Agilent 6890N Vs Perkin Elmer

[Jana P]

Hello all

I was wondering if anyone can provide advice in regards to the difference in chromatography that I am seeing between analysis performed on a Perkin Elmer Autosystem XL GC with Headspace Sampler HS40 and our newer Agilent 6890N with G1888 Headspace Sampler.

I am running a method for determination of ethanol in drug product. This method has been in use for a number of years and has been run on the Perkin Elmer with out problems.

The conditions currently employed for analysis on both systems are as follows:

N2 gas at 8 psi
Loop temp 90C
Equilibration 30 min
Pressurise time 2.0 min
Injection time 0.1 min
Air flow 450 mL
Hydrogen flow 45 mL
Inlet 90C
FID temp 260C
Column temp 40C
Hold 10 min
Ramp rate 15C/min
Final 220C
Hold 2 min

The main problem we are encountering is in regards to the recovery we are seeing on the Agilent system. Our recoveries on the Agilent are consistently around 130%. Where as our analysis performed on the Perkin Elmer will consistently give us recoveries of around 110%. All of these are consistent regardless of sample size (weight). What could account for this difference in recoveries? We have run numerous accuracy samples and always have the same result. Does anyone have any advice as to what could be causing this difference in recoveries? We are in a position where altering parameters and conditions is an option and are open to all suggestions.

If any further information is required for you to help answer the question I am happy to provide more information.

On a side note I have noticed that the Agilent system results in poor standard precision (approximately 3.0% where as the Perkin Elmer is approximately 1.0% at worst) has anyone else working with an Agilent system encountered this?

Thank you all

[Peter Apps]

We need some more information; what column are you using, and what is the split ratio (if any), and how are you connecting the headspacers to the GCs ?. Also what procedures and calculations do you use to determine recovery ?

3% rsd is about what you can expect from a G1888 out of the box - I worked on one intensively and it had some problems with temperature and gas flow control. When those were solved by some quite drastic modifications the repeatability got down to about 0.3%.

Peter

[Jana P]

Hi

The column being used is: SGE-50QC5, BP-1, fused silica bonded phase, film thickness, 5 mm, ID 0.53 mm, 50 m Length.

Our headspace is connected to our GC (for the Perkin Elmer and the Agilent) via a transfer line (temp. @ 90C).

To calculate recovery:

Recovery = (Actual Conc. / Theoretical Conc.) x 100

Where: Actual Conc. = (PA test / PA reference) x Conc. reference

Currently the method is run as a Splitless injection. I am in the process of running a few injections on the Agilent system with a split ratio, as much to improve peak shape as anything else.

If any more information is required let me know.

Thank you for taking the time to help

[Peter Apps]

Nothing in the analytical set up jumps out as a probable cause for recoveries above 100%.

How are you calibrating peak area against concentration - from your recovery calculation it looks as if you are using a single point calibration, which would be vulnerable to all sorts of problems. If you are usng multi-point what do the calibrations look like on each instrument ?

How are you connecting the transfer lines to the GC inlet - the two standards methods are to connect them into the carrier gas input line, or to go in through the inlet septum ?

Peter

[Jana P]

Our transfer line is connected through the carrier gas input line. Reconfiguration of the instrument is something i wish to avoid...

We are calculating using a single point. However this method we have been using, with little to no problems, for many years and the single point has always been satisfactory. Additionally the method has been shown to be linear over the range. Not to say that this may not be the cause of our problems now and is somethign that i will have to consider.

I have managed to improve the chromatography that we see on the Agilent with the adjustment of gas flows and introducing the sample using a split injection (5:1) opposed to a Splitless injection. i am planning to re-run the accuracy with the improved chromatography tomorrow. Although my hopes are not high it may have slight effect on the recoveries, otherwise it may be that we need to look at introducing a multipoint standard.

I am hoping improved chromatography will have a "turn it on and off" affect and I will see some improvement...

[Peter Apps]

As a trouble shooting step it would definitely be worthwhile running a multipoint calibration on both instruments - the most likely difference is in the slope of the calibration line and/or the intercept.

Are you using an internal standard ?, and is your standard made up by spiking blank matix ?

Peter

[mcbart]

I will try split the problem in smaller parts to understand where the problem is.

So I will do a test first on Agilent 7890 using HS40 to understand if the problem is autosampler related (at least partially) or not.

[AICMM]

Can I suggest you turn your injector temperature up? 90C seems a bit on the cool side even though your transfer line is at 90. I would suggest something like 150 or even 200 so that you don't get a lot of condensation (such as residual water and ethanol) especially on the underside of the septum.

Just a suggestion.

Best regards.

[krickos]

Can I suggest you turn your injector temperature up? 90C seems a bit on the cool side even though your transfer line is at 90. I would suggest something like 150 or even 200 so that you don't get a lot of condensation (such as residual water and ethanol) especially on the underside of the septum.

Just a suggestion.

Best regards.

I would as a general recommendation agree with AICMM here. I have trasfered a fair amount of methods from older PE HS 40 to Agilent headspace samplers. And at least one physical difference between some PE headspace samplers and The Agilent ones if memory serves are that the Agilent has a loop and sample valve while older PE variants had a flow controlled system. As such the Agilent system is generally more sensitive to contamination and you may get carry over or even discrimination. So I would generally set the valve/loop temperature atleast at the sample solvents temperature or at the temperatue of the solvent with higest bp if it is higher, consequently the transferline is set 10°C higher.

Hmm runing splitless you say. At least for Agilents earlier model before G1888 I seem to rember than runiing splitless was not an good idea (impossible unless total flow was like 10ml/min (split+column)??!!). So going for a split injection may be a good idea.

For the G1888 the Loop fill time is important normally you need to stay around 0,15-0,40min.

Also the injection time of 0,1min can be a problem. 0,1min times a total flow of 2ml/min would give 0,2ml and the typical sample loop is 1ml. So check that the sample loop volume is completly injected on the column.

Hope some helped

Cheers