Liquid Chromatography - Science method

Chiranjeev Sharma If you just need a secondary detection before the mass spec detector, I would think about derivatizing the sample. If amino sugars is all there is (and other other small amines), think about derivatization with FDNB or by the AccQ-Tag platform from Waters. Both ways result in added chromophore to the amine with ability to analyze by UV or DAD, or fluorescence detection, either detection able to go down to ng/mL. High resolution is achieved on standard RP columns with TFA or FA/ACN mobile phase. Check the attach literature below and think if you need to alter any of the acids/salts to make it more LC-MS friendly.

If simple sugars are also present, consider running orthogonal (additional) method by PAD (ion chromatography) detection for the simple sugars. Alternatively, if you are open to outsourcing your samples, reach out to a contract testing lab or university with capabilities to run ion-chromatography with in-line mass spec detection. That will be able to analyze both amino acids, amino sugars, and free sugars together.

FDNB Derivatization:

AccQ-Tag Platform:

Ion-Chromatography with mass spec:

Methods for analyzing trihalomethanes (THMs) in water, besides gas chromatography and liquid chromatography, include ion chromatography, mass spectrometry, solid phase microextraction, enzyme-linked immunosorbent assay (ELISA), UV-visible spectroscopy, and colorimetric methods.

Hello Mariano,

It is pretty common to have carry-over analyzing plasma/serum samples. Someone recommended me the use of Trifluoroethanol to wash out all retained peptides. Basically you just use it as a sample and inject 1 or 2ul direct onto the column with a relative short gradient. It does not damage the column in my experience, so maybe you can give it a try in the future.

Good luck!

Check for the solubility of diselenide and monoselenide. Monoselenides are usually soluble in methanol or ethanol and diselenides are soluble in chloroform. Recrystallize in suitable solvent.

Have you tried mass spectrometry with electrospray ionization, possibly in combination with ion mobility like TIMS or TWIMS?

Check the drive unit.

There are several causes for the presence of outlyers. Impurities, errors during the extraction, the needle that for some reason did not inject the same amount of sample, and samples that are naturally different from the other ones.

You can solve the problem using this checklist.

- check the look of the chromatograms and verify that all of them look more or less the same.

- check IS intensities and verify that their peak area is within an acceptable range for all the samples.

- normalize the data across samples.

- scale the data using UV scaling.

- evaluate the presence of outlyers using RobustPCA.

good luck

This confusing question is now 8 years old, but in case someone finds this old question here are a few comments that may help them.

hi Dharmendra Kumar , my HPLC does not have temperature control. I have a dedicated room with air condition (door and windows are shut), so I assume fluctuation in temperature could be low

yes bubbles can affect the reproducibility of the peak because when there are bubbles in the syringe, the volume of mobile phase that is aspirated from the syringe (which is used to inject and aspirate the sample) is not correct due to the bubbles of air so the volume of the sample is corrupted.

To monitor the carry over try to run a sample and then do another one by injecting only the mobile phase: you should not see peak or see a very small one

Hi Tamjid, according to my understading, these limits of detection and quantification should be determined for every method, as every method will have different conditions (the column, the analytes to analyze, flow, mobile phase, injection volume, etc). These limits are determined through the validation of the analytical method. After method validation. you can continue and quantify your samples in a reliable way. As an example of method development and validation, you can see this publication. I hope this helps

Dear Davy,

If contaminations or sample degradation are excluded, shouldered peaks are the sign of an LC issue. Your peptide is probably retained somewhere in the LC, which explained the slight RT shift inducing the second peak.

As you’ve changed the column, it’s likely not a column problem. Therefore, I suggest you to :

Points 1-3 will help understand the problem origin.

Hope this will help! I'm looking forward to see what was the trouble!

Best,

Jonathan Maurer

Kindly check..

I am detecting the polysaccharides by resorcinol-sulfuric acid-assay.

In my opinion try to use Methanol instead of Acetonitrile.....

Dear Christina Oettmeier although I'm absolutely no specialist in this field, I just came across the following potentially useful literature reference which might help you in your analysis:

Fortunately this paper is freely available as public full text on RG. Also please have a look at the attached article in which the authors use a concentration of 0.01% BHT:

I hope this helps. Good luck with your research and please stay safe and healthy!

I used MassLynx about 4.5 years and that ended up two years ago, so bare with me here.

I guess that you have an LC/MSMS system, am I right? When you double click the MassLynx icon, the first window that opens the main window as shown in the first attachment.

Then, you must open two sub-windows within MassLynx to be able to control the instrument. They are MS Tune and MS Console.

From MS Tune window, you can turn On the MS detector, and from the MS Console you can turn On the LC flow and set your A and B percentages of flow as well as the flow rate.

You are saying that the MS Tune option can't be found in the Instrument tab. To me, this means that the communication between the LC and the MS is lost and you need to restart the communications.

First, there is a reset button on the MS that you can press through a hole using a paper clip or another narrow and long object. Before resetting the MS, exit all MassLynx applications and restart the computer. See if this solves the communications problem and brings back the MS Tune back to the Instrument tab.

One more thing you can do to check if the MS is online or not, is to access some MS functions in the MS Console window. Try to run IntelliStart for example. You can turn ON/OFF the nitrogen gas to the MS from this window. You can also fill up the infusion syringe with some liquid through the lines A and B. If these commands from the MS Console IntelliStart can control the MS, then your MS is online and communications aren't lost. In this case, try restarting MassLynx to see if this helps. If not, try uninstalling MassLynx and reinstalling it again.

Remember, always call the technical support line before doing something that you think it is going to ruin your instrument. Never take your chances with analytical instruments, they are so delicate and moody.

Hi Kamil,

You may make an inquiry at Alfa Chemistry, they offer kinds of good-quality chemicals.

Hi there,

Neutralizing the pH as quick as possible limits the denaturating effect of low pH on proteins. It's not about denaturation/renaturation processes, it's rather about preventing denaturation.

Wolfgang Schechinger Thank you for your answer.

Yes, I for the protein nice layer but don't know what happened the after steps because I couldn't see any proteins bands in the gel.

Of course, searching the " detection of PLA2 in stool" was my 1st thing.

As there are many HPLC methods for AA, it is rather important that you share with us which of the many procedures and methods you will use. For the very commonly used OPA/FMOC procedure, many choices for ISTD exist. Examples would include: sarcosine and/or norvaline.

Water probably absorbs more than solvent if you use a water-to-organic solvent gradient.

No, we do not rely on MS signal alone for concentration. So called "abundance" is related to how well the various samples ionized under the specific HPLC method AND the MS parameters you selected (VERY SUBJECTIVE). Change the settings and you get different results. Some samples exist in high concentrations, but ionize poorly. Some in low concentration, but ionize strongly. Some compounds exist, are never seen. Hundreds of variation in signal detection lie between as this is a complex topic, esp for a web forum. 'Areas' or 'ion abundance' counts only have meaning when properly developed methods are used with formal standards. After that, complete calibration tables are created for quantitation (best done using a non-MS detector as MS is poor at quant). Isotopically labeled standards are best, but true "standards" in some form would be needed.

there is a signal to noise ratio which can be calculated by masshunter automatically. You can calculate your LOD and LOQ based on that.

You have not done your homework to understand the basic fundamentals of how an HPLC or LC-MS system operate yet (or maybe have been watching too many of those TV shows or Movies where the person inserts a sample into the system and out comes the structure and name on screen!). There is no similar library ability with HPLC-MS as their is with electron impact detectors (hard ionization) used in GC-MS. When a proper GC method is coupled to a GC based MS (EI) detector, then one or more applicable libraries may be used, along with actual standards, to make qualitative assignments. For HPLC, there are far too many variables involved in the methods themselves (which determine selectivity for the compounds), the detectors themselves and the settings used (esp. for Mass detectors which have many critical setting that must be optimized, for each sample) which can create, change or make data disappear. And, as each instrument is different, similar methods may generate different data too.

Principal Component Analysis (PCA) is normally able to gather similar data and discriminate between different sets. We successfully did it with coffee aroma analysis using GC-MS.

ACN is a poor choice as a 'wash solvent' for most methods. You need to use a wash solution (and perhaps a mobile phase too) that will fully (fully = 100%) dissolve the types of compounds you are running through this instrument. *This is a basic fundamental of training in the use of the system.

As you stated, FIND the cause of the clog first. Maybe you are overloading the system with sample? Maybe it is not fully soluble in the solution or when it mixes with mobile phase, it falls out of solution. This is clearly a training issue. In the meantime, replace the clogged tubing with fresh new tubing (tubing is cheap vs the time and solvents spent trying to clean it). Dissolve all samples in an appropriate mobile phase (one that they dissolve it). FILTER all samples with a chemically compatible filter before injecting. Insure that your HPLC method and mobile phase choices are appropriate for the samples (do they dissolve fully in the mobile phase?). Make sure you include a separate column Wash "step" after each run (this should be programmed as a 'Wash method" to be run after your analysis method). The "Column Wash" should utilize solvent mixtures that are fully compatible with your column and detector AND which are stronger than your mobile phase so they wash any potential late eluters or matrix off the column and out of the HPLC system and detector BEFORE the next analysis. Again, ACN is a very poor choice. Select solvents which will dissolve your materials and use a gradient during the wash.

How did you delete the data? Did you delete it from within Empower, or delete the files from the drive?

If you deleted it from the local drive:

Some of the HPLC software I know of uses some kind of relational database to keep track of the data, so deleting it from within Empower may have deleted the files and the way to access them via their user interface. I don't have many suggestions in this case as they tend to lock those databases down to maintain data integrity.

You have answered your own question. Proper chromatography methods require detection methods which are applicable to the samples under analysis. You may want to review the types of instruments, detectors, columns and materials available at your school before deciding on a final approach. Have someone in the school's analytical instrument lab assist you.

For HPLC samples with weak or no chromaphore, you will need to use a detector that does not rely on wavelength based light absorbance such as an RID, ELSD, CAD or MS, to name a few. Put a little time into researching this basic question so you will better understand the uses and limitations of these methods and detectors. I often suggest my students start with a keyword search on the web (GOOGLE or BING) to find examples and articles to review. This is one of the best ways to learn. Once you have familiarized yourself with the basics, be sure to ask your teacher for help in setting up and running some example methods.

BTW: You may also be able to derivatize some of your samples and use GC/FID or GC/MSD as an alternative to HPLC (only if the samples are appropriate).

Mr. Asali,

LC-MS can be used to quantify highly reliably analytees in mixture without internal standard. Please consider, the following discussion, and the references therein, together with the references to the corresponding literature mentioned. This can be carried out by means of our innovative quantitative model formulast developed more recently. The method improves significantly the method performances of MS up to r = 0.99995 studying steroids in mixture, for example.

Noel W Davies Thank you again. This makes so much more sense now.

Dear Alik Kandhita Febriani,

I think you could recrystallized from ethanol or water. Also, maybe you could perform a sublimation for your coumarin, you will get a very pure crystal.

Hi,

Did you finally manage to solve this question? I am struggling with the same problem.

Hi,

You may make an inquiry at Alfa Chemistry, they offer kinds of chemicals for research use.

Isam, that is a sales brochure, not a manual.

Let us try and help people by showing them how to solve problems (a process of learning) rather than trying to do their work for them and give them answers.

Please look at the following below links which may help you for the two tehniques:

Thanks

The detector response is based on concentration C=m/V if you reduce de volume the relative concentrtion increases and thus sensitivity.

And you still have not provided any information about the method or answered the most basic questions...

Yes, some mixing is done whenever two lines intersect, as in the proportioning module (which has four valves and acts as a gate, allowing different proportions in at a time or inside the pump head), but most HPLC systems have an addition mixer. *For the Accela, they have in internal inlet mixer plus the pulse dampener also acts as a mixer too. Optionally (or std on some), a dynamic mixer can be installed too. However, as noted earlier increasing retention times are NOT an indication of a mixing problem.

Using only two of the four inlet lines can indeed result in problems. The two unused channels dry up inside and the valves can leak (resulting in cross contamination). Do not plug the unused channels. Instead, double up on each solvent line and place each of the unused lines into the complementary MP bottles (same as you use). Flush each one with MP to purge them fully. Then go back to using your two lines. Periodically re-flush each of those lines (~ 2 days) to maintain the degassed liquid inside them.

If you would be so kind to share your actual running conditions, column dimensions (basic HPLC info) and also provide numbers (Rt's) of some of the significant peaks for say, 6 runs in a row, then we could provide some free insight.

A typical effect of to high elution power of the injection solvent. Try a lower MeOH ratio - e.g. by diluting your sample until your analytes are bound to the column. good luck

The profile it similar to coluomns full of air for which you obtain strong oscillation when bubbles come out or in case of high pressure coloumns with out the flow restriction after the coloumn.

Which supedex 200 is it? 10/30? 16/10? the just to pass a lot of water at the flow reccomended on the mamual for at least 5-10 CV and see if it return normal.

Just as a practical suggestion. you can routinelly wash your coloumn just by injecting some ml of naoh concentrated and i think is it better wash out it with water than buffers.

good luck

Manuele

Best advice I can provide (having hired and trained many in the field over the past decades): Get a job in industry to learn on the jobs (hands-on training) using a variety of different brands, models and types of modes. This is the only true way to gain practical experience with real-world compounds. Courses are great for learning the fundamentals and concepts (something that few spend the needed time to learn), but you will never gain any practical experience working at school, reading books or just taking more classes.

First, "Peak Purity Determination By HPLC" (specifically by use of a diode array detector) is a null test at best. It can only show possible impurities which may be present, but can NOT show true purity. As used by most HPLC instrument vendors, it is subject to many limitation and assumptions (such as all possible compounds absorb light with similar extinction coefficients, which is of course not realistic).

I teach the use of this feature on several software platforms and perhaps the two most important aspects of using it are: (1) That you have a great deal of experience and training in HPLC analysis and operation of the HPLC system and (2) that the analysis method used has been fully optimized and reviewed that is in compliance with following good chromatography fundamentals (e.g. Proper K prime's for peaks). This software feature is an advanced feature and can be easily manipulated to show any purity value you want. It requires formal training to use. Good documentation of the method settings and proof the method is selective for the given compound is required before use.

For more information on this topic, here is a link to a free, short article I wrote on the topic from my classes. Perhaps it will assist you with your questions.

See attached:

Hi. Yes, you can use a peristaltic pump, as long as you respect the flow rate suggested by the manufacturer. Calibrate your pump before connecting it to the column and you should be fine. Watch for bubbles.

I think that is difficult to say and will require some trial and error. It partially depends on your method, both from an LC perspective (what kind of gradient are you running over what time frame at which flow rate?), as well as from an MS (acquisition) perspective.

For what it's worth, a rough guess would be in the few micrograms range.

Disappear? Not likely.

This question comes up frequently and progress has been slow. The technique of HPLC is mature (~ 53 years old) with few major advances in the past years. So called "UHPLC" is nothing more than HPLC, by a different marketing name. It is not new. We first starting using small particles packed in narrow diameter commercial HPLC columns coupled to low delay volume HPLC systems in the early to mid 1980's (e.g. Back then, I used one of the early HP 1090M HPLC systems with a DAD and DR5 pumps with great success that many modern HPLC systems have only recently equaled and surpassed). Low, solvent saving flow rates were used with super high-efficiency columns (1.0, 2.1 and 3.0 mm ID columns with 3u particles) to produce some great separations, but the market was just not ready for it. It was also in the 1980's that many of the HPLC systems became more reliable and added fully computerized automation for data acquisition and data analysis which were huge improvements over what we had before (strip chart records and cutting peaks out with scissors is not missed by me). In the 1990's, many said CE and CZE would replace all or most HPLC methods! That was nonsense. A decade later in the 2000's, SFC was advertised to eclipse HPLC as the most widely used technique. Another misleading prediction. None of these came true and SFC and CE are still niche techniques which are not likely to replace HPLC at all, but instead find their own niche applications (*Variants of CE found a great application area in DNA analysis).

Perhaps the most significant advance seen over the past decade has been in increased column peak capacity using columns packed with smaller particles, though this was developed and marketed back in the late 1980's (and few scientists were interested back then. It took a re-boot of the idea many decades later, when better instrument reliability was achieved and esp higher quality column packing techniques were developed to improve reproducibility, a challenge we still face).

In my view, as a professional scientific consultant who teaches scientists to use the latest techniques and methods, detectors and detection techniques for chromatography have not significantly changed. So much misleading detector sales literature and marketing would have us believe much has changed in the last 5 years, but really, little has. Detectors advertised with super high-speed data collection rates often far exceed the calculated need to collect accurate data by 10, 100 or 1000 fold. No recent order-of-magnitude changes (though the sales bias product literature from the instrument manufacturer's often claims each new system they introduce is orders of magnitude more sensitive than the one it replaced, LOL!). Sensitivity for some detection methods has certainly improved, but many of the improvements have been from other aspects of the technique or method (column choice, flow stability, better sample prep), not just the detector (e.g. longer path length flow cells provide significant signal improvement, but not 10x). Bench-top MS systems coupled to DAD units took time to perfect and now they are more commonly used. This is still an area where we will see improvements in both the technology and also in cost reduction and shrinking of size.

3D printing of columns and even instruments, "Lab on a Chip" solutions for specific application areas and the miniaturization of instruments are all areas where I think we can expect to see change in the coming years. Improvements have already been made and given time, hold promise.

Dear Pszczolkowski,

I think attached papers are useful for your request.

Best wishes, Homayoon.

I see, Thanks Ludwig Gruber .

As the question is about the most "accurate" way of preparing the eluents, the only acceptable answer is to weigh all the constituents of your mobile phase. However, even with weighing in all the components make sure, volatile solvents are degassed before the final composition. For the preparation of an eluent, think about how to reliably and accurately prepare any liquid mixture. In other words: Do what we learned during the first practical lessons of our lab education.

Please allow me to talk the obvious: In gradient LC it is a very practical hint never to mix 100% pure solvents. Always have a small portion of the second eluent (in case of a binary gradient) present. It avoids any unwanted salt precipitations on the column or any significant volumetric contractions (or expansions) due to the mixing.

Now, did I always use the weighing method? Nope. I only weighed in the constituents, for ultimate performance. For most of our daily work, the faster volumetric approach is more than acceptable.

What made me stumble - several times in the past - was the imprecise description of eluent preparation in articles, posters, etc. I wished that people would accurately write what they did. For example instead of "5% (v/v)", one could write "add 50 mL of solvent A to 950 mL of solvent B". A couple of letters more, and everyone can reproduce the recipe...

My two cents.

Depends on the target compounds

Honey adulteration is a relatively hot area in food chemistry. As you point out, there are multiple published methods using NMR. I have done some work using HPLC-MS/MS, which I find to be vastly superior to the NMR methods (but slower, obviously) both in adulterated honey as well as adulterated aloe vera products, mostly concentrating on mannose. I can definitely see the advantage of looking for the rare sugars as adulterant markers (lower concentrations indicates adulteration, for example).

Solid phase extraction

You clean and preconcentrate

It is the best way to clean dirty samples

Check the AOAC. They had HPLC methods.

Two features of Hypercarb must be considered when using it for applications involving carbohydrates and anionic analytes. First, Hypercarb contains weak anion exchange sites on its surface. This aspect is primarily responsible for Hypercarb carbohydrate retention and selectivity. Second, these anion exchange sites are sensitive to oxidizing and reducing agents. The symptoms suggest the latter may be your problem. Try treating your column with ascorbic acid to restore the native oxidation state of Hypercarb.

The retention time is very low, I would suggest to lower the methanol content to 30-40 % in the mobile phase as well as to lower the acetic acid concentration to 0,1% and compare the results.

regards,

G

I thank you should used buffer solution less than 4( 0.02 MKH2PO4) with 1% trethylamine , such as mobile phase Methanol: buffer: triethylamine :( 55:44:1) increase the retention time of metformin .

First of all, you have to equilibrate for 15 min for every run.

Second, do you use gradient or isocratic system. In gradient system fluctuation of pressure is normal.

If you don't use serum-free media as Alexey indicates, serum proteins and serum-derived exosomes overwhelm the amount of protein coming from your cells. Rather than using acetone precipitation, we concentrated protein by filter centrifugation using low MWCO Amicon concentrators. Using either method, you should calculate protein concentration in the media prior to concentrating, and in the case of filter centrifugation, after concentrating as well. If you're comparing across conditions or samples, you should compare based on the protein from equivalent initial volume of media, not equivalent amount of protein when running MS analysis.

Keep in mind that the amount of protein you obtain will be extremely low. In our studies, 48 hrs of cell culture in serum-free conditions yielded low ng per mL concentrations of protein (15-40 ng/mL, depending on experimental condition), a range that clearly requires concentration prior to MS analysis.

You could repack the column. Get the gel out of the column, give it a good washing, remove any fines. Make sure the filters and lines are clean. Then pack the gel into the column again, following manufacturers instruction (https://www.sigmaaldrich.com/content/dam/sigma-aldrich/docs/Sigma-Aldrich/General_Information/1/ge-gel-filtration.pdf). This will take a whole day and require a gel reservoire that matches the column.

*Mohamed's comment above is misleading and inaccurate.

There is no such thing as a true Universal Detector for HPLC (If there was one, RID would not be it.). RID is close and sometimes you will find misinformation about RID being "universal" (especially in vendor product literature) because it only relies on one physical property for detection (refractive index). However, as your sample's RI approaches the RI of the mobile phase, the sensitivity gets lower until it is finally zero. So that is not a Universal detector at all.

For lipid analysis, there are thousands of papers and articles showing the use of RID, ELSD, CAD and MS units for detection. ELSD is by far the most popular and RID is the simplest to use, least expensive and most limiting (very poor sensitivity).

Try a web search for example application notes and papers.

Try to change the prefilter of the column. It is on the inlet of the column.

we used different HPLC machine for the whole project. Started everything from scratch and completed the project successfully . Thank you all for sharing your expertice

HILIC is not a column (though some vendors may place a HILIC label on it, they are only doing so to fool you), it is a mode of chromatography.

Column length alone has nothing to do with suitability for LC-MS analysis. First, develop the HPLC method and make sure it is compatible with the detector type and ion mode desired (e.g. ESI pos/neg; highest purity volatile additives and mobile phase ). For most types of MS systems, highest sensitivity will be obtained with the lowest flow rates. If sensitivity or LOD are not critical, then even 1.00 ml/min rates will be fine, otherwise, low flow rates (depending on the exact MS type) are preferred so the use of narrow ID columns helps to achieve this by maintaining linear velocity (e.g. for conventional methods) without loss of efficiency. You should get some help selecting the column when you first start to develop the actual method used. If LC-MS (ESI) is the eventual goal and sensitivity is important, then start with a column type and dimension which result in the needed separation (Have someone experienced in LC-MS method development help you). You will not know what the ideal column dimensions are until AFTER you start to develop the method. Work out the method on an HPLC system, then scale it ) to the LC-MS system so it works well (change column dimensions, flow rate, gradient conditions, inj volume and so on, as needed).

Hi Mel, maybe this will be helpful to you:

Greetings Markus

It's a solid in crystalline form that can be dissolved in highly organic solvent.  I was thinking to do a melting point analysis on just the solid.  

 There is no need to measure total antioxidants content by HPLC. There are a number pf simple photometric tests, e.g. DPPH and etc. Why you want to use HPLC?  

Hello,

I am sure you are aware of MaConda: http://www.maconda.bham.ac.uk/downloads.php and the data downloadable in .csv etc. formats for specific contaminants  and the paper: https://academic.oup.com/bioinformatics/article/28/21/2856/236679/MaConDa-a-publicly-accessible-mass-spectrometry

Would be good to write to Ralf Weber for the source files.

Unsure if people put such data on DBs as .raw files but then you can go to the "Blank" files in submitted large scale metabolomics studies at MetaboLights:http://www.ebi.ac.uk/metabolights/  or MetabolomicsWorkBench:http://www.metabolomicsworkbench.org/  for example: http://www.ebi.ac.uk/metabolights/MTBLS218#protocols and so on.....

Because the "Blank" or "Solvent runs" in such untargeted data sets would constitute of impurities of various sources during sample preparation, chromatography and mass-spec runs.

Thanks,

Biswa

My printer does not work in the laboratory. Converting min to mm does provide a number only

High pH (>pH 9) mobile phases have proven useful for efficient retention and resolution of ionizable compounds especially in combination with negative MS. I suggest you look up studies by

David Watson et al. at Strathclyde, UK.

Journal of Chromatography A, 1362 (2014) 168–179

You may also look at the study by Lucy et al.

Journal of Chromatography A, 1458 (2016) 82–89

and Teleki and co-workers

Analytical Biochemistry 475 (2015) 4–13

You will need to work with a basematerial that is pH stable at high pH and here is the polymeric phases recommended (keep in mind though that a polymeric particle is more sensitive to over-pressure and you should make sure not to work at pressures over 200 bar). You should also keep in mind that charged phases are pH dependent and they will alter their behavior with pH, thus not only your ionizable analytes. As a recommendation for buffer salt, a carbonate system will provide better buffer capacity than using only ammonia/ammonium hydroxide but both have been used in the literature.

Be sure both your mobile phases and samples are ultrafiltered (~0.1 um).