Why is low-level oxygen analysis so difficult?

As customers attempt to develop GC methods for lower detection limits of fixed/permanent gases, they often discover that low-level oxygen analysis is more challenging than expected.  They call Restek and ask “Why?”. In this post, I will list several possible issues as to why analyzing for low levels of oxygen can be difficult.

Detectors

First, let’s discuss detectors.  In order to detect oxygen, one would likely use a Universal detector like a TCD (Thermal Conductivity Detector), HID (Helium Ionization Detector), or other similar detectors (BID, DID, PDD, PDID, PDHID).  Detection limits are different for each detector.  Several are listed by the manufacturer as approximately <1ppm.  Therefore, if you are trying to detect lower levels of oxygen, make sure the on-column concentration is greater than the lower detection limit of the detector.  If it isn’t, you may need to inject larger than ideal volumes to achieve the required limits but if you notice the peak becoming very broad or tail, you will likely need to decrease the injection volume for the peak to become sharper.

A final thought on detectors; if you are considering using a mass spectrometer as the detector, contact your instrument manufacturer and ask if they have any data on the detection limits of oxygen and the reproducibility of area counts for this gas from injection to injection.  Often there are so many interfering ions at this lower range that the detection limits are usually much higher than you would think, and area count reproducibility is generally poor.

GC Columns

Next, let’s discuss GC columns for this type of analysis.  In most cases, the column used for light gas analysis is either a zeolite molecular sieve, a carbon molecular sieve or a porous polymer column.  These columns are discussed below.

Zeolite Molecular Sieve columns (5A & 13X)

These columns are often the best choice for separating most light gases, like oxygen from nitrogen, at room temperature.  However, if the surface of the particles is not completely oxidized, various amounts of oxygen adsorption may take place. 

How would you know if the column’s particles need to be reoxidized (also may be referred to as reactivated/regenerated)?  Two common signs are that the CO (carbon monoxide) peak has begun to tail and the response/peak area/ sensitivity of oxygen has decreased or disappeared.

How does one proceed to reoxidze the particles?  A clean (hydrocarbon-free) dry air supply needs to be substituted for carrier gas.  Once the column has been purged with air, begin increasing the GC oven temperature by 10°C/min.  You should confirm that plenty of air will be flowing through the column by increasing the air’s head-pressure to compensate for the decrease in flow as the oven temperature increases. Once the GC oven has reached 300C, the column needs to remain at 300C for 3-hours for the air (as carrier gas) to fully oxidize the surface of the particles.  Verify that the detector is OFF during this process.

To help prevent the surface of the particles from having their oxidation layer removed/stripped, do not condition or dry a zeolite molecular sieve columns at or near their maximum temperatures without using clean (hydrocarbon-free) dry air or nitrogen.  DO NOT use helium, argon, hydrogen, or any other gas as carrier gas during this process.

For additional details, please review the information contained in the link below.

Restek - Molecular Sieve 5A 13X packed columns Installation Conditioning Helpful Hints 

 

Carbon Molecular Sieve columns (ShinCarbon)

These columns are also very good for separating light gases.  Unfortunately, obtaining baseline separation between oxygen and nitrogen at room temperature can be difficult, but unlike the 5A & 13X, CO2 (carbon dioxide) will not become trapped in the pores.  As a result, these remain as the “go-to” columns when using a methanizer to detect low levels of CO and CO2 (as CH4) when using a GC/FID.  

Restek - Using a Methanizer for CO and CO2 Analysis at Low Levels

The link below will provide a general overview of the Molecular Sieve 5A & 13X and the ShinCarbon.

Restek - Molecular Sieve Packed Columns and Fixed Permanent Gas Analysis

As you will see, unlike the 5A & 13X, ShinCarbon columns can be dried at 250C in 30-minutes using most carrier gases (including nitrogen, helium, argon, and hydrogen) without the risk of creating surface activity/oxygen adsorption.  However, according to customers, extensive conditioning at higher temperatures (at temperatures greater than 275C and/or for longer than 3-hours) may cause surface activity.  Therefore, only dry or condition the column long enough to remove moisture and other impurities to obtain a stable baseline.  Do not dry or condition these columns overnight.  Additional details can be found in the link below.

Restek - ShinCarbon columns Installation Conditioning Helpful Hints

 

Porous Polymer columns (HayeSep, Porapak)

Many years ago, very long columns packed with very small particles (100/120 mesh) of porous polymers were used for the separation of light gases.  Often cryo-cooling was involved.  These columns worked OK when the standard TCD (not µTCD) was the most popular detector and the detection limits for gases were >250ppm.   However, as detection limits decreased, two issues became apparent.  Older TCDs were not able to detect these lower levels, and it was observed that porous polymers adsorbed low levels of oxygen.  In other words, these long columns packed with small particles made low oxygen detection limits almost impossible, even when using the more sensitive detectors like a HID and BID.  Therefore, these columns were often replaced by molecular sieve columns as the go-to columns for low-level oxygen (or fixed gas) analysis.  

If you had reviewed one or both of the two links above containing “Installation/Conditioning/Helpful Hints” in their title, you would have noticed a comment about porous polymer columns (see below).

If you experience a decrease in response for the oxygen peak, but the carbon monoxide peak does not tail, the problem may be with the pre-column (stripper column), especially if this pre-column is a porous polymer.

To check if the pre-column is to blame, install the primary analytical column into a GC which has no pre-column (or bypass the existing pre-column in the same GC) and inject an oxygen-containing standard.   Compare the results.

If you are only using one column, and it is a porous polymer, and the column had previously been able to produce an oxygen peak of reasonable sensitivity, there is one last troubleshooting step you may want to try before replacing the column; turn the detector off and cool the detector, GC oven and injection port to 50°C.  Turn off the carrier gas and let the column/instrument pressure decrease to atmospheric.  Switch the carrier gas supply to clean, dry air, and purge the column for 1-hour at 50°C (or less).  Do not increase the GC oven temperature while using air as the carrier gas.  Make sure the detector is OFF during this process.  After you switch back to the regular carrier gas, thoroughly purge the column to remove all traces of oxygen (air) before heating the injection port, oven/column and detector.  Once the baseline is stable, inject a standard containing oxygen.  Has the response for the oxygen peak improved?  If not, try replacing the column.

One last comment concerning porous polymer packed columns; shorter columns are not able to separate air from other light gases.  In most cases nitrogen, oxygen, argon and carbon monoxide will all co-elute.  It may even be difficult to separate methane and carbon dioxide from the air peak.  As a result, these columns are often only used for the separation of carbon dioxide and larger gas molecules.

Reproducibility/Variability

Because air is approximately 21% oxygen, it is very easy to contaminate a sample (or even a standard) if proper precautions are not taken.  The set-up is often a “closed system” which will prevent air from getting into the sample stream or standards.  An example is a valve and gas sample loop(s) which introduce the sample/standard directly into the column or injection port.  A closed system will eliminate any exposure of the samples/standards to air.  This is critical because even the smallest amount of cross-contamination will prevent accurate results.  These closed systems are usually designed by an engineer/chemist/scientist familiar with these systems and a full understanding of the process/analysis involved.

Conclusion

If you need to analyze for low-level oxygen, consider using a molecular sieve column (PLOT, packed or micropacked) and avoid porous polymer columns.  Select an appropriate detector, one which be able to meet the necessary detection limits.  If needed, you may be able to adjust/increase the injection volume of the standards/samples to meet these limits, but if you notice that any peaks are broad or tail, you will likely need to decrease the injection volume.  Finally, make sure your sample collection, sample introduction and analysis are designed prevent any possibility of cross-contamination with air.  By following these suggestions, we hope your low-level oxygen analysis is a success.  Let us know if you have any questions.