number of injections and calibration strategies

[LCFan]

Dear all,

I am interested in common industrial practices. I would like to know:

- how many injections of standard and sample solutions do you perform within routine sequences (ext std methods)?
- in which order do you inject (how does a routine sequence looks like)?
- what are your calibration strategies within routine analysis (3 levels, 1 level, more levels, through zero?)

Thank you so much in advance for sharing your practices.

Florian

[Don_Hilton]

This depends on the application. From when I was back in industry: Routine analysis in R&D may run differently from routine analysis in the QC department. I was in R&D. In R&D with product development and competitive product monitoring going on and with analyses that did not have product waiting on results to ship, we would run large numbers of samples in batches that could be prepared convienently together. Several batches of samples might be queued together for a run on the GC. Each batch would be bracketed with a standard solution to verify correct operation of the instrument. And, each batch would include a sample of at least one reference material to ensure correct analysis from preparation through chrmatography. Calibrations tended to be 5 levels, but if the need for dynamic range was large enough, there could be additional levels for a given method. Many instruments were recalibrated on a daily basis, but some methods were being evaluated for running a high and low range check and, if within acceptable limits, certiving the instruments to still be in calibaration - and calibrations being repeated on a less frequent basis, such as weekly.

With method precision known through validation studies, calibration and sample vials were sampled once each. A five level curve was run with five injections, followed by a control sample to confirm everything. Ocasionally two controll sampels would be used - one at the higher end of the range of interest and one at the low end.

Typical order for samples:

Calibration run as a sequence and confirmed before running sample sets.

Samples sets, run as batches and may include several batches in a sequence on the instrument. Each batch had gracketing control samples. Samples in the batch included at least one reference material - in a random spot in the sample sequence. (Samples were randomly drawn, and randomly placed in batches.)

Replication of samples through sample prep - or from a GC vial should we have decided to do that was based on validation data. If analysis of variance placed the error in sample prep, we needed multiple sample preps to average out the noise in the prep procedure. Multiple injections from a single vial would give a very precise, but inaccurate result.

Randomization of all samples, including reference materials, was to avoid being confounded by non-random factors.

Placement of controls besfore and after sampels is to let you know where the instrument has failed, so you know where you have to start the analysis again when you find the fault. You want to be sure the isntrument is workign at the end of the queue and it is worth checking at the beginning. And on long queues it is good to have check samples spaced here and there to be able to keep saying good so far as you continue through the data.

On method which could produce very large numbers of samples in a batch, control samples would be spaced at even intervals through the batch.

[BHolmes]

Our application is multiresidue pesticide analysis on fruits and vegetables (extracted using modified QuEChERS) using LC-MS/MS and GC-MS/MS. A sample sequence for a batch of 21 samples for either instrument is as follows:
2-4 condition injections of reagent blank, matrix blank, low level standard
calibration standards x4 levels (1x, 2x, 5x, & 10x LOD), 1 inj each
1 inj solvent blank (MeOH for LC-MS/MS, none for GC-MS/MS)
1 inj reagent blank*
1 inj matrix blank*
1 inj spiked matrix* (for recovery check)
1 inj standard that matches spike levels in spiked matrix recovery, for comparision only
1 inj sample 1
1 inj sample 2
....
1 inj sample 8
calibration standards x4 levels (1x, 2x, 5x, & 10x LOD), 1 inj each
1 inj solvent blank (MeOH for LC-MS/MS, none for GC-MS/MS)
1 inj sample 9
...
1 inj sample 21
calibration standard x4 levels (1x, 2x, 5x, & 10x LOD), 1 inj each
1 inj reagent blank (MeOH for LC-MS/MS, none for GC-MS/MS)

*extracted using the same modified QuEChERS that used to extract samples

For our purposes we bracket the sequence after every 10-15 inj of non-calibration standard injections to show that we can show calibration integrity throughout the entire run. For LC-MS/MS we use ALL standard injections to generate calibration curve, but for GC-MS/MS only the first set of standard injections are used - don't know why, I'm not the primary chemist for this instrument. For both instruments, we do NOT force through origin (zero) and use simple linear regression analysis to obtain calibration equation in the form of y = mx + b. For LC-MS/MS the curve is weighted as 1/x. For GC-MS/MS the curve is NOT weighted.

Hope this helps! What are your current lab practices for setting up sequence?

[LCFan]

Dear both,

thanks a lot for your detailed input.

BHolmes, you asked about my practice. Below, i describe our practice for assay determination in pharmaceuticals:

1 Injection of SST solution (for resolution from impurities, tailing, ...)
1 Injection of 100% standard No. 1
1 Injection of 100% standard No. 2
1 Injection of sample solution No. 1
1 Injection of sample solution No. 2
1 Injection of sample solution No. 3
1 Injection of 100% standard No. 1
1 Injection of 100% standard No. 2
1 Injection of sample solution No. 1
1 Injection of sample solution No. 2
1 Injection of sample solution No. 3
1 Injection of 100% standard No. 1
1 Injection of 100% standard No. 2
1 Injection of SST solution (for resolution from impurities, tailing, ...)

The peak area of each standard injection is corrected to its nominal concentration. Thus, a total of six theoretical identical areas results; the %RSD is calculated in order to assess precision.

The mean of these six areas is the 100% area => 1-point-calibration through origin.

The mean area of each sample solution is converted to concentration and by considering the sample weight, the amount in the sample is calculated.

Regards

Florian