negative intercept in calibration curve

[kaka_munda]

Hi everybody,

I am having negative intercepts for some of my analytes I injected on GC-uECD. I am wondering if I should force the origin to zero or just add the absolute value of intercept in the calculation?


regards,
kaka

[Don_Hilton]

More details are required. The negative intercept can indicate real things going on in the chromatographic system - in which case it is telling you things that can not be ignored - or it may be the result of noise - in which case the better option is reconsideration of calibration levels and number of injections at those levels. (Or the difference actually is so small that deviation from zero can be ignored.)

How are you making up standards and in what matrix? What kinds of compounds are you running? Are you using an internal standard?

[kaka_munda]

More details are required. The negative intercept can indicate real things going on in the chromatographic system - in which case it is telling you things that can not be ignored - or it may be the result of noise - in which case the better option is reconsideration of calibration levels and number of injections at those levels. (Or the difference actually is so small that deviation from zero can be ignored.)

How are you making up standards and in what matrix? What kinds of compounds are you running? Are you using an internal standard?

Thanks for your reply. Standards are hexachlorobenzene, PCB101, penta/hexa-chloroethanes and mirex and they are in isooctane. I made 5 levels by serial dilution spanning three order of magnitude. The intercept value is significant and not near to zero.

[Peter Apps]

Hi Kaka

An ECD is not quite linear in response, if you are calibrating over a wide range, or using high concentrations and then extrapolating below the range for which you have actual measurements this might be why you get a negative intercept.

Peter

[kaka_munda]

Hi Kaka

An ECD is not quite linear in response, if you are calibrating over a wide range, or using high concentrations and then extrapolating below the range for which you have actual measurements this might be why you get a negative intercept.

Peter

Hi Peter,

I see your point but the regression among different levels is pretty linear (R2= 0.99) and my sample points fall between the calibration levels. So, in this case, should I ignore intercept?

best,
Kaka

[Peter Apps]

Hi Kaka

Just to be certain - you are talking about the intercept on the Y axis ? So you have a positive intercept on the X-axis.

What is the calibration equation ?

Peter

[kaka_munda]

Hi Kaka

Just to be certain - you are talking about the intercept on the Y axis ? So you have a positive intercept on the X-axis.

What is the calibration equation ?

Peter

Yes, it is Y intercept (value is - 2E+09). The equation for pentachloroethane is:

y=(3E+11)x - 2E+09


Kaka

[Peter Apps]

So the intercept is 2/3 of a percent of the slope.

What are x and y for your lowest calibration point ?

Peter

[kaka_munda]

So the intercept is 2/3 of a percent of the slope.

What are x and y for your lowest calibration point ?

Peter

For lowest point, x is 3.68E-02 and y is 8.4E+09.

[Peter Apps]

The intercept is too big in comparison to your data points to just ignore it. You need to trouble-shoot the analysis; either your calibration standards have lower concentrations than you think that they have, or you have a constant bias in your instrumental response.

Peter

[kaka_munda]

The intercept is too big in comparison to your data points to just ignore it. You need to trouble-shoot the analysis; either your calibration standards have lower concentrations than you think that they have, or you have a constant bias in your instrumental response.

Peter

Thanks but I got negative intercept with this compound only and not for others. I injected them as a mix with equal concentrations.

[Peter Apps]

If the other compounds are OK then there is probably nothing wrong with the instrument. How do you make up you standards ?, at what stage do you have all the compounds in solution together ?

Peter

[kaka_munda]

If the other compounds are OK then there is probably nothing wrong with the instrument. How do you make up you standards ?, at what stage do you have all the compounds in solution together ?

Peter

right from the beginning. I made a stock solution containing the mix of 5 components and then the stock solution was serially diluted to 5 levels.

[Peter Apps]

If the other compounds are OK then there is probably nothing wrong with the instrument. How do you make up you standards ?, at what stage do you have all the compounds in solution together ?

Peter

right from the beginning. I made a stock solution containing the mix of 5 components and then the stock solution was serially diluted to 5 levels.

Step one is to make the stock solution again - by far the most likely explanation is operator errror at that stage.

Peter

[Don_Hilton]

Do you use an internal standard? And if so, the same internal standard for all compounds? And does an injection of solvents with no added internal standard or analyte give any peak - particularly in the location for the internal standard?