Chromatography Forum

For the He usage: Do you have the Gas Saver option selected in your method? Go to "File" "Edit Method" while it it loaded in Chemstation and click "OK" through the options until you see a window that should show injection parameters (e.g., Inlet Temp, Injection Mode (Split, splitless etc.), Inlet Pressure etc.). At the bottom of all this (layout may depend on your version of Chemstation) there will be a Check Box that says "Gas Saver" with parameters near it saying "Gas Saver Time" and "Flow Rate". I usually go down to 15 ml/min at about 2 min. You want Gas Saver to turn on after all your sample has had time to transfer from the injector (liner) to the column. This option just reduces the "Total Flow" of He during the rest of the analysis while maintaining a constant column flow rate or pressure (which ever option you are using).

Maybe you should let your advisor know the issues and concerns you have relating to the possible poor shape of the GC as well as requesting some formal GC training. A good research advisor will give you the tools you need to get good defensible data. In this case I would say they should send you to a GC maintenance class where you learn the basics of operating and maintaining a GC. It will cost money, but so does time wasted having a student try to operate an potentially poorly functioning instrument on their own with little guidance.

This is just my $0.02 as I have met some research advisors that don't really seem to have any involvement in what their student is doing. This can be very frustrating.

My research advisor sent me to a chromatography forum to take a course on HPLC troubleshooting as an undergrad researcher. It turned out to be money well spent.

If you don't mind me asking, what is your location? There may be some people in your area that you could contact for some guidance.

For the He usage: Do you have the Gas Saver option selected in your method?

I've noticed the gas saver option, but it was never turned on. I guess I will have to use it so I don't use up all the helium. Getting new gas cylinders is big hassle.

This is just my $0.02 as I have met some research advisors that don't really seem to have any involvement in what their student is doing. This can be very frustrating.

If you don't mind me asking, what is your location? There may be some people in your area that you could contact for some guidance.

This is actually my case. My research advisor got a new appointment at the university and is never available. Even before the appointment he was not much help. I had picked him based on the project he was running since it was the only one appealing to me in my department. I was under the impression that he had a proper lab setup ready which is actually for gas phase experiments rather than what I am doing which is liquid phase and off track from the main purpose of the project which is what got me attracted to it.

I first tried GC to analyze my solutions in beginning of the year but after sitting with the lab techs we failed to get any peaks. I then started trying UV-Vis spectroscopy and I had a huge issue with one machine where the lamps would shut off and after turning it back on, the readings were useless. I found another machine in a different department and got what I thought were good readings initially for my calibrations. But after analyzing my experiment samples the concentrations were not logical. The concentrations would increase much higher than the original concentration. I talked to my advisor about the issues with the GC and UV-Vis but he just told me to look around the lab and find something I could use. This when I realized I am on my own and had no choice but to try the GC one more time and I started looking up whatever I could find on GC and calibration methods. It's very difficult finding proper guides on analysis techniques even books were not much help, at least to me. I was lucky to find a lab procedure for students which helped me understand the idea of internal standard.

I am trying to calculate statistics for my method: method detection limit, limit of detection, limit of quantification. I noticed that a lot of the calculations are based on the EPA MDL procedure. The calculations are applied on the peak area of the analyte which they are using in their calibration. In my case I am using the ratio of peak area of my analyte to peak area of internal standard. So can I still apply the EPA procedure for my calculations?

I am trying to calculate statistics for my method: method detection limit, limit of detection, limit of quantification. I noticed that a lot of the calculations are based on the EPA MDL procedure. The calculations are applied on the peak area of the analyte which they are using in their calibration. In my case I am using the ratio of peak area of my analyte to peak area of internal standard. So can I still apply the EPA procedure for my calculations?

We usually applied it to calculated concentration. Make sure and read the MDL procedure carefully as it is not so straight forward. You have to "find" a spike amount that is detectable after extraction but not too high (I forget the actual calculation for "too high" signal). Do at least 7 reps of this spike level and then calculate your MDL from these. I believe it's the Student T-value at your n-reps (7) × the SD in concentration for the 7 reps.

Obviously if you use a spike amount that is too high, your SD will be small and will result in a falsely low MDL. Be careful of this.

The EPA method is for running water samples but can this method be applied to different systems? I am using methanol as solvent with benzene, toluene and xylene as analytes.

In the first step of the EPA MDL procedure I am supposed to estimate the MDL using one of four ways.

The first way is concentration value which has a signal to noise ratio between 2.5-5. I have been using a ratio of area of analyte to area of internal standard (y-axis) versus concentration of analyte (x-axis) to get a linear calibration curve. The simplest to calculate the signal to noise ratio is the average divided by the standard deviation. My question is should I use the average and standard deviation of my ratio of peak areas to calculate S/N ?

The second way to estimate the MDL is : "The concentration equivalent of three times the standard deviation of replicate instrumental measurements of the analyte in reagent water". So I run several injections of analyte free reagent and calculate the standard deviation of the peak areas of the reagent?? How many injections are usually run?

fangs,

In chromatography it can be hard to use the second method as there is usually "no signal" to integrate in the blank (reagent) runs. I would use the S/N method and use Chemstation to calculate the S/N using the RMS method for the Area. Keep running dilute standards until you are getting a calculated RMS S/N around 5 (I think you can actually go a little higher with this but you might want to check the method for sure).

Use this concentration to spike into a set of 7 (or more) blanks and then measure the replicates. You should get about the same S/N on these.

Calculate the concentration of each of these 7 reps using a calibration curve that includes this concentration level in it. You may need to adjust the curve type to produce a good calculation at this low level. I used to use linear w/ inverse concentration weighting as I was always running very low concentration samples but I was calibrated over 3 orders of magnitude in concentration.

Use the student T-value at n-reps × the SD in Conc. of the reps to get your MDL.

I think there is also a rule in the MDL calc where the resulting MDL must be > than 10% of the spike concentration. I'm not remembering exactly how this rule was but I think this was similar. If your MDL is lower than this than you need to use a lower spike level for your MDL calcs. Unfortunately you have to redo the whole procedure using a lower level.

The plus side is that once you find a good spike level for doing the MDLs you can usually use this level each year when MDLs are required to be rechecked. It usually works out each year to reuse the spike levels from before.

We usually used our Reporting Limit as a spike level and the calculated MDL usually came to be about 4 times lower than the RL. This works per the MDL method and the RL level was already in our calibration curve.

Hope this helps a little more. Let me know if you have more questions. You can email me if you like at ty (dot) lahren (at) gmail (dot) com

Shouldn't the amount of benzene in the injections be the same as in my solution? So if my solution is 50ppm = 50mg/L then I have 50mg in my injection, correct? So it is much higher than 0.1ng.

Since you are injecting the solution, the concentration of benzene in the injection has to be the same as the concentration in the solution - but this is not what directly determines the size of the peak. The mass of benzene in the 0.2 ul that you inject, not the mass of benzene in 1l of solution is what you need to calculate.

Peter

Hello. I'm posting here because I thought it would be more on point since it is regarding the same topic rather than posting a new thread.

I've been using the GC using methanol as my solvent but seeing as I have quite a large number of experiments I would be using up a lot of the methanol and I don't think the current stocks will be enough. Purchasing new bottles is a hassle and can delay my experiments. So I was thinking of using distilled water as my solvent rather than methanol.

My concern is whether using distilled water in GC injections is fine and it won't damage the equipment. I've never used distilled water in gc injections and even in my undergraduate experience all the injections consisted of organic solutions only.

No, you should not use ditilled water as solvent. It will cause all sorts of problems with injection repeatability, and probably damage the column. Also, you will have to throw out all the results using methanol and start again with the method development.

Peter

No, you should not use ditilled water as solvent. It will cause all sorts of problems with injection repeatability, and probably damage the column. Also, you will have to throw out all the results using methanol and start again with the method development.

Peter

By method development, do you mean I should make new batches of solutions with the new methanol and do the calibrations again?

No, you should not use ditilled water as solvent. It will cause all sorts of problems with injection repeatability, and probably damage the column. Also, you will have to throw out all the results using methanol and start again with the method development.

Peter

By method development, do you mean I should make new batches of solutions with the new methanol and do the calibrations again?

I thought that you were going to use water as solvent because you could not get new methanol. Just to be clear - if you can get more methanol so that you can carry on using methanol as your solvent, as you do at present, then you do not need to redevelop the method, but you may need to do the calibration again depending on how stable your exisintg method is. If you switch to water as solvent you will need to redeveop the method, especially with regard to inlet conditions, and you will have to re-do the calibrations, and then probably update them more frequently than you would need to with methanol as solvent becuase instrument performance is very unlikely to be stable with water as solvent

Peter

Thank you for the reply. Since there is chance of damaging the column I will continue using methanol.

I have a maybe silly question regarding calibration. What I've done is preparing standard solutions of different concentrations by diluting a stock solution of higher concentration. And using the standard solutions along with an internal standard for calibrations (internal standard is only added when analyzing and not added to the experiment).
Is the calibration curve only valid for the solutions prepared? I mean, if I use up the standard solutions or even if I use up the internal standard solution or if I prepare a new standard solution is the calibration curve still usable?

Since you are a student I will pose the answer in the form of a question: The calibration plots relative detector response against quantity (mass or concentration for example) of analyte. If the calibration is linear, then relative response per unit quantity is constant over the range of the calibration. By relative response I mean the response to the analyte divided by the response to theinternal standard.

Do you expect the relative detector response per unit quantity to change if you make up new solutions ?

Peter

Since you are a student I will pose the answer in the form of a question: The calibration plots relative detector response against quantity (mass or concentration for example) of analyte. If the calibration is linear, then relative response per unit quantity is constant over the range of the calibration. By relative response I mean the response to the analyte divided by the response to theinternal standard.

Do you expect the relative detector response per unit quantity to change if you make up new solutions ?

Peter

If I were to assume that the solutions are perfectly the same AND the GC is good, would I expect them to change? No. But since I've been seeing variability in my previous injections when injecting the same standard 3 times I suppose this means I have to use the same solution for calibration and experimentation.

Since you are a student I will pose the answer in the form of a question: The calibration plots relative detector response against quantity (mass or concentration for example) of analyte. If the calibration is linear, then relative response per unit quantity is constant over the range of the calibration. By relative response I mean the response to the analyte divided by the response to theinternal standard.

Do you expect the relative detector response per unit quantity to change if you make up new solutions ?

Peter

If I were to assume that the solutions are perfectly the same AND the GC is good, would I expect them to change? No. But since I've been seeing variability in my previous injections when injecting the same standard 3 times I suppose this means I have to use the same solution for calibration and experimentation.

You are confusing the poosr repeatability of your method with a change in detector response. Let me pose another question - will the repeatability change if you make up a different set of solutions ?

Peter