Using the standard curve, as was described in document: “Validation of Analytical Procedures: Text and Methodology Q2 (R1)”, complementary Guidelines on Methodology dated 6 November 1996, incorporated in November 2005., pages11-12.

LOD = 3,3 *σ/S

LOD =10 *σ/S

σ– the standard deviation of the response using an analyte concentration close to LOQ.

S – the slope of the calibration curve.

This explanation seems more practical than scientific. But in the gas chromatographic methods sometimes is thus more accurate and significantly better insured in the low values than known popular scientific approach, using 3σ and 6σ of sign/noise for 10 blank samples.

Found values are satisfied through additional analyzes with samples containing these concentrations, mentioned as LOD and LOQ.

Thanks for your answers.

Effectively, GC/MS detected some analytes that S/N ratio is below 3.

Best regards

Determining limit of detection (LOD) and lower limit of quantification (LLOQ) can be a surprisingly subtle and difficult problem. There is not complete agreement on the best ways to do this.

As was just pointed out, one traditional approach is to set the LOD to the concentration where the S/N ratio is 3:1.

Other approaches for LOD may also be used. Pages 359-360 of the fourth edition of Tietz Textbook of Clinical Chemistry and Molecular Diagnostics has a fairly complete discussion of LOD. (Other editions of that book do not necessarily have such a complete discussion.) The discussion in Tietz is based on some ISO recommendations from years 1999 and 2000 I believe.

The method recommended in Tietz is fairly involved. First one does a set of replicate measurements on the blank. Fromn this one then determines the concentration corresponding to the 95th percentile. This is done using non-parametric statistics, which means that a lot of replicates need to be run. This is call "limit of blank" (LOB).

Then one runs a series of replicates on a sample of concentration near the LOD. (Since one hasn't yet determined the LOD, I suppose the best one can do is guess at the LOD, prepare one or more solutions at that concentration, and proceed from there.) Determine the 5th percentile of this set of samples. You can usually assume this distribution is Gaussian and use parametric statistics for this step. The LOD is defined as the concentration where the LOB equals the 5th percentile of sample replicates whose concentration is the LOD. Obviously, if someone guessed grossly wrong for the LOD that person would need to repeat this process again with an improved guess of the LOD.

LOQ is determined based on an acceptable value for total error and probabilities. (See page 361 of Tietz.) For example, if you were to accept a 45% error at the LOQ, and if the bias of your method was 15%, and if your coefficient of variation (CV) were 15%, and if you could accept that you might be outside of the 45% total error specification by about 2.5% of the time, then you could use bias+2*CV as your cutoff criteria for LOQ, i.e. the concentration of your LOQ would be that point at which the total error was no more than 45%, where the total error is given by bias +2*CV.

In addition to other answers, if your analysis of PAH comprises extraction of PAHs from a liquid or solid sample, do not forget to evaluate the extraction efficiency for each PAH and correct your results. Also, you need to assess losses due to sorption on the glass wall of the vial and the septum cap.

Injecting in GC/MS low concentrations of PAHs and calculating signal-to-noise ratio. PAHs concentrations with S/N=3 are LD and S/N=5 are LC.

Besides you cal calculate both parameters from calibration curve

The sample with low concentration of PAH maybe more concentrated before GC/MS analysis by use of Solid Phase Extraction method.

The LODs and LOQs for all analytes were determined from free (blank) samples (n = 5) as the lowest concentrations yielding S/N ratios of at least 3:1 and 10:1, respectively. They were calculated for average baseline noise. The LOQ was subsequently determined by analysis of five spiked samples prepared at their respective concentrations. The LOD and LOQ were determined as published previously [Papoutsis, I., Nikolaou, P., Athanaselis, S., Alevizopoulos,

G., Pistos, C., Paraskevopoulou, C., Spiliopoulou, C., J. Sep. Sci. 2011, 34, 3037–3042; and other publication Karlonas, N., Padarauskas, A., Ramanavicius, A., Ramanaviciene, A., J. Sep. Sci. 2013, 36, 1437–1445.]

Using the standard curve, as was described in document: “Validation of Analytical Procedures: Text and Methodology Q2 (R1)”, complementary Guidelines on Methodology dated 6 November 1996, incorporated in November 2005., pages11-12.

LOD = 3,3 *σ/S

LOD =10 *σ/S

σ– the standard deviation of the response using an analyte concentration close to LOQ.

S – the slope of the calibration curve.

This explanation seems more practical than scientific. But in the gas chromatographic methods sometimes is thus more accurate and significantly better insured in the low values than known popular scientific approach, using 3σ and 6σ of sign/noise for 10 blank samples.

Found values are satisfied through additional analyzes with samples containing these concentrations, mentioned as LOD and LOQ.

making seven replicate measurements at a concentration near the expected detection limit. computing the standard deviation (SD) for the seven replicate concentrations, and multiplying SD by 3.14 if n=7

In addition to the answer of V. Christova-Bagdassarian. The fourmula indicated in many answers is based on the standard deviation (sd) of a blank. The use of a S/N ratio is very common in chromatography due to its simplicity but it leads many times to overoptimistic detection limit values because this method does not have into acount the variability of the method (precisión) and the posibility that the blank gives a signal. The best method to determine real detection limits is, as indicated by Christova, to prepare a fortified blank at a concentration that should be between expected LOD and LOQ values, to measure this fortified blank at least 10 times (the S/N ratio has to be always >3) and to determine the standard deviation of the obtained signal for each compound. This value can be used as the standard deviation of the blank and it can be used to calculate 3sd (or 3.3sd if we are very accurate and want to take into account type I and type II errors) and 10 sd to determine LOD and LOQ respectively. If the concentration of the fortified blank that was prepared for the test falls between the obtained LOD and LOQ, we can accept the values. Otherwise, we should prepare a new fortified blank taking into account the values obtained and repeat the experimental.Doing this, we take into account the S/N ratio and also the precisión of the method and the obtained values are more realistic. The main problema is that the values tend to be larger than those obtained using the S/N ratio and the methodology is time consuming and requires many experiments and time..

Spiking a low lever of target compound and do the same procedure and try to calculate

Dear Cristian,

Attached a true validation study excel sheet of PAH derermination by HPLC. you can use the same procedure in GC (the study includ LOD,LOC , etc.)

are you doing just instrumental analysis?

but anyway, you will also do real sample analysis.

for the instrument, it is usually instrumental detection limitation (IDL), which just use pure standard.

but for real samples, they are pretreatment, matrix effect. etc, therefore it is method detection limitation (MDL), which incudes all precedures, filtration, SPE(for water),... and instrumental analysis.

After extracting the sum of the ion currents of both precursor and fragment ions relative to each analyte, the peak height-to-averaged background noise ratio has to be measured. The background noise estimate has to be based on the peak-to-peak baseline near the analyte peak. LOQs could be then calculated on the basis of a minimal accepted value of the signal-to-noise ratio (S/N) of 10. For those analytes whose detection could be performed by monitoring more than three ions, LOQ estimation has to be calculated by considering only the three most abundant signals. For the LOD definition, it’s important to remind that when using a MS detector, the first condition to be satisfied for ascertaining the presence of a targeted compound is that the precursor ion and at least two product ions produce signals distinguishable from the background ion current. Therefore, when more than three ions could be selected for analyte identification, LODs have to be estimated selecting signals for the parent ion and, among fragment ions, the two ones giving the best S/N ratios. Accordingly, LOD (S/N 3) of each analyte could be estimated considering among the three selected signals the ion giving the worst S/N. Finally, as it makes any sense to quantify an analyte only if its presence is ascertained, it follows that the LOQ of the method, likewise the LOD, is dictated by the ion signal intensity of the least abundant ion. In other words, a LOQ is correctly estimated only if its value is larger or equal to the LOD estimated as mentioned above.

Quantification limit: simply the lowest spiking concentration that gave acceptable accuracy (70-120%) and precision (

Detection (DL) and quantitation (QL) limits can be calculated as explained in the TO 13A method of USEPA (1999) . DL is defined as 3.3 times the mean of the standard deviation of the replicates of the lowest point of the curve (for example: 5 μg•L−1) divided by the slope of the regression eq., and QL as ten times the DL

Hi Cristian,

there is a very useful technical specification from ISO/TS 13530- Quality assurance in water analysis, dealing also with your question. This ISO allows for a comparison of DL's and LOQ's with other labs as well. The basis is the experimental verification of estimated values in matrix.

Best success

Sabine

The limits of detection (LOD) and quantification (LOQ) for analytes (PAH) were determined from free samples (n=5) as the lowest concentrations yielding signal-to-noise ratios of at least 3:1 and 10:1, respectively. They were calculated for average baseline noise. The LOQ was subsequently determined by analysis of five spiked samples prepared at their respective concentrations.

Dear Cristiàn,

all the method that you and several others cited for calculating LOD and LOQ on the basis of noise level or standard devition fron blank replicates are undoubtly correct. But I prefer to be sure of what my method is  assuring in terms of sensitivity and reliability. With the continue analysisi the column becomes more and more dirty and this cause a peak broadening expecially with heaviest PAH i.e. 1,2,3 c,d indenopyrene og dibenzo-a,h.antracene or benzo, g,h,i-perylene. This cause severe losses in senstinvity. Because of this iprefere determine the lowest level that my analysis method can assure by spikinng my sample and checking the recoveries, with replicate analysis.  Best regards. MV.