Hi Nicoletta,

I have come across this problem in my analysis with phytohormones. There are three main things to consider in this. 

1. The range of concentrations should be carefully chosen so that we do not exceed the capacity of the column. Seems like you have taken this step already. With samples,  when concentration is unknown, dilution might also eliminate the overloading issues.

2. The second approach is use split injection and this will decrease the load of the internal standard and the sample and or standards. Split injections should be a mini-experiment for identifying the right amount and ensure there is no decrease in your detection limits.

3. The time for the whole cycle should be carefully planned so that there are no residues of previous run. This can be done by running a high concentration split and splitless and make sure you get most of it and the oven temperature is high enough to elute all of the residues. 

Good luck Nicolette. Hope this helps.

Thank you Aru for your answer... And nice to hear from you!

I have considered the "overload issue" and I have decreased the range of concentration used. Furthermore, I have already tried the split injection (trying different ratio combinations). But unfortunately it did not solve my problem... And I am sure there are no residues coming from the previous run because the T reached for the cleaning/post run is 250 degree Celsius and my samples are C14 acetate, so quite easy and volatile molecules... 

Hey Nicoletta,

Just have seen your question and I already had these problems with a GC/PCI-MS and an GC/EI-MS system. 

Sometimes it happens, if EI ionization is used, the amount of given electrons of the filament are not enough to ionize the whole amount of both, the internal and the native substance...so to say:

You reach a kind of "saturation" in the MS and less electrons for a complete ionization are available. This means in consequence, you will not see all the injected molecules under your peak, as they could not have been ionized and do not reach the mass specific detector...

Is the problem still given, if you dilute all the samples by a factor of 10?

Maybe this is another idea to check your problems?

Good luck from bavaria!

Dominik

Hi Dominik!

Thank you for your answer... I have actually tried to decrease the concentrations of my samples... the highest went from 4 millimolar to 2 millimolar. And I have also tried the split mode (1:50) for avoiding any possible overloading in the detection system (the machine has a FID)... but the problem is still there.

I will try to use a different column (INNOWax), even though the one that I am using is perfect for my samples (HP-5) and I want to check an intermediate concentration between the highest (2 millimolar) and the second highest (1 millimolar), if the trend is again observed.

Cheers from Sweden!

Nicoletta

When you say internal standard, do you mean a second compound, not the analyte, that you are adding to all the samples, and then you are plotting the ratio of the analyte/IS against known analyte concentration? From what you say I assume so. For any given sample the analyte/IS must remain constant for any volume of that mixture that you inject. If it isn't then the IS is not suitable. Have you tested to see if the IS response is linear when you inject different amounts?

Hello Nicoletta

Save yourself some time and trouble and don't change column chemistry/polarity if you are already using the optimal one. Instead and on top of the previous suggestions, consider using a column with a thicker film, a higher split ratio (1:200), and consider splitting your calibration into "low-mid" and "high" range, even if that means that you might have to reinject "high" samples.

This may sound like you've exceeded the linear range of your detection system. By trying to calibrate within the range of 15-2000 mM, you're jumping between two magnitudes. At these points of the calibration curve, you're hitting a point where detector response is no longer linear and the concentration ranges should be reconsidered. Even if you remove the highest concentrations one by one, the slope is still not linear and reflected in the regression. 

Attached are webpages from chromacademy and wikipedia discussing this concept.

http://www.chromacademy.com/lms/sco35/hplc_3_5_3.asp

https://en.wikipedia.org/wiki/Calibration_curve

http://www.sepscience.com/Techniques/GC/Articles/189-/GC-Solutions-20-Calibration-Curves--Part-2-Internal-Standard-Approach

Is it saturation of the FID? It doesn't seem so because it is the IS peak which is getting smaller. Perhaps the solution is to run the assay without an IS as that is the compound causing the problem.  Validate the assay without the IS responses included in the calculation.

Hi Robin,

Thank you for your answer! The volume and concentration of IS is always constant in all my samples (500 uM)... the only variation is observed for the highest concentration when the IS peak area seems dropping down while the sample peak area remains constant... this is was for the 4 milliM (when it was the highest conc) and then for the 2 milliM (when I reduced the conc)... As you suggested, I will try to run different concentrations of IS and different concentrations of samples. I think that maybe there is some interaction between the IS ( 12:OH) and the sample (14:OAc) that causes the problem... The graph in attached...

Cheers!

Nicoletta

Hey Patrick,

Thanks for all the links! Unfortunately my range of concentrations must be quite wide because I will do kinetic studies and I need to have such big window... and probably your observation is right and the detector is having trouble after the jump to the highest concentration. I will try to run an analysis with a single compound at the time (different concentrations of IS and samples) and checking if I see such problem for the highest one again!

Cheers!

Nicoletta

We run samples with very wide concentration ranges ( 90% to sub ppm), so we use 2 columns ,a 10 port valve, 2 columns  and 2 detector sets simultaneously. In complicated cases, we use 2 10 port valves, 4 columns and 6 detectors. A talk will discuss this at Pittcon in March  and the SME meeting in Feb.. Total cost for our system was under $25K in hardware!, We run one sample into 4 columns via 2 10 port valves. On 2 columns there are 2 detectors (mini pdd and TCD) and on the other 2 there are just TCD's. That gives us 6 channels of data in real time.

Hi Nicoletta,

When calibrating our GC/MS for environmental testing, some of the compounds are calibrated with a quadratic curve with weighting 1/x2 as allowed by the US EPA instead of a linear curve.  When we do this, the lowest calibration point will calculate to within 50% to 150% of its actual value.  The US EPA calls this the Upper and Lower Prediction Interval of Results (LPIR and UPIR). This is found in Method 524.3 which is attached. It also allows all standards to meet the acceptance criteria of each calibration standard, calculated as an unknown, must be within 30% of their true values.

Another thought, is when preparing our standards, methanol is used to dissolve the compounds to create water standards.  We purge and trap the compounds out of water and then desorb the trapped compounds to the GC/MS.  Methanol is also trapped and sent to the MS, except it is not "seen" when scanning from 35 to 260 m/z, it is still there.  Sometimes we have to balance the amount of methanol to be the same in all of the standards, as this methanol concentration difference may change the ionization.  This was very important when we were using Ion Trap type of MS instruments.  We had to create three different levels of standards, so we didn't exceed 50 uL of methanol added to create any of the standards.

Best regards,

Roger Bardsley

Applications Chemist

Teledyne Tekmar

BUT, if I have understood correctly, Nicoletta is using FID. Is that correct? If so, much of the discussion about ionization in MS isn't of much use. It is important to see if there is a linear resonse to each compound when they are injected individually. If the IS does not give a linear response then the IS is useless.

An FID has a very wide linear range, so detector overload is unlikely, and would have been resolved by running with split injections. You are putting approximately 100 ng of IS onto the column if you run splitless - this is ten times more than you need for a very good signal:noise ratio from an FID.

You do not give the concentration of the target analyte in your highest standard, or the ratio of concentrations between IS and analyte across the calibration. The IS should give a peak area approximately equal to the analyte at the centre point of the calibration.

I see two possible causes for the problem; an alchohol, with its active hydroxyl group is a poor choice of IS for an acetate ester - a propanoate or butanoate ester would be better, and your calibration points are very unevenly spread. For some reason I cannot enlarge the graph, but at the least you need a point in the big gap between the highest and second highest point, and fewer points cramped into the lower left corner. 

If you have not done so already you should make up a new calibration series from scratch to eliminate the possibility of operator error in making dilutions.

Hi Peter,

Thank you for your suggestions! As Robin said, I am using FID. I believe that the problem is related to the big gap between my highest and second highest concentrations... unfortunately I need to have this big window because my unknown can fall in these big range... I did serial dilutions in an exponential way (4mM, 2mM, 1mM, 0.5mM, 0.25 mM, 0.125 mM, 0.0625 mM, 0.03 mM, 0.015mM) but as you said the trick will be including some extra data points in the middle. Another option can be be including a blank rub between the highest and second highest to see if the detector has "memory" problem.

And, of course, I will consider using a different standard... the reason why we selected the alcohol is because, in the studied reaction, we want to quantify not just the acetate (substrate) but also the product which is an alcohol... but I guess the best iwill be switching to something else... I was thinking about a completely inert hydrocarbon chain (C15 or C17) which should not give me trouble...

Cheers!

Nicoletta

Hi Nicoletta,

Based on your published work on this site and this thread, I am going to make a few assumptions:-

Your sample is a tetradecylacetate, the internal standard is a dodecylacetate. You may be dissolving it in heptane solvent and you are using a 50:1 split injection with a GC-FID. The amount of C14 acetate ranges from 15-2,500 µM/L and your internal standard is at 500µM/L.

If we work out what should be happening here, the amount of C14 acetate that you are using for your calibration curve varies from 15  x (256/1,000,000) = 0.00384 g/L  to  2500 x (256/1,000,000) = 0.64 g/L or 3.84 ng/µL to 640 ng/L. Assuming a 1µL injection at a split ratio of 50:1 that would mean that somewhere between 0.077ng and 13ng goes on to the column. Your internal standard should be constant at 500 x (228/1,000,000) = 0.114g/L or 2.3ng on the column.

Assuming that your HP/DB5 column is 0.25mm id with a 0.25µm film, the optimum maximum concentration for a single peak of a low-middle polarity analyte is ~100ng, so you should be well within that for your maximum range. Assuming that your analyte peaks are

HeyTim,

Thank you!!! All your assumptions are quite right!!!

I am using heptane, but splitless injection (I have tried the 50:1 but no success), and the column is HP-5 with diameter/thickness that you mentioned...

1) I am not using polar solvents (i.e. methanol).

2) Peaks look fine, nice and sharp and good baseline.

For the other points, I assume that if I was experiencing some of the issues that you listed, I should have seen irregularity in all the other samples and not just for my highest concentrations...

But I will try for sure your suggestion analyzing the IS alone... I am actually running the IS in the range 0.06-4 mM doing 3 reps for each concentrations!

Cheers!

Nicoletta

Hi Nicoletta

This brings up a common problem with technical enquiries on this site and many others. It is very common for enquiries and pleas for help to include only the barest  information about samples, analytes, intrumental setups and methods. People who are trying to help then have to guess what the problem really is - and troubleshooting by guesswork is not usually efficient.

The more information you provide, the better the answers will be; to quote one of the experts on Chromforum "the more you don't tell us, the more we can't help you".

What is the repeatability of replicate injections you are doing now ?, deviations from linearity can be due  to random errors alone.

Hey Nicoletta,

If you use a splitless injection you will overload the column with the 2500µM standard (640ng on the column) that you are using. The 1000µM standard should will overload but might "just about" be OK as you are "only" putting 250ng on column. I really would not exceed 50-100ng on column - You say that the peaks are sharp but I suspect that the high concentration standards are "fronting" which means you won't be able to quantitate them properly - Also the IS at 114ng is on or above the upper sample amount limit of your column. I have attached a file with further information.

To 'fix' this, dilute all of your samples by 10 (1:9) with heptane; or if your autosampler is capable, tell it to inject 0.1µL (unlikely to work - unless it uses a

Hi Tim,

Thank you to get back to me!

According to your table, it should be an overloading because the actual highest amount that we injected was 508 ng (for the 4 mM). However, I have tried the split ratio (50:1) without no success... again I have observed for my highest concentration (4 mM) the almost disappearance of the IS peak and the same area peak for the sample, compare to the previous measurement (2 mM). But I guess it is a good procedure performing the split injection anyway for purging the inlet and avoiding any post-releasing residuals in the column... 

Cheers!

Nicoletta

Hi All,

Thanks for Peter Apps comments about limited information on these boards and Tim Strutt excellent comments on the analysis.  As an applications chemist for our instruments, we often need to ask more questions of the scientist to fully formulate a successful solution.

My sincere apology for the MS comments.  I made an assumption from the topics listed that there might be a MS involved.

Best regards,

Roger

Hi Peter,

thank you for your observation and I apologize if I was not clear with the info that I have provided... Regarding the question that you ask, my repeatability of the three reps is good and quite consistent with very small fluctuations. And for example, the measurements for the IS peak area are always consistent, between reps and between samples... the outsiders are just the three reps for the highest concentration.

Here the info:

- internal standard: 1-dodecanol at 500 uM (constant in all my preparations)

- sample: E11-14:OAc diluted in n-heptane in a range 15-4000 uM

- column HP-5 (0.25mm id, 0.25µm film)

- method - oven: 80°C hold for 5min, ramp: 10°C/min to 240°C, then hold for 5 min; inlet 240°C; splitless mode; volume injected 2µl;

I have tried the split mode 50:1, because I have taken into account the overloading problem, but I experienced again the same issue for the highest concentration...

Cheers

Nicoletta

Hi Nicoletta,

Sorry, I based by remarks on a 1µL splitless injection - As you are using 2µL you will need to dilute your sample by at least 20:1 :-) 

Cheers, Tim

Hi Tim,

We performed the dilutions. Samples were diluted 10 folds and I have run separately IS and acetate. Well, still the same problem... actually, chromatograms look very confusing because of the too low concentrations and it is hard to distinguish the peaks. And again for the highest concentration, the analyte disappears completely... And this is observed both for the IS alone (1-dodecanol) and for the sample (11-14:OAc)... at this point I don't know what to think!

Thanks again to get back to me!

Cheers, Nicoletta

Hi Nicoletta

'Regarding the question that you ask, my repeatability of the three reps is good and quite consistent with very small fluctuations. And for example, the measurements for the IS peak area are always consistent, between reps and between samples..'

How good is good ?, how consistent is consistent ? can you give actual values for relative standard deviation ?

There is nothing in your method that stands out as a likely cause for the problem, and the dilutions and split experiments rule out detector non-linearity. As a side note, column overloading spoils the shape of peaks but does not affect their area.

This leaves only the process of making up the standards and of adding the internal standards. Please tell us exactly how you do that. Are there any stock solutions that have been used for all these troubleshooting tests ? I notice that on the calibration graph the points for 500 and 100 uM are further below the line than the top point is above it - maybe it is these two points that are wrong, not the top point. If this is so it is almost certainly a problem with dilutions. This kind of kink in a calibration is very often due to using syringes of different capacities for different concentrations.

What you are trying to do is a pretty standard analysis. There is a lot of literature on it using a standard GC-FID with similar columns to yours. About 30 years ago I did some work on long chain alcohols with on older version of your column coupled to a magnetic sector mass spectrometer. After the method was developed on the GC-MS we were able to run it as a routine analysis on a GC-FID. The main difference from your method was that I used trimethylsilyl derivatives, but I am confident that your acetyl derivatization is fine.

OK, some ground rules:-

1) The amount of your internal standard injected on column **must not exceed 100ng**. If you do, the chromatography will be bad - Your peak will front, the geometry and retention time of your internal standard will change (dramatically) with only small variations in solvent composition, polarity and concentration. The 'perfect amount' will be between 10ng and 20ng. Please trust me on this :-) Going over 100ng will not give you any better results, honestly. If it does not work at 10-20ng your method or equipment is badly broken...

2) I am assuming that you will use a splitless injection - If you have a standard split/splitless injector **do not inject more than 1µL of sample** - You can use 2µL if you really, really, have to - But that will only be useful if you are doing trace analysis (you are not).

3) Use an initial column temperature 20C lower than the boiling point of your solvent (heptane b.p. 99C) so your initial temp of 80C should be OK. Ramp the column oven quickly to a temperature of 20C above the solvent boiling point - 30C/min should be fine. Then continue with your normal 10C/min ramp. I would go a higher than 240C (maybe 300C) to make sure than any high boilers elute, otherwise they will stay on the column and come out as random wide peaks in later runs.

4) Set your carrier gas flow velocity to 25cm/sec for helium, or 35cm/sec for hydrogen (1.2 or 1.8mL/min).

4) Do not complicate the method any more than you need to. Avoid Pulsed Electronic Pressure Control. You don't need to use it as you only are injecting 1µL of solvent. The heptane solvent will only produce ~210µL of solvent vapour inside your injection liner (A 2mm ID liner will hold ~250µL, a 4mmID liner will hold >1000µL).

So how do we find out what is going on? If we use simple logical steps it should isolate the problem:-

1) Inject 1µL of pure heptane solvent as a blank. Do you get large peaks coming out in the region of your analyte or IS? If you do they might be retained on the column as dirt or as carry-over from a previous injection.

2) Repeat the solvent blank injection. Have most of the peaks gone? If yes - It is now OK. If not consider cleaning the liner and bake the column off at 300C

3) Now make up a 100µM/L C12Acetate internal standard solution - This is equivalent to 23ng/µL. Run 3 replicates of your 23ng/µL internal. Yes I do mean only 1µL of just the C12 internal standard :-) **Do not add any C14 analytical standard**. Your internal standard defines the analyte quantitation retention times. On your 3 separate runs, your internal standard retention time should not vary by more than a couple of seconds and your peak areas should not vary by more than 15%.

If you can't meet these requirements you equipment or method is broken. Likely problems include: "The large peak that you **think** is your Internal standard isn't" (grease from your fingers. plasticiser, detergents, general lab contamination, dirt in the system and/or contamination in the carrier gas?). Or, (less likely): leaks in the injector; or your column pushed too far into the injector.

Assuming that everything is OK now, you can make up the equivalent of a 100µM/L C14Acetate standard (25.6ng/µL) containing 23ng/µL of your internal standard. Now run this sample 3 times. The retention time window for your internal standard should still be within a couple of seconds. Your integration software will then use this to accurately lock onto your analyte. Your replicates for your analyte calculated from the internal standard should be better than 3%.

Now that you have got this to work, the dynamic range of your system for C15Acetate should go from 0.1µM/L (0.026ng on column) up to the maximum capacity of the system for C14Acetate which will be a maximum of

In my previous answer; 

'I notice that on the calibration graph the points for 500 and 100 uM are further below the line than the top point is above it - maybe it is these two points that are wrong, not the top point.'

Should read

'I notice that on the calibration graph the points for 500 and 1000 uM are further below the line than the top point is above it - maybe it is these two points that are wrong, not the top point.'

I trained many Entomology Grad students (They like insects way more than chemistry). I found that Injections vs syringe cleanliness was a problem. A concentrated sample lives for a long time in your syringe, if only one syringe is being used for everything. Thus residue from more concentrated samples or even standards caused problems.  Damage to syringes (bent plunger wires) was a problem.  Keep analytes in the proper dynamic range as Tim states above.  Do you know how your detector measures symmetrical vs non-symmetrical peaks?  We used to have to measure them with a ruler.

 just a little question Nicoletta: Is the solvent composition 100% the same in the high standard as in the other ones?

slight differences might induce differences in boiling point and thus transfer fro split to column.

Hello, has this problem been solved? I can see the question goes back to February, but I have just seen it. If it has been solved, could you please give a quick account of what the problem was? I have experienced some problems with GC calibration curves as well. From the several replies I found some interesting suggestions, but perhaps you can narrow down which solved your specific problem. Thank you, Andrés