Question: How to interpret the results?

amir129 is asking a question about spectrometry
Follow this topic

by amir129 | January 23, 2018 19:48 | #15594


Hi,

I have a question. I am seeing so many graph sets/ comparison posted on the https://spectralworkbench.org website.

How do you interpret the result? let's say I collected three readings of beet root water with different dilution and uploaded the images to spectral workbench website. I will probably show the graph, but how will I interpret the result of which one has higher concentration? or diluted one?

I don't know much about Science but just trying to help out my daughter for her school project, so please let me where can I get some information about interpreting the graph.

Thanks in advance.



10 Comments

The posted graphs are, mostly, just individual experimental plots. Assuming 1) working spectrometer, 2) has been wavelength-calibrated using a CFL light source, the first step is defining what you are attempting to detect/measure. Eg: The concentration of Betanin (red color of beets and similar veggies) in different veg sources? If you can define what factor is of interest, then you can figure out if there is a way to measure it.

Is this a question? Click here to post it to the Questions page.

Thanks for your reply. I am trying to measure the percent transmission of light as a measure of pigment concentration in the water.

So if i capture screenshots using spectrometer using different concentration then how will the graph tell me concentration. Is it the lower the value in the graphs for red is less concentration?

Thanks

Is this a question? Click here to post it to the Questions page.


Reply to this comment...


Without a "reference standard" the measurements will all be relative; not absolute (like ppm [parts per mission]). So, the relative concentration would be roughly linearly related to the value of the 'peak' in the spectrum for the color associated with the pigment color.

Making such measurements is dependent on 1) having a mechnically stable setup where nothing changes except a single factor (the pigmented fluid), 2) a broad-spectrum light source (not a CFL but at least an incandescent; halogen can be better (Solux 4700K Halogen is 'ideal') so the spectrum isn't dramatically distorted by the light source and 3) setting the system to have a max (but non-clipped) response at the pigment color using clear water as a 'max response' reference (i.e. zero concentration). [Note, however, if ALL of your samples have significant color concentration that reduces the signal, then you might substitute the 'lightest'/'clearest' sample as a 'reference'.] [Also: the reverse can work as well -- with the 'darkest'/'highest concentration' as the reference where all other samples are more dilute..]

One of the measurement issues is the limited dynamic range of the typical PLab spectrometer. You must keep the 'brightest' signal from clipping and the 'dimmest'(most concentrated) from dropping down into the noise where measurement values become very difficult to read. You will have to adjust the light source distance (thus intensity) while trying samples to find the best setup.

Given all of that, the concentration is 'roughly'(for this device) linearly related to the measured value at the color (wavelength) associated with the pigment color. A good test is to start with red food coloring and water. If you measure the peak 'red' wavelength spectral value at 100 and then dilute by 50%, the new spectral value at the same wavelength would double (200) because the concentration dropped by half so twice as much light can get through. By extension, a 1000:1 dilution would be essentially clear (but it might then clip the signal due to the dynamic range limit) and pure food coloring might drop the signal into the noise where it is too dark to measure. So, within the measurement range and simple setup limits concentration and signal have a simple relationship. If you needed to "extend the dynamic range" you could do so using dilution to establish an intermediate reference, but that's an additional kettle of fish.

Thank you very much for such details explanation. I never thought about getting the reference to compare with the result. That is a good idea.

Since this is a school project for my daughter and my background is in IT, not science I'll post more questions as I progress. Specially when we start measuring it and seeing the result.

I am trying to find a cheaper spectrometer. The DIY specotmeter 3.0 is sold out. Any other alternative under $100.

Really appreciate your help!


We're going to start selling the Lego spectrometer with a raspberry pi camera too -- could be a good option: http://publiclab.org/lego


Reply to this comment...


You could buy a webcam from Amazon, use a spare DVD some razor blades, black paper, double-stick tape and build one for about $50 (most of the expense is the webcam).

Here are a few notes on devices I've built.

https://publiclab.org/notes/stoft/09-22-2016/all-paper-n-tape-mock-up-of-plab-3-spectrometer

https://publiclab.org/notes/stoft/09-16-2016/stability-upgrade-mockup-for-plab-spectrometer-3-0

https://publiclab.org/notes/stoft/10-03-2016/plab-3-proto-slit-upgrade

https://publiclab.org/notes/stoft/03-02-2015/plab-3-spectrometer-upgrade-prototype

https://publiclab.org/notes/stoft/10-23-2016/high-rez-webcam-spectrometers

Note: While the above projects were designed to be mechanically compatible with the PLab approach to product, many details can be simplified in a purely experimental setup.

1) The webcam-DVD needs to be stable and the DVD to webcam distance generally as short as possible. 2) The slit should be mechanically 'tied' to the webcam-DVD assembly so the optical path remains stable 3) use good double-stick and glue, not velcro and wood (for the rigid parts) rather than paper 4) Otherwise, housing cover parts can be replaced with a black cloth just to keep the light out

The basics are really quite simple and generally the two biggest factors are 1) keeping the 3 parts (webcam, DVD and slit) as rigid as possible and 2) keeping out any stray light. Next is just the alignment of the slit to the DVD/webcam -- but that's a simple 45-deg angle.

Thank you! I'll try to make one and will post the progress.


Reply to this comment...


Great!

1) I'd also suggest using stiff cardboard, Exacto knife, hot-melt glue (or fast-set 'yellow' wood glue

2) use double-stick for the DVD attach so you can place it several times until the alignment to the camera is right (nice horizontal spectral line in the camera image)

3) Build the DVD+Cam as a sub-assembly

4) Build the slit as a sub-assembly (separate them by about 8-10 in ... ideally at the 'focus' but not a very critical distance because the webcam lenses have such a large DOF (depth of field) )

5) Use stiff corrugated cardboard (or wood) as a 'base'

6) Leave the slit and dvd-cam assemblies adjustable on the 'base' until you get a good spectrum centered in the webcam image (you can do this all 'offline' w/o the PLab's software).

7) Once aligned, glue the assemblies down and 'hold/clamp/other' the whole assembly so it can't move with respect to the light source and the samples

Reply to this comment...


I have one more question. The experiment says to use the light setting 530nm:

"Use the spectrophotometer to measure the Absorption and Transmittance percentage for each of the beet jars.Use the light setting 530 nm.Use the set of jars with no beets in them as the blank samples"

Why do they say to use 530nm light setting? If we make the spectrophotometer using DVD using the above steps you suggested, will the result be similar? or it will totally invalidate the test results?

I think if we have solution with different concentration we should see different curve in graph, that should be good enough to prove Absorption and Transmittance percentage. What do you think?

Thanks,

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Questions are good ;-)

Reading between the lines of your description, the "blank" is the Transmission Reference (i.e. the clear container with only water). The light source + reference + spectrometer setup is then adjusted to give a maximum measurement value -- but w/o signal clipping (no signal distortion). They further appear to specify taking the measurement data at 530nm wavelength (you'll know this value once the spectrometer has been calibrated for measurement wavelength using the sharp spectral peaks of the CFL light source spectra). [The CFL source is ONLY used for wavelength-cal, NOT for measurements -- for that, use a broadband source.]

The spectrometer has a usable range of about 400-650nm so the "light red" color at 580 is probably a good fit to the beet juice. By starting with water (the reference with max signal) then introducing a beet juice sample, a reduction in measured signal would be expected. You can collect the data of the entire spectrum so you can see how various colors are affected but only the value at 580 will be reported I assume. If the signal from a sample drops by 90%, then the Transmission is 10%.

Absorbance (A) is calculated as: A = 2 - log10(T%). So, for 10% transmission, the Absorbance = 2 - log10(10) = 2 - 1 = 1. The Absorbance is a Log scale which is useful for scientific plots. [ The '2' is there because for water, T=100% and log10(100) = 2 which means A=0; zero absorbance.] However, it sounds like they are asking for Absorption Percentage and Transmission Percentage so it would be good to check what calculation for Absorption is being requested.

The Lambert-Beers Law states the Absorbance of light through a uniform fluid is A = ELC (E (epsilon) is a unique coefficient, L = length of the optical path and C = concentration. So this means that if the container is twice as thick, or the sample concentration were twice as dense, the Absorption would be doubled -- a linear relationship.

Again, this is just some related information. Hopefully the experiment's documentation will help clarify what calculation is expected.

Reply to this comment...


Log in to comment