RSD problem

[erdno]

Hi all!

I have to measure VPH with HS-GC-FID. I have problems with RSD (and SD).
I make 6 same concentration solution (from a normal alkan-mix, 75 ug/l for each compound) in headspace vials. I am using 20 ml vials with 10 ml water and 4g NaCl.

Headspace parameteres:
7694A HP headspace
1ml loop, 30 min equilibration time, 0,2 min loop pressure time, 0,2 min loop fill time, 0,8 min inject time, 70°oven temp and 85°loop temp, 20 PSI vial pressure, 12 ml/min carry gas from headspace
The carrier and vial pressure gas are N2
GC parameteres:
6890N Agilent GC
N2 carrier gas, 250°injector, 1:1 split, 1ml flow on column, oven: 35°for 1 min, then 12°/min to 140, 40°/min to 220.
I am using at the moment 25*0,2*0,5 column

I am getting for C5 to c7 5,2-6,2% RSD, and with C8 -C9 9-18% RSD.
It is like to be, i am getting discrimination, beacuse for c9, i get 42,8-61,1 aera, compare to c7 77,5-86,3 aera.

I tried to change liner, septum, o-ring, i changed the vial pressure (from 15 to 20PSI), loop fill time (0,1-0,2min) lowered the oven temp (on HS 80 to70°and on GC 40 to 35°), but it is still same.

Thanks for help in advance!

erdno

[kubowicz.tomasz]

Hello

You should check:
-inlet (perhaps there is a leak)
-pressurization valve in HS sampler
-vent valve in HS sampler
Defective valves in HS sampler can cause problems with injection precision.

Regards

Tomasz Kubowicz

[Peter Apps]

Because alkanes are very nearly insoluble in water it is very difficult to set up replicate standard, so the poor repeatability my be nothing to do with the analysis.

How are you making your standard vials ?

If you have 12 ml/min coming to the inlet from the headspacer and 1 ml/min going down the column you cannot have the 1:1 split ratio that you report. What is the flow out of the split vent and the septum purge vent.

How are you connecting the headspacer transfer line to the GC inlet ?

Peter

[erdno]

Dear Peter!

Yes, u are right, certainly i dont have 1:1 split with this configuration. I only do 1:1 split with the GC inlet. When i connect the transfer line (with a needle through the septum), i add about 12 ml flow from the HS with a manometer on the HS. I measure the split flow on the GC ( about 13 ml) with a bubble flow meter so i know i get about 12 ml flow from HS (since i have 1ml split flow from the GC inlet). I have 3ml/min septum purge with this.

About the standard vials. First i put 4g NaCl then 10ml water in the vial, and after it i add 75 ul working solution mix ( i dilute supelco mix in methanol), and finally i add 50 ul fluorbenzene ISTD. (I forget ISTD from my first post, but ISTD has 6-7% RSD too). Then i cap it.

Erdno

[Peter Apps]

Hi Erdno

Are the RSDs that you give for raw areas or for areas ratioed to internal standard ?

Where does the IS elute relative to the alkanes ?

When adding the alkane mix and the IS to the water do you put the tip of the needle down to the bottom of the vial ? Have you tried chilling the water in the vial before you add the IS and alkanes ?

What is the signal:noise like for the light and heavy alkanes ?

Peter

[erdno]

Good morning!

All RSD are for raw areas, without IS. I didnt calculate RSD for area ratio since RSD for IS isnt too good.

The IS come between C6 and C6 alkanes. i tried long time ago using diclorobenzene too, but it was wronger. It eluted after c9 alkane.

When i add mix and IS to the water i usually put the needle under 2-3 cm from the surface of the water. I have never tried chilling the water, the water temperature about 20°C. Why would it be good to chilling the water?

Noise is 0,07 pA, and i get for
c5: 13,5-14,2 pA
c6: 76,8-82 pA
c7: 41,3-46,4 pA
C8: 37-44,9 pA
c9: 28,1-40,2 pA
and fluorobenzene: 1152-1255

Erdno

[Peter Apps]

The whole idea of the IS is to correct for fluctuations. so you need to calculate the RSDs of the areas ratioed to the IS.

You are using 100 times too much IS - the peak for the IS needs to be about the same size as the alkane peaks.

Chilling will reduce the evaporation and diffusion losses of the more volatile alkanes, which is probably the main source of variability.

Are the RSDs that you already told us for peak height or peak area ? If peak height, what are the RSDs for area ?

Peter

[erdno]

When i calibrate, i am using IS to get area ratio for alkanes. I did 3 calibration on last week, but all of them was horrible. For c5-c7 i get 0,9995-9 r2 but for c8 and c9 it is bad. I calibrate it from 10 ug/l to 250 ug/l for each compound. For example: i get almost same area for 50 and 100 ug/l standards (and certainly for c9 too) however we are using a mix and add different volumes to the headspace vials.

The RSDs are for area, i didnt calculate it for height.

Erdno

[Peter Apps]

Hi Erdno

A calibration linear fit can be bad because of poor repeatability (high RSD), or because the relationship is not linear, or because there is no relationship.

You need to sort out the repeatability first.

Have you checked that the vial caps are leak tight ? If you checked, how did you do it ?

You certainly have some molecular weight discrimination. What kind of liner is in the inlet ?

What is the application for this analysis - why do you want to analyse volatile alkanes in water by equilibrium HS ?

[erdno]

Dear Peter!

I always look if vials have leaks. I am doing it by trying to rotate caps by finger, but i am unable to do it.
I tried out 2 different liner. First Singel taper spitless liner with wool(5062-3587 agilent), then without wool. Now i am using straight split liner (19251-60540).

Well i dont have any other instrument to analyse volatile alkanes from water, just HS. This is a new instrument here, and for this measure it should be perfect (cheap, dont need any solvent, fast, and it is perfect for needed LOD).

Thx
Erdno

[Peter Apps]

Hi Erdno

To check your crimping technique put some pentane into a vial, cap it as normal and then submerge it in nearly boiling water. A leak will give a steady stream of bubbles.

What kind of samples do you need to analyse using this method ?

Adding standards to the vials is still the most likely source of variability - did you try chilling the water ?.

The discrimination against C8 and C9 might be due to adsorption. Try increasing the loop and transfer line temperatures by 20 or 30C. Also reduce the inlet temperature to 180C - at 250C the gas exiting the transfer line has to expand suddenly and this might cause variable split ratios.

How variable are the area ratios between adjacent alkanes ?

Peter