[TestTubeWeigher]

Hi Folks,

I'm new here and hoping I can get some help.  I did a forum search and didn't find a good match to my question.  I'm wanting to determine LOD of an UV-vis instrument for Cr+6 analysis using a diphenylcarbazide method.  I'm also fairly new to this type of analysis but am trying to get up to speed quickly.  Statistical methodology that I'm finding seem to conform to using the linear regression of a calibration curve :  LOD = 3.3 x std-dev / slope.  In Excel: LOD = 3.3 x STEYX / slope  Equation is simple enough but I find resulting LOD value overly generous?

Here's an example of a recent calibration curve with our instrument for Cr+6 @ 542nm:

ppb 0 : abs 0.000318
ppb 10 : abs 0.001067
ppb 58 : abs 0.014104
ppb 116 : abs 0.024182
ppb 257 :  abs 0.076003
ppb 403 : abs .122031
ppb 598 : abs 0.188325
ppb 1263 : abs 0.415965

Linear regression of calibration curve: y=3089x
steyx=16.7228
LOD=18ppt

Visually, I'd say LOD is about ~1ppm but in any-case, 18 ppt seems overly optimistic.  Am I doing this correctly?  Would very much appreciate advice and any recommendations on statistical parameters of validation.

[Corribus]

I didn't reproduce your numbers, but be aware that the standard deviation being referred to is the standard deviation of a blank measured usually about 7 times.  It's not the standard deviation of your concentration standards.

http://en.wikipedia.org/wiki/Detection_limit
http://water.usgs.gov/owq/OFR_99-193/detection.html

[TestTubeWeigher]

Very much appreciate your help Corribus!  Thank you also for the links.  Should have thought of Wikipedia. 

I made a blank and ran seven analysis.  I also removed the cuvette and replaced as I find abs values will change appreciably compared to leaving in place and since all the samples are placed in the holder manually, thought this would give a better representation for LOD.

I then took the resulting abs value and converted to ppb concentration based on the calibration curve.  Then I took the average ppb of the seven analysis of the blank and added this to 3.3 times the standard deviation.  This method gave a LOD of 3.9ppb and a LOQ of  9.5ppb.  Those values I have far better confidence in believing.

Did I do this correctly?

Concentrations I got from abs values of the blank in ppb were:
-0.565, 1.01, 1.644, 1.807, 1.641, 1.653, 1.023.

Avg blank conc: 1.17 ppb
stdev: 0.83