[selenium23]

Hi. I'm doing absorbance measurements using Ocean Optics QE65000 spectrometer. The spectrum generated is too noisy. I basically have no idea how to troubleshoot the instrument. So far, I only tried manipulating variables I can tweak from the software (e.g., boxcar width and scans to average) but those didn't give significant improvement.

[Arkcon]

From the scale of the image, it seems like you have very little sample.  So, you may have to accept this noise level.  Try increasing the amount of analyate first.  Also check that your cuvette is clean, and you have enough sample to reach the optics.  Then you can try to work with the system options.

[Corribus]

From the scale on the negative end the peak height appears about around 0.4 OD, which is more than enough to get a good, noise-free spectrum (in fact, you shouldn't be too much higher than this - usually around 0.1 OD is a good range). Moreover the shape of the peak looks strange to me. What is this sample?

I suggest running a sample that has a clear and strong absorption peak with a well known extinction coefficient and peak maximum. Do you have any commercial dyes or pigments in your lab? If not, you can buy something pretty cheap to validate the instrument. Also run a blank (solvent only) to make sure the baseline is good (flat).

UV-Vis instruments are pretty low maintenance but lamps do eventually die. If your lamp is near the end of its life, it may not be putting out enough light, which could cause a noisy spectrum even for a strong absorber. Alternatively, if your sample has any turbidity to it, this can also cause noise. This is why I recommend using a standard and a blank to see what those spectra look like - this will help you diagnose whether it is your sample or your instrument.

[Babcock_Hall]

As one moves from larger to smaller wavelengths, the baseline seems to increase slowly, which I think is odd.

[selenium23]

Hi Arkcon. The sample I have in the photo is graphene oxide. I was trying to replicate the concentration we're using in other analyses.

[selenium23]

Hi Corribus. We have an aqueous solution of crystal violet in the lab and I did what you suggested. Shown are the spectra of the blank, 0.02 mM, and 0.002 mM of the solution, respectively.

The 200-400 nm range seem to be very noisy especially at low concentration while long, eratic peaks can be seen at >900 nm. The entire setup (lamp and spectrometer) is quite old so maybe the lamp is almost at the end of its life.

[Corribus]

The shape of your bottom spectrum looks at least reasonable, so that's good.

http://omlc.org/spectra/PhotochemCAD/html/048.html

Don't worry about noise at 900 nm. Very few spectrometers are useful beyond 750. The monochrometer grating and detector are both often very inefficient that far into the NIR. This is normal behavior. In the 200 to 300 nm range usually you are using a different lamp altogether (most UV Vis have 2 - deuterium and halogen - that cover different ranges). It is possible for one to be working well and the other not to be.

I would agree that the spectra in the visible range have more noise than I'm used to. What is the y-axis scale in those images? Can you set an integration time in the software? If so, what is yours set at?

With an isolated time point, it is sometimes hard to know how close you are to the end of a lamps' lives. However many spectrometers have internal validation procedures that you can use to see if the noise level falls within specification, and some also record total lamp hour usage. If you have the instrument's manual, you can look in it to see if there is a function in your software. If in complete doubt, you may consider ordering new lamps and seeing if this helps to improve the situation. If the lamps have been used for several thousand hours, it's probably time to order new ones. They typically do not run more than a few hundred dollars and are usually fairly easy to install. The manufacturer should be able to help you order the right part (or troubleshoot your problem if it comes to that).

[selenium23]

I tried 100 ms and 315000 μs (as recommended) integration times. Scale is at 0-1 OD or up to 3 OD for concentrated solution.

We're using D and W-He lamps. I just looked up the reference spectrum (light source and blank, as shown). It's the D lamp then.

[Corribus]

Just curious what is you resolution set to?