how to calibrate the injector

[slimshady]

I want to verify the injection volume of my chromatographic system. How can I do that in usual process?

Now I inject 10ul 25ul 50ul 100ul with the same solution(same concentration). My sample loop is 100uL.

the area of the injections are:
10uL 36.65
25uL 80.37
50uL 159.35
100uL 323.92




and today i did the same injections again for 50uL and 100ul, the area becomes larger.( 50uL 182.18, 100uL 345.39). And the whole peaks elute later than yesterday.

Any possible reasons>>??
And how does everyone do such kind of calibration of the injection volume?

[Rolandas Plausinaitis]

Hi,
If peaks elute later they will have slightly different shape. This is most likely case of peak area change.

To check injection volume You have to weight the vial with the sample, make 10 or 20 (or other suitable number) of injections from that vial. After second weighing You will know weight of sample injected which can be recalculated into volume. Of course easiest way is to use water (column can be removed, just use restriction capillary after the injector).

[DR]

Hi,
If peaks elute later they will have slightly different shape. This is most likely case of peak area change.

To check injection volume You have to weight the vial with the sample, make 10 or 20 (or other suitable number) of injections from that vial. After second weighing You will know weight of sample injected which can be recalculated into volume. Of course easiest way is to use water (column can be removed, just use restriction capillary after the injector).

^ only works on some injectors. If your injector does an overfill on partial injections (as many brands do), this approach will not work. Most universal way is to do a series of injections like you have done and compute r² (linear regression, volume injected vs. peak areas (or heights).

Different retention times are probably the cause of different areas. Different retention times are caused by variations in mobile phase composition, temperature, column chemistry/equilibration...

[dave.cheng]

^ only works on some injectors. If your injector does an overfill on partial injections (as many brands do), this approach will not work. Most universal way is to do a series of injections like you have done and compute r² (linear regression, volume injected vs. peak areas (or heights).

Different retention times are probably the cause of different areas. Different retention times are caused by variations in mobile phase composition, temperature, column chemistry/equilibration...[/quote]

the peak area variations issue generally come from three aspects:
1) the column issue:the column stability and repeatablity etc. you can used a connector to replacement the column to check.
2) the chromatographic system issue: including the injector and other extra-column effects. maybe you need check and verify.
3) the detector issue: the different issue with different type detector such as ELSD,UV,MSD
good luck!

[Mary Carson]

If your sample loop is 100 uL, your maximum injection should really only be 50 uL.

Weighing vial might be ok, if you don't have evaporative effects, but you should check your autosampler manual. Agilent 1100s take an additional 3 uL (that go to waste) for each injection, if I remember correctly.

[Alfred88]

Dear slimshady:
We prefer to "verify" rather than "calibrate" the Injector (Autosampler).

First, you need to check if your sampling needle has less than 5000 injections on it. Else: replace it. Then there are three things to check (or "verify"):

1. Check for accuracy: Inject water ten times, 50uL each. Record the mass of vial before and after each injection. The difference between the two successive mass values is the amount withdrawn (i.e. injected, plus or minus a constant!). Calculate ave., std, and RSD.

2. Check for reproducibility (of pump & injector): You need a BPR to generate a pressure of about 1500psi. You can use Caffeine (140-200 ug/mL) as sample. MP is water, and FR =0.3mL/min. WL = 272nm. Injection volume: 15 uL. Run 10 times, and calculate RSD for Area and RT.

3. Check for linearity: Conditions are as in item 2. Inject 5uL, 10uL, 15uL, 25uL, and 35uL (or another combination, but the highest volume is 50uL). Plot area vs. volume injected, and calculate correlation coeff.

Note: Item 1 is not on our current SOP for PQ of HPLC, but I added because Rolandas mentioned about it.

Alfred.

[Bruce Hamilton]

If your sample loop is 100 uL, your maximum injection should really only be 50 uL.

Weighing vial might be ok, if you don't have evaporative effects, but you should check your autosampler manual. Agilent 1100s take an additional 3 uL (that go to waste) for each injection, if I remember correctly.

I would assume that if the sample loop was 100ul, a person would inject, and calibrate, up to that volume, or ask for a discount from the instrument manufacturer.

I'm not completely familiar with the Agilent 1100 sampling flow diagram, but, as there is no waste position, I assume that the vial is placed, and sampled, in the normal solvent flow path for both switching valve and metering device, ensuring that 100% of sample is taken and injected. I could be mistaken, so somebody should check..

Please keep having fun,

Bruce Hamilton

[HW Mueller]

It seems that Mary had laminar flow in mind when she suggested not to fill the loop? A probably more certain way to get around the laminar flow problem would be to overfill the loop, apparently done by Agilent (according to some contributions above).
Can one weigh delivered H2O ("injected" H2O) in these autosampler systems by using air as "mobile phase" (no column of course)?

[Peter Apps]

How to do an injection volume calibration will depend on what design the injector is - with the good old fashioed six port valve and loop you just remove the loop, weigh it, fill it with water and re-weigh.

But why do you need to calibrate ? if you inject the same volume of sample and standard (which would always be the case surely) the injection volume falls out of the calculation.

Peter

[Bruce Hamilton]

As the orginal poster was using multiple volumes, I assume he was using a variable-volume metering device type ( Agilent 1100 ), or a partially-filled loop type injector, but it probably doesn't matter....

Incidentally, a quick look at the flow path in the Agilent 1100 autoasampler manual suggests all the sample is injected.

i wonder how many analytical HPLCs are routinely injecting more than 20ul, so weighing will not be a good option. Multiple injections will only yield an average, not a good idea unless you know the repeatability.

I suppose the main reason for wanting to know that 5 ul is 5 ul would be for method transfer between instruments, so the best way could be just to inject caffeine solutions - as described in the Agilent injector IQ/OQ. They sell expensive test solution kits, which came be used to test linearity, precision, and probably injection volume, assuming a known extinction co-efficient in a transparent mobile phase.

I'm sure most other brands offer a similar IQ/OQ testing, but if they don't, the details of the Agilent method are proably on their web site. If not, I can probably hunt them out and post.

Bruce Hamilton

[tom jupille]

The most likely explanation for an increase in retention times and an increase in area is a flow rate deficit (usually a leak).

You all have raised a number of important issues that come up frequently when we do troubleshooting courses:

• 1. You should always prepare your calibration curve by dilution rather than by varying the injection volume.
  
  2. You should always inject the same volume for samples and calibrators.
  
  3. If you follow rules 1 & 2, then you really don't care exactly how much you are injecting, so long as you always inject the same volume (i.e., repeatability is more important than accuracy).
  
  5. Quantitation depends not only on the volume injected, but on the chromatography and detection
  
  6. A calibration curve for quantitative analysis should be re-run each day!
  
  7. It is a very good idea to include check samples periodically during a run to verify repeatability over that time period.
  
  8. In a "partial fill" injector (i.e., most autosamplers), repeatability is affected by the combined performance of the syringe, the loop, and the plumbing. Repeatability will deteriorate at two extremes:
  - at small percentages of the syringe volume (due to stepper motor resolution)
  - at large percentages of the loop volume (due to laminar flow effects)

[Alfred88]

Dear tom jupille:

Thank you for your great response (& tips). The question in the original posting has 2 parts: area values not consistent and how to calibrate the injector.

If I follow you (Tom) correctly, then in a PQ for the Autosampler, we only need to check reproducibility but not linearity (or anything else)? In an old FDA web page since 1993 (but still accessible, see: http://www.fda.gov/ora/science_ref/priv ... /jaoac.htm), the relevant question for the Autosampler is “Are injected volumes accurate and reproducible?â€

[prashant.pqa]

Hi dear
I think you can see the injector linearity in this case. If you will not get regression coefficient more than 0.999 you can contact to instrument supplier.

[Uwe Neue]

The original calibration is OK, with a correlation coefficient of .9996 with an intercept of under 5% of the lowest value.

The values on the second day are higher. If injected from the same vial, I suspect evaporation of the sample solvent from the pierced vial caps. Of course, this depends on the way the injections were done and the vial caps and maybe other things.

A change in retention should not affect the peak area, unless some really strange things are going on. For the small day-to-day changes in retention that one can expect, I do not think that it is reasonable to see a change in the peak shape.