This month I talked with an ex-colleague who leads a team at a lab in the Greater Toronto Area. His lab has a high-resolution mass spectrometer and a triple quadrupole LC-MS. Both are idle most of the time. The instruments are fine. There’s nobody free who can run them at full capability.
I’ve heard a version of this from more than one lab since I started calling around the GTA this summer. The instrument is paid for. The samples exist. The person who can connect the two is missing.
I have 14 years in analytical chemistry, and a good part of that on GTA benches: Canadian Analytical Laboratories, Diteba, Dalriada, the Ontario Ministry of Environment. Before that I trained 200+ students on SPME and LC-MS in the Pawliszyn lab at Waterloo. So I’ve seen both ends of this pipeline: the graduates coming in, and the labs trying to hire.
The two ends don’t meet.
Universities stopped teaching this
In July, Tony Edge, president of the Chromatographic Society in the UK, published an article in LCGC called “The Silent Crisis”. His numbers: some UK BSc chemistry programs now average about 3 hours of chromatography teaching across a 3-year degree. Top European universities teach roughly 20 times that.
Canada is not the UK, and I have no Canadian numbers to put beside his. But the direction sounds the same from the graduates I meet. Chromatography compressed into one analytical methods course, taught by someone whose own research is in another area, with a lab that runs like a recipe.
Meanwhile the GTA is full of labs that depend on chromatography every day: pharma QC, CROs, cannabis testing, natural health products, food and environmental labs. Their instruments get more sophisticated every year. The training behind the person at the keyboard gets thinner.
And the modern instruments hide the damage. An autosampler, automatic integration, self-diagnostic software: a new analyst can load the vial and press start without understanding the separation. Edge calls this the black box effect. The analyst watches a monitor. The partition happening inside the column is theory from a course they barely had.
Edge draws the difference between the two kinds of analyst in a table. My version of it, from what I’ve seen on GTA benches:
| Trained on fundamentals | Trained on software | |
|---|---|---|
| Method development | Picks the column and buffer pH from pKa and log P | Copies the last method, or clicks auto-optimize |
| Troubleshooting | Reads a failing check valve from the pressure trace | Waits for an error message, then calls service |
| Data interpretation | Checks peak shape and baseline placement before accepting a result | Accepts the integrated area as truth |
| Run time | Sets the flow from the van Deemter optimum for that column | Runs 1.0 mL/min because it is the default |
| Instrument care | Replaces ferrules, seals, and check valves; cleans the source | Treats the instrument as a sealed unit |
Adapted from Tony Edge’s comparison in LCGC, July 2026.
What the gap looks like on the bench
I can make this concrete, because the failure modes repeat.
The analyst runs every method at 1.0 mL/min. That is what the software opens with, and the analyst does not check the van Deemter equation to optimize it. So the run is longer than it needs to be, on every sample in every stability study.
The pump pressure trace shows a ripple that could mean a check valve is failing. The analyst may not know how to read pressure traces. So the first sign of trouble is a failed sequence on a Friday night.
The software integrates a peak and the analyst accepts the number. It was not checked where the baseline was placed. In a GMP lab this is worse than slow. If your analyst can’t explain to an auditor why an integration parameter was chosen, that data is hard to defend. I’ve sat in enough audits to know this question comes.
And method development becomes guessing. Change one factor, run, change another factor, run again. Edge calls this one factor at a time, and he puts the cost at a method that runs in 30 minutes where a well-optimized one would run in 5. A chemist who can use pKa and log P picks the stationary phase and the buffer pH from first principles and gets there in days. Guessing takes weeks, and it often lands on a method that fails robustness a year later.
I made these mistakes too
I’m not writing this from above. At one Mississauga lab, I lost 2 days to a fronting UPLC peak. I changed the column. Same peak. I re-prepped every sample. Same peak. The problem was my sample diluent, which was stronger than my mobile phase. A 30-second check, once you know to make it.
Edge’s article lists poor sample diluent, along with injection solvent strength too high for retention, among the classic mistakes of graduates who never got the fundamentals. Reading that stung a little. It also proves a point: these gaps are specific, and they are teachable.
Build the people you already have
The senior LC-MS people every lab wants were built on the bench over 10 or more years, and most of them are employed. When one leaves your lab, the posting can stay open for months, and the knowledge leaves the same day.
The realistic option is to build the analysts you already have. Edge estimates a new hire needs 6 to 12 months of internal retraining before working without supervision, mostly because senior scientists must teach basics a degree should have covered. Focused hands-on training compresses exactly that part.
The Chromatographic Society reached the same conclusion. Their answer to the crisis is a 4-day hands-on course where inexperienced chromatographers take instruments apart and learn the separation variables on real hardware.
I do the same thing inside your lab: your instruments, your methods, your analysts. We work on the misbehaving method you already have, and I explain the why behind every step until your analyst can defend it without me.
If your lab has an instrument running below its capability, or a method that only one person understands, or an analyst who runs the system but can’t troubleshoot it, that is fixable. It takes days.
Questions, you can reach me: [email protected].