
Re: [ICPOES] Method concentration vs calculated concentration  large errors
tmcsweeney Jul 19, 2018 5:32 AM (in response to to.a.stranger)It depends on the values that you have in the "weight", "volume", and "dilution" columns on the Sequence page. I'm assuming you have a 5110 ICPOES. Unadjusted concentration values are typically in ppm in the solution or ug/ml. This is then multiplied by the value for the sample in the "volume" column (ml), and divided by what is in the "weight" column (g). If there is a dilution factor, this value is multiplied by that factor. This results in an adjusted concentration in ppm in the original solid sample (ug/g). It also depends on the fit you have selected on the "Standards" page, whether linear, rational, or quadratic and whether you have selected weighted fit and whether you have forced through the blank or origin.

Re: [ICPOES] Method concentration vs calculated concentration  large errors
to.a.stranger Jul 24, 2018 10:22 PM (in response to tmcsweeney)Thank you for your reply. I’m using 700 series with ICP Expert II software. I kept the “weight” “volume” and “dilution” columns in the “Sequence” page blank. It appears that deleting outliers in the calibration curve decreases the discrepancy between the “method concentration” and “calculated concentration.” That, as well as changing the conditions in the “Standard” page.

Re: [ICPOES] Method concentration vs calculated concentration  large errors
blhahesy Aug 1, 2018 6:34 PM (in response to to.a.stranger)When a calibration is run a line of best fit (or calibration curve) is generated, based on the calibration parameters (fit type, weighting, through blank), for the standard solution measured intensities and defined (or "method") concentrations. This results in an equation for the line that can be used to calculate concentrations from intensity measurements. Once the calibration is complete the measured intensities for each of the calibration solutions are then used to calculate a concentration based on the equation for the line, resulting in the "calculated" concentration in the software.
In an ideal world the measured points will exactly match the line of fit and the calculated concentration will be the same as the method or defined concentration, but this will rarely (if ever) be the case when the number of calibration points is greater than the minimum required for the particular fit type for a range of reasons.
The difference between the "method" and "calculated" concentrations gives an indication of how well the line of fit matches each measured point.
Assuming that there are no issues in the standard preparation and sample introduction, small differences or errors indicate that the line of fit is a good representation of the instrument response, while large errors would indicate that the line of fit is not a good representation of the response.
Different fit types and parameters will produce different calculated concentrations and associated errors.
Other issues in standard preparation, sample introduction components, selection of inappropriate concentrations and a number of other factors will also affect how well the calculated concentrations match the method concentrations.

Re: [ICPOES] Method concentration vs calculated concentration  large errors
to.a.stranger Aug 1, 2018 6:46 PM (in response to blhahesy)Thanks for your reply. So the instrument recalculates based on the calibration curve – I think I understand now.
For some unknown reason, the users in our lab always select the “Quadratic” type curve instead of “Linear.” Pardon my ignorance in this matter, but what is the minimum number of points for this type? TIA for your response.

Re: [ICPOES] Method concentration vs calculated concentration  large errors
blhahesy Aug 1, 2018 11:00 PM (in response to to.a.stranger)A quadratic fit requires a minimum of 3 points, just as a linear fit requires a minimum of 2 points. In each of these cases the line of fit would pass through all of the points, so there would be no error in the calculated concentration and parameters such as weighting and forcing through blank would have no effect.
For a quadratic fit the three points may be a blank and two standards, but the blank could be omitted of the fit is forced through the origin, i.e. the zero concentration and intensity point on the curve (which would be used as the 3rd point), and the same would apply to a linear fit.
From memory, if only two points are supplied when a quadratic fit is selected in ICP Expert II it will automatically switch back to calculating a linear fit.

Re: [ICPOES] Method concentration vs calculated concentration  large errors
to.a.stranger Aug 2, 2018 10:45 PM (in response to blhahesy)I see, I have indeed noticed that more than 3 standard points causes deviation in method concentration and calculated concentration in the quadratic fit. And also, with 2 points the software reverts to linear fit. Thank you very much for sharing. Much appreciated.




