Highly variable peak areas (and ratios)

[Badsocref2]

I'm running samples on an old Varian 3900 (FID). This machine has worked like a charm for years, but is now giving me highly variable peak areas for replicate samples. Moreover, the relative peak areas are also varying significantly. For example...

RT Peak Area 1 Peak Area 2 Peak Area 3
4.36 137978 (9.9%) 321234 (23.0%) 939288 (74.5%)
6.24 187362 (12.4%) 287981 (19.0%) 1039281 (78.3%)
7.17 287132 (19.4%) 317039 (21.5%) 873918 (73.38%)

I've replaced the normal consumables, but that hasn't helped at all. I figure something is dirty or leaky or broken, but don't really know where to start.

[Peter Apps]

I'm running samples on an old Varian 3900 (FID). This machine has worked like a charm for years, but is now giving me highly variable peak areas for replicate samples. Moreover, the relative peak areas are also varying significantly. For example...

RT Peak Area 1 Peak Area 2 Peak Area 3
4.36 137978 (9.9%) 321234 (23.0%) 939288 (74.5%)
6.24 187362 (12.4%) 287981 (19.0%) 1039281 (78.3%)
7.17 287132 (19.4%) 317039 (21.5%) 873918 (73.38%)

I've replaced the normal consumables, but that hasn't helped at all. I figure something is dirty or leaky or broken, but don't really know where to start.

Welcome to the forum.

The place to start is by telling us what you are doing; what are you analysing i.e. what is in your samples, what are they dissolved in, concentration, all the settings on the GC - flows, temperatures, pressures, what kind and dimensions of column etc etc etc. Also, what maintenance have you done recently, and has anything been changed; liner, septum, column, etc. The more you tell us, the more we can help.

Peter

[Bigbear]

The main rule of thumb for issues is to look at the last thing that was changed.

[Badsocref2]

fwiw - these are terpine runs using a 624 column.

Parameters
- solvent is methanol
- FID - 280C
- Injector - 250C
- Split - 100

Oven:
75C (hold 4')
-> 120 @ 20°/min
-> 250 @ 2.5°/min

4mm x 6.3 x 78.5 split liner w/ glass frit

I pulled the injector this morning and gave it a thorough cleaning w/ hexane. New rings, seals (I've replaced these before and it didn't make a difference). Signal (while idling) is very steady. About to do a few replicate injections to see if anything got better.

What confuses me is that some peak areas increase while others decrease (for duplicate runs). If they all increased or decreased, I might blame it on the syringe (which is also fairly new).

[Peter Apps]

The more you don't tell us the more we can't help you.

Column dimensions ?, concentration of sample ?, which peaks get bigger and which get smaller and is this consistent from run to run ?, injection volume ?, autosampler or manual injections ?

Hexane does not do a very good job of cleaning inlet liners, especially when you are using methanol as a sample solvent.

Peter

[Badsocref2]

Columns: I've used several - a 60 meter 624, a 30 meter 5, a 30 meter 35. 5 and 35 columns are 0.25 mm; 624 is 0.33 mm.

Concentration: variable, but no greater than 1 mg/ml. I'm not even close to maxing out the system. Peak areas are generally in the 10's to 100s of thousands.

Autosampler, 10 µl syringe - I've tried several injection sizes from 1 to 8 µl. I've replaced the syringe (twice).

There is no consistency. Sometimes two consecutive runs will pretty closely match. Next run, however, might have the third peak increase by 50% while the fifth peak decreases by 20%. I do not see a pattern. I'm using a couple of purchased standard sets. One include a & b-pinenes, linalool, terpinene, myrcene, humulene, caryophyllene, carene, limonene, bisabolol. The second contains these and another 10. I do not think this is about the analytes or their interaction with the columns.

[Peter Apps]

1 mg/ml at 100:1 split puts 10 ng on the column. This is plenty for an FID to see, but if this is your maximum then with more dilute samples or standards you are getting peaks that are small enough to be bothered by baseline deviations, noise and ghost peaks.

What does a blank of just methanol look like ?

An a temperature programme blank with no injection ?

How consistently is the integration finding peak starts and ends ?

Peter

[Badsocref2]

Methanol blank is showing lots of little tiny random peaks that are being picked up by the integrator, and I suspect this is a (and possible the) source of the error. I need to find the source for that noise.

Integrator is good at finding peaks reproducibly. Obviously, larger peaks show a slight lag in their RT.

I don't have an air blank. Will shoot a couple this evening.

I'd lower the split, but some of the peaks that I'm searching for are very close together. Either I slow down the temp ramp (it's already pretty slow), or stay with the 100 split. Will have to see if varying the split varies the noise.

[Peter Apps]

Methanol blank is showing lots of little tiny random peaks that are being picked up by the integrator, and I suspect this is a (and possible the) source of the error. I need to find the source for that noise.

Integrator is good at finding peaks reproducibly. Obviously, larger peaks show a slight lag in their RT.

I don't have an air blank. Will shoot a couple this evening.

I'd lower the split, but some of the peaks that I'm searching for are very close together. Either I slow down the temp ramp (it's already pretty slow), or stay with the 100 split. Will have to see if varying the split varies the noise.

Two red flags here:

If larger peaks have longer retentions that implies column overloading, but your sample concentration and split ratio puts much less on the column than would overload it. So either your samples are more concentrated than you think, or the split is not running at 100:1.

Except under extreme conditions, changing the split ratio will not affect the separation of peaks, so if decreasing the split rally does affect separation (of the peak apices not the overlap at baseline) then that also points to a problem with the splitter.

Your split vent line might be partially or completely blocked. NB that you cannot diagnose this by measuring the flow rate at the split exit. What model of inlet do you have, and does it have filters on the carrier gas and split lines ? If there are filters have they ever been changed ?

Peter

[Bigbear]

Has this method ever worked, or is it new?
If new, try lowering the initial temp of the oven to 40 or so.
Be careful regarding the amount you inject so that you don't overload your liner. Check the expansion volume of methanol at your injection parameters.