October 23, 2021, 05:03:56 PM
Forum Rules: Read This Before Posting


Topic: LOD calculating from 10 blanks  (Read 252 times)

0 Members and 1 Guest are viewing this topic.

Offline xshadow

  • Full Member
  • ****
  • Posts: 409
  • Mole Snacks: +1/-0
LOD calculating from 10 blanks
« on: September 01, 2021, 10:18:53 AM »
I don't understand very well how to calculate the LOD  having 10 blanks (matrix without the analyte).

The formula I know is : 
LOD = X + 3s

Now X should be the mean  concentration of 10 blanks
3s the standard deviation

BUt my doubt is that LOD is expressed with concentration... in my formula I have the signal instead : so I should find the corresponding concentration that gives the signal X.

But how can I do this?
With a calibration curve?  Do I have to put in these calibrators also the analyte??  And the matrix should be the same I have in the blanks? or for calibrators can I use only the analyte and a solvent

THanks

Offline Corribus

  • Chemist
  • Sr. Member
  • *
  • Posts: 3187
  • Mole Snacks: +485/-22
  • Gender: Male
  • A lover of spectroscopy and chocolate.
Re: LOD calculating from 10 blanks
« Reply #1 on: September 01, 2021, 02:02:16 PM »
Obviously, you need a calibration curve to convert the signal axis to a concentration axis.

It bears also mentioning that the LOD value needs to be defined for a certain confidence interval, which will determine both the number of blank replicates and the multiplier. 10 blanks and a multiplier of 3 is a commonly used quick rule of thumb, but actual multipliers should be determined by statistical t-tests using a desired confidence level and degrees of freedom, and specific organizations may require higher or lower confidence intervals for certain applications or methods. Certain assumptions are also implicit, such as a linear relationship between the signal and concentration, Gaussian distribution of the data points about the mean, and so forth. Finally, note that the LOD can change depending on how the blanks are treated. E.g., detection limits calculated from method blanks vs. instrument blanks are completely different things. I'm also sure you know that the limit of detection does not imply that a measured quantitative value is accurate or reliable at the LOD value, but it doesn't hurt to state it as a reminder.

Unfortunately these analytical terms are not frequently well-defined, and people often use the "10 blanks, multiplier of 3" rule of thumb without understanding where it comes from or what it means statistically. I think for most "casual" use (basic research, e.g.) it doesn't matter a whole lot but if you're doing analysis with regulatory or legal implications, and you're anywhere near the limits of your method in terms of analyte concentration, you can't ignore those kinds of things.
What men are poets who can speak of Jupiter if he were like a man, but if he is an immense spinning sphere of methane and ammonia must be silent?  - Richard P. Feynman

Offline xshadow

  • Full Member
  • ****
  • Posts: 409
  • Mole Snacks: +1/-0
Re: LOD calculating from 10 blanks
« Reply #2 on: September 01, 2021, 03:21:50 PM »
Obviously, you need a calibration curve to convert the signal axis to a concentration axis.

It bears also mentioning that the LOD value needs to be defined for a certain confidence interval, which will determine both the number of blank replicates and the multiplier. 10 blanks and a multiplier of 3 is a commonly used quick rule of thumb, but actual multipliers should be determined by statistical t-tests using a desired confidence level and degrees of freedom, and specific organizations may require higher or lower confidence intervals for certain applications or methods. Certain assumptions are also implicit, such as a linear relationship between the signal and concentration, Gaussian distribution of the data points about the mean, and so forth. Finally, note that the LOD can change depending on how the blanks are treated. E.g., detection limits calculated from method blanks vs. instrument blanks are completely different things. I'm also sure you know that the limit of detection does not imply that a measured quantitative value is accurate or reliable at the LOD value, but it doesn't hurt to state it as a reminder.

Unfortunately these analytical terms are not frequently well-defined, and people often use the "10 blanks, multiplier of 3" rule of thumb without understanding where it comes from or what it means statistically. I think for most "casual" use (basic research, e.g.) it doesn't matter a whole lot but if you're doing analysis with regulatory or legal implications, and you're anywhere near the limits of your method in terms of analyte concentration, you can't ignore those kinds of things.

thanks Corribus.

But how could I draw the calibration curve?

Are the calibrators   (which they cointain the analyte) at different concentrations  prepared using the same matrix that I have used in the 10 blanks?  Or I have to set up a solution with an analyte and solvent  but nothing more ( so very weak matrix effect compared with the matrix I have in the 10 "sperimental" blankk)

thanks :)

Offline Corribus

  • Chemist
  • Sr. Member
  • *
  • Posts: 3187
  • Mole Snacks: +485/-22
  • Gender: Male
  • A lover of spectroscopy and chocolate.
Re: LOD calculating from 10 blanks
« Reply #3 on: September 01, 2021, 09:53:36 PM »
Yes you would prepare several concentration standards in the same matrix your blanks are in, preferably over the same concentration range you will measuring with your method. Minimum of 3, weighted if appropriate.
What men are poets who can speak of Jupiter if he were like a man, but if he is an immense spinning sphere of methane and ammonia must be silent?  - Richard P. Feynman

Offline xshadow

  • Full Member
  • ****
  • Posts: 409
  • Mole Snacks: +1/-0
Re: LOD calculating from 10 blanks
« Reply #4 on: September 02, 2021, 04:55:35 AM »
Yes you would prepare several concentration standards in the same matrix your blanks are in, preferably over the same concentration range you will measuring with your method. Minimum of 3, weighted if appropriate.


thaaaanks Corribus
 :)

Sponsored Links