The spectrum you show is the result of dividing the spectrum of light transmitted through the sample with a reference spectrum. The noise in your spectrum is very typical of a reference spectrum that is nearly 0 at UV wavelengths--essentially you are dividing by noise at short end of the spectrum, which is why you get those wild gyrations. There could be four reasons for this problem:

1. The light source is very weak in the UV. Check that the UV lamp (usually a deuterium lamp) is on and not broken. The lamp could also be close to the end of its life, which would make it emit much less light than when it was new.

2. The detector you are using is not sufficiently sensitive in the UV. This is more likely to be a problem at long wavelengths, but I mention it here anyway to be complete. Some spectrophotometers have multiple detectors, and you need to make sure you have selected the right one in the software.

3. The spectrophotometer is not selecting the right filter when scanning through UV wavelengths. The internal filters are there to suppress higher order diffractions from the monochromator grating, and need to change as the wavelength changes. In most cases this should be done automatically, so if this is the problem your instrument may need to be serviced.

3. The reference sample has very low transmission in the UV. In my experience, this is the most likely reason, at least if you are supplying a blank sample in the reference beam. Make sure your cuvette (or substrate if you are looking at a film or a solid) and any solvent you are using actually transmit light below 300 nm.

I suppose that the signal is too low for the spectrophotometer sensibility. The problem could be enhanced by a correction curve.

Really the only way to verify is to measure the raw intensity in that spectral range, either in the power transmission curve with nothing in the way vs. a previous measurement or just doing longer collection as your signal:noise ratio will go up linearly with acquisition time and go up by the square root of the number of signal averages. So increase your acquistion time per scan until you are 2/3-3/4 of the way to the maximum (saturation) point of the detector and then just increase signal averages. However, judging from your curves, I would say that your sample is just too opaque and therefore you effectively have no signal to reference too.

I agree, its an equipment issue. Check the specifications of your detector. The noisy region might be beyond the range of your detector. Also, since this is %R, check how your spectrometer computes this. In the spectrometer I use (Ocean Optics USB 2000) it uses an equation using reference values. Sometimes the denominator becomes zero so the graph becomes noisy.

The noise comes very likely from the equipment. Is there any change of detector/lamp at 300 nm? I have no experience with the Varian spectrophotometer, but the Perkin-Elmer Lambda series also show this behaviour spuriously

You can open the spectral bandwidth, increasing the step size. If the concentration is too low, you can increase the pathlength.

do you have one or two bems in your spectrometer (%R suggest that you measured absorption / transmission)? Precise suggestion is possible if you answer my question.

Seeing your data I can be sure that your sample is too concentrated ! Reduce the pathlength or dilute your sample to be able to go down below 300nm. Remember data only be credited when A= 1 or 1 you have stray light which is an artifact. What is the molecular structure of the sample you are measuring ? What functional group absorbs below 300nm ? This should be your first step before measuring its UV !

I don't wish to advertise but Varian is the best commercial instrument if one know how to operate it :D

This might sound very simplistic, but it is a quick check: is your deuterium lamp on and working? I have seen this before when I have a lamp change at 300 nm and my D2 lamp is burnt out.

To examine, whether the noise came from the sample, you can take the spectra using pure silica or phosphate glasses (make sure in such glasses there is no any alkali components). If you found no noise at around 300 nm with these glasses, then the noise come from your sample. If you found noise at 300 nm, then the noise comes from the equipment.

The spectrum you show is the result of dividing the spectrum of light transmitted through the sample with a reference spectrum. The noise in your spectrum is very typical of a reference spectrum that is nearly 0 at UV wavelengths--essentially you are dividing by noise at short end of the spectrum, which is why you get those wild gyrations. There could be four reasons for this problem:

1. The light source is very weak in the UV. Check that the UV lamp (usually a deuterium lamp) is on and not broken. The lamp could also be close to the end of its life, which would make it emit much less light than when it was new.

2. The detector you are using is not sufficiently sensitive in the UV. This is more likely to be a problem at long wavelengths, but I mention it here anyway to be complete. Some spectrophotometers have multiple detectors, and you need to make sure you have selected the right one in the software.

3. The spectrophotometer is not selecting the right filter when scanning through UV wavelengths. The internal filters are there to suppress higher order diffractions from the monochromator grating, and need to change as the wavelength changes. In most cases this should be done automatically, so if this is the problem your instrument may need to be serviced.

3. The reference sample has very low transmission in the UV. In my experience, this is the most likely reason, at least if you are supplying a blank sample in the reference beam. Make sure your cuvette (or substrate if you are looking at a film or a solid) and any solvent you are using actually transmit light below 300 nm.

this noise is very similar to glass noise. do you use glass in your set up? if yes, you can replace it with quartz

of course it may be because of UV light source, but it's less probability.

The noise in your sample looks very much like the cuvette may be causing it. Glass or plastic tend to absorb in the lower regions of the spectrum, and it does not get corrected by the blank sample either. To test this theory, just run a water sample blank, then use the same sample as an analyte.

Second, if you are using high concentrations or really low concentrations, this noise can happen even with a quartz cuvette.

Also, your blank and sample should both use the same cuvette, or optically matched cuvettes.

A final thought is the lamps. The UV-lamps tend to blow faster than the other one. That means you may be getting the noise from a very weak intensity from the lamp.

I am not familliar with %R as a unit of absorbance or transmission. Could someone explain that for me?

I'm agree with Hans that your reference probably has too little transmittance in the UV.

I'm agree with Charles Sharpless, the reference have to little transmitance in uv, you can try with a zaphire substrate, if you have biological samples, the integrating time must be incremented, and the SBW should be close to 2.5 nm.

Good luck

The similar results have been shown to A% and T%

check the cuvette. I would run a UV for a dye and see the spectra in order to troubleshoot the instrument.

Your sample (solute + solvent) is fully absorbing below 300 nm.

Why not restrict yourself to the range 300 - 400 nm.

To study the range below you need an UV transparent solvent and lower concentration of the aromatic polyimide.

1. Run a measurement when both "sample" and "reference" beams are empty - if you still have the same noise, it is the spectrometer (lamp/detector/filter). Otherwise, go to #2

2. Put the empty cuvettes in both beams, and rerun the check - if the noise, absent in #1, appears, it is the cuvette. Otherwise, go to #3

3. Add the solvent and rerun your test

If all the three cases above fail, dilute your sample.

If the spectrometer, cuvette and solvent are OK, but the noise is not, AND you need to measure that high concentration, try to remeasure several and average (in digital form). The S/N (for stochastic noise) is increased as a square root of the number of measurements, and re-measuring 9 times should decrease your noise by a factor of 3. If it does not, your noise is non-stochastic, i.e. systematic.

Going beyond the average of ~10 is not practical, unless you can automate the process. In the latter case, average until you are satisfied. Keep in mind that this is the game of diminishing returns. However, in my own practice, I did averaging in the thousands (not on a spectrophotometer of course), which took me several days of experiments, with a need to change twice a day the (slowly degrading) sample.

We had similar issues at one point. We replaced the lamp and everything was fine.

I am curious: given you're looking at aeromatics I expect substantial absorbance below 300 nm. What is your absolute absorbance? I know that our instrument can register up to 5 (5 units in absorbance is a ridiculous and really meaningless number) we really only trust it up to about 2 (but it's an uncomfortable trust. Ideally you want absorbance between 0 and 1, e. g. transmittance greater than ~0%). The point is if your noise is part of your absorbance being beyond a reasonable range then try a smaller path length if you can (We have cells ranging from 10 cm to 0.01 cm, and they're not TOO expensive).

There have also been some very good points brought up about your spectrophotometric cell and solvent. What is the composition of your cuvette? What is your solvent? Acetonitrile has a peak in the UV and running a blank with acetonitrile then a sample that absorbs intensely in the same region will have a higher absolute absorbance, and cause the spectrum to look noisy around the solvent peak if the amount of transmitted light it is near the lower limit of the detector.

One more thing. I'm not sure if this will help at all, but you can play with your scan rate as well. Too slow and productivity is hampered but too fast might give messier spectra, especially if there are other contributing factors like many of those discussed above.

As said, their is absolutely no use to measure in this range with this sample (concentration) and solvent combination. Even if it would be without any noise. Absorbance should be in the range max 2, better 1.

Hi muhammed, do your pbm has solved

Goddness me ! Any solution giving absorbance >1 then you should stop immediately ! Theoretically A=0.8 !

If you use a glass vessel to hold your solvent then glass might have absorbed the UV portion below 340 nm. Please replace with quartz vessel and try.

Dear all,

1. Is this noise proportional to the signal or constant ?

2. Maybe you know some articles on this issue ?

The other way of determining the origin of the noise is to measure the transmittance on the open beam just after performing the baseline scan. One would expect to get a smooth 100%  line all the way through to the UV region. If you can still pick up the noise, I would conclude that is coming from the instrument itself not the sample you are trying to measure.

Look: Chemometriics in Spectroscopy by H. Mark & J. Workman,

Chapters: 40-53. Very useful and simple mathematics.

Joseph: addition; for free on the web

Analysis of Noise

Part XII

Howard Mark and Jerome Workman Jr.

http://www.spectroscopyonline.com/spectroscopy/data/articlecolumn/spectroscopy/402002/34212/article.pdf

What factor(s) accounts for the increase in the IR absorption frequency of  carbonyl of amide relative to its corresponding acid carbonyl. The C=O of acid (CO-OH) absorb at lower frequency in my spectra relative to the C=O(CO-NH). I think this is not right because, based on known principle oxygen is more electronegative than nitrogen, thus expected to enhance the SP3 character in the CO of COOH and lead to higher avsorption frequency than CONH... please i need possible suggestion on on how to explian this

Thanks

I am very grateful for your valuable feedback.I thank you for your cooperation