|
Addressing concerns in QC tests for GC columns
By Walt Jennings
Professor Emeritus, University of California, Davis, CA
Batch testing is the practice of testing a single column and assuming those test results are representative of an entire production batch. It appeals to some manufacturers because it is much less expensive, requiring a fraction of the labor, equipment, and resources that must be committed to the more exacting process of testing each column individually.
It is patently obvious that results obtained by individual testing are far more veracious due to established truth that within each production batch, every column quality, including bleed, inertness, and efficiency, follows a Gaussian distribution. With batch tested columns, some customers are predestined to receive columns that should have been – and in individual testing, would have been – rejected as scrap.
To minimize the possibility of rejecting good columns, most manufacturers employing individual testing scrupulously maintain their test instruments in mint condition. As a result, packaging each column’s actual test chromatogram with that specific column gives the column purchaser an opportunity that should be – but is not always – exploited. By injecting the same test mixture under the same test conditions, the instrument should produce a chromatogram that duplicates the test chromatogram. If the chromatographic traces agree, the analyst is ready to proceed. However, if the chromatographic traces differ, those differences can be used to diagnose problems in the instrument or operational procedures (e.g., injection technique).
In short, the cause of any variations between the chromatogram that was shipped with the column and the one generated by the purchaser should be tracked down and corrected before putting the new column into service.
New QC tests needed to keep pace with column advances
In a plenary lecture at the 2004 International Symposium for Capillary Chromatography and Electrophoretic Separations, I suggested that the stringency of our current QC tests had failed to keep pace with the advances achieved in our column deactivation procedures. When pass rates gain and failure rates drop, it may be time to design a more critical QC test. I went on to describe a new test procedure utilizing a more demanding test mixture that exposed column deficiencies that had been undetected by a typical “standard” QC test.
- First, tests should be run at lower isothermal temperatures where sorptive forces are stronger.
- Second, probes should be small and sterically unhindered to facilitate their access to the column surface.
- Third, the use of large amounts of low boiling solvents that might drench (and thus shield) active sites during the passage of the probes should be avoided, or at least minimized.
Proposed QC test parameters are explored
The proposed test temperature was 65° C, isothermal, well below that normally used in conventional tests, and the probes included propionic acid, octane, nitrobutane, 4-picoline, trimethyl phosphate, 1,2-pentanediol, propylbenzene, 1-heptanol, 3-octanone, and decane. Even on a poor column, the hydrocarbons (octane and decane) should generate well formed peaks that serve as standards for comparison. When hydrocarbons tail, it is rarely the fault of the column. Tailing hydrocarbons indicate vagaries in the flow path of the carrier gas – faulty column installation, inadequate split ratio, or insufficient or misdirected make-up gas – problems that should be corrected before proceeding further. The solvent used in this test was diisopropylbenzene, which eluted last and required a final temperature sweep.
The response from manufacturers in that audience was muted, but several sophisticated users that were experiencing column activity problems in their more demanding analyses expressed interest. Most notably, this group included Jim Luong1 of Dow Chemical in Canada, who began using the test, and soon faulted the use of diisopropylbenzene, because it often contains impurities, some of which elute long after the solvent. Consequently, its use can prolong test times to an hour or more.
Dow was also interested in detecting column activity at nanogram levels. Luong’s new test eliminated the need for solvent by utilizing a plunger-in-needle micro syringe, and using the gas saver feature of the Agilent GC as a dynamic splitter. The latter was accomplished by employing a high split ratio (1:900) during the injection, followed by activation of the gas saver feature. This resulted in reproducible injections of nanolitre amounts of neat probes.
More rigorous testing leads to breakthroughs in coating and deactivation
Initially, Agilent’s purpose in exploring these new test procedures was directed toward differentiating between “good” and “excellent” columns, none of which exhibited any flaws using our conventional QC test, i.e., peaks were symmetrical, with no tailing, and at full intensity. But on our new more rigorous test, the pass rate on these same columns (all of which had passed the conventional test) dropped to ca. 70%.
This precipitated serious discussions: Could we afford to begin using the new test while competing manufacturers continued to publish their test results using a far less demanding conventional test? Should we institute an educational program so that less sophisticated users would recognize that columns showing defects on the new test could actually be better than columns showing no flaws on conventional tests? While this argument was going on, we were also re-examining our surface pretreatment and deactivation procedures. Breakthroughs in these latter areas yielded gains in column inertness and overall quality that increased pass rates (with the new test) to acceptable levels, rendering both of these questions moot.
Agilent adjusts parameters for high volume testing
While the procedural changes employed by Luong et al. greatly improved the test as originally proposed, the dynamic dilution step in particular would be considered somewhat risky in a high volume manufacturing facility. Agilent authorities deemed it undesirable to use such high split ratios (ca. 1:900) on twenty-five plus test chromatographs simultaneously, and it was decided to use a minimal amount of solvent to dilute the injection, followed by auto injection under normal conditions.
To avoid interferences between solutes, the test mixture was also altered. For DB-5ms Ultra Inert columns the mixture (Über One mixture) was propionic acid, 1-octene, n-octane, 3-methyl pyridine, n-nonane, trimethyl phosphate, 1,2-pentane diol, n-propylbenzene, 1-heptanol, 3-octanone, and n-decane. Figure 1 illustrates a typical chromatogram using the newly designed test report, specific to this individual column.
Figure 2 illustrates results of our first tests on an HP-5ms Ultra Inert column using this same test mixture. HP-5ms and DB-5ms have always exhibited slightly different selectivities because there are subtle differences between their manufacturing processes. With the 2000 merger of J&W Scientific and Agilent Technologies, Agilent’s column production was moved and assimilated into the J&W operation. There were discussions about discontinuing one of these products because of their close similarity; after all, they use the same stationary phase. But we realized that the slight differences in their manufacture do cause slight differences in their selectivities, and different customers have established methods on one or the other. We felt – and still feel – an obligation to continue to offer both columns.
As shown in Figure 2, the resolution for trimethyl phosphate and 1,2-pentanediol is minimal. These are both important probes: Is the tailing entirely attributable to the phosphate, or is the diol also tailing? Because of the close proximity of these two solutes and the desirability of unmasking any defects in the shape of the trimethylphosphate peak (one of the most stringent probes), it was decided to substitute 3-propanediol for 1,2-pentanediol in our QC tests for HP-5 ms Ultra Inert (Über Two mixture). A typical chromatogram with this altered mixture is shown in Figure 3.
Besides listing the test conditions and test probes, the new test sheets illustrated in the figures list Agilent’s specifications for several criteria and the results obtained on each particular column, e.g., theoretical plates per meter, retention indices of n-propylbenzene and 1-heptanol (a check of the column’s selectivity), and the peak height ratio for the highly active trimethylphosphate versus the inactive probe n-nonane.
Ultra Inert column series quality “exceeds my wildest dreams”
In retrospect, when I first broached this subject at that 2004 meeting, I thought it might eventually be possible for both column manufacturers and column users to make a distinction between “good” and “excellent” columns. The history of column development led to the logical assumption that there would be renewed efforts in the areas of surface preparation and deactivation, which I felt would be painstakingly slow. The breakthroughs in surface pretreatments and improvements in surface deactivation came much more rapidly than I had anticipated. The quality of the new Ultra Inert series of columns exceeds my wildest dreams. I am satisfied that customers with the most demanding analyses of active analytes can have confidence that the DB-5ms Ultra Inert and HP-5ms Ultra Inert columns will provide the highest level of performance. I can only congratulate my Agilent colleagues who recognize the importance of continuing column development and innovation to further improve the state of the art going forward.
- J, Luong, R Gras, and W. Jennings. J. Separation Sci. vol 30, No 15, Oct 2007 pp 2480-2492. “An Advanced Solventless Column Test for Capillary GC Columns”
|