My task is the measurement of the reflection spectrum of the surface of a semiconductor structure. I am using a Xenon lamp as the light source. I am using an Ocean Optics USB4000 spectrometer as the detector. The spectral range I am interested in is 350 - 650 nm.
1 The light from the xenon lamp is chopped at 1 kHz.
2. there is a DC component in the spectrum I am analysing - I want to remove this (unfortunately I have no control over the formation of an additional light source).
Could I ask for help/proposal for a measuring system or configuration of an optical spectrometer to analyse only the ground reflection spectrum over time (removing the DC component)?
Dear Piotr Pokryszka ,
to my opinion, it might be a matter of shielding;
some ideas/hints....
a) put your set up into a black box ( i.e. no other light source to be present in the vicinity of the spectrometer);
b) please avoid any (multiple) reflections from the Xe lamp being impinged onto the spectrometer input port:
- shielding the Xe lamp; (especially the ventilation slots)
- shielding of the spectrometer itself (enclosure may not be light-tight; e.g. wrap the spectrometer in Al foil)
- shielding of any back reflection from the chopper --> cover of Xe lamp --> sample surface (or elsewhere)
c) check the transmission of the chopper blades...
d) any EM interference ?
Good luck and
best regards
G.M.
You will need to implement a kind of lock-in detection to remove the background. One possibility would be to use the same chopping also at the spectrometer side. If you acquire data with and without this additional chopping, a weighted difference should remove the background and provide the desired data.
The abovementioned method will work only if the light intensities are stable over the measurement time. A real lock-in detection, that does not have this drawback, would require another detector than the one installed in the spectrometer USB4000.
I understand that in the case of a measurement with a spectrometer, measurements would have to be taken alternately (fixed signal + variable signal) and (fixed signal). Later, these signals would be subtracted from each other, for example, and the resulting signal added together to obtain a higher count.
Can this be solved differently?
A different and better approach would be a light detector providing a signal proportional to the incoming light intensity with a bandwidth of at least 1 kHz, i.e. your copping frequency. By multiplying this signal with the copper signal, or by any other known method, the component proportional to the investigated light signal can be obtained. You may look at this page: https://en.wikipedia.org/wiki/Lock-in_amplifier, this explains the method in detail. You can find also other articles about the lock-in technique or synchronous detection, the latter name is more common in the electronics.
As an example, we designed such system a longer time ago, see https://www.cgc-instruments.com/Optoelectronics. The light source is a laser diode modulated by 50 or 100 kHz and a detector with a band-pass amplifier with a narrow frequency region around the laser frequency, i.e. 50 or 100 kHz ±5-10 kHz. The signal detection is made digitally, an FPGA reads data from ADCs digitizing the outputs of the photodiode and calculates output signals by multiplying the input by a sine wave with the laser frequency and a proper phase. The result is filtered by digital filters and the output provides a clean response w/o virtually any background. The noise is given just by the photodiode, however, if it is exposed to a strong background light (for instance from an ionization vacuum gauge), there is also a contribution due to shot noise.
Dear Piotr,
Your question is very interesting. When doing optical spectroscopy, there is always the need to remove background components. These backgrounds may come from stray light, from electrical offsets in the read-out electronics of the spectrometer sensor, and from the light source itself when it is partially modulated (dc light + ac light).
My laboratory has developed a general purpose optical lock-in spectrometer using one or several compact array spectrometers. Our goal was exactly what you described : remove all the dc components, whatever their origin, and measure the ac spectral components only. It is like having a lock-in amplifier at each spectral output channel.
We published several papers describing this work. We have several good prototypes in our lab. The first one uses two compact array spectrometers working in phase quadrature. The second prototype just uses one compact array spectrometer with triggers and acquisition times adjusted to perform the lock-in demodulation in real-time.
The results are very efficient. There is a huge gain in signal to noise ratio and a powerful rejection of background spectral components. No dark corrections needed any more !
You can find more details in the following papers:
Martinsons et al. “A Fiber-Optic Array Spectrometer With Parallel Multichannel Optical Lock-in Detection and Its Application to in Situ Lighting Measurements.” Journal of Lightwave Technology 41, no. 13 (July 2023): 4489–95. https://doi.org/10.1109/JLT.2023.3244652.
Martinsons et al. “Optical Lock-in Spectrometry Reveals Useful Spectral Features of Temporal Light Modulation in Several Light Source Technologies.” LEUKOS, June 22, 2022, 1–19. https://doi.org/10.1080/15502724.2022.2077754.
Don't hesitate to contact me if you want further information.
Dear Christophe Martinsons
Thank you very much for your reply, I will study your articles. Questions will certainly still show up :).
- Badges
- Science topic