LC Troubleshooting Odds and Ends from 2020

For most installments of “LC Troubleshooting” over the past couple of years, I have focused each installment on a single topic or a group of closely related topics. I’ll continue with this approach in 2021, but, for this particular installment, I’ve decided to take a different approach and discuss a handful of apparently unrelated, but nevertheless useful, topics pertaining to LC troubleshooting. These include a theme observed in online high performance liquid chromatography (HPLC) discussion boards and my interactions with users in 2020, a refresh on the very old but very important topic of the chemical stability of reversed-phase columns, and a technical update on the hydrophobic subtraction model (HSM) of reversed-phase selectivity.

One “LC Troubleshooting” Theme from 2020: Degassers

Degassing technology for HPLC has changed tremendously over the past three decades. My current students are amused to learn that in the “old days” we actually did things like strap a tank of helium to the bench and bubble helium into the mobile phase to remove other dissolved gases from the liquid (this is referred to as helium sparging) prior to pumping it into an HPLC column. Degassing units are integral parts of most modern HPLC pumps, and are so reliable that I doubt most users give them much thought. And yet, looking at online discussion boards in 2020, it is clear that users are going to be dealing with problems due to older and aging degassers for a long time.

I’d like to highlight three things here, but first we need to understand how these older degassers work. A major step forward in degassing technology for HPLC was the development of “inline” or flow-through degassers. The basic idea is that the mobile phase is drawn from a bottle through a piece of polymeric tubing situated inside of a vacuum chamber. If the polymer is chosen that it is semipermeable to gases, but not liquids, then, as the liquid moves though this tubing, dissolved gases in the liquid are drawn through the wall of the tubing into the vacuum chamber before being exhausted to the atmosphere by the vacuum pump. Based on known permeability of the polymer for different gases, one can calculate the length of tubing needed to remove enough of the dissolved gas from the liquid before it exits the vacuum chamber to support reliable operating of the pumping system (generally speaking, HPLC pumps really do not like bubbles). Now, onto my points.

The first point I’d like to make is that although modern inline degassers are based on the same principle described above, the amount of tubing inside the degassing unit can vary dramatically between older and newer units. This is a place where a lot of people get tripped up. The amount of tubing inside the degasser becomes relevant as soon as you want to change the solvent running through a particular channel of the degasser and pump, or you want to know that you have completely replaced the liquid inside the degasser channel with “fresh” solvent. In older inline degassing units, the volume of tubing inside the degasser per channel is approximately 15 mL. This means that, when changing solvents, one should draw a minimum of two times that volume of the new solvent through the degasser before actually using that channel for chromatography, and closer to three or four times that volume if a small amount of the prior solvent could have a significant impact on subsequent analyses (for example, changing from triethylamine buffer and UV detection to formic acid solution and MS detection). This can be done either by gently pulling solvent through using a syringe connected to the outlet of the degasser, or by simply running the pump at a high flow rate (typically about 5 mL/min) for several minutes with the purge valve open.

Secondly, one of the weaknesses of older inline degassers is the tubing that connects the vacuum pump itself to the vacuum chamber associated with each of the solvent channels. This polymeric tubing ages over time, especially when exposed to vapors of solvents drawn out of the mobile phase into the vacuum chamber and vacuum pump. If the tubing cracks or breaks, there is no way the vacuum can be maintained, and the degasser will not work. The bad news is that this is likely to happen at some point in the life of these degassers, but the good news is that replacing this tubing can be a relatively simple fix that may keep the degasser chugging along for more years.

Thirdly, a reasonable question is “Does the degassing step affect the composition of my mobile phase, especially if the solvent sits in the degasser for an extended period of time?” Gross changes in the organic solvent:water ratio in a premixed mobile phase are unlikely. However, small changes in organic solvent:water ratio that could impact very sensitive applications, or changes in the concentrations of buffering components such as ammonia, can occur. In this case, the best way to eliminate these changes as a possible effect on a separation is to purge the degasser following a period where the flow has been turned off, so that any mobile phase that might have been compromised is removed and replaced with “fresh solvent.”

Newer style inline degassers tend to have internal volumes well under 1 mL. This means that they can be purged quickly, and that the volume in solvent lines connecting the bottle to the degasser is the more important volume to worry about when changing to a new solvent.

Refresh on the Topic of Chemical Stability of Reversed-Phase Columns

Since the emergence of silica-based stationary phases for reversed-phase LC in the early days of HPLC, the chemical stability of these phases has been a limitation, and researchers across the globe have worked to improve their stability and develop alternatives. In spite of significant advances in the development of stationary phases based on organic polymers and other metal oxides including alumina, titania, and zirconia, silica-based phases are still the dominant type of stationary phase for reversed-phase LC in use today by far. Stationary phases prepared by covalently bonding stationary phase ligands to the silica surface through siloxane (Si–O–Si) bonds have liabilities at both ends of the pH scale. On the acidic side, the stationary phase ligands themselves can be lost from the column as a result of acid catalyzed hydrolysis of the Si–O–Si bond between the silica particle itself and the ligand. On the alkaline end of the scale, the major liability is dissolution of the silica particle backbone (SiO₂) itself. Historically, the conventional teaching has been that when using silica-based reversed-phase LC phases, one should stick to a pH range of 2–7. I see this emerge occasionally on discussion boards, and many bookshelves contain very good, but dated, books that carry this message. Mobile phases buffered below a pH of 2 will result in the slow loss of stationary phase ligands from the column, mobile phases buffered above pH 7 will cause dissolution of particles themselves, and both of these mechanisms can lead to changes in retention, selectivity, efficiency, and mechanical failure of the particle bed. Given that there are good chromatographic reasons to work in mobile phases buffered above pH 7 (for example, increased retention of aliphatic amines), understanding the factors affecting the stability of silica-based phases under these conditions, and development of novel materials to overcome this limitation, have been pursued with great interest over the past few decades. Two frequently cited articles on the topic of buffer and pH effects on silica-based phases above pH 7 are those from Jack Kirkland and co-workers; indeed, these two papers have been cited a remarkable 246 times (1,2).

But things change.

On the acidic side of the pH scale, bulky side chains at the base of silanes used to derivatize silica particles for reversed-phase LC can be used to slow the rate of acid hydrolysis of the siloxane bond tethering the stationary phase ligand to the silica surface (3). These have become known as “sterically-protected” phases, and are quite popular because they enable separations in mobile phases buffered around pH 2 for extended periods of time, even at elevated temperatures. On the alkaline end of the pH spectrum, progress has been slower but still significant. Multiple approaches have been used to improve the stability of silica-based phases in mobile phases buffered above pH 7, including the use of multidentate silanes that provide more than one point of attachment of the stationary phase ligand to the silica surface (3), and changes to the silica particle material itself, such as silica-organic hybrid materials (4,5). Here too, reversed-phase LC stationary phases based on these technologies have become quite popular because they extend the range of pH above 7 that can reasonably be used without significant concerns about method robustness.

Just how far above pH 7 one can reliably work with these materials depends, of course, on other parameters to some degree (for example, buffer concentration and column temperature). However, at a minimum, these materials enable the use of mobile phases buffered around pH 10 (for example, dilute ammonia) for extended periods of time. Users should always follow the recommendations of manufacturers regarding conditions that are suitable for use of a particular column. Legacy stationary phases that are not made using these new approaches for silica-based phases still suffer the same limitations with respect to pH that they did 20 years ago. As a community, we can move on from the old advice to stay within the range of pH 2–7 to updated advice that says silica-based phases can be used well above pH 7 provided that we choose materials designed for this purpose.

Technical Update on the Hydrophobic Subtraction Model (HSM) of Reversed-Phase Selectivity

The hydrophobic subtraction model (HSM) of the selectivity of reversed-phase LC columns was developed in the early 2000s by Lloyd Snyder, John Dolan, and several other collaborators. The HSM assumes that selectivity in reversed-phase LC can be described in quantitative terms using five pairs of analyte–stationary phase descriptors that are associated with specific types of analyte-stationary phase interactions. These include: hydrophobic interactions (the main determinant of retention in reversed-phase LC), resistance to penetration of the stationary phase by bulky analytes, hydrogen bonding interactions (with analytes acting as hydrogen bond acceptors or donors), and electro- static interactions (typically discussed as cationic analytes interacting with anionic stationary phase sites). To date, about 750 commercially available columns have been characterized using the HSM, and the resulting data are publicly available for free through two different websites (https://apps.usp.org/app/USPNF/columnsDB.html and http://www.hplccolumns.org). The principles of the model, its development, and applications to practical problems have been discussed in LCGC at multiple points over the past several years (6–9). One of the most important practical applications of the model—and the initial driver for the development of the model—is using the database of descriptors to find columns with very similar selectivities (“equivalent” for practical purposes) that could be used as a replacement for a column currently in use, or as a backup column in case problems with a column currently in use are encountered in the future. Most recently, I wrote a perspective for this “LC Troubleshooting” column sharing observations about the selectivities of commercial columns released over the past two years, which have been characterized using the HSM (10).

The type of question I’ve been asked most frequently about the HSM in recent years is focused on the extent to which HSM can be used to predict selectivity for a given analyte on a given column. Although the model was not originally developed with this purpose in mind, it certainly is tempting to think that the massive database of descriptors for 750 columns must have some use along these lines. A couple of years ago, my collaborators Sarah Rutan and Tony Taylor began a jour- ney to take a serious look at this question, and try to develop a quantitative sense for what is and is not possible with the HSM in this regard. As part of this effort, along with collaborators Josep Serret and Tina Dahlseid, we carried out a kind of retrospec- tive analysis and refinement of the HSM using all of the retention data collected over the past 18 years in the process of populating the HSM database with descriptors for commercially available columns. Whereas the HSM was initially developed in the early 2000s using data from only type B C18-silica columns, only about half of the columns currently in the database fit this description. The other half of the database is populated by diverse chemistries ranging from short chain alkyl phases to fluorophenyl phases, for example. The results of this retrospective analysis, and refinement of the model to build what we refer to as HSM2, have been published recently in the Journal of Chromatography, A (11). The principal finding of this study is that by considering selectivity data from many types of reversed- phase LC phases commercialized over the past 15 years, rather than just focusing on type B C18 phases to build the HSM, the predictive accuracy (that is, the ability to accurately predict selectivity on column using analyte descriptors measured using other columns, along with the database of column descriptors) of the model can be improved substantially. For example, we found that for the full data set of 8816 retention measurements (551 column x 16 analytes) the number of predictions with errors greater than 10% when using the original HSM was 235; when using the refined HSM2 for the same predictions, this number was reduced to just 25. To me, this result is very encouraging because it shows us that selectivity models like the HSM can be refined to accommodate diverse reversed-phase LC stationary phases. Moreover, further refinement of such models may yield predictive abilities that are useful in method development, albeit with a significant amount of effort to acquire the data required to refine and parameterize these models. Let’s see what the future holds!

References

(1) H.A. Claessens, M.A. van Straten, and J.J. Kirkland, J. Chromatogr. A. 728, 259–270 (1996). https://doi.org/10.1016/0021-9673(95)00904-3.

(2) J.J. Kirkland, M.A. van Straten, and H.A. Claessens, J. Chromatogr. A. 691, 3–19 (1995). https://doi.org/10.1016/0021-9673(94)00631-I.

(3) J.J. Kirkland, J.L. Glajch, and R.D. Farlee, Anal. Chem. 61, 2–11 (1989). https://doi.org/10.1021/ac00176a003.

(4) J.S. Mellors and J.W. Jorgenson, Anal. Chem. 76, 5441–5450 (2004). https://doi.org/10.1021/ac049643d.

(5) Y.F. Cheng, T. Walter, Z. Lu, P. Iraneta, B. Alden, C. Gendreau, U. Neue, J. Grassi, J. Carmody, J. O’Gara, and R. Fisk, LCGC North Am. 18, 1162–1172 (2000).

(6) J.W. Dolan, D.R. Stoll, and L.R. Snyder, LCGC North Am. 35, 660–667 (2017).

(7) J.W. Dolan and L.R. Snyder, LCGC North Am. 34, 730–741 (2016).

(8) P. Carr, L. Snyder, J.W. Dolan, and R.E. Majors, LCGC North Am. 28, 418–430 (2010).

(9) L.R. Snyder and J.W. Dolan, LCGC North Am. 20, 1016–1026 (2002).

(10) D.R.Stoll, LCGC North Am. 37, 168–172 (2019).

(11) D.R. Stoll, T.A. Dahlseid, S.C. Rutan, T. Taylor, and J.M. Serret, J. Chromatogr. A. (2020) In Press. https://doi.org/10.1016/j.chroma.2020.461682.

ABOUT THE COLUMN EDITOR

Dwight R. Stoll is the editor of “LC Troubleshooting.” Stoll is a professor and the co-chair of chemistry at Gustavus Adolphus College in St. Peter, Minnesota. His primary research focus is on the development of 2D-LC for both targeted and untargeted analyses. He has authored or coauthored more than 60 peer-reviewed publications and four book chapters in separation science and more than 100 conference presentations. He is also a member of LCGC’s editorial advisory board. Direct correspondence to: LCGCedit@mmhgroup.com