Did you know you can manage jagged bleed with a controlled cooling program?

GC column bleed sounds like one of those old problems people used to have. The XLB was developed to lower detection limits by minimizing bleed. Now, virtually all polysiloxane GC columns are low bleed, with thick film and high polarity phases being the exception. Modern GC-MS instruments are so sensitive that I’m rarely concerned about bleed interfering with my analysis. Sure, if your quant ion shares the same m/z as a major bleed ion, the baseline may be elevated, but it is rarely a problem.

Bleed becomes a problem for GC-MS analysis when you are looking at low levels of analyte and the baseline is jagged. Stationary phases with elevated phenyl content are more prone to jagged bleed, like this Rtx-PCB shown in Figure 1. Acquiring selected ion monitoring (SIM) data, as I have done for all these examples, exacerbates the effect of the jagged bleed if some of your target ions happen to share the same mass as a bleed ion.

Figure 1 - Jagged bleed on the Rtx-PCB column taken up to 340 ⁰C

You may be wondering, “What causes jagged bleed?”

I believe column bleed condensing unevenly throughout the column at the end of a GC run when the oven cools in an uncontrolled manner causes jagged bleed. The vents on the back of the oven open and hot air is blown out, drawing cool ambient air into the oven. This rapid uncontrolled cooling causes random sections of the column to cool faster than others, abruptly condensing the column bleed unevenly in discrete bands throughout the GC column. During the subsequent run, each discrete band of condensed column bleed reenters the mobile phase and chromatographs like any other analyte. The result is random spikes in the baseline with varying heights and widths, each one attributable to the nature of a discrete band of condensed column bleed.

Jaap de Zeeuw wrote an article on ghost peaks for SeparationScience that covers this same topic, but for high cyanopropyl- phases installed in an FID (2013, Volume 5, Issue 7); He also condensed his entire series on ghost peaks into a poster presentation which can be found here on our blog. He came to the same conclusion as me on the root cause, though I think the temperature gradient situation in the oven during uncontrolled cooling is much more chaotic than what he describes as a difference in temperature between the top and bottom of the oven. We know that the four corners of a GC oven are cooler than the center, and that there is a temperature gradient from back to front, but once the oven vent flap opens and the high speed fan turns on, the oven becomes filled with air currents of varying temperature.

Assuming we are right about the cause of jagged bleed, the solution should be controlling the cooling rate of the oven so the GC column cools more evenly, reducing the amount of uneven bleed condensation. Figure 2 shows this is precisely what happened when we added a 20 ⁰C/min cooling program to the end of the oven program for the second run (turquoise). The subsequent run (purple trace) had a much smoother baseline.

Figure 2 - Three consecutive runs showing the effect of controlled cooling on the jagged bleed. The first run (black trace) ended with uncontrolled oven cooling. The second run (turquoise trace) shows the jagged bleed resulting from the first run's uncontrolled cooling, and the descending baseline a result of 20 ⁰C/min cooling from 340 ⁰C to 200 ⁰C added to the end of the GC oven program. The third run (purple trace) shows a much smoother baseline resulting from the previous run’s 20 ⁰C/min cooling.

Jaap’s example implemented a 10 ⁰C/min cooling rate from 260 ⁰C all the way down to 60 ⁰C, but the cyanopropyl phases he was dealing with (1701 and 1301) experienced the region of jagged bleed at a much lower temperature range than the Rtx-PCB.

Figure 3 highlights the difference in data quality for low level Aroclor analysis. Injecting 20 pg of Aroclor 1268 on column, each of the individual congeners are at much lower level. The blue trace highlights how difficult it can be to discern real peaks from the jagged baseline. The black trace, which follows a run with a controlled oven-cooling program, has a much smoother baseline, and the PCB congeners are much more obvious and easier to integrate.

Figure 3 - 20 pg Aroclor 1268 on a 60 m x 0.25mm x 0.25 µm Rtx-PCB column. The overlay shows a relatively smooth baseline in the run following a controlled 20 C/min cooldown to 200C (black trace), and jagged bleed in the run following an uncontrolled oven cooling (turquoise trace).

The effect of the controlled cooling rate added to end of the GC oven program is very reproducible, allowing you to turn a jagged baseline on and off (Figure 4). There are even fingerprint regions where the jagged bleed looks the same in every run with jagged bleed, supporting Jaap’s claim about the influence of positional temperature gradients.

Figure 4 - Sequential Run on an Rtx-PCB illustrating the immediate effect a controlled oven-cooling rate has on jagged bleed. The first run after controlled cooling (3) has a significantly improved baseline. Ending the controlled cooling causes the jagged bleed to return in the following run (6).

I am currently working on a low level PCB congener application by GC-QqQ using the Rtx-PCB. I do not expect jagged bleed to be an issue because of the addition selectivity provided by the collision cell and second mass analyzer, but I will update either way.