Chromatography Forum

Here I am again with more questions.... We have an Agilent 1100 system with an RID detector. We are trying to build a solid calibration curve with good repeatability. We performed 5 injections of each calibration point but I am not getting good repeatability between each injection. I have attached some screenshots and am hoping someone can tell me if my baseline noise is expected or is way too high for what I am attempting. My samples are a 5% solution and the column/RID are running at 40C @ 1.4mL/min flow rate. 5ul injection (I have tried 10ul and 20ul injections and both produced some very ugly peaks). The column is a Waters Amino (NH2) Column, 125Å, 10 µm, 3.9 mm X 300mm with a 75% ACN/25% H2O mobile phase.

I really appreciate the help and direction... We are in this bad situation where the company purchased the instrument used for $5000 from a university and while it looks like everything is functioning/putting out good resolution, my manager and I do not have enough experience to know a few of the basic questions such as what to expect a baseline to look like on an RI detector? Prior to this position I held a position at another company that used a calcium column on a waters system and they functioned well enough to achieve results with a ~0.1% repeatability but with this instrument I am seeing swings of 2-5% between injections and I can't decide if we need to increase our sample concentration to get us away from the baseline or if the baseline itself looks bad/needs help. We unfortunately could not simply clone the system I am used to running because our outside customer produces mixes that have sugars that co-elute on a calcium column and this amino column was the only one I could find that would produce all of the peaks that they are wanting.

Link to some photos of my chromatograms/baseline noise...
https://imgur.com/a/DUK7wY5

Link to column
https://www.waters.com/nextgen/us/en/sh ... -1-pk.html

Link to the waters powerpoint that helped us make the decision on column/system settings
https://www.waters.com/webassets/cms/li ... foodjk.pdf

Thank you for providing many of the basic details in your post. Very helpful. Please, you really need to hire someone with practical HPLC experience as this is a very easy method (novice) to run, esp with just standards and no matrix. *RID use can be challenging as detection signal will be low and noise will be high. You must learn to optimize ALL settings and use a HIGH QUALITY HPLC pump running perfectly under stable conditions.
Here are some basic comments based on what you posted. Maybe some will help?
(1) Start by scaling your chromatograms to the standard peaks, not the injection peak.
(2) RID plus sugars on an amino column = WIDE peaks. That is normal.
(3) The "noise" you describe appear to be temperature cycling. You must setup the system to prevent this (or minimize it with RID). Make sure you stop flushing the RID reference cell just before you start an analysis (it should be 100% stable). Equil will probably take one hour (normal) IF your system and temps are also stable. Read: https://hplctips.blogspot.com/2018/02/h ... s-for.html
(4) USE continuous mobile phase degassing. RID is very unstable and you must start with a perfectly functioning and running pump PLUS degassed solvents at a stable temp to get reasonable results.
(5) TEMPERATURE stability includes insulating the lines, and maintaining room temperature at a stable temp too.
(6) Any HPLC system purchased for $5K is bound to be damaged. Invest in having it professionally evaluated and serviced. Using a system with unknown service history that has not been evaluated for condition and performance equals wasted time.
(7) Samples or standards should be dissolved in mobile phase.
(8) Get rid of the horrible Waters 10u NH2 column you have and just use a high quality AMINO in size 4.6 x 250 mm with 5u particles (e.g. Zorbax or Inertsil). Better peak shape, better resolution, better performance. KEEP IT SIMPLE.
(9) With sugars, the higher the concentration the greater the co-elution you will see by RID.
(10)Try a gradient of ACN/Water (15% ACN to start) for slightly better resolution. USE LOW concentrations to start with (that you can still detect), then run cal tables.

Thank you for providing many of the basic details in your post. Very helpful. Please, you really need to hire someone with practical HPLC experience as this is a very easy method (novice) to run, esp with just standards and no matrix. *RID use can be challenging as detection signal will be low and noise will be high. You must learn to optimize ALL settings and use a HIGH QUALITY HPLC pump running perfectly under stable conditions.
Here are some basic comments based on what you posted. Maybe some will help?
(1) Start by scaling your chromatograms to the standard peaks, not the injection peak.
(2) RID plus sugars on an amino column = WIDE peaks. That is normal.
(3) The "noise" you describe appear to be temperature cycling. You must setup the system to prevent this (or minimize it with RID). Make sure you stop flushing the RID reference cell just before you start an analysis (it should be 100% stable). Equil will probably take one hour (normal) IF your system and temps are also stable. Read: https://hplctips.blogspot.com/2018/02/h ... s-for.html
(4) USE continuous mobile phase degassing. RID is very unstable and you must start with a perfectly functioning and running pump PLUS degassed solvents at a stable temp to get reasonable results.
(5) TEMPERATURE stability includes insulating the lines, and maintaining room temperature at a stable temp too.
(6) Any HPLC system purchased for $5K is bound to be damaged. Invest in having it professionally evaluated and serviced. Using a system with unknown service history that has not been evaluated for condition and performance equals wasted time.
(7) Samples or standards should be dissolved in mobile phase.
(8) Get rid of the horrible Waters 10u NH2 column you have and just use a high quality AMINO in size 4.6 x 250 mm with 5u particles (e.g. Zorbax or Inertsil). Better peak shape, better resolution, better performance. KEEP IT SIMPLE.
(9) With sugars, the higher the concentration the greater the co-elution you will see by RID.
(10)Try a gradient of ACN/Water (15% ACN to start) for slightly better resolution. USE LOW concentrations to start with (that you can still detect), then run cal tables.

My manager and I both have ~5 years HPLC experience between us but both of us were not involved in method development whatsoever. They were very established/older systems that just ran. I really appreciate all the information. Our unfortunate situation is that this was essentially pressed on us and we have to work with what we have, including this column. The old Waters systems at my previous job seemed so unbelievably robust and had no issues even when the lab HVAC would go out the RI detector would just keep plugging away providing great results... it's why I'm so thrown for a loop with this system. The only reason we did such a high concentration of ACN was due to the Waters recommendation for that column... What reasoning would there be to have such a high concentration of ACN? At my previous job with the Ca+ columns, we ran straight heated water with no degasser whatsoever. They ran perfectly. I will definitely look into the temperature and the reference cell purging. I like to tackle one thing at a time. I think for starters I am going to work on the baseline.

Thank you so much!
I realize you are working with a lot of nonsensicals... why purchase a $5k instrument and expect a miracle. That portion of the decision and discussion had nothing to do with the two people that work here that have experience with HPLCs

5 years or less of full-time experience is considered "novice" in chromatography. Time alone is not the best measure, but a lot of hands-on time, with different systems, modes and sample types are needed. Method development, many more years. We regularly meet clients with 10-20 years of "experience", but in reality have the equivalent of just a few weeks of training because they worked in one or more labs where they were not provided any professional training, just told to "figure it out using what they have".

On a personal note, any company that can not afford to purchase a $300 consumable column (for better results) and/or can not afford to pay someone to professionally evaluate and test a used HPLC before use, scares me. I would not want to work in a place that does not support the employees in their work by either providing them with professional help and/or the needed services to insure success. Find an employer who invests in their employees and you will learn so many new skills (increasing your value to the employer).

When you change makes and models of HPLC systems (i.e. Waters to Agilent), everyone needs new training in how to use the new systems. Many things are different in how a Waters system operates (flow path, A/I etc) then would be found in any HP/Agilent 1050/1100/1200-series system. If you try to use an Agilent system like it was your Waters system, then many mistakes will be made. Odds are the pump in the 1100-series is far superior to what you had before, but it may be broken or damaged so may never offer good performance until it is evaluated. Paying a pro to spend a day going over your system may cost more than what you paid for the used system, but would save you tens of thousands of dollars in time and materials by letting you know what it needs so the remaining time could be spent on training to use it generate quality data.
You wrote: "The only reason we did such a high concentration of ACN was due to the Waters recommendation for that column... What reasoning would there be to have such a high concentration of ACN " - Sorry, no idea what you refer to. Who suggested higher conc of ACN?

You wrote: "The only reason we did such a high concentration of ACN was due to the Waters recommendation for that column... What reasoning would there be to have such a high concentration of ACN " - Sorry, no idea what you refer to. Who suggested higher conc of ACN?

I was referring to this powerpoint set up by Waters that has a very similar system to the one that we have set up at this point which uses a 75% ACN solution

https://www.waters.com/webassets/cms/li ... foodjk.pdf

Unfortunately the situation is that we are in a small community. I simply cannot select a new employer and am just hoping for some help which you folks have graciously provided. It may very well be that we need to level with them on the system but I need as much information as possible/go down every avenue that I possibly can before I approach them with the reality of the situation. Believe me when I say my own prior experience tells me how crazy it is to try to find success given such limited resources.

Hi Depizzymo,

This Waters application note is using a completely different column that you all are. This is arguably one of the most important comparisons and I would argue that this is not a similar system due to that fact alone. Ideally, successful and validated experiments can be performed across a wide range of LC's (at least ones that are capable of doing the work) as long as you have the same conditions and columns. But in this case, although it's a sugar analysis, you should focus on one that is specific to the type of amino column you are using. Depending on the column (and column chemistry specifically) your analytes will interact and retain differently due to this bonded/stationary phase. If you all are unfamiliar with method development, is your existing method based on these application notes? Even if so, as Multi has mentioned, there's likely a slew of issues with a $5k instrument and, like he said, should see an inspection. It can be lucky to get such a cheap LC, especially if it's in working condition, but with almost any used instrument it must be verified by professionals either before or after purchase to verify if it is accurate and in working condition. Like Multi said, this service will cost you probably around double what the LC costs, but, if you want publishable data and be confident in the work you produce, then getting it serviced/inspected is of utmost importance and will save you headaches in the future.

I'm sorry to hear that you all are struggling with this "new" instrument, but I am still hoping that it can be salvaged and you all can get this work done. I frequently travel to the local university in my city to use their LC-MSQTOF in their core facilities, but when I see graduate-student labs I am flabbergasted with the LC's and GC's that the graduate students have to use to generate data. Most of them exist as frankenstein instruments and if they can get them to work (which is a question in of itself), then I have hope for you and your boss.

Good luck!

I've posted this before, and this is what I did with the same equipment as you when I used 1100 quaternary system with RID (and also on our 1050 systems using 1047A).

1. I bypassed the multichannel mixing valve using the little Agilent adapter. IMPORTANT !

2. I set up the instrument the night before and had the mobile phase pump all night through the column and system, then back into the solvent reservoir. IMPORTANT !

I also did similar when using conductivity detector...

I don't agree that a $5K 1100 system is necessarily bad; companies have gone out of business, and some - like my former employer - have downsized support departments because they couldn't understand how they contribute to overall profits.

I've posted this before, and this is what I did with the same equipment as you when I used 1100 quaternary system with RID (and also on our 1050 systems using 1047A).

1. I bypassed the multichannel mixing valve using the little Agilent adapter. IMPORTANT !

2. I set up the instrument the night before and had the mobile phase pump all night through the column and system, then back into the solvent reservoir. IMPORTANT !

I also did similar when using conductivity detector...

I don't agree that a $5K 1100 system is necessarily bad; companies have gone out of business, and some - like my former employer - have downsized support departments because they couldn't understand how they contribute to overall profits.

I almost asked if these artifacts and issues might potentially be issues related to using a quat pump with the RID detector. Do you happen to know the item number for that bypass valve? We did not receive it with our pump.

some more inputs on your analysis/system

1) with sugars on amino column you're actually doing HILIC separation, which is kind of "reversed-reversed phase" or "aqueous NP" chromatography. So in this case, water is your strong eluent. That's why ACN is in that high concentration.
You may search for more literature and primers on HILIC separations.

Link to Waters: HILIC Method Development Strategy Wall Chart:
https://www.waters.com/waters/library.h ... d=10170486

With this in mind, adjust and exchange the samples solvent, as well as the injector wash solvents, to match as close or having less of water than the eluent. Of course, sometimes there are compromises to be made, e.g. because of the solubility of the sample and cleaning efficiency. But then, it's better to inject less of a more concentrated solution than more volume (you already observed peaks going worse with higher injection volumes).

2) RI only works in isocratic mode. But also do not mix the solvent online but pump 100% of premixed solvent on a single channel. Maybe this is already enough to not need the Agilent Adapter (but I'm not familiar with that; but 100% of a single channel is mandatory, even on Waters systems.)

3) add some back-pressure device on the outlet of the RI will also help with the baseline. But be careful to stay below the maximum pressure of the RI cell (which is often quite low, only about 10 bar or so; check the spec of your model).
You may use a short piece of narrow-bore tubing or a designated back-pressure regulator (bpr). You may use the tubing as a first try and if it helps, invest in a bpr.
If making your own tubing, include some safety margin for higher flow rates and higher viscosity solvents. e.g. if you normally run at 1.4 ml/min of ACN/H2O mixtures, nevertheless calculate in the way that even 2 ml/min of MeOH/H2O would not crack the RI-cell...

Link to backpressure calculator for tubings from Waters:
https://www.waters.com/webassets/other/ ... itters.htm
Link to optimizetech:
https://www.optimizetech.com/shop/unive ... egulators/

4) give enough time of equilibration and purging the RI reference cell before doing the injections. Having the RI running some hours or overnight may be a good advice

5) before blaming your system or having it serviced (which may still be a good option, because that's what is often lacking in universities...) you may carry out some basic performance checks on your own. Maybe Agilent has some SOPs on what they normally do (or someone from this forum can provide more specific tips, how they do it on their 1100er):
- check the injector's repeatability gravimetrically: place 6-10 vials, filled with water and tared, in the autosampler, do some 1 or multiples injections from each vial, weigh them back and compare if their masses differ a lot or are acceptable. Depending on the injector design, the mass loss will be exactly what you've programmed or it may be more if some needle overfill are withdrawn too.
But I would expect it to be always the same amount that will be withdrawn with each injection.

- replace the column by a piece of tubing (e.g. 1m x 0.1 mm, to have some backpressure on the pump so they work properly), run and equilibrate with 100% MeOH (or ACN, or water); then make multiple injections from of e.g. caffeine in the same solvent and compare the peak areas.
-- do some variation of the injection volume to check also injection linearity.
- if the repeatability is still not ok, then probably you need to service the system first, then cary on with your column and separation.

I almost asked if these artifacts and issues might potentially be issues related to using a quat pump with the RID detector. Do you happen to know the item number for that bypass valve? We did not receive it with our pump.

Agilent Part Number:0100-1847
https://www.agilent.com/store/en_US/Pro ... lsrc=aw.ds

Deppizzymo wrote:: "I almost asked if these artifacts and issues might potentially be issues related to using a quat pump with the RID detector" -
No, none at all. Using a quat pump presents no issues of any kind. When using a RID with ANY type of HPLC pump, from any manufacturer, use isocratic mode, pre-mixed mobile phase solution solution, run through a clean and serviced degasser. Direct the flow through ONE PUMP channel (on a Quat Pump, that could be 'A', 'B', 'C' or 'D'). It makes no difference which channel you use as long as the pump has been professionally serviced.

No need for any kind for any "by-pass valve" (who makes up these names ?). The adapter fitting (that is what it is called) mentioned is found on isocratic only versions of the pump which have no solvent selector installed (the quat has a solvent selector, the MCGV). No need to ever use any adapter. As noted above, run one channel of your four-channel (or two channel system if you have a binary pump) system to the pump head.

Why make things difficult? Keep it simple. Do not wast money on parts you do not need. Put the money into training and purchasing parts that you DO need.

run through a clean and serviced degasser

Sometimes, with ion-exclusion/ligand-exchange columns and RID, there may be a need to bypass (to avoid using) the in-line degasser in order to get rid of the negative peaks resulting from the air dissolved in the sample (but eliminated from the strongly degassed eluent) and overlapping with the positive peak(s) of the analyte(s).

Peaks from Dissolved Air in Sample Solvent
https://www.shimadzu.com/an/service-sup ... peaks.html

some more inputs on your analysis/system

1) with sugars on amino column you're actually doing HILIC separation, which is kind of "reversed-reversed phase" or "aqueous NP" chromatography. So in this case, water is your strong eluent. That's why ACN is in that high concentration.
You may search for more literature and primers on HILIC separations.

Link to Waters: HILIC Method Development Strategy Wall Chart:
https://www.waters.com/waters/library.h ... d=10170486

With this in mind, adjust and exchange the samples solvent, as well as the injector wash solvents, to match as close or having less of water than the eluent. Of course, sometimes there are compromises to be made, e.g. because of the solubility of the sample and cleaning efficiency. But then, it's better to inject less of a more concentrated solution than more volume (you already observed peaks going worse with higher injection volumes).

2) RI only works in isocratic mode. But also do not mix the solvent online but pump 100% of premixed solvent on a single channel. Maybe this is already enough to not need the Agilent Adapter (but I'm not familiar with that; but 100% of a single channel is mandatory, even on Waters systems.)

3) add some back-pressure device on the outlet of the RI will also help with the baseline. But be careful to stay below the maximum pressure of the RI cell (which is often quite low, only about 10 bar or so; check the spec of your model).
You may use a short piece of narrow-bore tubing or a designated back-pressure regulator (bpr). You may use the tubing as a first try and if it helps, invest in a bpr.
If making your own tubing, include some safety margin for higher flow rates and higher viscosity solvents. e.g. if you normally run at 1.4 ml/min of ACN/H2O mixtures, nevertheless calculate in the way that even 2 ml/min of MeOH/H2O would not crack the RI-cell...

Link to backpressure calculator for tubings from Waters:
https://www.waters.com/webassets/other/ ... itters.htm
Link to optimizetech:
https://www.optimizetech.com/shop/unive ... egulators/

4) give enough time of equilibration and purging the RI reference cell before doing the injections. Having the RI running some hours or overnight may be a good advice

5) before blaming your system or having it serviced (which may still be a good option, because that's what is often lacking in universities...) you may carry out some basic performance checks on your own. Maybe Agilent has some SOPs on what they normally do (or someone from this forum can provide more specific tips, how they do it on their 1100er):
- check the injector's repeatability gravimetrically: place 6-10 vials, filled with water and tared, in the autosampler, do some 1 or multiples injections from each vial, weigh them back and compare if their masses differ a lot or are acceptable. Depending on the injector design, the mass loss will be exactly what you've programmed or it may be more if some needle overfill are withdrawn too.
But I would expect it to be always the same amount that will be withdrawn with each injection.

- replace the column by a piece of tubing (e.g. 1m x 0.1 mm, to have some backpressure on the pump so they work properly), run and equilibrate with 100% MeOH (or ACN, or water); then make multiple injections from of e.g. caffeine in the same solvent and compare the peak areas.
-- do some variation of the injection volume to check also injection linearity.
- if the repeatability is still not ok, then probably you need to service the system first, then cary on with your column and separation.

This is beyond helpful, Thank you so much! I definitely had been using a single channel but I have got to figure out what is causing the ~500nRIU fluctuations. I will definitely replace that column with tubing to try this asap.

welcome.

And add a bit of backpressure to the RI-cell.
This helped me in the past and is mentioned and explained by John Dolan in the article posted by TylerSmith in your other thread about the equlibration time