although that might still be quieter than building up from noise.

Cheers,

Phil Hobbs
.

Relevant Pages

  • Re: Standing-Wave Current vs Traveling-Wave Current
    ... explanation of why it would cause waves to bounce off each other. ... During the initial turn-on of the system, energy does ... And yet before the pulses collide ... collide had to reverse direction in order to conserve power, energy, ...
    (rec.radio.amateur.antenna)
  • Re: Energy conservation in moving frames
    ... Suppose there is a spacecraft headed towards earth. ... These pulses have a duration T when measured by an on-board ... the total energy of the pulses is ...
    (sci.physics)
  • Standard deviation of beamsplitter ratio
    ... as a function of light intensity at different wavelengths (the name of the ... I spatially filter the OPA beam using a 15 ... instability in the pulses makes the std of S/R worse. ... not a constant but is a function of the spatial shape of the pulse, ...
    (sci.optics)
  • Re: Standing-Wave Current vs Traveling-Wave Current
    ... arriving at the right end will be similar to the ... The difficulty with saying THE pulses passed ... through each other arises with the energy. ... > charge: like charge repel. ...
    (rec.radio.amateur.antenna)
  • Re: Standing-Wave Current vs Traveling-Wave Current
    ... explanation means that Pis not equal to Vtimes I, ... current waves travel down the line superpose, ... that the wave moves energy. ... And yet before the pulses collide ...
    (rec.radio.amateur.antenna)