quantification using a calibration curve

[kouroshh1]

I have made my calibration curve based on the reference standard, which I prepared at the concentration from 0.0975 ppb, 0.195 ppb, 0.39 ppb, 0.78 ppb, 1,56 ppb, 3.125 ppb, 6.25 ppb, 12.5 ppb, 25 ppb, 50 ppb, 100 ppb and 200 ppb. Based on response of our instrument detector, I will make a calibration curve to be able to calculate the concentration of my analyte in my extracted sample.

The problem I have is when I do the extraction then i should dilute my sample in a way that the concentration of my analyte falls into the specific range in my calibration curve. I have this big problem that I do not know normally the response of my analyte should be fallen into what concentration in my calibration curve? lower concentration range for instance 6.25 ppb or higher concentration range for instance 100 ppb. What is more acceptable?

Thank you for your great help

[mattmullaney]

Hi kouroshh1,

I have made my calibration curve based on the reference standard, which I prepared at the concentration from 0.0975 ppb, 0.195 ppb, 0.39 ppb, 0.78 ppb, 1,56 ppb, 3.125 ppb, 6.25 ppb, 12.5 ppb, 25 ppb, 50 ppb, 100 ppb and 200 ppb. Based on the response of our instrument detector, I will make a calibration curve to be able to calculate the concentration of my analyte in my extracted sample.

Okey-dokey. Please, may I make two suggestions for the next curves...not quite so many calibrant levels (seven is more than sufficient) and to spread them out a bit more evenly (say 0.0975, 30, 60, 90 120, 150, and 180 ppb for example). Not much advantage in having > six levels, I prefer seven to evaluate the scedascity of the calibration curve, and the equal spacing of levels helps with the scedascity evaluation as well--no calibration point has an excess leverage.

As to the dilution...it depends. If you have a good idea of what level that 3-indole carbinol extracts out as, you'd want to design your calibration curve so that the extracts "hit" the mid-point of the curve, wherever that is.

My two cents worth.

[James_Ball]

Agreed, mid point of the calibration, 20-50ppb most likely for the curve you have listed. Unless you expect undiluted samples to be less than 1ppb, you should not need to go that low. If the instrument becomes less linear at the upper end, then drop the higher points and focus on lower ones.

[kouroshh1]

Thank you Mattmullaney and James for your comments. Actually, I do not know why I could have not received a notification by email or on top of this page where the notification sign located. Please, accept my apologize for a late reply.

Actually, currently I am not working with indole-3-carbinol. I have a standard mixture compromising of 12 different standards mixing together at concentration of 10 ppm and from that I make 1 ppm to be able to do my dilution.

In addition, the concentration of my analytes in the extracted sample is very high so it is above the maximum number in my calibration curve. Thus, I had to dilute my extracted sample 150 times to be able to fall my analyte peak intensity in a middle range of calibration curve (however after dilution the peak shape has lost it gaussian shape and the shape of the peak is not that good). But for the next time I will try to dilute in a way that my analyte of interest falls into the middle of the calibration curve as you said.

But may I ask what should I do with a peak shape when I dilute many times? After diluting my sample 150 times the intensity is in an acceptable range but the peak shape has lost its uniformity.

Sure, as you recommend I will try to spread the points equally in my calibration curve.

[mattmullaney]

Hi kouroshh1,

Good to hear you have a break from indole-3 carbinol!

Weird, usually peak shape improves upon dilution to a degree. Please, what is the diluent and eluent composition...and inj. vol.?

[kouroshh1]

Hello again,

If you can please open the following link you can see my chromatogram

http://tinypic.com/r/feklms/9

I do not know if the quantification after 150 times for the left peak is acceptable or not? or should I do quantification based on 100 times dilution?

My diluent is water 100 %
my eluent is:
A: H2O 100 % + 20 mM acetic acid
B: MeOH 100 % + 20 mM acetic acid


In addition, if you open the following link

http://tinypic.com/r/28a3o86/9

You can see my calibration curve based on the concentration I made below:

0.0975 ppb, 0.195 ppb, 0.39 ppb, 0.78 ppb, 1.56 ppb, 3.125 ppb, 6.25 ppb, 12.5 ppb, 25 ppb, 50 ppb, 100 ppb, 200 ppb, and 400 ppb

So, do you think my range is not spread well? One of the main reason I am using many points in lower concentration is that I need to determine the limit of detection and quantification for my method validation. What do you recommend me to do? If my unknown sample specific analyte concentration falls into the middle of this calibration curve do you think it is acceptable?

Thank you for your reply

[mattmullaney]

Hi kouroshh1

"I do not know if the quantification after 150 times for the left peak is acceptable or not? or should I do quantification based on 100 times dilution?"

I forget that sometimes the chromatographic peak with MS detection can be ugly at low concentrations. Injection solvent is okay. The 8X dilution looks good (best) in terms of peak shape. If you know the retention factor(s) for the peaks and they are > 2 all would seem to be well. Where does the 8X dilution fall on the calibration curve? I see that you may need a wide calibration range, anyway.

So, do you think my range is not spread well? One of the main reason I am using many points in lower concentration is that I need to determine the limit of detection and quantification for my method validation. What do you recommend me to do? If my unknown sample specific analyte concentration falls into the middle of this calibration curve do you think it is acceptable?

Yes...there are a number of standard levels that are bunched together at low concentration while fewer calibration points are spread out widely at higher concentration(s). The highest concentration level in this situation is said to have "leverage", an inordinately large effect on the slope of the calibration curve. To avoid this, it is better to evenly spread out the calibration levels in concentration. LOD and LOQ need not be determined from the curve (at least not LOD, typically a S/N value of 3 or so is used to assess this--can't quantify reliably between LOD and LOQ anyway). If you want to include LOQ in the curve, that is okay.

Yes--it is the middle of the (linear) calibration curve that affords the least error. It is desirable, if possible, to perform quantification of samples at the midpoint of the calibration curve.

I can send references next time if my help isn't adequate for now. Got this commitment to keep here.

[kouroshh1]

Hi Mattmullaney,

Actually, I have not calculated my retention factor but the analyte of my interests is very well retained into the column. I remember you calculated the void volume for my column last time.

Column ID is 2 mm, by 100 mm, so the column void volume is ca. 0.214 mL, divided by 0.3 mL/min affords t0 = 0.713 min.

So, retention factor would be K= 20.16-0.713/0.713= 27.27

The 8X dilution falls on between 100 to 200 ppb in my calibration curve. Yes it is strange the peak shape is getting worsen after dilution. Am I thinking correctly that after we dilute the matrix effect should be reduced so the peak shape should be improved or there is no correlation between dilution and matrix effect?

Thank you for sending me the reference.

[mattmullaney]

Hi Again,

Retention factor is not a problem...maybe even "too much".

Okay, if the samples after 8X are 100 - 200 ppb (w/v), you could calibrate to:

0.200 ppb (ca. LOQ I'm guessing), 40 ppb, 80 ppb, 120 ppb, 160 ppb, 200 ppb, 240 ppb, 300 ppb...sorry, eight levels instead of seven. Must be tired or something...

http://pubs.rsc.org/en/content/ebook/9780854046713

Statistics for the Quality Control Chemistry Laboratory
Author: Eamonn Mullins

http://pubs.rsc.org/en/content/ebook/9780854041312

Practical Statistics for the Analytical Scientist: A Bench Guide: Edition 2
Authors: Stephen L R Ellison, Vicki J Barwick, Trevor J Duguid Farrant

There are some others, Francisco Raposo comes to mind (Elseveir journals).

Last question is good...like I said, I'm not well-versed in MS following LC separation, naively I'd think peak shape would improve as the analyte's matrix is diluted. doesn't seem to work that way. Maybe if the diluent was the mobile phase composition at the point of which the elution takes place, the peak shape of the more dilute solution(s) would improve?

[kouroshh1]

Hello again,

Thank you for the links. I will go through them very soon. Regarding your comment of using diluent same as mobile phase composition at the point of which the elution takes place, I will give it a try very soon. Thank you for your very helpful feedback.

[James_Ball]

What I have found for best peak shapes is to have the dilution made in the beginning mobile phase, this makes sure there is no mixing problems with the injected sample and the mobile phase that is entering the column at that time.

The more aqueous the better normally. If you have a beginning mobile phase of 90% aqueous and inject a sample that is 75% methanol, some of the analytes will ride that solvent slug through the column and give a large peak early in the chromatogram, then the portion that is retained will give another peak in the gradient. This is known as "Peak Surfing" as a portion of the analyte rides through the stationary phase like a surfer on a wave.

If you always do a dilution on the sample, then the LOD and LOQ will have to take into account that dilution amount, so there is no need to calibrate to the lowest possible amount that can be detected on the instrument if the sample will always have a high level detect. If the sample will always be 100ppm then you don't need to calibrate to 0.000001ppm just because the instrument can see it(for an example). You could easily calibrate as 10ppm, 50ppm, 100ppm, 200ppm, and 500ppm to satisfy the needs of the test. If there could be some samples that will be non-detect even when not diluted then you need to know the limits of the instrument and procedure.

[mattmullaney]

Geez, James_Ball is correct! On all counts. As I recall, these samples were all prepared in dI water, though (gradient begins with water + 20 mM HoAc).

Perhaps could some MeOH be added to the A solvent? Kouroshh1 has much retention anyhow already.

[kouroshh1]

Thank you all for your very useful information. Actually, my mobile phase starting point is A: 95% DI H2O B: MeOH 5 %. I could retain my analyte better while I am using this solvent composition at the beginning of LC method.
Should i dilute my sample in 5 % MeOH then?

Now with your explanation, I am very glad that I understood about my mistake. I was always thinking that the the compound of my interest should be fallen at lower level of calibration curve. So, now i learned that i can have a wider range and fit my analyte into the middle of the curve.

[mattmullaney]

Hi again,

I think you're okay with your current prep in water for the calibration standard solutions--James_Ball is absolutely right about that. Your direction is the proper one.

[James_Ball]

5% difference is usually not enough to make a difference, but I have found that once you have more than 30% MeOH in the sample and only 5% in the mobile phase you begin to have peak shape problems, and they get worse the higher the concentration of MeOH and the larger the injection volume. With 80% MeOH in the sample you have to inject 1-2ul max into 5% MeOH mobile phase.