OR WAIT 15 SECS
How much chemistry do you really know?
How much chemistry do you really know?
In 2012 I wrote a column entitled “Know your Analyte…” (1), which highlighted the necessity of knowing something of your analyte chemistry to help with method development or troubleshooting experiences. I feel I need to revisit this topic, especially when considering the ability to troubleshoot problems with high performance liquid chromatography (HPLC) and gas chromatography (GC) separations.
I realize that for many of us, the “test” is the fundamental premise on which we work, that is, the assay of product X for release testing, the stability indicating assay of API XYZ, the measurement of impurity A in formulation B. My point is that when things go wrong, some knowledge of the chemistry of the analytes and matrix components alongside-and this is critical-the chemistry of the separation being undertaken is fundamental in being able to troubleshoot the issues at hand. Of course, here we are not discussing the type of analysis where the structure of unknowns is being elucidated via chromatographic and mass spectrometric or spectroscopic techniques; here you really do need a more thorough grasp of organic chemistry and of data interpretation. I guess this is also a good time to once again confess that my organic chemistry really isn’t too strong, but I’ve battled this through my career and pushed myself to better understand the chemistry-because I feel I only know half of the story if I don’t have that chemical insight into the separation.
So, do you know where to find the structure and physicochemical properties of your analytes? Could you write their formulae from memory? Could you write the structures of the stationary phase being used (HPLC or GC) and do you know anything of the physicochemical properties of the sample diluent solvents being used? Frankly, it’s too easy to become distracted with the instrument hardware and the “method” and lose sight that we are chemists, with a need to understand the chemistry of chromatography to work effectively, troubleshoot problems, and look with a critical eye on the data we are producing.
If you are thinking, “Well we get along fine without knowing all of this stuff-I don’t see the need!” that’s like me asking you if you are happy with your current car. Most people answer yes-until someone points out that it could travel twice as far on half the amount of fuel. Would you still be happy? Essentially, it’s knowing what could be, rather than what is.
So, how can a little analyte and separation chemistry knowledge help us to become better citizens of the analytical laboratory? I’ve recounted below a small selection of examples from my own recent experience in which troubleshooting a problem relied on some knowledge of the sample chemistry, or separation chemistry was key to finding the answer to the problem. Of course, I’ve précised these tales for the purposes of brevity, however the nature of the problem and its underlying cause have been preserved.
A colleague had an issue with a separation where certain peaks within the chromatogram had shifted relative to others and showed poor peak shape. The peaks were broader than expected and one showed a shouldering type behaviour. The clues here were that only certain analytes were affected, which pointed to a chemical rather than a physical issue-a key observation when troubleshooting. On learning more about the analyte chemistry we discovered that the analytes that had shifted in retention time were acidic, with pKa values in the range 3.6 to 4.5. The pKa or partial acid dissociation constant is the pH at which the analyte will spend 50% of its time in the ionized form and 50% in the unionized (charged and more polar) form. The eluent used was 0.1% formic acid and on further investigation it was discovered that the eluent pH was around 4.8-so clearly the aqueous portion of the eluent had not been properly prepared. Actually the operator had prepared the eluent with 0.01% formic acid and the partially ionized acidic species had shifted to earlier retention, the poor peak shape being attributable to the analyte whose pKa was very close to that to the eluent pH-a classic symptom.
In another case, there was a method in the laboratory that seemed very susceptible to changes in operator and lacked robustness, with retention times and resolution changing markedly between operators and instruments. Of course, this type of method is not ideal and not well designed, but I’m sure we all have difficult methods such as this. On further investigation of the analyte chemistry, it was noted that there were three ionized functional groups, two of which were basic and one acidic. We used a freeware program to estimate the pKa values of the functional groups by inputting the analyte structure. The eluent was made up with ammonium acetate and the pH adjusted to 5.0 with dilute acetic acid, which puzzled us greatly because the pKa of the acidic functional group was 5.2; on further investigation of the chromatogram, it was in fact this peak whose resolution appeared to change quite markedly between different eluent batches. We implemented a gravimetric method for preparing the eluent, which reduced the problems significantly, based on the premise that slight variations in the eluent pH were leading to fairly major changes in the ionization state of the analyte and hence its retention relative to other sample components. This was ultimately a poor method, which would have been better dealt with by using an ion pairing approach (perhaps with 0.1% v/v TFA aqueous eluent dare I say!).
Another problematic HPLC method had analytes whose retention times decreased during a long campaign of analysis-the variation being as much as 10% in some cases. This one was fairly straightforward to remedy and on examining the HPLC system with uncapped eluent reservoirs, we were able to quickly pinpoint the issue to the loss of a volatile additive (formic acid) from the aqueous eluent portion. By standing in the warm laboratory the formic acid was preferentially evaporated, the eluent pH increased, and the acidic analytes eluted earlier than they should as their degree of ionization increased. Golden rule: always properly cap your eluent reservoirs-especially in the summer months.
A separation showed poor peak shape with several of the analytes showing an appreciable amount of peak tailing, which was attributed to the age of the column. However, on replacing the column with a new one from the same vendor, the peak shape actually worsened! It became apparent that the analytes were predominantly basic and were fully ionized at the eluent pH used; again we used free software to estimate the pKa of each of the analytes and measured the pH of the eluent. The column was an older type of C18 and had been on the market for a long while and we supposed that the silica used as the stationary phase support was a Type A silica, which contains a higher proportion of acidic (lone or isolated) silanol species. Silanol groups can produce secondary interactions with basic analyte functional groups, which leads to peak tailing, and we proposed that as the original column had been in use for some time, spurious matrix components had become irreversible adsorbed to the support surface and had effectively “endâcapped” the phase for us, leaving a smaller proportion of the silanol groups to interact with the basic analytes. The new column had not been deactivated in this way (that is, coated with sample matrix components) and so the problem of secondary silanol interactions with the analytes had been amplified. We tried the separation again using a more modern Type B column, which contained fewer acidic silanol groups, and the problem was very much reduced.
A liquid chromatography–mass spectrometry (LC–MS) separation suddenly began to show a much reduced response for several analytes within the chromatogram, whilst others remained relatively unaffected. After much head scratching, checking of eluent pH, and cleaning of the instrument ion source, we realized that the affected analytes were in fact basic in nature. What could this tell us about the problem? It turned out that the instrument had inadvertently been switched from positive ion mode to negative ion mode, making it much less sensitive to analytes whose natural charge state is positive. Another salutary lesson in locking methods to guard against “finger trouble”.
There are also many examples in my notebook regarding problems with gas chromatography methods.
One classic “gotcha” that we suffered was a sudden retention shift for selected analytes when using a batch of 5% phenyl polydimethyl siloxane columns (the 5% phenyl phases as they are sometimes known). The manufacturers had sold us a lower bleed version of the phase, which they claimed was much more amenable to MS analysis because of less fouling of the ion source and ion optics from the column’s bleed products. What they failed to tell us was that the phase chemistry had changed and instead of the phenyl functional groups being pendant to the polymeric siloxane backbone, they were now part of the polymeric backbone using an arylene-type chemistry. We never managed to draw any correlation between the type of analyte whose retention time had been affected and the new stationary phase chemistry, but on changing back to our original columns the separations were as they had been before. Again, knowing something of the stationary phase chemistry had helped us to identify the problem.
Another problem involved a sudden change in the chromatographic peak shape and the response of certain analytes within a separation when setting up the method. The problem was quickly rectified by changing to a “new” column of the same stationary phase chemistry, but why the sudden change for the old column and why were only certain analytes affected? It turned out that the affected analytes were mainly polar species and we postulated that the column had been stripped of stationary phase at the head of the column. This exposure of bare (uncoated) silica was a great place for the polar analytes to irreversibly adsorb or suffer secondary (unwanted) retention depending upon their relative polarities. And the reason for the stripping of the phase? Well the previous operator had been developing a new method and was investigating the use of THF as a sample solvent-a solvent notorious for its ability to solubilize the immobilized stationary phase polymer. Removing around a metre of the old (30 m) column showed a much improved response and peak shape-although annoyingly the retention times had shifted out of their designated “windows” in the data system. At least we could use the column, but the operator who used the THF was told in no uncertain terms to stop using the dreaded solvent.
A very similar problem is seen when injection liners age and the peak shapes begin to deteriorate. Effectively the deactivation reagent (typically a chlorosilane) will be removed from the quartz glass of the liner and any packing material via hydrolysis, leaving silanol groups exposed, which again leads to peak tailing and adsorption of polar analytes. Keeping one’s liner in a good state of deactivation is always a good idea when dealing with polar analytes.
My final example is a case of peak shape deformation that eventually led to a shift in retention time of a single peak within a chromatogram. The phenomenon was most pronounced on re-injecting the sample after prolonged standing on the autosampler. On investigating the analyte chemistry we noted that there was an acidic functional group on the molecule (modern GC columns are capable of chromatographing acidic species without prior derivatization providing the liner is new and the column cut is satisfactory). Again, after much head scratching and a few conversations with colleagues in the synthetic laboratory (those folks who really know something of organic chemistry) it was realized that the methanol that was being used as the sample diluent was in fact reacting with the acidic functional group in a classic Fisher esterification, rendering the once fairly polar analyte much less polar!
I know this has been a lengthy instalment, and well done for making it this far. However, I did want to cite several different examples where some knowledge of chemistry was vital in troubleshooting a wide range of problems in HPLC and GC. As I said back in April 2012 (1), learning a little more of the analyte and stationary phase chemistries can lead to a much more insightful and ultimately rewarding career in chromatography. If you are using a method in the laboratory today and you aren’t sure of the chemistry of the separation I challenge you to make time to find out a little more-even though the original method development may have been done by someone else, the job title of analytical chemist tells the world that you should know something of the science.
Contact author: Incognito