Quadrupole Scan Speed and the 8270 Instrument Checkout Mix

In my last blog (here), I promised an update on the impact of the detector scan speed on the tailing factor. I had speculated that a pentachlorophenol tailing factor value of 0.94 was more likely < 1.0 because of the scan rate, rather than column overload.

The examples I put forward here were collected on a system with a different configuration (Table 1), but the basic principles hold true. While most sources will tell you that you need 6 to 7 scans to accurately define a peak, this is only sufficient for reproducible peak areas. 6 to 7 scans are not necessarily sufficient for evaluating peak symmetry.

Instrument Configuration

Table 1 – Agilent 7890A – 5975C GC-MS configuration (Fast and Slow scanning)

Figure 1 and Figure 2 are examples of pentachlorophenol peaks acquired at 3.1 Hz (Slow on Table 1) that show asymmetric peak apex assignment and a resulting bias in the calculated tailing factors. Figure 3 is an overlay of 9 injections that highlights the variability in peak apex assignment when a slow acquisition rate is used.

PCP slow 1

Figure 1 – 5.55 ng Pentachlorophenol (acquired at 3.1 Hz) Tailing Factor Evaluation showing peak apex assignment biased to the rear of the peak, yielding a tailing factor result biased low.

PCP slow 2

Figure 2 – 5.55 ng Pentachlorophenol (acquired at 3.1 Hz) Tailing Factor Evaluation showing peak apex assignment biased to the front of the peak yielding a tailing factor result biased high.

PCP slow overlay

Figure 3 – Overlay of 9 Pentachlorophenol chromatograms acquired at 3.1 Hz showing high variability in peak apex assignment.

We can nearly double the acquisition rate by halving the number of samples that are averaged for each data point. Figure 4 is an example pentachlorophenol peak acquired at 5.9 Hz (Fast on Table 1) that shows little bias in the peak apex assignment. Figure 5 is an overlay of 8 injections that highlights the low variability in peak apex assignment when a fast acquisition rate is used.

PCP fast

Figure 4 – 5.55 ng Pentachlorophenol (acquired at 5.9 Hz) Tailing Factor Evaluation showing minimal room for peak apex assignment bias.

PCP fast overlay

Figure 5 – Overlay of 8 Pentachlorophenol chromatograms acquired at 5.9 Hz showing little variability in peak apex assignment.

Figure 6 and Figure 7 are graphical representations of the tailing factors calculated from the overlays in Figure 3 and Figure 5.

PCP BENZ slow TF graph

Figure 6 – Set of 9 Pentachlorophenol (Figure 3 overlay) and Benzidine tailing factors acquired at 3.1 Hz.

PCP Benz Fast TF graph

Figure 7 – Set of 8 Pentachlorophenol (Figure 5 overlay) and Benzidine tailing factors acquired at 5.9 Hz.

EPA 8270D has initial tailing factor criteria of ≤ 2.0 for both Pentachlorophenol (PCP) and Benzidine (BENZ). As we saw in Figures 1 and 2, using a slow scan speed of 3.1 Hz yielded a range of 0.94 to 1.69 for PCP on a new column, inlet liner and inlet seal combination. It is easy to envision a situation where analyzing a few extraction batches may leave the instrument unsuitable for further sample analysis, even after inlet maintenance. Figure 8 and Figure 9 are statistical demonstrations of reduced variability in the tailing factor calculation for both Pentachlorophenol and Benzidine when the faster acquisition speed is used.

Box Plot Slow Fast

Figure 8 – Boxplot (with labeled medians) of Pentachlorophenol and Benzidine tailing factors under slow and fast acquisition speeds.

ANOVA slow fast

Figure 9 – ANOVA showing reduced variances for both compounds under fast acquisition speed

 

2 Responses to “Quadrupole Scan Speed and the 8270 Instrument Checkout Mix”

  1. Jack Cochran says:

    Excellent work and eye-opening data, Chris. Great distinction between what is needed for accurate quantification via number of points for peak area and the number of points needed for asymmetry determinations. How does peak smoothing fit into the picture?

  2. Aside from using the on the fly noise reduction filter (aka Trace Ion Detection), I don’t run any peak smoothing algorithms on my MS data, so I don’t know.

Leave a Reply