To diagnose the problem, either wash the column and rerun a test w/o a sample or remove the column and place a restriction capillary in the same position. Your pump is loosing prime at that time. Probably a sticking check valve, bad proportioning valve, leaking piston or piston seal(s). You did not provide any information about the HPLC system used. Is this a high pressure binary pump or low pressure mixing pump with proportioning valve? *Monitor backpressure with the % change over time and run a ramp from 5% to 95% with a 10 minute hold. Fix the pump and then re-run the gradient without a sample to confirm correct operation.

Separate Items: What you refer to as "flushing" between 20 and 30 minutes should be what we call a "hold time" where the composition is left unchanged for ten minutes (not "flushing" which is different). END the method at 30 minutes. The actual flushing and re-equilibration should be separate steps, not part of the analysis method (a separate method to wash the column and get it ready for the next sample.) Please don't ramp down to 10% ACN in 0.2 minutes anly only give it 5 minutes to re-equilibrate (that is too quick a drop and not enough time for a re-equilibration)

At which wavelength was your chromatogramm measured? Triethalamin has an extinction coefficient of 400 /cm at 240 nm. Perhaps these artefact is produced by low light intensity at the beginning of your chromatogramm and refractive change of the eluent during the gradient.

@ Mr. Letter:

Im currently using a Shimadzu LC10 System. It has a high pressure (not UHPLC) trinary pump system of which i am only using 2 for the binary gradient. I have checked for leakage and found no problem. I have also run a stepwise gradient to observe the pressure/baseline: With increasing concentration the Absorbance also changes according to the percentage of ACN. When changing from 50 to 70 % of ACN the Baseline starts to increase and after having reached the expected plateau drops down as shown in the graph. After the drop it increases to the plateau of 70% ACN again.

@ Mr. Elend: I have measured the absorbance at 220 nm at which i'm usually measuring my samples. I've also looked for the absorbance of TEA at 220 nm and recognized it being quite close to the maximum. Since the buffer concentration is the same in ACN and the dd Water I didn't expect such a big increase of baseline as I wittnesed. Nevertheless this does not explain the massive drop while ramping to 100% of ACN.

Hello Ferenc Liedl,

try a much lower concentration of buffer or use a wavelength higher than 250nm. It is possible that your UV-signal is not 220nm. Normal HPLC-UV detectors with deuterium lamps do not have special stray-light filters like UV-spectrometers so if you have total absorption (Extinction >3) the detector monitors the stray-light from other wavelenght from zero order or second order of the holographic grid-mirror. So if the refractive index changes of your eluent (H2O to CH3CN) there are surprising signals on your detector. Before the chromatogramm starts the system makes a zero baseline.

Lots of problems and also missing information. Sounds like this may be one of your first trips into the world of HPLC. It has a steep learning curve for newbies as it takes years to learn. We rely on basic information to assist in these matters (details of method).

First, mixtures of ACN vs water will show a large backpressure change when the ACN reaches ~ 55%. *Run a 5% ACN to 95% ACN gradient at 225/10nm w/o additives and obs the back pressure (This effect is far more dramatic with MeOH in place of ACN). This is a normal change, but does not account 100% for your observed data, and also, it should show a more Gaussian profile over time as the pressure slowly changes, maxes and then drops off. Yours does not do this. Your data shows an abrupt change which normally indicates a mechanical loss of prime problem (as I addressed earlier). You need to verify correct operation of your pump using standard methods as it is not working correctly  (pressure seal test, gradient mix test and so on. Links below with troubleshooting examples). I am not aware of any Shimadzu LC10 pump which is ternary and of a high pressure mixing design. I think you have a low-pressure mixing, ternary system so I would check the pump's proportioning valve for problems (as mentioned earlier). "UHPLC" has nothing to do with what we are discussing.

The wavelength used of 220 nm is definitely a problem by itself. You did not specify the bandwidth used, which is a critical #. I hope it is less than 8nm (see link for more info). You are going to pick up so much interference using that low wavelength with an absorbing compound IN the mobile phase. Perhaps you could choose a better buffering system (see link)?

There are many issues with the method. Is there someone with practical experience running and using an HPLC system at your school? I am sure that they could provide you with on-site assistance to work this out. These types of problems are very difficult to solve over the web. Training is the key and you will learn faster with the help of someone local to train you and help you work through all of the basic troubleshooting steps to develop a better quality method and also service your HPLC system so it runs well.

http://hplctips.blogspot.com/2014/06/popular-hplc-volatile-mobile-phase.html

http://hplctips.blogspot.com/2014/01/diagnosing-troubleshooting-hplc.html

http://hplctips.blogspot.com/2011/09/uv-vis-hplc-detector-signal-bandwidth.html

http://hplctips.blogspot.com/2014/05/gradient-mixing-test-for-your-hplc-pump.html

Background: I'm trying to establish a buffered mobile Phase on our HPLC System since im actually a synthetic chemist who wants to create reproducible retentiontimes for kinetic measurements with molecules bearing pH-sensitive groups (in my case phosphates). I noticed that most people use the TEAA buffer for oligonucleotide synthesis in the literature (mostly 100 mM). 

@Mr. Elend: I've tried reducing the buffer concentration down to 20 mM by diluting the stock solution. Since I'm using a DAD I've also observed the baseline at 254 and 280 nm. At 254 nm I see the same Pattern but much weaker (i assume because I'm measuring further from the buffer substances absorption maximum). At 280 nm the baseline is flat. I also agree with your theory of my Detector having no stray light filter, since it's not one of the state of the art devices.

@Mr. Letter: Running another gradient works perfectly fine. For example: Starting at 10% ACN, ramping to 95 % of ACN over 10 minutes with a subsequent hold time of 10 minutes W/O buffer shows almost a flat baseline at 220 nm (just the usual UV Drift from the increasing concentration of ACN which is very small). Running the same gradient WITH the buffer shows the drop even in the other method. Also looking at bufferless mobile phase's retention times, I notice a narrow reproducibility due to pH-sensitive groups as mentioned in the beginning. At my side i have a trained HPLC teacher who also has never witnessed such a behaviour. She has a lot of experience in the field of HPLC unfurtonately not in using buffered mobile phases. Usually we use acidic mobile phases to sharpen the peaks. Unfortunately my molecules do not like the acidic environment and decompose. I will read through your links.

I thank both of you for your detailed answers and your time! If you have additional thoughts on my subject let me know. Ill update this post as soon as i got any new results on the problem.