All’s Well That Ends Well

John W. Dolan, LC Troubleshooting Editor

In his final column before retirement, John Dolan reminisces about his years as the author of “LC Troubleshooting”, and some of the changes that have taken place in liquid chromatography (LC) technology during that time.

I was 34 years old when I wrote my first instalment of “LC Troubleshooting”-in October 1983, 34 years ago-so I’ve spent half my life writing for LCGC. LCGC, then known as LC Magazine, was dedicated to liquid chromatography (LC) and related topics. LC Magazine had published its first issue at Pittcon in March 1983, with Dennis Runser writing the “Troubleshooting” column, as it was known then. Runser wrote three columns and then decided that it wasn’t his calling. I was approached in the summer by the editor as a candidate to take over the “Troubleshooting” article. I was a little unsure about taking on the responsibility of writing a 2000‑word article every month, so I asked a few colleagues for advice. After receiving their recommendations, I recruited the late Vern Berry, a professor at Salem State College, to join me at the task. Vern and I wrote 10 columns together over the next year, but by then we had discovered that we weren’t ideally matched-it took longer to write a column together than to write it alone (those were the days of typed manuscripts sent back and forth through the mail)-so I took over as the sole column editor in October 1984 and Vern moved on to reviewing various scientific meetings.

If my count is correct, this is the 390th column I’ve been involved with for LCGC. After the first year of partial publication, LCGC published 11 issues of editorial content each year, with one additional issue (August) dedicated as the “Buyer’s Guide.” In 2005, a regular August issue was added and the format changed to 12 full issues per year, as it is today. 

My writing colleague for most of my LCGC experience was Ron Majors, who wrote “Column Watch” and “Sample Prep Perspectives” before his retirement about two years ago. Ron had the skill of using many different contributors-it seems like half his columns drew on contributions from other scientists. However, I found it more work to recruit contributors, edit content, and manage deadlines than to take charge of the content myself. That is not to say I didn’t rely on others. There were approximately 60 others who contributed columns or coauthored them with me; over a third of those were from the staff at LC Resources, the company that Lloyd Snyder and I started in 1984. For about 15 years, part of my responsibility at LC Resources was to manage our contract research laboratory. The laboratory seemed to be a bottomless pit of case studies of LC-related problems and their solutions. The joke in the laboratory for many years was that if you really messed up, you didn’t get fired-you became a coauthor on an “LC Troubleshooting” article. 

More recently, the source of many column topics came from reader questions or questions that came up in my short-course business. These days I typically field one or two reader questions a week via email. 

One thing that is quite clear is that the frequency and variety of problems encountered by typical LC users have diminished greatly since I made my first injection in 1972. As a graduate student, I did not have a budget to buy a LC system, so I did as many others were doing at the time-I bought components, created my own system, packed my own columns, and used a ruler and handheld calculator to manually compile the results from chromatograms traced in ink on a strip-chart recorder. 

For the rest of this column, I’d like to look at some of those early LC systems and their related problems and contrast them with today’s equipment and processes. Some of the companies mentioned below were purchased by other vendors or simply went out of business. A disclaimer is appropriate here: I’m writing from my perspective and relying on memory for dates, so you old-timers might remember things a little differently than I’m reporting.

Pumps and Degassing

The early LC pumps were simple reciprocating-piston pumps modified from other applications. Single-piston pumps, by design, spent half their time filling and the other half delivering solvent. This design resulted in pulses of flow and pressure that were hard on columns and often resulted in wavy detector baselines. Mechanical damping of the pulses was helpful, but it didn’t cure the problem and added significant volume to the liquid path of the system. Early versions of dual-piston pumps improved things, because the two pistons worked 180° out of phase, resulting in smoother flow and pressure. Today’s pumps use two pistons, either in parallel or tandem, but add sophisticated electronic control of the piston speed that makes them able to produce a very steady output stream with very small-volume pulse dampers.

Two pump problems that were common until the 1990s were premature pump-seal wear and check-valve failure. The early pump seals were simply parts that were in current production for other industrial applications-often automotive. The polymers were not necessarily matched to the solvents used for LC separations, so solvent compatibility was sometimes a problem. Pump piston alignment in early pumps was pretty crude, with the piston held rigidly in place. Unless the piston was aligned perfectly, the fixed alignment put additional strain on one edge of the seal, causing premature wear. Later, the design was changed to a floating piston mount, in which the piston self-aligned with the seal, greatly reducing seal wear. As pump seals wear, they can shed small particles of debris that can cause check valves to leak and block column frits. Seals were often changed on a monthly basis, and I remember an LC setup in the lab where I worked in the late 1970s that required the pump seal to be changed every week. Today’s improved polymers and highly polished pistons mean that pump seal lifetimes seem to be indefinite, so blocked columns and contaminated check valves are much less of a problem, too. Most workers change the seals during an annual maintenance session, which is a good idea, but it is likely that the seals might last much longer.

Most of the early LC work was in the isocratic mode using premixed mobile phases, so mobile-phase outgassing was generally not a big problem. The first online solvent mixing was done under high-pressure-mixing conditions, where the output of two pumps was combined. Because mixing was done under high-pressure conditions, bubble formation was minimal. When degassing was required, most workers used vacuum degassing to lower the dissolved air content of the mobile phase to reduce bubble problems. Degassing was accomplished by placing the solvent in a vacuum flask and applying a vacuum from a water aspirator or vane-type vacuum pump for a few minutes. In many cases, the process of filtering the mobile phase using a vacuum filter provided sufficient degassing for reliable operation. 

With the introduction of low‑pressure mixing by Spectra‑Physics in the late 1970s, degassing problems became much more serious. Mixtures of water and acetonitrile or methanol have a much lower capacity for dissolved air relative to the pure solvents, and the result was that solvent out‑gassing often occurred on mixing. Bubbles from the solvent interfered with reliable pump operation, causing pulses in flow and pressure, and often creating an air lock in the pump. Bubbles passing through the detector caused large spikes in the chromatogram. A clever solution to this problem was helium sparging of the mobile-phase components, either batchwise or on a continuous basis. This process was patented by Spectra-Physics and enabled the company to introduce a reliable four-solvent, low-pressure-mixing LC system in the late 1970s. Helium sparging displaces enough of the dissolved air that solvent outgassing does not occur when water and organic solvent is mixed. Helium is much more soluble in such mixtures, so it stays in solution. Helium sparging resulted in a huge advance in system reliability. Because the patent was nearly impossible to enforce on an individual basis, many workers built their own helium degassing apparatus from simple parts.

In the 1990s, on-line vacuum degassing was introduced. The early systems comprised thin‑walled polymer tubing routed through a vacuum chamber. When solvent was pumped through the tubing, dissolved gas passed through the walls of the tubing and liquid continued downstream. These degassers were not very efficient, but later development of systems using specialized polymer membranes allowed an increase in degassing efficiency and lowered the volume of the tubing from as much as 10 mL per liquid channel to less than 1 mL. Today, nearly all LC systems incorporate online degassers and many workers have never experienced regular problems with air bubbles in the pump. To give you a feeling of the significance of the impact, consider that LCGC user surveys in the 1980s indicated that bubble problems were the number one problem with LC systems. This survey result is validated by the fact that more than half the questions I fielded from readers were related to bubble problems. I’m not sure when I last received a question about bubble problems, but it has been years, so I know that bubble problems are not an issue today.

Columns

When I started doing LC in 1972, it was just at the transition point from dry-packing 50–100-cm long columns with 37–44-µm silica particles. This technique was quickly displaced with the introduction of slurry packing techniques, where the particles were suspended in a viscous solvent and pumped into the column. With that development, 10-µm particles could be packed, and higher plate numbers generated by smaller particles meant that a 250-mm-long column was sufficient for most work. Although commercially packed columns were available, many workers purchased 10-µm silica with the C18 phase pre-bonded or added their own bonded phase and packed columns in their own labs. Column packing became a highly refined art, and proprietary column packing techniques became closely guarded company secrets. A lab‑packed column is a rarity today.

Column particles progressed from irregular particles of crushed and sieved, naturally occurring silica in the early 1970s to synthetic particles that gradually evolved from 10 µm to 5 µm to 3 µm to the sub-2-µm particles we see today. In this process, particle geometry changed from irregular to spherical and the size distribution of some of today’s particles is nearly monodisperse with ≤6% relative standard deviation in diameter. The nature of the silica also changed from what is referred to as “type-A” silica, characterized by acidic silanol groups and metal contamination, to today’s high‑purity “type-B” silica synthesized from highly purified reagents and containing a low population of acidic silanols and practically no trace metals. 

The practical impact of these changes in column technology has been to reduce or eliminate many problems of the past. Well into the 1980s, column stability was a problem-even for new columns. It was an observation in the industry that approximately 15% of new columns that passed specifications at the manufacturer failed when tested under the same conditions by the user. As a result, nearly everyone tested every column as soon as it was received-the manufacturers routinely replaced failed columns. Further problems resulted from difficulties generating consistent column chemistry from batch to batch, especially over a period of years. As a result, some users reserved extra packing material so that columns for a particular method could be ordered from a specific packing batch. This column inconsistency also resulted in the recommendation that new methods be tested during validation with columns from different batches. Today, neither the stability nor selectivity is a problem for most columns, so users seldom test new columns-they assume with confidence that each column will perform the same as the last one.

Settling or bed collapse in the early columns was common, and was caused by mechanical instability in the packed bed. The column bed was disturbed during the shock of shipping or by the pressure pulses the column experienced from the pump or when the injection valve was rotated. To help mitigate pressure shock during injections, new valve designs were introduced. Waters developed the U6K manual injector, which included a bypass channel such that part of the flow to the column was never shut off during the injection, so pressure shocks were reduced. The bypass design had its own problems, especially when partial blockage of the valve occurred. Another solution to this pressure shock problem was introduced with Rheodyne’s make‑before-break (MBB) valve, which prepressurized the sample before it was injected. Other solutions included faster valve actuators. Today’s columns are sufficiently pressure-stable that none of these techniques are needed anymore.

The type-A silica was prone to exhibit strong peak tailing, especially for basic analytes. Peak tailing was such a problem that triethylamine (TEA) was an almost universal component of reversed‑phase mobile phases. TEA is a low-molecular-weight amine that binds strongly to acidic silanol sites, providing a sort of dynamic endcapping. After the acidic silanols were blocked, tailing was greatly reduced. Today’s high-purity type-B silica columns are much less susceptible to peak tailing, so tail‑suppressing additives are seldom used.

The instability of columns meant that it was not uncommon that the column would “settle” a millimetre or more during use. This settling caused unacceptable band broadening, so it was common to remove the inlet frit and fill the void with packing from an old column. Frit replacement because of blockage caused by particulates from pump seals, samples, and other sources was so common that most manufacturers shipped new columns with a few extra frits. Now frit replacement is strongly discouraged. Columns are packed so tightly today that if the frit is removed, the column packing will usually ooze from the column and ruin it.

Best Practices

We’ve looked at some of the changes in instrumentation and column technology that have reduced LC-related problems over time. One of the benefits that workers in the 1970s and 1980s had was that LC systems and columns were meta-stable-they might malfunction at any time and often without warning. As a result anyone using LC on a regular basis became pretty good at troubleshooting and repair-it was a survival skill! Today’s equipment and columns are so reliable that it is possible to run routine methods for weeks or months with only minor maintenance and seldom experience any surprise failures. This reliability is good news in terms of productivity, but can be bad news if we have to rely on specialists to fix any problem we encounter. 

One way to further reduce the chance of encountering LC problems is to follow a few “best practices” when using the LC system. I’ll list just four here that will go a long way towards improving system reliability:

Keep It Clean: Keep the system clean by using high-quality mobile-phase components and filtering them when appropriate. Flush the system when you are finished with a batch of samples to remove strongly retained materials from the column and any salts or buffers from the hardware before it is shut off. Practice appropriate sample pretreatment to prevent easily removed contaminants from entering the column and use a guard column if it makes sense for your method.

Dedicate the Column: Use a single column for every method-don’t share a column between methods. You will find that your column budget will be lower in the long run and that problems caused by cross-contamination of the column are eliminated.

Preventive Maintenance: Put together a preventive maintenance plan. It should include the system-flushing steps mentioned above as well as regular replacement of wear parts, such as seals and filters.

Remember the Economics: When it comes time to repair or replace, consider whether it makes economic sense to repair. We violate this general rule most commonly by trying to extend column lifetimes. I figure that a column doesn’t owe me anything after 500 injections, and 1000–2000 injections is a reasonable lifetime even for a method with dirty samples, such as precipitated plasma. On a per-sample basis, the column is one of the least expensive components of LC analysis. It doesn’t make sense to spend much time trying to fix a problem column, and if you do “fix” it, it may still not perform as well as a new one.

Conclusions

I’ve highlighted a few changes in LC systems and columns that I’ve observed over the time I’ve been involved in this technology. Fortunately, almost all of the changes have been for the good, resulting in much more reliable systems and methods. However, not all is perfect. I love the quote attributed to Jesse Ventura, “You can’t legislate against stupidity”. Although many improvements have been made, the users-you and me both-can be a weak point in the process if we don’t pay attention. 

I’ve enjoyed working with many individual readers of this column over the last 34 years, and I’ll miss that. I’ve already received some messages that you’ll be missing me, too. If you feel like wandering more down memory lane, all of my columns are available as pdfs on our website: www.lcresources.com/tsbible.

And don’t despair-you are in very good hands as Dwight Stoll takes responsibility for “LC Troubleshooting” next month. Dwight may be a relative newcomer as a column editor, but he’s no newcomer to LC. He’s been a practical worker in the lab for many years and is a world-respected leader in two‑dimensional LC. His perspective will be different than mine, but that’s a good thing. I’m confident in the transition. Shakespeare’s title “All’s Well That Ends Well” is apropos for the completion of my contributions to this column.

“LC Troubleshooting” Editor John Dolan has been writing “LC Troubleshooting” for LCGC for more than 30 years. One of the industry’s most respected professionals, John is currently a principal instructor for LC Resources in McMinnville, Oregon, USA. He is also a member of LCGC Europe’s editorial advisory board. Direct correspondence about this column via e-mail to LCGCedit@ubm.com

Dwight Stoll is the incoming editor for “LC Troubleshooting”, effective November 2017. Stoll is an associate professor and co-chair of chemistry at Gustavus Adolphus College in St. Peter, Minnesota, USA. His primary research focus is on the development of 2D-LC for both targeted and untargeted analyses. He has authored or coauthored 48 peer-reviewed publications and three book chapters in separation science and more than 95 conference presentations. He is also a member of LCGC’s editorial advisory board.