At what point are you seeing the negative CPSs?

If this is after the software has carried out a background correction, then it may be that the background measurement is incorrect.  We have occasionally seen this on our ICP-MS when the filters on the ultapure water system have begun to fail and the blank solution is not as blank as it could be.

Dear Tharanga:

It is a great pleasure to contribute to your valid question.

Please remember that there are two types of signals:

1. Gross signal:   Which can never be negative.

2.  Net signal:     Net Signal  = Gross Signal – Correction

The above “correction” (subtraction) to obtain a Net Signal can generally be either, (a) background correction, or (b) the blank subtraction.  If the instrument software is setup to do either of the above automatically, you can end-up with a negative signal, as a result of the overestimation of the background or the blank, respectively.

To make things clear, I have attached a diagram using data that I simulated for a situation where you can have negative intensities for the net intensity of a valid peak, when there is a concave spectral background, or structured background, and try to do a background correction. If such over-estimation of the background has occurred, as long as that correction (structure of the background) remains almost constant for all the samples and standards, it can still give a perfect calibration line, but with a negative intercept. Although you have a negative intercept, it can still produce correct concentration values for the samples as long as the intercept remains the same for all the samples and standards.  For example, in the simulated diagram, 0.1 ppm gives negative intensity. 

Similarly, you may do such an over correction also for the blank (not a spectral issue).  For example, it can happen when you estimate the blank using a contaminated acid, but your samples and standards are prepared using a cleaner acid. Then, if you do a blank subtraction using the contaminated blank, you will be subtracting a bigger blank than necessary, but it will be constant for all of them.  Then you still get the correct data, even though your calibration may have a big negative intercept, similar to the over-correction for the background.

I have included some simulated data on the attached spreadsheet (a second attachment) to show that you can have a nice straight line calibration with some negative intensities.

The above explanation should not be taken as an advice to ignore the importance of  proper definition of the spectral background or proper blank correction during an analysis, as it can still add errors if it varies from one sample to the other.  

Hope, this will give the correct answer to your question.

Best wishes, Nimal

Dear Professor Nimal,

Thank you very much for your brilliant answer!