Random peaks, spectrophotometer & iridescence

[Samuels-art]

Dear PH,

Recently I have been trying to determine if a surface is iridescence or not. The most quantitative method to achieve this is to find the maximum reflective peak, this is also highly repeatable and worked well.

The idea is to move a sample at all angles of possible viewing geometry and find the maximum reflection peak and that measurement (nm) is the so called colour! This is accepted in biological systems, the field I am working in.

Having measured the surface I found the data to have a very sharp peak where I can clearly see there shouldn't be one. You can see in the attached file there are two peaks that I believe should not be there as I can see that there is no represented colour on the surface no matter what angle. The peaks that are 'weird' are at 490nm and 660nm. The integration time for the Spectrophotometer was at 100, the average scans at 5 and the Box CAr Width was also at 5.

My question is this, is there any known reason why I should be getting these random peaks? Could it be the intergration time, something I don't fully understand anyway or could it be something to do with iridescence as a phenomenon?

Thank you in advance and kind regards

Sam

 

[Baluncore]

Maybe you are seeing a specular reflection from a surface somewhere in the field.

 

[Samuels-art]

Hey, thank you for your reply, would the specular highlight not be the seen at the same wavelengths as the actual color of the surface, in this case blue or green?

 

[ehild]

Those sharp peaks are artifacts. Ozone and water vapour in the atmosphere have sharp peaks in that range of wavelengths. The spectrometer output is the ratio of detected intensity with and without the sample. Because of the sample, the light travels different length as without the sample, and also some reflected radiation also reaches the detector: so these peaks do not quite cancel. Just ignore them. ehild

 

[Andy Resnick]

Recently I have been trying to determine if a surface is iridescence or not. The most quantitative method to achieve this is to find the maximum reflective peak, this is also highly repeatable and worked well.

I agree with others, the narrow peaks are spurious and can be ignored. Measuring iridescence is more complicated than I thought- the origin of color is due to physical structure as opposed to (say) pigment, and I could not find an ASTM standard that applies.

Because the origin of iridescence is due to interference and diffraction, the details of your illumination geometry and coherence may affect the measurements. Interesting problem!

 

[Samuels-art]

Dear all,
Thank you for the replies, if anyone is interested further in the quantification of iridescence in a biological field then please refer to the attached paper. Does anyone know if there is a process either an R function or something where I can remove this noise seen in the graph.

I am working with a beetle species that I have just identified as being iridescent which means I can publish quite soon.

I have another problem that may interest some of you guys :). The measurements in the paper are taken from a flat surface, I am trying to think of a way to quantify structural colours that have a curve like my beetles. If anyone has an idea how this could be possible I would be very interested to hear from you.

Kind Regards

Sam

 

[ehild]

Try to change the measuring parameters. Slower scan, longer integration time, lower resolution. I do not know what your data are as you did not give the units. Your spectra look very noisy.

ehild

 

[Samuels-art]

Hi there again, and thank you for the interest and replies.

I am attaching some data so you can see the problem, I am not sure if it is the exact same data set as I have a lot of them, but they all have the same noise and peaks.

units are % reflection and nm.


Kind Regards

Sam

 

[ehild]

I would try to average more spectra, say 20 or so, to decrease the noise.
I think you use a small diaphragm for the incident light when taking the reflectance?
Anyway, the spectral range should not exceed 350-800 nm. The artifacts can be cut out from the spectrum. Or you can cheat a bit, replacing the reflectance data with more reliable ones.
With your data processing program, Excel or Origin, you can also smooth out the noise by "adjacent averaging".
See attached spectra. The first is is the original one, the second is the one where I replaced the impossible data with some value in range, and the third is the smoothed one. I used Origin.

ehild

 

[Samuels-art]

Dear ehild, sorry I never thanked you earlier, I was distracted by yet another set of issues from the same data set and have been working on that before I could consider smoothing. I thought it would take a lot less time to sort my problem but unfortunately not.

Anyway, thank you very much for taking the time and suggesting the smoothing techniques. Very much appreciated.

Kind Regards

Sam