Dear all,
I'd like to know, in your experience, how long DB-5ms (or equivalent) columns (30m x 0.25mm x 0.25um) should last without loosing performance (in a simple test we inject naphtalene using a isotermic run. The peak shows tailing instead of being symmetrical as in new columns). I've had to move from LC (10 years) to GCMS field and I'd like to get some more 'direct' experience.
Some average data of our biomethods:
CHROMATOGRAPY: GGMS Shimadzu QP-5050, splitless injection (always) split ratio 1:30 - 1:50, inj. temp. 250-300ºC, temp. programm starting at 100-160ºC (depending on the sample solvent, toluene, n-heptane, tol/py, DMSO), temp. rates 10ºC-40ºC, final temp. 280-300ºC, carrrier He (0.8 - 1.5 ml/min), Detection SIM,
SAMPLES: drug/metabolite determination in plasma samples from PK studies. Analyte concentration in plasma: 0.1-200 ng/ml. We mostly use acid-base liq-liq extraction of analytes (mostly basic drugs), sometimes with solvent evaporation-concentration steps. We remove as much lipids as possible (cholesterol and glyceril+fatty acids esters) but they always are presents so that we need to increase column temp. to 300ºC after each run to avoid changes in perfomance after a long sequence of inj. Finally we usually derivatize amines using HFBA/Py or TFA/Py and OH (acids) using TSIM.
We have seen (5 columns) an average length of 600-1000 injections. Sometimes we've observed that columns showing bad peak shapes for naphtalene or neutral substances give nice peaks for the original method (column surface contamination?)
Also, I'm looking for a simple test for controlling column performance that you can recommend.
Thanks in advance.

Marcelo B.

You are shooting samples with a significant percentage of non-volatile material. I would use a guard column for this application, and clip the end off the guard column on a regular basis. The samples you are shooting will also tend to build up residue in the liner relatively quickly. You didn't mention the frequency of liner changes. A plug of quartz wool or deactivated fused silica wool will also help retain residues in the liner instead of getting on column.

The column life you are getting does seem a little short, but I have seen columns fail sooner than that when shooting very dirty samples. Call your column supplier and see if they have an estimate for normal column life for the sample type you are shooting.

You're right, I forgot to mention that we perform 40-80 inj/day during routine analysis of samples from a single pharmacokinetic study (usually consiting of 600-1000 plasma samples). Septum is usually changed every day. Liners are always silanized and we use deactivated silica wool in them. They need to be changed every 100-200 injections depending on the sample. One goal to achieve when we're developing methods is to avoid the need to change the liner during the daily sequence. Neutral drugs are particularly difficult to isolate from lipids. We tried solid phase extraction (SPE) using Si columns but although it was possible to isolate the analyte from lipids, to use 600-1000 SPE columns was far more expensive than a GC column (at least in our country). Sometimes we've also decided to sacrifice some of the sample clean up because of the time-consuming procedure.
Has anybody tried to clean a capillary column with solvent rinsing? does it work?
I'll keep in mind the idea of using guard columns.
Thanks again.

Marcelo

Good idea to archive chromatogram from injection when column was new, for comparison. Our similar columns can last two years or more, depending on type of sample and number of injections. I'll assume your liner and inlet are clean. What we'd try here would be to remove about 20 cm column from inlet end or to rinse the column backwards with 10 ml methylene chloride CH2Cl2.

Solvent rinsing can do wonders for cleaning up nonvolatile or semivolatile residues in the column. It takes some time to do and normally would not be considered very "convenient", however I have seen rinsing revive columns that appeared very dead. It will not work for everything, however it can be something to try.

Estimating column lifetime is very difficult indeed. I usually make it a point not to give estimates on lifetime because they can vary so widely. I know some people who are used to getting
a few weeks of lifetime out of a column for some very difficult methods (high residue content and/or chemical damage by inorganic acids or bases). Then, on the other end of the spectrum, there are people who can get up to 6-7+ years out of columns used for volatiles analysis (low residue content, no chemical damage). Lifetime also is affected by the degree of thermal cycling that the column experiences (how hot and for how long), and also the purity of the carrier gas being used. Thus, there are many variables that go into column lifetime. Biological sample matrices tend to be very harsh on columns, and relatively short lifetimes are somewhat expected.

Jason Ellis
Technical Support Engineer
Agilent Technologies

I'm not sure what do you mean by "splitless injection (always) split ratio 1:30 - 1:50".

Do you inject splitless and then set split to values indicated (after what time?) or "splitless" mode is "on" always (which makes "split" value not valid)?

I'm sorry, what I tried to say with "always" was that my routine injection method is splitless (split valve closed 0.5-1.2 min, depending on the particular method and then, the valve is opened with a split ratio 1:30-1:50, in order to vent the last tail of the sample).

Marcelo