The LCGC Blog: Chromatography Technical Tips - Sweating the Small Stuff

My grandfather was a mining engineer and had a love of working with all things mechanical, especially steam trains.  It seemed to be the harmonious way in which many, often intricate, parts needed to work together in order to allow complex machinery to function that resonated so strongly with him. I remember him telling me that very often, when something mechanical goes wrong, rather than being the result of the catastrophic failure of one component, it's often the accumulation of smaller issues and problems that result in the output being unacceptable – in his case the train not moving or the trepanning machine not cutting coal!

I've often reflected back on this during my working career, whilst recalling one of my grandfather's favourite phrases - "someone who can't be bothered to care for smallest bolts, can't be trusted to care for the biggest engines".

In this issue, I hope to outline some "small bolts" that are commonly ignored in everyday chromatography and mass spectrometry - instrument settings that I have to come call "lock-and-leave" parameters. We can generate data without changing them or paying them any attention, but how much better could our data be if we bothered to optimise them. As I've said so many times to folks who say "why bother to change this variable if we don't have a problem" - "who defined the word problem in your previous sentence?"

1. Capillary voltage in Electrospray LC–MS: Above the onset potential – it's possible to generate a signal using a wide variety of voltages in electrospray mode.  There are many primary and secondary modes of charging which will ensure that charged droplets are fired across the API interface, ultimately resulting in charged analyte ions in the gas phase being samples in the first vacuum region of the mass spectrometer.  Hence the oft used phase – "Oh, that instrument works best at 4kV" – does it?  There are regions in the applied potential v's signal intensity curve which are highly sensitive to small changes in eluent flow or eluent chemistry, as well as pseudo spray modes, such as rim emission mode, which are inherently less robust that a "proper" electrospray emanating from the tip of a Taylor cone formed at the end of the sprayer capillary.  Each time you change the eluent chemistry, flow rate, organic modifier, analyte type or sample matrix, you should consider re-optimising your sprayer potential.  You may increase your sensitivity, reduce baseline noise and improve your experimental robustness by doing this.  Also consider optimising the nebulising gas flow rate at the same time

2. Solvent compressibility settings in your HPLC pump: A pump doesn't know what solvent it's pumping until you tell it the compressibility of the liquid you are feeding into it.  It can't therefore adjust the piston stroke length to mix the most reproducible and accurate gradients possible.  It will also result in baseline ripple – which may not show up with a UV detector monitoring higher wavelengths, but you can be sure to see it in your RI detector or if you happen to get a bubble in the flow cell of your UV.  To improve gradient reproducibility and reduce baseline noise, set the correct solvent compressibility into your data system or hardware.  If you are using pre-mixed solvents, follow your manufacturer's instructions on how best to estimate the compressibility.

3. Correct column dimensions in your HPLC and GC system: Your HPLC system will use the column dimensions and hold up time (t0) to calculate several of the important chromatographic descriptors (efficiency, retention factor, selectivity etc.).  Only by accurately entering these details will the system be able to calculate these values accurately for you.

GC column dimensions (accurate to the nearest cm in column length) are used to calculate the correct flow rate via pressure measurements.  Without accurate column dimensions, the correct flow and linear velocity cannot be established.  The actual column length can be calculated by injecting a non-retained solute (or use the first baseline disturbance created by the eluting solvent) and entering this value along with the column nominal internal diameter into one of the many free GC calculators available.

4. UV detector settings: I wrote in this blog recently about the importance of the 'minor' settings in UV detection. Parameters such as reference wavelength, bandwidth, slit width and response time are very important in determining the sensitivity and reproducibility of your results as well as their validity in, for example, determining peak purity.  If you don't know how to set the sample wavelength and bandwidth, the reference wavelength and bandwidth (or even if you need one these!), the slit width, sampling rate or response time – I strongly urge you to read the blog above.  If you have never changed any of these things – you are certainly not payng enough attention to your methods!

5. Injection volume in GC and HPLC: Whilst this isn't such a 'lock and leave' parameter, injection volume in both techniques can certainly be the cause of method problems.  In HPLC, using large solvent injection volumes in combination with highly eluotropic sample diluents can cause real problems with peak shape and resolution, especially for early eluting compounds.  SO the injection volume and the nature of the sample diluent need to be carefully considered.

In GC, the liquid sample expands manifold as it vaporises and an injection volume of 1 ml may expand to 1 mL or more depending upon the solvent used and the conditions within the inlet.  If this volume is greater than the internal volume of the liner, then 'backlash' can occur which can ultimately result in carry-over and poor peak area reproducibility.  If you haven't checked your method for the potential of blackflash using a vapour volume calculator – then you should do!  You never know – you may be able to gain extra sensitivity by injecting more sample!

6. HPLC pump flow gradient: Every time you start your HPLC pump, the pressure increase can cause mechanical damage to the packing material in the column.  The silica substrate particles are relatively fragile and can fracture, ultimately disrupting the packed bed, forming voids at the column head which lead to shouldered and split peaks, and increased column back pressure as the fines migrate towards and block the column outlet frit.  Some modern pumps have 'flow ramp' settings (usually in mL/min/min).  This is the maximum ramp rate at which the flow, and hence the back pressure, are allowed to increased.  By having the pressure build up gradually, the mechanical shock on the packing material is reduced and columns last longer.  If you pump does not have this setting, try to increase the flow gradually over time in order to maximise column lifetime.  Going from 0 to 1.0 mL/min. flow should take at least one minute to achieve!

7. GC Column Equilibration Time: So – this isn't the time taken to cool the oven or the time at which the oven is held at the top temperature.  It's the time which elapses between when the instrument reports that the GC oven is at the correct initial temperature to start the next run and when the next injection is actually made.  The column and the carrier gas have a finite thermal mass, and as such they need time to cool and heat to the correct temperature.  So – just because the instrument says that the column oven is at the correct temperature, doesn't mean that the column and carrier will be.  It's very wise to have a column re-equilibration time in the method to allow the column to reach the same temperature as the oven – each and every time.  This avoids irreproducible retention times and problems with resolution between earlier eluting analytes.  I usually find that 30 secs. to 1 min. are enough to give good retention time reproducibility.

8. Threshold and Sampling Number in GC–MS: These are parameters which are called different things by different manufactures – so you should check your particular manufacturers nomenclature.  The threshold is the minimum number of ion counts above which a signal will be recorded at that mass to charge ratio value.  It's essentially a good way to reject noise in order to gain sensitivity and reduce background spectral contribution in trace analysis.  Care is needed when optimising this setting to ensure that valid data is not being rejected, however optimising this setting can bring many benefits in terms of spectral quality and signal to noise improvements.

Sampling number (sometimes called Oversampling number) is the number of measurements taken at each mass to charge ratio.  For qualitative work, its better to set this number higher, so that although fewer spectra will be take across the peak, the quality of the spectra will be higher with less evidence of spectral tilting.  Remember to take a half height peak averaged spectrum for good library matches!  For quantitative work the sampling number should be lower in order to obtain enough data points across the peak for a valid and reproducible quantitative measurement.

As mentioned above, you may not have a 'problem' as such – but do you have room for improvement?  Are you paying attention to the small bolts?  Can you be trusted to care for and operate the big machines?

For more information, contact either Bev (bev@crawfordscientific.com) or Colin (colin@crawfordscientific.com). For more tutorials on LC, GC, or MS, or to try a free LC or GC troubleshooting took, please visit www.chromacademy.com